From 7ce29019f7b42403d2f9d9122e48bfda252ad05e Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Wed, 1 Apr 2026 15:06:26 +0800 Subject: [PATCH 001/245] feat(memory): Add memory config API controller and end user info endpoints - Create new memory_config_api_controller.py for dedicated memory configuration management - Add /end_user/info GET endpoint to retrieve end user information (aliases, metadata) - Add /end_user/info/update POST endpoint to update end user details - Move /memory/configs endpoint from memory_api_controller to memory_config_api_controller - Extract _get_current_user helper function to build user context from API key auth - Support optional app_id parameter in end user creation with UUID validation - Update service controller imports with alphabetical ordering and multi-line formatting - Register memory_config_api_controller router in service module initialization - Refactor memory_api_controller imports for consistency and clarity --- api/app/controllers/service/__init__.py | 13 +- .../service/end_user_api_controller.py | 80 ++++++- .../service/memory_api_controller.py | 60 +----- .../service/memory_config_api_controller.py | 39 ++++ .../storage_services/search/__init__.py | 197 ++++++++---------- api/app/schemas/memory_api_schema.py | 2 + api/app/services/memory_api_service.py | 3 +- 7 files changed, 219 insertions(+), 175 deletions(-) create mode 100644 api/app/controllers/service/memory_config_api_controller.py diff --git a/api/app/controllers/service/__init__.py b/api/app/controllers/service/__init__.py index 96da0949..52d4b732 100644 --- a/api/app/controllers/service/__init__.py +++ b/api/app/controllers/service/__init__.py @@ -4,7 +4,17 @@ 认证方式: API Key """ from fastapi import APIRouter -from . import app_api_controller, rag_api_knowledge_controller, rag_api_document_controller, rag_api_file_controller, rag_api_chunk_controller, memory_api_controller, end_user_api_controller + +from . import ( + app_api_controller, + end_user_api_controller, + memory_api_controller, + memory_config_api_controller, + rag_api_chunk_controller, + rag_api_document_controller, + rag_api_file_controller, + rag_api_knowledge_controller, +) # 创建 V1 API 路由器 service_router = APIRouter() @@ -17,5 +27,6 @@ service_router.include_router(rag_api_file_controller.router) service_router.include_router(rag_api_chunk_controller.router) service_router.include_router(memory_api_controller.router) service_router.include_router(end_user_api_controller.router) +service_router.include_router(memory_config_api_controller.router) __all__ = ["service_router"] diff --git a/api/app/controllers/service/end_user_api_controller.py b/api/app/controllers/service/end_user_api_controller.py index 9d410bd2..658cef15 100644 --- a/api/app/controllers/service/end_user_api_controller.py +++ b/api/app/controllers/service/end_user_api_controller.py @@ -5,6 +5,7 @@ import uuid from fastapi import APIRouter, Body, Depends, Request from sqlalchemy.orm import Session +from app.controllers import user_memory_controllers from app.core.api_key_auth import require_api_key from app.core.error_codes import BizCode from app.core.exceptions import BusinessException @@ -13,13 +14,31 @@ from app.core.response_utils import success from app.db import get_db from app.repositories.end_user_repository import EndUserRepository from app.schemas.api_key_schema import ApiKeyAuth +from app.schemas.end_user_info_schema import EndUserInfoUpdate from app.schemas.memory_api_schema import CreateEndUserRequest, CreateEndUserResponse +from app.services import api_key_service from app.services.memory_config_service import MemoryConfigService router = APIRouter(prefix="/end_user", tags=["V1 - End User API"]) logger = get_business_logger() +def _get_current_user(api_key_auth: ApiKeyAuth, db: Session): + """Build a current_user object from API key auth + + Args: + api_key_auth: Validated API key auth info + db: Database session + + Returns: + User object with current_workspace_id set + """ + api_key = api_key_service.ApiKeyService.get_api_key(db, api_key_auth.api_key_id, api_key_auth.workspace_id) + current_user = api_key.creator + current_user.current_workspace_id = api_key_auth.workspace_id + return current_user + + @router.post("/create") @require_api_key(scopes=["memory"]) async def create_end_user( @@ -37,6 +56,7 @@ async def create_end_user( Optionally accepts a memory_config_id to connect the end user to a specific memory configuration. If not provided, falls back to the workspace default config. + Optionally accepts an app_id to bind the end user to a specific app. """ body = await request.json() payload = CreateEndUserRequest(**body) @@ -71,9 +91,20 @@ async def create_end_user( else: logger.warning(f"No default memory config found for workspace: {workspace_id}") + # Resolve app_id: explicit from payload, otherwise None + app_id = None + if payload.app_id: + try: + app_id = uuid.UUID(payload.app_id) + except ValueError: + raise BusinessException( + f"Invalid app_id format: {payload.app_id}", + BizCode.INVALID_PARAMETER + ) + end_user_repo = EndUserRepository(db) end_user = end_user_repo.get_or_create_end_user_with_config( - app_id=api_key_auth.resource_id, + app_id=app_id, workspace_id=workspace_id, other_id=payload.other_id, memory_config_id=memory_config_id, @@ -90,3 +121,50 @@ async def create_end_user( } return success(data=CreateEndUserResponse(**result).model_dump(), msg="End user created successfully") + + +@router.get("/info") +@require_api_key(scopes=["memory"]) +async def get_end_user_info( + request: Request, + end_user_id: str, + api_key_auth: ApiKeyAuth = None, + db: Session = Depends(get_db), +): + """ + Get end user info. + + Retrieves the info record (aliases, meta_data, etc.) for the specified end user. + Delegates to the manager-side controller for shared logic. + """ + current_user = _get_current_user(api_key_auth, db) + return await user_memory_controllers.get_end_user_info( + end_user_id=end_user_id, + current_user=current_user, + db=db, + ) + + +@router.post("/info/update") +@require_api_key(scopes=["memory"]) +async def update_end_user_info( + request: Request, + api_key_auth: ApiKeyAuth = None, + db: Session = Depends(get_db), + message: str = Body(None, description="Request body"), +): + """ + Update end user info. + + Updates the info record (other_name, aliases, meta_data) for the specified end user. + Delegates to the manager-side controller for shared logic. + """ + body = await request.json() + payload = EndUserInfoUpdate(**body) + + current_user = _get_current_user(api_key_auth, db) + return await user_memory_controllers.update_end_user_info( + info_update=payload, + current_user=current_user, + db=db, + ) diff --git a/api/app/controllers/service/memory_api_controller.py b/api/app/controllers/service/memory_api_controller.py index dc5e0408..d1229205 100644 --- a/api/app/controllers/service/memory_api_controller.py +++ b/api/app/controllers/service/memory_api_controller.py @@ -1,22 +1,20 @@ """Memory 服务接口 - 基于 API Key 认证""" +from fastapi import APIRouter, Body, Depends, Request +from sqlalchemy.orm import Session + from app.core.api_key_auth import require_api_key from app.core.logging_config import get_business_logger from app.core.response_utils import success from app.db import get_db from app.schemas.api_key_schema import ApiKeyAuth from app.schemas.memory_api_schema import ( - CreateEndUserRequest, - CreateEndUserResponse, - ListConfigsResponse, MemoryReadRequest, MemoryReadResponse, MemoryWriteRequest, MemoryWriteResponse, ) from app.services.memory_api_service import MemoryAPIService -from fastapi import APIRouter, Body, Depends, Request -from sqlalchemy.orm import Session router = APIRouter(prefix="/memory", tags=["V1 - Memory API"]) logger = get_business_logger() @@ -91,55 +89,3 @@ async def read_memory_api_service( logger.info(f"Memory read successful for end_user: {payload.end_user_id}") return success(data=MemoryReadResponse(**result).model_dump(), msg="Memory read successfully") - - -@router.get("/configs") -@require_api_key(scopes=["memory"]) -async def list_memory_configs( - request: Request, - api_key_auth: ApiKeyAuth = None, - db: Session = Depends(get_db), -): - """ - List all memory configs for the workspace. - - Returns all available memory configurations associated with the authorized workspace. - """ - logger.info(f"List configs request - workspace_id: {api_key_auth.workspace_id}") - - memory_api_service = MemoryAPIService(db) - - result = memory_api_service.list_memory_configs( - workspace_id=api_key_auth.workspace_id, - ) - - logger.info(f"Listed {result['total']} configs for workspace: {api_key_auth.workspace_id}") - return success(data=ListConfigsResponse(**result).model_dump(), msg="Configs listed successfully") - - -@router.post("/end_users") -@require_api_key(scopes=["memory"]) -async def create_end_user( - request: Request, - api_key_auth: ApiKeyAuth = None, - db: Session = Depends(get_db), -): - """ - Create an end user. - - Creates a new end user for the authorized workspace. - If an end user with the same other_id already exists, returns the existing one. - """ - body = await request.json() - payload = CreateEndUserRequest(**body) - logger.info(f"Create end user request - other_id: {payload.other_id}, workspace_id: {api_key_auth.workspace_id}") - - memory_api_service = MemoryAPIService(db) - - result = memory_api_service.create_end_user( - workspace_id=api_key_auth.workspace_id, - other_id=payload.other_id, - ) - - logger.info(f"End user ready: {result['id']}") - return success(data=CreateEndUserResponse(**result).model_dump(), msg="End user created successfully") diff --git a/api/app/controllers/service/memory_config_api_controller.py b/api/app/controllers/service/memory_config_api_controller.py new file mode 100644 index 00000000..bfc60510 --- /dev/null +++ b/api/app/controllers/service/memory_config_api_controller.py @@ -0,0 +1,39 @@ +"""Memory Config 服务接口 - 基于 API Key 认证""" + +from fastapi import APIRouter, Depends, Request +from sqlalchemy.orm import Session + +from app.core.api_key_auth import require_api_key +from app.core.logging_config import get_business_logger +from app.core.response_utils import success +from app.db import get_db +from app.schemas.api_key_schema import ApiKeyAuth +from app.schemas.memory_api_schema import ListConfigsResponse +from app.services.memory_api_service import MemoryAPIService + +router = APIRouter(prefix="/memory_config", tags=["V1 - Memory Config API"]) +logger = get_business_logger() + + +@router.get("/configs") +@require_api_key(scopes=["memory"]) +async def list_memory_configs( + request: Request, + api_key_auth: ApiKeyAuth = None, + db: Session = Depends(get_db), +): + """ + List all memory configs for the workspace. + + Returns all available memory configurations associated with the authorized workspace. + """ + logger.info(f"List configs request - workspace_id: {api_key_auth.workspace_id}") + + memory_api_service = MemoryAPIService(db) + + result = memory_api_service.list_memory_configs( + workspace_id=api_key_auth.workspace_id, + ) + + logger.info(f"Listed {result['total']} configs for workspace: {api_key_auth.workspace_id}") + return success(data=ListConfigsResponse(**result).model_dump(), msg="Configs listed successfully") diff --git a/api/app/core/memory/storage_services/search/__init__.py b/api/app/core/memory/storage_services/search/__init__.py index c12c39b0..49154e19 100644 --- a/api/app/core/memory/storage_services/search/__init__.py +++ b/api/app/core/memory/storage_services/search/__init__.py @@ -4,11 +4,6 @@ 本模块提供统一的搜索服务接口，支持关键词搜索、语义搜索和混合搜索。 """ -from typing import TYPE_CHECKING - -if TYPE_CHECKING: - from app.schemas.memory_config_schema import MemoryConfig - from app.core.memory.storage_services.search.hybrid_search import HybridSearchStrategy from app.core.memory.storage_services.search.keyword_search import KeywordSearchStrategy from app.core.memory.storage_services.search.search_strategy import ( @@ -29,115 +24,87 @@ __all__ = [ # ============================================================================ -# 向后兼容的函数式API +# 向后兼容的函数式API (DEPRECATED - 未被使用) # ============================================================================ -# 为了兼容旧代码，提供与 src/search.py 相同的函数式接口 +# 所有调用方均直接使用 app.core.memory.src.search.run_hybrid_search +# 保留注释以备参考 - -async def run_hybrid_search( - query_text: str, - search_type: str = "hybrid", - end_user_id: str | None = None, - apply_id: str | None = None, - user_id: str | None = None, - limit: int = 50, - include: list[str] | None = None, - alpha: float = 0.6, - use_forgetting_curve: bool = False, - memory_config: "MemoryConfig" = None, - **kwargs -) -> dict: - """运行混合搜索（向后兼容的函数式API） - - 这是一个向后兼容的包装函数，将旧的函数式API转换为新的基于类的API。 - - Args: - query_text: 查询文本 - search_type: 搜索类型（"hybrid", "keyword", "semantic"） - end_user_id: 组ID过滤 - apply_id: 应用ID过滤 - user_id: 用户ID过滤 - limit: 每个类别的最大结果数 - include: 要包含的搜索类别列表 - alpha: BM25分数权重（0.0-1.0） - use_forgetting_curve: 是否使用遗忘曲线 - memory_config: MemoryConfig object containing embedding_model_id - **kwargs: 其他参数 - - Returns: - dict: 搜索结果字典，格式与旧API兼容 - """ - from app.core.memory.llm_tools.openai_embedder import OpenAIEmbedderClient - from app.core.models.base import RedBearModelConfig - from app.db import get_db_context - from app.repositories.neo4j.neo4j_connector import Neo4jConnector - from app.services.memory_config_service import MemoryConfigService - - if not memory_config: - raise ValueError("memory_config is required for search") - - # 初始化客户端 - connector = Neo4jConnector() - with get_db_context() as db: - config_service = MemoryConfigService(db) - embedder_config_dict = config_service.get_embedder_config(str(memory_config.embedding_model_id)) - embedder_config = RedBearModelConfig(**embedder_config_dict) - embedder_client = OpenAIEmbedderClient(embedder_config) - - try: - # 根据搜索类型选择策略 - if search_type == "keyword": - strategy = KeywordSearchStrategy(connector=connector) - elif search_type == "semantic": - strategy = SemanticSearchStrategy( - connector=connector, - embedder_client=embedder_client - ) - else: # hybrid - strategy = HybridSearchStrategy( - connector=connector, - embedder_client=embedder_client, - alpha=alpha, - use_forgetting_curve=use_forgetting_curve - ) - - # 执行搜索 - result = await strategy.search( - query_text=query_text, - end_user_id=end_user_id, - limit=limit, - include=include, - alpha=alpha, - use_forgetting_curve=use_forgetting_curve, - **kwargs - ) - - # 转换为旧格式 - result_dict = result.to_dict() - - # 保存到文件（如果指定了output_path） - output_path = kwargs.get('output_path', 'search_results.json') - if output_path: - import json - import os - from datetime import datetime - - try: - # 确保目录存在 - out_dir = os.path.dirname(output_path) - if out_dir: - os.makedirs(out_dir, exist_ok=True) - - # 保存结果 - with open(output_path, "w", encoding="utf-8") as f: - json.dump(result_dict, f, ensure_ascii=False, indent=2, default=str) - print(f"Search results saved to {output_path}") - except Exception as e: - print(f"Error saving search results: {e}") - return result_dict - - finally: - await connector.close() - - -__all__.append("run_hybrid_search") +# async def run_hybrid_search( +# query_text: str, +# search_type: str = "hybrid", +# end_user_id: str | None = None, +# apply_id: str | None = None, +# user_id: str | None = None, +# limit: int = 50, +# include: list[str] | None = None, +# alpha: float = 0.6, +# use_forgetting_curve: bool = False, +# memory_config: "MemoryConfig" = None, +# **kwargs +# ) -> dict: +# """运行混合搜索（向后兼容的函数式API）""" +# from app.core.memory.llm_tools.openai_embedder import OpenAIEmbedderClient +# from app.core.models.base import RedBearModelConfig +# from app.db import get_db_context +# from app.repositories.neo4j.neo4j_connector import Neo4jConnector +# from app.services.memory_config_service import MemoryConfigService +# +# if not memory_config: +# raise ValueError("memory_config is required for search") +# +# connector = Neo4jConnector() +# with get_db_context() as db: +# config_service = MemoryConfigService(db) +# embedder_config_dict = config_service.get_embedder_config(str(memory_config.embedding_model_id)) +# embedder_config = RedBearModelConfig(**embedder_config_dict) +# embedder_client = OpenAIEmbedderClient(embedder_config) +# +# try: +# if search_type == "keyword": +# strategy = KeywordSearchStrategy(connector=connector) +# elif search_type == "semantic": +# strategy = SemanticSearchStrategy( +# connector=connector, +# embedder_client=embedder_client +# ) +# else: +# strategy = HybridSearchStrategy( +# connector=connector, +# embedder_client=embedder_client, +# alpha=alpha, +# use_forgetting_curve=use_forgetting_curve +# ) +# +# result = await strategy.search( +# query_text=query_text, +# end_user_id=end_user_id, +# limit=limit, +# include=include, +# alpha=alpha, +# use_forgetting_curve=use_forgetting_curve, +# **kwargs +# ) +# +# result_dict = result.to_dict() +# +# output_path = kwargs.get('output_path', 'search_results.json') +# if output_path: +# import json +# import os +# from datetime import datetime +# +# try: +# out_dir = os.path.dirname(output_path) +# if out_dir: +# os.makedirs(out_dir, exist_ok=True) +# with open(output_path, "w", encoding="utf-8") as f: +# json.dump(result_dict, f, ensure_ascii=False, indent=2, default=str) +# print(f"Search results saved to {output_path}") +# except Exception as e: +# print(f"Error saving search results: {e}") +# return result_dict +# +# finally: +# await connector.close() +# +# __all__.append("run_hybrid_search") diff --git a/api/app/schemas/memory_api_schema.py b/api/app/schemas/memory_api_schema.py index ff62355f..2229c540 100644 --- a/api/app/schemas/memory_api_schema.py +++ b/api/app/schemas/memory_api_schema.py @@ -141,10 +141,12 @@ class CreateEndUserRequest(BaseModel): other_id: External user identifier (required) other_name: Display name for the end user memory_config_id: Optional memory config ID. If not provided, uses workspace default. + app_id: Optional app ID to bind the end user to. """ other_id: str = Field(..., description="External user identifier (required)") other_name: Optional[str] = Field("", description="Display name") memory_config_id: Optional[str] = Field(None, description="Memory config ID. Falls back to workspace default if not provided.") + app_id: Optional[str] = Field(None, description="App ID to bind the end user to") @field_validator("other_id") @classmethod diff --git a/api/app/services/memory_api_service.py b/api/app/services/memory_api_service.py index f62f526c..bea313fc 100644 --- a/api/app/services/memory_api_service.py +++ b/api/app/services/memory_api_service.py @@ -8,6 +8,8 @@ This service validates inputs and delegates to MemoryAgentService for core memor import uuid from typing import Any, Dict, Optional +from sqlalchemy.orm import Session + from app.core.error_codes import BizCode from app.core.exceptions import BusinessException, ResourceNotFoundException from app.core.logging_config import get_logger @@ -15,7 +17,6 @@ from app.models.app_model import App from app.models.end_user_model import EndUser from app.schemas.memory_config_schema import ConfigurationError from app.services.memory_agent_service import MemoryAgentService -from sqlalchemy.orm import Session logger = get_logger(__name__) From 010eff17cfad8d7fa47568e493ef5932233f2950 Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Thu, 2 Apr 2026 14:47:36 +0800 Subject: [PATCH 002/245] feat(memory): Refactor memory API to support async task-based and sync operations - Rename endpoints from write_api_service/read_api_service to write/read for clarity - Add async task-based endpoints (/write, /read) that dispatch to Celery with fair locking - Add task status polling endpoints (/write/status, /read/status) to check async operation results - Add synchronous endpoints (/write/sync, /read/sync) for blocking operations with direct results - Introduce TaskStatusResponse schema for task status polling responses - Add MemoryWriteSyncResponse and MemoryReadSyncResponse schemas for sync operations - Implement write_memory_sync and read_memory_sync methods in MemoryAPIService - Remove await from async service calls in task-based endpoints (now handled by Celery) - Add Query parameter import for task_id in status endpoints - Update docstrings to clarify async vs sync behavior and task polling workflow - Integrate task_service for retrieving Celery task results --- .../service/memory_api_controller.py | 160 ++++++++++++--- api/app/schemas/memory_api_schema.py | 43 +++- api/app/services/memory_api_service.py | 186 ++++++++++++++---- 3 files changed, 320 insertions(+), 69 deletions(-) diff --git a/api/app/controllers/service/memory_api_controller.py b/api/app/controllers/service/memory_api_controller.py index d1229205..34135fec 100644 --- a/api/app/controllers/service/memory_api_controller.py +++ b/api/app/controllers/service/memory_api_controller.py @@ -1,6 +1,6 @@ """Memory 服务接口 - 基于 API Key 认证""" -from fastapi import APIRouter, Body, Depends, Request +from fastapi import APIRouter, Body, Depends, Query, Request from sqlalchemy.orm import Session from app.core.api_key_auth import require_api_key @@ -11,8 +11,11 @@ from app.schemas.api_key_schema import ApiKeyAuth from app.schemas.memory_api_schema import ( MemoryReadRequest, MemoryReadResponse, + MemoryReadSyncResponse, MemoryWriteRequest, MemoryWriteResponse, + MemoryWriteSyncResponse, + TaskStatusResponse, ) from app.services.memory_api_service import MemoryAPIService @@ -26,26 +29,27 @@ async def get_memory_info(): return success(data={}, msg="Memory API - Coming Soon") -@router.post("/write_api_service") +@router.post("/write") @require_api_key(scopes=["memory"]) -async def write_memory_api_service( +async def write_memory( request: Request, api_key_auth: ApiKeyAuth = None, db: Session = Depends(get_db), message: str = Body(..., description="Message content"), ): """ - Write memory to storage. - - Stores memory content for the specified end user using the Memory API Service. + Submit a memory write task. + + Validates the end user, then dispatches the write to a Celery background task + with per-user fair locking. Returns a task_id for status polling. """ body = await request.json() payload = MemoryWriteRequest(**body) logger.info(f"Memory write request - end_user_id: {payload.end_user_id}, workspace_id: {api_key_auth.workspace_id}") - + memory_api_service = MemoryAPIService(db) - - result = await memory_api_service.write_memory( + + result = memory_api_service.write_memory( workspace_id=api_key_auth.workspace_id, end_user_id=payload.end_user_id, message=payload.message, @@ -53,31 +57,53 @@ async def write_memory_api_service( storage_type=payload.storage_type, user_rag_memory_id=payload.user_rag_memory_id, ) - - logger.info(f"Memory write successful for end_user: {payload.end_user_id}") - return success(data=MemoryWriteResponse(**result).model_dump(), msg="Memory written successfully") + + logger.info(f"Memory write task submitted: task_id={result['task_id']}, end_user_id: {payload.end_user_id}") + return success(data=MemoryWriteResponse(**result).model_dump(), msg="Memory write task submitted") -@router.post("/read_api_service") +@router.get("/write/status") @require_api_key(scopes=["memory"]) -async def read_memory_api_service( +async def get_write_task_status( + request: Request, + task_id: str = Query(..., description="Celery task ID"), + api_key_auth: ApiKeyAuth = None, + db: Session = Depends(get_db), +): + """ + Check the status of a memory write task. + + Returns the current status and result (if completed) of a previously submitted write task. + """ + logger.info(f"Write task status check - task_id: {task_id}") + + from app.services.task_service import get_task_memory_write_result + result = get_task_memory_write_result(task_id) + + return success(data=TaskStatusResponse(**result).model_dump(), msg="Task status retrieved") + + +@router.post("/read") +@require_api_key(scopes=["memory"]) +async def read_memory( request: Request, api_key_auth: ApiKeyAuth = None, db: Session = Depends(get_db), message: str = Body(..., description="Query message"), ): """ - Read memory from storage. - - Queries and retrieves memories for the specified end user with context-aware responses. + Submit a memory read task. + + Validates the end user, then dispatches the read to a Celery background task. + Returns a task_id for status polling. """ body = await request.json() payload = MemoryReadRequest(**body) logger.info(f"Memory read request - end_user_id: {payload.end_user_id}") - + memory_api_service = MemoryAPIService(db) - - result = await memory_api_service.read_memory( + + result = memory_api_service.read_memory( workspace_id=api_key_auth.workspace_id, end_user_id=payload.end_user_id, message=payload.message, @@ -86,6 +112,94 @@ async def read_memory_api_service( storage_type=payload.storage_type, user_rag_memory_id=payload.user_rag_memory_id, ) - - logger.info(f"Memory read successful for end_user: {payload.end_user_id}") - return success(data=MemoryReadResponse(**result).model_dump(), msg="Memory read successfully") + + logger.info(f"Memory read task submitted: task_id={result['task_id']}, end_user_id: {payload.end_user_id}") + return success(data=MemoryReadResponse(**result).model_dump(), msg="Memory read task submitted") + + +@router.get("/read/status") +@require_api_key(scopes=["memory"]) +async def get_read_task_status( + request: Request, + task_id: str = Query(..., description="Celery task ID"), + api_key_auth: ApiKeyAuth = None, + db: Session = Depends(get_db), +): + """ + Check the status of a memory read task. + + Returns the current status and result (if completed) of a previously submitted read task. + """ + logger.info(f"Read task status check - task_id: {task_id}") + + from app.services.task_service import get_task_memory_read_result + result = get_task_memory_read_result(task_id) + + return success(data=TaskStatusResponse(**result).model_dump(), msg="Task status retrieved") + + +@router.post("/write/sync") +@require_api_key(scopes=["memory"]) +async def write_memory_sync( + request: Request, + api_key_auth: ApiKeyAuth = None, + db: Session = Depends(get_db), + message: str = Body(..., description="Message content"), +): + """ + Write memory synchronously. + + Blocks until the write completes and returns the result directly. + For async processing with task polling, use /write instead. + """ + body = await request.json() + payload = MemoryWriteRequest(**body) + logger.info(f"Memory write (sync) request - end_user_id: {payload.end_user_id}") + + memory_api_service = MemoryAPIService(db) + + result = await memory_api_service.write_memory_sync( + workspace_id=api_key_auth.workspace_id, + end_user_id=payload.end_user_id, + message=payload.message, + config_id=payload.config_id, + storage_type=payload.storage_type, + user_rag_memory_id=payload.user_rag_memory_id, + ) + + logger.info(f"Memory write (sync) successful for end_user: {payload.end_user_id}") + return success(data=MemoryWriteSyncResponse(**result).model_dump(), msg="Memory written successfully") + + +@router.post("/read/sync") +@require_api_key(scopes=["memory"]) +async def read_memory_sync( + request: Request, + api_key_auth: ApiKeyAuth = None, + db: Session = Depends(get_db), + message: str = Body(..., description="Query message"), +): + """ + Read memory synchronously. + + Blocks until the read completes and returns the answer directly. + For async processing with task polling, use /read instead. + """ + body = await request.json() + payload = MemoryReadRequest(**body) + logger.info(f"Memory read (sync) request - end_user_id: {payload.end_user_id}") + + memory_api_service = MemoryAPIService(db) + + result = await memory_api_service.read_memory_sync( + workspace_id=api_key_auth.workspace_id, + end_user_id=payload.end_user_id, + message=payload.message, + search_switch=payload.search_switch, + config_id=payload.config_id, + storage_type=payload.storage_type, + user_rag_memory_id=payload.user_rag_memory_id, + ) + + logger.info(f"Memory read (sync) successful for end_user: {payload.end_user_id}") + return success(data=MemoryReadSyncResponse(**result).model_dump(), msg="Memory read successfully") diff --git a/api/app/schemas/memory_api_schema.py b/api/app/schemas/memory_api_schema.py index 2229c540..4523d02d 100644 --- a/api/app/schemas/memory_api_schema.py +++ b/api/app/schemas/memory_api_schema.py @@ -110,6 +110,30 @@ class MemoryReadRequest(BaseModel): class MemoryWriteResponse(BaseModel): """Response schema for memory write operation. + Attributes: + task_id: Celery task ID for status polling + status: Initial task status (PENDING) + end_user_id: End user ID the write was submitted for + """ + task_id: str = Field(..., description="Celery task ID for polling") + status: str = Field(..., description="Task status: PENDING") + end_user_id: str = Field(..., description="End user ID") + + +class TaskStatusResponse(BaseModel): + """Response schema for task status check. + + Attributes: + status: Task status (PENDING, STARTED, SUCCESS, FAILURE, SKIPPED) + result: Task result data (available when status is SUCCESS or FAILURE) + """ + status: str = Field(..., description="Task status") + result: Optional[Dict[str, Any]] = Field(None, description="Task result when completed") + + +class MemoryWriteSyncResponse(BaseModel): + """Response schema for synchronous memory write. + Attributes: status: Operation status (success or failed) end_user_id: End user ID that was written to @@ -118,8 +142,8 @@ class MemoryWriteResponse(BaseModel): end_user_id: str = Field(..., description="End user ID") -class MemoryReadResponse(BaseModel): - """Response schema for memory read operation. +class MemoryReadSyncResponse(BaseModel): + """Response schema for synchronous memory read. Attributes: answer: Generated answer from memory retrieval @@ -128,12 +152,25 @@ class MemoryReadResponse(BaseModel): """ answer: str = Field(..., description="Generated answer") intermediate_outputs: List[Dict[str, Any]] = Field( - default_factory=list, + default_factory=list, description="Intermediate retrieval outputs" ) end_user_id: str = Field(..., description="End user ID") +class MemoryReadResponse(BaseModel): + """Response schema for memory read operation. + + Attributes: + task_id: Celery task ID for status polling + status: Initial task status (PENDING) + end_user_id: End user ID the read was submitted for + """ + task_id: str = Field(..., description="Celery task ID for polling") + status: str = Field(..., description="Task status: PENDING") + end_user_id: str = Field(..., description="End user ID") + + class CreateEndUserRequest(BaseModel): """Request schema for creating an end user. diff --git a/api/app/services/memory_api_service.py b/api/app/services/memory_api_service.py index bea313fc..330b84ad 100644 --- a/api/app/services/memory_api_service.py +++ b/api/app/services/memory_api_service.py @@ -125,7 +125,7 @@ class MemoryAPIService: except Exception as e: logger.warning(f"Failed to update memory_config_id for end_user {end_user_id}: {e}") - async def write_memory( + def write_memory( self, workspace_id: uuid.UUID, end_user_id: str, @@ -134,27 +134,28 @@ class MemoryAPIService: storage_type: str = "neo4j", user_rag_memory_id: Optional[str] = None, ) -> Dict[str, Any]: - """Write memory with validation. - + """Submit a memory write task via Celery. + Validates end_user exists and belongs to workspace, updates the end user's - memory_config_id, then delegates to MemoryAgentService.write_memory. - + memory_config_id, then dispatches write_message_task to Celery for async + processing with per-user fair locking. + Args: workspace_id: Workspace ID for resource validation - end_user_id: End user identifier (used as end_user_id) + end_user_id: End user identifier message: Message content to store config_id: Memory configuration ID (required) storage_type: Storage backend (neo4j or rag) user_rag_memory_id: Optional RAG memory ID - + Returns: - Dict with status and end_user_id - + Dict with task_id, status, and end_user_id + Raises: ResourceNotFoundException: If end_user not found - BusinessException: If end_user not in authorized workspace or write fails + BusinessException: If validation fails """ - logger.info(f"Writing memory for end_user: {end_user_id}, workspace: {workspace_id}") + logger.info(f"Submitting memory write for end_user: {end_user_id}, workspace: {workspace_id}") # Validate end_user exists and belongs to workspace self.validate_end_user(end_user_id, workspace_id) @@ -162,9 +163,120 @@ class MemoryAPIService: # Update end user's memory_config_id self._update_end_user_config(end_user_id, config_id) + # Convert to message list format expected by write_message_task + messages = message if isinstance(message, list) else [{"role": "user", "content": message}] + + from app.tasks import write_message_task + task = write_message_task.delay( + end_user_id, + messages, + config_id, + storage_type, + user_rag_memory_id or "", + ) + + logger.info(f"Memory write task submitted: task_id={task.id}, end_user_id={end_user_id}") + + return { + "task_id": task.id, + "status": "PENDING", + "end_user_id": end_user_id, + } + + def read_memory( + self, + workspace_id: uuid.UUID, + end_user_id: str, + message: str, + search_switch: str = "0", + config_id: str = "", + storage_type: str = "neo4j", + user_rag_memory_id: Optional[str] = None, + ) -> Dict[str, Any]: + """Submit a memory read task via Celery. + + Validates end_user exists and belongs to workspace, updates the end user's + memory_config_id, then dispatches read_message_task to Celery for async processing. + + Args: + workspace_id: Workspace ID for resource validation + end_user_id: End user identifier + message: Query message + search_switch: Search mode (0=deep search with verification, 1=deep search, 2=fast search) + config_id: Memory configuration ID (required) + storage_type: Storage backend (neo4j or rag) + user_rag_memory_id: Optional RAG memory ID + + Returns: + Dict with task_id, status, and end_user_id + + Raises: + ResourceNotFoundException: If end_user not found + BusinessException: If validation fails + """ + logger.info(f"Submitting memory read for end_user: {end_user_id}, workspace: {workspace_id}") + + # Validate end_user exists and belongs to workspace + self.validate_end_user(end_user_id, workspace_id) + + # Update end user's memory_config_id + self._update_end_user_config(end_user_id, config_id) + + from app.tasks import read_message_task + task = read_message_task.delay( + end_user_id, + message, + [], # history + search_switch, + config_id, + storage_type, + user_rag_memory_id or "", + ) + + logger.info(f"Memory read task submitted: task_id={task.id}, end_user_id={end_user_id}") + + return { + "task_id": task.id, + "status": "PENDING", + "end_user_id": end_user_id, + } + + async def write_memory_sync( + self, + workspace_id: uuid.UUID, + end_user_id: str, + message: str, + config_id: str, + storage_type: str = "neo4j", + user_rag_memory_id: Optional[str] = None, + ) -> Dict[str, Any]: + """Write memory synchronously (inline, no Celery). + + Validates end_user, then calls MemoryAgentService.write_memory directly. + Blocks until the write completes. Use for cases where the caller needs + immediate confirmation. + + Args: + workspace_id: Workspace ID for resource validation + end_user_id: End user identifier + message: Message content to store + config_id: Memory configuration ID (required) + storage_type: Storage backend (neo4j or rag) + user_rag_memory_id: Optional RAG memory ID + + Returns: + Dict with status and end_user_id + + Raises: + ResourceNotFoundException: If end_user not found + BusinessException: If write fails + """ + logger.info(f"Writing memory (sync) for end_user: {end_user_id}, workspace: {workspace_id}") + + self.validate_end_user(end_user_id, workspace_id) + self._update_end_user_config(end_user_id, config_id) + try: - # Delegate to MemoryAgentService - # Convert string message to list[dict] format expected by MemoryAgentService messages = message if isinstance(message, list) else [{"role": "user", "content": message}] result = await MemoryAgentService().write_memory( end_user_id=end_user_id, @@ -175,11 +287,8 @@ class MemoryAPIService: user_rag_memory_id=user_rag_memory_id or "", ) - logger.info(f"Memory write successful for end_user: {end_user_id}") + logger.info(f"Memory write (sync) successful for end_user: {end_user_id}") - # result may be a string "success" or a dict with a "status" key - # Preserve the full dict so callers don't silently lose extra fields - # (e.g. error codes, metadata) returned by MemoryAgentService. if isinstance(result, dict): return { **result, @@ -193,20 +302,17 @@ class MemoryAPIService: except ConfigurationError as e: logger.error(f"Memory configuration error for end_user {end_user_id}: {e}") - raise BusinessException( - message=str(e), - code=BizCode.MEMORY_CONFIG_NOT_FOUND - ) + raise BusinessException(message=str(e), code=BizCode.MEMORY_CONFIG_NOT_FOUND) except BusinessException: raise except Exception as e: - logger.error(f"Memory write failed for end_user {end_user_id}: {e}") + logger.error(f"Memory write (sync) failed for end_user {end_user_id}: {e}") raise BusinessException( message=f"Memory write failed: {str(e)}", code=BizCode.MEMORY_WRITE_FAILED ) - async def read_memory( + async def read_memory_sync( self, workspace_id: uuid.UUID, end_user_id: str, @@ -216,37 +322,34 @@ class MemoryAPIService: storage_type: str = "neo4j", user_rag_memory_id: Optional[str] = None, ) -> Dict[str, Any]: - """Read memory with validation. - - Validates end_user exists and belongs to workspace, updates the end user's - memory_config_id, then delegates to MemoryAgentService.read_memory. - + """Read memory synchronously (inline, no Celery). + + Validates end_user, then calls MemoryAgentService.read_memory directly. + Blocks until the read completes. Use for cases where the caller needs + the answer immediately. + Args: workspace_id: Workspace ID for resource validation - end_user_id: End user identifier (used as end_user_id) + end_user_id: End user identifier message: Query message search_switch: Search mode (0=deep search with verification, 1=deep search, 2=fast search) config_id: Memory configuration ID (required) storage_type: Storage backend (neo4j or rag) user_rag_memory_id: Optional RAG memory ID - + Returns: Dict with answer, intermediate_outputs, and end_user_id - + Raises: ResourceNotFoundException: If end_user not found - BusinessException: If end_user not in authorized workspace or read fails + BusinessException: If read fails """ - logger.info(f"Reading memory for end_user: {end_user_id}, workspace: {workspace_id}") + logger.info(f"Reading memory (sync) for end_user: {end_user_id}, workspace: {workspace_id}") - # Validate end_user exists and belongs to workspace self.validate_end_user(end_user_id, workspace_id) - - # Update end user's memory_config_id self._update_end_user_config(end_user_id, config_id) try: - # Delegate to MemoryAgentService result = await MemoryAgentService().read_memory( end_user_id=end_user_id, message=message, @@ -258,7 +361,7 @@ class MemoryAPIService: user_rag_memory_id=user_rag_memory_id or "" ) - logger.info(f"Memory read successful for end_user: {end_user_id}") + logger.info(f"Memory read (sync) successful for end_user: {end_user_id}") return { "answer": result.get("answer", ""), @@ -268,14 +371,11 @@ class MemoryAPIService: except ConfigurationError as e: logger.error(f"Memory configuration error for end_user {end_user_id}: {e}") - raise BusinessException( - message=str(e), - code=BizCode.MEMORY_CONFIG_NOT_FOUND - ) + raise BusinessException(message=str(e), code=BizCode.MEMORY_CONFIG_NOT_FOUND) except BusinessException: raise except Exception as e: - logger.error(f"Memory read failed for end_user {end_user_id}: {e}") + logger.error(f"Memory read (sync) failed for end_user {end_user_id}: {e}") raise BusinessException( message=f"Memory read failed: {str(e)}", code=BizCode.MEMORY_READ_FAILED From 2f0bb793d8ea48cabea42e478c04b4ea71140e02 Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Thu, 2 Apr 2026 14:49:46 +0800 Subject: [PATCH 003/245] feat(memory): Add task result sanitization for JSON serialization - Remove unused TaskStatusResponse import from memory_api_schema - Add _sanitize_task_result() helper function to convert non-serializable types (UUID, datetime) to strings - Update get_write_task_status endpoint to use sanitization instead of TaskStatusResponse validation - Update get_read_task_status endpoint to use sanitization instead of TaskStatusResponse validation - Ensures Celery task results are properly JSON-serializable before returning to clients --- .../service/memory_api_controller.py | 33 +++++++++++++++++-- 1 file changed, 30 insertions(+), 3 deletions(-) diff --git a/api/app/controllers/service/memory_api_controller.py b/api/app/controllers/service/memory_api_controller.py index 34135fec..9acd865f 100644 --- a/api/app/controllers/service/memory_api_controller.py +++ b/api/app/controllers/service/memory_api_controller.py @@ -15,7 +15,6 @@ from app.schemas.memory_api_schema import ( MemoryWriteRequest, MemoryWriteResponse, MemoryWriteSyncResponse, - TaskStatusResponse, ) from app.services.memory_api_service import MemoryAPIService @@ -23,6 +22,34 @@ router = APIRouter(prefix="/memory", tags=["V1 - Memory API"]) logger = get_business_logger() +def _sanitize_task_result(result: dict) -> dict: + """Make Celery task result JSON-serializable. + + Converts UUID and other non-serializable values to strings. + + Args: + result: Raw task result dict from task_service + + Returns: + JSON-safe dict + """ + import uuid as _uuid + from datetime import datetime + + def _convert(obj): + if isinstance(obj, dict): + return {k: _convert(v) for k, v in obj.items()} + if isinstance(obj, list): + return [_convert(i) for i in obj] + if isinstance(obj, _uuid.UUID): + return str(obj) + if isinstance(obj, datetime): + return obj.isoformat() + return obj + + return _convert(result) + + @router.get("") async def get_memory_info(): """获取记忆服务信息（占位）""" @@ -80,7 +107,7 @@ async def get_write_task_status( from app.services.task_service import get_task_memory_write_result result = get_task_memory_write_result(task_id) - return success(data=TaskStatusResponse(**result).model_dump(), msg="Task status retrieved") + return success(data=_sanitize_task_result(result), msg="Task status retrieved") @router.post("/read") @@ -135,7 +162,7 @@ async def get_read_task_status( from app.services.task_service import get_task_memory_read_result result = get_task_memory_read_result(task_id) - return success(data=TaskStatusResponse(**result).model_dump(), msg="Task status retrieved") + return success(data=_sanitize_task_result(result), msg="Task status retrieved") @router.post("/write/sync") From 0afc38e7eff8043b847a860ca94ecd7918b38f5a Mon Sep 17 00:00:00 2001 From: lanceyq <1982376970@qq.com> Date: Fri, 10 Apr 2026 18:45:35 +0800 Subject: [PATCH 004/245] refactor(memory): flatten meta_data fields in update_end_user_info response Align update response with get_end_user_info by extracting profile, knowledge_tags, and behavioral_hints to top-level keys instead of returning raw meta_data dict. --- api/app/services/user_memory_service.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/api/app/services/user_memory_service.py b/api/app/services/user_memory_service.py index ab51d922..79f8fa05 100644 --- a/api/app/services/user_memory_service.py +++ b/api/app/services/user_memory_service.py @@ -398,12 +398,16 @@ class UserMemoryService: } # 构建响应数据（转换时间为毫秒时间戳） + # 将 meta_data 中的 profile、knowledge_tags、behavioral_hints 平铺到顶层 + meta = end_user_info_record.meta_data or {} response_data = { "end_user_info_id": str(end_user_info_record.id), "end_user_id": str(end_user_info_record.end_user_id), "other_name": end_user_info_record.other_name, "aliases": end_user_info_record.aliases, - "meta_data": end_user_info_record.meta_data, + "profile": meta.get("profile"), + "knowledge_tags": meta.get("knowledge_tags"), + "behavioral_hints": meta.get("behavioral_hints"), "created_at": datetime_to_timestamp(end_user_info_record.created_at), "updated_at": datetime_to_timestamp(end_user_info_record.updated_at) } From e1cf3bb3d20da491be31646aa25d3490a68e8c5b Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 13 Apr 2026 10:21:35 +0800 Subject: [PATCH 005/245] fix(web): i18n update --- web/src/i18n/en.ts | 2 +- web/src/i18n/zh.ts | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/web/src/i18n/en.ts b/web/src/i18n/en.ts index d813f40f..e78964d5 100644 --- a/web/src/i18n/en.ts +++ b/web/src/i18n/en.ts @@ -116,7 +116,7 @@ export const en = { prompt: 'Prompt Engineering', skills: 'Skill Library', workbench: 'Workbench', - memoryRelated: 'Memory-Related', + memoryRelated: 'Memory Hub', advancedSettings: 'Advanced Settings', promptHistory: 'My history', platformManagement: 'Platform Management', diff --git a/web/src/i18n/zh.ts b/web/src/i18n/zh.ts index fc846dcd..d206a1c6 100644 --- a/web/src/i18n/zh.ts +++ b/web/src/i18n/zh.ts @@ -116,7 +116,7 @@ export const zh = { prompt: '提示词工程', skills: '技能库', workbench: '工作台', - memoryRelated: '记忆相关', + memoryRelated: '记忆中枢', advancedSettings: '高级设置', promptHistory: '我的历史', platformManagement: '平台管理', From 2b52b32b96f2ee55f20c992978d52941442a38ec Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 13 Apr 2026 11:36:14 +0800 Subject: [PATCH 006/245] fix(web): variable ui update --- web/src/i18n/en.ts | 3 + web/src/i18n/zh.ts | 3 + .../components/Editor/nodes/VariableNode.tsx | 16 +-- .../Editor/plugin/AutocompletePlugin.tsx | 83 ++++++------ .../components/Properties/VariableSelect.tsx | 124 +++++++++--------- .../Properties/hooks/useVariableList.ts | 4 +- .../Workflow/components/Properties/index.tsx | 8 +- 7 files changed, 121 insertions(+), 120 deletions(-) diff --git a/web/src/i18n/en.ts b/web/src/i18n/en.ts index e78964d5..59a303f1 100644 --- a/web/src/i18n/en.ts +++ b/web/src/i18n/en.ts @@ -2552,6 +2552,9 @@ Memory Bear: After the rebellion, regional warlordism intensified for several re 'list-operator.input_list': 'Input list', }, checkListHasErrors: 'Please resolve all issues in the checklist before publishing', + variableSelect: { + empty: 'No variables available', + }, }, emotionEngine: { emotionEngineConfig: 'Emotion Engine Configuration', diff --git a/web/src/i18n/zh.ts b/web/src/i18n/zh.ts index d206a1c6..1c3791d4 100644 --- a/web/src/i18n/zh.ts +++ b/web/src/i18n/zh.ts @@ -2516,6 +2516,9 @@ export const zh = { 'list-operator.input_list': '输入变量', }, checkListHasErrors: '发布前确认检查清单中所有问题均已解决', + variableSelect: { + empty: '暂无变量', + }, }, emotionEngine: { emotionEngineConfig: '情感引擎配置', diff --git a/web/src/views/Workflow/components/Editor/nodes/VariableNode.tsx b/web/src/views/Workflow/components/Editor/nodes/VariableNode.tsx index 5688342c..72e73220 100644 --- a/web/src/views/Workflow/components/Editor/nodes/VariableNode.tsx +++ b/web/src/views/Workflow/components/Editor/nodes/VariableNode.tsx @@ -48,17 +48,13 @@ const VariableComponent: React.FC<{ nodeKey: NodeKey; data: Suggestion }> = ({ return ( - {data.isContext ? ( - 📄 - ) : data.group !== 'CONVERSATION' && !data.value.includes('conv') ? ( - - ) : } + {!data.isContext && data.group !== 'CONVERSATION' && !data.value.includes('conv') + ?
+ : null + } {!data.isContext && data.group !== 'CONVERSATION' && ( <> {!data.value.includes('conv') && <> @@ -73,7 +69,7 @@ const VariableComponent: React.FC<{ nodeKey: NodeKey; data: Suggestion }> = ({ )} )} - {data.label} + {data.label} ); }; diff --git a/web/src/views/Workflow/components/Editor/plugin/AutocompletePlugin.tsx b/web/src/views/Workflow/components/Editor/plugin/AutocompletePlugin.tsx index f9537032..9f718826 100644 --- a/web/src/views/Workflow/components/Editor/plugin/AutocompletePlugin.tsx +++ b/web/src/views/Workflow/components/Editor/plugin/AutocompletePlugin.tsx @@ -2,12 +2,13 @@ * @Author: ZhaoYing * @Date: 2025-12-23 16:22:51 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-07 16:51:04 + * @Last Modified time: 2026-04-13 11:12:18 */ import { useEffect, useLayoutEffect, useState, useRef, type FC } from 'react'; import { useLexicalComposerContext } from '@lexical/react/LexicalComposerContext'; import { $getSelection, $isRangeSelection, COMMAND_PRIORITY_HIGH, KEY_ENTER_COMMAND, KEY_ARROW_DOWN_COMMAND, KEY_ARROW_UP_COMMAND, KEY_ESCAPE_COMMAND } from 'lexical'; import { Space, Flex } from 'antd'; +import clsx from 'clsx'; import { INSERT_VARIABLE_COMMAND, CLOSE_AUTOCOMPLETE_COMMAND } from '../commands'; import type { NodeProperties } from '../../../types' @@ -284,23 +285,23 @@ const AutocompletePlugin: FC<{ options: Suggestion[] }> = ({ options }) => { ref={popupRef} data-autocomplete-popup="true" onMouseDown={(e) => e.preventDefault()} - className="rb:fixed rb:z-1000 rb:bg-white rb:rounded-xl rb:shadow-[0px_2px_12px_0px_rgba(23,23,25,0.12)]" + className="rb:min-w-70 rb:max-h-57.5 rb:overflow-y-auto rb:fixed rb:z-1000 rb:bg-white rb:rounded-lg rb:border-[0.5px] rb:border-[#EBEBEB] rb:shadow-[0px_2px_6px_0px_rgba(0,0,0,0.1)] rb:py-3 rb:px-2" style={{ top: popupPosition.top, left: popupPosition.left, }} > -
- - {Object.entries(groupedSuggestions).map(([nodeId, nodeOptions]) => { - const nodeName = nodeOptions[0]?.nodeData?.name || nodeId; - const nodeIcon = nodeOptions[0]?.nodeData?.icon; - return ( -
- {nodeName !== 'undefined' && - {nodeIcon &&
} + + {Object.entries(groupedSuggestions).map(([nodeId, nodeOptions]) => { + const nodeName = nodeOptions[0]?.nodeData?.name || nodeId; + return ( +
+ {nodeName !== 'undefined' && +
{nodeName} - } +
+ } + {nodeOptions.map((option) => { const globalIndex = flatOptions.indexOf(option); const isExpanded = expandedParent?.key === option.key; @@ -310,14 +311,13 @@ const AutocompletePlugin: FC<{ options: Suggestion[] }> = ({ options }) => { key={option.key} ref={(el) => { if (el) itemRefs.current.set(option.key, el); }} data-selected={selectedIndex === globalIndex} - className="rb:pl-6! rb:pr-3! rb:py-2!" + className={clsx("rb:px-2! rb:py-0.75! rb:rounded-sm rb:leading-4.5 rb:text-[#5B6167] rb:hover:bg-[#F6F6F6]", { + 'rb:bg-[#F6F6F6]': selectedIndex === globalIndex || isExpanded, + 'rb:cursor-not-allowed rb:opacity-65': option.disabled, + 'rb:cursor-pointer': !option.disabled, + })} align="center" justify="space-between" - style={{ - cursor: option.disabled ? 'not-allowed' : 'pointer', - background: (selectedIndex === globalIndex || isExpanded) ? '#f0f8ff' : 'white', - opacity: option.disabled ? 0.5 : 1, - }} onClick={() => { if (option.disabled) return; insertMention(option); @@ -337,26 +337,27 @@ const AutocompletePlugin: FC<{ options: Suggestion[] }> = ({ options }) => { } }} > - {option.label && - {option.isContext ? '📄' : `{x}`} - {option.label} - } - - {option.dataType && {option.dataType}} - {hasChildren && ›} + {option.label && +
+ {`{x}`} {option.label} +
+ } + + {option.dataType && {option.dataType}} + {hasChildren &&
} ); })} -
- ); - })} - -
+ +
+ ); + })} + {/* Child variables panel - floats to the left */} {expandedParent?.children?.length && (
= ({ options }) => { }} onMouseEnter={() => setExpandedParent(expandedParent)} > - {/* Header */} -
- +
+ {expandedParent.nodeData.name}.{expandedParent.label} {expandedParent.dataType} @@ -377,19 +377,18 @@ const AutocompletePlugin: FC<{ options: Suggestion[] }> = ({ options }) => { !child.disabled && insertMention(child)} onMouseEnter={() => setSelectedIndex(childIndex)} > - {child.label} - {child.dataType && {child.dataType}} + {child.label} + {child.dataType && {child.dataType}} ); })} diff --git a/web/src/views/Workflow/components/Properties/VariableSelect.tsx b/web/src/views/Workflow/components/Properties/VariableSelect.tsx index b28d7b4f..c0207cb5 100644 --- a/web/src/views/Workflow/components/Properties/VariableSelect.tsx +++ b/web/src/views/Workflow/components/Properties/VariableSelect.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 15:40:13 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-08 10:48:21 + * @Last Modified time: 2026-04-13 11:25:40 */ import { useState, useRef, useEffect, useLayoutEffect, type FC } from 'react' import { createPortal } from 'react-dom' @@ -190,20 +190,30 @@ const VariableSelect: FC = ({ {/* Trigger */}
setOpen(o => !o)} > {multiple ? ( selectedValues.length > 0 ? ( - + {selectedValues.map(v => { const s = suggestionMap.get(v); if (!s) return null; @@ -214,11 +224,11 @@ const VariableSelect: FC = ({ return ( - {!isConv && nd?.icon &&
} + {!isConv && nd?.icon &&
} {!isConv && nd?.name && {nd.name}{sep}} - + {parent ? <>{parent.label}{sep}{s.label} : s.label} = ({ ); })} - + ) : ( - {placeholder} + {placeholder} ) ) : selectedSuggestion ? (
- - {!isConversation && nodeData?.icon &&
} - {!isConversation && nodeData?.name && {nodeData.name}} - {!isConversation && nodeData?.name && {sep}} - + + {!isConversation && nodeData?.icon &&
} + {!isConversation && nodeData?.name && {nodeData.name}} + {!isConversation && nodeData?.name && {sep}} + {parentOfSelected ? <>{parentOfSelected.label}{sep}{selectedSuggestion.label} : selectedSuggestion.label} @@ -266,18 +278,19 @@ const VariableSelect: FC = ({ {open && createPortal(
-
- {Object.entries(filteredGroups).map(([nodeId, suggestions]) => { +
+ {Object.entries(filteredGroups).map(([nodeId, suggestions], index) => { const nd = suggestions[0].nodeData; return ( -
- - {nd.icon &&
} +
+
{nd.name} - +
{suggestions.map(s => { const isSelected = multiple ? selectedValues.includes(`{{${s.value}}}`) @@ -288,11 +301,9 @@ const VariableSelect: FC = ({ { if (el) itemRefs.current.set(s.key, el); }} - className={clsx("rb:pl-6! rb:pr-3! rb:py-1.25! rb:rounded-lg!", { - 'rb:bg-[#e6f4ff]': isSelected || isExpanded, - 'rb:bg-white rb:hover:bg-[#F6F6F6]!': !(isSelected || isExpanded), - 'rb:opacity-60': s.disabled, - 'rb:cursor-not-allowed': s.disabled, + className={clsx("rb:px-2! rb:py-0.75! rb:rounded-sm rb:leading-4.5 rb:text-[#5B6167] rb:hover:bg-[#F6F6F6]", { + 'rb:bg-[#F6F6F6]': isSelected || isExpanded, + 'rb:cursor-not-allowed rb:opacity-65': s.disabled, 'rb:cursor-pointer': !s.disabled, })} align="center" @@ -314,17 +325,16 @@ const VariableSelect: FC = ({ } }} > - +
{multiple && ( - + )} - {`{x}`} - {s.label} - - - {s.dataType && {s.dataType}} + {`{x}`} {s.label} +
- {hasChildren &&
} + + {s.dataType && {s.dataType}} + {hasChildren &&
} ); @@ -334,7 +344,7 @@ const VariableSelect: FC = ({ })} {Object.keys(filteredGroups).length === 0 && (
- {t('workflow.variableSelect.empty', '暂无变量')} + {t('workflow.variableSelect.empty')}
)}
@@ -346,18 +356,13 @@ const VariableSelect: FC = ({ {open && expandedParent?.children?.length && createPortal(
setExpandedParent(expandedParent)} > -
!expandedParent.disabled && handleSelect(expandedParent)} - > +
- - {expandedParent.nodeData.name}.{expandedParent.label} - + {expandedParent.nodeData.name}.{expandedParent.label} {expandedParent.dataType}
@@ -365,32 +370,27 @@ const VariableSelect: FC = ({ const isSelected = multiple ? selectedValues.includes(`{{${child.value}}}`) : `{{${child.value}}}` === value; - const hasGrandChildren = !!child.children?.length; return ( !child.disabled && handleSelect(child)} > - + {multiple && ( )} - {child.label} - - - {child.dataType && {child.dataType}} - {hasGrandChildren && ›} + {child.label} + + {child.dataType && {child.dataType}} + ); })} diff --git a/web/src/views/Workflow/components/Properties/hooks/useVariableList.ts b/web/src/views/Workflow/components/Properties/hooks/useVariableList.ts index 3c4ea6f7..14dcced2 100644 --- a/web/src/views/Workflow/components/Properties/hooks/useVariableList.ts +++ b/web/src/views/Workflow/components/Properties/hooks/useVariableList.ts @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-01-19 17:00:26 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-08 10:12:27 + * @Last Modified time: 2026-04-13 10:44:17 */ /** * useVariableList Hook @@ -414,7 +414,7 @@ export const useVariableList = ( const pd = parentLoop.getData(); const pid = pd.id; if (pd.type === 'loop') { - (pd.cycle_vars || []).forEach((cv: any) => addVariable(list, keys, `${pid}_cycle_${cv.name}`, cv.name, cv.type || 'String', `${pid}.${cv.name}`, pd)); + (pd.cycle_vars || []).forEach((cv: any) => addVariable(list, keys, `${pid}_cycle_${cv.name}`, cv.name, cv.type || 'string', `${pid}.${cv.name}`, pd)); } else if (pd.type === 'iteration' && pd.config.input.defaultValue) { let itemType = 'object'; const iv = list.find(v => `{{${v.value}}}` === pd.config.input.defaultValue); diff --git a/web/src/views/Workflow/components/Properties/index.tsx b/web/src/views/Workflow/components/Properties/index.tsx index b5bc2d2e..f826edd9 100644 --- a/web/src/views/Workflow/components/Properties/index.tsx +++ b/web/src/views/Workflow/components/Properties/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 15:39:59 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-10 17:24:19 + * @Last Modified time: 2026-04-13 10:44:19 */ import { type FC, useEffect, useState, useMemo } from "react"; import clsx from 'clsx' @@ -266,7 +266,7 @@ const Properties: FC = ({ key, label: cycleVar.name, type: 'variable', - dataType: cycleVar.type || 'String', + dataType: cycleVar.type || 'string', value: `${parentNodeId}.${cycleVar.name}`, nodeData: parentData, }); @@ -643,7 +643,7 @@ const Properties: FC = ({ key: contextKey, label: 'context', type: 'variable', - dataType: 'String', + dataType: 'string', value: `context`, nodeData: selectedNode.getData(), isContext: true, @@ -791,7 +791,7 @@ const Properties: FC = ({ key: `${selectedNode.id}_cycle_${cycleVar.name}`, label: cycleVar.name, type: 'variable', - dataType: cycleVar.type || 'String', + dataType: cycleVar.type || 'string', value: `${selectedNode.getData().id}.${cycleVar.name}`, nodeData: selectedNode.getData(), })); From 72be9f75f92ae6c2f59306bf9a4a9629e94c637a Mon Sep 17 00:00:00 2001 From: wxy Date: Mon, 13 Apr 2026 11:58:14 +0800 Subject: [PATCH 007/245] feat: Add quota check decorator and implement tenant-level API rate limiting - Add quota check decorator module quota_stub.py, providing community edition stub implementation - Add quota check decorators to multiple controllers - Implement tenant-level API call rate limiting - Remove redundant plan fields from tenant_model.py - Optimize user permission check logic with added error handling --- api/app/controllers/app_controller.py | 2 + api/app/controllers/file_controller.py | 2 + api/app/controllers/knowledge_controller.py | 2 + .../controllers/memory_storage_controller.py | 2 + api/app/controllers/model_controller.py | 6 +++ api/app/controllers/ontology_controller.py | 3 ++ .../service/end_user_api_controller.py | 2 + .../service/memory_api_controller.py | 2 + api/app/controllers/skill_controller.py | 2 + api/app/controllers/user_controller.py | 13 +++--- api/app/controllers/workspace_controller.py | 2 + api/app/core/api_key_auth.py | 32 ++++++++++++++ api/app/core/quota_stub.py | 44 +++++++++++++++++++ api/app/models/tenant_model.py | 7 +-- api/app/services/api_key_service.py | 29 ++++++++++++ 15 files changed, 140 insertions(+), 10 deletions(-) create mode 100644 api/app/core/quota_stub.py diff --git a/api/app/controllers/app_controller.py b/api/app/controllers/app_controller.py index db3c7536..3b7240de 100644 --- a/api/app/controllers/app_controller.py +++ b/api/app/controllers/app_controller.py @@ -28,6 +28,7 @@ from app.services.app_statistics_service import AppStatisticsService from app.services.workflow_import_service import WorkflowImportService from app.services.workflow_service import WorkflowService, get_workflow_service from app.services.app_dsl_service import AppDslService +from app.core.quota_stub import check_app_quota router = APIRouter(prefix="/apps", tags=["Apps"]) logger = get_business_logger() @@ -35,6 +36,7 @@ logger = get_business_logger() @router.post("", summary="创建应用（可选创建 Agent 配置）") @cur_workspace_access_guard() +@check_app_quota def create_app( payload: app_schema.AppCreate, db: Session = Depends(get_db), diff --git a/api/app/controllers/file_controller.py b/api/app/controllers/file_controller.py index f7bd0e7a..6f8b1b97 100644 --- a/api/app/controllers/file_controller.py +++ b/api/app/controllers/file_controller.py @@ -19,6 +19,7 @@ from app.models.user_model import User from app.schemas import file_schema, document_schema from app.schemas.response_schema import ApiResponse from app.services import file_service, document_service +from app.core.quota_stub import check_knowledge_capacity_quota # Obtain a dedicated API logger @@ -131,6 +132,7 @@ async def create_folder( @router.post("/file", response_model=ApiResponse) +@check_knowledge_capacity_quota async def upload_file( kb_id: uuid.UUID, parent_id: uuid.UUID, diff --git a/api/app/controllers/knowledge_controller.py b/api/app/controllers/knowledge_controller.py index afda7cce..5cd87647 100644 --- a/api/app/controllers/knowledge_controller.py +++ b/api/app/controllers/knowledge_controller.py @@ -27,6 +27,7 @@ from app.schemas import knowledge_schema from app.schemas.response_schema import ApiResponse from app.services import knowledge_service, document_service from app.services.model_service import ModelConfigService +from app.core.quota_stub import check_knowledge_capacity_quota # Obtain a dedicated API logger api_logger = get_api_logger() @@ -179,6 +180,7 @@ async def get_knowledges( @router.post("/knowledge", response_model=ApiResponse) +@check_knowledge_capacity_quota async def create_knowledge( create_data: knowledge_schema.KnowledgeCreate, db: Session = Depends(get_db), diff --git a/api/app/controllers/memory_storage_controller.py b/api/app/controllers/memory_storage_controller.py index 76eed50f..545f8302 100644 --- a/api/app/controllers/memory_storage_controller.py +++ b/api/app/controllers/memory_storage_controller.py @@ -34,6 +34,7 @@ from app.services.memory_storage_service import ( search_entity, search_statement, ) +from app.core.quota_stub import check_memory_engine_quota from fastapi import APIRouter, Depends, Header from fastapi.responses import StreamingResponse from sqlalchemy.orm import Session @@ -76,6 +77,7 @@ async def get_storage_info( @router.post("/create_config", response_model=ApiResponse) # 创建配置文件，其他参数默认 +@check_memory_engine_quota def create_config( payload: ConfigParamsCreate, current_user: User = Depends(get_current_user), diff --git a/api/app/controllers/model_controller.py b/api/app/controllers/model_controller.py index 71fd41ad..6105c3d8 100644 --- a/api/app/controllers/model_controller.py +++ b/api/app/controllers/model_controller.py @@ -15,6 +15,7 @@ from app.core.response_utils import success from app.schemas.response_schema import ApiResponse, PageData from app.services.model_service import ModelConfigService, ModelApiKeyService, ModelBaseService from app.core.logging_config import get_api_logger +from app.core.quota_stub import check_model_quota, check_model_activation_quota # 获取API专用日志器 api_logger = get_api_logger() @@ -236,6 +237,7 @@ def delete_model_base( @router.post("/model_plaza/{model_base_id}/add", response_model=ApiResponse) +@check_model_quota def add_model_from_plaza( model_base_id: uuid.UUID, db: Session = Depends(get_db), @@ -273,6 +275,7 @@ def get_model_by_id( @router.post("", response_model=ApiResponse) +@check_model_quota async def create_model( model_data: model_schema.ModelConfigCreate, db: Session = Depends(get_db), @@ -303,6 +306,7 @@ async def create_model( @router.post("/composite", response_model=ApiResponse) +@check_model_quota async def create_composite_model( model_data: model_schema.CompositeModelCreate, db: Session = Depends(get_db), @@ -329,6 +333,7 @@ async def create_composite_model( @router.put("/composite/{model_id}", response_model=ApiResponse) +@check_model_activation_quota async def update_composite_model( model_id: uuid.UUID, model_data: model_schema.CompositeModelCreate, @@ -370,6 +375,7 @@ def delete_composite_model( @router.put("/{model_id}", response_model=ApiResponse) +@check_model_activation_quota def update_model( model_id: uuid.UUID, model_data: model_schema.ModelConfigUpdate, diff --git a/api/app/controllers/ontology_controller.py b/api/app/controllers/ontology_controller.py index fe6b3598..83f75888 100644 --- a/api/app/controllers/ontology_controller.py +++ b/api/app/controllers/ontology_controller.py @@ -28,6 +28,8 @@ from fastapi import APIRouter, Depends, HTTPException, File, UploadFile, Form, H from fastapi.responses import StreamingResponse, JSONResponse from sqlalchemy.orm import Session +from app.core.quota_stub import check_ontology_project_quota + from app.core.config import settings from app.core.error_codes import BizCode from app.core.language_utils import get_language_from_header @@ -287,6 +289,7 @@ async def extract_ontology( # ==================== 本体场景管理接口 ==================== @router.post("/scene", response_model=ApiResponse) +@check_ontology_project_quota async def create_scene( request: SceneCreateRequest, db: Session = Depends(get_db), diff --git a/api/app/controllers/service/end_user_api_controller.py b/api/app/controllers/service/end_user_api_controller.py index df9996c2..92a9d7c8 100644 --- a/api/app/controllers/service/end_user_api_controller.py +++ b/api/app/controllers/service/end_user_api_controller.py @@ -9,6 +9,7 @@ from app.core.api_key_auth import require_api_key from app.core.error_codes import BizCode from app.core.exceptions import BusinessException from app.core.logging_config import get_business_logger +from app.core.quota_stub import check_end_user_quota from app.core.response_utils import success from app.db import get_db from app.repositories.end_user_repository import EndUserRepository @@ -22,6 +23,7 @@ logger = get_business_logger() @router.post("/create") @require_api_key(scopes=["memory"]) +@check_end_user_quota async def create_end_user( request: Request, api_key_auth: ApiKeyAuth = None, diff --git a/api/app/controllers/service/memory_api_controller.py b/api/app/controllers/service/memory_api_controller.py index dc5e0408..16f1e223 100644 --- a/api/app/controllers/service/memory_api_controller.py +++ b/api/app/controllers/service/memory_api_controller.py @@ -2,6 +2,7 @@ from app.core.api_key_auth import require_api_key from app.core.logging_config import get_business_logger +from app.core.quota_stub import check_end_user_quota from app.core.response_utils import success from app.db import get_db from app.schemas.api_key_schema import ApiKeyAuth @@ -119,6 +120,7 @@ async def list_memory_configs( @router.post("/end_users") @require_api_key(scopes=["memory"]) +@check_end_user_quota async def create_end_user( request: Request, api_key_auth: ApiKeyAuth = None, diff --git a/api/app/controllers/skill_controller.py b/api/app/controllers/skill_controller.py index 6e673679..4ee07c7d 100644 --- a/api/app/controllers/skill_controller.py +++ b/api/app/controllers/skill_controller.py @@ -11,11 +11,13 @@ from app.schemas import skill_schema from app.schemas.response_schema import PageData, PageMeta from app.services.skill_service import SkillService from app.core.response_utils import success +from app.core.quota_stub import check_skill_quota router = APIRouter(prefix="/skills", tags=["Skills"]) @router.post("", summary="创建技能") +@check_skill_quota def create_skill( data: skill_schema.SkillCreate, db: Session = Depends(get_db), diff --git a/api/app/controllers/user_controller.py b/api/app/controllers/user_controller.py index cc16a6b4..5a329165 100644 --- a/api/app/controllers/user_controller.py +++ b/api/app/controllers/user_controller.py @@ -114,11 +114,14 @@ def get_current_user_info( # 设置权限：如果用户来自 SSO Source，则使用该 Source 的 permissions；否则返回 "all" 表示拥有所有权限 if current_user.external_source: - from premium.sso.models import SSOSource - source = db.query(SSOSource).filter(SSOSource.source_code == current_user.external_source).first() - if source and source.permissions: - result_schema.permissions = source.permissions - else: + try: + from premium.sso.models import SSOSource + source = db.query(SSOSource).filter(SSOSource.source_code == current_user.external_source).first() + if source and source.permissions: + result_schema.permissions = source.permissions + else: + result_schema.permissions = [] + except ModuleNotFoundError: result_schema.permissions = [] else: result_schema.permissions = ["all"] diff --git a/api/app/controllers/workspace_controller.py b/api/app/controllers/workspace_controller.py index 6f4a4fa8..47068288 100644 --- a/api/app/controllers/workspace_controller.py +++ b/api/app/controllers/workspace_controller.py @@ -35,6 +35,7 @@ from app.schemas.workspace_schema import ( WorkspaceUpdate, ) from app.services import workspace_service +from app.core.quota_stub import check_workspace_quota # 获取API专用日志器 api_logger = get_api_logger() @@ -106,6 +107,7 @@ def get_workspaces( @router.post("", response_model=ApiResponse) +@check_workspace_quota def create_workspace( workspace: WorkspaceCreate, language_type: str = Header(default="zh", alias="X-Language-Type"), diff --git a/api/app/core/api_key_auth.py b/api/app/core/api_key_auth.py index 342405b8..91d6bd8a 100644 --- a/api/app/core/api_key_auth.py +++ b/api/app/core/api_key_auth.py @@ -96,6 +96,38 @@ def require_api_key( resource_id=api_key_obj.resource_id, ) + # ── Tenant 级别限速（来自套餐配额 api_ops_rate_limit）────────── + try: + from app.models.workspace_model import Workspace + from premium.platform_admin.package_plan_service import TenantSubscriptionService + + workspace = db.query(Workspace).filter( + Workspace.id == api_key_obj.workspace_id + ).first() + if workspace: + quota = TenantSubscriptionService(db).get_effective_quota(workspace.tenant_id) + tenant_qps_limit = quota.get("api_ops_rate_limit") if quota else None + if tenant_qps_limit: + rate_limiter = RateLimiterService() + tenant_ok, tenant_info = await rate_limiter.check_tenant_rate_limit( + workspace.tenant_id, tenant_qps_limit + ) + if not tenant_ok: + raise RateLimitException( + "租户 API 调用速率超限", + BizCode.API_KEY_QPS_LIMIT_EXCEEDED, + rate_headers={ + "X-RateLimit-Tenant-Limit": str(tenant_info["limit"]), + "X-RateLimit-Tenant-Remaining": str(tenant_info["remaining"]), + "X-RateLimit-Tenant-Reset": str(tenant_info["reset"]), + } + ) + except RateLimitException: + raise + except Exception as e: + logger.warning(f"Tenant 限速检查异常，跳过: {e}") + # ───────────────────────────────────────────────────────────── + rate_limiter = RateLimiterService() is_allowed, error_msg, rate_headers = await rate_limiter.check_all_limits(api_key_obj) if not is_allowed: diff --git a/api/app/core/quota_stub.py b/api/app/core/quota_stub.py new file mode 100644 index 00000000..b8f82e75 --- /dev/null +++ b/api/app/core/quota_stub.py @@ -0,0 +1,44 @@ +""" +配额检查 stub - 社区版使用，所有检查直接放行。 +企业版通过 premium.platform_admin.quota_decorator 提供真实实现。 +""" +from functools import wraps +from typing import Callable + + +def _noop_decorator(func: Callable) -> Callable: + """空装饰器，直接放行""" + return func + + +def _noop_check(*args, **kwargs): + """空检查函数，直接放行""" + pass + + +try: + from premium.platform_admin.quota_decorator import ( + check_workspace_quota, + check_skill_quota, + check_app_quota, + check_knowledge_capacity_quota, + check_memory_engine_quota, + check_end_user_quota, + check_ontology_project_quota, + check_model_quota, + check_model_activation_quota, + get_quota_usage, + _check_quota, + ) +except ModuleNotFoundError: + check_workspace_quota = _noop_decorator + check_skill_quota = _noop_decorator + check_app_quota = _noop_decorator + check_knowledge_capacity_quota = _noop_decorator + check_memory_engine_quota = _noop_decorator + check_end_user_quota = _noop_decorator + check_ontology_project_quota = _noop_decorator + check_model_quota = _noop_decorator + check_model_activation_quota = _noop_decorator + get_quota_usage = lambda db, tenant_id: {} + _check_quota = _noop_check diff --git a/api/app/models/tenant_model.py b/api/app/models/tenant_model.py index a92b5629..c3fd82df 100644 --- a/api/app/models/tenant_model.py +++ b/api/app/models/tenant_model.py @@ -29,11 +29,8 @@ class Tenants(Base): contact_email = Column(String(255), nullable=True) # 联系人邮箱 contact_phone = Column(String(50), nullable=True) # 联系人电话 - # 租户套餐信息 - plan = Column(String(50), nullable=True) # 套餐类型 - plan_expired_at = Column(DateTime, nullable=True) # 套餐到期时间 - api_ops_rate_limit = Column(String(100), nullable=True) # API 调用频率限制 - status = Column(String(50), nullable=True, default='active') # 租户状态 + # 租户套餐信息（只读，从 tenant_subscriptions 动态获取） + status = Column(String(50), nullable=True, default='active', server_default='active') # 租户状态 # Relationship to users - one tenant has many users users = relationship("User", back_populates="tenant") diff --git a/api/app/services/api_key_service.py b/api/app/services/api_key_service.py index a49e8fe0..07d55198 100644 --- a/api/app/services/api_key_service.py +++ b/api/app/services/api_key_service.py @@ -248,6 +248,35 @@ class RateLimiterService: def __init__(self): self.redis = aio_redis + async def check_tenant_rate_limit(self, tenant_id: uuid.UUID, limit: int) -> Tuple[bool, dict]: + """ + 按 tenant_id 做 1 秒滑动窗口限速，限制值来自套餐配额 api_ops_rate_limit + """ + now = time.time() + window_start = now - 1 # 1 秒窗口 + key = f"rate_limit:tenant_qps:{tenant_id}" + + async with self.redis.pipeline() as pipe: + # 清理 1 秒前的旧记录 + pipe.zremrangebyscore(key, 0, window_start) + # 加入当前请求（score=时间戳，member=时间戳+随机数保证唯一） + pipe.zadd(key, {f"{now}:{uuid.uuid4().hex}": now}) + # 统计窗口内请求数 + pipe.zcard(key) + # 设置 key 过期（2 秒后自动清理） + pipe.expire(key, 2) + results = await pipe.execute() + + current = results[2] + remaining = max(0, limit - current) + reset_time = int(now) + 1 + + return current <= limit, { + "limit": limit, + "remaining": remaining, + "reset": reset_time, + } + async def check_qps(self, api_key_id: uuid.UUID, limit: int) -> Tuple[bool, dict]: """ 检查QPS限制 From 520ee7c132b3ed4bad8bd3c619e8a4ffb7f12097 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 13 Apr 2026 12:01:37 +0800 Subject: [PATCH 008/245] fix(web): sub node connected --- .../views/Workflow/hooks/useWorkflowGraph.ts | 51 ++++++++++++------- 1 file changed, 33 insertions(+), 18 deletions(-) diff --git a/web/src/views/Workflow/hooks/useWorkflowGraph.ts b/web/src/views/Workflow/hooks/useWorkflowGraph.ts index f385acf3..5d0bb9c6 100644 --- a/web/src/views/Workflow/hooks/useWorkflowGraph.ts +++ b/web/src/views/Workflow/hooks/useWorkflowGraph.ts @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 15:17:48 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-07 23:17:50 + * @Last Modified time: 2026-04-13 12:00:09 */ import { Clipboard, Graph, Keyboard, MiniMap, Node, Snapline, type Edge } from '@antv/x6'; import { register } from '@antv/x6-react-shape'; @@ -1022,24 +1022,39 @@ export const useWorkflowGraph = ({ graphRef.current.on('node:removed', blankClick) // When edge connected, bring connected nodes' ports to front - graphRef.current.on('edge:connected', ({ isNew }) => { - // Bring edge to front first, then bring child nodes above edges - // Parent (loop/iteration) nodes stay behind to avoid covering edges - // Reset any port hover state left from dragging + graphRef.current.on('edge:connected', ({ isNew, edge }) => { if (isNew) { - graphRef.current?.getNodes().forEach(node => { - if (!node.getData()?.cycle) node.toFront(); - }); - graphRef.current?.getEdges().forEach(edge => { - const sourceCell = graphRef.current?.getCellById(edge.getSourceCellId()); - const targetCell = graphRef.current?.getCellById(edge.getTargetCellId()); - if (sourceCell?.getData()?.cycle || targetCell?.getData()?.cycle) { - edge.toFront(); - } - }); - graphRef.current?.getNodes().forEach(node => { - if (node.getData()?.cycle) node.toFront(); - }); + const sourceCellId = edge.getSourceCellId() + const targetCellId = edge.getTargetCellId() + const sourceCell = graphRef.current?.getCellById(sourceCellId); + const targetCell = graphRef.current?.getCellById(targetCellId); + + sourceCell?.toFront(); + targetCell?.toFront() + if (['loop', 'iteration'].includes(sourceCell?.getData()?.type)) { + graphRef.current?.getEdges().forEach(edge => { + const edgeSourceCell = graphRef.current?.getCellById(edge.getSourceCellId()); + const edgeTargetCell = graphRef.current?.getCellById(edge.getTargetCellId()); + if (edgeSourceCell?.getData()?.cycle === sourceCellId || edgeTargetCell?.getData()?.cycle === sourceCellId) { + edge.toFront(); + } + }); + graphRef.current?.getNodes().forEach(node => { + if (node.getData()?.cycle === sourceCellId) node.toFront(); + }); + } + if (['loop', 'iteration'].includes(targetCell?.getData()?.type)) { + graphRef.current?.getEdges().forEach(edge => { + const edgeSourceCell = graphRef.current?.getCellById(edge.getSourceCellId()); + const edgeTargetCell = graphRef.current?.getCellById(edge.getTargetCellId()); + if (edgeSourceCell?.getData()?.cycle === targetCellId || edgeTargetCell?.getData()?.cycle === targetCellId) { + edge.toFront(); + } + }); + graphRef.current?.getNodes().forEach(node => { + if (node.getData()?.cycle === targetCellId) node.toFront(); + }); + } } }); From 988d101e935de0b6898612cbec23edc45f2d366d Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 13 Apr 2026 12:12:49 +0800 Subject: [PATCH 009/245] fix(web): tool checklist --- .../Workflow/components/CheckList/index.tsx | 77 ++++--------------- 1 file changed, 13 insertions(+), 64 deletions(-) diff --git a/web/src/views/Workflow/components/CheckList/index.tsx b/web/src/views/Workflow/components/CheckList/index.tsx index 636ae9a9..0256416a 100644 --- a/web/src/views/Workflow/components/CheckList/index.tsx +++ b/web/src/views/Workflow/components/CheckList/index.tsx @@ -79,7 +79,6 @@ const specialValidators: Record boolean> = { } function isEmpty(val: any): boolean { - console.log('validateNode isEmpty', val, val === undefined || val === null || val === '') if (val === undefined || val === null || val === '') return true if (Array.isArray(val)) return val.length === 0 return false @@ -98,7 +97,6 @@ function validateNode(type: string, config: Record): CheckError[] { const specialKey = `${type}.${field}` const specialValidator = specialValidators[specialKey] const isInvalid = specialValidator ? specialValidator(val) : isEmpty(val) - console.log('validateNode', val, specialKey, specialValidator, isEmpty(val)) if (isInvalid) errors.push({ key: specialKey, message: '' }) }) @@ -114,62 +112,6 @@ function validateNode(type: string, config: Record): CheckError[] { return errors } -export async function runCheckOnGraph( - graph: import('@antv/x6').Graph, - t: (key: string) => string -): Promise { - const nodes = graph.getNodes() - const edges = graph.getEdges() - const targetIds = new Set() - const childTargetIds = new Set() - edges.forEach(e => { - targetIds.add(e.getTargetCellId()) - const srcData = graph.getCellById(e.getSourceCellId())?.getData() - const tgtData = graph.getCellById(e.getTargetCellId())?.getData() - if (srcData?.cycle && tgtData?.cycle && srcData.cycle === tgtData.cycle) { - childTargetIds.add(e.getTargetCellId()) - } - }) - - const checked: NodeCheckResult[] = [] - for (const node of nodes) { - const data = node.getData() - if (!data || ['add-node', 'notes', 'cycle-start', 'break'].includes(data.type)) continue - - const errors: CheckError[] = [] - const isChildNode = !!data.cycle - const hasIncoming = isChildNode ? childTargetIds.has(node.id) : !['start', 'cycle-start'].includes(data.type) ? targetIds.has(node.id) : true - if (!hasIncoming) errors.push({ key: 'notConnected', message: t('workflow.notConnected') }) - - const configErrors = validateNode(data.type, data.config ?? {}) - configErrors.forEach(e => { - errors.push({ key: e.key, message: `${t(`workflow.checkListErrors.${e.key}`)} ${t('workflow.cannotBeEmpty')}`.trim() }) - }) - - if (data.type === 'tool') { - const toolId = data.config?.tool_id?.defaultValue ?? data.config?.tool_id - const toolParameters = data.config?.tool_parameters?.defaultValue ?? data.config?.tool_parameters ?? {} - if (toolId) { - try { - const methods = await getToolMethods(toolId) as Array<{ name: string; parameters: Array<{ name: string; required: boolean }> }> - const operation = toolParameters?.operation - const method = operation ? methods.find(m => m.name === operation) : methods[0] - if (method) { - method.parameters - .filter(p => p.required && (toolParameters[p.name] === undefined || toolParameters[p.name] === null || toolParameters[p.name] === '')) - .forEach(p => errors.push({ key: 'tool.tool_parameters', message: `${p.name} ${t('workflow.cannotBeEmpty')}` })) - } - } catch { /* ignore */ } - } - } - - if (errors.length) { - checked.push({ id: node.id, name: data.name || t(`workflow.${data.type}`), type: data.type, icon: nodeIconMap[data.type] ?? '', errors }) - } - } - return checked -} - const CheckList: FC = ({ workflowRef, appId }) => { const { t } = useTranslation() const [open, setOpen] = useState(false) @@ -222,7 +164,8 @@ const CheckList: FC = ({ workflowRef, appId }) => { if (data.type === 'tool') { const toolId = data.config?.tool_id?.defaultValue ?? data.config?.tool_id const toolParameters = data.config?.tool_parameters?.defaultValue ?? data.config?.tool_parameters ?? {} - if (toolId) { + + if (typeof toolId === 'string') { try { const methods = await getToolMethods(toolId) as Array<{ name: string; parameters: Array<{ name: string; required: boolean }> }> const operation = toolParameters?.operation @@ -251,21 +194,27 @@ const CheckList: FC = ({ workflowRef, appId }) => { return checked }, [workflowRef.current?.graphRef?.current, t]) + const scheduleCheckRef = useRef<() => void>() + const scheduleCheck = useCallback(() => { clearTimeout(timerRef.current) timerRef.current = setTimeout(async () => { setCheckResults(appId, await runCheck()) - }, 500) + }, 300) }, [runCheck]) + scheduleCheckRef.current = scheduleCheck + useEffect(() => { const graph = workflowRef.current?.graphRef?.current + console.log('graph') if (!graph) return - const events = ['node:added', 'node:removed', 'node:change:data', 'edge:added', 'edge:removed'] - events.forEach(e => graph.on(e, scheduleCheck)) - scheduleCheck() + const handler = () => scheduleCheckRef.current?.() + const events = ['node:added', 'node:removed', 'node:change:data', 'edge:added', 'edge:removed', 'edge:connected', 'edge:changed'] + events.forEach(e => graph.on(e, handler)) + scheduleCheckRef.current?.() return () => { - events.forEach(e => graph.off(e, scheduleCheck)) + events.forEach(e => graph.off(e, handler)) clearTimeout(timerRef.current) } }, [workflowRef.current?.graphRef?.current]) From efdee32f8566e9aadcaaa4a21347e62f9c6585f6 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 13 Apr 2026 12:16:32 +0800 Subject: [PATCH 010/245] fix(web): update chat variable defaultValue validate rule --- .../components/AddChatVariable/ChatVariableModal.tsx | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) diff --git a/web/src/views/Workflow/components/AddChatVariable/ChatVariableModal.tsx b/web/src/views/Workflow/components/AddChatVariable/ChatVariableModal.tsx index 5666f3ab..5c80fab1 100644 --- a/web/src/views/Workflow/components/AddChatVariable/ChatVariableModal.tsx +++ b/web/src/views/Workflow/components/AddChatVariable/ChatVariableModal.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2025-12-30 13:59:36 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-08 11:05:34 + * @Last Modified time: 2026-04-13 12:16:00 */ import { forwardRef, useImperativeHandle, useState, useRef, useMemo } from 'react'; import { Form, Input, Select, InputNumber, Button, Row, Col, Flex } from 'antd'; @@ -345,15 +345,16 @@ const ChatVariableModal = forwardRef { if (!value) return Promise.resolve(); try { JSON.parse(value); return Promise.resolve(); } catch { return Promise.reject(t('workflow.invalidJSON')); } } - } : {} - ]} + }] + : undefined + } > {type === 'number' ? From 62355186ef582b1dfcc4d7ad6c5f252296fd41da Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 13 Apr 2026 13:38:10 +0800 Subject: [PATCH 011/245] fix(web): echarts grid --- web/src/views/UserMemoryDetail/components/InterestAreas.tsx | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/web/src/views/UserMemoryDetail/components/InterestAreas.tsx b/web/src/views/UserMemoryDetail/components/InterestAreas.tsx index 91554880..4d9be5b5 100644 --- a/web/src/views/UserMemoryDetail/components/InterestAreas.tsx +++ b/web/src/views/UserMemoryDetail/components/InterestAreas.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 18:32:53 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-16 14:27:12 + * @Last Modified time: 2026-04-13 13:37:43 */ import { useEffect, useState, forwardRef, useImperativeHandle, useRef } from 'react' import { useTranslation } from 'react-i18next' @@ -93,7 +93,7 @@ const InterestAreas = forwardRef<{ handleRefresh: () => void; }>((_props, ref) = ref={chartRef} option={{ color: Colors, - grid: { top: 8, left: 38, right: 8, bottom: 24 }, + grid: { top: 14, left: 38, right: 8, bottom: 24 }, xAxis: { type: 'category', data: keys.map(k => t(`implicitDetail.${k}`)), From 7a2a941ac4d94734cec11170de68b5a4ad240202 Mon Sep 17 00:00:00 2001 From: lanceyq <1982376970@qq.com> Date: Mon, 13 Apr 2026 13:47:59 +0800 Subject: [PATCH 012/245] refactor(neo4j): rename execute_query parameter from query to cypher Improves readability by making the parameter name explicitly reflect that it expects a Cypher query string rather than a generic query. --- api/app/repositories/neo4j/neo4j_connector.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/api/app/repositories/neo4j/neo4j_connector.py b/api/app/repositories/neo4j/neo4j_connector.py index ea8fa917..d20bf75f 100644 --- a/api/app/repositories/neo4j/neo4j_connector.py +++ b/api/app/repositories/neo4j/neo4j_connector.py @@ -77,11 +77,11 @@ class Neo4jConnector: """ await self.driver.close() - async def execute_query(self, query: str, json_format=False, **kwargs: Any) -> List[Dict[str, Any]]: + async def execute_query(self, cypher: str, json_format=False, **kwargs: Any) -> List[Dict[str, Any]]: """执行Cypher查询 Args: - query: Cypher查询语句 + cypher: Cypher查询语句 json_format: json格式化 **kwargs: 查询参数，将作为参数传递给Cypher查询 @@ -92,7 +92,7 @@ class Neo4jConnector: """ result = await self.driver.execute_query( - query, + cypher, database="neo4j", **kwargs ) From dca3173ed97a8dd0fd4b8fe501bd87ff210d8044 Mon Sep 17 00:00:00 2001 From: Eternity <1533512157@qq.com> Date: Fri, 10 Apr 2026 17:42:57 +0800 Subject: [PATCH 013/245] refactor(memory): restructure memory search architecture - Replace storage_services/search with new read_services/memory_search structure - Implement content_search and perceptual_search strategies - Add query_preprocessor for search optimization - Create memory_service as unified interface - Update celery_app and graph_search for new architecture - Add enums for memory operations - Implement base_pipeline and memory_read pipeline patterns --- api/app/celery_app.py | 1 - api/app/core/memory/enums.py | 18 + api/app/core/memory/memory_service.py | 57 +++ api/app/core/memory/pipelines/__init__.py | 0 .../core/memory/pipelines/base_pipeline.py | 29 ++ api/app/core/memory/pipelines/memory_read.py | 26 ++ .../read_services/memory_search/__init__.py | 0 .../memory_search/content_search.py | 14 + .../memory_search/perceptual_search.py | 228 ++++++++++ .../read_services/query_preprocessor.py | 18 + .../storage_services/search/__init__.py | 143 ------ .../storage_services/search/hybrid_search.py | 408 ------------------ .../storage_services/search/keyword_search.py | 122 ------ .../search/search_strategy.py | 125 ------ .../search/semantic_search.py | 166 ------- api/app/repositories/neo4j/cypher_queries.py | 45 +- api/app/repositories/neo4j/graph_search.py | 57 ++- 17 files changed, 463 insertions(+), 994 deletions(-) create mode 100644 api/app/core/memory/enums.py create mode 100644 api/app/core/memory/memory_service.py create mode 100644 api/app/core/memory/pipelines/__init__.py create mode 100644 api/app/core/memory/pipelines/base_pipeline.py create mode 100644 api/app/core/memory/pipelines/memory_read.py create mode 100644 api/app/core/memory/read_services/memory_search/__init__.py create mode 100644 api/app/core/memory/read_services/memory_search/content_search.py create mode 100644 api/app/core/memory/read_services/memory_search/perceptual_search.py create mode 100644 api/app/core/memory/read_services/query_preprocessor.py delete mode 100644 api/app/core/memory/storage_services/search/__init__.py delete mode 100644 api/app/core/memory/storage_services/search/hybrid_search.py delete mode 100644 api/app/core/memory/storage_services/search/search_strategy.py delete mode 100644 api/app/core/memory/storage_services/search/semantic_search.py diff --git a/api/app/celery_app.py b/api/app/celery_app.py index e44001d9..b0894eb8 100644 --- a/api/app/celery_app.py +++ b/api/app/celery_app.py @@ -101,7 +101,6 @@ celery_app.conf.update( 'app.core.memory.agent.read_message_priority': {'queue': 'memory_tasks'}, 'app.core.memory.agent.read_message': {'queue': 'memory_tasks'}, 'app.core.memory.agent.write_message': {'queue': 'memory_tasks'}, - 'app.tasks.write_perceptual_memory': {'queue': 'memory_tasks'}, # Long-term storage tasks → memory_tasks queue (batched write strategies) 'app.core.memory.agent.long_term_storage.window': {'queue': 'memory_tasks'}, diff --git a/api/app/core/memory/enums.py b/api/app/core/memory/enums.py new file mode 100644 index 00000000..d0baf732 --- /dev/null +++ b/api/app/core/memory/enums.py @@ -0,0 +1,18 @@ +from enum import StrEnum + + +class StorageType(StrEnum): + NEO4J = 'neo4j' + RAG = 'rag' + + +class Neo4jStorageStrategy(StrEnum): + WINDOW = 'window' + TIMELINE = 'timeline' + AGGREGATE = "aggregate" + + +class SearchStrategy(StrEnum): + DEEP = "0" + NORMAL = "1" + QUICK = "2" diff --git a/api/app/core/memory/memory_service.py b/api/app/core/memory/memory_service.py new file mode 100644 index 00000000..2f0d8bb3 --- /dev/null +++ b/api/app/core/memory/memory_service.py @@ -0,0 +1,57 @@ +from sqlalchemy.orm import Session +from pydantic import BaseModel, ConfigDict + +from app.core.memory.enums import StorageType +from app.schemas import MemoryConfig +from app.services.memory_config_service import MemoryConfigService + + +class MemoryContext(BaseModel): + model_config = ConfigDict(frozen=True, arbitrary_types_allowed=True) + + end_user_id: str + memory_config: MemoryConfig + storage_type: StorageType = StorageType.NEO4J + user_rag_memory_id: str | None = None + language: str = "zh" + + +class MemoryService: + def __init__( + self, + db: Session, + config_id: str, + end_user_id: str, + workspace_id: str | None = None, + storage_type: str = "neo4j", + user_rag_memory_id: str | None = None, + language: str = "zh", + ): + config_service = MemoryConfigService(db) + memory_config = config_service.load_memory_config( + config_id=config_id, + workspace_id=workspace_id, + service_name="MemoryService", + ) + self.ctx = MemoryContext( + end_user_id=end_user_id, + memory_config=memory_config, + storage_type=StorageType(storage_type), + user_rag_memory_id=user_rag_memory_id, + language=language, + ) + + async def write(self, messages: list[dict]) -> str: + raise NotImplementedError + + async def read(self, query: str, history: list, search_switch: str) -> dict: + raise NotImplementedError + + async def forget(self, max_batch: int = 100, min_days: int = 30) -> dict: + raise NotImplementedError + + async def reflect(self) -> dict: + raise NotImplementedError + + async def cluster(self, new_entity_ids: list[str] = None) -> None: + raise NotImplementedError diff --git a/api/app/core/memory/pipelines/__init__.py b/api/app/core/memory/pipelines/__init__.py new file mode 100644 index 00000000..e69de29b diff --git a/api/app/core/memory/pipelines/base_pipeline.py b/api/app/core/memory/pipelines/base_pipeline.py new file mode 100644 index 00000000..d423aef2 --- /dev/null +++ b/api/app/core/memory/pipelines/base_pipeline.py @@ -0,0 +1,29 @@ +# -*- coding: UTF-8 -*- +# Author: Eternity +# @Email: 1533512157@qq.com +# @Time : 2026/4/3 11:44 +import uuid +from abc import ABC, abstractmethod +from typing import Any + +from sqlalchemy.orm import Session + +from demo.memory_alpha import MemoryContext + + +class ModelClientMixin(ABC): + def get_llm_client(self, db: Session, model_id: uuid.UUID): + pass + + def get_embedding_client(self, db: Session, model_id: uuid.UUID): + pass + + +class BasePipeline(ABC): + def __init__(self, ctx: MemoryContext, db: Session): + self.ctx = ctx + self.db = db + + @abstractmethod + async def run(self, *args, **kwargs) -> Any: + pass diff --git a/api/app/core/memory/pipelines/memory_read.py b/api/app/core/memory/pipelines/memory_read.py new file mode 100644 index 00000000..f6ccd210 --- /dev/null +++ b/api/app/core/memory/pipelines/memory_read.py @@ -0,0 +1,26 @@ +from app.core.memory.enums import SearchStrategy +from app.core.memory.pipelines.base_pipeline import BasePipeline +from app.core.memory.read_services.query_preprocessor import QueryPreprocessor + + +class ReadPipeLine(BasePipeline): + async def run(self, query, search_switch, memory_config): + query = QueryPreprocessor.process(query) + match search_switch: + case SearchStrategy.DEEP: + return await self._deep_read() + case SearchStrategy.NORMAL: + return await self._normal_read(query) + case SearchStrategy.QUICK: + return await self._quick_read() + case _: + raise RuntimeError("Unsupported search strategy") + + async def _deep_read(self): + pass + + async def _normal_read(self, query): + pass + + async def _quick_read(self): + pass diff --git a/api/app/core/memory/read_services/memory_search/__init__.py b/api/app/core/memory/read_services/memory_search/__init__.py new file mode 100644 index 00000000..e69de29b diff --git a/api/app/core/memory/read_services/memory_search/content_search.py b/api/app/core/memory/read_services/memory_search/content_search.py new file mode 100644 index 00000000..f5e58696 --- /dev/null +++ b/api/app/core/memory/read_services/memory_search/content_search.py @@ -0,0 +1,14 @@ +# -*- coding: UTF-8 -*- +# Author: Eternity +# @Email: 1533512157@qq.com +# @Time : 2026/4/9 16:48 +from app.core.memory.llm_tools import OpenAIEmbedderClient +from app.core.memory.memory_service import MemoryContext + + +class ContentSearch: + def __init__(self, ctx: MemoryContext): + self.ctx = ctx + + async def search(self, query): + pass \ No newline at end of file diff --git a/api/app/core/memory/read_services/memory_search/perceptual_search.py b/api/app/core/memory/read_services/memory_search/perceptual_search.py new file mode 100644 index 00000000..db81e2f8 --- /dev/null +++ b/api/app/core/memory/read_services/memory_search/perceptual_search.py @@ -0,0 +1,228 @@ +import asyncio +import logging +from typing import Any + +from pydantic import BaseModel + +from app.core.memory.llm_tools import OpenAIEmbedderClient +from app.core.memory.memory_service import MemoryContext +from app.core.memory.utils.data import escape_lucene_query +from app.repositories.neo4j.graph_search import search_perceptual, search_perceptual_by_embedding +from app.repositories.neo4j.neo4j_connector import Neo4jConnector + +logger = logging.getLogger(__name__) + + +class PerceptualResult(BaseModel): + memories: list[dict[str, Any]] = [] + content: str = "" + keyword_raw: int = 0 + embedding_raw: int = 0 + + +class PerceptualRetrieverService: + DEFAULT_ALPHA = 0.6 + DEFAULT_FULLTEXT_SCORE_THRESHOLD = 0.5 + DEFAULT_COSINE_SCORE_THRESHOLD = 0.7 + + def __init__( + self, + ctx: MemoryContext, + embedder: OpenAIEmbedderClient, + alpha: float = DEFAULT_ALPHA, + fulltext_score_threshold: float = DEFAULT_FULLTEXT_SCORE_THRESHOLD, + cosine_score_threshold: float = DEFAULT_COSINE_SCORE_THRESHOLD + ): + self.ctx = ctx + self.alpha = alpha + self.fulltext_score_threshold = fulltext_score_threshold + self.cosine_score_threshold = cosine_score_threshold + + self.embedder: OpenAIEmbedderClient = embedder + self.connector = Neo4jConnector() + + async def search( + self, + query: str, + keywords: list[str] | None = None, + limit: int = 10 + ) -> PerceptualResult: + if keywords is None: + keywords = [query] if query else [] + + try: + kw_task = self._keyword_search(keywords, limit) + emb_task = self._embedding_search(query, limit) + kw_results, emb_results = await asyncio.gather(kw_task, emb_task, return_exceptions=True) + if isinstance(kw_results, Exception): + logger.warning(f"[PerceptualSearch] keyword search error: {kw_results}") + kw_results = [] + if isinstance(emb_results, Exception): + logger.warning(f"[PerceptualSearch] embedding search error: {emb_results}") + emb_results = [] + + reranked = self._rerank(kw_results, emb_results, limit) + + memories = [] + content_parts = [] + for record in reranked: + fmt = self._format_result(record) + fmt["score"] = round(record.get("content_score", 0), 4) + memories.append(fmt) + content_parts.append(self._build_content_text(fmt)) + + logger.info( + f"[PerceptualSearch] {len(memories)} results after rerank " + f"(keyword_raw={len(kw_results)}, embedding_raw={len(emb_results)})" + ) + return PerceptualResult( + memories=memories, + content="\n\n".join(content_parts), + keyword_raw=len(kw_results), + embedding_raw=len(emb_results), + ) + except Exception as e: + logger.error(f"[PerceptualSearch] search failed: {e}", exc_info=True) + return PerceptualResult() + finally: + await self.connector.close() + + async def _keyword_search( + self, + keywords: list[str], + limit: int + ) -> list[dict]: + seen_ids: set = set() + all_results: list[dict] = [] + + async def _one(kw: str): + escaped = escape_lucene_query(kw) + if not escaped.strip(): + return [] + r = await search_perceptual( + connector=self.connector, q=escaped, + end_user_id=self.ctx.end_user_id, limit=limit + ) + perceptuals = r.get("perceptuals", []) + return [perceptual for perceptual in perceptuals if perceptual["score"] > self.fulltext_score_threshold] + + tasks = [_one(kw) for kw in keywords] + batch = await asyncio.gather(*tasks, return_exceptions=True) + + for result in batch: + if isinstance(result, Exception): + logger.warning(f"[PerceptualSearch] keyword sub-query error: {result}") + continue + for rec in result: + rid = rec.get("id", "") + if rid and rid not in seen_ids: + seen_ids.add(rid) + all_results.append(rec) + all_results.sort(key=lambda x: float(x.get("score", 0)), reverse=True) + return all_results[:limit] + + async def _embedding_search( + self, + query: str, + limit: int + ) -> list[dict]: + r = await search_perceptual_by_embedding( + connector=self.connector, + embedder_client=self.embedder, + query_text=query, + end_user_id=self.ctx.end_user_id, + limit=limit + ) + perceptuals = r.get("perceptuals", []) + return [perceptual for perceptual in perceptuals if perceptual["score"] > self.cosine_score_threshold] + + def _rerank( + self, + keyword_results: list[dict], + embedding_results: list[dict], + limit: int, + ) -> list[dict]: + keyword_results = self._normalize_scores(keyword_results) + embedding_results = self._normalize_scores(embedding_results) + + kw_norm_map = {} + for item in keyword_results: + item_id = item["id"] + kw_norm_map[item_id] = float(item.get("normalized_score", 0)) + + emb_norm_map = {} + for item in embedding_results: + item_id = item["id"] + emb_norm_map[item_id] = float(item.get("normalized_score", 0)) + + combined = {} + for item in keyword_results: + item_id = item["id"] + combined[item_id] = item.copy() + combined[item_id]["kw_score"] = kw_norm_map.get(item_id, 0) + combined[item_id]["embedding_score"] = emb_norm_map.get(item_id, 0) + + for item in embedding_results: + item_id = item["id"] + if item_id in combined: + combined[item_id]["embedding_score"] = emb_norm_map.get(item_id, 0) + else: + combined[item_id] = item.copy() + combined[item_id]["kw_score"] = kw_norm_map.get(item_id, 0) + combined[item_id]["embedding_score"] = emb_norm_map.get(item_id, 0) + + for item in combined.values(): + kw = float(item.get("kw_score", 0) or 0) + emb = float(item.get("embedding_score", 0) or 0) + item["content_score"] = self.alpha * emb + (1 - self.alpha) * kw + + results = list(combined.values()) + results.sort(key=lambda x: x["content_score"], reverse=True) + results = results[:limit] + + logger.info( + f"[PerceptualSearch] rerank: merged={len(combined)}, after_threshold={len(results)} " + f"(alpha={self.alpha})" + ) + return results + + @staticmethod + def _normalize_scores(items: list[dict], field: str = "score") -> list[dict]: + """Min-max 归一化，将分数线性映射到 [0, 1]。""" + if not items: + return items + scores = [float(it.get(field, 0) or 0) for it in items] + min_s = min(scores) + max_s = max(scores) + diff = max_s - min_s + for it, s in zip(items, scores): + it[f"normalized_{field}"] = (s - min_s) / diff if diff > 0 else 1.0 + return items + + @staticmethod + def _format_result(record: dict) -> dict: + return { + "id": record.get("id", ""), + "perceptual_type": record.get("perceptual_type", ""), + "file_name": record.get("file_name", ""), + "file_path": record.get("file_path", ""), + "summary": record.get("summary", ""), + "topic": record.get("topic", ""), + "domain": record.get("domain", ""), + "keywords": record.get("keywords", []), + "created_at": str(record.get("created_at", "")), + "file_type": record.get("file_type", ""), + "score": record.get("score", 0), + } + + @staticmethod + def _build_content_text(formatted: dict) -> str: + content_text = (f"" + f"{formatted["file_name"]}" + f"{formatted["file_path"]}" + f"{formatted["file_type"]}" + f"{formatted["topic"]}" + f"{formatted["keywords"]}" + f"{formatted["summary"]}" + f"") + return content_text diff --git a/api/app/core/memory/read_services/query_preprocessor.py b/api/app/core/memory/read_services/query_preprocessor.py new file mode 100644 index 00000000..02d757c9 --- /dev/null +++ b/api/app/core/memory/read_services/query_preprocessor.py @@ -0,0 +1,18 @@ +# -*- coding: UTF-8 -*- +# Author: Eternity +# @Email: 1533512157@qq.com +# @Time : 2026/4/8 18:11 +import re + +from app.schemas.memory_agent_schema import AgentMemoryDataset + + +class QueryPreprocessor: + @staticmethod + def process(query: str) -> str: + text = query.strip() + if not text: + return text + + text = re.sub(rf"{"|".join(AgentMemoryDataset.PRONOUN)}", AgentMemoryDataset.NAME, text) + return text diff --git a/api/app/core/memory/storage_services/search/__init__.py b/api/app/core/memory/storage_services/search/__init__.py deleted file mode 100644 index c12c39b0..00000000 --- a/api/app/core/memory/storage_services/search/__init__.py +++ /dev/null @@ -1,143 +0,0 @@ -# -*- coding: utf-8 -*- -"""搜索服务模块 - -本模块提供统一的搜索服务接口，支持关键词搜索、语义搜索和混合搜索。 -""" - -from typing import TYPE_CHECKING - -if TYPE_CHECKING: - from app.schemas.memory_config_schema import MemoryConfig - -from app.core.memory.storage_services.search.hybrid_search import HybridSearchStrategy -from app.core.memory.storage_services.search.keyword_search import KeywordSearchStrategy -from app.core.memory.storage_services.search.search_strategy import ( - SearchResult, - SearchStrategy, -) -from app.core.memory.storage_services.search.semantic_search import ( - SemanticSearchStrategy, -) - -__all__ = [ - "SearchStrategy", - "SearchResult", - "KeywordSearchStrategy", - "SemanticSearchStrategy", - "HybridSearchStrategy", -] - - -# ============================================================================ -# 向后兼容的函数式API -# ============================================================================ -# 为了兼容旧代码，提供与 src/search.py 相同的函数式接口 - - -async def run_hybrid_search( - query_text: str, - search_type: str = "hybrid", - end_user_id: str | None = None, - apply_id: str | None = None, - user_id: str | None = None, - limit: int = 50, - include: list[str] | None = None, - alpha: float = 0.6, - use_forgetting_curve: bool = False, - memory_config: "MemoryConfig" = None, - **kwargs -) -> dict: - """运行混合搜索（向后兼容的函数式API） - - 这是一个向后兼容的包装函数，将旧的函数式API转换为新的基于类的API。 - - Args: - query_text: 查询文本 - search_type: 搜索类型（"hybrid", "keyword", "semantic"） - end_user_id: 组ID过滤 - apply_id: 应用ID过滤 - user_id: 用户ID过滤 - limit: 每个类别的最大结果数 - include: 要包含的搜索类别列表 - alpha: BM25分数权重（0.0-1.0） - use_forgetting_curve: 是否使用遗忘曲线 - memory_config: MemoryConfig object containing embedding_model_id - **kwargs: 其他参数 - - Returns: - dict: 搜索结果字典，格式与旧API兼容 - """ - from app.core.memory.llm_tools.openai_embedder import OpenAIEmbedderClient - from app.core.models.base import RedBearModelConfig - from app.db import get_db_context - from app.repositories.neo4j.neo4j_connector import Neo4jConnector - from app.services.memory_config_service import MemoryConfigService - - if not memory_config: - raise ValueError("memory_config is required for search") - - # 初始化客户端 - connector = Neo4jConnector() - with get_db_context() as db: - config_service = MemoryConfigService(db) - embedder_config_dict = config_service.get_embedder_config(str(memory_config.embedding_model_id)) - embedder_config = RedBearModelConfig(**embedder_config_dict) - embedder_client = OpenAIEmbedderClient(embedder_config) - - try: - # 根据搜索类型选择策略 - if search_type == "keyword": - strategy = KeywordSearchStrategy(connector=connector) - elif search_type == "semantic": - strategy = SemanticSearchStrategy( - connector=connector, - embedder_client=embedder_client - ) - else: # hybrid - strategy = HybridSearchStrategy( - connector=connector, - embedder_client=embedder_client, - alpha=alpha, - use_forgetting_curve=use_forgetting_curve - ) - - # 执行搜索 - result = await strategy.search( - query_text=query_text, - end_user_id=end_user_id, - limit=limit, - include=include, - alpha=alpha, - use_forgetting_curve=use_forgetting_curve, - **kwargs - ) - - # 转换为旧格式 - result_dict = result.to_dict() - - # 保存到文件（如果指定了output_path） - output_path = kwargs.get('output_path', 'search_results.json') - if output_path: - import json - import os - from datetime import datetime - - try: - # 确保目录存在 - out_dir = os.path.dirname(output_path) - if out_dir: - os.makedirs(out_dir, exist_ok=True) - - # 保存结果 - with open(output_path, "w", encoding="utf-8") as f: - json.dump(result_dict, f, ensure_ascii=False, indent=2, default=str) - print(f"Search results saved to {output_path}") - except Exception as e: - print(f"Error saving search results: {e}") - return result_dict - - finally: - await connector.close() - - -__all__.append("run_hybrid_search") diff --git a/api/app/core/memory/storage_services/search/hybrid_search.py b/api/app/core/memory/storage_services/search/hybrid_search.py deleted file mode 100644 index 4111b09c..00000000 --- a/api/app/core/memory/storage_services/search/hybrid_search.py +++ /dev/null @@ -1,408 +0,0 @@ -# # -*- coding: utf-8 -*- -# """混合搜索策略 - -# 结合关键词搜索和语义搜索的混合检索方法。 -# 支持结果重排序和遗忘曲线加权。 -# """ - -# from typing import List, Dict, Any, Optional -# import math -# from datetime import datetime -# from app.core.logging_config import get_memory_logger -# from app.repositories.neo4j.neo4j_connector import Neo4jConnector -# from app.core.memory.storage_services.search.search_strategy import SearchStrategy, SearchResult -# from app.core.memory.storage_services.search.keyword_search import KeywordSearchStrategy -# from app.core.memory.storage_services.search.semantic_search import SemanticSearchStrategy -# from app.core.memory.llm_tools.openai_embedder import OpenAIEmbedderClient -# from app.core.memory.models.variate_config import ForgettingEngineConfig -# from app.core.memory.storage_services.forgetting_engine.forgetting_engine import ForgettingEngine - -# logger = get_memory_logger(__name__) - - -# class HybridSearchStrategy(SearchStrategy): -# """混合搜索策略 - -# 结合关键词搜索和语义搜索的优势： -# - 关键词搜索：精确匹配，适合已知术语 -# - 语义搜索：语义理解，适合概念查询 -# - 混合重排序：综合两种搜索的结果 -# - 遗忘曲线：根据时间衰减调整相关性 -# """ - -# def __init__( -# self, -# connector: Optional[Neo4jConnector] = None, -# embedder_client: Optional[OpenAIEmbedderClient] = None, -# alpha: float = 0.6, -# use_forgetting_curve: bool = False, -# forgetting_config: Optional[ForgettingEngineConfig] = None -# ): -# """初始化混合搜索策略 - -# Args: -# connector: Neo4j连接器 -# embedder_client: 嵌入模型客户端 -# alpha: BM25分数权重（0.0-1.0），1-alpha为嵌入分数权重 -# use_forgetting_curve: 是否使用遗忘曲线 -# forgetting_config: 遗忘引擎配置 -# """ -# self.connector = connector -# self.embedder_client = embedder_client -# self.alpha = alpha -# self.use_forgetting_curve = use_forgetting_curve -# self.forgetting_config = forgetting_config or ForgettingEngineConfig() -# self._owns_connector = connector is None - -# # 创建子策略 -# self.keyword_strategy = KeywordSearchStrategy(connector=connector) -# self.semantic_strategy = SemanticSearchStrategy( -# connector=connector, -# embedder_client=embedder_client -# ) - -# async def __aenter__(self): -# """异步上下文管理器入口""" -# if self._owns_connector: -# self.connector = Neo4jConnector() -# self.keyword_strategy.connector = self.connector -# self.semantic_strategy.connector = self.connector -# return self - -# async def __aexit__(self, exc_type, exc_val, exc_tb): -# """异步上下文管理器出口""" -# if self._owns_connector and self.connector: -# await self.connector.close() - -# async def search( -# self, -# query_text: str, -# end_user_id: Optional[str] = None, -# limit: int = 50, -# include: Optional[List[str]] = None, -# **kwargs -# ) -> SearchResult: -# """执行混合搜索 - -# Args: -# query_text: 查询文本 -# end_user_id: 可选的组ID过滤 -# limit: 每个类别的最大结果数 -# include: 要包含的搜索类别列表 -# **kwargs: 其他搜索参数（如alpha, use_forgetting_curve） - -# Returns: -# SearchResult: 搜索结果对象 -# """ -# logger.info(f"执行混合搜索: query='{query_text}', end_user_id={end_user_id}, limit={limit}") - -# # 从kwargs中获取参数 -# alpha = kwargs.get("alpha", self.alpha) -# use_forgetting = kwargs.get("use_forgetting_curve", self.use_forgetting_curve) - -# # 获取有效的搜索类别 -# include_list = self._get_include_list(include) - -# try: -# # 并行执行关键词搜索和语义搜索 -# keyword_result = await self.keyword_strategy.search( -# query_text=query_text, -# end_user_id=end_user_id, -# limit=limit, -# include=include_list -# ) - -# semantic_result = await self.semantic_strategy.search( -# query_text=query_text, -# end_user_id=end_user_id, -# limit=limit, -# include=include_list -# ) - -# # 重排序结果 -# if use_forgetting: -# reranked_results = self._rerank_with_forgetting_curve( -# keyword_result=keyword_result, -# semantic_result=semantic_result, -# alpha=alpha, -# limit=limit -# ) -# else: -# reranked_results = self._rerank_hybrid_results( -# keyword_result=keyword_result, -# semantic_result=semantic_result, -# alpha=alpha, -# limit=limit -# ) - -# # 创建元数据 -# metadata = self._create_metadata( -# query_text=query_text, -# search_type="hybrid", -# end_user_id=end_user_id, -# limit=limit, -# include=include_list, -# alpha=alpha, -# use_forgetting_curve=use_forgetting -# ) - -# # 添加结果统计 -# metadata["keyword_results"] = keyword_result.metadata.get("result_counts", {}) -# metadata["semantic_results"] = semantic_result.metadata.get("result_counts", {}) -# metadata["total_keyword_results"] = keyword_result.total_results() -# metadata["total_semantic_results"] = semantic_result.total_results() -# metadata["total_reranked_results"] = reranked_results.total_results() - -# reranked_results.metadata = metadata - -# logger.info(f"混合搜索完成: 共找到 {reranked_results.total_results()} 条结果") -# return reranked_results - -# except Exception as e: -# logger.error(f"混合搜索失败: {e}", exc_info=True) -# # 返回空结果但包含错误信息 -# return SearchResult( -# metadata=self._create_metadata( -# query_text=query_text, -# search_type="hybrid", -# end_user_id=end_user_id, -# limit=limit, -# error=str(e) -# ) -# ) - -# def _normalize_scores( -# self, -# results: List[Dict[str, Any]], -# score_field: str = "score" -# ) -> List[Dict[str, Any]]: -# """使用z-score标准化和sigmoid转换归一化分数 - -# Args: -# results: 结果列表 -# score_field: 分数字段名 - -# Returns: -# List[Dict[str, Any]]: 归一化后的结果列表 -# """ -# if not results: -# return results - -# # 提取分数 -# scores = [] -# for item in results: -# if score_field in item: -# score = item.get(score_field) -# if score is not None and isinstance(score, (int, float)): -# scores.append(float(score)) -# else: -# scores.append(0.0) - -# if not scores or len(scores) == 1: -# # 单个分数或无分数，设置为1.0 -# for item in results: -# if score_field in item: -# item[f"normalized_{score_field}"] = 1.0 -# return results - -# # 计算均值和标准差 -# mean_score = sum(scores) / len(scores) -# variance = sum((score - mean_score) ** 2 for score in scores) / len(scores) -# std_dev = math.sqrt(variance) - -# if std_dev == 0: -# # 所有分数相同，设置为1.0 -# for item in results: -# if score_field in item: -# item[f"normalized_{score_field}"] = 1.0 -# else: -# # z-score标准化 + sigmoid转换 -# for item in results: -# if score_field in item: -# score = item[score_field] -# if score is None or not isinstance(score, (int, float)): -# score = 0.0 -# z_score = (score - mean_score) / std_dev -# normalized = 1 / (1 + math.exp(-z_score)) -# item[f"normalized_{score_field}"] = normalized - -# return results - -# def _rerank_hybrid_results( -# self, -# keyword_result: SearchResult, -# semantic_result: SearchResult, -# alpha: float, -# limit: int -# ) -> SearchResult: -# """重排序混合搜索结果 - -# Args: -# keyword_result: 关键词搜索结果 -# semantic_result: 语义搜索结果 -# alpha: BM25分数权重 -# limit: 结果限制 - -# Returns: -# SearchResult: 重排序后的结果 -# """ -# reranked_data = {} - -# for category in ["statements", "chunks", "entities", "summaries"]: -# keyword_items = getattr(keyword_result, category, []) -# semantic_items = getattr(semantic_result, category, []) - -# # 归一化分数 -# keyword_items = self._normalize_scores(keyword_items, "score") -# semantic_items = self._normalize_scores(semantic_items, "score") - -# # 合并结果 -# combined_items = {} - -# # 添加关键词结果 -# for item in keyword_items: -# item_id = item.get("id") or item.get("uuid") -# if item_id: -# combined_items[item_id] = item.copy() -# combined_items[item_id]["bm25_score"] = item.get("normalized_score", 0) -# combined_items[item_id]["embedding_score"] = 0 - -# # 添加或更新语义结果 -# for item in semantic_items: -# item_id = item.get("id") or item.get("uuid") -# if item_id: -# if item_id in combined_items: -# combined_items[item_id]["embedding_score"] = item.get("normalized_score", 0) -# else: -# combined_items[item_id] = item.copy() -# combined_items[item_id]["bm25_score"] = 0 -# combined_items[item_id]["embedding_score"] = item.get("normalized_score", 0) - -# # 计算组合分数 -# for item_id, item in combined_items.items(): -# bm25_score = item.get("bm25_score", 0) -# embedding_score = item.get("embedding_score", 0) -# combined_score = alpha * bm25_score + (1 - alpha) * embedding_score -# item["combined_score"] = combined_score - -# # 排序并限制结果 -# sorted_items = sorted( -# combined_items.values(), -# key=lambda x: x.get("combined_score", 0), -# reverse=True -# )[:limit] - -# reranked_data[category] = sorted_items - -# return SearchResult( -# statements=reranked_data.get("statements", []), -# chunks=reranked_data.get("chunks", []), -# entities=reranked_data.get("entities", []), -# summaries=reranked_data.get("summaries", []) -# ) - -# def _parse_datetime(self, value: Any) -> Optional[datetime]: -# """解析日期时间字符串""" -# if value is None: -# return None -# if isinstance(value, datetime): -# return value -# if isinstance(value, str): -# s = value.strip() -# if not s: -# return None -# try: -# return datetime.fromisoformat(s) -# except Exception: -# return None -# return None - -# def _rerank_with_forgetting_curve( -# self, -# keyword_result: SearchResult, -# semantic_result: SearchResult, -# alpha: float, -# limit: int -# ) -> SearchResult: -# """使用遗忘曲线重排序混合搜索结果 - -# Args: -# keyword_result: 关键词搜索结果 -# semantic_result: 语义搜索结果 -# alpha: BM25分数权重 -# limit: 结果限制 - -# Returns: -# SearchResult: 重排序后的结果 -# """ -# engine = ForgettingEngine(self.forgetting_config) -# now_dt = datetime.now() - -# reranked_data = {} - -# for category in ["statements", "chunks", "entities", "summaries"]: -# keyword_items = getattr(keyword_result, category, []) -# semantic_items = getattr(semantic_result, category, []) - -# # 归一化分数 -# keyword_items = self._normalize_scores(keyword_items, "score") -# semantic_items = self._normalize_scores(semantic_items, "score") - -# # 合并结果 -# combined_items = {} - -# for src_items, is_embedding in [(keyword_items, False), (semantic_items, True)]: -# for item in src_items: -# item_id = item.get("id") or item.get("uuid") -# if not item_id: -# continue - -# if item_id not in combined_items: -# combined_items[item_id] = item.copy() -# combined_items[item_id]["bm25_score"] = 0 -# combined_items[item_id]["embedding_score"] = 0 - -# if is_embedding: -# combined_items[item_id]["embedding_score"] = item.get("normalized_score", 0) -# else: -# combined_items[item_id]["bm25_score"] = item.get("normalized_score", 0) - -# # 计算分数并应用遗忘权重 -# for item_id, item in combined_items.items(): -# bm25_score = float(item.get("bm25_score", 0) or 0) -# embedding_score = float(item.get("embedding_score", 0) or 0) -# combined_score = alpha * bm25_score + (1 - alpha) * embedding_score - -# # 计算时间衰减 -# dt = self._parse_datetime(item.get("created_at")) -# if dt is None: -# time_elapsed_days = 0.0 -# else: -# time_elapsed_days = max(0.0, (now_dt - dt).total_seconds() / 86400.0) - -# memory_strength = 1.0 # 默认强度 -# forgetting_weight = engine.calculate_weight( -# time_elapsed=time_elapsed_days, -# memory_strength=memory_strength -# ) - -# final_score = combined_score * forgetting_weight -# item["combined_score"] = final_score -# item["forgetting_weight"] = forgetting_weight -# item["time_elapsed_days"] = time_elapsed_days - -# # 排序并限制结果 -# sorted_items = sorted( -# combined_items.values(), -# key=lambda x: x.get("combined_score", 0), -# reverse=True -# )[:limit] - -# reranked_data[category] = sorted_items - -# return SearchResult( -# statements=reranked_data.get("statements", []), -# chunks=reranked_data.get("chunks", []), -# entities=reranked_data.get("entities", []), -# summaries=reranked_data.get("summaries", []) -# ) diff --git a/api/app/core/memory/storage_services/search/keyword_search.py b/api/app/core/memory/storage_services/search/keyword_search.py index 2458cf30..e69de29b 100644 --- a/api/app/core/memory/storage_services/search/keyword_search.py +++ b/api/app/core/memory/storage_services/search/keyword_search.py @@ -1,122 +0,0 @@ -# -*- coding: utf-8 -*- -"""关键词搜索策略 - -实现基于关键词的全文搜索功能。 -使用Neo4j的全文索引进行高效的文本匹配。 -""" - -from typing import List, Optional -from app.core.logging_config import get_memory_logger -from app.repositories.neo4j.neo4j_connector import Neo4jConnector -from app.core.memory.storage_services.search.search_strategy import SearchStrategy, SearchResult -from app.repositories.neo4j.graph_search import search_graph - -logger = get_memory_logger(__name__) - - -class KeywordSearchStrategy(SearchStrategy): - """关键词搜索策略 - - 使用Neo4j全文索引进行关键词匹配搜索。 - 支持跨陈述句、实体、分块和摘要的搜索。 - """ - - def __init__(self, connector: Optional[Neo4jConnector] = None): - """初始化关键词搜索策略 - - Args: - connector: Neo4j连接器，如果为None则创建新连接 - """ - self.connector = connector - self._owns_connector = connector is None - - async def __aenter__(self): - """异步上下文管理器入口""" - if self._owns_connector: - self.connector = Neo4jConnector() - return self - - async def __aexit__(self, exc_type, exc_val, exc_tb): - """异步上下文管理器出口""" - if self._owns_connector and self.connector: - await self.connector.close() - - async def search( - self, - query_text: str, - end_user_id: Optional[str] = None, - limit: int = 50, - include: Optional[List[str]] = None, - **kwargs - ) -> SearchResult: - """执行关键词搜索 - - Args: - query_text: 查询文本 - end_user_id: 可选的组ID过滤 - limit: 每个类别的最大结果数 - include: 要包含的搜索类别列表 - **kwargs: 其他搜索参数 - - Returns: - SearchResult: 搜索结果对象 - """ - logger.info(f"执行关键词搜索: query='{query_text}', end_user_id={end_user_id}, limit={limit}") - - # 获取有效的搜索类别 - include_list = self._get_include_list(include) - - # 确保连接器已初始化 - if not self.connector: - self.connector = Neo4jConnector() - - try: - # 调用底层的关键词搜索函数 - results_dict = await search_graph( - connector=self.connector, - query=query_text, - end_user_id=end_user_id, - limit=limit, - include=include_list - ) - - # 创建元数据 - metadata = self._create_metadata( - query_text=query_text, - search_type="keyword", - end_user_id=end_user_id, - limit=limit, - include=include_list - ) - - # 添加结果统计 - metadata["result_counts"] = { - category: len(results_dict.get(category, [])) - for category in include_list - } - metadata["total_results"] = sum(metadata["result_counts"].values()) - - # 构建SearchResult对象 - search_result = SearchResult( - statements=results_dict.get("statements", []), - chunks=results_dict.get("chunks", []), - entities=results_dict.get("entities", []), - summaries=results_dict.get("summaries", []), - metadata=metadata - ) - - logger.info(f"关键词搜索完成: 共找到 {search_result.total_results()} 条结果") - return search_result - - except Exception as e: - logger.error(f"关键词搜索失败: {e}", exc_info=True) - # 返回空结果但包含错误信息 - return SearchResult( - metadata=self._create_metadata( - query_text=query_text, - search_type="keyword", - end_user_id=end_user_id, - limit=limit, - error=str(e) - ) - ) diff --git a/api/app/core/memory/storage_services/search/search_strategy.py b/api/app/core/memory/storage_services/search/search_strategy.py deleted file mode 100644 index 3a670dd6..00000000 --- a/api/app/core/memory/storage_services/search/search_strategy.py +++ /dev/null @@ -1,125 +0,0 @@ -# -*- coding: utf-8 -*- -"""搜索策略基类 - -定义搜索策略的抽象接口和统一的搜索结果数据结构。 -遵循策略模式（Strategy Pattern）和开放-关闭原则（OCP）。 -""" - -from abc import ABC, abstractmethod -from typing import List, Dict, Any, Optional -from pydantic import BaseModel, Field -from datetime import datetime - - -class SearchResult(BaseModel): - """统一的搜索结果数据结构 - - Attributes: - statements: 陈述句搜索结果列表 - chunks: 分块搜索结果列表 - entities: 实体搜索结果列表 - summaries: 摘要搜索结果列表 - metadata: 搜索元数据（如查询时间、结果数量等） - """ - statements: List[Dict[str, Any]] = Field(default_factory=list, description="陈述句搜索结果") - chunks: List[Dict[str, Any]] = Field(default_factory=list, description="分块搜索结果") - entities: List[Dict[str, Any]] = Field(default_factory=list, description="实体搜索结果") - summaries: List[Dict[str, Any]] = Field(default_factory=list, description="摘要搜索结果") - metadata: Dict[str, Any] = Field(default_factory=dict, description="搜索元数据") - - def total_results(self) -> int: - """返回所有类别的结果总数""" - return ( - len(self.statements) + - len(self.chunks) + - len(self.entities) + - len(self.summaries) - ) - - def to_dict(self) -> Dict[str, Any]: - """转换为字典格式""" - return { - "statements": self.statements, - "chunks": self.chunks, - "entities": self.entities, - "summaries": self.summaries, - "metadata": self.metadata - } - - -class SearchStrategy(ABC): - """搜索策略抽象基类 - - 定义所有搜索策略必须实现的接口。 - 遵循依赖反转原则（DIP）：高层模块依赖抽象而非具体实现。 - """ - - @abstractmethod - async def search( - self, - query_text: str, - end_user_id: Optional[str] = None, - limit: int = 50, - include: Optional[List[str]] = None, - **kwargs - ) -> SearchResult: - """执行搜索 - - Args: - query_text: 查询文本 - end_user_id: 可选的组ID过滤 - limit: 每个类别的最大结果数 - include: 要包含的搜索类别列表（statements, chunks, entities, summaries） - **kwargs: 其他搜索参数 - - Returns: - SearchResult: 统一的搜索结果对象 - """ - pass - - def _create_metadata( - self, - query_text: str, - search_type: str, - end_user_id: Optional[str] = None, - limit: int = 50, - **kwargs - ) -> Dict[str, Any]: - """创建搜索元数据 - - Args: - query_text: 查询文本 - search_type: 搜索类型 - end_user_id: 组ID - limit: 结果限制 - **kwargs: 其他元数据 - - Returns: - Dict[str, Any]: 元数据字典 - """ - metadata = { - "query": query_text, - "search_type": search_type, - "end_user_id": end_user_id, - "limit": limit, - "timestamp": datetime.now().isoformat() - } - metadata.update(kwargs) - return metadata - - def _get_include_list(self, include: Optional[List[str]] = None) -> List[str]: - """获取要包含的搜索类别列表 - - Args: - include: 用户指定的类别列表 - - Returns: - List[str]: 有效的类别列表 - """ - default_include = ["statements", "chunks", "entities", "summaries"] - if include is None: - return default_include - - # 验证并过滤有效的类别 - valid_categories = set(default_include) - return [cat for cat in include if cat in valid_categories] diff --git a/api/app/core/memory/storage_services/search/semantic_search.py b/api/app/core/memory/storage_services/search/semantic_search.py deleted file mode 100644 index 8d4eb05f..00000000 --- a/api/app/core/memory/storage_services/search/semantic_search.py +++ /dev/null @@ -1,166 +0,0 @@ -# -*- coding: utf-8 -*- -"""语义搜索策略 - -实现基于向量嵌入的语义搜索功能。 -使用余弦相似度进行语义匹配。 -""" - -from typing import Any, Dict, List, Optional - -from app.core.logging_config import get_memory_logger -from app.core.memory.llm_tools.openai_embedder import OpenAIEmbedderClient -from app.core.memory.storage_services.search.search_strategy import ( - SearchResult, - SearchStrategy, -) -from app.core.memory.utils.config import definitions as config_defs -from app.core.models.base import RedBearModelConfig -from app.db import get_db_context -from app.repositories.neo4j.graph_search import search_graph_by_embedding -from app.repositories.neo4j.neo4j_connector import Neo4jConnector -from app.services.memory_config_service import MemoryConfigService - -logger = get_memory_logger(__name__) - - -class SemanticSearchStrategy(SearchStrategy): - """语义搜索策略 - - 使用向量嵌入和余弦相似度进行语义搜索。 - 支持跨陈述句、分块、实体和摘要的语义匹配。 - """ - - def __init__( - self, - connector: Optional[Neo4jConnector] = None, - embedder_client: Optional[OpenAIEmbedderClient] = None - ): - """初始化语义搜索策略 - - Args: - connector: Neo4j连接器，如果为None则创建新连接 - embedder_client: 嵌入模型客户端，如果为None则根据配置创建 - """ - self.connector = connector - self.embedder_client = embedder_client - self._owns_connector = connector is None - self._owns_embedder = embedder_client is None - - async def __aenter__(self): - """异步上下文管理器入口""" - if self._owns_connector: - self.connector = Neo4jConnector() - if self._owns_embedder: - self.embedder_client = self._create_embedder_client() - return self - - async def __aexit__(self, exc_type, exc_val, exc_tb): - """异步上下文管理器出口""" - if self._owns_connector and self.connector: - await self.connector.close() - - def _create_embedder_client(self) -> OpenAIEmbedderClient: - """创建嵌入模型客户端 - - Returns: - OpenAIEmbedderClient: 嵌入模型客户端实例 - """ - try: - # 从数据库读取嵌入器配置 - with get_db_context() as db: - config_service = MemoryConfigService(db) - embedder_config_dict = config_service.get_embedder_config(config_defs.SELECTED_EMBEDDING_ID) - rb_config = RedBearModelConfig( - model_name=embedder_config_dict["model_name"], - provider=embedder_config_dict["provider"], - api_key=embedder_config_dict["api_key"], - base_url=embedder_config_dict["base_url"], - type="llm" - ) - return OpenAIEmbedderClient(model_config=rb_config) - except Exception as e: - logger.error(f"创建嵌入模型客户端失败: {e}", exc_info=True) - raise - - async def search( - self, - query_text: str, - end_user_id: Optional[str] = None, - limit: int = 50, - include: Optional[List[str]] = None, - **kwargs - ) -> SearchResult: - """执行语义搜索 - - Args: - query_text: 查询文本 - end_user_id: 可选的组ID过滤 - limit: 每个类别的最大结果数 - include: 要包含的搜索类别列表 - **kwargs: 其他搜索参数 - - Returns: - SearchResult: 搜索结果对象 - """ - logger.info(f"执行语义搜索: query='{query_text}', end_user_id={end_user_id}, limit={limit}") - - # 获取有效的搜索类别 - include_list = self._get_include_list(include) - - # 确保连接器和嵌入器已初始化 - if not self.connector: - self.connector = Neo4jConnector() - if not self.embedder_client: - self.embedder_client = self._create_embedder_client() - - try: - # 调用底层的语义搜索函数 - results_dict = await search_graph_by_embedding( - connector=self.connector, - embedder_client=self.embedder_client, - query_text=query_text, - end_user_id=end_user_id, - limit=limit, - include=include_list - ) - - # 创建元数据 - metadata = self._create_metadata( - query_text=query_text, - search_type="semantic", - end_user_id=end_user_id, - limit=limit, - include=include_list - ) - - # 添加结果统计 - metadata["result_counts"] = { - category: len(results_dict.get(category, [])) - for category in include_list - } - metadata["total_results"] = sum(metadata["result_counts"].values()) - - # 构建SearchResult对象 - search_result = SearchResult( - statements=results_dict.get("statements", []), - chunks=results_dict.get("chunks", []), - entities=results_dict.get("entities", []), - summaries=results_dict.get("summaries", []), - metadata=metadata - ) - - logger.info(f"语义搜索完成: 共找到 {search_result.total_results()} 条结果") - return search_result - - except Exception as e: - logger.error(f"语义搜索失败: {e}", exc_info=True) - # 返回空结果但包含错误信息 - return SearchResult( - metadata=self._create_metadata( - query_text=query_text, - search_type="semantic", - end_user_id=end_user_id, - limit=limit, - error=str(e) - ) - ) diff --git a/api/app/repositories/neo4j/cypher_queries.py b/api/app/repositories/neo4j/cypher_queries.py index 4b5273ac..9bd46f94 100644 --- a/api/app/repositories/neo4j/cypher_queries.py +++ b/api/app/repositories/neo4j/cypher_queries.py @@ -1452,6 +1452,30 @@ ON CREATE SET r.end_user_id = edge.end_user_id, RETURN elementId(r) AS uuid """ +SEARCH_PERCEPTUAL_BY_USER_ID = """ +MATCH (p:Perceptual) +WHERE p.end_user_id = $end_user_id +RETURN p.id AS id, + p.summary_embedding AS summary_embedding +""" + +SEARCH_PERCEPTUAL_BY_IDS = """ +MATCH (p:Perceptual) +WHERE p.id IN $ids +RETURN p.id AS id, + p.end_user_id AS end_user_id, + p.perceptual_type AS perceptual_type, + p.file_path AS file_path, + p.file_name AS file_name, + p.file_ext AS file_ext, + p.summary AS summary, + p.keywords AS keywords, + p.topic AS topic, + p.domain AS domain, + p.created_at AS created_at, + p.file_type AS file_type +""" + SEARCH_PERCEPTUAL_BY_KEYWORD = """ CALL db.index.fulltext.queryNodes("perceptualFulltext", $query) YIELD node AS p, score WHERE p.end_user_id = $end_user_id @@ -1471,24 +1495,3 @@ RETURN p.id AS id, ORDER BY score DESC LIMIT $limit """ - -PERCEPTUAL_EMBEDDING_SEARCH = """ -CALL db.index.vector.queryNodes('perceptual_summary_embedding_index', $limit * 100, $embedding) -YIELD node AS p, score -WHERE p.summary_embedding IS NOT NULL AND p.end_user_id = $end_user_id -RETURN p.id AS id, - p.end_user_id AS end_user_id, - p.perceptual_type AS perceptual_type, - p.file_path AS file_path, - p.file_name AS file_name, - p.file_ext AS file_ext, - p.summary AS summary, - p.keywords AS keywords, - p.topic AS topic, - p.domain AS domain, - p.created_at AS created_at, - p.file_type AS file_type, - score -ORDER BY score DESC -LIMIT $limit -""" diff --git a/api/app/repositories/neo4j/graph_search.py b/api/app/repositories/neo4j/graph_search.py index a191dad6..665b68ff 100644 --- a/api/app/repositories/neo4j/graph_search.py +++ b/api/app/repositories/neo4j/graph_search.py @@ -3,13 +3,15 @@ import logging from typing import Any, Dict, List, Optional from app.core.memory.utils.data.text_utils import escape_lucene_query +import numpy as np + +from app.core.memory.llm_tools import OpenAIEmbedderClient from app.repositories.neo4j.cypher_queries import ( CHUNK_EMBEDDING_SEARCH, COMMUNITY_EMBEDDING_SEARCH, ENTITY_EMBEDDING_SEARCH, EXPAND_COMMUNITY_STATEMENTS, MEMORY_SUMMARY_EMBEDDING_SEARCH, - PERCEPTUAL_EMBEDDING_SEARCH, SEARCH_CHUNK_BY_CHUNK_ID, SEARCH_CHUNKS_BY_CONTENT, SEARCH_COMMUNITIES_BY_KEYWORD, @@ -17,7 +19,6 @@ from app.repositories.neo4j.cypher_queries import ( SEARCH_ENTITIES_BY_NAME, SEARCH_ENTITIES_BY_NAME_OR_ALIAS, SEARCH_MEMORY_SUMMARIES_BY_KEYWORD, - SEARCH_PERCEPTUAL_BY_KEYWORD, SEARCH_STATEMENTS_BY_CREATED_AT, SEARCH_STATEMENTS_BY_KEYWORD, SEARCH_STATEMENTS_BY_KEYWORD_TEMPORAL, @@ -28,14 +29,41 @@ from app.repositories.neo4j.cypher_queries import ( SEARCH_STATEMENTS_L_CREATED_AT, SEARCH_STATEMENTS_L_VALID_AT, STATEMENT_EMBEDDING_SEARCH, + SEARCH_PERCEPTUAL_BY_KEYWORD, + SEARCH_PERCEPTUAL_BY_IDS, + SEARCH_PERCEPTUAL_BY_USER_ID, ) - # 使用新的仓储层 from app.repositories.neo4j.neo4j_connector import Neo4jConnector logger = logging.getLogger(__name__) +def cosine_similarity_search( + query: list[float], + vectors: list[list[float]], + limit: int +) -> dict[int, float]: + if not vectors: + return {} + vectors: np.ndarray = np.array(vectors, dtype=np.float32) + vectors_norm = vectors / np.linalg.norm(vectors, axis=1, keepdims=True) + query: np.ndarray = np.array(query, dtype=np.float32) + query_norm = query / np.linalg.norm(query) + + similarities = vectors_norm @ query_norm + similarities = (similarities + 1) / 2 + top_k = min(limit, similarities.shape[0]) + if top_k <= 0: + return {} + top_indices = np.argpartition(-similarities, top_k - 1)[-top_k:] + top_indices = top_indices[np.argsort(-similarities[top_indices])] + result = {} + for idx in top_indices: + result[idx] = similarities[idx] + return result + + async def _update_activation_values_batch( connector: Neo4jConnector, nodes: List[Dict[str, Any]], @@ -352,7 +380,7 @@ async def search_graph_by_embedding( query_text: str, end_user_id: Optional[str] = None, limit: int = 50, - include: List[str] = ["statements", "chunks", "entities", "summaries"], + include=None, ) -> Dict[str, List[Dict[str, Any]]]: """ Embedding-based semantic search across Statements, Chunks, and Entities. @@ -365,6 +393,8 @@ async def search_graph_by_embedding( - Filters by end_user_id if provided - Returns up to 'limit' per included type """ + if include is None: + include = ["statements", "chunks", "entities", "summaries"] import time # Get embedding for the query @@ -1011,7 +1041,7 @@ async def search_perceptual( async def search_perceptual_by_embedding( connector: Neo4jConnector, - embedder_client, + embedder_client: OpenAIEmbedderClient, query_text: str, end_user_id: Optional[str] = None, limit: int = 10, @@ -1040,11 +1070,22 @@ async def search_perceptual_by_embedding( try: perceptuals = await connector.execute_query( - PERCEPTUAL_EMBEDDING_SEARCH, - embedding=embedding, + SEARCH_PERCEPTUAL_BY_USER_ID, end_user_id=end_user_id, - limit=limit, ) + ids = [item['id'] for item in perceptuals] + vectors = [item['summary_embedding'] for item in perceptuals] + sim_res = cosine_similarity_search(embedding, vectors, limit=limit) + perceptual_res = { + ids[idx]: score + for idx, score in sim_res.items() + } + perceptuals = await connector.execute_query( + SEARCH_PERCEPTUAL_BY_IDS, + ids=list(perceptual_res.keys()) + ) + for perceptual in perceptuals: + perceptual["score"] = perceptual_res[perceptual["id"]] except Exception as e: logger.warning(f"search_perceptual_by_embedding: vector search failed: {e}") perceptuals = [] From ac51ccaf1f23b75d79ae98172a323791b3d5c821 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 13 Apr 2026 14:04:31 +0800 Subject: [PATCH 014/245] fix(web): ui fix --- .../Editor/plugin/AutocompletePlugin.tsx | 46 ++++++++++--------- .../components/Properties/VariableSelect.tsx | 2 +- 2 files changed, 26 insertions(+), 22 deletions(-) diff --git a/web/src/views/Workflow/components/Editor/plugin/AutocompletePlugin.tsx b/web/src/views/Workflow/components/Editor/plugin/AutocompletePlugin.tsx index 9f718826..6d3b7a4f 100644 --- a/web/src/views/Workflow/components/Editor/plugin/AutocompletePlugin.tsx +++ b/web/src/views/Workflow/components/Editor/plugin/AutocompletePlugin.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2025-12-23 16:22:51 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-13 11:12:18 + * @Last Modified time: 2026-04-13 14:00:07 */ import { useEffect, useLayoutEffect, useState, useRef, type FC } from 'react'; import { useLexicalComposerContext } from '@lexical/react/LexicalComposerContext'; @@ -285,23 +285,24 @@ const AutocompletePlugin: FC<{ options: Suggestion[] }> = ({ options }) => { ref={popupRef} data-autocomplete-popup="true" onMouseDown={(e) => e.preventDefault()} - className="rb:min-w-70 rb:max-h-57.5 rb:overflow-y-auto rb:fixed rb:z-1000 rb:bg-white rb:rounded-lg rb:border-[0.5px] rb:border-[#EBEBEB] rb:shadow-[0px_2px_6px_0px_rgba(0,0,0,0.1)] rb:py-3 rb:px-2" + className="rb:fixed rb:z-1000 rb:bg-white rb:rounded-lg rb:border-[0.5px] rb:border-[#EBEBEB] rb:shadow-[0px_2px_6px_0px_rgba(0,0,0,0.1)] rb:py-3 rb:px-2" style={{ top: popupPosition.top, left: popupPosition.left, }} > - - {Object.entries(groupedSuggestions).map(([nodeId, nodeOptions]) => { - const nodeName = nodeOptions[0]?.nodeData?.name || nodeId; - return ( -
- {nodeName !== 'undefined' && -
- {nodeName} -
- } - +
+ + {Object.entries(groupedSuggestions).map(([nodeId, nodeOptions]) => { + const nodeName = nodeOptions[0]?.nodeData?.name || nodeId; + return ( +
+ {nodeName !== 'undefined' && +
+ {nodeName} +
+ } + {nodeOptions.map((option) => { const globalIndex = flatOptions.indexOf(option); const isExpanded = expandedParent?.key === option.key; @@ -344,20 +345,21 @@ const AutocompletePlugin: FC<{ options: Suggestion[] }> = ({ options }) => { } {option.dataType && {option.dataType}} - {hasChildren &&
} + {hasChildren &&
} ); })} - -
- ); - })} - + +
+ ); + })} + +
{/* Child variables panel - floats to the left */} {expandedParent?.children?.length && (
= ({ options }) => { onClick={() => !child.disabled && insertMention(child)} onMouseEnter={() => setSelectedIndex(childIndex)} > - {child.label} + + {`{x}`} {child.label} + {child.dataType && {child.dataType}} ); diff --git a/web/src/views/Workflow/components/Properties/VariableSelect.tsx b/web/src/views/Workflow/components/Properties/VariableSelect.tsx index c0207cb5..5523c06e 100644 --- a/web/src/views/Workflow/components/Properties/VariableSelect.tsx +++ b/web/src/views/Workflow/components/Properties/VariableSelect.tsx @@ -334,7 +334,7 @@ const VariableSelect: FC = ({ {s.dataType && {s.dataType}} - {hasChildren &&
} + {hasChildren &&
} ); From 5bb9ce9018eee47a8b8277c446d4f2d9090ffae4 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Mon, 13 Apr 2026 14:40:57 +0800 Subject: [PATCH 015/245] fix(user): add user retrieval regardless of active status and update DSL config enrichment Added `get_user_by_id_regardless_active` in user repository to support activation/deactivation workflows, updated `user_service` to use it, and refactored `_enrich_release_config` in `app_dsl_service` to accept `default_model_config_id` as a parameter instead of reading from config dict. --- api/app/repositories/user_repository.py | 4 ++++ api/app/services/app_dsl_service.py | 10 ++++------ api/app/services/user_service.py | 2 +- 3 files changed, 9 insertions(+), 7 deletions(-) diff --git a/api/app/repositories/user_repository.py b/api/app/repositories/user_repository.py index 2dd76b04..6874f9bf 100644 --- a/api/app/repositories/user_repository.py +++ b/api/app/repositories/user_repository.py @@ -297,6 +297,10 @@ def get_user_by_id(db: Session, user_id: uuid.UUID) -> Optional[User]: """根据ID获取用户""" return UserRepository(db).get_user_by_id(user_id) +def get_user_by_id_regardless_active(db: Session, user_id: uuid.UUID) -> Optional[User]: + """根据ID获取用户（不过滤 is_active，用于启用/禁用场景）""" + return db.query(User).filter(User.id == user_id).first() + def get_user_by_email(db: Session, email: str) -> Optional[User]: """根据邮箱获取用户""" return UserRepository(db).get_user_by_email(email) diff --git a/api/app/services/app_dsl_service.py b/api/app/services/app_dsl_service.py index 8c198be4..3a897109 100644 --- a/api/app/services/app_dsl_service.py +++ b/api/app/services/app_dsl_service.py @@ -73,15 +73,14 @@ class AppDslService: AppType.MULTI_AGENT: "multi_agent_config", AppType.WORKFLOW: "workflow" }.get(app.type, "config") - config_data = self._enrich_release_config(app.type, release.config or {}) + config_data = self._enrich_release_config(app.type, release.config or {}, release.default_model_config_id) dsl = {**meta, "app": app_meta, config_key: config_data} return yaml.dump(dsl, default_flow_style=False, allow_unicode=True), f"{release.name}_v{release.version_name}.yaml" - def _enrich_release_config(self, app_type: str, cfg: dict) -> dict: + def _enrich_release_config(self, app_type: str, cfg: dict, default_model_config_id=None) -> dict: if app_type == AppType.AGENT: enriched = {**cfg} - if "default_model_config_id" in cfg: - enriched["default_model_config_ref"] = self._model_ref(cfg["default_model_config_id"]) + enriched["default_model_config_ref"] = self._model_ref(default_model_config_id) if "knowledge_retrieval" in cfg: enriched["knowledge_retrieval"] = self._enrich_knowledge_retrieval(cfg["knowledge_retrieval"]) if "tools" in cfg: @@ -91,8 +90,7 @@ class AppDslService: return enriched if app_type == AppType.MULTI_AGENT: enriched = {**cfg} - if "default_model_config_id" in cfg: - enriched["default_model_config_ref"] = self._model_ref(cfg["default_model_config_id"]) + enriched["default_model_config_ref"] = self._model_ref(default_model_config_id) if "master_agent_id" in cfg: enriched["master_agent_ref"] = self._release_ref(cfg["master_agent_id"]) if "sub_agents" in cfg: diff --git a/api/app/services/user_service.py b/api/app/services/user_service.py index 3122d282..43a58c5f 100644 --- a/api/app/services/user_service.py +++ b/api/app/services/user_service.py @@ -285,7 +285,7 @@ def activate_user(db: Session, user_id_to_activate: uuid.UUID, current_user: Use try: # 查找用户 business_logger.debug(f"查找待激活用户: {user_id_to_activate}") - db_user = user_repository.get_user_by_id(db, user_id=user_id_to_activate) + db_user = user_repository.get_user_by_id_regardless_active(db, user_id=user_id_to_activate) if not db_user: business_logger.warning(f"用户不存在: {user_id_to_activate}") raise BusinessException("用户不存在", code=BizCode.USER_NOT_FOUND) From 2b067ce08a013654fb875b09fbbb1eb04617061b Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 13 Apr 2026 15:35:45 +0800 Subject: [PATCH 016/245] fix(web): third variable --- .../OpenStatementSettingModal.tsx | 4 +-- .../AddChatVariable/ChatVariableModal.tsx | 4 +-- .../components/Nodes/ConditionNode.tsx | 2 ++ .../Properties/AssignmentList/index.tsx | 27 ++++++++++++++++--- .../components/Properties/CaseList/index.tsx | 4 ++- .../Properties/ConditionList/index.tsx | 4 ++- .../Properties/GroupVariableList/index.tsx | 8 ++++-- .../views/Workflow/hooks/useWorkflowGraph.ts | 3 ++- 8 files changed, 43 insertions(+), 13 deletions(-) diff --git a/web/src/views/ApplicationConfig/components/FeaturesConfig/OpenStatementSettingModal.tsx b/web/src/views/ApplicationConfig/components/FeaturesConfig/OpenStatementSettingModal.tsx index 91d0d19f..a46d973a 100644 --- a/web/src/views/ApplicationConfig/components/FeaturesConfig/OpenStatementSettingModal.tsx +++ b/web/src/views/ApplicationConfig/components/FeaturesConfig/OpenStatementSettingModal.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-03-05 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-07 16:58:10 + * @Last Modified time: 2026-04-13 15:13:36 */ import { forwardRef, useImperativeHandle, useState } from 'react'; import { Button, Form, Input, Flex, App } from 'antd'; @@ -36,8 +36,6 @@ const OpenStatementSettingModal = forwardRef(); - console.log('chatVariables', chatVariables) - const handleClose = () => { setVisible(false); form.resetFields(); diff --git a/web/src/views/Workflow/components/AddChatVariable/ChatVariableModal.tsx b/web/src/views/Workflow/components/AddChatVariable/ChatVariableModal.tsx index 5c80fab1..e4f62432 100644 --- a/web/src/views/Workflow/components/AddChatVariable/ChatVariableModal.tsx +++ b/web/src/views/Workflow/components/AddChatVariable/ChatVariableModal.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2025-12-30 13:59:36 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-13 12:16:00 + * @Last Modified time: 2026-04-13 15:26:33 */ import { forwardRef, useImperativeHandle, useState, useRef, useMemo } from 'react'; import { Form, Input, Select, InputNumber, Button, Row, Col, Flex } from 'antd'; @@ -136,7 +136,7 @@ const ChatVariableModal = forwardRef { const defaultValue = Array.isArray(values.defaultValue) ? values.defaultValue.filter((v: any) => v !== undefined && v !== null && v !== '') - : values.type.includes('object') + : values.type.includes('object') && values.defaultValue ? JSON.parse(values.defaultValue) : values.defaultValue; refresh({ ...values, defaultValue }, editIndex); diff --git a/web/src/views/Workflow/components/Nodes/ConditionNode.tsx b/web/src/views/Workflow/components/Nodes/ConditionNode.tsx index 996ae5dd..625a1b4d 100644 --- a/web/src/views/Workflow/components/Nodes/ConditionNode.tsx +++ b/web/src/views/Workflow/components/Nodes/ConditionNode.tsx @@ -31,6 +31,8 @@ const ConditionNode: ReactShapeConfig['component'] = ({ node }) => { }; const labelRender = (value: string) => { const filterOption = variableList.find(vo => `{{${vo.value}}}` === value) + ?? variableList.flatMap(vo => vo.children ?? []).find(child => `{{${child.value}}}` === value) + ?? variableList.flatMap(vo => vo.children ?? []).flatMap((child: any) => child.children ?? []).find((grandchild: any) => `{{${grandchild.value}}}` === value) if (filterOption) { return ( diff --git a/web/src/views/Workflow/components/Properties/AssignmentList/index.tsx b/web/src/views/Workflow/components/Properties/AssignmentList/index.tsx index 98f86ecf..e24d531d 100644 --- a/web/src/views/Workflow/components/Properties/AssignmentList/index.tsx +++ b/web/src/views/Workflow/components/Properties/AssignmentList/index.tsx @@ -30,6 +30,25 @@ const operationsObj = { ], } +const filterByDataType = (options: Suggestion[], dataType: string): Suggestion[] => + options.reduce((acc, vo) => { + if (vo.children?.length) { + const children = vo.children.reduce((cacc, child) => { + if (child.children?.length) { + const grandchildren = child.children.filter(gc => gc.dataType === dataType); + if (grandchildren.length) cacc.push({ ...child, children: grandchildren }); + } else if (child.dataType === dataType) { + cacc.push(child); + } + return cacc; + }, []); + if (children.length) acc.push({ ...vo, children }); + } else if (vo.dataType === dataType) { + acc.push(vo); + } + return acc; + }, []); + const AssignmentList: FC = ({ parentName, options = [], @@ -59,7 +78,9 @@ const AssignmentList: FC = ({ {fields.map(({ key, name, ...restField }) => { const variableSelector = form.getFieldValue([parentName, name, 'variable_selector']); - const selectedOption = options.find(option => `{{${option.value}}}` === variableSelector); + const selectedOption = options.find(option => `{{${option.value}}}` === variableSelector) + ?? options.flatMap(o => o.children ?? []).find(child => `{{${child.value}}}` === variableSelector) + ?? options.flatMap(o => o.children ?? []).flatMap((c: any) => c.children ?? []).find((gc: any) => `{{${gc.value}}}` === variableSelector); const dataType = selectedOption?.dataType; const operationOptions = dataType === 'number' ? operationsObj.number : operationsObj.default; @@ -119,7 +140,7 @@ const AssignmentList: FC = ({ {dataType === 'number' && operation === 'cover' ? vo.dataType === dataType) : options} + options={dataType ? filterByDataType(options, dataType) : options} size={size} className="rb:flex-1!" variant="filled" @@ -150,7 +171,7 @@ const AssignmentList: FC = ({ : vo.dataType === dataType) : options} + options={dataType ? filterByDataType(options, dataType) : options} size={size} className="rb:flex-1!" variant="filled" diff --git a/web/src/views/Workflow/components/Properties/CaseList/index.tsx b/web/src/views/Workflow/components/Properties/CaseList/index.tsx index 40353f64..f0a58517 100644 --- a/web/src/views/Workflow/components/Properties/CaseList/index.tsx +++ b/web/src/views/Workflow/components/Properties/CaseList/index.tsx @@ -329,7 +329,9 @@ const CaseList: FC = ({ const currentExpression = currentCase.expressions?.[conditionIndex] || {}; const currentOperator = currentExpression.operator; const leftFieldValue = currentExpression.left; - const leftFieldOption = options.find(option => `{{${option.value}}}` === leftFieldValue); + const leftFieldOption = options.find(option => `{{${option.value}}}` === leftFieldValue) + ?? options.flatMap(o => o.children ?? []).find(child => `{{${child.value}}}` === leftFieldValue) + ?? options.flatMap(o => o.children ?? []).flatMap((c: any) => c.children ?? []).find((gc: any) => `{{${gc.value}}}` === leftFieldValue); const leftFieldType = leftFieldOption?.dataType; const hideRightField = currentOperator === 'empty' || currentOperator === 'not_empty' || leftFieldType === 'file' || leftFieldType === 'array[object]' || leftFieldType === 'array[file]'; const operatorList = leftFieldType && operatorsObj[leftFieldType] diff --git a/web/src/views/Workflow/components/Properties/ConditionList/index.tsx b/web/src/views/Workflow/components/Properties/ConditionList/index.tsx index ddf92971..3e9f3261 100644 --- a/web/src/views/Workflow/components/Properties/ConditionList/index.tsx +++ b/web/src/views/Workflow/components/Properties/ConditionList/index.tsx @@ -155,7 +155,9 @@ const ConditionList: FC = ({ const currentExpression = expressions[index] || {}; const currentOperator = currentExpression.operator; const leftFieldValue = currentExpression.left; - const leftFieldOption = options.find(option => `{{${option.value}}}` === leftFieldValue); + const leftFieldOption = options.find(option => `{{${option.value}}}` === leftFieldValue) + ?? options.flatMap(o => o.children ?? []).find(child => `{{${child.value}}}` === leftFieldValue) + ?? options.flatMap(o => o.children ?? []).flatMap((c: any) => c.children ?? []).find((gc: any) => `{{${gc.value}}}` === leftFieldValue); const leftFieldType = leftFieldOption?.dataType; const hideRightField = currentOperator === 'empty' || currentOperator === 'not_empty' || ['array[object]', 'object'].includes(leftFieldType as string); const operatorList = leftFieldType && ['array[object]', 'object'].includes(leftFieldType) diff --git a/web/src/views/Workflow/components/Properties/GroupVariableList/index.tsx b/web/src/views/Workflow/components/Properties/GroupVariableList/index.tsx index 0100707c..24cdc89a 100644 --- a/web/src/views/Workflow/components/Properties/GroupVariableList/index.tsx +++ b/web/src/views/Workflow/components/Properties/GroupVariableList/index.tsx @@ -62,14 +62,18 @@ const GroupVariableList: FC = ({ */ useEffect(() => { if (!isCanAdd && value[0]) { - const firstVariable = options.find(opt => `{{${opt.value}}}` === value[0]); + const firstVariable = options.find(opt => `{{${opt.value}}}` === value[0]) + ?? options.flatMap(o => o.children ?? []).find(c => `{{${c.value}}}` === value[0]) + ?? options.flatMap(o => o.children ?? []).flatMap((c: any) => c.children ?? []).find((gc: any) => `{{${gc.value}}}` === value[0]); if (firstVariable) { form.setFieldValue(['group_type', 'output'], firstVariable.dataType); } } else if (isCanAdd) { value.forEach((item: any, index: number) => { if (item?.value?.[0]) { - const firstVariable = options.find(opt => `{{${opt.value}}}` === item.value[0]); + const firstVariable = options.find(opt => `{{${opt.value}}}` === item.value[0]) + ?? options.flatMap(o => o.children ?? []).find(c => `{{${c.value}}}` === item.value[0]) + ?? options.flatMap(o => o.children ?? []).flatMap((c: any) => c.children ?? []).find((gc: any) => `{{${gc.value}}}` === item.value[0]); if (firstVariable) { form.setFieldValue(['group_type', index], firstVariable.dataType); } diff --git a/web/src/views/Workflow/hooks/useWorkflowGraph.ts b/web/src/views/Workflow/hooks/useWorkflowGraph.ts index 5d0bb9c6..ac82b230 100644 --- a/web/src/views/Workflow/hooks/useWorkflowGraph.ts +++ b/web/src/views/Workflow/hooks/useWorkflowGraph.ts @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 15:17:48 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-13 12:00:09 + * @Last Modified time: 2026-04-13 15:33:58 */ import { Clipboard, Graph, Keyboard, MiniMap, Node, Snapline, type Edge } from '@antv/x6'; import { register } from '@antv/x6-react-shape'; @@ -111,6 +111,7 @@ export const useWorkflowGraph = ({ graphRef.current.getNodes().forEach(node => { const data = node.getData() if (data?.type === 'if-else' || data?.type === 'question-classifier') { + console.log('chatVariables', chatVariables) node.setData({ ...data, chatVariables }, { silent: true }) } }) From ef8c7093b5aec0c87e49c72e561c98acc2c11c41 Mon Sep 17 00:00:00 2001 From: lanceyq <1982376970@qq.com> Date: Mon, 13 Apr 2026 18:32:43 +0800 Subject: [PATCH 017/245] refactor(memory): use MemorySummary node count for implicit memory metrics - Replace Statement-based implicit memory count (count/3) with actual MemorySummary node count filtered by DERIVED_FROM_STATEMENT relationship - Add minimum threshold of 5 MemorySummary nodes before reporting data - Add _build_empty_profile() to return structured empty profile when insufficient data exists, skipping unnecessary LLM calls --- api/app/services/implicit_memory_service.py | 61 +++++++++++++++++++++ api/app/services/user_memory_service.py | 22 ++++---- 2 files changed, 72 insertions(+), 11 deletions(-) diff --git a/api/app/services/implicit_memory_service.py b/api/app/services/implicit_memory_service.py index 4bd11deb..10504fe7 100644 --- a/api/app/services/implicit_memory_service.py +++ b/api/app/services/implicit_memory_service.py @@ -379,12 +379,59 @@ class ImplicitMemoryService: raise + def _build_empty_profile(self) -> dict: + """构建 MemorySummary 不足时返回的固定空白画像数据""" + now_ms = int(datetime.now().timestamp() * 1000) + insufficient = "Insufficient data for analysis" + + def _empty_dimension(name: str) -> dict: + return { + "evidence": [insufficient], + "reasoning": f"No clear evidence found for {name} dimension", + "percentage": 0.0, + "dimension_name": name, + "confidence_level": 20, + } + + def _empty_category(name: str) -> dict: + return { + "evidence": [insufficient], + "percentage": 25.0, + "category_name": name, + "trending_direction": None, + } + + return { + "habits": [], + "portrait": { + "aesthetic": _empty_dimension("aesthetic"), + "creativity": _empty_dimension("creativity"), + "literature": _empty_dimension("literature"), + "technology": _empty_dimension("technology"), + "historical_trends": None, + "analysis_timestamp": now_ms, + "total_summaries_analyzed": 0, + }, + "preferences": [], + "interest_areas": { + "art": _empty_category("art"), + "tech": _empty_category("tech"), + "music": _empty_category("music"), + "lifestyle": _empty_category("lifestyle"), + "analysis_timestamp": now_ms, + "total_summaries_analyzed": 0, + }, + } + async def generate_complete_profile( self, user_id: str ) -> dict: """生成完整的用户画像（包含所有4个模块） + 需要该用户的 MemorySummary 节点数量 >= 5 才会真正调用 LLM 生成画像， + 否则返回固定的空白画像数据。 + Args: user_id: 用户ID @@ -394,6 +441,20 @@ class ImplicitMemoryService: logger.info(f"生成完整用户画像: user={user_id}") try: + # 前置检查：查询该用户有效的 MemorySummary 节点数量（排除孤立节点） + query = """ + MATCH (n:MemorySummary)-[:DERIVED_FROM_STATEMENT]->(:Statement) + WHERE n.end_user_id = $end_user_id + RETURN count(DISTINCT n) as count + """ + result = await self.neo4j_connector.execute_query(query, end_user_id=user_id) + memory_summary_count = result[0]["count"] if result and len(result) > 0 else 0 + logger.info(f"用户 MemorySummary 节点数量: {memory_summary_count} (user={user_id})") + + if memory_summary_count < 5: + logger.info(f"MemorySummary 数量不足 5（当前 {memory_summary_count}），返回空白画像: user={user_id}") + return self._build_empty_profile() + # 并行调用4个分析方法 preferences, portrait, interest_areas, habits = await asyncio.gather( self.get_preference_tags(user_id=user_id), diff --git a/api/app/services/user_memory_service.py b/api/app/services/user_memory_service.py index ab51d922..dcbedba6 100644 --- a/api/app/services/user_memory_service.py +++ b/api/app/services/user_memory_service.py @@ -1500,7 +1500,7 @@ async def analytics_memory_types( 2. 工作记忆 (WORKING_MEMORY) = 会话数量（通过 ConversationRepository.get_conversation_by_user_id 获取） 3. 短期记忆 (SHORT_TERM_MEMORY) = /short_term 接口返回的问答对数量 4. 显性记忆 (EXPLICIT_MEMORY) = 情景记忆 + 语义记忆（通过 MemoryBaseService.get_explicit_memory_count 获取） - 5. 隐性记忆 (IMPLICIT_MEMORY) = Statement 节点数量的三分之一 + 5. 隐性记忆 (IMPLICIT_MEMORY) = MemorySummary 节点数量（需 >= 5 才显示，否则为 0） 6. 情绪记忆 (EMOTIONAL_MEMORY) = 情绪标签统计总数（通过 MemoryBaseService.get_emotional_memory_count 获取） 7. 情景记忆 (EPISODIC_MEMORY) = memory_summary（通过 MemoryBaseService.get_episodic_memory_count 获取） 8. 遗忘记忆 (FORGET_MEMORY) = 激活值低于阈值的节点数（通过 MemoryBaseService.get_forget_memory_count 获取） @@ -1557,23 +1557,23 @@ async def analytics_memory_types( logger.warning(f"获取会话数量失败，工作记忆数量设为0: {str(e)}") work_count = 0 - # 获取隐性记忆数量（基于 Statement 节点数量的三分之一） + # 获取隐性记忆数量（基于有关联关系的 MemorySummary 节点数量，需 >= 5 才计入） implicit_count = 0 if end_user_id: try: - # 查询 Statement 节点数量 + # 只统计有 DERIVED_FROM_STATEMENT 关系的 MemorySummary 节点，排除孤立节点 query = """ - MATCH (n:Statement) + MATCH (n:MemorySummary)-[:DERIVED_FROM_STATEMENT]->(:Statement) WHERE n.end_user_id = $end_user_id - RETURN count(n) as count + RETURN count(DISTINCT n) as count """ result = await _neo4j_connector.execute_query(query, end_user_id=end_user_id) - statement_count = result[0]["count"] if result and len(result) > 0 else 0 - # 取三分之一作为隐性记忆数量 - implicit_count = round(statement_count / 3) - logger.debug(f"隐性记忆数量（Statement数量的1/3）: {implicit_count} (Statement总数={statement_count}, end_user_id={end_user_id})") + memory_summary_count = result[0]["count"] if result and len(result) > 0 else 0 + # 仅当 MemorySummary 节点数量 >= 5 时才显示数量，否则为 0 + implicit_count = memory_summary_count if memory_summary_count >= 5 else 0 + logger.debug(f"隐性记忆数量（有效MemorySummary节点数）: {implicit_count} (有效MemorySummary总数={memory_summary_count}, end_user_id={end_user_id})") except Exception as e: - logger.warning(f"获取Statement数量失败，隐性记忆数量设为0: {str(e)}") + logger.warning(f"获取MemorySummary数量失败，隐性记忆数量设为0: {str(e)}") implicit_count = 0 # 原有的基于行为习惯的统计方式（已注释） @@ -1639,7 +1639,7 @@ async def analytics_memory_types( "WORKING_MEMORY": work_count, # 工作记忆（基于会话数量） "SHORT_TERM_MEMORY": short_term_count, # 短期记忆（基于问答对数量） "EXPLICIT_MEMORY": explicit_count, # 显性记忆（情景记忆 + 语义记忆） - "IMPLICIT_MEMORY": implicit_count, # 隐性记忆（Statement数量的1/3） + "IMPLICIT_MEMORY": implicit_count, # 隐性记忆（MemorySummary节点数，需>=5） "EMOTIONAL_MEMORY": emotion_count, # 情绪记忆（使用情绪标签统计） "EPISODIC_MEMORY": episodic_count, # 情景记忆 "FORGET_MEMORY": forget_count # 遗忘记忆（激活值低于阈值） From 095f4e3001529862e0bf2f9e074a3c5b1d3df619 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 13 Apr 2026 18:33:45 +0800 Subject: [PATCH 018/245] feat(web): app import and Overwrite --- web/src/i18n/en.ts | 2 ++ web/src/i18n/zh.ts | 2 ++ .../components/ConfigHeader.tsx | 18 ++++++++++--- .../components/UploadModal.tsx | 27 ++++++++++++------- 4 files changed, 36 insertions(+), 13 deletions(-) diff --git a/web/src/i18n/en.ts b/web/src/i18n/en.ts index c878476b..fc3a041d 100644 --- a/web/src/i18n/en.ts +++ b/web/src/i18n/en.ts @@ -1522,6 +1522,8 @@ export const en = { "version":"app_release_id" // string, optional, application version ID; specify a historical release version ID, or omit to use the currently active version; }`, + uploadCover: 'Import and Overwrite', + refresh: 'Refresh Current Page', }, userMemory: { userMemory: 'User Memory', diff --git a/web/src/i18n/zh.ts b/web/src/i18n/zh.ts index da80fed2..01c766b8 100644 --- a/web/src/i18n/zh.ts +++ b/web/src/i18n/zh.ts @@ -857,6 +857,8 @@ export const zh = { "version":"app_release_id" //string，可选，应用版本ID；指定历史发布版本ID，不传则使用当前生效版本； }`, + uploadCover: '导入并覆盖', + refresh: '刷新当前页', }, table: { totalRecords: '共 {{total}} 条记录' diff --git a/web/src/views/ApplicationConfig/components/ConfigHeader.tsx b/web/src/views/ApplicationConfig/components/ConfigHeader.tsx index d38a657a..a3a9bd7b 100644 --- a/web/src/views/ApplicationConfig/components/ConfigHeader.tsx +++ b/web/src/views/ApplicationConfig/components/ConfigHeader.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 16:27:52 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-07 16:28:33 + * @Last Modified time: 2026-04-13 18:19:27 */ import { type FC, useRef, useMemo } from 'react'; import { useNavigate, useParams } from 'react-router-dom'; @@ -12,13 +12,14 @@ import { useTranslation } from 'react-i18next'; import clsx from 'clsx'; import styles from '../index.module.css' -import type { Application, ApplicationModalRef } from '@/views/ApplicationManagement/types'; +import type { Application, ApplicationModalRef, UploadWorkflowModalRef } from '@/views/ApplicationManagement/types'; import ApplicationModal from '@/views/ApplicationManagement/components/ApplicationModal' import type { CopyModalRef, AgentRef, ClusterRef, WorkflowRef, FeaturesConfigForm } from '../types' import { deleteApplication, appExport } from '@/api/application' import CopyModal from './CopyModal' import PageHeader from '@/components/Layout/PageHeader' import CheckList from '@/views/Workflow/components/CheckList' +import UploadModal from '@/views/ApplicationManagement/components/UploadModal' /** * Tab keys for application configuration @@ -77,6 +78,7 @@ const ConfigHeader: FC = ({ const { id, source } = useParams(); const applicationModalRef = useRef(null); const copyModalRef = useRef(null); + const uploadModalRef = useRef(null); /** * Format tab items for display @@ -111,6 +113,9 @@ const ConfigHeader: FC = ({ case 'delete': handleDelete() break; + case 'uploadCover': + uploadModalRef.current?.handleOpen() + break } } /** @@ -165,11 +170,11 @@ const ConfigHeader: FC = ({ * Format dropdown menu items */ const formatMenuItems = useMemo(() => { - const items = (application?.type !== 'multi_agent' ? ['edit', 'copy', 'export', 'delete'] : ['edit', 'copy', 'delete']).map(key => ({ + const items = (application?.type !== 'multi_agent' ? ['edit', 'copy', 'export', 'uploadCover', 'delete'] : ['edit', 'copy', 'delete']).map(key => ({ key, icon:
, danger: key === 'delete', - label: t(`common.${key}`), + label: key === 'uploadCover' ? t('application.uploadCover') : t(`common.${key}`), })) return items }, [t, handleClick, application]) @@ -261,6 +266,11 @@ const ConfigHeader: FC = ({ refresh={refresh} /> + ); }; diff --git a/web/src/views/ApplicationManagement/components/UploadModal.tsx b/web/src/views/ApplicationManagement/components/UploadModal.tsx index a7acc093..4211e72b 100644 --- a/web/src/views/ApplicationManagement/components/UploadModal.tsx +++ b/web/src/views/ApplicationManagement/components/UploadModal.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-28 14:08:14 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-12 17:19:46 + * @Last Modified time: 2026-04-13 18:17:32 */ /** * UploadModal Component @@ -28,6 +28,7 @@ import { appImport } from '@/api/application' interface UploadModalProps { /** Function to refresh the parent component after workflow import */ refresh: () => void; + id?: string; } @@ -46,10 +47,11 @@ const steps = [ * @param {React.Ref} ref - Ref for imperative methods */ const UploadModal = forwardRef(({ - refresh + refresh, + id }, ref) => { const { t } = useTranslation(); - + // State management const [visible, setVisible] = useState(false); // Modal visibility const [form] = Form.useForm<{ file: File[] }>(); // Form instance @@ -87,8 +89,8 @@ const UploadModal = forwardRef(({ */ const handleSave = () => { const values = form.getFieldsValue(); - - switch(current) { + + switch (current) { case 0: // Step 1: Upload file if (!values.file || values.file.length === 0) { message.warning(t('application.pleaseUploadFile')); @@ -96,6 +98,9 @@ const UploadModal = forwardRef(({ } const formData = new FormData(); formData.append('file', values.file[0]); + if (id) { + formData.append('app_id', id) + } setLoading(true) // Call import API @@ -134,8 +139,12 @@ const UploadModal = forwardRef(({ setTimeout(() => { switch (type) { case 'detail': - // Open application detail page in new tab - window.open(`/#/application/config/${appId}`, '_blank'); + if (id) { + window.location.reload(); + } else { + // Open application detail page in new tab + window.open(`/#/application/config/${appId}`, '_blank'); + } break; } }, 100) @@ -171,7 +180,7 @@ const UploadModal = forwardRef(({ loading={loading} onClick={() => handleJump('detail')} > - {t('application.gotoDetail')} + {id ? t('application.refresh') : t('application.gotoDetail')} ] default: @@ -244,7 +253,7 @@ const UploadModal = forwardRef(({ loading={loading} onClick={() => handleJump('detail')} > - {t('application.gotoDetail')} + {id ? t('application.refresh') : t('application.gotoDetail')} ]} /> From 10f1089198a52b377439c971cc8ef6f621a6c380 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Mon, 13 Apr 2026 18:38:12 +0800 Subject: [PATCH 019/245] feat(workflow): refactor iteration runtime to support independent subgraph per task feat(app): support file metadata in chat messages and DSL app overwrite - Extended chat message file objects with `name`, `size`, and `file_type` fields across app_chat_service and workflow_service - Added ability to overwrite existing app configurations via DSL import in app_dsl_service, including type validation and config update logic for AgentConfig, MultiAgentConfig, and WorkflowConfig --- api/app/controllers/app_controller.py | 10 +- .../workflow/nodes/cycle_graph/iteration.py | 172 ++++++++++++------ .../core/workflow/nodes/cycle_graph/node.py | 17 +- api/app/schemas/app_schema.py | 2 + api/app/services/app_chat_service.py | 30 ++- api/app/services/app_dsl_service.py | 137 +++++++++++--- api/app/services/draft_run_service.py | 15 +- api/app/services/workflow_service.py | 5 +- 8 files changed, 288 insertions(+), 100 deletions(-) diff --git a/api/app/controllers/app_controller.py b/api/app/controllers/app_controller.py index db3c7536..b4a209f8 100644 --- a/api/app/controllers/app_controller.py +++ b/api/app/controllers/app_controller.py @@ -1250,9 +1250,11 @@ async def export_app( async def import_app( file: UploadFile = File(...), db: Session = Depends(get_db), - current_user: User = Depends(get_current_user) + current_user: User = Depends(get_current_user), + app_id: Optional[str] = Form(None), ): """从 YAML 文件导入 agent / multi_agent / workflow 应用。 + 传入 app_id 时覆盖该应用的配置（类型必须一致），否则创建新应用。 跨空间/跨租户导入时，模型/工具/知识库会按名称匹配，匹配不到则置空并返回 warnings。 """ if not file.filename.lower().endswith((".yaml", ".yml")): @@ -1263,13 +1265,15 @@ async def import_app( if not dsl or "app" not in dsl: return fail(msg="YAML 格式无效，缺少 app 字段", code=BizCode.BAD_REQUEST) - new_app, warnings = AppDslService(db).import_dsl( + target_app_id = uuid.UUID(app_id) if app_id else None + result_app, warnings = AppDslService(db).import_dsl( dsl=dsl, workspace_id=current_user.current_workspace_id, tenant_id=current_user.tenant_id, user_id=current_user.id, + app_id=target_app_id, ) return success( - data={"app": app_schema.App.model_validate(new_app), "warnings": warnings}, + data={"app": app_schema.App.model_validate(result_app), "warnings": warnings}, msg="应用导入成功" + ("，但部分资源需手动配置" if warnings else "") ) diff --git a/api/app/core/workflow/nodes/cycle_graph/iteration.py b/api/app/core/workflow/nodes/cycle_graph/iteration.py index cf7ac976..7652e21f 100644 --- a/api/app/core/workflow/nodes/cycle_graph/iteration.py +++ b/api/app/core/workflow/nodes/cycle_graph/iteration.py @@ -28,86 +28,135 @@ class IterationRuntime: def __init__( self, - start_id: str, stream: bool, - graph: CompiledStateGraph, node_id: str, config: dict[str, Any], state: WorkflowState, variable_pool: VariablePool, - child_variable_pool: VariablePool, + cycle_nodes: list, + cycle_edges: list, ): """ Initialize the iteration runtime. Args: - graph: Compiled workflow graph capable of async invocation. - node_id: Unique identifier of the loop node. - config: Dictionary containing iteration node configuration. - state: Current workflow state at the point of iteration. + stream: Whether to run in streaming mode. When True, each iteration + uses graph.astream and emits cycle_item events in real time. + When False, graph.ainvoke is used instead. + node_id: The unique identifier of the iteration node in the workflow. + Also used as the variable namespace for item/index inside + the subgraph (e.g. {{ node_id.item }}). + config: Raw configuration dict for the iteration node, parsed into + IterationNodeConfig. Controls input/output variable selectors, + parallel execution settings, and output flattening. + state: The parent workflow state at the point the iteration node is + entered. Each task receives a copy of this state as its + starting point. + variable_pool: The parent VariablePool containing all variables available + at the time the iteration node executes, including sys.*, + conv.*, and outputs from upstream nodes. Used as the source + for deep-copying into each task's independent child pool. + cycle_nodes: List of node config dicts belonging to this iteration's + subgraph (i.e. nodes whose cycle field equals node_id). + Passed to GraphBuilder when constructing each task's subgraph. + cycle_edges: List of edge config dicts connecting nodes within the subgraph. + Passed to GraphBuilder alongside cycle_nodes. """ - self.start_id = start_id self.stream = stream - self.graph = graph self.state = state self.node_id = node_id self.typed_config = IterationNodeConfig(**config) self.looping = True self.variable_pool = variable_pool - self.child_variable_pool = child_variable_pool + self.cycle_nodes = cycle_nodes + self.cycle_edges = cycle_edges self.event_write = get_stream_writer() - self.checkpoint = RunnableConfig( - configurable={ - "thread_id": uuid.uuid4() - } - ) self.output_value = None self.result: list = [] - async def _init_iteration_state(self, item, idx): + def _build_child_graph(self) -> tuple[CompiledStateGraph, VariablePool, str]: """ - Initialize a per-iteration copy of the workflow state. + Build an independent compiled subgraph for a single iteration task. - Args: - item: Current element from the input array for this iteration. - idx: Index of the element in the input array. + Each call creates a brand-new VariablePool by deep-copying the parent pool, + then passes it to GraphBuilder. GraphBuilder binds this pool to every node's + execution closure at build time, so the pool and the subgraph always reference + the same object. This is the key design invariant: item/index written into the + pool after build will be visible to all nodes inside the subgraph. Returns: - A copy of the workflow state with iteration-specific variables set. + graph: The compiled LangGraph subgraph ready for invocation. + child_pool: The VariablePool bound to this subgraph's node closures. + Callers must write item/index into this pool before invoking + the graph, and read output from it after invocation. + start_node_id: The ID of the CYCLE_START node inside the subgraph, + used to set the initial activation signal in workflow state. """ - loopstate = WorkflowState( - **self.state + from app.core.workflow.engine.graph_builder import GraphBuilder + child_pool = VariablePool() + child_pool.copy(self.variable_pool) + builder = GraphBuilder( + {"nodes": self.cycle_nodes, "edges": self.cycle_edges}, + stream=self.stream, + variable_pool=child_pool, + cycle=self.node_id, ) - self.child_variable_pool.copy(self.variable_pool) - await self.child_variable_pool.new(self.node_id, "item", item, VariableType.type_map(item), mut=True) - await self.child_variable_pool.new(self.node_id, "index", item, VariableType.type_map(item), mut=True) - loopstate["node_outputs"][self.node_id] = { - "item": item, - "index": idx, - } + graph = builder.build() + return graph, builder.variable_pool, builder.start_node_id + + async def _init_iteration_state(self, item, idx, child_pool: VariablePool, start_id: str): + """ + Initialize the workflow state for a single iteration. + + Writes the current item and its index into child_pool under the iteration + node's namespace (e.g. iteration_xxx.item, iteration_xxx.index), making them + accessible to downstream nodes inside the subgraph via variable selectors. + + Also prepares a copy of the parent workflow state with: + - node_outputs[node_id] set to {item, index} so the state snapshot is consistent + with the pool values. + - looping flag set to 1 (active) to signal the subgraph is inside a cycle. + - activate[start_id] set to True to trigger the CYCLE_START node. + + Args: + item: The current element from the input array. + idx: The zero-based index of this element in the input array. + child_pool: The VariablePool bound to this iteration's subgraph. + Must be the same object returned by _build_child_graph. + start_id: The ID of the CYCLE_START node inside the subgraph. + + Returns: + A WorkflowState instance ready to be passed to graph.ainvoke or graph.astream. + """ + loopstate = WorkflowState(**self.state) + await child_pool.new(self.node_id, "item", item, VariableType.type_map(item), mut=True) + await child_pool.new(self.node_id, "index", idx, VariableType.type_map(idx), mut=True) + loopstate["node_outputs"][self.node_id] = {"item": item, "index": idx} loopstate["looping"] = 1 - loopstate["activate"][self.start_id] = True + loopstate["activate"][start_id] = True return loopstate - def merge_conv_vars(self): - self.variable_pool.variables["conv"].update( - self.child_variable_pool.variables["conv"] - ) + def _merge_conv_vars(self, child_pool: VariablePool): + self.variable_pool.variables["conv"].update(child_pool.variables["conv"]) async def run_task(self, item, idx): """ Execute a single iteration asynchronously. + Each task builds its own subgraph so the variable pool closure is independent. - Args: - item: The input element for this iteration. - idx: The index of this iteration. + Returns: + Tuple of (idx, output, result, child_pool, stopped) """ + graph, child_pool, start_id = self._build_child_graph() + checkpoint = RunnableConfig(configurable={"thread_id": uuid.uuid4()}) + init_state = await self._init_iteration_state(item, idx, child_pool, start_id) + if self.stream: - async for event in self.graph.astream( - await self._init_iteration_state(item, idx), + async for event in graph.astream( + init_state, stream_mode=["debug"], - config=self.checkpoint + config=checkpoint ): if isinstance(event, tuple) and len(event) == 2: mode, data = event @@ -117,7 +166,6 @@ class IterationRuntime: event_type = data.get("type") payload = data.get("payload", {}) node_name = payload.get("name") - if node_name and node_name.startswith("nop"): continue if event_type == "task_result": @@ -140,17 +188,13 @@ class IterationRuntime: "token_usage": result.get("node_outputs", {}).get(node_name, {}).get("token_usage") } }) - result = self.graph.get_state(config=self.checkpoint).values + result = graph.get_state(config=checkpoint).values else: - result = await self.graph.ainvoke(await self._init_iteration_state(item, idx)) - output = self.child_variable_pool.get_value(self.output_value) - if isinstance(output, list) and self.typed_config.flatten: - self.result.extend(output) - else: - self.result.append(output) - if result["looping"] == 2: - self.looping = False - return result + result = await graph.ainvoke(init_state) + + output = child_pool.get_value(self.output_value) + stopped = result["looping"] == 2 + return idx, output, result, child_pool, stopped def _create_iteration_tasks(self, array_obj, idx): """ @@ -196,16 +240,32 @@ class IterationRuntime: tasks = self._create_iteration_tasks(array_obj, idx) logger.info(f"Iteration node {self.node_id}: running, concurrency {len(tasks)}") idx += self.typed_config.parallel_count - child_state.extend(await asyncio.gather(*tasks)) - self.merge_conv_vars() + batch = await asyncio.gather(*tasks) + # Sort by idx to preserve order, then collect results + batch_sorted = sorted(batch, key=lambda x: x[0]) + for _, output, result, child_pool, stopped in batch_sorted: + if isinstance(output, list) and self.typed_config.flatten: + self.result.extend(output) + else: + self.result.append(output) + child_state.append(result) + if stopped: + self.looping = False + self._merge_conv_vars(batch_sorted[-1][3]) else: # Execute iterations sequentially while idx < len(array_obj) and self.looping: logger.info(f"Iteration node {self.node_id}: running") item = array_obj[idx] - result = await self.run_task(item, idx) - self.merge_conv_vars() + _, output, result, child_pool, stopped = await self.run_task(item, idx) + if isinstance(output, list) and self.typed_config.flatten: + self.result.extend(output) + else: + self.result.append(output) + self._merge_conv_vars(child_pool) child_state.append(result) + if stopped: + self.looping = False idx += 1 logger.info(f"Iteration node {self.node_id}: execution completed") return { diff --git a/api/app/core/workflow/nodes/cycle_graph/node.py b/api/app/core/workflow/nodes/cycle_graph/node.py index 68c83025..002c34df 100644 --- a/api/app/core/workflow/nodes/cycle_graph/node.py +++ b/api/app/core/workflow/nodes/cycle_graph/node.py @@ -123,7 +123,7 @@ class CycleGraphNode(BaseNode): return cycle_nodes, cycle_edges - def build_graph(self): + def build_graph(self, variable_pool: VariablePool): """ Build and compile the internal subgraph for this cycle node. @@ -135,6 +135,7 @@ class CycleGraphNode(BaseNode): from app.core.workflow.engine.graph_builder import GraphBuilder self.child_variable_pool = VariablePool() + self.child_variable_pool.copy(variable_pool) builder = GraphBuilder( { "nodes": self.cycle_nodes, @@ -165,8 +166,8 @@ class CycleGraphNode(BaseNode): Raises: RuntimeError: If the node type is unsupported. """ - self.build_graph() if self.node_type == NodeType.LOOP: + self.build_graph(variable_pool) return await LoopRuntime( start_id=self.start_node_id, stream=False, @@ -179,20 +180,19 @@ class CycleGraphNode(BaseNode): ).run() if self.node_type == NodeType.ITERATION: return await IterationRuntime( - start_id=self.start_node_id, stream=False, - graph=self.graph, node_id=self.node_id, config=self.config, state=state, variable_pool=variable_pool, - child_variable_pool=self.child_variable_pool + cycle_nodes=self.cycle_nodes, + cycle_edges=self.cycle_edges, ).run() raise RuntimeError("Unknown cycle node type") async def execute_stream(self, state: WorkflowState, variable_pool: VariablePool): - self.build_graph() if self.node_type == NodeType.LOOP: + self.build_graph(variable_pool) yield { "__final__": True, "result": await LoopRuntime( @@ -211,14 +211,13 @@ class CycleGraphNode(BaseNode): yield { "__final__": True, "result": await IterationRuntime( - start_id=self.start_node_id, stream=True, - graph=self.graph, node_id=self.node_id, config=self.config, state=state, variable_pool=variable_pool, - child_variable_pool=self.child_variable_pool + cycle_nodes=self.cycle_nodes, + cycle_edges=self.cycle_edges, ).run() } return diff --git a/api/app/schemas/app_schema.py b/api/app/schemas/app_schema.py index 5f73cde1..1ed98f68 100644 --- a/api/app/schemas/app_schema.py +++ b/api/app/schemas/app_schema.py @@ -44,6 +44,8 @@ class FileInput(BaseModel): upload_file_id: Optional[uuid.UUID] = Field(None, description="已上传文件ID（local_file时必填）") url: Optional[str] = Field(None, description="远程URL（remote_url时必填）") file_type: Optional[str] = Field(None, description="具体文件格式（如image/jpg、audio/wav、document/docx、video/mp4）") + name: Optional[str] = Field(None, description="文件名") + size: Optional[int] = Field(None, description="文件大小（字节）") _content = None diff --git a/api/app/services/app_chat_service.py b/api/app/services/app_chat_service.py index ec0c4b79..94362176 100644 --- a/api/app/services/app_chat_service.py +++ b/api/app/services/app_chat_service.py @@ -26,6 +26,7 @@ from app.services.model_service import ModelApiKeyService from app.services.multi_agent_orchestrator import MultiAgentOrchestrator from app.services.multimodal_service import MultimodalService from app.services.workflow_service import WorkflowService +from app.models.file_metadata_model import FileMetadata logger = get_business_logger() @@ -219,10 +220,21 @@ class AppChatService: } if files: for f in files: - # url = await MultimodalService(self.db).get_file_url(f) + name, size = f.name, f.size + if f.transfer_method.value == "local_file" and f.upload_file_id and (not name or not size): + meta = self.db.query(FileMetadata).filter( + FileMetadata.id == f.upload_file_id, + FileMetadata.status == "completed" + ).first() + if meta: + name = name or meta.file_name + size = size or meta.file_size human_meta["files"].append({ "type": f.type, - "url": f.url + "url": f.url, + "name": name, + "size": size, + "file_type": f.file_type, }) if processed_files: @@ -510,9 +522,21 @@ class AppChatService: if files: for f in files: + name, size = f.name, f.size + if f.transfer_method.value == "local_file" and f.upload_file_id and (not name or not size): + meta = self.db.query(FileMetadata).filter( + FileMetadata.id == f.upload_file_id, + FileMetadata.status == "completed" + ).first() + if meta: + name = name or meta.file_name + size = size or meta.file_size human_meta["files"].append({ "type": f.type, - "url": f.url + "url": f.url, + "name": name, + "size": size, + "file_type": f.file_type, }) if processed_files: human_meta["history_files"] = { diff --git a/api/app/services/app_dsl_service.py b/api/app/services/app_dsl_service.py index 8c198be4..612b93fe 100644 --- a/api/app/services/app_dsl_service.py +++ b/api/app/services/app_dsl_service.py @@ -229,8 +229,11 @@ class AppDslService: workspace_id: uuid.UUID, tenant_id: uuid.UUID, user_id: uuid.UUID, + app_id: Optional[uuid.UUID] = None, ) -> tuple[App, list[str]]: - """解析 DSL，创建应用及配置，返回 (new_app, warnings)""" + """解析 DSL，创建或覆盖应用配置，返回 (app, warnings)。 + app_id 不为空时：校验类型一致后覆盖配置；为空时创建新应用。 + """ app_meta = dsl.get("app", {}) app_type = app_meta.get("type") if app_type not in (AppType.AGENT, AppType.MULTI_AGENT, AppType.WORKFLOW): @@ -239,6 +242,9 @@ class AppDslService: warnings: list[str] = [] now = datetime.datetime.now() + if app_id is not None: + return self._overwrite_dsl(dsl, app_id, app_type, workspace_id, tenant_id, warnings, now) + new_app = App( id=uuid.uuid4(), workspace_id=workspace_id, @@ -258,11 +264,57 @@ class AppDslService: self.db.add(new_app) self.db.flush() + self._write_config(new_app.id, app_type, dsl, workspace_id, tenant_id, warnings, now, create=True) + + self.db.commit() + self.db.refresh(new_app) + return new_app, warnings + + def _overwrite_dsl( + self, + dsl: dict, + app_id: uuid.UUID, + app_type: str, + workspace_id: uuid.UUID, + tenant_id: uuid.UUID, + warnings: list, + now: datetime.datetime, + ) -> tuple[type(App), list]: + """覆盖已有应用的配置，类型不一致时抛出异常""" + app = self.db.query(App).filter( + App.id == app_id, + App.workspace_id == workspace_id, + App.is_active.is_(True) + ).first() + if not app: + raise ResourceNotFoundException("应用", str(app_id)) + if app.type != app_type: + raise BusinessException( + f"YAML 类型 '{app_type}' 与应用类型 '{app.type}' 不一致，无法导入", + BizCode.BAD_REQUEST + ) + + self._write_config(app_id, app_type, dsl, workspace_id, tenant_id, warnings, now, create=False) + + self.db.commit() + self.db.refresh(app) + return app, warnings + + def _write_config( + self, + app_id: uuid.UUID, + app_type: str, + dsl: dict, + workspace_id: uuid.UUID, + tenant_id: uuid.UUID, + warnings: list, + now: datetime.datetime, + create: bool, + ) -> None: + """写入（新建或覆盖）应用配置""" if app_type == AppType.AGENT: cfg = dsl.get("agent_config") or {} - self.db.add(AgentConfig( - id=uuid.uuid4(), - app_id=new_app.id, + fields = dict( system_prompt=cfg.get("system_prompt"), model_parameters=cfg.get("model_parameters"), default_model_config_id=self._resolve_model(cfg.get("default_model_config_ref"), tenant_id, warnings), @@ -272,16 +324,21 @@ class AppDslService: tools=self._resolve_tools(cfg.get("tools", []), tenant_id, warnings), skills=self._resolve_skills(cfg.get("skills", {}), tenant_id, warnings), features=cfg.get("features", {}), - is_active=True, - created_at=now, updated_at=now, - )) + ) + if create: + self.db.add(AgentConfig(id=uuid.uuid4(), app_id=app_id, is_active=True, created_at=now, **fields)) + else: + existing = self.db.query(AgentConfig).filter(AgentConfig.app_id == app_id).first() + if existing: + for k, v in fields.items(): + setattr(existing, k, v) + else: + self.db.add(AgentConfig(id=uuid.uuid4(), app_id=app_id, is_active=True, created_at=now, **fields)) elif app_type == AppType.MULTI_AGENT: cfg = dsl.get("multi_agent_config") or {} - self.db.add(MultiAgentConfig( - id=uuid.uuid4(), - app_id=new_app.id, + fields = dict( orchestration_mode=cfg.get("orchestration_mode", "collaboration"), master_agent_name=cfg.get("master_agent_name"), model_parameters=cfg.get("model_parameters"), @@ -291,10 +348,17 @@ class AppDslService: routing_rules=self._resolve_routing_rules(cfg.get("routing_rules"), warnings), execution_config=cfg.get("execution_config", {}), aggregation_strategy=cfg.get("aggregation_strategy", "merge"), - is_active=True, - created_at=now, updated_at=now, - )) + ) + if create: + self.db.add(MultiAgentConfig(id=uuid.uuid4(), app_id=app_id, is_active=True, created_at=now, **fields)) + else: + existing = self.db.query(MultiAgentConfig).filter(MultiAgentConfig.app_id == app_id).first() + if existing: + for k, v in fields.items(): + setattr(existing, k, v) + else: + self.db.add(MultiAgentConfig(id=uuid.uuid4(), app_id=app_id, is_active=True, created_at=now, **fields)) elif app_type == AppType.WORKFLOW: adapter = MemoryBearAdapter(dsl) @@ -306,20 +370,39 @@ class AppDslService: for w in result.warnings: warnings.append(f"[节点警告] {w.node_name or w.node_id}: {w.detail}") wf = dsl.get("workflow") or {} - WorkflowService(self.db).create_workflow_config( - app_id=new_app.id, - nodes=[n.model_dump() for n in result.nodes], - edges=[e.model_dump() for e in result.edges], - variables=[v.model_dump() for v in result.variables], - execution_config=wf.get("execution_config", {}), - features=wf.get("features", {}), - triggers=wf.get("triggers", []), - validate=False, - ) - - self.db.commit() - self.db.refresh(new_app) - return new_app, warnings + wf_service = WorkflowService(self.db) + if create: + wf_service.create_workflow_config( + app_id=app_id, + nodes=[n.model_dump() for n in result.nodes], + edges=[e.model_dump() for e in result.edges], + variables=[v.model_dump() for v in result.variables], + execution_config=wf.get("execution_config", {}), + features=wf.get("features", {}), + triggers=wf.get("triggers", []), + validate=False, + ) + else: + existing = self.db.query(WorkflowConfig).filter(WorkflowConfig.app_id == app_id).first() + if existing: + existing.nodes = [n.model_dump() for n in result.nodes] + existing.edges = [e.model_dump() for e in result.edges] + existing.variables = [v.model_dump() for v in result.variables] + existing.execution_config = wf.get("execution_config", {}) + existing.features = wf.get("features", {}) + existing.triggers = wf.get("triggers", []) + existing.updated_at = now + else: + wf_service.create_workflow_config( + app_id=app_id, + nodes=[n.model_dump() for n in result.nodes], + edges=[e.model_dump() for e in result.edges], + variables=[v.model_dump() for v in result.variables], + execution_config=wf.get("execution_config", {}), + features=wf.get("features", {}), + triggers=wf.get("triggers", []), + validate=False, + ) def _unique_app_name(self, name: str, workspace_id: uuid.UUID, app_type: AppType) -> str: """生成唯一应用名称，同时检查本空间自有应用和共享到本空间的应用""" diff --git a/api/app/services/draft_run_service.py b/api/app/services/draft_run_service.py index 5c10e4f8..52c3d7f3 100644 --- a/api/app/services/draft_run_service.py +++ b/api/app/services/draft_run_service.py @@ -1299,10 +1299,23 @@ class AgentRunService: "history_files": {} } if files: + from app.models.file_metadata_model import FileMetadata for f in files: + name, size = f.name, f.size + if f.transfer_method.value == "local_file" and f.upload_file_id and (not name or not size): + meta = self.db.query(FileMetadata).filter( + FileMetadata.id == f.upload_file_id, + FileMetadata.status == "completed" + ).first() + if meta: + name = name or meta.file_name + size = size or meta.file_size human_meta["files"].append({ "type": f.type, - "url": f.url + "url": f.url, + "file_type": f.file_type, + "name": name, + "size": size }) # 保存 history_files，包含 provider 和 is_omni 信息 diff --git a/api/app/services/workflow_service.py b/api/app/services/workflow_service.py index b771c639..0d282d78 100644 --- a/api/app/services/workflow_service.py +++ b/api/app/services/workflow_service.py @@ -957,7 +957,10 @@ class WorkflowService: for file in message["content"]: human_meta["files"].append({ "type": file.get("type"), - "url": file.get("url") + "url": file.get("url"), + "file_type": file.get("origin_file_type"), + "name": file.get("name"), + "size": file.get("size") }) if message["role"] == "assistant": assistant_message = message["content"] From 9470dd2f1e882d218da46d09b4cb717a30591a62 Mon Sep 17 00:00:00 2001 From: lanceyq <1982376970@qq.com> Date: Mon, 13 Apr 2026 18:47:56 +0800 Subject: [PATCH 020/245] refactor(memory): extract shared MemorySummary count query and replace magic number MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit - Move duplicated Neo4j MemorySummary count query into MemoryBaseService.get_valid_memory_summary_count() - Introduce MIN_MEMORY_SUMMARY_COUNT constant to replace hardcoded 5 - Fix import ordering in implicit_emotions_storage_repository - Use UTC consistently for date calculations (remove CST offset, datetime.now → datetime.utcnow) --- .../implicit_emotions_storage_repository.py | 20 +++++----- api/app/services/implicit_memory_service.py | 17 ++++----- api/app/services/memory_base_service.py | 38 +++++++++++++++++++ api/app/services/user_memory_service.py | 20 +++------- 4 files changed, 60 insertions(+), 35 deletions(-) diff --git a/api/app/repositories/implicit_emotions_storage_repository.py b/api/app/repositories/implicit_emotions_storage_repository.py index b6c40b40..b665924d 100644 --- a/api/app/repositories/implicit_emotions_storage_repository.py +++ b/api/app/repositories/implicit_emotions_storage_repository.py @@ -5,16 +5,9 @@ Implicit Emotions Storage Repository 事务由调用方控制，仓储层只使用 flush/refresh """ import logging -from datetime import date, datetime, timezone +from datetime import datetime, timedelta, timezone from typing import Generator, Optional - -class TimeFilterUnavailableError(Exception): - """redis_client 不可用，无法执行时间轴筛选。 - - 调用方捕获此异常后可选择回退到 get_all_user_ids 进行全量处理。 - """ - import redis from sqlalchemy import exists, not_, select from sqlalchemy.orm import Session @@ -25,6 +18,13 @@ from app.models.implicit_emotions_storage_model import ImplicitEmotionsStorage logger = logging.getLogger(__name__) +class TimeFilterUnavailableError(Exception): + """redis_client 不可用，无法执行时间轴筛选。 + + 调用方捕获此异常后可选择回退到 get_all_user_ids 进行全量处理。 + """ + + class ImplicitEmotionsStorageRepository: """隐性记忆和情绪存储仓储类""" @@ -216,9 +216,7 @@ class ImplicitEmotionsStorageRepository: """ from sqlalchemy import String as SAString from sqlalchemy import cast - CST = timezone(timedelta(hours=8)) - now_cst = datetime.now(CST) - today_start = now_cst.replace(hour=0, minute=0, second=0, microsecond=0).astimezone(timezone.utc).replace(tzinfo=None) + today_start = datetime.utcnow().replace(hour=0, minute=0, second=0, microsecond=0) tomorrow_start = today_start + timedelta(days=1) offset = 0 while True: diff --git a/api/app/services/implicit_memory_service.py b/api/app/services/implicit_memory_service.py index 10504fe7..7a186f33 100644 --- a/api/app/services/implicit_memory_service.py +++ b/api/app/services/implicit_memory_service.py @@ -34,6 +34,7 @@ from app.schemas.implicit_memory_schema import ( UserMemorySummary, ) from app.schemas.memory_config_schema import MemoryConfig +from app.services.memory_base_service import MIN_MEMORY_SUMMARY_COUNT from sqlalchemy.orm import Session logger = logging.getLogger(__name__) @@ -381,7 +382,7 @@ class ImplicitMemoryService: def _build_empty_profile(self) -> dict: """构建 MemorySummary 不足时返回的固定空白画像数据""" - now_ms = int(datetime.now().timestamp() * 1000) + now_ms = int(datetime.utcnow().timestamp() * 1000) insufficient = "Insufficient data for analysis" def _empty_dimension(name: str) -> dict: @@ -442,17 +443,13 @@ class ImplicitMemoryService: try: # 前置检查：查询该用户有效的 MemorySummary 节点数量（排除孤立节点） - query = """ - MATCH (n:MemorySummary)-[:DERIVED_FROM_STATEMENT]->(:Statement) - WHERE n.end_user_id = $end_user_id - RETURN count(DISTINCT n) as count - """ - result = await self.neo4j_connector.execute_query(query, end_user_id=user_id) - memory_summary_count = result[0]["count"] if result and len(result) > 0 else 0 + from app.services.memory_base_service import MemoryBaseService + base_service = MemoryBaseService() + memory_summary_count = await base_service.get_valid_memory_summary_count(user_id) logger.info(f"用户 MemorySummary 节点数量: {memory_summary_count} (user={user_id})") - if memory_summary_count < 5: - logger.info(f"MemorySummary 数量不足 5（当前 {memory_summary_count}），返回空白画像: user={user_id}") + if memory_summary_count < MIN_MEMORY_SUMMARY_COUNT: + logger.info(f"MemorySummary 数量不足 {MIN_MEMORY_SUMMARY_COUNT}（当前 {memory_summary_count}），返回空白画像: user={user_id}") return self._build_empty_profile() # 并行调用4个分析方法 diff --git a/api/app/services/memory_base_service.py b/api/app/services/memory_base_service.py index bc647752..e615af8b 100644 --- a/api/app/services/memory_base_service.py +++ b/api/app/services/memory_base_service.py @@ -265,12 +265,50 @@ async def Translation_English(modid, text, fields=None): # 其他类型（数字、布尔值、None等）：原样返回 else: return text +# 隐性记忆画像生成所需的最低 MemorySummary 节点数量 +MIN_MEMORY_SUMMARY_COUNT = 5 + + class MemoryBaseService: """记忆服务基类，提供共享的辅助方法""" def __init__(self): self.neo4j_connector = Neo4jConnector() + async def get_valid_memory_summary_count( + self, + end_user_id: str + ) -> int: + """获取用户有效的 MemorySummary 节点数量（排除孤立节点）。 + + 只统计存在 DERIVED_FROM_STATEMENT 关系的 MemorySummary 节点。 + + Args: + end_user_id: 终端用户ID + + Returns: + 有效 MemorySummary 节点数量 + """ + try: + query = """ + MATCH (n:MemorySummary)-[:DERIVED_FROM_STATEMENT]->(:Statement) + WHERE n.end_user_id = $end_user_id + RETURN count(DISTINCT n) as count + """ + result = await self.neo4j_connector.execute_query( + query, end_user_id=end_user_id + ) + count = result[0]["count"] if result and len(result) > 0 else 0 + logger.debug( + f"有效 MemorySummary 节点数量: {count} (end_user_id={end_user_id})" + ) + return count + except Exception as e: + logger.error( + f"获取有效 MemorySummary 数量失败: {str(e)}", exc_info=True + ) + return 0 + @staticmethod def parse_timestamp(timestamp_value) -> Optional[int]: """ diff --git a/api/app/services/user_memory_service.py b/api/app/services/user_memory_service.py index dcbedba6..cc18447e 100644 --- a/api/app/services/user_memory_service.py +++ b/api/app/services/user_memory_service.py @@ -21,7 +21,7 @@ from app.repositories.end_user_repository import EndUserRepository from app.repositories.neo4j.cypher_queries import Graph_Node_query from app.repositories.neo4j.neo4j_connector import Neo4jConnector from app.schemas.memory_episodic_schema import EmotionSubject, EmotionType, type_mapping -from app.services.memory_base_service import MemoryBaseService +from app.services.memory_base_service import MemoryBaseService, MIN_MEMORY_SUMMARY_COUNT from app.services.memory_config_service import MemoryConfigService from app.services.memory_perceptual_service import MemoryPerceptualService from app.services.memory_short_service import ShortService @@ -1500,7 +1500,7 @@ async def analytics_memory_types( 2. 工作记忆 (WORKING_MEMORY) = 会话数量（通过 ConversationRepository.get_conversation_by_user_id 获取） 3. 短期记忆 (SHORT_TERM_MEMORY) = /short_term 接口返回的问答对数量 4. 显性记忆 (EXPLICIT_MEMORY) = 情景记忆 + 语义记忆（通过 MemoryBaseService.get_explicit_memory_count 获取） - 5. 隐性记忆 (IMPLICIT_MEMORY) = MemorySummary 节点数量（需 >= 5 才显示，否则为 0） + 5. 隐性记忆 (IMPLICIT_MEMORY) = MemorySummary 节点数量（需 >= MIN_MEMORY_SUMMARY_COUNT 才显示，否则为 0） 6. 情绪记忆 (EMOTIONAL_MEMORY) = 情绪标签统计总数（通过 MemoryBaseService.get_emotional_memory_count 获取） 7. 情景记忆 (EPISODIC_MEMORY) = memory_summary（通过 MemoryBaseService.get_episodic_memory_count 获取） 8. 遗忘记忆 (FORGET_MEMORY) = 激活值低于阈值的节点数（通过 MemoryBaseService.get_forget_memory_count 获取） @@ -1557,20 +1557,12 @@ async def analytics_memory_types( logger.warning(f"获取会话数量失败，工作记忆数量设为0: {str(e)}") work_count = 0 - # 获取隐性记忆数量（基于有关联关系的 MemorySummary 节点数量，需 >= 5 才计入） + # 获取隐性记忆数量（基于有关联关系的 MemorySummary 节点数量，需 >= MIN_MEMORY_SUMMARY_COUNT 才计入） implicit_count = 0 if end_user_id: try: - # 只统计有 DERIVED_FROM_STATEMENT 关系的 MemorySummary 节点，排除孤立节点 - query = """ - MATCH (n:MemorySummary)-[:DERIVED_FROM_STATEMENT]->(:Statement) - WHERE n.end_user_id = $end_user_id - RETURN count(DISTINCT n) as count - """ - result = await _neo4j_connector.execute_query(query, end_user_id=end_user_id) - memory_summary_count = result[0]["count"] if result and len(result) > 0 else 0 - # 仅当 MemorySummary 节点数量 >= 5 时才显示数量，否则为 0 - implicit_count = memory_summary_count if memory_summary_count >= 5 else 0 + memory_summary_count = await base_service.get_valid_memory_summary_count(end_user_id) + implicit_count = memory_summary_count if memory_summary_count >= MIN_MEMORY_SUMMARY_COUNT else 0 logger.debug(f"隐性记忆数量（有效MemorySummary节点数）: {implicit_count} (有效MemorySummary总数={memory_summary_count}, end_user_id={end_user_id})") except Exception as e: logger.warning(f"获取MemorySummary数量失败，隐性记忆数量设为0: {str(e)}") @@ -1639,7 +1631,7 @@ async def analytics_memory_types( "WORKING_MEMORY": work_count, # 工作记忆（基于会话数量） "SHORT_TERM_MEMORY": short_term_count, # 短期记忆（基于问答对数量） "EXPLICIT_MEMORY": explicit_count, # 显性记忆（情景记忆 + 语义记忆） - "IMPLICIT_MEMORY": implicit_count, # 隐性记忆（MemorySummary节点数，需>=5） + "IMPLICIT_MEMORY": implicit_count, # 隐性记忆（MemorySummary节点数，需>=MIN_MEMORY_SUMMARY_COUNT） "EMOTIONAL_MEMORY": emotion_count, # 情绪记忆（使用情绪标签统计） "EPISODIC_MEMORY": episodic_count, # 情景记忆 "FORGET_MEMORY": forget_count # 遗忘记忆（激活值低于阈值） From 7ca80b5d0120a19b01ceeb47b0885aa154af5fed Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Mon, 13 Apr 2026 18:52:43 +0800 Subject: [PATCH 021/245] perf(app): optimize FileMetadata queries by batching lookups Multiple services were performing individual database queries for FileMetadata when resolving missing file names/sizes. This change batches the queries using `in_()` to reduce database round trips and improve performance. --- api/app/services/app_chat_service.py | 30 ++++++++++++++++++++------- api/app/services/app_dsl_service.py | 2 +- api/app/services/draft_run_service.py | 15 ++++++++++---- 3 files changed, 34 insertions(+), 13 deletions(-) diff --git a/api/app/services/app_chat_service.py b/api/app/services/app_chat_service.py index 94362176..2d10ed44 100644 --- a/api/app/services/app_chat_service.py +++ b/api/app/services/app_chat_service.py @@ -219,13 +219,20 @@ class AppChatService: "reasoning_content": result.get("reasoning_content") } if files: + local_ids = [f.upload_file_id for f in files + if f.transfer_method.value == "local_file" and f.upload_file_id + and (not f.name or not f.size)] + meta_map = {} + if local_ids: + rows = self.db.query(FileMetadata).filter( + FileMetadata.id.in_(local_ids), + FileMetadata.status == "completed" + ).all() + meta_map = {str(r.id): r for r in rows} for f in files: name, size = f.name, f.size if f.transfer_method.value == "local_file" and f.upload_file_id and (not name or not size): - meta = self.db.query(FileMetadata).filter( - FileMetadata.id == f.upload_file_id, - FileMetadata.status == "completed" - ).first() + meta = meta_map.get(str(f.upload_file_id)) if meta: name = name or meta.file_name size = size or meta.file_size @@ -521,13 +528,20 @@ class AppChatService: } if files: + local_ids = [f.upload_file_id for f in files + if f.transfer_method.value == "local_file" and f.upload_file_id + and (not f.name or not f.size)] + meta_map = {} + if local_ids: + rows = self.db.query(FileMetadata).filter( + FileMetadata.id.in_(local_ids), + FileMetadata.status == "completed" + ).all() + meta_map = {str(r.id): r for r in rows} for f in files: name, size = f.name, f.size if f.transfer_method.value == "local_file" and f.upload_file_id and (not name or not size): - meta = self.db.query(FileMetadata).filter( - FileMetadata.id == f.upload_file_id, - FileMetadata.status == "completed" - ).first() + meta = meta_map.get(str(f.upload_file_id)) if meta: name = name or meta.file_name size = size or meta.file_size diff --git a/api/app/services/app_dsl_service.py b/api/app/services/app_dsl_service.py index 612b93fe..16aa8a3a 100644 --- a/api/app/services/app_dsl_service.py +++ b/api/app/services/app_dsl_service.py @@ -279,7 +279,7 @@ class AppDslService: tenant_id: uuid.UUID, warnings: list, now: datetime.datetime, - ) -> tuple[type(App), list]: + ) -> tuple[App, list[str]]: """覆盖已有应用的配置，类型不一致时抛出异常""" app = self.db.query(App).filter( App.id == app_id, diff --git a/api/app/services/draft_run_service.py b/api/app/services/draft_run_service.py index 52c3d7f3..b47bd4cd 100644 --- a/api/app/services/draft_run_service.py +++ b/api/app/services/draft_run_service.py @@ -1300,13 +1300,20 @@ class AgentRunService: } if files: from app.models.file_metadata_model import FileMetadata + local_ids = [f.upload_file_id for f in files + if f.transfer_method.value == "local_file" and f.upload_file_id + and (not f.name or not f.size)] + meta_map = {} + if local_ids: + rows = self.db.query(FileMetadata).filter( + FileMetadata.id.in_(local_ids), + FileMetadata.status == "completed" + ).all() + meta_map = {str(r.id): r for r in rows} for f in files: name, size = f.name, f.size if f.transfer_method.value == "local_file" and f.upload_file_id and (not name or not size): - meta = self.db.query(FileMetadata).filter( - FileMetadata.id == f.upload_file_id, - FileMetadata.status == "completed" - ).first() + meta = meta_map.get(str(f.upload_file_id)) if meta: name = name or meta.file_name size = size or meta.file_size From 52e726eabcc6b879e33b0037f7e871a3f2dc2ecb Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Mon, 13 Apr 2026 18:53:49 +0800 Subject: [PATCH 022/245] ci: add release notification workflow for merged PRs - Add GitHub Actions workflow to notify on merged release branch PRs - Implement HEAD sync check to ensure branch is up-to-date before notification - Fetch commit messages from merged PR for AI summarization - Integrate Alibaba Qwen AI to generate Chinese release summaries for QA team - Send formatted Markdown notifications to WeChat webhook with PR details and AI summary - Workflow triggers only on final PR merge to release branches to avoid duplicate notifications --- .github/workflows/release-notify.yml | 107 +++++++++++++++++++++++++++ 1 file changed, 107 insertions(+) create mode 100644 .github/workflows/release-notify.yml diff --git a/.github/workflows/release-notify.yml b/.github/workflows/release-notify.yml new file mode 100644 index 00000000..6b86db96 --- /dev/null +++ b/.github/workflows/release-notify.yml @@ -0,0 +1,107 @@ +name: Release Notify (Ali AI Final) + +on: + pull_request: + types: [closed] + +jobs: + notify: + if: > + github.event.pull_request.merged == true && + startsWith(github.event.pull_request.base.ref, 'release') + runs-on: ubuntu-latest + + steps: + # 防止 GitHub HEAD 未同步 + - name: Wait for ref sync + run: sleep 3 + + # 1️⃣ 获取分支 HEAD + - name: Get HEAD + id: head + env: + GH_TOKEN: ${{ secrets.GITHUB_TOKEN }} + REPO: ${{ github.repository }} + BASE_REF: ${{ github.event.pull_request.base.ref }} + run: | + HEAD_SHA=$(curl -s \ + -H "Authorization: Bearer $GH_TOKEN" \ + "https://api.github.com/repos/$REPO/git/ref/heads/$BASE_REF" \ + | jq -r '.object.sha') + echo "head_sha=$HEAD_SHA" >> $GITHUB_OUTPUT + + # 2️⃣ 判断是否最终PR + - name: Check Latest + id: check + env: + MERGE_SHA: ${{ github.event.pull_request.merge_commit_sha }} + HEAD_SHA: ${{ steps.head.outputs.head_sha }} + run: | + if [ "$MERGE_SHA" = "$HEAD_SHA" ]; then + echo "ok=true" >> $GITHUB_OUTPUT + else + echo "ok=false" >> $GITHUB_OUTPUT + fi + + # 3️⃣ 获取 commits + - name: Get Commits + if: steps.check.outputs.ok == 'true' + id: commits + env: + GH_TOKEN: ${{ secrets.GITHUB_TOKEN }} + COMMITS_URL: ${{ github.event.pull_request.commits_url }} + run: | + curl -s \ + -H "Authorization: Bearer $GH_TOKEN" \ + "$COMMITS_URL" \ + | jq -r '.[].commit.message' | head -n 20 > commits.txt + + # 4️⃣ 阿里 AI 总结（通义千问） + - name: AI Summary (Qwen) + if: steps.check.outputs.ok == 'true' + id: ai + env: + DASHSCOPE_API_KEY: ${{ secrets.DASHSCOPE_API_KEY }} + run: | + COMMIT_MESSAGES=$(cat commits.txt) + + jq -n --arg msgs "请用中文总结以下代码提交，输出3-5条，面向测试人员： + $COMMIT_MESSAGES" \ + '{"model": "qwen-plus", "input": {"prompt": $msgs}}' > ai_payload.json + + SUMMARY=$(curl -s https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d @ai_payload.json | jq -r '.output.text') + + echo "summary<> $GITHUB_OUTPUT + echo "$SUMMARY" >> $GITHUB_OUTPUT + echo "EOF" >> $GITHUB_OUTPUT + + # 5️⃣ 企业微信通知（Markdown） + - name: Notify WeChat + if: steps.check.outputs.ok == 'true' + env: + WECHAT_WEBHOOK: ${{ secrets.WECHAT_WEBHOOK }} + BRANCH: ${{ github.event.pull_request.base.ref }} + AUTHOR: ${{ github.event.pull_request.user.login }} + PR_TITLE: ${{ github.event.pull_request.title }} + PR_URL: ${{ github.event.pull_request.html_url }} + AI_SUMMARY: ${{ steps.ai.outputs.summary }} + run: | + jq -n \ + --arg branch "$BRANCH" \ + --arg author "$AUTHOR" \ + --arg title "$PR_TITLE" \ + --arg url "$PR_URL" \ + --arg summary "$AI_SUMMARY" \ + '{ + msgtype: "markdown", + markdown: { + content: "## 🚀 Release 发布通知\n> 📦 **分支**: \($branch)\n> 👤 **提交人**: \($author)\n> 📝 **标题**: \($title)\n\n### 🧠 AI变更摘要\n\($summary)\n\n---\n🔗 [查看PR详情](\($url))" + } + }' > wechat_payload.json + + curl -s "$WECHAT_WEBHOOK" \ + -H 'Content-Type: application/json' \ + -d @wechat_payload.json From 2450fe3afea44aca3d78d8334edb0b50a4fdf4b6 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Mon, 13 Apr 2026 19:00:36 +0800 Subject: [PATCH 023/245] refactor(workflow): move _merge_conv_vars call inside iteration loop for consistent state updates --- api/app/core/workflow/nodes/cycle_graph/iteration.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/api/app/core/workflow/nodes/cycle_graph/iteration.py b/api/app/core/workflow/nodes/cycle_graph/iteration.py index 7652e21f..1633b9c7 100644 --- a/api/app/core/workflow/nodes/cycle_graph/iteration.py +++ b/api/app/core/workflow/nodes/cycle_graph/iteration.py @@ -249,9 +249,9 @@ class IterationRuntime: else: self.result.append(output) child_state.append(result) + self._merge_conv_vars(child_pool) if stopped: self.looping = False - self._merge_conv_vars(batch_sorted[-1][3]) else: # Execute iterations sequentially while idx < len(array_obj) and self.looping: From a3f0415cd3e028f1b156d71cc167a44f46937e2c Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Mon, 13 Apr 2026 19:02:28 +0800 Subject: [PATCH 024/245] ci(workflow): add release notification workflow for WeChat - Add new GitHub Actions workflow to notify WeChat on release branch merges - Implement HEAD sync check to prevent race conditions with GitHub API - Add commit validation to ensure PR is the latest merge to release branch - Fetch PR commits and generate AI summary using Alibaba Qwen API - Send formatted Markdown notification to WeChat webhook with release details - Include branch, author, PR title, and AI-generated change summary in notification --- .github/workflows/release-notify-wechat.yml | 107 ++++++++++++++++++++ 1 file changed, 107 insertions(+) create mode 100644 .github/workflows/release-notify-wechat.yml diff --git a/.github/workflows/release-notify-wechat.yml b/.github/workflows/release-notify-wechat.yml new file mode 100644 index 00000000..6b86db96 --- /dev/null +++ b/.github/workflows/release-notify-wechat.yml @@ -0,0 +1,107 @@ +name: Release Notify (Ali AI Final) + +on: + pull_request: + types: [closed] + +jobs: + notify: + if: > + github.event.pull_request.merged == true && + startsWith(github.event.pull_request.base.ref, 'release') + runs-on: ubuntu-latest + + steps: + # 防止 GitHub HEAD 未同步 + - name: Wait for ref sync + run: sleep 3 + + # 1️⃣ 获取分支 HEAD + - name: Get HEAD + id: head + env: + GH_TOKEN: ${{ secrets.GITHUB_TOKEN }} + REPO: ${{ github.repository }} + BASE_REF: ${{ github.event.pull_request.base.ref }} + run: | + HEAD_SHA=$(curl -s \ + -H "Authorization: Bearer $GH_TOKEN" \ + "https://api.github.com/repos/$REPO/git/ref/heads/$BASE_REF" \ + | jq -r '.object.sha') + echo "head_sha=$HEAD_SHA" >> $GITHUB_OUTPUT + + # 2️⃣ 判断是否最终PR + - name: Check Latest + id: check + env: + MERGE_SHA: ${{ github.event.pull_request.merge_commit_sha }} + HEAD_SHA: ${{ steps.head.outputs.head_sha }} + run: | + if [ "$MERGE_SHA" = "$HEAD_SHA" ]; then + echo "ok=true" >> $GITHUB_OUTPUT + else + echo "ok=false" >> $GITHUB_OUTPUT + fi + + # 3️⃣ 获取 commits + - name: Get Commits + if: steps.check.outputs.ok == 'true' + id: commits + env: + GH_TOKEN: ${{ secrets.GITHUB_TOKEN }} + COMMITS_URL: ${{ github.event.pull_request.commits_url }} + run: | + curl -s \ + -H "Authorization: Bearer $GH_TOKEN" \ + "$COMMITS_URL" \ + | jq -r '.[].commit.message' | head -n 20 > commits.txt + + # 4️⃣ 阿里 AI 总结（通义千问） + - name: AI Summary (Qwen) + if: steps.check.outputs.ok == 'true' + id: ai + env: + DASHSCOPE_API_KEY: ${{ secrets.DASHSCOPE_API_KEY }} + run: | + COMMIT_MESSAGES=$(cat commits.txt) + + jq -n --arg msgs "请用中文总结以下代码提交，输出3-5条，面向测试人员： + $COMMIT_MESSAGES" \ + '{"model": "qwen-plus", "input": {"prompt": $msgs}}' > ai_payload.json + + SUMMARY=$(curl -s https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d @ai_payload.json | jq -r '.output.text') + + echo "summary<> $GITHUB_OUTPUT + echo "$SUMMARY" >> $GITHUB_OUTPUT + echo "EOF" >> $GITHUB_OUTPUT + + # 5️⃣ 企业微信通知（Markdown） + - name: Notify WeChat + if: steps.check.outputs.ok == 'true' + env: + WECHAT_WEBHOOK: ${{ secrets.WECHAT_WEBHOOK }} + BRANCH: ${{ github.event.pull_request.base.ref }} + AUTHOR: ${{ github.event.pull_request.user.login }} + PR_TITLE: ${{ github.event.pull_request.title }} + PR_URL: ${{ github.event.pull_request.html_url }} + AI_SUMMARY: ${{ steps.ai.outputs.summary }} + run: | + jq -n \ + --arg branch "$BRANCH" \ + --arg author "$AUTHOR" \ + --arg title "$PR_TITLE" \ + --arg url "$PR_URL" \ + --arg summary "$AI_SUMMARY" \ + '{ + msgtype: "markdown", + markdown: { + content: "## 🚀 Release 发布通知\n> 📦 **分支**: \($branch)\n> 👤 **提交人**: \($author)\n> 📝 **标题**: \($title)\n\n### 🧠 AI变更摘要\n\($summary)\n\n---\n🔗 [查看PR详情](\($url))" + } + }' > wechat_payload.json + + curl -s "$WECHAT_WEBHOOK" \ + -H 'Content-Type: application/json' \ + -d @wechat_payload.json From 70d4e79de1546b1cb36e856575f4270fa34eb356 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 13 Apr 2026 19:05:32 +0800 Subject: [PATCH 025/245] fix(web): breadcrumb ui --- web/src/components/Header/index.module.css | 8 +++++ web/src/components/Header/index.tsx | 37 +++++++++++----------- 2 files changed, 27 insertions(+), 18 deletions(-) diff --git a/web/src/components/Header/index.module.css b/web/src/components/Header/index.module.css index d39c91ec..525a2432 100644 --- a/web/src/components/Header/index.module.css +++ b/web/src/components/Header/index.module.css @@ -12,6 +12,14 @@ font-weight: 500; font-style: normal; } +.breadcrumbTitle { + display: inline-block; + max-width: 200px; + overflow: hidden; + text-overflow: ellipsis; + white-space: nowrap; + vertical-align: bottom; +} .header :global(.ant-breadcrumb) { line-height: 31px; } diff --git a/web/src/components/Header/index.tsx b/web/src/components/Header/index.tsx index 49988223..f2eff014 100644 --- a/web/src/components/Header/index.tsx +++ b/web/src/components/Header/index.tsx @@ -14,7 +14,7 @@ */ import { type FC, useRef, useState } from 'react'; -import { Layout, Dropdown, Breadcrumb, Flex } from 'antd'; +import { Layout, Dropdown, Breadcrumb, Flex, Tooltip } from 'antd'; import type { MenuProps, BreadcrumbProps } from 'antd'; import { useTranslation } from 'react-i18next'; import { useLocation } from 'react-router-dom'; @@ -136,27 +136,28 @@ const AppHeader: FC<{source?: 'space' | 'manage';}> = ({source = 'manage'}) => { */ const formatBreadcrumbNames = () => { return breadcrumbs.filter(item => item.type !== 'group').map((menu, index) => { + const label = menu.i18nKey ? t(menu.i18nKey) : menu.label; + const isLast = index === breadcrumbs.length - 1; const item: any = { - title: menu.i18nKey ? t(menu.i18nKey) : menu.label, + title: ( + + {label} + + ), }; - // If it's the last item, don't set path - if (index === breadcrumbs.length - 1) { - return item; + if (!isLast) { + if ((menu as any).onClick) { + item.onClick = (e: React.MouseEvent) => { + e.preventDefault(); + (menu as any).onClick(e); + }; + item.href = '#'; + } else if (menu.path && menu.path !== '#') { + item.path = menu.path; + } } - - // If has custom onClick, use onClick and set href to '#' to show pointer cursor - if ((menu as any).onClick) { - item.onClick = (e: React.MouseEvent) => { - e.preventDefault(); - (menu as any).onClick(e); - }; - item.href = '#'; - } else if (menu.path && menu.path !== '#') { - // Only set path when path is not '#' - item.path = menu.path; - } - + return item; }); } From 60124e323222b95fd48cdf0255d2879eb57b3235 Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Mon, 13 Apr 2026 19:06:18 +0800 Subject: [PATCH 026/245] ci(workflow): simplify WeChat notification payload generation - Rename workflow from "Release Notify (Ali AI Final)" to "Release Notify Workflow" for clarity - Replace jq multi-line argument construction with printf for better readability - Simplify payload generation by building content string separately before passing to jq - Reduce complexity of nested jq arguments while maintaining identical output format --- .github/workflows/release-notify-wechat.yml | 19 ++++++------------- 1 file changed, 6 insertions(+), 13 deletions(-) diff --git a/.github/workflows/release-notify-wechat.yml b/.github/workflows/release-notify-wechat.yml index 6b86db96..7b3378b0 100644 --- a/.github/workflows/release-notify-wechat.yml +++ b/.github/workflows/release-notify-wechat.yml @@ -1,4 +1,4 @@ -name: Release Notify (Ali AI Final) +name: Release Notify Workflow on: pull_request: @@ -89,18 +89,11 @@ jobs: PR_URL: ${{ github.event.pull_request.html_url }} AI_SUMMARY: ${{ steps.ai.outputs.summary }} run: | - jq -n \ - --arg branch "$BRANCH" \ - --arg author "$AUTHOR" \ - --arg title "$PR_TITLE" \ - --arg url "$PR_URL" \ - --arg summary "$AI_SUMMARY" \ - '{ - msgtype: "markdown", - markdown: { - content: "## 🚀 Release 发布通知\n> 📦 **分支**: \($branch)\n> 👤 **提交人**: \($author)\n> 📝 **标题**: \($title)\n\n### 🧠 AI变更摘要\n\($summary)\n\n---\n🔗 [查看PR详情](\($url))" - } - }' > wechat_payload.json + CONTENT=$(printf '## 🚀 Release 发布通知\n> 📦 **分支**: %s\n> 👤 **提交人**: %s\n> 📝 **标题**: %s\n\n### 🧠 AI变更摘要\n%s\n\n---\n🔗 [查看PR详情](%s)' \ + "$BRANCH" "$AUTHOR" "$PR_TITLE" "$AI_SUMMARY" "$PR_URL") + + jq -n --arg content "$CONTENT" \ + '{"msgtype": "markdown", "markdown": {"content": $content}}' > wechat_payload.json curl -s "$WECHAT_WEBHOOK" \ -H 'Content-Type: application/json' \ From 1ff3748935b31ebc9a1ae3978ddbeaf3d1de1301 Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Mon, 13 Apr 2026 19:11:15 +0800 Subject: [PATCH 027/245] ci: remove release notification workflow - Delete release-notify.yml GitHub Actions workflow - Remove AI-powered release summary generation via Qwen API - Remove WeChat enterprise notification integration - Simplify CI/CD pipeline by consolidating notification logic --- .github/workflows/release-notify.yml | 107 --------------------------- 1 file changed, 107 deletions(-) delete mode 100644 .github/workflows/release-notify.yml diff --git a/.github/workflows/release-notify.yml b/.github/workflows/release-notify.yml deleted file mode 100644 index 6b86db96..00000000 --- a/.github/workflows/release-notify.yml +++ /dev/null @@ -1,107 +0,0 @@ -name: Release Notify (Ali AI Final) - -on: - pull_request: - types: [closed] - -jobs: - notify: - if: > - github.event.pull_request.merged == true && - startsWith(github.event.pull_request.base.ref, 'release') - runs-on: ubuntu-latest - - steps: - # 防止 GitHub HEAD 未同步 - - name: Wait for ref sync - run: sleep 3 - - # 1️⃣ 获取分支 HEAD - - name: Get HEAD - id: head - env: - GH_TOKEN: ${{ secrets.GITHUB_TOKEN }} - REPO: ${{ github.repository }} - BASE_REF: ${{ github.event.pull_request.base.ref }} - run: | - HEAD_SHA=$(curl -s \ - -H "Authorization: Bearer $GH_TOKEN" \ - "https://api.github.com/repos/$REPO/git/ref/heads/$BASE_REF" \ - | jq -r '.object.sha') - echo "head_sha=$HEAD_SHA" >> $GITHUB_OUTPUT - - # 2️⃣ 判断是否最终PR - - name: Check Latest - id: check - env: - MERGE_SHA: ${{ github.event.pull_request.merge_commit_sha }} - HEAD_SHA: ${{ steps.head.outputs.head_sha }} - run: | - if [ "$MERGE_SHA" = "$HEAD_SHA" ]; then - echo "ok=true" >> $GITHUB_OUTPUT - else - echo "ok=false" >> $GITHUB_OUTPUT - fi - - # 3️⃣ 获取 commits - - name: Get Commits - if: steps.check.outputs.ok == 'true' - id: commits - env: - GH_TOKEN: ${{ secrets.GITHUB_TOKEN }} - COMMITS_URL: ${{ github.event.pull_request.commits_url }} - run: | - curl -s \ - -H "Authorization: Bearer $GH_TOKEN" \ - "$COMMITS_URL" \ - | jq -r '.[].commit.message' | head -n 20 > commits.txt - - # 4️⃣ 阿里 AI 总结（通义千问） - - name: AI Summary (Qwen) - if: steps.check.outputs.ok == 'true' - id: ai - env: - DASHSCOPE_API_KEY: ${{ secrets.DASHSCOPE_API_KEY }} - run: | - COMMIT_MESSAGES=$(cat commits.txt) - - jq -n --arg msgs "请用中文总结以下代码提交，输出3-5条，面向测试人员： - $COMMIT_MESSAGES" \ - '{"model": "qwen-plus", "input": {"prompt": $msgs}}' > ai_payload.json - - SUMMARY=$(curl -s https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation \ - -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ - -H "Content-Type: application/json" \ - -d @ai_payload.json | jq -r '.output.text') - - echo "summary<> $GITHUB_OUTPUT - echo "$SUMMARY" >> $GITHUB_OUTPUT - echo "EOF" >> $GITHUB_OUTPUT - - # 5️⃣ 企业微信通知（Markdown） - - name: Notify WeChat - if: steps.check.outputs.ok == 'true' - env: - WECHAT_WEBHOOK: ${{ secrets.WECHAT_WEBHOOK }} - BRANCH: ${{ github.event.pull_request.base.ref }} - AUTHOR: ${{ github.event.pull_request.user.login }} - PR_TITLE: ${{ github.event.pull_request.title }} - PR_URL: ${{ github.event.pull_request.html_url }} - AI_SUMMARY: ${{ steps.ai.outputs.summary }} - run: | - jq -n \ - --arg branch "$BRANCH" \ - --arg author "$AUTHOR" \ - --arg title "$PR_TITLE" \ - --arg url "$PR_URL" \ - --arg summary "$AI_SUMMARY" \ - '{ - msgtype: "markdown", - markdown: { - content: "## 🚀 Release 发布通知\n> 📦 **分支**: \($branch)\n> 👤 **提交人**: \($author)\n> 📝 **标题**: \($title)\n\n### 🧠 AI变更摘要\n\($summary)\n\n---\n🔗 [查看PR详情](\($url))" - } - }' > wechat_payload.json - - curl -s "$WECHAT_WEBHOOK" \ - -H 'Content-Type: application/json' \ - -d @wechat_payload.json From 77ed9faea102c365241221b24208f916362dba44 Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Mon, 13 Apr 2026 19:13:23 +0800 Subject: [PATCH 028/245] chore(.gitignore): add redbear-mem-benchmark to ignored paths - Add redbear-mem-benchmark directory to .gitignore - Prevents benchmark artifacts from being tracked in version control - Aligns with existing pattern of ignoring redbear-mem-metrics directory --- .gitignore | 1 + 1 file changed, 1 insertion(+) diff --git a/.gitignore b/.gitignore index 0ec6822c..a1896da7 100644 --- a/.gitignore +++ b/.gitignore @@ -27,6 +27,7 @@ time.log celerybeat-schedule.db search_results.json redbear-mem-metrics/ +redbear-mem-benchmark/ pitch-deck/ api/migrations/versions From 7a4a02b2bb722c2e3b522d8bc7e0984abdf10c55 Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Mon, 13 Apr 2026 19:15:54 +0800 Subject: [PATCH 029/245] ci: add WeChat release notification workflow - Add GitHub Actions workflow to notify WeChat on release branch merges - Implement multi-step pipeline: sync ref, verify latest PR, fetch commits - Integrate Aliyun Qwen AI for automated Chinese commit message summarization - Send formatted Markdown notifications to WeChat webhook with release details - Include branch, author, PR title, AI summary, and PR link in notifications --- .github/workflows/release-notify-wechat.yml | 100 ++++++++++++++++++++ 1 file changed, 100 insertions(+) create mode 100644 .github/workflows/release-notify-wechat.yml diff --git a/.github/workflows/release-notify-wechat.yml b/.github/workflows/release-notify-wechat.yml new file mode 100644 index 00000000..7b3378b0 --- /dev/null +++ b/.github/workflows/release-notify-wechat.yml @@ -0,0 +1,100 @@ +name: Release Notify Workflow + +on: + pull_request: + types: [closed] + +jobs: + notify: + if: > + github.event.pull_request.merged == true && + startsWith(github.event.pull_request.base.ref, 'release') + runs-on: ubuntu-latest + + steps: + # 防止 GitHub HEAD 未同步 + - name: Wait for ref sync + run: sleep 3 + + # 1️⃣ 获取分支 HEAD + - name: Get HEAD + id: head + env: + GH_TOKEN: ${{ secrets.GITHUB_TOKEN }} + REPO: ${{ github.repository }} + BASE_REF: ${{ github.event.pull_request.base.ref }} + run: | + HEAD_SHA=$(curl -s \ + -H "Authorization: Bearer $GH_TOKEN" \ + "https://api.github.com/repos/$REPO/git/ref/heads/$BASE_REF" \ + | jq -r '.object.sha') + echo "head_sha=$HEAD_SHA" >> $GITHUB_OUTPUT + + # 2️⃣ 判断是否最终PR + - name: Check Latest + id: check + env: + MERGE_SHA: ${{ github.event.pull_request.merge_commit_sha }} + HEAD_SHA: ${{ steps.head.outputs.head_sha }} + run: | + if [ "$MERGE_SHA" = "$HEAD_SHA" ]; then + echo "ok=true" >> $GITHUB_OUTPUT + else + echo "ok=false" >> $GITHUB_OUTPUT + fi + + # 3️⃣ 获取 commits + - name: Get Commits + if: steps.check.outputs.ok == 'true' + id: commits + env: + GH_TOKEN: ${{ secrets.GITHUB_TOKEN }} + COMMITS_URL: ${{ github.event.pull_request.commits_url }} + run: | + curl -s \ + -H "Authorization: Bearer $GH_TOKEN" \ + "$COMMITS_URL" \ + | jq -r '.[].commit.message' | head -n 20 > commits.txt + + # 4️⃣ 阿里 AI 总结（通义千问） + - name: AI Summary (Qwen) + if: steps.check.outputs.ok == 'true' + id: ai + env: + DASHSCOPE_API_KEY: ${{ secrets.DASHSCOPE_API_KEY }} + run: | + COMMIT_MESSAGES=$(cat commits.txt) + + jq -n --arg msgs "请用中文总结以下代码提交，输出3-5条，面向测试人员： + $COMMIT_MESSAGES" \ + '{"model": "qwen-plus", "input": {"prompt": $msgs}}' > ai_payload.json + + SUMMARY=$(curl -s https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d @ai_payload.json | jq -r '.output.text') + + echo "summary<> $GITHUB_OUTPUT + echo "$SUMMARY" >> $GITHUB_OUTPUT + echo "EOF" >> $GITHUB_OUTPUT + + # 5️⃣ 企业微信通知（Markdown） + - name: Notify WeChat + if: steps.check.outputs.ok == 'true' + env: + WECHAT_WEBHOOK: ${{ secrets.WECHAT_WEBHOOK }} + BRANCH: ${{ github.event.pull_request.base.ref }} + AUTHOR: ${{ github.event.pull_request.user.login }} + PR_TITLE: ${{ github.event.pull_request.title }} + PR_URL: ${{ github.event.pull_request.html_url }} + AI_SUMMARY: ${{ steps.ai.outputs.summary }} + run: | + CONTENT=$(printf '## 🚀 Release 发布通知\n> 📦 **分支**: %s\n> 👤 **提交人**: %s\n> 📝 **标题**: %s\n\n### 🧠 AI变更摘要\n%s\n\n---\n🔗 [查看PR详情](%s)' \ + "$BRANCH" "$AUTHOR" "$PR_TITLE" "$AI_SUMMARY" "$PR_URL") + + jq -n --arg content "$CONTENT" \ + '{"msgtype": "markdown", "markdown": {"content": $content}}' > wechat_payload.json + + curl -s "$WECHAT_WEBHOOK" \ + -H 'Content-Type: application/json' \ + -d @wechat_payload.json From 8495aa5dde784b423f5fbe9522be34a9df910e38 Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Mon, 13 Apr 2026 19:18:11 +0800 Subject: [PATCH 030/245] ci(wechat-notify): replace shell string formatting with Python - Replace printf and jq command chain with Python script for payload generation - Improve readability by using Python string concatenation instead of nested printf format specifiers - Ensure proper JSON encoding with ensure_ascii=False to preserve Chinese characters - Simplify environment variable interpolation using os.environ dictionary access --- .github/workflows/release-notify-wechat.yml | 21 ++++++++++++++++----- 1 file changed, 16 insertions(+), 5 deletions(-) diff --git a/.github/workflows/release-notify-wechat.yml b/.github/workflows/release-notify-wechat.yml index 7b3378b0..ae5f3f6e 100644 --- a/.github/workflows/release-notify-wechat.yml +++ b/.github/workflows/release-notify-wechat.yml @@ -89,11 +89,22 @@ jobs: PR_URL: ${{ github.event.pull_request.html_url }} AI_SUMMARY: ${{ steps.ai.outputs.summary }} run: | - CONTENT=$(printf '## 🚀 Release 发布通知\n> 📦 **分支**: %s\n> 👤 **提交人**: %s\n> 📝 **标题**: %s\n\n### 🧠 AI变更摘要\n%s\n\n---\n🔗 [查看PR详情](%s)' \ - "$BRANCH" "$AUTHOR" "$PR_TITLE" "$AI_SUMMARY" "$PR_URL") - - jq -n --arg content "$CONTENT" \ - '{"msgtype": "markdown", "markdown": {"content": $content}}' > wechat_payload.json + python3 -c " + import json, os + content = ( + '## 🚀 Release 发布通知\n' + '> 📦 **分支**: ' + os.environ['BRANCH'] + '\n' + '> 👤 **提交人**: ' + os.environ['AUTHOR'] + '\n' + '> 📝 **标题**: ' + os.environ['PR_TITLE'] + '\n\n' + '### 🧠 AI变更摘要\n' + + os.environ['AI_SUMMARY'] + '\n\n' + '---\n' + '🔗 [查看PR详情](' + os.environ['PR_URL'] + ')' + ) + payload = {'msgtype': 'markdown', 'markdown': {'content': content}} + with open('wechat_payload.json', 'w') as f: + json.dump(payload, f, ensure_ascii=False) + " curl -s "$WECHAT_WEBHOOK" \ -H 'Content-Type: application/json' \ From b20971dc9565fb91925dbc4f461f607fb112f3ca Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Mon, 13 Apr 2026 19:20:53 +0800 Subject: [PATCH 031/245] ci(wechat-notify): extract payload building logic to Python script - Create new `.github/scripts/build_wechat_payload.py` to handle WeChat payload generation - Replace inline Python string concatenation with dedicated script for better maintainability - Add checkout step to access the script during workflow execution - Simplify workflow by delegating payload construction to external script - Improve code readability and reusability for future notification enhancements --- .github/scripts/build_wechat_payload.py | 24 +++++++++++++++++++++ .github/workflows/release-notify-wechat.yml | 21 +++++------------- 2 files changed, 29 insertions(+), 16 deletions(-) create mode 100644 .github/scripts/build_wechat_payload.py diff --git a/.github/scripts/build_wechat_payload.py b/.github/scripts/build_wechat_payload.py new file mode 100644 index 00000000..5e292ee9 --- /dev/null +++ b/.github/scripts/build_wechat_payload.py @@ -0,0 +1,24 @@ +import json +import os + +branch = os.environ.get("BRANCH", "") +author = os.environ.get("AUTHOR", "") +pr_title = os.environ.get("PR_TITLE", "") +pr_url = os.environ.get("PR_URL", "") +ai_summary = os.environ.get("AI_SUMMARY", "") + +content = ( + "## 🚀 Release 发布通知\n" + f"> 📦 **分支**: {branch}\n" + f"> 👤 **提交人**: {author}\n" + f"> 📝 **标题**: {pr_title}\n\n" + "### 🧠 AI变更摘要\n" + f"{ai_summary}\n\n" + "---\n" + f"🔗 [查看PR详情]({pr_url})" +) + +payload = {"msgtype": "markdown", "markdown": {"content": content}} + +with open("wechat_payload.json", "w", encoding="utf-8") as f: + json.dump(payload, f, ensure_ascii=False) diff --git a/.github/workflows/release-notify-wechat.yml b/.github/workflows/release-notify-wechat.yml index ae5f3f6e..b6c8a04d 100644 --- a/.github/workflows/release-notify-wechat.yml +++ b/.github/workflows/release-notify-wechat.yml @@ -79,6 +79,10 @@ jobs: echo "EOF" >> $GITHUB_OUTPUT # 5️⃣ 企业微信通知（Markdown） + - name: Checkout for script + if: steps.check.outputs.ok == 'true' + uses: actions/checkout@v4 + - name: Notify WeChat if: steps.check.outputs.ok == 'true' env: @@ -89,22 +93,7 @@ jobs: PR_URL: ${{ github.event.pull_request.html_url }} AI_SUMMARY: ${{ steps.ai.outputs.summary }} run: | - python3 -c " - import json, os - content = ( - '## 🚀 Release 发布通知\n' - '> 📦 **分支**: ' + os.environ['BRANCH'] + '\n' - '> 👤 **提交人**: ' + os.environ['AUTHOR'] + '\n' - '> 📝 **标题**: ' + os.environ['PR_TITLE'] + '\n\n' - '### 🧠 AI变更摘要\n' - + os.environ['AI_SUMMARY'] + '\n\n' - '---\n' - '🔗 [查看PR详情](' + os.environ['PR_URL'] + ')' - ) - payload = {'msgtype': 'markdown', 'markdown': {'content': content}} - with open('wechat_payload.json', 'w') as f: - json.dump(payload, f, ensure_ascii=False) - " + python3 .github/scripts/build_wechat_payload.py curl -s "$WECHAT_WEBHOOK" \ -H 'Content-Type: application/json' \ From 99559621c5aabc5861c3ef5d1487e5a3ec43d45e Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Mon, 13 Apr 2026 19:24:50 +0800 Subject: [PATCH 032/245] ci(wechat-notify): inline payload building logic into workflow - Remove build_wechat_payload.py script and consolidate payload construction directly in workflow - Eliminate intermediate environment variables and file I/O operations for cleaner execution - Inline AI summary payload generation into curl request - Inline WeChat notification payload generation into curl request - Remove unnecessary checkout step since script is no longer needed - Simplify workflow by reducing file dependencies and improving readability --- .github/scripts/build_wechat_payload.py | 24 -------- .github/workflows/release-notify-wechat.yml | 62 +++++++-------------- 2 files changed, 21 insertions(+), 65 deletions(-) delete mode 100644 .github/scripts/build_wechat_payload.py diff --git a/.github/scripts/build_wechat_payload.py b/.github/scripts/build_wechat_payload.py deleted file mode 100644 index 5e292ee9..00000000 --- a/.github/scripts/build_wechat_payload.py +++ /dev/null @@ -1,24 +0,0 @@ -import json -import os - -branch = os.environ.get("BRANCH", "") -author = os.environ.get("AUTHOR", "") -pr_title = os.environ.get("PR_TITLE", "") -pr_url = os.environ.get("PR_URL", "") -ai_summary = os.environ.get("AI_SUMMARY", "") - -content = ( - "## 🚀 Release 发布通知\n" - f"> 📦 **分支**: {branch}\n" - f"> 👤 **提交人**: {author}\n" - f"> 📝 **标题**: {pr_title}\n\n" - "### 🧠 AI变更摘要\n" - f"{ai_summary}\n\n" - "---\n" - f"🔗 [查看PR详情]({pr_url})" -) - -payload = {"msgtype": "markdown", "markdown": {"content": content}} - -with open("wechat_payload.json", "w", encoding="utf-8") as f: - json.dump(payload, f, ensure_ascii=False) diff --git a/.github/workflows/release-notify-wechat.yml b/.github/workflows/release-notify-wechat.yml index b6c8a04d..4264570f 100644 --- a/.github/workflows/release-notify-wechat.yml +++ b/.github/workflows/release-notify-wechat.yml @@ -13,31 +13,23 @@ jobs: steps: # 防止 GitHub HEAD 未同步 - - name: Wait for ref sync - run: sleep 3 + - run: sleep 3 # 1️⃣ 获取分支 HEAD - name: Get HEAD id: head - env: - GH_TOKEN: ${{ secrets.GITHUB_TOKEN }} - REPO: ${{ github.repository }} - BASE_REF: ${{ github.event.pull_request.base.ref }} run: | HEAD_SHA=$(curl -s \ - -H "Authorization: Bearer $GH_TOKEN" \ - "https://api.github.com/repos/$REPO/git/ref/heads/$BASE_REF" \ + -H "Authorization: Bearer ${{ secrets.GITHUB_TOKEN }}" \ + https://api.github.com/repos/${{ github.repository }}/git/ref/heads/${{ github.event.pull_request.base.ref }} \ | jq -r '.object.sha') echo "head_sha=$HEAD_SHA" >> $GITHUB_OUTPUT # 2️⃣ 判断是否最终PR - name: Check Latest id: check - env: - MERGE_SHA: ${{ github.event.pull_request.merge_commit_sha }} - HEAD_SHA: ${{ steps.head.outputs.head_sha }} run: | - if [ "$MERGE_SHA" = "$HEAD_SHA" ]; then + if [ "${{ github.event.pull_request.merge_commit_sha }}" = "${{ steps.head.outputs.head_sha }}" ]; then echo "ok=true" >> $GITHUB_OUTPUT else echo "ok=false" >> $GITHUB_OUTPUT @@ -47,54 +39,42 @@ jobs: - name: Get Commits if: steps.check.outputs.ok == 'true' id: commits - env: - GH_TOKEN: ${{ secrets.GITHUB_TOKEN }} - COMMITS_URL: ${{ github.event.pull_request.commits_url }} run: | curl -s \ - -H "Authorization: Bearer $GH_TOKEN" \ - "$COMMITS_URL" \ + -H "Authorization: Bearer ${{ secrets.GITHUB_TOKEN }}" \ + ${{ github.event.pull_request.commits_url }} \ | jq -r '.[].commit.message' | head -n 20 > commits.txt # 4️⃣ 阿里 AI 总结（通义千问） - name: AI Summary (Qwen) if: steps.check.outputs.ok == 'true' id: ai - env: - DASHSCOPE_API_KEY: ${{ secrets.DASHSCOPE_API_KEY }} run: | - COMMIT_MESSAGES=$(cat commits.txt) - - jq -n --arg msgs "请用中文总结以下代码提交，输出3-5条，面向测试人员： - $COMMIT_MESSAGES" \ - '{"model": "qwen-plus", "input": {"prompt": $msgs}}' > ai_payload.json + CONTENT=$(cat commits.txt | sed ':a;N;$!ba;s/\n/\\n/g') SUMMARY=$(curl -s https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation \ - -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Authorization: Bearer ${{ secrets.DASHSCOPE_API_KEY }}" \ -H "Content-Type: application/json" \ - -d @ai_payload.json | jq -r '.output.text') + -d "{ + \"model\": \"qwen-plus\", + \"input\": { + \"prompt\": \"请用中文总结以下代码提交，输出3-5条，面向测试人员：\\n$CONTENT\" + } + }" | jq -r '.output.text') echo "summary<> $GITHUB_OUTPUT echo "$SUMMARY" >> $GITHUB_OUTPUT echo "EOF" >> $GITHUB_OUTPUT # 5️⃣ 企业微信通知（Markdown） - - name: Checkout for script - if: steps.check.outputs.ok == 'true' - uses: actions/checkout@v4 - - name: Notify WeChat if: steps.check.outputs.ok == 'true' - env: - WECHAT_WEBHOOK: ${{ secrets.WECHAT_WEBHOOK }} - BRANCH: ${{ github.event.pull_request.base.ref }} - AUTHOR: ${{ github.event.pull_request.user.login }} - PR_TITLE: ${{ github.event.pull_request.title }} - PR_URL: ${{ github.event.pull_request.html_url }} - AI_SUMMARY: ${{ steps.ai.outputs.summary }} run: | - python3 .github/scripts/build_wechat_payload.py - - curl -s "$WECHAT_WEBHOOK" \ + curl '${{ secrets.WECHAT_WEBHOOK }}' \ -H 'Content-Type: application/json' \ - -d @wechat_payload.json + -d "{ + \"msgtype\": \"markdown\", + \"markdown\": { + \"content\": \"## 🚀 Release 发布通知\n> 📦 **分支**: ${{ github.event.pull_request.base.ref }}\n> 👤 **提交人**: ${{ github.event.pull_request.user.login }}\n> 📝 **标题**: ${{ github.event.pull_request.title }}\n\n### 🧠 AI变更摘要\n${{ steps.ai.outputs.summary }}\n\n---\n🔗 [查看PR详情](${{ github.event.pull_request.html_url }})\" + } + }" From 90365cd026dee4de15ccedae5b63ece2bbd97354 Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Mon, 13 Apr 2026 19:28:10 +0800 Subject: [PATCH 033/245] ci(wechat-notify): refactor payload building to Python script - Extract WeChat notification payload construction from inline curl command - Move environment variables to explicit env section for clarity - Build JSON payload using Python for better string handling and readability - Write payload to temporary file and pass to curl via -d @wechat.json - Improves maintainability and reduces shell string escaping complexity --- .github/workflows/release-notify-wechat.yml | 33 ++++++++++++++++----- 1 file changed, 26 insertions(+), 7 deletions(-) diff --git a/.github/workflows/release-notify-wechat.yml b/.github/workflows/release-notify-wechat.yml index 4264570f..0c1ad7ca 100644 --- a/.github/workflows/release-notify-wechat.yml +++ b/.github/workflows/release-notify-wechat.yml @@ -69,12 +69,31 @@ jobs: # 5️⃣ 企业微信通知（Markdown） - name: Notify WeChat if: steps.check.outputs.ok == 'true' + env: + WECHAT_WEBHOOK: ${{ secrets.WECHAT_WEBHOOK }} + BRANCH: ${{ github.event.pull_request.base.ref }} + AUTHOR: ${{ github.event.pull_request.user.login }} + PR_TITLE: ${{ github.event.pull_request.title }} + PR_URL: ${{ github.event.pull_request.html_url }} + AI_SUMMARY: ${{ steps.ai.outputs.summary }} run: | - curl '${{ secrets.WECHAT_WEBHOOK }}' \ + python3 << 'PYEOF' + import json, os + content = ( + "## 🚀 Release 发布通知\n" + "> 📦 **分支**: " + os.environ["BRANCH"] + "\n" + "> 👤 **提交人**: " + os.environ["AUTHOR"] + "\n" + "> 📝 **标题**: " + os.environ["PR_TITLE"] + "\n\n" + "### 🧠 AI变更摘要\n" + + os.environ["AI_SUMMARY"] + "\n\n" + "---\n" + "🔗 [查看PR详情](" + os.environ["PR_URL"] + ")" + ) + payload = {"msgtype": "markdown", "markdown": {"content": content}} + with open("wechat.json", "w", encoding="utf-8") as f: + json.dump(payload, f, ensure_ascii=False) + PYEOF + + curl -s "$WECHAT_WEBHOOK" \ -H 'Content-Type: application/json' \ - -d "{ - \"msgtype\": \"markdown\", - \"markdown\": { - \"content\": \"## 🚀 Release 发布通知\n> 📦 **分支**: ${{ github.event.pull_request.base.ref }}\n> 👤 **提交人**: ${{ github.event.pull_request.user.login }}\n> 📝 **标题**: ${{ github.event.pull_request.title }}\n\n### 🧠 AI变更摘要\n${{ steps.ai.outputs.summary }}\n\n---\n🔗 [查看PR详情](${{ github.event.pull_request.html_url }})\" - } - }" + -d @wechat.json From bafcb5c5453fcd92a9cb9c6a3d0c3f8629b84338 Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Mon, 13 Apr 2026 19:30:15 +0800 Subject: [PATCH 034/245] ci(wechat-notify): replace curl with urllib for webhook request - Replace curl command with Python urllib.request for direct HTTP POST - Remove intermediate wechat.json file write, send payload directly - Add urllib.request import to Python script - Simplify workflow by eliminating file I/O and shell command dependency - Improves reliability by keeping notification logic entirely within Python --- .github/workflows/release-notify-wechat.yml | 16 +++++++++------- 1 file changed, 9 insertions(+), 7 deletions(-) diff --git a/.github/workflows/release-notify-wechat.yml b/.github/workflows/release-notify-wechat.yml index 0c1ad7ca..52fd0e44 100644 --- a/.github/workflows/release-notify-wechat.yml +++ b/.github/workflows/release-notify-wechat.yml @@ -78,7 +78,7 @@ jobs: AI_SUMMARY: ${{ steps.ai.outputs.summary }} run: | python3 << 'PYEOF' - import json, os + import json, os, urllib.request content = ( "## 🚀 Release 发布通知\n" "> 📦 **分支**: " + os.environ["BRANCH"] + "\n" @@ -90,10 +90,12 @@ jobs: "🔗 [查看PR详情](" + os.environ["PR_URL"] + ")" ) payload = {"msgtype": "markdown", "markdown": {"content": content}} - with open("wechat.json", "w", encoding="utf-8") as f: - json.dump(payload, f, ensure_ascii=False) + data = json.dumps(payload, ensure_ascii=False).encode("utf-8") + req = urllib.request.Request( + os.environ["WECHAT_WEBHOOK"], + data=data, + headers={"Content-Type": "application/json"} + ) + resp = urllib.request.urlopen(req) + print(resp.read().decode()) PYEOF - - curl -s "$WECHAT_WEBHOOK" \ - -H 'Content-Type: application/json' \ - -d @wechat.json From c614bb5be7402051ef8035179ac9502373f4e741 Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Mon, 13 Apr 2026 19:33:30 +0800 Subject: [PATCH 035/245] ci(wechat-notify): refine AI prompt for commit summarization MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit - Update prompt instruction to request numbered list format - Remove title and preamble from AI output for cleaner formatting - Improve clarity by specifying "要点" (key points) in prompt - Enhance consistency of release notification messages --- .github/workflows/release-notify-wechat.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.github/workflows/release-notify-wechat.yml b/.github/workflows/release-notify-wechat.yml index 52fd0e44..6894db15 100644 --- a/.github/workflows/release-notify-wechat.yml +++ b/.github/workflows/release-notify-wechat.yml @@ -58,7 +58,7 @@ jobs: -d "{ \"model\": \"qwen-plus\", \"input\": { - \"prompt\": \"请用中文总结以下代码提交，输出3-5条，面向测试人员：\\n$CONTENT\" + \"prompt\": \"请用中文总结以下代码提交，输出3-5条要点，面向测试人员。直接输出编号列表，不要输出标题或前言：\\n$CONTENT\" } }" | jq -r '.output.text') From 0d16e168e75ebe611ec6baab938ea4f7ef222881 Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Mon, 13 Apr 2026 19:36:27 +0800 Subject: [PATCH 036/245] ci(wechat-notify): refactor AI summary generation to Python - Replace curl with urllib.request for API calls to improve portability - Move API key to environment variable for better security practices - Inline Python script using heredoc for cleaner workflow definition - Add intermediate file (ai_summary.txt) to separate concerns between API call and output handling - Simplify JSON payload construction using Python's json module - Improve error handling with fallback message for failed AI generation --- .github/workflows/release-notify-wechat.yml | 38 +++++++++++++++------ 1 file changed, 28 insertions(+), 10 deletions(-) diff --git a/.github/workflows/release-notify-wechat.yml b/.github/workflows/release-notify-wechat.yml index 6894db15..d8f4e5aa 100644 --- a/.github/workflows/release-notify-wechat.yml +++ b/.github/workflows/release-notify-wechat.yml @@ -49,19 +49,37 @@ jobs: - name: AI Summary (Qwen) if: steps.check.outputs.ok == 'true' id: ai + env: + DASHSCOPE_API_KEY: ${{ secrets.DASHSCOPE_API_KEY }} run: | - CONTENT=$(cat commits.txt | sed ':a;N;$!ba;s/\n/\\n/g') + python3 << 'PYEOF' + import json, os, urllib.request - SUMMARY=$(curl -s https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation \ - -H "Authorization: Bearer ${{ secrets.DASHSCOPE_API_KEY }}" \ - -H "Content-Type: application/json" \ - -d "{ - \"model\": \"qwen-plus\", - \"input\": { - \"prompt\": \"请用中文总结以下代码提交，输出3-5条要点，面向测试人员。直接输出编号列表，不要输出标题或前言：\\n$CONTENT\" - } - }" | jq -r '.output.text') + with open("commits.txt", "r") as f: + commits = f.read().strip() + prompt = "请用中文总结以下代码提交，输出3-5条要点，面向测试人员。直接输出编号列表，不要输出标题或前言：\n" + commits + payload = {"model": "qwen-plus", "input": {"prompt": prompt}} + data = json.dumps(payload, ensure_ascii=False).encode("utf-8") + + req = urllib.request.Request( + "https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation", + data=data, + headers={ + "Authorization": "Bearer " + os.environ["DASHSCOPE_API_KEY"], + "Content-Type": "application/json" + } + ) + resp = urllib.request.urlopen(req) + result = json.loads(resp.read().decode()) + summary = result.get("output", {}).get("text", "AI 摘要生成失败") + print(summary) + + with open("ai_summary.txt", "w", encoding="utf-8") as f: + f.write(summary) + PYEOF + + SUMMARY=$(cat ai_summary.txt) echo "summary<> $GITHUB_OUTPUT echo "$SUMMARY" >> $GITHUB_OUTPUT echo "EOF" >> $GITHUB_OUTPUT From 06075ffef504c0b58e286b5cf434cefc7633b68f Mon Sep 17 00:00:00 2001 From: zhaoying Date: Tue, 14 Apr 2026 09:57:36 +0800 Subject: [PATCH 037/245] fix(web): calculate using the filtered breadcrumbs length --- web/src/components/Header/index.tsx | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/web/src/components/Header/index.tsx b/web/src/components/Header/index.tsx index f2eff014..d85a84b0 100644 --- a/web/src/components/Header/index.tsx +++ b/web/src/components/Header/index.tsx @@ -135,9 +135,10 @@ const AppHeader: FC<{source?: 'space' | 'manage';}> = ({source = 'manage'}) => { * - Disables navigation for the last breadcrumb item */ const formatBreadcrumbNames = () => { - return breadcrumbs.filter(item => item.type !== 'group').map((menu, index) => { + const filtered = breadcrumbs.filter(item => item.type !== 'group'); + return filtered.map((menu, index) => { const label = menu.i18nKey ? t(menu.i18nKey) : menu.label; - const isLast = index === breadcrumbs.length - 1; + const isLast = index === filtered.length - 1; const item: any = { title: ( From 2d9986f9025dd577108dc1a4f7572e7e2a75fe89 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Tue, 14 Apr 2026 10:03:46 +0800 Subject: [PATCH 038/245] fix(web): header user name --- web/src/components/Header/index.tsx | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/web/src/components/Header/index.tsx b/web/src/components/Header/index.tsx index d85a84b0..23a89894 100644 --- a/web/src/components/Header/index.tsx +++ b/web/src/components/Header/index.tsx @@ -77,7 +77,7 @@ const AppHeader: FC<{source?: 'space' | 'manage';}> = ({source = 'manage'}) => { { key: '1', icon: - {/[\u4e00-\u9fa5]/.test(user.username) ? user.username.slice(0, 2) : user.username?.[0]} + {/[\u4e00-\u9fa5]/.test(user.username) ? user.username.slice(-2) : user.username[0]} , label: (<>
{user.username}
@@ -182,7 +182,7 @@ const AppHeader: FC<{source?: 'space' | 'manage';}> = ({source = 'manage'}) => { > - {/[\u4e00-\u9fa5]/.test(user.username) ? user.username.slice(user.username.length, -2) : user.username[0]} + {/[\u4e00-\u9fa5]/.test(user.username) ? user.username.slice(-2) : user.username[0]} {user.username}
Date: Tue, 14 Apr 2026 10:20:50 +0800 Subject: [PATCH 039/245] feat(web): chat add file info --- web/src/components/Chat/ChatContent.tsx | 5 +++-- web/src/views/Conversation/index.tsx | 9 ++++++--- 2 files changed, 9 insertions(+), 5 deletions(-) diff --git a/web/src/components/Chat/ChatContent.tsx b/web/src/components/Chat/ChatContent.tsx index 5c722e45..f28b5dce 100644 --- a/web/src/components/Chat/ChatContent.tsx +++ b/web/src/components/Chat/ChatContent.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2025-12-10 16:46:17 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-10 18:46:57 + * @Last Modified time: 2026-04-14 10:13:56 */ import { type FC, useRef, useEffect, useState } from 'react' import clsx from 'clsx' @@ -174,6 +174,7 @@ const ChatContent: FC = ({ ) } + const documentType = (file.file_type || file.type)?.split('/') return ( = ({ >
{file.name}
-
{file.type?.split('/')[file.type?.split('/').length - 1]} · {file.size}
+
{documentType?.[documentType.length - 1]} · {file.size}
 ) diff --git a/web/src/views/Conversation/index.tsx b/web/src/views/Conversation/index.tsx index 8432b1c2..778279d3 100644 --- a/web/src/views/Conversation/index.tsx +++ b/web/src/views/Conversation/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 16:58:03 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-07 21:21:52 + * @Last Modified time: 2026-04-13 18:32:58 */ /** * Conversation Page @@ -397,7 +397,10 @@ const Conversation: FC = () => { return { type: file.type, transfer_method: 'local_file', - upload_file_id: file.response.data.file_id + upload_file_id: file.response.data.file_id, + file_type: file.response.data.file_type, + size: file.response.data.file_size, + name: file.response.data.file_name } } }), @@ -444,7 +447,7 @@ const Conversation: FC = () => { }) } - console.log('chatList', chatList, streamLoadingRef.current) + console.log('chatList', fileList, streamLoadingRef.current) return ( From 7f8765b815b30e3cafaf75daba9c9c455c6ba445 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Tue, 14 Apr 2026 14:51:02 +0800 Subject: [PATCH 040/245] feat(web): package --- web/src/api/package.ts | 14 +++ web/src/i18n/en.ts | 64 +++++++++++++ web/src/i18n/zh.ts | 64 +++++++++++++ web/src/routes/index.tsx | 5 +- web/src/routes/routes.json | 3 +- web/src/store/menu.json | 16 +--- web/src/utils/request.ts | 7 +- web/src/views/Package/constant.ts | 34 +++++++ web/src/views/Package/index.tsx | 145 ++++++++++++++++++++++++++++++ web/src/views/Package/types.ts | 61 +++++++++++++ 10 files changed, 392 insertions(+), 21 deletions(-) create mode 100644 web/src/api/package.ts create mode 100644 web/src/views/Package/constant.ts create mode 100644 web/src/views/Package/index.tsx create mode 100644 web/src/views/Package/types.ts diff --git a/web/src/api/package.ts b/web/src/api/package.ts new file mode 100644 index 00000000..da52d355 --- /dev/null +++ b/web/src/api/package.ts @@ -0,0 +1,14 @@ +import { request } from '@/utils/request' + +import type { Package } from '@/views/Package/types' + +export const SYS_API_PREFIX = '/sys'; +// 套餐列表 +export const getPackageListUrl = `${SYS_API_PREFIX}/package-plans` +export const getPackageList = (query: { category: Package['category']; status: boolean; }) => { + return request.get(getPackageListUrl, query) +} +// 获取套餐详情 +export const getPackageDetail = (package_plan_id: string) => { + return request.get(`${SYS_API_PREFIX}/package-plans/${package_plan_id}`) +} \ No newline at end of file diff --git a/web/src/i18n/en.ts b/web/src/i18n/en.ts index fc3a041d..6bcc5034 100644 --- a/web/src/i18n/en.ts +++ b/web/src/i18n/en.ts @@ -3016,5 +3016,69 @@ Memory Bear: After the rebellion, regional warlordism intensified for several re apply: 'Apply', tools: 'Tools', }, + package: { + package: 'Package Management', + saas_personal: 'SaaS Personal', + commercial_deployment: 'Commercial Deployment', + noCommercialPackages: 'No commercial deployment packages available', + + addPackage: 'Add Plan', + packageName: 'Plan Name', + packageNameZh: 'Plan Name (中文)', + packageNameEn: 'Plan Name (English)', + packageNamePlaceholder: '中文, 例如：记忆体验版', + packageNamePlaceholderEn: 'English, e.g. Memory Trial Plan', + packageCategory: 'Package Category', + price: 'Price', + pricePlaceholder: 'e.g. 0, 19, 299 or Contact Us', + billingPeriod: 'Billing Period', + monthly: 'Monthly', + yearly: 'Yearly', + permanent_free: 'Permanent Free', + local_deployment: 'Local Deployment', + coreValue: 'Core Value', + coreValueZh: 'Core Value (中文)', + coreValueEn: 'Core Value (English)', + coreValuePlaceholder: '中文, 一句话描述核心价值', + coreValuePlaceholderEn: 'EngLish, describe the core value in one sentence', + tech_support: 'Technical Support', + tech_support_zh: 'Technical Support (中文)', + tech_support_en: 'Technical Support (English)', + technicalSupportPlaceholder: '中文, 例如：社群交流、工单支持', + technicalSupportPlaceholderEn: 'English, e.g. Community support, ticket support', + sla: 'SLA & Compliance', + slaZh: 'SLA & Compliance (中文)', + slaEn: 'SLA & Compliance (English)', + slaPlaceholder: '中文, 例如：无、验证力加强+审计日志', + slaPlaceholderEn: 'English, e.g. None, dedicated compute pool + audit logs', + customPage: 'Chat Page Customization', + customPageZh: 'Chat Page Customization (中文)', + customPageEn: 'Chat Page Customization (English)', + customPagePlaceholder: '中文, 例如：LOGO定制', + customPagePlaceholderEn: 'English, e.g. Logo customization', + primaryColor: 'Primary Color', + status: 'Status', + active: 'Active', + inactive: 'Inactive', + api_ops_rate_limit: 'API OPS Rate Limit', + ops: 'req/s', + pcs: 'pcs', + GB: 'GB', + tier_level: 'Tier Level', + numberPlaceholder: 'e.g. 10', + + packageDetail: 'Package Detail', + basicInfo: 'Basic Info', + featureConfig: 'Billing Unit Quota', + workspace_quota: 'Workspace Quota', + skill_quota: 'Skill Library Quota', + app_quota: 'App Quota', + knowledge_capacity_quota: 'Knowledge Base Capacity', + memory_engine_quota: 'Memory Engine Quota', + end_user_quota: 'Memorable End Users', + ontology_project_quota: 'Ontology Project', + model_quota: 'Model Quota', + editPackage: 'Edit Package', + }, }, }; diff --git a/web/src/i18n/zh.ts b/web/src/i18n/zh.ts index 01c766b8..fff8c1af 100644 --- a/web/src/i18n/zh.ts +++ b/web/src/i18n/zh.ts @@ -2980,5 +2980,69 @@ export const zh = { apply: '应用', tools: '工具', }, + package: { + package: '套餐管理', + saas_personal: 'SaaS 个人版', + commercial_deployment: '商业化部署', + noCommercialPackages: '暂无商业化部署套餐', + + addPackage: '添加套餐', + packageName: '套餐名称', + packageNameZh: '套餐名称 (中文)', + packageNameEn: '套餐名称 (English)', + packageNamePlaceholder: '中文, 例如：记忆体验版', + packageNamePlaceholderEn: 'English, e.g. Memory Trial Plan', + packageCategory: '套餐分类', + price: '价格', + pricePlaceholder: '例如: 0, 19, 299 或联系我们', + billingPeriod: '计费周期', + monthly: '月', + yearly: '年', + permanent_free: '永久免费', + local_deployment: '本地化部署', + coreValue: '核心价值', + coreValueZh: '核心价值 (中文)', + coreValueEn: '核心价值 (English)', + coreValuePlaceholder: '中文, 一句话描述核心价值', + coreValuePlaceholderEn: 'EngLish, describe the core value in one sentence', + tech_support: '技术支持', + tech_support_zh: '技术支持 (中文)', + tech_support_en: '技术支持 (English)', + technicalSupportPlaceholder: '中文, 例如：社群交流、工单支持', + technicalSupportPlaceholderEn: 'English, e.g. Community support, ticket support', + sla: 'SLA与合规', + slaZh: 'SLA与合规 (中文)', + slaEn: 'SLA与合规 (English)', + slaPlaceholder: '中文, 例如：无、验证力加强+审计日志', + slaPlaceholderEn: 'English, e.g. None, dedicated compute pool + audit logs', + customPage: '对应页面个性化配置', + customPageZh: '对应页面个性化配置 (中文)', + customPageEn: '对应页面个性化配置 (English)', + customPagePlaceholder: '中文, 例如：LOGO定制', + customPagePlaceholderEn: 'English, e.g. Logo customization', + primaryColor: '主题色', + status: '状态', + active: '启用', + inactive: '停用', + api_ops_rate_limit: 'API OPS 频次', + ops: '次/秒', + pcs: '个', + GB: 'GB', + tier_level: '层级', + numberPlaceholder: '如: 10', + + packageDetail: '套餐详情', + basicInfo: '基础信息', + featureConfig: '计费单元配额', + workspace_quota: '空间数量', + skill_quota: '技能库数量', + app_quota: '应用数量', + knowledge_capacity_quota: '知识库容量', + memory_engine_quota: '记忆引擎数量', + end_user_quota: '可记忆终端用户数', + ontology_project_quota: '本体工程', + model_quota: '可负载模型数量', + editPackage: '编辑套餐', + }, }, } \ No newline at end of file diff --git a/web/src/routes/index.tsx b/web/src/routes/index.tsx index 92f7a5cf..7b940068 100644 --- a/web/src/routes/index.tsx +++ b/web/src/routes/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-02 16:33:11 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-02-04 18:11:34 + * @Last Modified time: 2026-04-13 16:53:15 */ /** * Route Configuration @@ -76,13 +76,12 @@ const componentMap: Record>> = SpaceManagement: lazy(() => import('@/views/SpaceManagement')), ApiKeyManagement: lazy(() => import('@/views/ApiKeyManagement')), EmotionEngine: lazy(() => import('@/views/EmotionEngine')), - StatementDetail: lazy(() => import('@/views/UserMemoryDetail/pages/StatementDetail')), ForgetDetail: lazy(() => import('@/views/UserMemoryDetail/pages/ForgetDetail')), MemoryNodeDetail: lazy(() => import('@/views/UserMemoryDetail/pages/index')), SelfReflectionEngine: lazy(() => import('@/views/SelfReflectionEngine')), OrderPayment: lazy(() => import('@/views/OrderPayment')), OrderHistory: lazy(() => import('@/views/OrderHistory')), - Pricing: lazy(() => import('@/views/Pricing')), + Package: lazy(() => import('@/views/Package')), ToolManagement: lazy(() => import('@/views/ToolManagement')), SpaceConfig: lazy(() => import('@/views/SpaceConfig')), Ontology: lazy(() => import('@/views/Ontology')), diff --git a/web/src/routes/routes.json b/web/src/routes/routes.json index 5ff1f90c..422387a7 100644 --- a/web/src/routes/routes.json +++ b/web/src/routes/routes.json @@ -7,7 +7,7 @@ { "path": "/model", "element": "ModelManagement" }, { "path": "/space", "element": "SpaceManagement" }, { "path": "/tool", "element": "ToolManagement" }, - { "path": "/pricing", "element": "Pricing" }, + { "path": "/pricing", "element": "Package" }, { "path": "/order-pay", "element": "OrderPayment" }, { "path": "/orders", "element": "OrderHistory" }, { "path": "/skills", "element": "Skills" }, @@ -48,7 +48,6 @@ { "path": "/application/config/:id", "element": "ApplicationConfig" }, { "path": "/application/config/:id/:source", "element": "ApplicationConfig" }, { "path": "/user-memory/neo4j/:id", "element": "Neo4jUserMemoryDetail" }, - { "path": "/statement/:id", "element": "StatementDetail" }, { "path": "/user-memory/detail/:id/:type", "element": "MemoryNodeDetail" }, { "path": "/ontology/:id", "element": "OntologyDetail" } ] diff --git a/web/src/store/menu.json b/web/src/store/menu.json index 8d30dcc4..90b546ca 100644 --- a/web/src/store/menu.json +++ b/web/src/store/menu.json @@ -421,21 +421,7 @@ "display": false, "level": 3, "sort": 0, - "subs": [ - { - "id": 2211, - "parent": 221, - "code": "statementDetail", - "label": "记忆详情", - "i18nKey": "menu.statementDetail", - "path": "/statement/:id", - "enable": true, - "display": false, - "level": 4, - "sort": 0, - "subs": null - } - ] + "subs": [] }, { "id": 222, diff --git a/web/src/utils/request.ts b/web/src/utils/request.ts index 80c12f85..318738dd 100644 --- a/web/src/utils/request.ts +++ b/web/src/utils/request.ts @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-02 16:35:15 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-06 10:39:00 + * @Last Modified time: 2026-04-14 14:43:54 */ /** * HTTP Request Utility Module @@ -23,6 +23,7 @@ import { clearAuthData } from './auth'; import { message } from 'antd'; import { refreshTokenUrl, refreshToken, loginUrl, logoutUrl } from '@/api/user' import i18n from '@/i18n' +import { SYS_API_PREFIX } from '@/api/package' /** * Standard API response structure @@ -74,6 +75,10 @@ let requests: RequestQueueItem[] = []; // Request interceptor service.interceptors.request.use( (config) => { + console.log('config', config, config.url?.startsWith(SYS_API_PREFIX)) + if (config.url?.startsWith(SYS_API_PREFIX)) { + config.baseURL = ''; + } if (!config.headers.Authorization) { const token = cookieUtils.get('authToken'); if (token) { diff --git a/web/src/views/Package/constant.ts b/web/src/views/Package/constant.ts new file mode 100644 index 00000000..e4b04719 --- /dev/null +++ b/web/src/views/Package/constant.ts @@ -0,0 +1,34 @@ +export const billingUnits = [ + { + key: 'workspace_quota', + unit: '个', placeholder: '如: 10', + }, + { + key: 'skill_quota', + unit: '个', placeholder: '如: 10', + }, + { + key: 'app_quota', + unit: '个', placeholder: '如: 10', + }, + { + key: 'knowledge_capacity_quota', + unit: 'GB', placeholder: '如: 10', + }, + { + key: 'memory_engine_quota', + unit: '个', placeholder: '如: 10', + }, + { + key: 'end_user_quota', + unit: '个', placeholder: '如: 10', + }, + { + key: 'ontology_project_quota', + unit: '个', placeholder: '如: 10', + }, + { + key: 'model_quota', + unit: '次/秒', placeholder: '如: 10', + }, +] \ No newline at end of file diff --git a/web/src/views/Package/index.tsx b/web/src/views/Package/index.tsx new file mode 100644 index 00000000..0e8a347f --- /dev/null +++ b/web/src/views/Package/index.tsx @@ -0,0 +1,145 @@ +/* + * @Author: ZhaoYing + * @Date: 2026-02-25 + * @Last Modified by: ZhaoYing + * @Last Modified time: 2026-04-14 14:48:13 + */ +/** + * Package Component + * + * Package management page with: + * - Tabs for SaaS Personal and Commercial Deployment + * - Package cards showing features and pricing + * - Edit and delete actions + * + * @component + */ + +import { useMemo, useState, useEffect, type FC } from 'react'; +import { useTranslation } from 'react-i18next'; +import { useNavigate } from 'react-router-dom'; +import { Flex, Row, Col, type SegmentedProps } from 'antd'; +import clsx from 'clsx'; + +import type { Package } from './types' +import { getPackageList } from '@/api/package'; +import PageTabs from '@/components/PageTabs' +import { billingUnits } from './constant' +import RbCard from '@/components/RbCard/Card' +import BodyWrapper from '@/components/Empty/BodyWrapper' +import { useI18n } from '@/store/locale' +import RbButton from '@/components/RbButton' + +const Package: FC = () => { + const { t } = useTranslation(); + const { language } = useI18n() + const navigate = useNavigate(); + const [data, setData] = useState([]) + + const [activeTab, setActiveTab] = useState('saas_personal'); + const formatTabItems = useMemo(() => { + return ['saas_personal', 'commercial_deployment'].map(value => ({ + value, + label: t(`package.${value}`), + })) + }, [t]) + /** Handle tab change */ + const handleChangeTab = (value: SegmentedProps['value']) => { + setActiveTab(value as string); + } + const getList = () => { + getPackageList({ category: activeTab as Package['category'], status: true }) + .then(res => { + setData(res as Package[] || []) + }) + } + + useEffect(() => { + getList() + }, [activeTab]) + + const getKeyWithLanguage = (key: string) => { + return (language === 'en' ? `${key}_en` : key) as keyof Package + } + /** Navigate to order history */ + const goToHistory = () => { + navigate('/orders'); + } + return ( + <> + + + +
+ {t('pricing.orderHistory')} + + + + + {data.map((pkg) => ( + + + +
+ {/* Header */} +
+

+ {String(pkg[getKeyWithLanguage('name')] ?? '')} +

+

{String(pkg[getKeyWithLanguage('core_value')] ?? '')}

+
+ {pkg.billing_cycle !== 'permanent_free' && <>¥{pkg.price}} + {pkg.billing_cycle && {pkg.billing_cycle !== 'permanent_free' && '/'}{t(`package.${pkg.billing_cycle}`)}} +
+
+ + {/* Features */} +
+ {billingUnits.map(({ key, unit }) => { + if (typeof pkg.quotas[key as keyof Package['quotas']] === 'number') { + return ( +
+ {t(`package.${key}`)} + {pkg.quotas[key as keyof Package['quotas']]}{unit} +
+ ) + } + })} + {pkg.tech_support && +
+ {t(`package.tech_support`)} + {String(pkg[getKeyWithLanguage('tech_support')] ?? '')} +
+ } + {pkg.api_ops_rate_limit && +
+ {t(`package.api_ops_rate_limit`)} + {pkg.api_ops_rate_limit}（次/秒) +
+ } +
+
+ + + + + ))} + + + + ); +}; + +export default Package; diff --git a/web/src/views/Package/types.ts b/web/src/views/Package/types.ts new file mode 100644 index 00000000..6517f63a --- /dev/null +++ b/web/src/views/Package/types.ts @@ -0,0 +1,61 @@ +/* + * @Author: ZhaoYing + * @Date: 2026-04-14 11:35:01 + * @Last Modified by: ZhaoYing + * @Last Modified time: 2026-04-14 14:28:46 + */ +export interface Package { + id: string; + // 名称 + name: string; + name_en: string; + // 类型 + category: "saas_personal" | "commercial_deployment"; + tier_level: number; + // 版本 + version: string; + // 状态 + status: boolean; + // 价格 + price: string; + // 计费周期 + billing_cycle: "monthly" | "yearly" | "permanent_free" | "local_deployment"; + // 核心价值 + core_value: string; + core_value_en: string; + // 技术支持 + tech_support: string; + tech_support_en: string; + // SLA与合规 + sla_compliance: string; + sla_compliance_en: string; + // 对话页面个性化配置 + page_customization: string; + page_customization_en: string; + // API OPS 频次（次/秒） + api_ops_rate_limit: number; + // 主题色 + theme_color: string; + quotas: { + // 空间数量 + workspace_quota: number; + // 技能库数量 + skill_quota: number; + // 应用数量 + app_quota: number; + // 知识库容量 + knowledge_capacity_quota: string; + // 记忆引擎数量 + memory_engine_quota: number; + // 可记忆终端用户数 + end_user_quota: number; + // 本体工程 + ontology_project_quota: number; + // 可负载模型数量 + model_quota: number; + }, + created_at: number; + updated_at: number; + created_by: string; + updated_by: string | null; +} From e539b3eeb7541291c45535e27623eadfe51b13b7 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Tue, 14 Apr 2026 14:59:32 +0800 Subject: [PATCH 041/245] fix(web): i18n --- web/src/views/Package/constant.ts | 22 ++++++++++++++-------- web/src/views/Package/index.tsx | 16 ++++++++-------- 2 files changed, 22 insertions(+), 16 deletions(-) diff --git a/web/src/views/Package/constant.ts b/web/src/views/Package/constant.ts index e4b04719..8d3b0d48 100644 --- a/web/src/views/Package/constant.ts +++ b/web/src/views/Package/constant.ts @@ -1,34 +1,40 @@ +/* + * @Author: ZhaoYing + * @Date: 2026-04-14 11:43:57 + * @Last Modified by: ZhaoYing + * @Last Modified time: 2026-04-14 14:55:20 + */ export const billingUnits = [ { key: 'workspace_quota', - unit: '个', placeholder: '如: 10', + unit: 'pcs', placeholder: 'numberPlaceholder', }, { key: 'skill_quota', - unit: '个', placeholder: '如: 10', + unit: 'pcs', placeholder: 'numberPlaceholder', }, { key: 'app_quota', - unit: '个', placeholder: '如: 10', + unit: 'pcs', placeholder: 'numberPlaceholder', }, { key: 'knowledge_capacity_quota', - unit: 'GB', placeholder: '如: 10', + unit: 'GB', placeholder: 'numberPlaceholder', }, { key: 'memory_engine_quota', - unit: '个', placeholder: '如: 10', + unit: 'pcs', placeholder: 'numberPlaceholder', }, { key: 'end_user_quota', - unit: '个', placeholder: '如: 10', + unit: 'pcs', placeholder: 'numberPlaceholder', }, { key: 'ontology_project_quota', - unit: '个', placeholder: '如: 10', + unit: 'pcs', placeholder: 'numberPlaceholder', }, { key: 'model_quota', - unit: '次/秒', placeholder: '如: 10', + unit: 'ops', placeholder: 'numberPlaceholder', }, ] \ No newline at end of file diff --git a/web/src/views/Package/index.tsx b/web/src/views/Package/index.tsx index 0e8a347f..64ce0c04 100644 --- a/web/src/views/Package/index.tsx +++ b/web/src/views/Package/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-25 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-14 14:48:13 + * @Last Modified time: 2026-04-14 14:59:11 */ /** * Package Component @@ -112,23 +112,23 @@ const Package: FC = () => { return (
{t(`package.${key}`)} - {pkg.quotas[key as keyof Package['quotas']]}{unit} + {pkg.quotas[key as keyof Package['quotas']]}{t(`package.${unit}`)}
) } })} + {pkg.api_ops_rate_limit && +
+ {t(`package.api_ops_rate_limit`)} + {pkg.api_ops_rate_limit}{t('package.ops')} +
+ } {pkg.tech_support &&
{t(`package.tech_support`)} {String(pkg[getKeyWithLanguage('tech_support')] ?? '')}
} - {pkg.api_ops_rate_limit && -
- {t(`package.api_ops_rate_limit`)} - {pkg.api_ops_rate_limit}（次/秒) -
- }
From b5ec5c2ceaa1c5fa00152564f6510576f2b01b2f Mon Sep 17 00:00:00 2001 From: zhaoying Date: Tue, 14 Apr 2026 15:08:07 +0800 Subject: [PATCH 042/245] fix(web): change http body key name --- .../components/Properties/HttpRequest/EditableTable.tsx | 4 ++-- web/src/views/Workflow/hooks/useWorkflowGraph.ts | 6 +++--- 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/web/src/views/Workflow/components/Properties/HttpRequest/EditableTable.tsx b/web/src/views/Workflow/components/Properties/HttpRequest/EditableTable.tsx index e0a27b47..e4b2cc29 100644 --- a/web/src/views/Workflow/components/Properties/HttpRequest/EditableTable.tsx +++ b/web/src/views/Workflow/components/Properties/HttpRequest/EditableTable.tsx @@ -85,9 +85,9 @@ const EditableTable: FC = ({ return [ { title: t('workflow.config.name'), - dataIndex: 'name', + dataIndex: 'key', render: (_: any, __: TableRow, index: number) => ( - + ).map(([key, value]) => ({ key, value })) : group_variables } else if (type === 'http-request' && (key === 'headers' || key === 'params') && config[key] && typeof config[key] === 'object' && !Array.isArray(config[key]) && nodeLibraryConfig.config && nodeLibraryConfig.config[key]) { - nodeLibraryConfig.config[key].defaultValue = Object.entries(config[key]).map(([name, value]) => ({ name, value })) + nodeLibraryConfig.config[key].defaultValue = Object.entries(config[key]).map(([key, value]) => ({ key, value })) } else if (type === 'code' && key === 'code' && config[key] && nodeLibraryConfig.config && nodeLibraryConfig.config[key]) { try { nodeLibraryConfig.config[key].defaultValue = decodeURIComponent(atob(config[key] as string)) @@ -1259,7 +1259,7 @@ export const useWorkflowGraph = ({ itemConfig[key] = {} if (value.length > 0) { value.forEach((vo: any) => { - itemConfig[key][vo.name] = vo.value + itemConfig[key][vo.key] = vo.value }) } } else if (data.config[key] && 'defaultValue' in data.config[key] && key !== 'knowledge_retrieval') { From 09650082103b1915369677603d4044313bcf9d95 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Tue, 14 Apr 2026 15:53:16 +0800 Subject: [PATCH 043/245] fix(http-request): support array and file variables in form-data files upload - Updated form-data handling to accept both single FileVariable and ArrayVariable containing FileVariable for file uploads - Fixed HTTP client redirect handling by enabling follow_redirects=True when downloading remote files - Adjusted config validation to correctly require list type for form-data fields instead of HttpFormData class --- .../workflow/nodes/http_request/config.py | 4 ++-- .../core/workflow/nodes/http_request/node.py | 19 ++++++++++++------- .../workflow/variable/variable_objects.py | 2 +- 3 files changed, 15 insertions(+), 10 deletions(-) diff --git a/api/app/core/workflow/nodes/http_request/config.py b/api/app/core/workflow/nodes/http_request/config.py index e1b84f0c..3473f666 100644 --- a/api/app/core/workflow/nodes/http_request/config.py +++ b/api/app/core/workflow/nodes/http_request/config.py @@ -72,8 +72,8 @@ class HttpContentTypeConfig(BaseModel): @classmethod def validate_data(cls, v, info): content_type = info.data.get("content_type") - if content_type == HttpContentType.FROM_DATA and not isinstance(v, HttpFormData): - raise ValueError("When content_type is 'form-data', data must be of type HttpFormData") + if content_type == HttpContentType.FROM_DATA and not isinstance(v, list): + raise ValueError("When content_type is 'form-data', data must be a list of HttpFormData") elif content_type in [HttpContentType.JSON] and not isinstance(v, str): raise ValueError("When content_type is JSON, data must be of type str") elif content_type in [HttpContentType.WWW_FORM] and not isinstance(v, dict): diff --git a/api/app/core/workflow/nodes/http_request/node.py b/api/app/core/workflow/nodes/http_request/node.py index 086bee4a..783c230b 100644 --- a/api/app/core/workflow/nodes/http_request/node.py +++ b/api/app/core/workflow/nodes/http_request/node.py @@ -260,17 +260,22 @@ class HttpRequestNode(BaseNode): )) case HttpContentType.FROM_DATA: data = {} - content["files"] = {} + files = [] for item in self.typed_config.body.data: + key = self._render_template(item.key, variable_pool) if item.type == "text": - data[self._render_template(item.key, variable_pool)] = self._render_template(item.value, - variable_pool) + data[key] = self._render_template(item.value, variable_pool) elif item.type == "file": - content["files"][self._render_template(item.key, variable_pool)] = ( - uuid.uuid4().hex, - await variable_pool.get_instance(item.value).get_content() - ) + file_instance = variable_pool.get_instance(item.value) + if isinstance(file_instance, ArrayVariable): + for v in file_instance.value: + if isinstance(v, FileVariable): + files.append((key, (uuid.uuid4().hex, await v.get_content()))) + elif isinstance(file_instance, FileVariable): + files.append((key, (uuid.uuid4().hex, await file_instance.get_content()))) content["data"] = data + if files: + content["files"] = files case HttpContentType.BINARY: content["files"] = [] file_instence = variable_pool.get_instance(self.typed_config.body.data) diff --git a/api/app/core/workflow/variable/variable_objects.py b/api/app/core/workflow/variable/variable_objects.py index 94f87287..2b849c94 100644 --- a/api/app/core/workflow/variable/variable_objects.py +++ b/api/app/core/workflow/variable/variable_objects.py @@ -84,7 +84,7 @@ class FileVariable(BaseVariable): total_bytes = 0 chunks = [] - async with httpx.AsyncClient() as client: + async with httpx.AsyncClient(follow_redirects=True) as client: async with client.stream("GET", self.value.url) as resp: resp.raise_for_status() async for chunk in resp.aiter_bytes(8192): From fa1e5ee43c962d66c6b20e4d12765fae996dc2c5 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Tue, 14 Apr 2026 16:06:03 +0800 Subject: [PATCH 044/245] fix(web): adjust the value of End User Name --- web/src/views/UserMemoryDetail/Neo4j.tsx | 10 ++++++---- .../UserMemoryDetail/components/EndUserProfile.tsx | 7 ++++--- web/src/views/UserMemoryDetail/types.ts | 3 ++- 3 files changed, 12 insertions(+), 8 deletions(-) diff --git a/web/src/views/UserMemoryDetail/Neo4j.tsx b/web/src/views/UserMemoryDetail/Neo4j.tsx index 51be7c8d..3fdaaed3 100644 --- a/web/src/views/UserMemoryDetail/Neo4j.tsx +++ b/web/src/views/UserMemoryDetail/Neo4j.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 17:57:26 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-26 18:59:53 + * @Last Modified time: 2026-04-14 16:04:08 */ /** * Neo4j User Memory Detail View @@ -22,7 +22,7 @@ import InterestDistribution from './components/InterestDistribution' import NodeStatistics from './components/NodeStatistics' import RelationshipNetwork from './components/RelationshipNetwork' import MemoryInsight from './components/MemoryInsight' -import type { EndUserProfileRef, MemoryInsightRef, AboutMeRef } from './types' +import type { EndUserProfileRef, MemoryInsightRef, AboutMeRef, EndUser } from './types' import { analyticsRefresh, } from '@/api/memory' @@ -39,8 +39,10 @@ const Neo4j: FC = () => { const [selectedKey, setSelectedKey] = useState(null) /** Update displayed name */ - const handleNameUpdate = (data: { other_name?: string; id: string }) => { - setName(data.other_name && data.other_name !== '' ? data.other_name : data.id) + const handleNameUpdate = (data?: EndUser) => { + if (!data) return + let name = data.other_name && data.other_name !== '' ? data.other_name : data.id || data.end_user_id + setName(name) } /** Navigate back */ diff --git a/web/src/views/UserMemoryDetail/components/EndUserProfile.tsx b/web/src/views/UserMemoryDetail/components/EndUserProfile.tsx index c689bf72..4dee9d4f 100644 --- a/web/src/views/UserMemoryDetail/components/EndUserProfile.tsx +++ b/web/src/views/UserMemoryDetail/components/EndUserProfile.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 18:33:30 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-27 11:11:09 + * @Last Modified time: 2026-04-14 16:03:41 */ /** * End User Profile Component @@ -27,11 +27,11 @@ import type { EndUser, EndUserProfileModalRef, EndUserProfileRef } from '../type * Component props */ interface EndUserProfileProps { - onDataLoaded?: (data: { other_name?: string; id: string }) => void; + onDataLoaded?: (data?: EndUser) => void; className?: string; } -const EndUserProfile = forwardRef(({ className }, ref) => { +const EndUserProfile = forwardRef(({ className, onDataLoaded }, ref) => { const { t } = useTranslation() const { id } = useParams() const endUserProfileModalRef = useRef(null) @@ -51,6 +51,7 @@ const EndUserProfile = forwardRef(({ cla const userData = res as EndUser setData(userData) setLoading(false) + onDataLoaded?.(userData as EndUser) }) .finally(() => { setLoading(false) diff --git a/web/src/views/UserMemoryDetail/types.ts b/web/src/views/UserMemoryDetail/types.ts index 9e56bb5d..d8bc6f23 100644 --- a/web/src/views/UserMemoryDetail/types.ts +++ b/web/src/views/UserMemoryDetail/types.ts @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 17:57:15 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-24 17:58:54 + * @Last Modified time: 2026-04-14 16:03:16 */ /** * User Memory Detail Types @@ -172,6 +172,7 @@ export interface EndUser { other_name: string; aliases: string | null; meta_data: Record; + id?: string; end_user_info_id: string; end_user_id: string; created_at: string; From e3265e4ba37982afabc8234060348412beedc994 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Tue, 14 Apr 2026 16:14:01 +0800 Subject: [PATCH 045/245] fix(http-request,embedding,naive): tighten form-data validation, reduce truncation length to 8000, and disable chunking for Excel The form-data validation now ensures all items in the list are of type HttpFormData. Truncation length for embedding inputs is reduced from 8191 to 8000 to accommodate tokenizer differences and avoid overflow. Excel parsing now disables chunking by setting chunk_token_num to 0, aligning with intended behavior for structured file ingestion. --- api/app/core/rag/app/naive.py | 2 +- api/app/core/rag/llm/embedding_model.py | 10 +++++++--- api/app/core/workflow/nodes/http_request/config.py | 3 ++- 3 files changed, 10 insertions(+), 5 deletions(-) diff --git a/api/app/core/rag/app/naive.py b/api/app/core/rag/app/naive.py index 72272347..93b96843 100644 --- a/api/app/core/rag/app/naive.py +++ b/api/app/core/rag/app/naive.py @@ -675,7 +675,7 @@ def chunk(filename, binary=None, from_page=0, to_page=100000, parser_config["chunk_token_num"] = 0 else: sections = [(_, "") for _ in excel_parser(binary) if _] - parser_config["chunk_token_num"] = 12800 + parser_config["chunk_token_num"] = 0 elif re.search(r"\.(txt|py|js|java|c|cpp|h|php|go|ts|sh|cs|kt|sql)$", filename, re.IGNORECASE): callback(0.1, "Start to parse.") diff --git a/api/app/core/rag/llm/embedding_model.py b/api/app/core/rag/llm/embedding_model.py index 22e35a15..59210054 100644 --- a/api/app/core/rag/llm/embedding_model.py +++ b/api/app/core/rag/llm/embedding_model.py @@ -50,7 +50,9 @@ class OpenAIEmbed(Base): def encode(self, texts: list): # OpenAI requires batch size <=16 batch_size = 16 - texts = [truncate(t, 8191) for t in texts] + # Use 8000 instead of 8191 to leave safety margin for tokenizer differences + # between cl100k_base (used by truncate) and the actual embedding model + texts = [truncate(t, 8000) for t in texts] ress = [] total_tokens = 0 for i in range(0, len(texts), batch_size): @@ -63,7 +65,7 @@ class OpenAIEmbed(Base): return np.array(ress), total_tokens def encode_queries(self, text): - res = self.client.embeddings.create(input=[truncate(text, 8191)], model=self.model_name, encoding_format="float",extra_body={"drop_params": True}) + res = self.client.embeddings.create(input=[truncate(text, 8000)], model=self.model_name, encoding_format="float",extra_body={"drop_params": True}) return np.array(res.data[0].embedding), self.total_token_count(res) @@ -79,6 +81,7 @@ class LocalAIEmbed(Base): def encode(self, texts: list): batch_size = 16 + texts = [truncate(t, 8000) for t in texts] ress = [] for i in range(0, len(texts), batch_size): res = self.client.embeddings.create(input=texts[i : i + batch_size], model=self.model_name) @@ -173,6 +176,7 @@ class XinferenceEmbed(Base): def encode(self, texts: list): batch_size = 16 + texts = [truncate(t, 8000) for t in texts] ress = [] total_tokens = 0 for i in range(0, len(texts), batch_size): @@ -188,7 +192,7 @@ class XinferenceEmbed(Base): def encode_queries(self, text): res = None try: - res = self.client.embeddings.create(input=[text], model=self.model_name) + res = self.client.embeddings.create(input=[truncate(text, 8000)], model=self.model_name) return np.array(res.data[0].embedding), self.total_token_count(res) except Exception as _e: log_exception(_e, res) diff --git a/api/app/core/workflow/nodes/http_request/config.py b/api/app/core/workflow/nodes/http_request/config.py index 3473f666..72474436 100644 --- a/api/app/core/workflow/nodes/http_request/config.py +++ b/api/app/core/workflow/nodes/http_request/config.py @@ -72,7 +72,8 @@ class HttpContentTypeConfig(BaseModel): @classmethod def validate_data(cls, v, info): content_type = info.data.get("content_type") - if content_type == HttpContentType.FROM_DATA and not isinstance(v, list): + if content_type == HttpContentType.FROM_DATA and ( + not isinstance(v, list) or not all(isinstance(item, HttpFormData) for item in v)): raise ValueError("When content_type is 'form-data', data must be a list of HttpFormData") elif content_type in [HttpContentType.JSON] and not isinstance(v, str): raise ValueError("When content_type is JSON, data must be of type str") From 4f0e5d08662c46c7e50862b2dc8b792bfd4f0d0f Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Tue, 14 Apr 2026 16:24:20 +0800 Subject: [PATCH 046/245] ci(wechat-notify): add Sourcery summary extraction with Qwen fallback - Extract Sourcery AI summary from PR body as primary source - Add fallback to Qwen AI summarization when Sourcery summary unavailable - Refactor notification payload to conditionally use Sourcery or Qwen summary - Update step conditions to skip Qwen processing when Sourcery summary found - Improve code formatting and indentation consistency in Python scripts - Reduce redundant file I/O by writing directly to GITHUB_OUTPUT --- .github/workflows/release-notify-wechat.yml | 88 +++++++++++++++------ 1 file changed, 63 insertions(+), 25 deletions(-) diff --git a/.github/workflows/release-notify-wechat.yml b/.github/workflows/release-notify-wechat.yml index d8f4e5aa..bc67518b 100644 --- a/.github/workflows/release-notify-wechat.yml +++ b/.github/workflows/release-notify-wechat.yml @@ -35,20 +35,52 @@ jobs: echo "ok=false" >> $GITHUB_OUTPUT fi - # 3️⃣ 获取 commits - - name: Get Commits + # 3️⃣ 尝试从 PR body 提取 Sourcery 摘要 + - name: Extract Sourcery Summary if: steps.check.outputs.ok == 'true' - id: commits + id: sourcery + env: + PR_BODY: ${{ github.event.pull_request.body }} + run: | + python3 << 'PYEOF' + import os, re + + body = os.environ.get("PR_BODY", "") or "" + match = re.search( + r"## Summary by Sourcery\s*\n(.*?)(?=\n## |\Z)", + body, + re.DOTALL + ) + + if match: + summary = match.group(1).strip() + found = "true" + else: + summary = "" + found = "false" + + with open("sourcery_summary.txt", "w", encoding="utf-8") as f: + f.write(summary) + + with open(os.environ["GITHUB_OUTPUT"], "a") as gh: + gh.write(f"found={found}\n") + gh.write("summary< commits.txt - # 4️⃣ 阿里 AI 总结（通义千问） - - name: AI Summary (Qwen) - if: steps.check.outputs.ok == 'true' - id: ai + - name: AI Summary (Qwen Fallback) + if: steps.check.outputs.ok == 'true' && steps.sourcery.outputs.found == 'false' + id: qwen env: DASHSCOPE_API_KEY: ${{ secrets.DASHSCOPE_API_KEY }} run: | @@ -56,34 +88,30 @@ jobs: import json, os, urllib.request with open("commits.txt", "r") as f: - commits = f.read().strip() + commits = f.read().strip() prompt = "请用中文总结以下代码提交，输出3-5条要点，面向测试人员。直接输出编号列表，不要输出标题或前言：\n" + commits payload = {"model": "qwen-plus", "input": {"prompt": prompt}} data = json.dumps(payload, ensure_ascii=False).encode("utf-8") req = urllib.request.Request( - "https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation", - data=data, - headers={ - "Authorization": "Bearer " + os.environ["DASHSCOPE_API_KEY"], - "Content-Type": "application/json" - } + "https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation", + data=data, + headers={ + "Authorization": "Bearer " + os.environ["DASHSCOPE_API_KEY"], + "Content-Type": "application/json" + } ) resp = urllib.request.urlopen(req) result = json.loads(resp.read().decode()) summary = result.get("output", {}).get("text", "AI 摘要生成失败") - print(summary) - with open("ai_summary.txt", "w", encoding="utf-8") as f: - f.write(summary) + with open(os.environ["GITHUB_OUTPUT"], "a") as gh: + gh.write("summary<> $GITHUB_OUTPUT - echo "$SUMMARY" >> $GITHUB_OUTPUT - echo "EOF" >> $GITHUB_OUTPUT - # 5️⃣ 企业微信通知（Markdown） - name: Notify WeChat if: steps.check.outputs.ok == 'true' @@ -93,17 +121,27 @@ jobs: AUTHOR: ${{ github.event.pull_request.user.login }} PR_TITLE: ${{ github.event.pull_request.title }} PR_URL: ${{ github.event.pull_request.html_url }} - AI_SUMMARY: ${{ steps.ai.outputs.summary }} + SOURCERY_FOUND: ${{ steps.sourcery.outputs.found }} + SOURCERY_SUMMARY: ${{ steps.sourcery.outputs.summary }} + QWEN_SUMMARY: ${{ steps.qwen.outputs.summary }} run: | python3 << 'PYEOF' import json, os, urllib.request + + if os.environ.get("SOURCERY_FOUND") == "true": + label = "Summary by Sourcery" + summary = os.environ.get("SOURCERY_SUMMARY", "") + else: + label = "AI变更摘要" + summary = os.environ.get("QWEN_SUMMARY", "AI 摘要生成失败") + content = ( "## 🚀 Release 发布通知\n" "> 📦 **分支**: " + os.environ["BRANCH"] + "\n" "> 👤 **提交人**: " + os.environ["AUTHOR"] + "\n" "> 📝 **标题**: " + os.environ["PR_TITLE"] + "\n\n" - "### 🧠 AI变更摘要\n" + - os.environ["AI_SUMMARY"] + "\n\n" + "### 🧠 " + label + "\n" + + summary + "\n\n" "---\n" "🔗 [查看PR详情](" + os.environ["PR_URL"] + ")" ) From 3c2a78a449c77892c6cda50752ce758c6431474d Mon Sep 17 00:00:00 2001 From: zhaoying Date: Tue, 14 Apr 2026 16:35:19 +0800 Subject: [PATCH 047/245] fix(web): Hide error message when workflow node error message equals empty string --- .../views/Workflow/components/Chat/Runtime.tsx | 17 ++++++++--------- 1 file changed, 8 insertions(+), 9 deletions(-) diff --git a/web/src/views/Workflow/components/Chat/Runtime.tsx b/web/src/views/Workflow/components/Chat/Runtime.tsx index 68bdc452..4a5be793 100644 --- a/web/src/views/Workflow/components/Chat/Runtime.tsx +++ b/web/src/views/Workflow/components/Chat/Runtime.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-24 17:57:08 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-07 14:05:50 + * @Last Modified time: 2026-04-14 16:33:33 */ /* * Runtime Component @@ -161,8 +161,7 @@ const Runtime: FC<{ item: ChatItem; index: number;}> = ({ children: ( {/* Display error message for failed nodes */} - - {item.error && + {vo.content?.error && vo.content?.error !== '' && @@ -219,11 +218,11 @@ const Runtime: FC<{ item: ChatItem; index: number;}> = ({
} - /** Copy value to clipboard and show success message */ - const handleCopy = (value: string) => { - copy(value) - message.success(t('common.copySuccess')) - } + /** Copy value to clipboard and show success message */ + const handleCopy = (value: string) => { + copy(value) + message.success(t('common.copySuccess')) + } return (
= ({
) :
- {item.error && + {item.error && item.error !== '' && } {renderChild(item.subContent)} From 1294aabbcc6260c5b3fe1a2a253371c75e8aa43b Mon Sep 17 00:00:00 2001 From: zhaoying Date: Tue, 14 Apr 2026 16:38:59 +0800 Subject: [PATCH 048/245] feat(web): update document title --- web/src/App.tsx | 55 ++++++++++++++++++++-- web/src/hooks/useBreadcrumbManager.ts | 10 ++-- web/src/routes/routes.json | 1 - web/src/store/menu.json | 15 +++--- web/src/views/ApplicationConfig/index.tsx | 11 ++++- web/src/views/MemoryManagement/index.tsx | 7 +-- web/src/views/Ontology/pages/Detail.tsx | 6 ++- web/src/views/Skills/pages/SkillConfig.tsx | 3 +- web/src/views/UserMemoryDetail/Neo4j.tsx | 17 ++++--- web/src/views/UserMemoryDetail/Rag.tsx | 6 ++- 10 files changed, 101 insertions(+), 30 deletions(-) diff --git a/web/src/App.tsx b/web/src/App.tsx index a10f9409..1af38372 100644 --- a/web/src/App.tsx +++ b/web/src/App.tsx @@ -16,7 +16,7 @@ import { ConfigProvider, App as AntdApp } from 'antd'; -import { useTranslation } from 'react-i18next'; +import i18n from 'i18next'; import { lightTheme } from './styles/antdThemeConfig.ts' import router from './routes'; @@ -29,11 +29,58 @@ import 'dayjs/plugin/utc' import { cookieUtils } from './utils/request'; import { useUser } from '@/store/user'; +import menuJson from '@/store/menu.json'; + +type MenuEntry = { path: string; i18nKey: string }; + +function flattenMenuEntries(list: any[]): MenuEntry[] { + const result: MenuEntry[] = []; + for (const item of list) { + if (item.path && item.i18nKey && item.type !== 'group') result.push({ path: item.path, i18nKey: item.i18nKey }); + if (item.subs?.length) result.push(...flattenMenuEntries(item.subs)); + } + return result; +} + +const menuEntries: MenuEntry[] = flattenMenuEntries([...menuJson.manage, ...menuJson.space]); + +function pathMatches(pattern: string, path: string): boolean { + if (pattern === path) return true; + if (pattern.includes(':')) { + return new RegExp('^' + pattern.replace(/:[\w-]+/g, '[^/]+') + '$').test(path); + } + return false; +} + +function getPageTitle(pathname: string): string { + const appName = i18n.t('memoryBear'); + const entry = menuEntries.find(e => pathMatches(e.path, pathname)); + if (!entry) return appName; + return `${i18n.t(entry.i18nKey)} - ${appName}`; +} + +const SKIP_TITLE_PATTERNS = [ + '/user-memory/detail/:id/:type', + '/forgetting-engine/:id', + '/memory-extraction-engine/:id', + '/emotion-engine/:id', + '/reflection-engine/:id', +]; + + + function App() { - const { t } = useTranslation(); const { locale, language, timeZone } = useI18n() const { checkJump } = useUser(); + useEffect(() => { + const unsubscribe = router.subscribe(({ location }) => { + if (SKIP_TITLE_PATTERNS.some(p => pathMatches(p, location.pathname))) return; + document.title = getPageTitle(location.pathname); + }); + return () => unsubscribe(); + }, []) + useEffect(() => { const authToken = cookieUtils.get('authToken') if (!authToken && !window.location.hash.includes('#/login') && !window.location.hash.includes('#/conversation/') && !window.location.hash.includes('#/jump') && !window.location.hash.includes('#/invite-register')) { @@ -44,7 +91,9 @@ function App() { }, []) useEffect(() => { - document.title = t('memoryBear') + if (!SKIP_TITLE_PATTERNS.some(p => pathMatches(p, router.state.location.pathname))) { + document.title = getPageTitle(router.state.location.pathname) + } dayjs.locale(language) localStorage.setItem('language', language) }, [language]) diff --git a/web/src/hooks/useBreadcrumbManager.ts b/web/src/hooks/useBreadcrumbManager.ts index 161fbb65..1b7cf4b2 100644 --- a/web/src/hooks/useBreadcrumbManager.ts +++ b/web/src/hooks/useBreadcrumbManager.ts @@ -1,8 +1,8 @@ /* * @Author: ZhaoYing * @Date: 2026-02-02 16:24:44 - * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-02-02 16:24:44 + * @Last Modified by: ZhaoYing + * @Last Modified time: 2026-04-14 15:52:57 */ /** * useBreadcrumbManager Hook @@ -18,6 +18,7 @@ import { useCallback } from 'react'; import { useNavigate } from 'react-router-dom'; +import { useTranslation } from 'react-i18next' import { useMenu } from '@/store/menu'; import type { MenuItem } from '@/store/menu'; @@ -53,6 +54,7 @@ export interface BreadcrumbOptions { export const useBreadcrumbManager = (options?: BreadcrumbOptions) => { const { allBreadcrumbs, setCustomBreadcrumbs } = useMenu(); const navigate = useNavigate(); + const { t } = useTranslation() /** Update breadcrumbs based on current path and type */ const updateBreadcrumbs = useCallback((breadcrumbPath: BreadcrumbPath) => { @@ -336,8 +338,8 @@ export const useBreadcrumbManager = (options?: BreadcrumbOptions) => { /** Use different keys based on breadcrumb type to implement independent breadcrumb paths */ const breadcrumbKey = breadcrumbType === 'list' ? 'space' : 'space-detail'; - - + const lastMenu = customBreadcrumbs[customBreadcrumbs.length - 1] + document.title = `${lastMenu.i18nKey ? t(lastMenu.i18nKey) : lastMenu.label} - ${t('memoryBear') }`; setCustomBreadcrumbs(customBreadcrumbs, breadcrumbKey); }, [setCustomBreadcrumbs, navigate, options?.breadcrumbType, options?.onKnowledgeBaseMenuClick, options?.onKnowledgeBaseFolderClick]); diff --git a/web/src/routes/routes.json b/web/src/routes/routes.json index 5ff1f90c..f6dc631d 100644 --- a/web/src/routes/routes.json +++ b/web/src/routes/routes.json @@ -48,7 +48,6 @@ { "path": "/application/config/:id", "element": "ApplicationConfig" }, { "path": "/application/config/:id/:source", "element": "ApplicationConfig" }, { "path": "/user-memory/neo4j/:id", "element": "Neo4jUserMemoryDetail" }, - { "path": "/statement/:id", "element": "StatementDetail" }, { "path": "/user-memory/detail/:id/:type", "element": "MemoryNodeDetail" }, { "path": "/ontology/:id", "element": "OntologyDetail" } ] diff --git a/web/src/store/menu.json b/web/src/store/menu.json index 8d30dcc4..ec80a384 100644 --- a/web/src/store/menu.json +++ b/web/src/store/menu.json @@ -6,7 +6,7 @@ "code": "workbench", "label": "workbench", "i18nKey": "menu.workbench", - "path": "/", + "path": null, "enable": true, "display": true, "level": 1, @@ -174,7 +174,7 @@ "code": "workbench", "label": "workbench", "i18nKey": "menu.workbench", - "path": "/", + "path": null, "enable": true, "display": true, "level": 1, @@ -425,15 +425,14 @@ { "id": 2211, "parent": 221, - "code": "statementDetail", + "code": "userMemoryDetail", "label": "记忆详情", - "i18nKey": "menu.statementDetail", - "path": "/statement/:id", + "i18nKey": "menu.userMemoryDetail", + "path": "/user-memory/detail/:id/:type", "enable": true, "display": false, - "level": 4, - "sort": 0, - "subs": null + "level": 3, + "sort": 0 } ] }, diff --git a/web/src/views/ApplicationConfig/index.tsx b/web/src/views/ApplicationConfig/index.tsx index a9e1df5d..fa6f9939 100644 --- a/web/src/views/ApplicationConfig/index.tsx +++ b/web/src/views/ApplicationConfig/index.tsx @@ -2,11 +2,12 @@ * @Author: ZhaoYing * @Date: 2026-02-03 16:29:37 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-26 15:37:18 + * @Last Modified time: 2026-04-14 10:01:05 */ import React, { useEffect, useState, useRef } from 'react'; import { useParams } from 'react-router-dom'; import { Flex } from 'antd' +import { useTranslation } from 'react-i18next' import ConfigHeader from './components/ConfigHeader' import type { AgentRef, ClusterRef, WorkflowRef, Config } from './types' @@ -30,6 +31,7 @@ import Logs from './Logs'; const ApplicationConfig: React.FC = () => { // Hooks const { id, source } = useParams(); + const { t } = useTranslation() // Refs for different application types const agentRef = useRef(null) @@ -95,6 +97,13 @@ const ApplicationConfig: React.FC = () => { getApplicationInfo() }, [id]) + useEffect(() => { + if (application?.name) { + const appName = t('memoryBear'); + document.title = `${application.name} - ${appName}`; + } + }, [application?.name]) + /** * Fetch application information */ diff --git a/web/src/views/MemoryManagement/index.tsx b/web/src/views/MemoryManagement/index.tsx index 1fb945b0..5bbbb8bd 100644 --- a/web/src/views/MemoryManagement/index.tsx +++ b/web/src/views/MemoryManagement/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 17:33:15 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-26 14:56:00 + * @Last Modified time: 2026-04-14 16:17:29 */ /** * Memory Management Page @@ -74,7 +74,8 @@ const MemoryManagement: React.FC = () => { }; /** Navigate to engine configuration page */ - const handleClick = (id: number, type: string) => { + const handleClick = (id: number, type: string, config_name: string) => { + document.title = `${config_name} - ${t('memoryBear')}`; switch (type) { case 'memoryExtractionEngine': navigate(`/memory-extraction-engine/${id}`) @@ -130,7 +131,7 @@ const MemoryManagement: React.FC = () => { align="center" justify="space-between" className="rb:cursor-pointer rb:bg-[#F6F6F6] rb:h-8 rb:rounded-lg rb:font-medium rb:leading-5 rb:pl-2! rb:pr-1! rb:hover:shadow-[0px_2px_8px_0px_rgba(23,23,25,0.16)]" - onClick={() => handleClick(item.config_id, key)} + onClick={() => handleClick(item.config_id, key, item.config_name)} > {t(`memory.${key}`)}
{ getData() }, [id, query]) + useEffect(() => { + document.title = `${data.scene_name} - ${t('memoryBear')}`; + }, [data.scene_name]) + /** * Fetch ontology class list data */ diff --git a/web/src/views/Skills/pages/SkillConfig.tsx b/web/src/views/Skills/pages/SkillConfig.tsx index 84c82378..91cd710a 100644 --- a/web/src/views/Skills/pages/SkillConfig.tsx +++ b/web/src/views/Skills/pages/SkillConfig.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-05 10:44:08 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-02-05 10:56:28 + * @Last Modified time: 2026-04-14 16:27:08 */ import { type FC, useEffect, useRef, useState } from "react"; import { useTranslation } from 'react-i18next'; @@ -71,6 +71,7 @@ const SkillConfig: FC = () => { .then(res => { form.setFieldsValue(res as SkillFormData) setData(res as SkillFormData) + document.title = `${(res as SkillFormData).name} - ${t('memoryBear')}`; }) .finally(() => { setLoading(false) diff --git a/web/src/views/UserMemoryDetail/Neo4j.tsx b/web/src/views/UserMemoryDetail/Neo4j.tsx index 51be7c8d..67ddc065 100644 --- a/web/src/views/UserMemoryDetail/Neo4j.tsx +++ b/web/src/views/UserMemoryDetail/Neo4j.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 17:57:26 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-26 18:59:53 + * @Last Modified time: 2026-04-14 16:38:21 */ /** * Neo4j User Memory Detail View @@ -22,7 +22,7 @@ import InterestDistribution from './components/InterestDistribution' import NodeStatistics from './components/NodeStatistics' import RelationshipNetwork from './components/RelationshipNetwork' import MemoryInsight from './components/MemoryInsight' -import type { EndUserProfileRef, MemoryInsightRef, AboutMeRef } from './types' +import type { EndUserProfileRef, MemoryInsightRef, AboutMeRef, EndUser } from './types' import { analyticsRefresh, } from '@/api/memory' @@ -39,8 +39,11 @@ const Neo4j: FC = () => { const [selectedKey, setSelectedKey] = useState(null) /** Update displayed name */ - const handleNameUpdate = (data: { other_name?: string; id: string }) => { - setName(data.other_name && data.other_name !== '' ? data.other_name : data.id) + const handleNameUpdate = (data?: EndUser) => { + if (!data) return + let name = data.other_name && data.other_name !== '' ? data.other_name : data.id || data.end_user_id + setName(name) + document.title = `${name} - ${t('memoryBear')}`; } /** Navigate back */ @@ -78,7 +81,7 @@ const Neo4j: FC = () => { - {
- {
- + diff --git a/web/src/views/UserMemoryDetail/Rag.tsx b/web/src/views/UserMemoryDetail/Rag.tsx index ff9069c7..a1da95d4 100644 --- a/web/src/views/UserMemoryDetail/Rag.tsx +++ b/web/src/views/UserMemoryDetail/Rag.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 17:57:11 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-31 15:29:45 + * @Last Modified time: 2026-04-14 15:56:15 */ /** * RAG User Memory Detail View @@ -97,6 +97,10 @@ const Rag: FC = () => { } const name = loading.detail ? '' : data?.name && data?.name !== '' ? data.name : id + useEffect(() => { + document.title = `${name} - ${t('memoryBear')}`; + }, [name]) + const [refreshLoading, setRefreshLoading] = useState(false) const handleRefresh = () => { if (refreshLoading || !id) return From 5e6490213d6b6b72624f6a986e3c7012d2d0b7fc Mon Sep 17 00:00:00 2001 From: zhaoying Date: Tue, 14 Apr 2026 17:03:22 +0800 Subject: [PATCH 049/245] fix(web): document title support i18n --- web/src/hooks/useBreadcrumbManager.ts | 6 ++++-- web/src/views/ApplicationConfig/index.tsx | 6 ++++-- web/src/views/EmotionEngine/index.tsx | 8 +++++++- web/src/views/ForgettingEngine/index.tsx | 9 ++++++++- .../views/MemoryExtractionEngine/components/Result.tsx | 1 + web/src/views/MemoryExtractionEngine/index.tsx | 8 +++++++- web/src/views/Ontology/pages/Detail.tsx | 6 ++++-- web/src/views/SelfReflectionEngine/index.tsx | 7 ++++++- web/src/views/Skills/pages/SkillConfig.tsx | 10 ++++++++-- web/src/views/UserMemoryDetail/Neo4j.tsx | 10 +++++++--- web/src/views/UserMemoryDetail/Rag.tsx | 6 ++++-- 11 files changed, 60 insertions(+), 17 deletions(-) diff --git a/web/src/hooks/useBreadcrumbManager.ts b/web/src/hooks/useBreadcrumbManager.ts index 1b7cf4b2..e2567cfd 100644 --- a/web/src/hooks/useBreadcrumbManager.ts +++ b/web/src/hooks/useBreadcrumbManager.ts @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-02 16:24:44 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-14 15:52:57 + * @Last Modified time: 2026-04-14 16:52:43 */ /** * useBreadcrumbManager Hook @@ -21,6 +21,7 @@ import { useNavigate } from 'react-router-dom'; import { useTranslation } from 'react-i18next' import { useMenu } from '@/store/menu'; import type { MenuItem } from '@/store/menu'; +import { useI18n } from '@/store/locale' /** Breadcrumb item interface */ export interface BreadcrumbItem { @@ -55,6 +56,7 @@ export const useBreadcrumbManager = (options?: BreadcrumbOptions) => { const { allBreadcrumbs, setCustomBreadcrumbs } = useMenu(); const navigate = useNavigate(); const { t } = useTranslation() + const { language } = useI18n() /** Update breadcrumbs based on current path and type */ const updateBreadcrumbs = useCallback((breadcrumbPath: BreadcrumbPath) => { @@ -341,7 +343,7 @@ export const useBreadcrumbManager = (options?: BreadcrumbOptions) => { const lastMenu = customBreadcrumbs[customBreadcrumbs.length - 1] document.title = `${lastMenu.i18nKey ? t(lastMenu.i18nKey) : lastMenu.label} - ${t('memoryBear') }`; setCustomBreadcrumbs(customBreadcrumbs, breadcrumbKey); - }, [setCustomBreadcrumbs, navigate, options?.breadcrumbType, options?.onKnowledgeBaseMenuClick, options?.onKnowledgeBaseFolderClick]); + }, [setCustomBreadcrumbs, navigate, options?.breadcrumbType, options?.onKnowledgeBaseMenuClick, options?.onKnowledgeBaseFolderClick, language]); return { updateBreadcrumbs, diff --git a/web/src/views/ApplicationConfig/index.tsx b/web/src/views/ApplicationConfig/index.tsx index fa6f9939..5b9bcb85 100644 --- a/web/src/views/ApplicationConfig/index.tsx +++ b/web/src/views/ApplicationConfig/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 16:29:37 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-14 10:01:05 + * @Last Modified time: 2026-04-14 16:53:27 */ import React, { useEffect, useState, useRef } from 'react'; import { useParams } from 'react-router-dom'; @@ -22,6 +22,7 @@ import Statistics from './Statistics' import TestChat from './TestChat' import type { WorkflowConfig } from '@/views/Workflow/types'; import Logs from './Logs'; +import { useI18n } from '@/store/locale' /** * Application configuration page component @@ -32,6 +33,7 @@ const ApplicationConfig: React.FC = () => { // Hooks const { id, source } = useParams(); const { t } = useTranslation() + const { language } = useI18n() // Refs for different application types const agentRef = useRef(null) @@ -102,7 +104,7 @@ const ApplicationConfig: React.FC = () => { const appName = t('memoryBear'); document.title = `${application.name} - ${appName}`; } - }, [application?.name]) + }, [application?.name, language]) /** * Fetch application information diff --git a/web/src/views/EmotionEngine/index.tsx b/web/src/views/EmotionEngine/index.tsx index 8c0188da..28933c32 100644 --- a/web/src/views/EmotionEngine/index.tsx +++ b/web/src/views/EmotionEngine/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 16:56:54 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-26 15:43:29 + * @Last Modified time: 2026-04-14 16:59:16 */ /** * Emotion Engine Configuration Page @@ -25,6 +25,7 @@ import DescWrapper from '@/components/FormItem/DescWrapper' import RbSlider from '@/components/RbSlider'; import RbAlert from '@/components/RbAlert'; import ModelSelect from '@/components/ModelSelect'; +import { useI18n } from '@/store/locale' /** * Configuration field definitions @@ -69,9 +70,14 @@ const EmotionEngine: React.FC = () => { const [form] = Form.useForm(); const { message: messageApi } = App.useApp(); const [loading, setLoading] = useState(false) + const { language } = useI18n() const values = Form.useWatch([], form); + useEffect(() => { + document.title = [document.title.split(' - ')[0], t('memoryBear')].join(' - ') + }, [language]) + useEffect(() => { getConfigData() }, [id]) diff --git a/web/src/views/ForgettingEngine/index.tsx b/web/src/views/ForgettingEngine/index.tsx index 0b15867e..bcd7dbc1 100644 --- a/web/src/views/ForgettingEngine/index.tsx +++ b/web/src/views/ForgettingEngine/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 17:00:12 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-26 15:47:37 + * @Last Modified time: 2026-04-14 16:54:38 */ /** * Forgetting Engine Configuration Page @@ -22,6 +22,7 @@ import type { ConfigForm } from './types' import SwitchFormItem from '@/components/FormItem/SwitchFormItem' import RbSlider from '@/components/RbSlider'; import DescWrapper from '@/components/FormItem/DescWrapper' +import { useI18n } from '@/store/locale' /** * Configuration field definitions @@ -109,9 +110,14 @@ const ForgettingEngine: React.FC = () => { const [form] = Form.useForm(); const { message: messageApi } = App.useApp(); const [loading, setLoading] = useState(false) + const { language } = useI18n() const values = Form.useWatch([], form); + useEffect(() => { + document.title = [document.title.split(' - ')[0], t('memoryBear')].join(' - ') + }, [language]) + useEffect(() => { getConfigData() }, []) @@ -182,6 +188,7 @@ const ForgettingEngine: React.FC = () => { if (config.type === 'button') { return ( {t(`forgettingEngine.type`)}: {config.type}} diff --git a/web/src/views/MemoryExtractionEngine/components/Result.tsx b/web/src/views/MemoryExtractionEngine/components/Result.tsx index 4d07aae9..2fa8788f 100644 --- a/web/src/views/MemoryExtractionEngine/components/Result.tsx +++ b/web/src/views/MemoryExtractionEngine/components/Result.tsx @@ -328,6 +328,7 @@ const Result: FC = ({ loading, handleSave }) => { {['processData', 'finalResult'].map(tab => (
{ const { t } = useTranslation(); const { message } = App.useApp(); const { id } = useParams() + const { language } = useI18n() const [expandedKeys, setExpandedKeys] = useState(keys) const [form] = Form.useForm() const values = Form.useWatch([], form) const [loading, setLoading] = useState(false) const [iterationPeriodDisabled, setIterationPeriodDisabled] = useState(false) + useEffect(() => { + document.title = [document.title.split(' - ')[0], t('memoryBear')].join(' - ') + }, [language]) + useEffect(() => { if (values?.reflexion_range === 'database') { form.setFieldValue('iteration_period', 24) diff --git a/web/src/views/Ontology/pages/Detail.tsx b/web/src/views/Ontology/pages/Detail.tsx index 80ce3a0f..b6ee5953 100644 --- a/web/src/views/Ontology/pages/Detail.tsx +++ b/web/src/views/Ontology/pages/Detail.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 14:10:20 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-14 15:54:05 + * @Last Modified time: 2026-04-14 16:55:26 */ import { type FC, useEffect, useState, useRef } from 'react' import { useParams, useNavigate } from 'react-router-dom'; @@ -18,6 +18,7 @@ import SearchInput from '@/components/SearchInput'; import OntologyClassExtractModal from '../components/OntologyClassExtractModal' import BodyWrapper from '@/components/Empty/BodyWrapper' import Tag from '@/components/Tag' +import { useI18n } from '@/store/locale' /** * Ontology detail page component @@ -29,6 +30,7 @@ const Detail: FC = () => { const navigate = useNavigate() const { id } = useParams() const { modal, message } = App.useApp() + const { language } = useI18n() // Refs const ontologyClassModalRef = useRef(null) @@ -48,7 +50,7 @@ const Detail: FC = () => { useEffect(() => { document.title = `${data.scene_name} - ${t('memoryBear')}`; - }, [data.scene_name]) + }, [data.scene_name, language]) /** * Fetch ontology class list data diff --git a/web/src/views/SelfReflectionEngine/index.tsx b/web/src/views/SelfReflectionEngine/index.tsx index bd1470c1..ef5e412c 100644 --- a/web/src/views/SelfReflectionEngine/index.tsx +++ b/web/src/views/SelfReflectionEngine/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 17:46:47 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-26 18:57:08 + * @Last Modified time: 2026-04-14 16:59:56 */ /** * Self Reflection Engine Configuration Page @@ -99,6 +99,10 @@ const SelfReflectionEngine: React.FC = () => { const values = Form.useWatch([], form); + useEffect(() => { + document.title = [document.title.split(' - ')[0], t('memoryBear')].join(' - ') + }, [language]) + useEffect(() => { getConfigData() }, [id]) @@ -242,6 +246,7 @@ const SelfReflectionEngine: React.FC = () => { return ( diff --git a/web/src/views/Skills/pages/SkillConfig.tsx b/web/src/views/Skills/pages/SkillConfig.tsx index 91cd710a..440356fd 100644 --- a/web/src/views/Skills/pages/SkillConfig.tsx +++ b/web/src/views/Skills/pages/SkillConfig.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-05 10:44:08 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-14 16:27:08 + * @Last Modified time: 2026-04-14 16:57:52 */ import { type FC, useEffect, useRef, useState } from "react"; import { useTranslation } from 'react-i18next'; @@ -17,6 +17,7 @@ import type { SkillFormData } from '../types' import { getSkillDetail, createSkill, updateSkill } from '@/api/skill' import { stringRegExp } from '@/utils/validator'; import PageHeader from '@/components/Layout/PageHeader' +import { useI18n } from '@/store/locale' /** * Skill Configuration Page Component @@ -43,6 +44,7 @@ const SkillConfig: FC = () => { const [loading, setLoading] = useState(false) const [form] = Form.useForm(); const [data, setData] = useState(null) + const { language } = useI18n() /** * Effect: Load skill data if editing existing skill @@ -71,12 +73,16 @@ const SkillConfig: FC = () => { .then(res => { form.setFieldsValue(res as SkillFormData) setData(res as SkillFormData) - document.title = `${(res as SkillFormData).name} - ${t('memoryBear')}`; }) .finally(() => { setLoading(false) }) } + + useEffect(() => { + if (!data) return; + document.title = `${data?.name} - ${t('memoryBear')}`; + }, [language, data?.name]) const aiPromptModalRef = useRef(null) diff --git a/web/src/views/UserMemoryDetail/Neo4j.tsx b/web/src/views/UserMemoryDetail/Neo4j.tsx index 67ddc065..4de310b3 100644 --- a/web/src/views/UserMemoryDetail/Neo4j.tsx +++ b/web/src/views/UserMemoryDetail/Neo4j.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 17:57:26 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-14 16:38:21 + * @Last Modified time: 2026-04-14 16:57:59 */ /** * Neo4j User Memory Detail View @@ -10,7 +10,7 @@ * Shows profile, interests, node statistics, relationships, and insights */ -import { type FC, useRef, useState, type MouseEvent } from 'react' +import { type FC, useRef, useState, type MouseEvent, useEffect } from 'react' import clsx from 'clsx' import { useParams, useNavigate } from 'react-router-dom' import { Flex, Popover } from 'antd' @@ -26,6 +26,7 @@ import type { EndUserProfileRef, MemoryInsightRef, AboutMeRef, EndUser } from '. import { analyticsRefresh, } from '@/api/memory' +import { useI18n } from '@/store/locale' const Neo4j: FC = () => { const { id } = useParams() @@ -33,6 +34,7 @@ const Neo4j: FC = () => { const navigate = useNavigate(); const [loading, setLoading] = useState(false) const [name, setName] = useState('') + const { language } = useI18n() const ref = useRef(null) const memoryInsightRef = useRef(null) const aboutMeRef = useRef(null) @@ -43,8 +45,10 @@ const Neo4j: FC = () => { if (!data) return let name = data.other_name && data.other_name !== '' ? data.other_name : data.id || data.end_user_id setName(name) - document.title = `${name} - ${t('memoryBear')}`; } + useEffect(() => { + document.title = `${name} - ${t('memoryBear')}`; + }, [name, language]) /** Navigate back */ const goBack = () => { diff --git a/web/src/views/UserMemoryDetail/Rag.tsx b/web/src/views/UserMemoryDetail/Rag.tsx index a1da95d4..07fb0af6 100644 --- a/web/src/views/UserMemoryDetail/Rag.tsx +++ b/web/src/views/UserMemoryDetail/Rag.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 17:57:11 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-14 15:56:15 + * @Last Modified time: 2026-04-14 16:56:36 */ /** * RAG User Memory Detail View @@ -26,6 +26,7 @@ import { } from '@/api/memory' import Empty from '@/components/Empty' import ConversationMemory from './components/ConversationMemory' +import { useI18n } from '@/store/locale' /** * Title component props @@ -45,6 +46,7 @@ const Title: FC = ({ title, iconClassName }) => ( const Rag: FC = () => { const { t } = useTranslation() const { id } = useParams() + const { language } = useI18n() const [data, setData] = useState(null) const [summary, setSummary] = useState('') const [loading, setLoading] = useState>({ @@ -99,7 +101,7 @@ const Rag: FC = () => { useEffect(() => { document.title = `${name} - ${t('memoryBear')}`; - }, [name]) + }, [name, language]) const [refreshLoading, setRefreshLoading] = useState(false) const handleRefresh = () => { From 75e95bab01239a465cb4cb123e07d9debda73b68 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Tue, 14 Apr 2026 17:10:52 +0800 Subject: [PATCH 050/245] refactor(rag): simplify Excel parsing logic and remove redundant chunk_token_num assignment --- api/app/core/rag/app/naive.py | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/api/app/core/rag/app/naive.py b/api/app/core/rag/app/naive.py index 93b96843..312216dd 100644 --- a/api/app/core/rag/app/naive.py +++ b/api/app/core/rag/app/naive.py @@ -672,10 +672,15 @@ def chunk(filename, binary=None, from_page=0, to_page=100000, excel_parser = ExcelParser() if parser_config.get("html4excel") and parser_config.get("html4excel").lower() == "true": sections = [(_, "") for _ in excel_parser.html(binary, 12) if _] - parser_config["chunk_token_num"] = 0 else: sections = [(_, "") for _ in excel_parser(binary) if _] - parser_config["chunk_token_num"] = 0 + callback(0.8, "Finish parsing.") + # Excel 每行直接作为一个 chunk，不经过 naive_merge 避免被 delimiter 拆分 + chunks = [s for s, _ in sections] + res.extend(tokenize_chunks(chunks, doc, is_english, None)) + res.extend(embed_res) + res.extend(url_res) + return res elif re.search(r"\.(txt|py|js|java|c|cpp|h|php|go|ts|sh|cs|kt|sql)$", filename, re.IGNORECASE): callback(0.1, "Start to parse.") From 811193dd756295c9ac89d3d54c3b03dac131c840 Mon Sep 17 00:00:00 2001 From: lanceyq <1982376970@qq.com> Date: Tue, 14 Apr 2026 17:28:24 +0800 Subject: [PATCH 051/245] fix(memory): make PgSQL the single source of truth for user entity aliases MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit - Skip alias merging for user entities during dedup (_merge_attribute and _merge_entities_with_aliases) to prevent dirty data from overwriting PgSQL authoritative aliases - Add PgSQL→Neo4j alias sync after Neo4j write in write_tools to ensure Neo4j user entities always reflect the PgSQL source - Remove deduped_aliases (Neo4j history) from alias sync in extraction_orchestrator, only append newly extracted aliases to PgSQL - Guard Neo4j MERGE cypher to preserve existing aliases for user entities (name IN ['用户','我','User','I']) - Fix emotion_analytics_service query to use ExtractedEntity label and entity_type property --- .../core/memory/agent/utils/write_tools.py | 29 ++++- .../deduplication/deduped_and_disamb.py | 109 +++++------------- .../extraction_orchestrator.py | 44 ++----- api/app/repositories/neo4j/cypher_queries.py | 2 + api/app/services/emotion_analytics_service.py | 4 +- 5 files changed, 71 insertions(+), 117 deletions(-) diff --git a/api/app/core/memory/agent/utils/write_tools.py b/api/app/core/memory/agent/utils/write_tools.py index bae4643e..5e51beba 100644 --- a/api/app/core/memory/agent/utils/write_tools.py +++ b/api/app/core/memory/agent/utils/write_tools.py @@ -191,15 +191,34 @@ async def write( if success: logger.info("Successfully saved all data to Neo4j") - # 使用 Celery 异步任务触发聚类（不阻塞主流程） if all_entity_nodes: + end_user_id = all_entity_nodes[0].end_user_id + + # Neo4j 写入完成后，用 PgSQL 权威 aliases 覆盖 Neo4j 用户实体 + try: + from app.repositories.end_user_info_repository import EndUserInfoRepository + if end_user_id: + with get_db_context() as db_session: + info = EndUserInfoRepository(db_session).get_by_end_user_id(uuid.UUID(end_user_id)) + pg_aliases = info.aliases if info and info.aliases else [] + if pg_aliases: + await neo4j_connector.execute_query( + """ + MATCH (e:ExtractedEntity) + WHERE e.end_user_id = $end_user_id AND e.name IN ['用户', '我', 'User', 'I'] + SET e.aliases = $aliases + """, + end_user_id=end_user_id, aliases=pg_aliases, + ) + logger.info(f"[AliasSync] Neo4j 用户实体 aliases 已用 PgSQL 权威源覆盖: {pg_aliases}") + except Exception as sync_err: + logger.warning(f"[AliasSync] PgSQL→Neo4j aliases 同步失败（不影响主流程）: {sync_err}") + + # 使用 Celery 异步任务触发聚类（不阻塞主流程） try: from app.tasks import run_incremental_clustering - end_user_id = all_entity_nodes[0].end_user_id new_entity_ids = [e.id for e in all_entity_nodes] - - # 异步提交 Celery 任务 task = run_incremental_clustering.apply_async( kwargs={ "end_user_id": end_user_id, @@ -207,7 +226,6 @@ async def write( "llm_model_id": str(memory_config.llm_model_id) if memory_config.llm_model_id else None, "embedding_model_id": str(memory_config.embedding_model_id) if memory_config.embedding_model_id else None, }, - # 设置任务优先级（低优先级，不影响主业务） priority=3, ) logger.info( @@ -215,7 +233,6 @@ async def write( f"task_id={task.id}, end_user_id={end_user_id}, entity_count={len(new_entity_ids)}" ) except Exception as e: - # 聚类任务提交失败不影响主流程 logger.error(f"[Clustering] 提交聚类任务失败（不影响主流程）: {e}", exc_info=True) break diff --git a/api/app/core/memory/storage_services/extraction_engine/deduplication/deduped_and_disamb.py b/api/app/core/memory/storage_services/extraction_engine/deduplication/deduped_and_disamb.py index 7e0976fe..8f659a27 100644 --- a/api/app/core/memory/storage_services/extraction_engine/deduplication/deduped_and_disamb.py +++ b/api/app/core/memory/storage_services/extraction_engine/deduplication/deduped_and_disamb.py @@ -82,51 +82,35 @@ def _merge_attribute(canonical: ExtractedEntityNode, ent: ExtractedEntityNode): canonical.connect_strength = next(iter(pair)) # 别名合并（去重保序，使用标准化工具） + # 用户实体的 aliases 由 PgSQL end_user_info 作为唯一权威源，去重合并时不修改 try: canonical_name = (getattr(canonical, "name", "") or "").strip() - incoming_name = (getattr(ent, "name", "") or "").strip() - - # 收集所有需要合并的别名 - all_aliases = [] - - # 1. 添加canonical现有的别名 - existing = getattr(canonical, "aliases", []) or [] - all_aliases.extend(existing) - - # 2. 添加incoming实体的名称（如果不同于canonical的名称） - if incoming_name and incoming_name != canonical_name: - all_aliases.append(incoming_name) - - # 3. 添加incoming实体的所有别名 - incoming = getattr(ent, "aliases", []) or [] - all_aliases.extend(incoming) - - # 4. 标准化并去重（优先使用alias_utils工具函数） - try: - from app.core.memory.utils.alias_utils import normalize_aliases - canonical.aliases = normalize_aliases(canonical_name, all_aliases) - except Exception: - # 如果导入失败，使用增强的去重逻辑 - seen_normalized = set() - unique_aliases = [] + if canonical_name.lower() not in _USER_PLACEHOLDER_NAMES: + incoming_name = (getattr(ent, "name", "") or "").strip() - for alias in all_aliases: - if not alias: - continue - - alias_stripped = str(alias).strip() - if not alias_stripped or alias_stripped == canonical_name: - continue - - # 标准化：转小写用于去重判断 - alias_normalized = alias_stripped.lower() - - if alias_normalized not in seen_normalized: - seen_normalized.add(alias_normalized) - unique_aliases.append(alias_stripped) + # 收集所有需要合并的别名 + all_aliases = list(getattr(canonical, "aliases", []) or []) + if incoming_name and incoming_name != canonical_name: + all_aliases.append(incoming_name) + all_aliases.extend(getattr(ent, "aliases", []) or []) - # 排序并赋值 - canonical.aliases = sorted(unique_aliases) + try: + from app.core.memory.utils.alias_utils import normalize_aliases + canonical.aliases = normalize_aliases(canonical_name, all_aliases) + except Exception: + seen_normalized = set() + unique_aliases = [] + for alias in all_aliases: + if not alias: + continue + alias_stripped = str(alias).strip() + if not alias_stripped or alias_stripped == canonical_name: + continue + alias_normalized = alias_stripped.lower() + if alias_normalized not in seen_normalized: + seen_normalized.add(alias_normalized) + unique_aliases.append(alias_stripped) + canonical.aliases = sorted(unique_aliases) except Exception: pass @@ -733,66 +717,37 @@ def fuzzy_match( def _merge_entities_with_aliases(canonical: ExtractedEntityNode, losing: ExtractedEntityNode): - """ 模糊匹配中的实体合并。 + """模糊匹配中的实体合并（别名部分）。 - 合并策略： - 1. 保留canonical的主名称不变 - 2. 将losing的主名称添加为alias（如果不同） - 3. 合并两个实体的所有aliases - 4. 自动去重（case-insensitive）并排序 - - Args: - canonical: 规范实体（保留） - losing: 被合并实体（删除） - - Note: - 使用alias_utils.normalize_aliases进行标准化去重 + 用户实体的 aliases 由 PgSQL end_user_info 作为唯一权威源，跳过合并。 """ - # 获取规范实体的名称 canonical_name = (getattr(canonical, "name", "") or "").strip() + if canonical_name.lower() in _USER_PLACEHOLDER_NAMES: + return + losing_name = (getattr(losing, "name", "") or "").strip() - # 收集所有需要合并的别名 - all_aliases = [] - - # 1. 添加canonical现有的别名 - current_aliases = getattr(canonical, "aliases", []) or [] - all_aliases.extend(current_aliases) - - # 2. 添加losing实体的名称（如果不同于canonical的名称） + all_aliases = list(getattr(canonical, "aliases", []) or []) if losing_name and losing_name != canonical_name: all_aliases.append(losing_name) + all_aliases.extend(getattr(losing, "aliases", []) or []) - # 3. 添加losing实体的所有别名 - losing_aliases = getattr(losing, "aliases", []) or [] - all_aliases.extend(losing_aliases) - - # 4. 标准化并去重（使用标准化后的字符串进行去重） try: from app.core.memory.utils.alias_utils import normalize_aliases canonical.aliases = normalize_aliases(canonical_name, all_aliases) except Exception: - # 如果导入失败，使用增强的去重逻辑 - # 使用标准化后的字符串作为key进行去重 seen_normalized = set() unique_aliases = [] - for alias in all_aliases: if not alias: continue - alias_stripped = str(alias).strip() if not alias_stripped or alias_stripped == canonical_name: continue - - # 标准化：转小写用于去重判断 alias_normalized = alias_stripped.lower() - if alias_normalized not in seen_normalized: seen_normalized.add(alias_normalized) unique_aliases.append(alias_stripped) - - # 排序并赋值 canonical.aliases = sorted(unique_aliases) # ========== 主循环：遍历所有实体对进行模糊匹配 ========== diff --git a/api/app/core/memory/storage_services/extraction_engine/extraction_orchestrator.py b/api/app/core/memory/storage_services/extraction_engine/extraction_orchestrator.py index 5636dcb5..75fc87d2 100644 --- a/api/app/core/memory/storage_services/extraction_engine/extraction_orchestrator.py +++ b/api/app/core/memory/storage_services/extraction_engine/extraction_orchestrator.py @@ -1391,18 +1391,18 @@ class ExtractionOrchestrator: """ 将本轮提取的用户别名同步到 end_user 和 end_user_info 表。 - 注意：此方法在 Neo4j 写入之前调用，因此不能依赖 Neo4j 作为别名的权威数据源。 - 改为直接使用内存中去重后的 entity_nodes 的 aliases，与 PgSQL 已有的 aliases 合并。 + PgSQL end_user_info.aliases 是用户别名的唯一权威源。 + 此方法仅将本轮 LLM 从对话中新提取的别名增量追加到 PgSQL， + 不再从 Neo4j 二层去重合并历史别名，避免脏数据反向污染 PgSQL。 策略： - 1. 从内存中的 entity_nodes 提取本轮用户别名（current_aliases） - 2. 从去重后的 entity_nodes 中提取完整别名（含 Neo4j 二层去重合并的历史别名） - 3. 从 PgSQL end_user_info 读取已有的 aliases（db_aliases） - 4. 合并 db_aliases + deduped_aliases + current_aliases，去重保序 - 5. 写回 PgSQL + 1. 从本轮对话原始发言中提取用户别名（current_aliases） + 2. 从 PgSQL end_user_info 读取已有的 aliases（db_aliases） + 3. 合并 db_aliases + current_aliases，去重保序 + 4. 写回 PgSQL Args: - entity_nodes: 去重后的实体节点列表（内存中，含二层去重合并结果） + entity_nodes: 去重后的实体节点列表（内存中） dialog_data_list: 对话数据列表 """ try: @@ -1418,11 +1418,6 @@ class ExtractionOrchestrator: # 1. 提取本轮对话的用户别名（保持 LLM 提取的原始顺序，不排序） current_aliases = self._extract_current_aliases(entity_nodes, dialog_data_list) - # 1.5 从去重后的 entity_nodes 中提取完整别名 - # 二层去重会将 Neo4j 中已有的历史别名合并到 entity_nodes 中， - # 这里提取出来确保 PgSQL 与 Neo4j 的别名保持同步 - deduped_aliases = self._extract_deduped_entity_aliases(entity_nodes) - # 1.6 从 Neo4j 查询已有的 AI 助手别名，作为额外的排除源 # （防止 LLM 未提取出 AI 助手实体时，AI 别名泄漏到用户别名中） neo4j_assistant_aliases = await self._fetch_neo4j_assistant_aliases(end_user_id) @@ -1434,19 +1429,12 @@ class ExtractionOrchestrator: ] if len(current_aliases) < before_count: logger.info(f"通过 Neo4j AI 助手别名排除了 {before_count - len(current_aliases)} 个误归属别名") - # 同样过滤 deduped_aliases - deduped_aliases = [ - a for a in deduped_aliases - if a.strip().lower() not in neo4j_assistant_aliases - ] - if not current_aliases and not deduped_aliases: + if not current_aliases: logger.debug(f"本轮未提取到用户别名，跳过同步: end_user_id={end_user_id}") return logger.info(f"本轮对话提取的 aliases: {current_aliases}") - if deduped_aliases: - logger.info(f"去重后实体的完整 aliases（含历史）: {deduped_aliases}") # 2. 同步到数据库 end_user_uuid = uuid.UUID(end_user_id) @@ -1457,21 +1445,15 @@ class ExtractionOrchestrator: logger.warning(f"未找到 end_user_id={end_user_id} 的用户记录") return - # 3. 从 PgSQL 读取已有 aliases 并与本轮合并 + # 3. 从 PgSQL 读取已有 aliases 并与本轮新增合并 info = EndUserInfoRepository(db).get_by_end_user_id(end_user_uuid) db_aliases = (info.aliases if info and info.aliases else []) # 过滤掉占位名称 db_aliases = [a for a in db_aliases if a.strip().lower() not in self.USER_PLACEHOLDER_NAMES] - # 合并：已有 + 去重后完整别名 + 本轮新增，去重保序 + # 合并：PgSQL 已有 + 本轮新增，去重保序（不再合并 Neo4j 历史别名） merged_aliases = list(db_aliases) seen_lower = {a.strip().lower() for a in merged_aliases} - # 先合并去重后实体的完整别名（含 Neo4j 历史别名） - for alias in deduped_aliases: - if alias.strip().lower() not in seen_lower: - merged_aliases.append(alias) - seen_lower.add(alias.strip().lower()) - # 再合并本轮新提取的别名 for alias in current_aliases: if alias.strip().lower() not in seen_lower: merged_aliases.append(alias) @@ -1505,9 +1487,7 @@ class ExtractionOrchestrator: info.aliases = merged_aliases logger.info(f"同步合并后 aliases 到 end_user_info: {merged_aliases}") else: - first_alias = current_aliases[0].strip() if current_aliases else ( - deduped_aliases[0].strip() if deduped_aliases else "" - ) + first_alias = current_aliases[0].strip() if current_aliases else "" # 确保 first_alias 不是占位名称 if first_alias and first_alias.lower() not in self.USER_PLACEHOLDER_NAMES: db.add(EndUserInfo( diff --git a/api/app/repositories/neo4j/cypher_queries.py b/api/app/repositories/neo4j/cypher_queries.py index 4b5273ac..daf04bcb 100644 --- a/api/app/repositories/neo4j/cypher_queries.py +++ b/api/app/repositories/neo4j/cypher_queries.py @@ -93,6 +93,8 @@ SET e.name = CASE WHEN entity.name IS NOT NULL AND entity.name <> '' THEN entity END, e.statement_id = CASE WHEN entity.statement_id IS NOT NULL AND entity.statement_id <> '' THEN entity.statement_id ELSE e.statement_id END, e.aliases = CASE + // 用户实体的 aliases 由 PgSQL end_user_info 作为唯一权威源，知识抽取完全不写入 + WHEN entity.name IN ['用户', '我', 'User', 'I'] THEN e.aliases WHEN entity.aliases IS NOT NULL AND size(entity.aliases) > 0 THEN CASE WHEN e.aliases IS NULL THEN entity.aliases diff --git a/api/app/services/emotion_analytics_service.py b/api/app/services/emotion_analytics_service.py index c226348e..9a215cd6 100644 --- a/api/app/services/emotion_analytics_service.py +++ b/api/app/services/emotion_analytics_service.py @@ -679,9 +679,9 @@ class EmotionAnalyticsService: # 查询用户的实体和标签 query = """ - MATCH (e:Entity) + MATCH (e:ExtractedEntity) WHERE e.end_user_id = $end_user_id - RETURN e.name as name, e.type as type + RETURN e.name as name, e.entity_type as type ORDER BY e.created_at DESC LIMIT 20 """ From 47c242e5135880c3a0e54ed2af2455b42f6323ff Mon Sep 17 00:00:00 2001 From: zhaoying Date: Tue, 14 Apr 2026 17:43:58 +0800 Subject: [PATCH 052/245] fix(web): Compatible with Windows whitespace --- .../components/Properties/HttpRequest/index.tsx | 7 +++---- web/src/views/Workflow/hooks/useWorkflowGraph.ts | 13 +++++++++---- 2 files changed, 12 insertions(+), 8 deletions(-) diff --git a/web/src/views/Workflow/components/Properties/HttpRequest/index.tsx b/web/src/views/Workflow/components/Properties/HttpRequest/index.tsx index b032016b..4cf7c150 100644 --- a/web/src/views/Workflow/components/Properties/HttpRequest/index.tsx +++ b/web/src/views/Workflow/components/Properties/HttpRequest/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-09 18:35:43 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-02 17:17:06 + * @Last Modified time: 2026-04-14 17:36:53 */ import { type FC, useMemo, useRef, useState } from "react"; import { useTranslation } from 'react-i18next' @@ -35,9 +35,8 @@ const HttpRequest: FC<{ options: Suggestion[]; selectedNode?: any; graphRef?: an form.setFieldsValue({ auth }) } - const handleChangeBodyContentType = (e: any) => { - const value = e.target.value || e.target.value - form.setFieldValue(['body', 'data'], ['form-data', 'x-www-form-urlencoded'].includes(value) ? [{}] : undefined) + const handleChangeBodyContentType = () => { + form.setFieldValue(['body', 'data'], undefined) } // Handle error handling method change and update node ports accordingly diff --git a/web/src/views/Workflow/hooks/useWorkflowGraph.ts b/web/src/views/Workflow/hooks/useWorkflowGraph.ts index 3280562a..535a4eb0 100644 --- a/web/src/views/Workflow/hooks/useWorkflowGraph.ts +++ b/web/src/views/Workflow/hooks/useWorkflowGraph.ts @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 15:17:48 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-14 15:05:33 + * @Last Modified time: 2026-04-14 17:43:14 */ import { Clipboard, Graph, Keyboard, MiniMap, Node, Snapline, type Edge } from '@antv/x6'; import { register } from '@antv/x6-react-shape'; @@ -1200,9 +1200,6 @@ export const useWorkflowGraph = ({ }) || []; const edges = graphRef.current?.getEdges() || [] - - console.log('config', config) - const params = { ...config, features: featuresRef.current, @@ -1262,6 +1259,14 @@ export const useWorkflowGraph = ({ itemConfig[key][vo.key] = vo.value }) } + } else if (data.type === 'http-request' && key === 'body' && data.config[key] && 'defaultValue' in data.config[key]) { + const value = data.config[key].defaultValue + itemConfig[key] = value + if (value.content_type === 'json' && value.data && value.data !== '') { + itemConfig[key].data = value.data.replace(/\u00a0/g, ' ') + } else { + itemConfig[key].data = value.data + } } else if (data.config[key] && 'defaultValue' in data.config[key] && key !== 'knowledge_retrieval') { itemConfig[key] = data.config[key].defaultValue } else if (key === 'knowledge_retrieval' && data.config[key] && 'defaultValue' in data.config[key]) { From 441b21774d8cdfb4aaf19033593a777ee257ff00 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Tue, 14 Apr 2026 17:56:30 +0800 Subject: [PATCH 053/245] fix(rag): replace semicolon separators with newlines in Excel parser output --- api/app/core/rag/deepdoc/parser/excel_parser.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/api/app/core/rag/deepdoc/parser/excel_parser.py b/api/app/core/rag/deepdoc/parser/excel_parser.py index d66a21a8..c3999be9 100644 --- a/api/app/core/rag/deepdoc/parser/excel_parser.py +++ b/api/app/core/rag/deepdoc/parser/excel_parser.py @@ -232,14 +232,14 @@ class RAGExcelParser: t = str(ti[i].value) if i < len(ti) else "" t += ("：" if t else "") + str(c.value) fields.append(t) - line = "; ".join(fields) + line = "\n".join(fields) if sheetname.lower().find("sheet") < 0: - line += " ——" + sheetname + line += "\n——" + sheetname res.append(line) else: # 只有表头的情况 if header_fields: - line = "; ".join(header_fields) + line = "\n".join(header_fields) if sheetname.lower().find("sheet") < 0: line += " ——" + sheetname res.append(line) From 3e48d620b2853b44d3b386ef50ebfbe1c62379bc Mon Sep 17 00:00:00 2001 From: zhaoying Date: Tue, 14 Apr 2026 17:59:24 +0800 Subject: [PATCH 054/245] feat(web): table support pagesize --- web/src/components/Table/index.tsx | 6 +++--- web/src/views/Index/index.tsx | 1 + 2 files changed, 4 insertions(+), 3 deletions(-) diff --git a/web/src/components/Table/index.tsx b/web/src/components/Table/index.tsx index bb79b4bc..d6cb3c68 100644 --- a/web/src/components/Table/index.tsx +++ b/web/src/components/Table/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-02 15:29:46 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-26 14:52:23 + * @Last Modified time: 2026-04-14 17:55:15 */ /** * RbTable Component @@ -27,7 +27,7 @@ import { useTranslation } from 'react-i18next'; import { request } from '@/utils/request'; import Empty from '@/components/Empty'; -interface TablePaginationConfig { pagesize: number; page: number; } +interface TablePaginationConfig { pagesize?: number; page?: number; } /** Props interface for Table component */ interface TableComponentProps, Q = Record> extends Omit, 'pagination'> { @@ -102,7 +102,7 @@ const RbTable = forwardRef(, Q = Record { rowKey="id" bordered={false} scrollY="100%" + pagination={{pagesize: 10}} />
From 49e0801d1582c08bc7d746edd20d8314abc8bf29 Mon Sep 17 00:00:00 2001 From: lanceyq <1982376970@qq.com> Date: Tue, 14 Apr 2026 18:06:56 +0800 Subject: [PATCH 055/245] refactor(memory): unify user placeholder names and harden alias sync logic - Replace hardcoded user placeholder name lists in write_tools and user_memory_service with shared _USER_PLACEHOLDER_NAMES constant - Filter user placeholder names during alias merging in _merge_attribute to prevent cross-role alias contamination on non-user entities - Use toLower() in Cypher query for case-insensitive name matching - Change PgSQL->Neo4j alias sync condition from 'if pg_aliases' to 'if info is not None' so empty aliases correctly clear stale data --- api/app/core/memory/agent/utils/write_tools.py | 8 ++++++-- .../deduplication/deduped_and_disamb.py | 9 ++++++--- api/app/services/user_memory_service.py | 3 ++- 3 files changed, 14 insertions(+), 6 deletions(-) diff --git a/api/app/core/memory/agent/utils/write_tools.py b/api/app/core/memory/agent/utils/write_tools.py index 5e51beba..3b0ea1ee 100644 --- a/api/app/core/memory/agent/utils/write_tools.py +++ b/api/app/core/memory/agent/utils/write_tools.py @@ -14,6 +14,7 @@ from dotenv import load_dotenv from app.core.logging_config import get_agent_logger from app.core.memory.agent.utils.get_dialogs import get_chunked_dialogs +from app.core.memory.storage_services.extraction_engine.deduplication.deduped_and_disamb import _USER_PLACEHOLDER_NAMES from app.core.memory.storage_services.extraction_engine.extraction_orchestrator import ExtractionOrchestrator from app.core.memory.storage_services.extraction_engine.knowledge_extraction.memory_summary import \ memory_summary_generation @@ -201,14 +202,17 @@ async def write( with get_db_context() as db_session: info = EndUserInfoRepository(db_session).get_by_end_user_id(uuid.UUID(end_user_id)) pg_aliases = info.aliases if info and info.aliases else [] - if pg_aliases: + if info is not None: + # 将 Python 侧占位名集合作为参数传入，避免 Cypher 硬编码 + placeholder_names = list(_USER_PLACEHOLDER_NAMES) await neo4j_connector.execute_query( """ MATCH (e:ExtractedEntity) - WHERE e.end_user_id = $end_user_id AND e.name IN ['用户', '我', 'User', 'I'] + WHERE e.end_user_id = $end_user_id AND toLower(e.name) IN $placeholder_names SET e.aliases = $aliases """, end_user_id=end_user_id, aliases=pg_aliases, + placeholder_names=placeholder_names, ) logger.info(f"[AliasSync] Neo4j 用户实体 aliases 已用 PgSQL 权威源覆盖: {pg_aliases}") except Exception as sync_err: diff --git a/api/app/core/memory/storage_services/extraction_engine/deduplication/deduped_and_disamb.py b/api/app/core/memory/storage_services/extraction_engine/deduplication/deduped_and_disamb.py index 8f659a27..715f190c 100644 --- a/api/app/core/memory/storage_services/extraction_engine/deduplication/deduped_and_disamb.py +++ b/api/app/core/memory/storage_services/extraction_engine/deduplication/deduped_and_disamb.py @@ -88,11 +88,14 @@ def _merge_attribute(canonical: ExtractedEntityNode, ent: ExtractedEntityNode): if canonical_name.lower() not in _USER_PLACEHOLDER_NAMES: incoming_name = (getattr(ent, "name", "") or "").strip() - # 收集所有需要合并的别名 + # 收集所有需要合并的别名，过滤掉用户占位名避免污染非用户实体 all_aliases = list(getattr(canonical, "aliases", []) or []) - if incoming_name and incoming_name != canonical_name: + if incoming_name and incoming_name != canonical_name and incoming_name.lower() not in _USER_PLACEHOLDER_NAMES: all_aliases.append(incoming_name) - all_aliases.extend(getattr(ent, "aliases", []) or []) + all_aliases.extend( + a for a in (getattr(ent, "aliases", []) or []) + if a and a.strip().lower() not in _USER_PLACEHOLDER_NAMES + ) try: from app.core.memory.utils.alias_utils import normalize_aliases diff --git a/api/app/services/user_memory_service.py b/api/app/services/user_memory_service.py index cc18447e..9389ecfa 100644 --- a/api/app/services/user_memory_service.py +++ b/api/app/services/user_memory_service.py @@ -14,6 +14,7 @@ from pydantic import BaseModel, Field from sqlalchemy.orm import Session from app.core.logging_config import get_logger +from app.core.memory.storage_services.extraction_engine.deduplication.deduped_and_disamb import _USER_PLACEHOLDER_NAMES from app.core.memory.utils.llm.llm_utils import MemoryClientFactory from app.db import get_db_context from app.repositories.conversation_repository import ConversationRepository @@ -473,7 +474,7 @@ class UserMemoryService: allowed_fields = {'other_name', 'aliases', 'meta_data'} # 用户占位名称黑名单，不允许作为 other_name 或出现在 aliases 中 - _user_placeholder_names = {'用户', '我', 'User', 'I'} + _user_placeholder_names = _USER_PLACEHOLDER_NAMES # 过滤 other_name：不允许设置为占位名称 if 'other_name' in update_data and update_data['other_name'] and update_data['other_name'].strip() in _user_placeholder_names: From 18be1a9f89dd0f90499dfed3dfeeb4eb26faa07f Mon Sep 17 00:00:00 2001 From: wxy Date: Tue, 14 Apr 2026 18:14:45 +0800 Subject: [PATCH 056/245] feat(tenant): add tenant package query endpoint Add tenant package query functionality. Regular users can access this endpoint to retrieve their tenant's package information. --- api/app/controllers/__init__.py | 4 +- .../tenant_subscription_controller.py | 53 +++++++++++++++++++ 2 files changed, 56 insertions(+), 1 deletion(-) create mode 100644 api/app/controllers/tenant_subscription_controller.py diff --git a/api/app/controllers/__init__.py b/api/app/controllers/__init__.py index 50e9e0b0..377205c4 100644 --- a/api/app/controllers/__init__.py +++ b/api/app/controllers/__init__.py @@ -47,7 +47,8 @@ from . import ( user_memory_controllers, workspace_controller, ontology_controller, - skill_controller + skill_controller, + tenant_subscription_controller, ) # 创建管理端 API 路由器 @@ -98,5 +99,6 @@ manager_router.include_router(file_storage_controller.router) manager_router.include_router(ontology_controller.router) manager_router.include_router(skill_controller.router) manager_router.include_router(i18n_controller.router) +manager_router.include_router(tenant_subscription_controller.router) __all__ = ["manager_router"] diff --git a/api/app/controllers/tenant_subscription_controller.py b/api/app/controllers/tenant_subscription_controller.py new file mode 100644 index 00000000..2629f7f1 --- /dev/null +++ b/api/app/controllers/tenant_subscription_controller.py @@ -0,0 +1,53 @@ +""" +租户套餐查询接口（普通用户可访问） +""" +from typing import Callable + +from fastapi import APIRouter, Depends +from fastapi.responses import JSONResponse +from sqlalchemy.orm import Session + +from app.core.logging_config import get_api_logger +from app.core.response_utils import success, fail +from app.db import get_db +from app.dependencies import get_current_user +from app.i18n.dependencies import get_translator +from app.models.user_model import User +from app.schemas.response_schema import ApiResponse + +logger = get_api_logger() + +router = APIRouter(prefix="/tenant", tags=["Tenant"]) + + +@router.get("/subscription", response_model=ApiResponse, summary="获取当前用户所属租户的套餐信息") +async def get_my_tenant_subscription( + current_user: User = Depends(get_current_user), + db: Session = Depends(get_db), + t: Callable = Depends(get_translator), +): + """ + 获取当前登录用户所属租户的有效套餐订阅信息。 + 包含套餐名称、版本、配额、到期时间等。 + """ + try: + from premium.platform_admin.package_plan_service import TenantSubscriptionService + + if not current_user.tenant: + return JSONResponse(status_code=404, content=fail(code=404, msg="用户未关联租户")) + + tenant_id = current_user.tenant.id + svc = TenantSubscriptionService(db) + sub = svc.get_subscription(tenant_id) + + if not sub: + return success(data=None, msg="暂无有效套餐") + + return success(data=svc.build_response(sub)) + + except ModuleNotFoundError: + # 社区版无 premium 模块，返回空 + return success(data=None, msg="套餐功能未启用") + except Exception as e: + logger.error(f"获取租户套餐信息失败: {e}", exc_info=True) + return JSONResponse(status_code=500, content=fail(code=500, msg="获取套餐信息失败")) From 017efdc3202b424dc48d483ecba3498cd50810ff Mon Sep 17 00:00:00 2001 From: Eternity <1533512157@qq.com> Date: Wed, 15 Apr 2026 11:03:44 +0800 Subject: [PATCH 057/245] fix(prompt-optimizer): support list content type in prompt optimizer --- api/app/services/prompt_optimizer_service.py | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/api/app/services/prompt_optimizer_service.py b/api/app/services/prompt_optimizer_service.py index fde8c4f9..b1de84d2 100644 --- a/api/app/services/prompt_optimizer_service.py +++ b/api/app/services/prompt_optimizer_service.py @@ -227,7 +227,14 @@ class PromptOptimizerService: content = getattr(chunk, "content", chunk) if not content: continue - buffer += content + if isinstance(content, str): + buffer += content + elif isinstance(content, list): + for _ in content: + buffer += _["text"] + else: + logger.error(f"Unsupported content type - {content}") + raise Exception("Unsupported content type") cache = buffer[:-20] # 尝试找到 "prompt": " 开始位置 From daaee63bd590ff069ed4d50bce45e38e976426cd Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Wed, 15 Apr 2026 12:15:16 +0800 Subject: [PATCH 058/245] refactor(custom-tools): coerce query and request body parameters to schema types --- api/app/core/tools/custom/base.py | 27 ++++++++++++++++++++------- 1 file changed, 20 insertions(+), 7 deletions(-) diff --git a/api/app/core/tools/custom/base.py b/api/app/core/tools/custom/base.py index c03fe206..93ad03a7 100644 --- a/api/app/core/tools/custom/base.py +++ b/api/app/core/tools/custom/base.py @@ -221,7 +221,7 @@ class CustomTool(BaseTool): query_params = {} for param_name, param_info in operation.get("parameters", {}).items(): if param_info.get("in") == "query" and param_name in params: - query_params[param_name] = params[param_name] + query_params[param_name] = self._coerce_param(params[param_name], param_info.get("type", "string")) if query_params: from urllib.parse import urlencode @@ -251,21 +251,34 @@ class CustomTool(BaseTool): return headers @staticmethod - def _build_request_data(operation: Dict[str, Any], params: Dict[str, Any]) -> Optional[Dict[str, Any]]: + def _coerce_param(value: Any, schema_type: str) -> Any: + """根据 schema 类型转换参数值""" + if value is None: + return value + try: + if schema_type == "integer": + return int(value) + elif schema_type == "number": + return float(value) + elif schema_type == "boolean": + if isinstance(value, str): + return value.lower() not in ("false", "0", "") + return bool(value) + except (ValueError, TypeError): + pass + return value + + def _build_request_data(self, operation: Dict[str, Any], params: Dict[str, Any]) -> Optional[Dict[str, Any]]: """构建请求数据""" if operation["method"] in ["POST", "PUT", "PATCH"]: request_body = operation.get("request_body") if request_body: - # 构建请求体数据 data = {} properties = request_body.get("properties", {}) - for prop_name, prop_schema in properties.items(): if prop_name in params: - data[prop_name] = params[prop_name] - + data[prop_name] = self._coerce_param(params[prop_name], prop_schema.get("type", "string")) return data if data else None - return None async def _send_http_request( From 2716a55c7f7844a0a4b054229ca0d06f9a6465c0 Mon Sep 17 00:00:00 2001 From: Eternity <1533512157@qq.com> Date: Wed, 15 Apr 2026 12:18:23 +0800 Subject: [PATCH 059/245] feat(memory): implement quick search pipeline with Neo4j integration --- .../nodes/perceptual_retrieve_node.py | 6 +- .../langgraph_graph/nodes/summary_nodes.py | 3 +- .../agent/langgraph_graph/read_graph.py | 17 +- .../memory/agent/services/search_service.py | 21 +- api/app/core/memory/enums.py | 11 + .../core/memory/llm_tools/chunker_client.py | 7 +- api/app/core/memory/memory_service.py | 22 +- api/app/core/memory/models/service_models.py | 26 + .../core/memory/pipelines/base_pipeline.py | 26 +- api/app/core/memory/pipelines/memory_read.py | 17 +- api/app/core/memory/read_services/__init__.py | 0 .../memory/read_services/content_search.py | 178 ++++++ .../memory/read_services/result_builder.py | 150 +++++ api/app/core/memory/src/search.py | 12 +- .../storage_services/short_engine/__init__.py | 0 api/app/models/memory_perceptual_model.py | 3 +- api/app/repositories/neo4j/cypher_queries.py | 530 +++++++++--------- api/app/repositories/neo4j/graph_search.py | 432 ++++++-------- api/app/repositories/neo4j/neo4j_connector.py | 12 +- 19 files changed, 899 insertions(+), 574 deletions(-) create mode 100644 api/app/core/memory/models/service_models.py create mode 100644 api/app/core/memory/read_services/__init__.py create mode 100644 api/app/core/memory/read_services/content_search.py create mode 100644 api/app/core/memory/read_services/result_builder.py create mode 100644 api/app/core/memory/storage_services/short_engine/__init__.py diff --git a/api/app/core/memory/agent/langgraph_graph/nodes/perceptual_retrieve_node.py b/api/app/core/memory/agent/langgraph_graph/nodes/perceptual_retrieve_node.py index 1cf5e291..64becc4c 100644 --- a/api/app/core/memory/agent/langgraph_graph/nodes/perceptual_retrieve_node.py +++ b/api/app/core/memory/agent/langgraph_graph/nodes/perceptual_retrieve_node.py @@ -15,7 +15,7 @@ from app.core.logging_config import get_agent_logger from app.core.memory.agent.utils.llm_tools import ReadState from app.core.memory.utils.data.text_utils import escape_lucene_query from app.repositories.neo4j.graph_search import ( - search_perceptual, + search_perceptual_by_fulltext, search_perceptual_by_embedding, ) from app.repositories.neo4j.neo4j_connector import Neo4jConnector @@ -152,7 +152,7 @@ class PerceptualSearchService: if not escaped.strip(): return [] try: - r = await search_perceptual( + r = await search_perceptual_by_fulltext( connector=connector, query=escaped, end_user_id=self.end_user_id, limit=limit * 5, # 多查一些以提高命中率 @@ -177,7 +177,7 @@ class PerceptualSearchService: escaped = escape_lucene_query(kw) if not escaped.strip(): return [] - r = await search_perceptual( + r = await search_perceptual_by_fulltext( connector=connector, query=escaped, end_user_id=self.end_user_id, limit=limit, ) diff --git a/api/app/core/memory/agent/langgraph_graph/nodes/summary_nodes.py b/api/app/core/memory/agent/langgraph_graph/nodes/summary_nodes.py index 1bf68966..eee98ac7 100644 --- a/api/app/core/memory/agent/langgraph_graph/nodes/summary_nodes.py +++ b/api/app/core/memory/agent/langgraph_graph/nodes/summary_nodes.py @@ -19,6 +19,7 @@ from app.core.memory.agent.utils.llm_tools import ( from app.core.memory.agent.utils.redis_tool import store from app.core.memory.agent.utils.session_tools import SessionService from app.core.memory.agent.utils.template_tools import TemplateService +from app.core.memory.enums import Neo4jNodeType from app.core.rag.nlp.search import knowledge_retrieval from app.db import get_db_context @@ -338,7 +339,7 @@ async def Input_Summary(state: ReadState) -> ReadState: "end_user_id": end_user_id, "question": data, "return_raw_results": True, - "include": ["summaries", "communities"] # MemorySummary 和 Community 同为高维度概括节点 + "include": [Neo4jNodeType.MEMORYSUMMARY, Neo4jNodeType.COMMUNITY] # MemorySummary 和 Community 同为高维度概括节点 } try: diff --git a/api/app/core/memory/agent/langgraph_graph/read_graph.py b/api/app/core/memory/agent/langgraph_graph/read_graph.py index d3ca4ea7..d3ec9ab6 100644 --- a/api/app/core/memory/agent/langgraph_graph/read_graph.py +++ b/api/app/core/memory/agent/langgraph_graph/read_graph.py @@ -1,15 +1,14 @@ #!/usr/bin/env python3 +import logging from contextlib import asynccontextmanager -from langchain_core.messages import HumanMessage from langgraph.constants import START, END from langgraph.graph import StateGraph -from app.db import get_db -from app.services.memory_config_service import MemoryConfigService - -from app.core.memory.agent.utils.llm_tools import ReadState from app.core.memory.agent.langgraph_graph.nodes.data_nodes import content_input_node +from app.core.memory.agent.langgraph_graph.nodes.perceptual_retrieve_node import ( + perceptual_retrieve_node, +) from app.core.memory.agent.langgraph_graph.nodes.problem_nodes import ( Split_The_Problem, Problem_Extension, @@ -17,9 +16,6 @@ from app.core.memory.agent.langgraph_graph.nodes.problem_nodes import ( from app.core.memory.agent.langgraph_graph.nodes.retrieve_nodes import ( retrieve_nodes, ) -from app.core.memory.agent.langgraph_graph.nodes.perceptual_retrieve_node import ( - perceptual_retrieve_node, -) from app.core.memory.agent.langgraph_graph.nodes.summary_nodes import ( Input_Summary, Retrieve_Summary, @@ -32,6 +28,9 @@ from app.core.memory.agent.langgraph_graph.routing.routers import ( Retrieve_continue, Verify_continue, ) +from app.core.memory.agent.utils.llm_tools import ReadState + +logger = logging.getLogger(__name__) @asynccontextmanager @@ -51,7 +50,7 @@ async def make_read_graph(): """ try: # Build workflow graph - workflow = StateGraph(ReadState) + workflow = StateGraph(ReadState) workflow.add_node("content_input", content_input_node) workflow.add_node("Split_The_Problem", Split_The_Problem) workflow.add_node("Problem_Extension", Problem_Extension) diff --git a/api/app/core/memory/agent/services/search_service.py b/api/app/core/memory/agent/services/search_service.py index eaa5f0ab..93d1ebee 100644 --- a/api/app/core/memory/agent/services/search_service.py +++ b/api/app/core/memory/agent/services/search_service.py @@ -7,6 +7,7 @@ and deduplication. from typing import List, Tuple, Optional from app.core.logging_config import get_agent_logger +from app.core.memory.enums import Neo4jNodeType from app.core.memory.src.search import run_hybrid_search from app.core.memory.utils.data.text_utils import escape_lucene_query @@ -111,13 +112,13 @@ class SearchService: content_parts = [] # Statements: extract statement field - if 'statement' in result and result['statement']: - content_parts.append(result['statement']) + if Neo4jNodeType.STATEMENT in result and result[Neo4jNodeType.STATEMENT]: + content_parts.append(result[Neo4jNodeType.STATEMENT]) # Community 节点：有 member_count 或 core_entities 字段，或 node_type 明确指定 # 用 "[主题：{name}]" 前缀区分，让 LLM 知道这是主题级摘要 is_community = ( - node_type == "community" + node_type == Neo4jNodeType.COMMUNITY or 'member_count' in result or 'core_entities' in result ) @@ -204,7 +205,7 @@ class SearchService: raw_results is None if return_raw_results=False """ if include is None: - include = ["statements", "chunks", "entities", "summaries", "communities"] + include = [Neo4jNodeType.STATEMENT, Neo4jNodeType.CHUNK, Neo4jNodeType.EXTRACTEDENTITY, Neo4jNodeType.MEMORYSUMMARY, Neo4jNodeType.COMMUNITY] # Clean query cleaned_query = self.clean_query(question) @@ -231,7 +232,7 @@ class SearchService: reranked_results = answer.get('reranked_results', {}) # Priority order: summaries first (most contextual), then communities, statements, chunks, entities - priority_order = ['summaries', 'communities', 'statements', 'chunks', 'entities'] + priority_order = [Neo4jNodeType.STATEMENT, Neo4jNodeType.CHUNK, Neo4jNodeType.EXTRACTEDENTITY, Neo4jNodeType.MEMORYSUMMARY, Neo4jNodeType.COMMUNITY] for category in priority_order: if category in include and category in reranked_results: @@ -241,7 +242,7 @@ class SearchService: else: # For keyword or embedding search, results are directly in answer dict # Apply same priority order - priority_order = ['summaries', 'communities', 'statements', 'chunks', 'entities'] + priority_order = [Neo4jNodeType.STATEMENT, Neo4jNodeType.CHUNK, Neo4jNodeType.EXTRACTEDENTITY, Neo4jNodeType.MEMORYSUMMARY, Neo4jNodeType.COMMUNITY] for category in priority_order: if category in include and category in answer: @@ -250,11 +251,11 @@ class SearchService: answer_list.extend(category_results) # 对命中的 community 节点展开其成员 statements（路径 "0"/"1" 需要，路径 "2" 不需要） - if expand_communities and "communities" in include: + if expand_communities and Neo4jNodeType.COMMUNITY in include: community_results = ( - answer.get('reranked_results', {}).get('communities', []) + answer.get('reranked_results', {}).get(Neo4jNodeType.COMMUNITY.value, []) if search_type == "hybrid" - else answer.get('communities', []) + else answer.get(Neo4jNodeType.COMMUNITY.value, []) ) cleaned_stmts, new_texts = await expand_communities_to_statements( community_results=community_results, @@ -266,7 +267,7 @@ class SearchService: content_list = [] for ans in answer_list: # community 节点有 member_count 或 core_entities 字段 - ntype = "community" if ('member_count' in ans or 'core_entities' in ans) else "" + ntype = Neo4jNodeType.COMMUNITY if ('member_count' in ans or 'core_entities' in ans) else "" content_list.append(self.extract_content_from_result(ans, node_type=ntype)) # Filter out empty strings and join with newlines diff --git a/api/app/core/memory/enums.py b/api/app/core/memory/enums.py index d0baf732..5c4c3a13 100644 --- a/api/app/core/memory/enums.py +++ b/api/app/core/memory/enums.py @@ -16,3 +16,14 @@ class SearchStrategy(StrEnum): DEEP = "0" NORMAL = "1" QUICK = "2" + + +class Neo4jNodeType(StrEnum): + CHUNK = "Chunk" + COMMUNITY = "Community" + DIALOGUE = "Dialogue" + EXTRACTEDENTITY = "ExtractedEntity" + MEMORYSUMMARY = "MemorySummary" + PERCEPTUAL = "Perceptual" + STATEMENT = "Statement" + diff --git a/api/app/core/memory/llm_tools/chunker_client.py b/api/app/core/memory/llm_tools/chunker_client.py index 51d15aab..fbac4cca 100644 --- a/api/app/core/memory/llm_tools/chunker_client.py +++ b/api/app/core/memory/llm_tools/chunker_client.py @@ -21,6 +21,7 @@ from chonkie import ( from app.core.memory.models.config_models import ChunkerConfig from app.core.memory.models.message_models import DialogData, Chunk + try: from app.core.memory.llm_tools.openai_client import OpenAIClient except Exception: @@ -32,6 +33,7 @@ logger = logging.getLogger(__name__) class LLMChunker: """LLM-based intelligent chunking strategy""" + def __init__(self, llm_client: OpenAIClient, chunk_size: int = 1000): self.llm_client = llm_client self.chunk_size = chunk_size @@ -46,7 +48,8 @@ class LLMChunker: """ messages = [ - {"role": "system", "content": "You are a professional text analysis assistant, skilled at splitting long texts into semantically coherent paragraphs."}, + {"role": "system", + "content": "You are a professional text analysis assistant, skilled at splitting long texts into semantically coherent paragraphs."}, {"role": "user", "content": prompt} ] @@ -311,7 +314,7 @@ class ChunkerClient: f.write("=" * 60 + "\n\n") for i, chunk in enumerate(dialogue.chunks): - f.write(f"Chunk {i+1}:\n") + f.write(f"Chunk {i + 1}:\n") f.write(f"Size: {len(chunk.content)} characters\n") if hasattr(chunk, 'metadata') and 'start_index' in chunk.metadata: f.write(f"Position: {chunk.metadata.get('start_index')}-{chunk.metadata.get('end_index')}\n") diff --git a/api/app/core/memory/memory_service.py b/api/app/core/memory/memory_service.py index 2f0d8bb3..67c814b1 100644 --- a/api/app/core/memory/memory_service.py +++ b/api/app/core/memory/memory_service.py @@ -1,21 +1,12 @@ from sqlalchemy.orm import Session -from pydantic import BaseModel, ConfigDict -from app.core.memory.enums import StorageType -from app.schemas import MemoryConfig +from app.core.memory.enums import StorageType, SearchStrategy +from app.core.memory.models.service_models import Memory, MemoryContext +from app.core.memory.pipelines.memory_read import ReadPipeLine +from app.db import get_db_context from app.services.memory_config_service import MemoryConfigService -class MemoryContext(BaseModel): - model_config = ConfigDict(frozen=True, arbitrary_types_allowed=True) - - end_user_id: str - memory_config: MemoryConfig - storage_type: StorageType = StorageType.NEO4J - user_rag_memory_id: str | None = None - language: str = "zh" - - class MemoryService: def __init__( self, @@ -44,8 +35,9 @@ class MemoryService: async def write(self, messages: list[dict]) -> str: raise NotImplementedError - async def read(self, query: str, history: list, search_switch: str) -> dict: - raise NotImplementedError + async def read(self, query: str, history: list, search_switch: SearchStrategy) -> list[Memory]: + with get_db_context() as db: + return await ReadPipeLine(self.ctx, db).run(query, search_switch, limit=10) async def forget(self, max_batch: int = 100, min_days: int = 30) -> dict: raise NotImplementedError diff --git a/api/app/core/memory/models/service_models.py b/api/app/core/memory/models/service_models.py new file mode 100644 index 00000000..82a867c7 --- /dev/null +++ b/api/app/core/memory/models/service_models.py @@ -0,0 +1,26 @@ +from pydantic import BaseModel, Field, field_serializer, ConfigDict + +from app.core.memory.enums import Neo4jNodeType, StorageType +from app.schemas.memory_config_schema import MemoryConfig + + +class MemoryContext(BaseModel): + model_config = ConfigDict(frozen=True, arbitrary_types_allowed=True) + + end_user_id: str + memory_config: MemoryConfig + storage_type: StorageType = StorageType.NEO4J + user_rag_memory_id: str | None = None + language: str = "zh" + + +class Memory(BaseModel): + source: Neo4jNodeType = Field(...) + score: float = Field(default=0.0) + content: str = Field(default="") + data: dict = Field(default_factory=dict) + query: str = Field(...) + + @field_serializer("source") + def serialize_source(self, v) -> str: + return v.value diff --git a/api/app/core/memory/pipelines/base_pipeline.py b/api/app/core/memory/pipelines/base_pipeline.py index d423aef2..322f6787 100644 --- a/api/app/core/memory/pipelines/base_pipeline.py +++ b/api/app/core/memory/pipelines/base_pipeline.py @@ -1,22 +1,32 @@ -# -*- coding: UTF-8 -*- -# Author: Eternity -# @Email: 1533512157@qq.com -# @Time : 2026/4/3 11:44 import uuid from abc import ABC, abstractmethod from typing import Any from sqlalchemy.orm import Session -from demo.memory_alpha import MemoryContext +from app.core.memory.llm_tools import OpenAIEmbedderClient +from app.core.memory.models.service_models import MemoryContext +from app.core.models import RedBearModelConfig +from app.services.memory_config_service import MemoryConfigService class ModelClientMixin(ABC): - def get_llm_client(self, db: Session, model_id: uuid.UUID): + @staticmethod + def get_llm_client(db: Session, model_id: uuid.UUID): pass - def get_embedding_client(self, db: Session, model_id: uuid.UUID): - pass + @staticmethod + def get_embedding_client(db: Session, model_id: uuid.UUID) -> OpenAIEmbedderClient: + config_service = MemoryConfigService(db) + embedder_client_config = config_service.get_embedder_config(str(model_id)) + return OpenAIEmbedderClient( + RedBearModelConfig( + model_name=embedder_client_config["model_name"], + provider=embedder_client_config["provider"], + api_key=embedder_client_config["api_key"], + base_url=embedder_client_config["base_url"], + ) + ) class BasePipeline(ABC): diff --git a/api/app/core/memory/pipelines/memory_read.py b/api/app/core/memory/pipelines/memory_read.py index f6ccd210..5f5a1a1f 100644 --- a/api/app/core/memory/pipelines/memory_read.py +++ b/api/app/core/memory/pipelines/memory_read.py @@ -1,10 +1,11 @@ from app.core.memory.enums import SearchStrategy -from app.core.memory.pipelines.base_pipeline import BasePipeline +from app.core.memory.pipelines.base_pipeline import BasePipeline, ModelClientMixin +from app.core.memory.read_services.content_search import Neo4jSearchService from app.core.memory.read_services.query_preprocessor import QueryPreprocessor -class ReadPipeLine(BasePipeline): - async def run(self, query, search_switch, memory_config): +class ReadPipeLine(ModelClientMixin, BasePipeline): + async def run(self, query: str, search_switch: SearchStrategy, limit: int = 10): query = QueryPreprocessor.process(query) match search_switch: case SearchStrategy.DEEP: @@ -12,7 +13,7 @@ class ReadPipeLine(BasePipeline): case SearchStrategy.NORMAL: return await self._normal_read(query) case SearchStrategy.QUICK: - return await self._quick_read() + return await self._quick_read(query, limit) case _: raise RuntimeError("Unsupported search strategy") @@ -22,5 +23,9 @@ class ReadPipeLine(BasePipeline): async def _normal_read(self, query): pass - async def _quick_read(self): - pass + async def _quick_read(self, query, limit): + search_service = Neo4jSearchService( + self.ctx, + self.get_embedding_client(self.db, self.ctx.memory_config.embedding_model_id) + ) + return await search_service.search(query, limit) diff --git a/api/app/core/memory/read_services/__init__.py b/api/app/core/memory/read_services/__init__.py new file mode 100644 index 00000000..e69de29b diff --git a/api/app/core/memory/read_services/content_search.py b/api/app/core/memory/read_services/content_search.py new file mode 100644 index 00000000..69ca6b11 --- /dev/null +++ b/api/app/core/memory/read_services/content_search.py @@ -0,0 +1,178 @@ +import asyncio +import logging +import math +import time + +from pydantic import BaseModel, Field + +from app.core.memory.enums import Neo4jNodeType +from app.core.memory.llm_tools import OpenAIEmbedderClient +from app.core.memory.memory_service import MemoryContext +from app.core.memory.models.service_models import Memory +from app.core.memory.read_services.result_builder import data_builder_factory +from app.repositories.neo4j.graph_search import search_graph, search_graph_by_embedding +from app.repositories.neo4j.neo4j_connector import Neo4jConnector + +logger = logging.getLogger(__name__) + + +class MemorySearchResult(BaseModel): + memories: dict[str, list[dict]] = Field(default_factory=dict) + content: str = Field(default="") + count: int = Field(default=0) + + +class Neo4jSearchService: + DEFAULT_ALPHA = 0.6 + DEFAULT_FULLTEXT_SCORE_THRESHOLD = 1 + DEFAULT_COSINE_SCORE_THRESHOLD = 0.5 + DEFAULT_CONTENT_SCORE_THRESHOLD = 0.5 + + def __init__( + self, + ctx: MemoryContext, + embedder: OpenAIEmbedderClient, + includes: list[Neo4jNodeType] | None = None, + alpha: float = DEFAULT_ALPHA, + fulltext_score_threshold: float = DEFAULT_FULLTEXT_SCORE_THRESHOLD, + cosine_score_threshold: float = DEFAULT_COSINE_SCORE_THRESHOLD, + content_score_threshold: float = DEFAULT_CONTENT_SCORE_THRESHOLD + ): + self.ctx = ctx + self.alpha = alpha + self.fulltext_score_threshold = fulltext_score_threshold + self.cosine_score_threshold = cosine_score_threshold + self.content_score_threshold = content_score_threshold + + self.embedder: OpenAIEmbedderClient = embedder + self.connector: Neo4jConnector | None = None + + self.includes = includes + if includes is None: + self.includes = [ + Neo4jNodeType.STATEMENT, + Neo4jNodeType.CHUNK, + Neo4jNodeType.EXTRACTEDENTITY, + Neo4jNodeType.MEMORYSUMMARY, + Neo4jNodeType.PERCEPTUAL, + Neo4jNodeType.COMMUNITY + ] + + async def _keyword_search( + self, + query: str, + limit: int + ): + return await search_graph( + connector=self.connector, + query=query, + end_user_id=self.ctx.end_user_id, + limit=limit, + include=self.includes + ) + + async def _embedding_search(self, query, limit): + return await search_graph_by_embedding( + connector=self.connector, + embedder_client=self.embedder, + query_text=query, + end_user_id=self.ctx.end_user_id, + limit=limit, + include=self.includes + ) + + def _rerank( + self, + keyword_results: list[dict], + embedding_results: list[dict], + limit: int, + ) -> list[dict]: + keyword_results = self._normalize_kw_scores(keyword_results) + embedding_results = embedding_results + + kw_norm_map = {} + for item in keyword_results: + item_id = item["id"] + kw_norm_map[item_id] = float(item.get("normalized_kw_score", 0)) + + emb_norm_map = {} + for item in embedding_results: + item_id = item["id"] + emb_norm_map[item_id] = float(item.get("score", 0)) + + combined = {} + for item in keyword_results: + item_id = item["id"] + combined[item_id] = item.copy() + combined[item_id]["kw_score"] = kw_norm_map.get(item_id, 0) + combined[item_id]["embedding_score"] = emb_norm_map.get(item_id, 0) + + for item in embedding_results: + item_id = item["id"] + if item_id in combined: + combined[item_id]["embedding_score"] = emb_norm_map.get(item_id, 0) + else: + combined[item_id] = item.copy() + combined[item_id]["kw_score"] = kw_norm_map.get(item_id, 0) + combined[item_id]["embedding_score"] = emb_norm_map.get(item_id, 0) + + for item in combined.values(): + item_id = item["id"] + kw = float(combined[item_id].get("kw_score", 0) or 0) + emb = float(combined[item_id].get("embedding_score", 0) or 0) + base = self.alpha * emb + (1 - self.alpha) * kw + combined[item_id]["content_score"] = base + min(1 - base, 0.1 * kw * emb) + results = sorted(combined.values(), key=lambda x: x["content_score"], reverse=True) + # results = [res for res in results if res["content_score"] > self.content_score_threshold] + results = results[:limit] + + logger.info( + f"[MemorySearch] rerank: merged={len(combined)}, after_threshold={len(results)} " + f"(alpha={self.alpha})" + ) + return results + + def _normalize_kw_scores(self, items: list[dict]) -> list[dict]: + if not items: + return items + scores = [float(it.get("score", 0) or 0) for it in items] + for it, s in zip(items, scores): + it[f"normalized_kw_score"] = 1 / (1 + math.exp(-(s - self.fulltext_score_threshold) / 2)) + return items + + async def search( + self, + query: str, + limit: int = 10, + ) -> list[Memory]: + async with Neo4jConnector() as connector: + self.connector = connector + kw_task = self._keyword_search(query, limit) + emb_task = self._embedding_search(query, limit) + kw_results, emb_results = await asyncio.gather(kw_task, emb_task, return_exceptions=True) + + if isinstance(kw_results, Exception): + logger.warning(f"[MemorySearch] keyword search error: {kw_results}") + kw_results = {} + if isinstance(emb_results, Exception): + logger.warning(f"[MemorySearch] embedding search error: {emb_results}") + emb_results = {} + + memories = [] + for node_type in self.includes: + reranked = self._rerank( + kw_results.get(node_type, []), + emb_results.get(node_type, []), + limit + ) + for record in reranked: + memory = data_builder_factory(node_type, record) + memories.append(Memory( + score=memory.score, + content=memory.content, + data=memory.data, + source=node_type, + query=query + )) + memories.sort(key=lambda x: x.score, reverse=True) + return memories[:limit] diff --git a/api/app/core/memory/read_services/result_builder.py b/api/app/core/memory/read_services/result_builder.py new file mode 100644 index 00000000..10ff8c86 --- /dev/null +++ b/api/app/core/memory/read_services/result_builder.py @@ -0,0 +1,150 @@ +from abc import ABC, abstractmethod +from typing import TypeVar + +from app.core.memory.enums import Neo4jNodeType + + +class BaseBuilder(ABC): + def __init__(self, records: dict): + self.record = records + + @property + @abstractmethod + def data(self) -> dict: + pass + + @property + @abstractmethod + def content(self) -> str: + pass + + @property + def score(self) -> float: + return self.record.get("content_score", 0.0) or 0.0 + + +T = TypeVar("T", bound=BaseBuilder) + + +class ChunkBuilder(BaseBuilder): + @property + def data(self) -> dict: + return { + "id": self.record.get("id"), + "content": self.record.get("content"), + "kw_score": self.record.get("kw_score", 0.0), + "emb_score": self.record.get("embedding_score", 0.0) + } + + @property + def content(self) -> str: + return self.record.get("content") + + +class StatementBuiler(BaseBuilder): + @property + def data(self) -> dict: + return { + "id": self.record.get("id"), + "content": self.record.get("statement"), + "kw_score": self.record.get("kw_score", 0.0), + "emb_score": self.record.get("embedding_score", 0.0) + } + + @property + def content(self) -> str: + return self.record.get("statement") + + +class EntityBuilder(BaseBuilder): + @property + def data(self) -> dict: + return { + "id": self.record.get("id"), + "content": self.record.get("name"), + "kw_score": self.record.get("kw_score", 0.0), + "emb_score": self.record.get("embedding_score", 0.0) + } + + @property + def content(self) -> str: + return self.record.get("name") + + +class SummaryBuilder(BaseBuilder): + @property + def data(self) -> dict: + return { + "id": self.record.get("id"), + "content": self.record.get("content"), + "kw_score": self.record.get("kw_score", 0.0), + "emb_score": self.record.get("embedding_score", 0.0) + } + + @property + def content(self) -> str: + return self.record.get("content") + + +class PerceptualBuilder(BaseBuilder): + @property + def data(self) -> dict: + return { + "id": self.record.get("id", ""), + "perceptual_type": self.record.get("perceptual_type", ""), + "file_name": self.record.get("file_name", ""), + "file_path": self.record.get("file_path", ""), + "summary": self.record.get("summary", ""), + "topic": self.record.get("topic", ""), + "domain": self.record.get("domain", ""), + "keywords": self.record.get("keywords", []), + "created_at": str(self.record.get("created_at", "")), + "file_type": self.record.get("file_type", ""), + "kw_score": self.record.get("kw_score", 0.0), + "emb_score": self.record.get("embedding_score", 0.0) + } + + @property + def content(self) -> str: + return ("" + f"{self.record.get('file_name')}" + f"{self.record.get('file_path')}" + f"{self.record.get('summary')}" + f"{self.record.get('topic')}" + f"{self.record.get('domain')}" + f"{self.record.get('keywords')}" + f"{self.record.get('file_type')}" + "") + + +class CommunityBuilder(BaseBuilder): + @property + def data(self) -> dict: + return { + "id": self.record.get("id"), + "content": self.record.get("content"), + "kw_score": self.record.get("kw_score", 0.0), + "emb_score": self.record.get("embedding_score", 0.0) + } + + @property + def content(self) -> str: + return self.record.get("content") + + +def data_builder_factory(node_type, data: dict) -> T: + match node_type: + case Neo4jNodeType.STATEMENT: + return StatementBuiler(data) + case Neo4jNodeType.CHUNK: + return ChunkBuilder(data) + case Neo4jNodeType.EXTRACTEDENTITY: + return EntityBuilder(data) + case Neo4jNodeType.MEMORYSUMMARY: + return SummaryBuilder(data) + case Neo4jNodeType.PERCEPTUAL: + return PerceptualBuilder(data) + case Neo4jNodeType.COMMUNITY: + return CommunityBuilder(data) + case _: + raise KeyError(f"Unknown node_type: {node_type}") diff --git a/api/app/core/memory/src/search.py b/api/app/core/memory/src/search.py index 4e2883d5..b58da0af 100644 --- a/api/app/core/memory/src/search.py +++ b/api/app/core/memory/src/search.py @@ -6,6 +6,8 @@ import time from datetime import datetime from typing import TYPE_CHECKING, Any, Dict, List, Optional +from app.core.memory.enums import Neo4jNodeType + if TYPE_CHECKING: from app.schemas.memory_config_schema import MemoryConfig @@ -131,7 +133,7 @@ def normalize_scores(results: List[Dict[str, Any]], score_field: str = "score") return results -def _deduplicate_results(items: List[Dict[str, Any]]) -> List[Dict[str, Any]]: +def deduplicate_results(items: List[Dict[str, Any]]) -> List[Dict[str, Any]]: """ Remove duplicate items from search results based on content. @@ -194,7 +196,7 @@ def rerank_with_activation( forgetting_config: ForgettingEngineConfig | None = None, activation_boost_factor: float = 0.8, now: datetime | None = None, - content_score_threshold: float = 0.5, + content_score_threshold: float = 0.1, ) -> Dict[str, List[Dict[str, Any]]]: """ 两阶段排序：先按内容相关性筛选，再按激活值排序。 @@ -239,7 +241,7 @@ def rerank_with_activation( reranked: Dict[str, List[Dict[str, Any]]] = {} - for category in ["statements", "chunks", "entities", "summaries", "communities"]: + for category in [Neo4jNodeType.STATEMENT, Neo4jNodeType.CHUNK, Neo4jNodeType.EXTRACTEDENTITY, Neo4jNodeType.MEMORYSUMMARY, Neo4jNodeType.COMMUNITY]: keyword_items = keyword_results.get(category, []) embedding_items = embedding_results.get(category, []) @@ -405,7 +407,7 @@ def rerank_with_activation( f"items below content_score_threshold={content_score_threshold}" ) - sorted_items = _deduplicate_results(sorted_items) + sorted_items = deduplicate_results(sorted_items) reranked[category] = sorted_items @@ -691,7 +693,7 @@ async def run_hybrid_search( search_type: str, end_user_id: str | None, limit: int, - include: List[str], + include: List[Neo4jNodeType], output_path: str | None, memory_config: "MemoryConfig", rerank_alpha: float = 0.6, diff --git a/api/app/core/memory/storage_services/short_engine/__init__.py b/api/app/core/memory/storage_services/short_engine/__init__.py new file mode 100644 index 00000000..e69de29b diff --git a/api/app/models/memory_perceptual_model.py b/api/app/models/memory_perceptual_model.py index ae8cc1bd..7610b79f 100644 --- a/api/app/models/memory_perceptual_model.py +++ b/api/app/models/memory_perceptual_model.py @@ -7,7 +7,8 @@ from sqlalchemy.dialects.postgresql import UUID from sqlalchemy.dialects.postgresql import JSONB from app.db import Base -from app.schemas import FileType +from app.schemas.app_schema import FileType + class PerceptualType(IntEnum): VISION = 1 diff --git a/api/app/repositories/neo4j/cypher_queries.py b/api/app/repositories/neo4j/cypher_queries.py index 9bd46f94..03d51a7c 100644 --- a/api/app/repositories/neo4j/cypher_queries.py +++ b/api/app/repositories/neo4j/cypher_queries.py @@ -1,3 +1,4 @@ +from app.core.memory.enums import Neo4jNodeType DIALOGUE_NODE_SAVE = """ UNWIND $dialogues AS dialogue @@ -147,57 +148,6 @@ SET r.predicate = rel.predicate, RETURN elementId(r) AS uuid """ -# 在 Neo4j 5及后续版本中，id() 函数已被标记为弃用，用elementId() 函数替代 - -# 保存弱关系实体，设置 e.is_weak = true；不维护 e.relations 聚合字段 -WEAK_ENTITY_NODE_SAVE = """ -UNWIND $weak_entities AS entity -MERGE (e:ExtractedEntity {id: entity.id, run_id: entity.run_id}) -SET e += { - name: entity.name, - end_user_id: entity.end_user_id, - run_id: entity.run_id, - description: entity.description, - chunk_id: entity.chunk_id, - dialog_id: entity.dialog_id -} -// Independent weak flag，仅标记弱关系，不再维护 relations 聚合字段 -SET e.is_weak = true -RETURN e.id AS id -""" - -# 为强关系三元组中的主语和宾语创建/更新实体节点，仅设置 e.is_strong = true，不维护 e.relations 字段 -SAVE_STRONG_TRIPLE_ENTITIES = """ -UNWIND $items AS item -MERGE (s:ExtractedEntity {id: item.source_id, run_id: item.run_id}) -SET s += {name: item.subject, end_user_id: item.end_user_id, run_id: item.run_id} -// Independent strong flag -SET s.is_strong = true -MERGE (o:ExtractedEntity {id: item.target_id, run_id: item.run_id}) -SET o += {name: item.object, end_user_id: item.end_user_id, run_id: item.run_id} -// Independent strong flag -SET o.is_strong = true -""" - - -DIALOGUE_STATEMENT_EDGE_SAVE = """ - UNWIND $dialogue_statement_edges AS edge - // 支持按 uuid 或 ref_id 连接到 Dialogue，避免因来源 ID 不一致而断链 - MATCH (dialogue:Dialogue) - WHERE dialogue.uuid = edge.source OR dialogue.ref_id = edge.source - MATCH (statement:Statement {id: edge.target}) - // 仅按端点去重，关系属性可更新 - MERGE (dialogue)-[e:MENTIONS]->(statement) - SET e.uuid = edge.id, - e.end_user_id = edge.end_user_id, - e.created_at = edge.created_at, - e.expired_at = edge.expired_at - RETURN e.uuid AS uuid -""" - -# 在 Neo4j 5及后续版本中，id() 函数已被标记为弃用，用elementId() 函数替代 - - CHUNK_STATEMENT_EDGE_SAVE = """ UNWIND $chunk_statement_edges AS edge MATCH (statement:Statement {id: edge.source, run_id: edge.run_id}) @@ -226,87 +176,6 @@ SET r.end_user_id = rel.end_user_id, RETURN elementId(r) AS uuid """ -ENTITY_EMBEDDING_SEARCH = """ -CALL db.index.vector.queryNodes('entity_embedding_index', $limit * 100, $embedding) -YIELD node AS e, score -WHERE e.name_embedding IS NOT NULL - AND ($end_user_id IS NULL OR e.end_user_id = $end_user_id) -RETURN e.id AS id, - e.name AS name, - e.end_user_id AS end_user_id, - e.entity_type AS entity_type, - COALESCE(e.activation_value, e.importance_score, 0.5) AS activation_value, - COALESCE(e.importance_score, 0.5) AS importance_score, - e.last_access_time AS last_access_time, - COALESCE(e.access_count, 0) AS access_count, - score -ORDER BY score DESC -LIMIT $limit -""" -# Embedding-based search: cosine similarity on Statement.statement_embedding -STATEMENT_EMBEDDING_SEARCH = """ -CALL db.index.vector.queryNodes('statement_embedding_index', $limit * 100, $embedding) -YIELD node AS s, score -WHERE s.statement_embedding IS NOT NULL - AND ($end_user_id IS NULL OR s.end_user_id = $end_user_id) -RETURN s.id AS id, - s.statement AS statement, - s.end_user_id AS end_user_id, - s.chunk_id AS chunk_id, - s.created_at AS created_at, - s.expired_at AS expired_at, - s.valid_at AS valid_at, - s.invalid_at AS invalid_at, - COALESCE(s.activation_value, s.importance_score, 0.5) AS activation_value, - COALESCE(s.importance_score, 0.5) AS importance_score, - s.last_access_time AS last_access_time, - COALESCE(s.access_count, 0) AS access_count, - score -ORDER BY score DESC -LIMIT $limit -""" - -# Embedding-based search: cosine similarity on Chunk.chunk_embedding -CHUNK_EMBEDDING_SEARCH = """ -CALL db.index.vector.queryNodes('chunk_embedding_index', $limit * 100, $embedding) -YIELD node AS c, score -WHERE c.chunk_embedding IS NOT NULL - AND ($end_user_id IS NULL OR c.end_user_id = $end_user_id) -RETURN c.id AS chunk_id, - c.end_user_id AS end_user_id, - c.content AS content, - c.dialog_id AS dialog_id, - COALESCE(c.activation_value, 0.5) AS activation_value, - c.last_access_time AS last_access_time, - COALESCE(c.access_count, 0) AS access_count, - score -ORDER BY score DESC -LIMIT $limit -""" - -SEARCH_STATEMENTS_BY_KEYWORD = """ -CALL db.index.fulltext.queryNodes("statementsFulltext", $query) YIELD node AS s, score -WHERE ($end_user_id IS NULL OR s.end_user_id = $end_user_id) -OPTIONAL MATCH (c:Chunk)-[:CONTAINS]->(s) -OPTIONAL MATCH (s)-[:REFERENCES_ENTITY]->(e:ExtractedEntity) -RETURN s.id AS id, - s.statement AS statement, - s.end_user_id AS end_user_id, - s.chunk_id AS chunk_id, - s.created_at AS created_at, - s.expired_at AS expired_at, - s.valid_at AS valid_at, - s.invalid_at AS invalid_at, - c.id AS chunk_id_from_rel, - collect(DISTINCT e.id) AS entity_ids, - COALESCE(s.activation_value, s.importance_score, 0.5) AS activation_value, - COALESCE(s.importance_score, 0.5) AS importance_score, - s.last_access_time AS last_access_time, - COALESCE(s.access_count, 0) AS access_count, - score -ORDER BY score DESC -LIMIT $limit -""" # 查询实体名称包含指定字符串的实体 SEARCH_ENTITIES_BY_NAME = """ CALL db.index.fulltext.queryNodes("entitiesFulltext", $query) YIELD node AS e, score @@ -338,73 +207,6 @@ ORDER BY score DESC LIMIT $limit """ -SEARCH_ENTITIES_BY_NAME_OR_ALIAS = """ -CALL db.index.fulltext.queryNodes("entitiesFulltext", $query) YIELD node AS e, score -WHERE ($end_user_id IS NULL OR e.end_user_id = $end_user_id) -WITH e, score -With collect({entity: e, score: score}) AS fulltextResults - -OPTIONAL MATCH (ae:ExtractedEntity) -WHERE ($end_user_id IS NULL OR ae.end_user_id = $end_user_id) - AND ae.aliases IS NOT NULL - AND ANY(alias IN ae.aliases WHERE toLower(alias) CONTAINS toLower($query)) -WITH fulltextResults, collect(ae) AS aliasEntities - -UNWIND (fulltextResults + [x IN aliasEntities | {entity: x, score: - CASE - WHEN ANY(alias IN x.aliases WHERE toLower(alias) = toLower($query)) THEN 1.0 - WHEN ANY(alias IN x.aliases WHERE toLower(alias) STARTS WITH toLower($query)) THEN 0.9 - ELSE 0.8 - END -}]) AS row -WITH row.entity AS e, row.score AS score -WITH DISTINCT e, MAX(score) AS score -OPTIONAL MATCH (s:Statement)-[:REFERENCES_ENTITY]->(e) -OPTIONAL MATCH (c:Chunk)-[:CONTAINS]->(s) -RETURN e.id AS id, - e.name AS name, - e.end_user_id AS end_user_id, - e.entity_type AS entity_type, - e.created_at AS created_at, - e.expired_at AS expired_at, - e.entity_idx AS entity_idx, - e.statement_id AS statement_id, - e.description AS description, - e.aliases AS aliases, - e.name_embedding AS name_embedding, - e.connect_strength AS connect_strength, - collect(DISTINCT s.id) AS statement_ids, - collect(DISTINCT c.id) AS chunk_ids, - COALESCE(e.activation_value, e.importance_score, 0.5) AS activation_value, - COALESCE(e.importance_score, 0.5) AS importance_score, - e.last_access_time AS last_access_time, - COALESCE(e.access_count, 0) AS access_count, - score -ORDER BY score DESC -LIMIT $limit -""" - - -SEARCH_CHUNKS_BY_CONTENT = """ -CALL db.index.fulltext.queryNodes("chunksFulltext", $query) YIELD node AS c, score -WHERE ($end_user_id IS NULL OR c.end_user_id = $end_user_id) -OPTIONAL MATCH (c)-[:CONTAINS]->(s:Statement) -OPTIONAL MATCH (s)-[:REFERENCES_ENTITY]->(e:ExtractedEntity) -RETURN c.id AS chunk_id, - c.end_user_id AS end_user_id, - c.content AS content, - c.dialog_id AS dialog_id, - c.sequence_number AS sequence_number, - collect(DISTINCT s.id) AS statement_ids, - collect(DISTINCT e.id) AS entity_ids, - COALESCE(c.activation_value, 0.5) AS activation_value, - c.last_access_time AS last_access_time, - COALESCE(c.access_count, 0) AS access_count, - score -ORDER BY score DESC -LIMIT $limit -""" - # 以下是关于第二层去重消歧与数据库进行检索的语句，在最近的规划中不再使用 # # 同组group_id下按“精确名字或别名+可选类型一致”来检索 @@ -677,49 +479,6 @@ MATCH (n:Statement {end_user_id: $end_user_id, id: $id}) SET n.invalid_at = $new_invalid_at """ -# MemorySummary keyword search using fulltext index -SEARCH_MEMORY_SUMMARIES_BY_KEYWORD = """ -CALL db.index.fulltext.queryNodes("summariesFulltext", $query) YIELD node AS m, score -WHERE ($end_user_id IS NULL OR m.end_user_id = $end_user_id) -OPTIONAL MATCH (m)-[:DERIVED_FROM_STATEMENT]->(s:Statement) -RETURN m.id AS id, - m.name AS name, - m.end_user_id AS end_user_id, - m.dialog_id AS dialog_id, - m.chunk_ids AS chunk_ids, - m.content AS content, - m.created_at AS created_at, - COALESCE(m.activation_value, m.importance_score, 0.5) AS activation_value, - COALESCE(m.importance_score, 0.5) AS importance_score, - m.last_access_time AS last_access_time, - COALESCE(m.access_count, 0) AS access_count, - score -ORDER BY score DESC -LIMIT $limit -""" - -# Embedding-based search: cosine similarity on MemorySummary.summary_embedding -MEMORY_SUMMARY_EMBEDDING_SEARCH = """ -CALL db.index.vector.queryNodes('summary_embedding_index', $limit * 100, $embedding) -YIELD node AS m, score -WHERE m.summary_embedding IS NOT NULL - AND ($end_user_id IS NULL OR m.end_user_id = $end_user_id) -RETURN m.id AS id, - m.name AS name, - m.end_user_id AS end_user_id, - m.dialog_id AS dialog_id, - m.chunk_ids AS chunk_ids, - m.content AS content, - m.created_at AS created_at, - COALESCE(m.activation_value, m.importance_score, 0.5) AS activation_value, - COALESCE(m.importance_score, 0.5) AS importance_score, - m.last_access_time AS last_access_time, - COALESCE(m.access_count, 0) AS access_count, - score -ORDER BY score DESC -LIMIT $limit -""" - MEMORY_SUMMARY_NODE_SAVE = """ UNWIND $summaries AS summary MERGE (m:MemorySummary {id: summary.id}) @@ -1030,8 +789,6 @@ RETURN DISTINCT e.statement AS statement; """ -'''获取实体''' - Memory_Space_User = """ MATCH (n)-[r]->(m) WHERE n.end_user_id = $end_user_id AND m.name="用户" @@ -1363,22 +1120,6 @@ WHERE c.name IS NULL OR c.name = '' RETURN c.community_id AS community_id """ -# Community keyword search: matches name or summary via fulltext index -SEARCH_COMMUNITIES_BY_KEYWORD = """ -CALL db.index.fulltext.queryNodes("communitiesFulltext", $query) YIELD node AS c, score -WHERE ($end_user_id IS NULL OR c.end_user_id = $end_user_id) -RETURN c.community_id AS id, - c.name AS name, - c.summary AS content, - c.core_entities AS core_entities, - c.member_count AS member_count, - c.end_user_id AS end_user_id, - c.updated_at AS updated_at, - score -ORDER BY score DESC -LIMIT $limit -""" - # Community 向量检索 ────────────────────────────────────────────────── # Community embedding-based search: cosine similarity on Community.summary_embedding COMMUNITY_EMBEDDING_SEARCH = """ @@ -1452,13 +1193,54 @@ ON CREATE SET r.end_user_id = edge.end_user_id, RETURN elementId(r) AS uuid """ +# ------------------- +# search by user id +# ------------------- SEARCH_PERCEPTUAL_BY_USER_ID = """ MATCH (p:Perceptual) WHERE p.end_user_id = $end_user_id RETURN p.id AS id, - p.summary_embedding AS summary_embedding + p.summary_embedding AS embedding """ +SEARCH_STATEMENTS_BY_USER_ID = """ +MATCH (s:Statement) +WHERE s.end_user_id = $end_user_id +RETURN s.id AS id, + s.statement_embedding AS embedding +""" + +SEARCH_ENTITIES_BY_USER_ID = """ +MATCH (e:ExtractedEntity) +WHERE e.end_user_id = $end_user_id +RETURN e.id AS id, + e.name_embedding AS embedding +""" + +SEARCH_CHUNKS_BY_USER_ID = """ +MATCH (c:Chunk) +WHERE c.end_user_id = $end_user_id +RETURN c.id AS id, + c.chunk_embedding AS embedding +""" + +SEARCH_MEMORY_SUMMARIES_BY_USER_ID = """ +MATCH (s:MemorySummary) +WHERE s.end_user_id = $end_user_id +RETURN s.id AS id, + s.summary_embedding AS embedding +""" + +SEARCH_COMMUNITIES_BY_USER_ID = """ +MATCH (c:Community) +WHERE c.end_user_id = $end_user_id +RETURN c.id AS id, + c.summary_embedding AS embedding +""" + +# ------------------- +# search by id +# ------------------- SEARCH_PERCEPTUAL_BY_IDS = """ MATCH (p:Perceptual) WHERE p.id IN $ids @@ -1476,7 +1258,79 @@ RETURN p.id AS id, p.file_type AS file_type """ -SEARCH_PERCEPTUAL_BY_KEYWORD = """ +SEARCH_STATEMENTS_BY_IDS = """ +MATCH (s:Statement) +WHERE s.id IN $ids +RETURN s.id AS id, + s.statement AS statement, + s.end_user_id AS end_user_id, + s.chunk_id AS chunk_id, + s.created_at AS created_at, + s.expired_at AS expired_at, + s.valid_at AS valid_at, + properties(s)['invalid_at'] AS invalid_at, + COALESCE(s.activation_value, s.importance_score, 0.5) AS activation_value, + COALESCE(s.importance_score, 0.5) AS importance_score, + s.last_access_time AS last_access_time, + COALESCE(s.access_count, 0) AS access_count +""" + +SEARCH_CHUNKS_BY_IDS = """ +MATCH (c:Chunk) +WHERE c.id IN $ids +RETURN c.id AS id, + c.end_user_id AS end_user_id, + c.content AS content, + c.dialog_id AS dialog_id, + COALESCE(c.activation_value, 0.5) AS activation_value, + c.last_access_time AS last_access_time, + COALESCE(c.access_count, 0) AS access_count +""" + +SEARCH_ENTITIES_BY_IDS = """ +MATCH (e:ExtractedEntity) +WHERE e.id IN $ids +RETURN e.id AS id, + e.name AS name, + e.end_user_id AS end_user_id, + e.entity_type AS entity_type, + COALESCE(e.activation_value, e.importance_score, 0.5) AS activation_value, + COALESCE(e.importance_score, 0.5) AS importance_score, + e.last_access_time AS last_access_time, + COALESCE(e.access_count, 0) AS access_count +""" + +SEARCH_MEMORY_SUMMARIES_BY_IDS = """ +MATCH (m:MemorySummary) +WHERE m.id IN $ids +RETURN m.id AS id, + m.name AS name, + m.end_user_id AS end_user_id, + m.dialog_id AS dialog_id, + m.chunk_ids AS chunk_ids, + m.content AS content, + m.created_at AS created_at, + COALESCE(m.activation_value, m.importance_score, 0.5) AS activation_value, + COALESCE(m.importance_score, 0.5) AS importance_score, + m.last_access_time AS last_access_time, + COALESCE(m.access_count, 0) AS access_count +""" + +SEARCH_COMMUNITIES_BY_IDS = """ +MATCH (c:Community) +WHERE c.id IN $ids +RETURN c.id AS id, + c.name AS name, + c.summary AS content, + c.core_entities AS core_entities, + c.member_count AS member_count, + c.end_user_id AS end_user_id, + c.updated_at AS updated_at +""" +# ------------------- +# search by fulltext +# ------------------- +SEARCH_PERCEPTUALS_BY_KEYWORD = """ CALL db.index.fulltext.queryNodes("perceptualFulltext", $query) YIELD node AS p, score WHERE p.end_user_id = $end_user_id RETURN p.id AS id, @@ -1495,3 +1349,155 @@ RETURN p.id AS id, ORDER BY score DESC LIMIT $limit """ + +SEARCH_STATEMENTS_BY_KEYWORD = """ +CALL db.index.fulltext.queryNodes("statementsFulltext", $query) YIELD node AS s, score +WHERE ($end_user_id IS NULL OR s.end_user_id = $end_user_id) +OPTIONAL MATCH (c:Chunk)-[:CONTAINS]->(s) +OPTIONAL MATCH (s)-[:REFERENCES_ENTITY]->(e:ExtractedEntity) +RETURN s.id AS id, + s.statement AS statement, + s.end_user_id AS end_user_id, + s.chunk_id AS chunk_id, + s.created_at AS created_at, + s.expired_at AS expired_at, + s.valid_at AS valid_at, + properties(s)['invalid_at'] AS invalid_at, + c.id AS chunk_id_from_rel, + collect(DISTINCT e.id) AS entity_ids, + COALESCE(s.activation_value, s.importance_score, 0.5) AS activation_value, + COALESCE(s.importance_score, 0.5) AS importance_score, + s.last_access_time AS last_access_time, + COALESCE(s.access_count, 0) AS access_count, + score +ORDER BY score DESC +LIMIT $limit +""" + +SEARCH_ENTITIES_BY_NAME_OR_ALIAS = """ +CALL db.index.fulltext.queryNodes("entitiesFulltext", $query) YIELD node AS e, score +WHERE ($end_user_id IS NULL OR e.end_user_id = $end_user_id) +WITH e, score +With collect({entity: e, score: score}) AS fulltextResults + +OPTIONAL MATCH (ae:ExtractedEntity) +WHERE ($end_user_id IS NULL OR ae.end_user_id = $end_user_id) + AND ae.aliases IS NOT NULL + AND ANY(alias IN ae.aliases WHERE toLower(alias) CONTAINS toLower($query)) +WITH fulltextResults, collect(ae) AS aliasEntities + +UNWIND (fulltextResults + [x IN aliasEntities | {entity: x, score: + CASE + WHEN ANY(alias IN x.aliases WHERE toLower(alias) = toLower($query)) THEN 1.0 + WHEN ANY(alias IN x.aliases WHERE toLower(alias) STARTS WITH toLower($query)) THEN 0.9 + ELSE 0.8 + END +}]) AS row +WITH row.entity AS e, row.score AS score +WITH DISTINCT e, MAX(score) AS score +OPTIONAL MATCH (s:Statement)-[:REFERENCES_ENTITY]->(e) +OPTIONAL MATCH (c:Chunk)-[:CONTAINS]->(s) +RETURN e.id AS id, + e.name AS name, + e.end_user_id AS end_user_id, + e.entity_type AS entity_type, + e.created_at AS created_at, + e.expired_at AS expired_at, + e.entity_idx AS entity_idx, + e.statement_id AS statement_id, + e.description AS description, + e.aliases AS aliases, + e.name_embedding AS name_embedding, + e.connect_strength AS connect_strength, + collect(DISTINCT s.id) AS statement_ids, + collect(DISTINCT c.id) AS chunk_ids, + COALESCE(e.activation_value, e.importance_score, 0.5) AS activation_value, + COALESCE(e.importance_score, 0.5) AS importance_score, + e.last_access_time AS last_access_time, + COALESCE(e.access_count, 0) AS access_count, + score +ORDER BY score DESC +LIMIT $limit +""" + +SEARCH_CHUNKS_BY_CONTENT = """ +CALL db.index.fulltext.queryNodes("chunksFulltext", $query) YIELD node AS c, score +WHERE ($end_user_id IS NULL OR c.end_user_id = $end_user_id) +OPTIONAL MATCH (c)-[:CONTAINS]->(s:Statement) +OPTIONAL MATCH (s)-[:REFERENCES_ENTITY]->(e:ExtractedEntity) +RETURN c.id AS id, + c.end_user_id AS end_user_id, + c.content AS content, + c.dialog_id AS dialog_id, + c.sequence_number AS sequence_number, + collect(DISTINCT s.id) AS statement_ids, + collect(DISTINCT e.id) AS entity_ids, + COALESCE(c.activation_value, 0.5) AS activation_value, + c.last_access_time AS last_access_time, + COALESCE(c.access_count, 0) AS access_count, + score +ORDER BY score DESC +LIMIT $limit +""" + +# MemorySummary keyword search using fulltext index +SEARCH_MEMORY_SUMMARIES_BY_KEYWORD = """ +CALL db.index.fulltext.queryNodes("summariesFulltext", $query) YIELD node AS m, score +WHERE ($end_user_id IS NULL OR m.end_user_id = $end_user_id) +OPTIONAL MATCH (m)-[:DERIVED_FROM_STATEMENT]->(s:Statement) +RETURN m.id AS id, + m.name AS name, + m.end_user_id AS end_user_id, + m.dialog_id AS dialog_id, + m.chunk_ids AS chunk_ids, + m.content AS content, + m.created_at AS created_at, + COALESCE(m.activation_value, m.importance_score, 0.5) AS activation_value, + COALESCE(m.importance_score, 0.5) AS importance_score, + m.last_access_time AS last_access_time, + COALESCE(m.access_count, 0) AS access_count, + score +ORDER BY score DESC +LIMIT $limit +""" + +# Community keyword search: matches name or summary via fulltext index +SEARCH_COMMUNITIES_BY_KEYWORD = """ +CALL db.index.fulltext.queryNodes("communitiesFulltext", $query) YIELD node AS c, score +WHERE ($end_user_id IS NULL OR c.end_user_id = $end_user_id) +RETURN c.id AS id, + c.name AS name, + c.summary AS content, + c.core_entities AS core_entities, + c.member_count AS member_count, + c.end_user_id AS end_user_id, + c.updated_at AS updated_at, + score +ORDER BY score DESC +LIMIT $limit +""" + +FULLTEXT_QUERY_CYPHER_MAPPING = { + Neo4jNodeType.STATEMENT: SEARCH_STATEMENTS_BY_KEYWORD, + Neo4jNodeType.EXTRACTEDENTITY: SEARCH_ENTITIES_BY_NAME_OR_ALIAS, + Neo4jNodeType.CHUNK: SEARCH_CHUNKS_BY_CONTENT, + Neo4jNodeType.MEMORYSUMMARY: SEARCH_MEMORY_SUMMARIES_BY_KEYWORD, + Neo4jNodeType.COMMUNITY: SEARCH_COMMUNITIES_BY_KEYWORD, + Neo4jNodeType.PERCEPTUAL: SEARCH_PERCEPTUALS_BY_KEYWORD +} +USER_ID_QUERY_CYPHER_MAPPING = { + Neo4jNodeType.STATEMENT: SEARCH_STATEMENTS_BY_USER_ID, + Neo4jNodeType.EXTRACTEDENTITY: SEARCH_ENTITIES_BY_USER_ID, + Neo4jNodeType.CHUNK: SEARCH_CHUNKS_BY_USER_ID, + Neo4jNodeType.MEMORYSUMMARY: SEARCH_MEMORY_SUMMARIES_BY_USER_ID, + Neo4jNodeType.COMMUNITY: SEARCH_COMMUNITIES_BY_USER_ID, + Neo4jNodeType.PERCEPTUAL: SEARCH_PERCEPTUAL_BY_USER_ID +} +NODE_ID_QUERY_CYPHER_MAPPING = { + Neo4jNodeType.STATEMENT: SEARCH_STATEMENTS_BY_IDS, + Neo4jNodeType.EXTRACTEDENTITY: SEARCH_ENTITIES_BY_IDS, + Neo4jNodeType.CHUNK: SEARCH_CHUNKS_BY_IDS, + Neo4jNodeType.MEMORYSUMMARY: SEARCH_MEMORY_SUMMARIES_BY_IDS, + Neo4jNodeType.COMMUNITY: SEARCH_COMMUNITIES_BY_IDS, + Neo4jNodeType.PERCEPTUAL: SEARCH_PERCEPTUAL_BY_IDS +} diff --git a/api/app/repositories/neo4j/graph_search.py b/api/app/repositories/neo4j/graph_search.py index 665b68ff..336f4134 100644 --- a/api/app/repositories/neo4j/graph_search.py +++ b/api/app/repositories/neo4j/graph_search.py @@ -1,26 +1,19 @@ import asyncio import logging -from typing import Any, Dict, List, Optional +import time +from typing import Any, Dict, List, Optional, Coroutine -from app.core.memory.utils.data.text_utils import escape_lucene_query import numpy as np +from app.core.memory.enums import Neo4jNodeType from app.core.memory.llm_tools import OpenAIEmbedderClient +from app.core.memory.utils.data.text_utils import escape_lucene_query from app.repositories.neo4j.cypher_queries import ( - CHUNK_EMBEDDING_SEARCH, - COMMUNITY_EMBEDDING_SEARCH, - ENTITY_EMBEDDING_SEARCH, EXPAND_COMMUNITY_STATEMENTS, - MEMORY_SUMMARY_EMBEDDING_SEARCH, SEARCH_CHUNK_BY_CHUNK_ID, - SEARCH_CHUNKS_BY_CONTENT, - SEARCH_COMMUNITIES_BY_KEYWORD, SEARCH_DIALOGUE_BY_DIALOG_ID, SEARCH_ENTITIES_BY_NAME, - SEARCH_ENTITIES_BY_NAME_OR_ALIAS, - SEARCH_MEMORY_SUMMARIES_BY_KEYWORD, SEARCH_STATEMENTS_BY_CREATED_AT, - SEARCH_STATEMENTS_BY_KEYWORD, SEARCH_STATEMENTS_BY_KEYWORD_TEMPORAL, SEARCH_STATEMENTS_BY_TEMPORAL, SEARCH_STATEMENTS_BY_VALID_AT, @@ -28,12 +21,14 @@ from app.repositories.neo4j.cypher_queries import ( SEARCH_STATEMENTS_G_VALID_AT, SEARCH_STATEMENTS_L_CREATED_AT, SEARCH_STATEMENTS_L_VALID_AT, - STATEMENT_EMBEDDING_SEARCH, - SEARCH_PERCEPTUAL_BY_KEYWORD, + SEARCH_PERCEPTUALS_BY_KEYWORD, SEARCH_PERCEPTUAL_BY_IDS, SEARCH_PERCEPTUAL_BY_USER_ID, + FULLTEXT_QUERY_CYPHER_MAPPING, + USER_ID_QUERY_CYPHER_MAPPING, + NODE_ID_QUERY_CYPHER_MAPPING ) -# 使用新的仓储层 + from app.repositories.neo4j.neo4j_connector import Neo4jConnector logger = logging.getLogger(__name__) @@ -52,7 +47,7 @@ def cosine_similarity_search( query_norm = query / np.linalg.norm(query) similarities = vectors_norm @ query_norm - similarities = (similarities + 1) / 2 + similarities = np.clip(similarities, 0, 1) top_k = min(limit, similarities.shape[0]) if top_k <= 0: return {} @@ -60,7 +55,7 @@ def cosine_similarity_search( top_indices = top_indices[np.argsort(-similarities[top_indices])] result = {} for idx in top_indices: - result[idx] = similarities[idx] + result[idx] = float(similarities[idx]) return result @@ -173,7 +168,10 @@ async def _update_search_results_activation( knowledge_node_types = { 'statements': 'Statement', 'entities': 'ExtractedEntity', - 'summaries': 'MemorySummary' + 'summaries': 'MemorySummary', + Neo4jNodeType.STATEMENT: Neo4jNodeType.STATEMENT.value, + Neo4jNodeType.EXTRACTEDENTITY: Neo4jNodeType.EXTRACTEDENTITY.value, + Neo4jNodeType.MEMORYSUMMARY: Neo4jNodeType.MEMORYSUMMARY.value, } # 并行更新所有类型的节点 @@ -250,12 +248,147 @@ async def _update_search_results_activation( return updated_results +async def search_perceptual_by_fulltext( + connector: Neo4jConnector, + query: str, + end_user_id: Optional[str] = None, + limit: int = 10, +) -> Dict[str, List[Dict[str, Any]]]: + try: + perceptuals = await connector.execute_query( + SEARCH_PERCEPTUALS_BY_KEYWORD, + query=escape_lucene_query(query), + end_user_id=end_user_id, + limit=limit, + ) + except Exception as e: + logger.warning(f"search_perceptual: keyword search failed: {e}") + perceptuals = [] + + # Deduplicate + from app.core.memory.src.search import deduplicate_results + perceptuals = deduplicate_results(perceptuals) + + return {"perceptuals": perceptuals} + + +async def search_perceptual_by_embedding( + connector: Neo4jConnector, + embedder_client: OpenAIEmbedderClient, + query_text: str, + end_user_id: Optional[str] = None, + limit: int = 10, +) -> Dict[str, List[Dict[str, Any]]]: + """ + Search Perceptual memory nodes using embedding-based semantic search. + + Uses cosine similarity on summary_embedding via the perceptual_summary_embedding_index. + + Args: + connector: Neo4j connector + embedder_client: Embedding client with async response() method + query_text: Query text to embed + end_user_id: Optional user filter + limit: Max results + + Returns: + Dictionary with 'perceptuals' key containing matched perceptual memory nodes + """ + embeddings = await embedder_client.response([query_text]) + if not embeddings or not embeddings[0]: + logger.warning(f"search_perceptual_by_embedding: embedding generation failed for '{query_text[:50]}'") + return {"perceptuals": []} + + embedding = embeddings[0] + + try: + perceptuals = await connector.execute_query( + SEARCH_PERCEPTUAL_BY_USER_ID, + end_user_id=end_user_id, + ) + ids = [item['id'] for item in perceptuals] + vectors = [item['summary_embedding'] for item in perceptuals] + sim_res = cosine_similarity_search(embedding, vectors, limit=limit) + perceptual_res = { + ids[idx]: score + for idx, score in sim_res.items() + } + perceptuals = await connector.execute_query( + SEARCH_PERCEPTUAL_BY_IDS, + ids=list(perceptual_res.keys()) + ) + for perceptual in perceptuals: + perceptual["score"] = perceptual_res[perceptual["id"]] + except Exception as e: + logger.warning(f"search_perceptual_by_embedding: vector search failed: {e}") + perceptuals = [] + + from app.core.memory.src.search import deduplicate_results + perceptuals = deduplicate_results(perceptuals) + + return {"perceptuals": perceptuals} + + +def search_by_fulltext( + connector: Neo4jConnector, + node_type: Neo4jNodeType, + end_user_id: str, + query: str, + limit: int = 10, +) -> Coroutine[Any, Any, list[dict[str, Any]]]: + cypher = FULLTEXT_QUERY_CYPHER_MAPPING[node_type] + return connector.execute_query( + cypher, + json_format=True, + end_user_id=end_user_id, + query=query, + limit=limit, + ) + + +async def search_by_embedding( + connector: Neo4jConnector, + node_type: Neo4jNodeType, + end_user_id: str, + query_embedding: list[float], + limit: int = 10, +) -> list[dict[str, Any]]: + try: + records = await connector.execute_query( + USER_ID_QUERY_CYPHER_MAPPING[node_type], + end_user_id=end_user_id, + ) + records = [record for record in records if record if record["embedding"] is not None] + ids = [item['id'] for item in records] + vectors = [item['embedding'] for item in records] + sim_res = cosine_similarity_search(query_embedding, vectors, limit=limit) + records_score_map = { + ids[idx]: score + for idx, score in sim_res.items() + } + records = await connector.execute_query( + NODE_ID_QUERY_CYPHER_MAPPING[node_type], + ids=list(records_score_map.keys()), + json_format=True + ) + for record in records: + record["score"] = records_score_map[record["id"]] + except Exception as e: + logger.warning(f"search_graph_by_embedding: vector search failed: {e}, node_type:{node_type.value}", + exc_info=True) + records = [] + + from app.core.memory.src.search import deduplicate_results + records = deduplicate_results(records) + return records + + async def search_graph( connector: Neo4jConnector, query: str, end_user_id: Optional[str] = None, limit: int = 50, - include: List[str] = None, + include: List[Neo4jNodeType] = None, ) -> Dict[str, List[Dict[str, Any]]]: """ Search across Statements, Entities, Chunks, and Summaries using a free-text query. @@ -279,7 +412,13 @@ async def search_graph( Dictionary with search results per category (with updated activation values) """ if include is None: - include = ["statements", "chunks", "entities", "summaries"] + include = [ + Neo4jNodeType.STATEMENT, + Neo4jNodeType.CHUNK, + Neo4jNodeType.EXTRACTEDENTITY, + Neo4jNodeType.MEMORYSUMMARY, + Neo4jNodeType.PERCEPTUAL + ] # Escape Lucene special characters to prevent query parse errors escaped_query = escape_lucene_query(query) @@ -288,55 +427,9 @@ async def search_graph( tasks = [] task_keys = [] - if "statements" in include: - tasks.append(connector.execute_query( - SEARCH_STATEMENTS_BY_KEYWORD, - json_format=True, - query=escaped_query, - end_user_id=end_user_id, - limit=limit, - )) - task_keys.append("statements") - - if "entities" in include: - tasks.append(connector.execute_query( - SEARCH_ENTITIES_BY_NAME_OR_ALIAS, - json_format=True, - query=escaped_query, - end_user_id=end_user_id, - limit=limit, - )) - task_keys.append("entities") - - if "chunks" in include: - tasks.append(connector.execute_query( - SEARCH_CHUNKS_BY_CONTENT, - json_format=True, - query=escaped_query, - end_user_id=end_user_id, - limit=limit, - )) - task_keys.append("chunks") - - if "summaries" in include: - tasks.append(connector.execute_query( - SEARCH_MEMORY_SUMMARIES_BY_KEYWORD, - json_format=True, - query=escaped_query, - end_user_id=end_user_id, - limit=limit, - )) - task_keys.append("summaries") - - if "communities" in include: - tasks.append(connector.execute_query( - SEARCH_COMMUNITIES_BY_KEYWORD, - json_format=True, - query=escaped_query, - end_user_id=end_user_id, - limit=limit, - )) - task_keys.append("communities") + for node_type in include: + tasks.append(search_by_fulltext(connector, node_type, end_user_id, escaped_query, limit)) + task_keys.append(node_type.value) # Execute all queries in parallel task_results = await asyncio.gather(*tasks, return_exceptions=True) @@ -352,16 +445,16 @@ async def search_graph( # Deduplicate results before updating activation values # This prevents duplicates from propagating through the pipeline - from app.core.memory.src.search import _deduplicate_results + from app.core.memory.src.search import deduplicate_results for key in results: if isinstance(results[key], list): - results[key] = _deduplicate_results(results[key]) + results[key] = deduplicate_results(results[key]) # 更新知识节点的激活值（Statement, ExtractedEntity, MemorySummary） # Skip activation updates if only searching summaries (optimization) needs_activation_update = any( key in include and key in results and results[key] - for key in ['statements', 'entities', 'chunks'] + for key in [Neo4jNodeType.STATEMENT, Neo4jNodeType.EXTRACTEDENTITY, Neo4jNodeType.MEMORYSUMMARY] ) if needs_activation_update: @@ -378,7 +471,7 @@ async def search_graph_by_embedding( connector: Neo4jConnector, embedder_client, query_text: str, - end_user_id: Optional[str] = None, + end_user_id: str, limit: int = 50, include=None, ) -> Dict[str, List[Dict[str, Any]]]: @@ -394,96 +487,32 @@ async def search_graph_by_embedding( - Returns up to 'limit' per included type """ if include is None: - include = ["statements", "chunks", "entities", "summaries"] - import time + include = [ + Neo4jNodeType.STATEMENT, + Neo4jNodeType.CHUNK, + Neo4jNodeType.EXTRACTEDENTITY, + Neo4jNodeType.MEMORYSUMMARY, + Neo4jNodeType.PERCEPTUAL + ] - # Get embedding for the query - embed_start = time.time() embeddings = await embedder_client.response([query_text]) - embed_time = time.time() - embed_start - logger.debug(f"[PERF] Embedding generation took: {embed_time:.4f}s") - if not embeddings or not embeddings[0]: - logger.warning( - f"search_graph_by_embedding: embedding 生成失败或为空，" - f"query='{query_text[:50]}', end_user_id={end_user_id}，向量检索跳过" - ) - return {"statements": [], "chunks": [], "entities": [], "summaries": [], "communities": []} + logger.warning(f"search_graph_by_embedding: embedding generation failed for '{query_text[:50]}'") + return {search_key: [] for search_key in include} embedding = embeddings[0] # Prepare tasks for parallel execution tasks = [] task_keys = [] - # Statements (embedding) - if "statements" in include: - tasks.append(connector.execute_query( - STATEMENT_EMBEDDING_SEARCH, - json_format=True, - embedding=embedding, - end_user_id=end_user_id, - limit=limit, - )) - task_keys.append("statements") + for node_type in include: + tasks.append(search_by_embedding(connector, node_type, end_user_id, embedding, limit)) + task_keys.append(node_type.value) - # Chunks (embedding) - if "chunks" in include: - tasks.append(connector.execute_query( - CHUNK_EMBEDDING_SEARCH, - json_format=True, - embedding=embedding, - end_user_id=end_user_id, - limit=limit, - )) - task_keys.append("chunks") - - # Entities - if "entities" in include: - tasks.append(connector.execute_query( - ENTITY_EMBEDDING_SEARCH, - json_format=True, - embedding=embedding, - end_user_id=end_user_id, - limit=limit, - )) - task_keys.append("entities") - - # Memory summaries - if "summaries" in include: - tasks.append(connector.execute_query( - MEMORY_SUMMARY_EMBEDDING_SEARCH, - json_format=True, - embedding=embedding, - end_user_id=end_user_id, - limit=limit, - )) - task_keys.append("summaries") - - # Communities (向量语义匹配) - if "communities" in include: - tasks.append(connector.execute_query( - COMMUNITY_EMBEDDING_SEARCH, - json_format=True, - embedding=embedding, - end_user_id=end_user_id, - limit=limit, - )) - task_keys.append("communities") - - # Execute all queries in parallel - query_start = time.time() task_results = await asyncio.gather(*tasks, return_exceptions=True) - query_time = time.time() - query_start - logger.debug(f"[PERF] Neo4j queries (parallel) took: {query_time:.4f}s") # Build results dictionary - results: Dict[str, List[Dict[str, Any]]] = { - "statements": [], - "chunks": [], - "entities": [], - "summaries": [], - "communities": [], - } + results: Dict[str, List[Dict[str, Any]]] = {} for key, result in zip(task_keys, task_results): if isinstance(result, Exception): @@ -494,16 +523,16 @@ async def search_graph_by_embedding( # Deduplicate results before updating activation values # This prevents duplicates from propagating through the pipeline - from app.core.memory.src.search import _deduplicate_results + from app.core.memory.src.search import deduplicate_results for key in results: if isinstance(results[key], list): - results[key] = _deduplicate_results(results[key]) + results[key] = deduplicate_results(results[key]) # 更新知识节点的激活值（Statement, ExtractedEntity, MemorySummary） # Skip activation updates if only searching summaries (optimization) needs_activation_update = any( key in include and key in results and results[key] - for key in ['statements', 'entities', 'chunks'] + for key in [Neo4jNodeType.STATEMENT, Neo4jNodeType.EXTRACTEDENTITY, Neo4jNodeType.MEMORYSUMMARY] ) if needs_activation_update: @@ -781,12 +810,12 @@ async def search_graph_community_expand( expanded.extend(result) # 按 activation_value 全局排序后去重 - from app.core.memory.src.search import _deduplicate_results + from app.core.memory.src.search import deduplicate_results expanded.sort( key=lambda x: float(x.get("activation_value") or 0), reverse=True, ) - expanded = _deduplicate_results(expanded) + expanded = deduplicate_results(expanded) logger.info(f"社区展开检索完成: community_ids={community_ids}, 展开 statements={len(expanded)}") return {"expanded_statements": expanded} @@ -999,98 +1028,3 @@ async def search_graph_l_valid_at( ) return results - - -async def search_perceptual( - connector: Neo4jConnector, - query: str, - end_user_id: Optional[str] = None, - limit: int = 10, -) -> Dict[str, List[Dict[str, Any]]]: - """ - Search Perceptual memory nodes using fulltext keyword search. - - Matches against summary, topic, and domain fields via the perceptualFulltext index. - - Args: - connector: Neo4j connector - query: Query text for full-text search - end_user_id: Optional user filter - limit: Max results - - Returns: - Dictionary with 'perceptuals' key containing matched perceptual memory nodes - """ - try: - perceptuals = await connector.execute_query( - SEARCH_PERCEPTUAL_BY_KEYWORD, - query=escape_lucene_query(query), - end_user_id=end_user_id, - limit=limit, - ) - except Exception as e: - logger.warning(f"search_perceptual: keyword search failed: {e}") - perceptuals = [] - - # Deduplicate - from app.core.memory.src.search import _deduplicate_results - perceptuals = _deduplicate_results(perceptuals) - - return {"perceptuals": perceptuals} - - -async def search_perceptual_by_embedding( - connector: Neo4jConnector, - embedder_client: OpenAIEmbedderClient, - query_text: str, - end_user_id: Optional[str] = None, - limit: int = 10, -) -> Dict[str, List[Dict[str, Any]]]: - """ - Search Perceptual memory nodes using embedding-based semantic search. - - Uses cosine similarity on summary_embedding via the perceptual_summary_embedding_index. - - Args: - connector: Neo4j connector - embedder_client: Embedding client with async response() method - query_text: Query text to embed - end_user_id: Optional user filter - limit: Max results - - Returns: - Dictionary with 'perceptuals' key containing matched perceptual memory nodes - """ - embeddings = await embedder_client.response([query_text]) - if not embeddings or not embeddings[0]: - logger.warning(f"search_perceptual_by_embedding: embedding generation failed for '{query_text[:50]}'") - return {"perceptuals": []} - - embedding = embeddings[0] - - try: - perceptuals = await connector.execute_query( - SEARCH_PERCEPTUAL_BY_USER_ID, - end_user_id=end_user_id, - ) - ids = [item['id'] for item in perceptuals] - vectors = [item['summary_embedding'] for item in perceptuals] - sim_res = cosine_similarity_search(embedding, vectors, limit=limit) - perceptual_res = { - ids[idx]: score - for idx, score in sim_res.items() - } - perceptuals = await connector.execute_query( - SEARCH_PERCEPTUAL_BY_IDS, - ids=list(perceptual_res.keys()) - ) - for perceptual in perceptuals: - perceptual["score"] = perceptual_res[perceptual["id"]] - except Exception as e: - logger.warning(f"search_perceptual_by_embedding: vector search failed: {e}") - perceptuals = [] - - from app.core.memory.src.search import _deduplicate_results - perceptuals = _deduplicate_results(perceptuals) - - return {"perceptuals": perceptuals} diff --git a/api/app/repositories/neo4j/neo4j_connector.py b/api/app/repositories/neo4j/neo4j_connector.py index ea8fa917..cd9dfe03 100644 --- a/api/app/repositories/neo4j/neo4j_connector.py +++ b/api/app/repositories/neo4j/neo4j_connector.py @@ -70,6 +70,12 @@ class Neo4jConnector: auth=basic_auth(username, password) ) + async def __aenter__(self): + return self + + async def __aexit__(self, exc_type, exc_val, exc_tb): + await self.close() + async def close(self): """关闭数据库连接 @@ -77,11 +83,11 @@ class Neo4jConnector: """ await self.driver.close() - async def execute_query(self, query: str, json_format=False, **kwargs: Any) -> List[Dict[str, Any]]: + async def execute_query(self, cypher: str, json_format=False, **kwargs: Any) -> List[Dict[str, Any]]: """执行Cypher查询 Args: - query: Cypher查询语句 + cypher: Cypher查询语句 json_format: json格式化 **kwargs: 查询参数，将作为参数传递给Cypher查询 @@ -92,7 +98,7 @@ class Neo4jConnector: """ result = await self.driver.execute_query( - query, + cypher, database="neo4j", **kwargs ) From 737858731b132d3dd92c418560d84cac098d0bd5 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Wed, 15 Apr 2026 12:24:11 +0800 Subject: [PATCH 060/245] fix(core): conditionally apply thinking parameters based on model support --- api/app/core/models/base.py | 33 +++++++++++++++++---------------- 1 file changed, 17 insertions(+), 16 deletions(-) diff --git a/api/app/core/models/base.py b/api/app/core/models/base.py index 1de4b120..89a7dcee 100644 --- a/api/app/core/models/base.py +++ b/api/app/core/models/base.py @@ -112,22 +112,23 @@ class RedBearModelFactory: params["stream_usage"] = True # 深度思考模式 is_streaming = bool(config.extra_params.get("streaming")) - if is_streaming and not config.is_omni: - if provider == ModelProvider.VOLCANO: - # 火山引擎深度思考仅流式调用支持，非流式时不传 thinking 参数 - thinking_config: Dict[str, Any] = { - "type": "enabled" if config.deep_thinking else "disabled" - } - if config.deep_thinking and config.thinking_budget_tokens: - thinking_config["budget_tokens"] = config.thinking_budget_tokens - params["extra_body"] = {"thinking": thinking_config} - else: - # 始终显式传递 enable_thinking，不支持该参数的模型（如 DeepSeek-R1）会直接忽略 - model_kwargs: Dict[str, Any] = config.extra_params.get("model_kwargs", {}) - model_kwargs["enable_thinking"] = config.deep_thinking - if config.deep_thinking and config.thinking_budget_tokens: - model_kwargs["thinking_budget"] = config.thinking_budget_tokens - params["model_kwargs"] = model_kwargs + if config.support_thinking: + if is_streaming and not config.is_omni: + if provider == ModelProvider.VOLCANO: + # 火山引擎深度思考仅流式调用支持，非流式时不传 thinking 参数 + thinking_config: Dict[str, Any] = { + "type": "enabled" if config.deep_thinking else "disabled" + } + if config.deep_thinking and config.thinking_budget_tokens: + thinking_config["budget_tokens"] = config.thinking_budget_tokens + params["extra_body"] = {"thinking": thinking_config} + else: + # 始终显式传递 enable_thinking，不支持该参数的模型（如 DeepSeek-R1）会直接忽略 + model_kwargs: Dict[str, Any] = config.extra_params.get("model_kwargs", {}) + model_kwargs["enable_thinking"] = config.deep_thinking + if config.deep_thinking and config.thinking_budget_tokens: + model_kwargs["thinking_budget"] = config.thinking_budget_tokens + params["model_kwargs"] = model_kwargs return params elif provider == ModelProvider.DASHSCOPE: params = { From 3018d186f73c26eda25b1acf6ba95d283c933cfc Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Wed, 15 Apr 2026 13:56:08 +0800 Subject: [PATCH 061/245] fix(custom-tools): remove parameter coercion in custom tool base class --- api/app/core/tools/custom/base.py | 27 +++++++-------------------- 1 file changed, 7 insertions(+), 20 deletions(-) diff --git a/api/app/core/tools/custom/base.py b/api/app/core/tools/custom/base.py index 93ad03a7..c03fe206 100644 --- a/api/app/core/tools/custom/base.py +++ b/api/app/core/tools/custom/base.py @@ -221,7 +221,7 @@ class CustomTool(BaseTool): query_params = {} for param_name, param_info in operation.get("parameters", {}).items(): if param_info.get("in") == "query" and param_name in params: - query_params[param_name] = self._coerce_param(params[param_name], param_info.get("type", "string")) + query_params[param_name] = params[param_name] if query_params: from urllib.parse import urlencode @@ -251,34 +251,21 @@ class CustomTool(BaseTool): return headers @staticmethod - def _coerce_param(value: Any, schema_type: str) -> Any: - """根据 schema 类型转换参数值""" - if value is None: - return value - try: - if schema_type == "integer": - return int(value) - elif schema_type == "number": - return float(value) - elif schema_type == "boolean": - if isinstance(value, str): - return value.lower() not in ("false", "0", "") - return bool(value) - except (ValueError, TypeError): - pass - return value - - def _build_request_data(self, operation: Dict[str, Any], params: Dict[str, Any]) -> Optional[Dict[str, Any]]: + def _build_request_data(operation: Dict[str, Any], params: Dict[str, Any]) -> Optional[Dict[str, Any]]: """构建请求数据""" if operation["method"] in ["POST", "PUT", "PATCH"]: request_body = operation.get("request_body") if request_body: + # 构建请求体数据 data = {} properties = request_body.get("properties", {}) + for prop_name, prop_schema in properties.items(): if prop_name in params: - data[prop_name] = self._coerce_param(params[prop_name], prop_schema.get("type", "string")) + data[prop_name] = params[prop_name] + return data if data else None + return None async def _send_http_request( From ed765b7c262712ba19a6a303c9838f72d857ff17 Mon Sep 17 00:00:00 2001 From: Eternity <1533512157@qq.com> Date: Wed, 15 Apr 2026 13:19:02 +0800 Subject: [PATCH 062/245] fix(prompt-optimizer): handle escaped quotes in JSON parsing --- api/app/controllers/prompt_optimizer_controller.py | 5 +++-- api/app/services/prompt_optimizer_service.py | 7 +++++-- 2 files changed, 8 insertions(+), 4 deletions(-) diff --git a/api/app/controllers/prompt_optimizer_controller.py b/api/app/controllers/prompt_optimizer_controller.py index 80f14cd3..b9fc697c 100644 --- a/api/app/controllers/prompt_optimizer_controller.py +++ b/api/app/controllers/prompt_optimizer_controller.py @@ -124,10 +124,11 @@ async def get_prompt_opt( skill=data.skill ): # chunk 是 prompt 的增量内容 - yield f"event:message\ndata: {json.dumps(chunk)}\n\n" + yield f"event:message\ndata: {json.dumps(chunk, ensure_ascii=False)}\n\n" except Exception as e: yield f"event:error\ndata: {json.dumps( - {"error": str(e)} + {"error": str(e)}, + ensure_ascii=False )}\n\n" yield "event:end\ndata: {}\n\n" diff --git a/api/app/services/prompt_optimizer_service.py b/api/app/services/prompt_optimizer_service.py index b1de84d2..30901111 100644 --- a/api/app/services/prompt_optimizer_service.py +++ b/api/app/services/prompt_optimizer_service.py @@ -236,8 +236,11 @@ class PromptOptimizerService: logger.error(f"Unsupported content type - {content}") raise Exception("Unsupported content type") cache = buffer[:-20] + last_idx = 19 + while cache and cache[-1] == '\\' and last_idx > 0: + cache = buffer[:-last_idx] + last_idx -= 1 - # 尝试找到 "prompt": " 开始位置 if prompt_finished: continue @@ -279,7 +282,7 @@ class PromptOptimizerService: def parser_prompt_variables(prompt: str): try: pattern = r'\{\{\s*([a-zA-Z_][a-zA-Z0-9_]*)\s*\}\}' - matches = re.findall(pattern, prompt) + matches = re.findall(pattern, str(prompt)) variables = list(set(matches)) return variables except Exception as e: From 71e5b6586a1186b88d81e90ba1d02bfe41db6b98 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 15 Apr 2026 14:38:40 +0800 Subject: [PATCH 063/245] fix(web): prompt editor --- web/src/components/Chat/PromptChatPanel.tsx | 39 ++++++++ .../components/AiPromptModal.tsx | 56 +++++------ .../components/Editor/index.tsx | 93 +++++++++++++------ web/src/views/Prompt/index.tsx | 59 ++++++------ 4 files changed, 165 insertions(+), 82 deletions(-) create mode 100644 web/src/components/Chat/PromptChatPanel.tsx diff --git a/web/src/components/Chat/PromptChatPanel.tsx b/web/src/components/Chat/PromptChatPanel.tsx new file mode 100644 index 00000000..51343ae1 --- /dev/null +++ b/web/src/components/Chat/PromptChatPanel.tsx @@ -0,0 +1,39 @@ +import { forwardRef, useImperativeHandle, useState } from 'react' +import ChatContent from './ChatContent' +import type { ChatItem } from './types' +import type { ReactNode } from 'react' + +export interface PromptChatPanelRef { + append: (item: ChatItem) => void + clear: () => void +} + +interface PromptChatPanelProps { + classNames?: string + contentClassNames?: string + empty: ReactNode + labelFormat: (item: ChatItem) => any +} + +const PromptChatPanel = forwardRef((props, ref) => { + const [chatList, setChatList] = useState([]) + + useImperativeHandle(ref, () => ({ + append: (item) => setChatList(prev => [...prev, item]), + clear: () => setChatList([]), + })) + + return ( + + ) +}) + +export default PromptChatPanel diff --git a/web/src/views/ApplicationConfig/components/AiPromptModal.tsx b/web/src/views/ApplicationConfig/components/AiPromptModal.tsx index 1666e075..fd2dc595 100644 --- a/web/src/views/ApplicationConfig/components/AiPromptModal.tsx +++ b/web/src/views/ApplicationConfig/components/AiPromptModal.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 16:26:44 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-20 13:53:05 + * @Last Modified time: 2026-04-15 14:21:55 */ /** * AI Prompt Assistant Modal @@ -20,10 +20,9 @@ import { updatePromptMessages, createPromptSessions } from '@/api/prompt' import type { AiPromptModalRef, AiPromptVariableModalRef, AiPromptForm } from '../types' import RbModal from '@/components/RbModal' import type { Model } from '@/views/ModelManagement/types' -import ChatContent from '@/components/Chat/ChatContent' import Empty from '@/components/Empty' import ConversationEmptyIcon from '@/assets/images/conversation/conversationEmpty.svg' -import type { ChatItem } from '@/components/Chat/types' +import PromptChatPanel, { type PromptChatPanelRef } from '@/components/Chat/PromptChatPanel' import ModelSelect from '@/components/ModelSelect' import AiPromptVariableModal from './AiPromptVariableModal' import { type SSEMessage } from '@/utils/stream' @@ -55,12 +54,14 @@ const AiPromptModal = forwardRef(({ const [visible, setVisible] = useState(false); const [loading, setLoading] = useState(false) const [form] = Form.useForm() - const [chatList, setChatList] = useState([]) const [variables, setVariables] = useState([]) const [promptSession, setPromptSession] = useState(null) + const [hasPrompt, setHasPrompt] = useState(false) const aiPromptVariableModalRef = useRef(null) + const chatPanelRef = useRef(null) const editorRef = useRef(null) const currentPromptValueRef = useRef('') + const isStreamingRef = useRef(false) const values = Form.useWatch([], form) @@ -68,12 +69,13 @@ const AiPromptModal = forwardRef(({ const handleClose = () => { setVisible(false); setLoading(false) - setChatList([]) + chatPanelRef.current?.clear() setVariables([]) form.setFieldsValue({ message: undefined, current_prompt: undefined, }) + setHasPrompt(false) }; /** Open modal and create new prompt session */ @@ -102,9 +104,7 @@ const AiPromptModal = forwardRef(({ } const messageContent = values.message setLoading(true) - setChatList(prev => { - return [...prev, { role: 'user', content: messageContent}] - }) + chatPanelRef.current?.append({ role: 'user', content: messageContent }) form.setFieldsValue({ message: undefined, current_prompt: undefined }) const handleStreamMessage = (data: SSEMessage[]) => { @@ -114,6 +114,8 @@ const AiPromptModal = forwardRef(({ switch (item.event) { case 'start': currentPromptValueRef.current = '' + isStreamingRef.current = true + setHasPrompt(true) if (editorRef.current?.clear) { editorRef.current.clear(); } @@ -123,15 +125,12 @@ const AiPromptModal = forwardRef(({ currentPromptValueRef.current += content; if (editorRef.current?.appendText) { editorRef.current.appendText(content); - editorRef.current.scrollToBottom(); } else { form.setFieldsValue({ current_prompt: currentPromptValueRef.current }) } } if (desc) { - setChatList(prev => { - return [...prev, { role: 'assistant', content: desc }] - }) + chatPanelRef.current?.append({ role: 'assistant', content: desc }) } if (variables) { setVariables(variables) @@ -139,6 +138,7 @@ const AiPromptModal = forwardRef(({ break; case 'end': setLoading(false) + isStreamingRef.current = false // Sync form value when stream ends form.setFieldsValue({ current_prompt: currentPromptValueRef.current }) break @@ -193,7 +193,6 @@ const AiPromptModal = forwardRef(({ setIsFocus(false) } - console.log(values) return ( (({ body: 'rb:p-0! rb:border-t rb:border-t-[#EBEBEB]' }} > -
+
(({ /> - } - data={chatList || []} - streamLoading={false} - labelPosition="top" labelFormat={(item) => item.role === 'user' ? t(`${source}.you`) : t(`${source}.ai`)} /> (({ - - {values?.current_prompt - ? + form.setFieldValue('current_prompt', value)} + height="rb:h-[calc(100vh-276px)]" + className={clsx('rb:bg-white! rb:border-none! rb:p-0!')} + onChange={(value) => { + if (!isStreamingRef.current) { + form.setFieldValue('current_prompt', value) + } + }} /> - : - } - + + : + }
diff --git a/web/src/views/ApplicationConfig/components/Editor/index.tsx b/web/src/views/ApplicationConfig/components/Editor/index.tsx index ab89a610..c3edc1fc 100644 --- a/web/src/views/ApplicationConfig/components/Editor/index.tsx +++ b/web/src/views/ApplicationConfig/components/Editor/index.tsx @@ -2,14 +2,14 @@ * @Author: ZhaoYing * @Date: 2026-02-03 16:25:17 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-02-26 11:18:04 + * @Last Modified time: 2026-04-15 14:00:07 */ /** * Rich text editor component using Lexical framework * Provides text editing with insert, append, clear, and scroll capabilities */ -import {forwardRef, useImperativeHandle } from 'react'; +import {forwardRef, useEffect, useImperativeHandle, useRef } from 'react'; import clsx from 'clsx'; import { LexicalComposer } from '@lexical/react/LexicalComposer'; import { RichTextPlugin } from '@lexical/react/LexicalRichTextPlugin'; @@ -50,7 +50,7 @@ interface LexicalEditorProps { /** Callback when content changes */ onChange?: (value: string) => void; /** Editor height in pixels */ - height?: number; + height?: string; disabled?: boolean; } @@ -73,9 +73,42 @@ const EditorContent = forwardRef(({ value, placeholder = "Please enter content...", onChange, - disabled + disabled, + height }, ref) => { const [editor] = useLexicalComposerContext(); + const scrollRef = useRef(null); + const pendingTextRef = useRef(''); + const rafRef = useRef(null); + const isAppendingRef = useRef(false); + const scrollTopRef = useRef(0); + + useEffect(() => { + const el = scrollRef.current; + if (!el) return; + const onPointerDown = () => { + if (!isAppendingRef.current) scrollTopRef.current = el.scrollTop; + }; + el.addEventListener('pointerdown', onPointerDown); + return () => el.removeEventListener('pointerdown', onPointerDown); + }, []); + + useEffect(() => { + return editor.registerUpdateListener(({ tags }) => { + if (!scrollRef.current) return; + if (tags.has('append-text')) { + scrollRef.current.scrollTop = scrollRef.current.scrollHeight; + } else { + scrollRef.current.scrollTop = scrollTopRef.current; + } + }); + }, [editor]); + + const scrollToBottom = () => { + if (scrollRef.current) { + scrollRef.current.scrollTop = scrollRef.current.scrollHeight; + } + }; /** * Expose editor methods to parent component @@ -94,24 +127,33 @@ const EditorContent = forwardRef(({ }); }, appendText: (text: string) => { - editor.update(() => { - const root = $getRoot(); - const lastChild = root.getLastChild(); - if (lastChild && $isParagraphNode(lastChild)) { - const lastTextNode = lastChild.getLastChild(); - if (lastTextNode && $isTextNode(lastTextNode)) { - const currentText = lastTextNode.getTextContent(); - lastTextNode.setTextContent(currentText + text); + pendingTextRef.current += text; + if (rafRef.current !== null) return; + rafRef.current = requestAnimationFrame(() => { + rafRef.current = null; + const batch = pendingTextRef.current; + pendingTextRef.current = ''; + if (scrollRef.current) scrollTopRef.current = scrollRef.current.scrollTop; + isAppendingRef.current = true; + editor.update(() => { + const root = $getRoot(); + const lastChild = root.getLastChild(); + if (lastChild && $isParagraphNode(lastChild)) { + const lastTextNode = lastChild.getLastChild(); + if (lastTextNode && $isTextNode(lastTextNode)) { + lastTextNode.setTextContent(lastTextNode.getTextContent() + batch); + } else { + lastChild.append($createTextNode(batch)); + } } else { - const textNode = $createTextNode(text); - lastChild.append(textNode); + const paragraph = $createParagraphNode(); + paragraph.append($createTextNode(batch)); + root.append(paragraph); } - } else { - const paragraph = $createParagraphNode(); - const textNode = $createTextNode(text); - paragraph.append(textNode); - root.append(paragraph); - } + }, { + tag: 'append-text', + onUpdate: () => { isAppendingRef.current = false; } + }); }); }, clear: () => { @@ -122,21 +164,16 @@ const EditorContent = forwardRef(({ root.append(paragraph); }); }, - scrollToBottom: () => { - const editorElement = editor.getRootElement(); - if (editorElement) { - editorElement.scrollTop = editorElement.scrollHeight; - } - } + scrollToBottom, }), [editor]); return ( -
+
{ const { message } = App.useApp() const [loading, setLoading] = useState(false) const [form] = Form.useForm() - const [chatList, setChatList] = useState([]) const [variables, setVariables] = useState([]) const [promptSession, setPromptSession] = useState(null) const aiPromptVariableModalRef = useRef(null) const promptSaveModalRef = useRef(null) + const chatPanelRef = useRef(null) const editorRef = useRef(null) const currentPromptValueRef = useRef(undefined) + const isStreamingRef = useRef(false) + const [hasPrompt, setHasPrompt] = useState(false) const values = Form.useWatch([], form) const [editVo, setEditVo] = useState(null) @@ -56,14 +57,14 @@ const Prompt: FC = () => { useEffect(() => { if (editVo?.id) { form.setFieldValue('current_prompt', editVo.prompt) - setChatList([]) + setHasPrompt(true) + chatPanelRef.current?.clear() } updateSession() }, [editVo]) /** Update session ID */ const updateSession = () => { - console.log('updateSession') createPromptSessions().then(res => { const response = res as { id: string } setPromptSession(response.id) @@ -83,9 +84,7 @@ const Prompt: FC = () => { } const messageContent = values.message setLoading(true) - setChatList(prev => { - return [...prev, { role: 'user', content: messageContent}] - }) + chatPanelRef.current?.append({ role: 'user', content: messageContent }) form.setFieldsValue({ message: undefined, current_prompt: undefined }) const handleStreamMessage = (data: SSEMessage[]) => { @@ -95,33 +94,35 @@ const Prompt: FC = () => { switch (item.event) { case 'start': currentPromptValueRef.current = '' + isStreamingRef.current = true + setHasPrompt(true) if (editorRef.current?.clear) { editorRef.current.clear(); } break; case 'message': - if (typeof content === 'string') { + if (content) { currentPromptValueRef.current += content; if (editorRef.current?.appendText) { editorRef.current.appendText(content); - editorRef.current.scrollToBottom(); } else { form.setFieldsValue({ current_prompt: currentPromptValueRef.current }) } } if (desc) { - setChatList(prev => { - return [...prev, { role: 'assistant', content: desc }] - }) + chatPanelRef.current?.append({ role: 'assistant', content: desc }) } if (variables) { setVariables(variables) } + console.log('currentPromptValueRef.current', currentPromptValueRef.current) break; case 'end': setLoading(false) + isStreamingRef.current = false // Sync form values when stream ends form.setFieldsValue({ current_prompt: currentPromptValueRef.current }) + console.log('currentPromptValueRef.current', currentPromptValueRef.current) break } }) @@ -164,7 +165,8 @@ const Prompt: FC = () => { const handleRefresh = () => { form.setFieldValue('current_prompt', undefined) currentPromptValueRef.current = undefined; - setChatList([]) + setHasPrompt(false) + chatPanelRef.current?.clear() setEditVo(null) updateSession() } @@ -193,13 +195,11 @@ const Prompt: FC = () => { headerType="borderless" bodyClassName="rb:px-4! rb:pt-0! rb:pb-3!" > - } - data={chatList || []} - streamLoading={false} - labelPosition="top" labelFormat={(item) => item.role === 'user' ? t(`prompt.you`) : t(`prompt.ai`)} /> { > } > - - {values?.current_prompt - ? + form.setFieldValue('current_prompt', value)} + height="rb:h-[calc(100vh-193px)]" + className="rb:bg-white! rb:border-none! rb:p-0! rb:text-[#212332] rb:leading-5" + onChange={(value) => { + if (!isStreamingRef.current) { + form.setFieldValue('current_prompt', value) + } + }} /> - : - } - + + : + }
From 5ac2d5602e5fa77530c9ad31adb4aac5363d55d5 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 15 Apr 2026 14:50:19 +0800 Subject: [PATCH 064/245] Revert "fix(web): prompt editor" This reverts commit 71e5b6586a1186b88d81e90ba1d02bfe41db6b98. --- web/src/components/Chat/PromptChatPanel.tsx | 39 -------- .../components/AiPromptModal.tsx | 56 ++++++----- .../components/Editor/index.tsx | 93 ++++++------------- web/src/views/Prompt/index.tsx | 59 ++++++------ 4 files changed, 82 insertions(+), 165 deletions(-) delete mode 100644 web/src/components/Chat/PromptChatPanel.tsx diff --git a/web/src/components/Chat/PromptChatPanel.tsx b/web/src/components/Chat/PromptChatPanel.tsx deleted file mode 100644 index 51343ae1..00000000 --- a/web/src/components/Chat/PromptChatPanel.tsx +++ /dev/null @@ -1,39 +0,0 @@ -import { forwardRef, useImperativeHandle, useState } from 'react' -import ChatContent from './ChatContent' -import type { ChatItem } from './types' -import type { ReactNode } from 'react' - -export interface PromptChatPanelRef { - append: (item: ChatItem) => void - clear: () => void -} - -interface PromptChatPanelProps { - classNames?: string - contentClassNames?: string - empty: ReactNode - labelFormat: (item: ChatItem) => any -} - -const PromptChatPanel = forwardRef((props, ref) => { - const [chatList, setChatList] = useState([]) - - useImperativeHandle(ref, () => ({ - append: (item) => setChatList(prev => [...prev, item]), - clear: () => setChatList([]), - })) - - return ( - - ) -}) - -export default PromptChatPanel diff --git a/web/src/views/ApplicationConfig/components/AiPromptModal.tsx b/web/src/views/ApplicationConfig/components/AiPromptModal.tsx index fd2dc595..1666e075 100644 --- a/web/src/views/ApplicationConfig/components/AiPromptModal.tsx +++ b/web/src/views/ApplicationConfig/components/AiPromptModal.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 16:26:44 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-15 14:21:55 + * @Last Modified time: 2026-03-20 13:53:05 */ /** * AI Prompt Assistant Modal @@ -20,9 +20,10 @@ import { updatePromptMessages, createPromptSessions } from '@/api/prompt' import type { AiPromptModalRef, AiPromptVariableModalRef, AiPromptForm } from '../types' import RbModal from '@/components/RbModal' import type { Model } from '@/views/ModelManagement/types' +import ChatContent from '@/components/Chat/ChatContent' import Empty from '@/components/Empty' import ConversationEmptyIcon from '@/assets/images/conversation/conversationEmpty.svg' -import PromptChatPanel, { type PromptChatPanelRef } from '@/components/Chat/PromptChatPanel' +import type { ChatItem } from '@/components/Chat/types' import ModelSelect from '@/components/ModelSelect' import AiPromptVariableModal from './AiPromptVariableModal' import { type SSEMessage } from '@/utils/stream' @@ -54,14 +55,12 @@ const AiPromptModal = forwardRef(({ const [visible, setVisible] = useState(false); const [loading, setLoading] = useState(false) const [form] = Form.useForm() + const [chatList, setChatList] = useState([]) const [variables, setVariables] = useState([]) const [promptSession, setPromptSession] = useState(null) - const [hasPrompt, setHasPrompt] = useState(false) const aiPromptVariableModalRef = useRef(null) - const chatPanelRef = useRef(null) const editorRef = useRef(null) const currentPromptValueRef = useRef('') - const isStreamingRef = useRef(false) const values = Form.useWatch([], form) @@ -69,13 +68,12 @@ const AiPromptModal = forwardRef(({ const handleClose = () => { setVisible(false); setLoading(false) - chatPanelRef.current?.clear() + setChatList([]) setVariables([]) form.setFieldsValue({ message: undefined, current_prompt: undefined, }) - setHasPrompt(false) }; /** Open modal and create new prompt session */ @@ -104,7 +102,9 @@ const AiPromptModal = forwardRef(({ } const messageContent = values.message setLoading(true) - chatPanelRef.current?.append({ role: 'user', content: messageContent }) + setChatList(prev => { + return [...prev, { role: 'user', content: messageContent}] + }) form.setFieldsValue({ message: undefined, current_prompt: undefined }) const handleStreamMessage = (data: SSEMessage[]) => { @@ -114,8 +114,6 @@ const AiPromptModal = forwardRef(({ switch (item.event) { case 'start': currentPromptValueRef.current = '' - isStreamingRef.current = true - setHasPrompt(true) if (editorRef.current?.clear) { editorRef.current.clear(); } @@ -125,12 +123,15 @@ const AiPromptModal = forwardRef(({ currentPromptValueRef.current += content; if (editorRef.current?.appendText) { editorRef.current.appendText(content); + editorRef.current.scrollToBottom(); } else { form.setFieldsValue({ current_prompt: currentPromptValueRef.current }) } } if (desc) { - chatPanelRef.current?.append({ role: 'assistant', content: desc }) + setChatList(prev => { + return [...prev, { role: 'assistant', content: desc }] + }) } if (variables) { setVariables(variables) @@ -138,7 +139,6 @@ const AiPromptModal = forwardRef(({ break; case 'end': setLoading(false) - isStreamingRef.current = false // Sync form value when stream ends form.setFieldsValue({ current_prompt: currentPromptValueRef.current }) break @@ -193,6 +193,7 @@ const AiPromptModal = forwardRef(({ setIsFocus(false) } + console.log(values) return ( (({ body: 'rb:p-0! rb:border-t rb:border-t-[#EBEBEB]' }} > -
+
(({ /> - } + data={chatList || []} + streamLoading={false} + labelPosition="top" labelFormat={(item) => item.role === 'user' ? t(`${source}.you`) : t(`${source}.ai`)} /> (({ - {hasPrompt - ? - + {values?.current_prompt + ? { - if (!isStreamingRef.current) { - form.setFieldValue('current_prompt', value) - } - }} + className="rb:h-119 rb:bg-white! rb:border-none! rb:p-0!" + onChange={(value) => form.setFieldValue('current_prompt', value)} /> - - : - } + : + } +
diff --git a/web/src/views/ApplicationConfig/components/Editor/index.tsx b/web/src/views/ApplicationConfig/components/Editor/index.tsx index c3edc1fc..ab89a610 100644 --- a/web/src/views/ApplicationConfig/components/Editor/index.tsx +++ b/web/src/views/ApplicationConfig/components/Editor/index.tsx @@ -2,14 +2,14 @@ * @Author: ZhaoYing * @Date: 2026-02-03 16:25:17 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-15 14:00:07 + * @Last Modified time: 2026-02-26 11:18:04 */ /** * Rich text editor component using Lexical framework * Provides text editing with insert, append, clear, and scroll capabilities */ -import {forwardRef, useEffect, useImperativeHandle, useRef } from 'react'; +import {forwardRef, useImperativeHandle } from 'react'; import clsx from 'clsx'; import { LexicalComposer } from '@lexical/react/LexicalComposer'; import { RichTextPlugin } from '@lexical/react/LexicalRichTextPlugin'; @@ -50,7 +50,7 @@ interface LexicalEditorProps { /** Callback when content changes */ onChange?: (value: string) => void; /** Editor height in pixels */ - height?: string; + height?: number; disabled?: boolean; } @@ -73,42 +73,9 @@ const EditorContent = forwardRef(({ value, placeholder = "Please enter content...", onChange, - disabled, - height + disabled }, ref) => { const [editor] = useLexicalComposerContext(); - const scrollRef = useRef(null); - const pendingTextRef = useRef(''); - const rafRef = useRef(null); - const isAppendingRef = useRef(false); - const scrollTopRef = useRef(0); - - useEffect(() => { - const el = scrollRef.current; - if (!el) return; - const onPointerDown = () => { - if (!isAppendingRef.current) scrollTopRef.current = el.scrollTop; - }; - el.addEventListener('pointerdown', onPointerDown); - return () => el.removeEventListener('pointerdown', onPointerDown); - }, []); - - useEffect(() => { - return editor.registerUpdateListener(({ tags }) => { - if (!scrollRef.current) return; - if (tags.has('append-text')) { - scrollRef.current.scrollTop = scrollRef.current.scrollHeight; - } else { - scrollRef.current.scrollTop = scrollTopRef.current; - } - }); - }, [editor]); - - const scrollToBottom = () => { - if (scrollRef.current) { - scrollRef.current.scrollTop = scrollRef.current.scrollHeight; - } - }; /** * Expose editor methods to parent component @@ -127,33 +94,24 @@ const EditorContent = forwardRef(({ }); }, appendText: (text: string) => { - pendingTextRef.current += text; - if (rafRef.current !== null) return; - rafRef.current = requestAnimationFrame(() => { - rafRef.current = null; - const batch = pendingTextRef.current; - pendingTextRef.current = ''; - if (scrollRef.current) scrollTopRef.current = scrollRef.current.scrollTop; - isAppendingRef.current = true; - editor.update(() => { - const root = $getRoot(); - const lastChild = root.getLastChild(); - if (lastChild && $isParagraphNode(lastChild)) { - const lastTextNode = lastChild.getLastChild(); - if (lastTextNode && $isTextNode(lastTextNode)) { - lastTextNode.setTextContent(lastTextNode.getTextContent() + batch); - } else { - lastChild.append($createTextNode(batch)); - } + editor.update(() => { + const root = $getRoot(); + const lastChild = root.getLastChild(); + if (lastChild && $isParagraphNode(lastChild)) { + const lastTextNode = lastChild.getLastChild(); + if (lastTextNode && $isTextNode(lastTextNode)) { + const currentText = lastTextNode.getTextContent(); + lastTextNode.setTextContent(currentText + text); } else { - const paragraph = $createParagraphNode(); - paragraph.append($createTextNode(batch)); - root.append(paragraph); + const textNode = $createTextNode(text); + lastChild.append(textNode); } - }, { - tag: 'append-text', - onUpdate: () => { isAppendingRef.current = false; } - }); + } else { + const paragraph = $createParagraphNode(); + const textNode = $createTextNode(text); + paragraph.append(textNode); + root.append(paragraph); + } }); }, clear: () => { @@ -164,16 +122,21 @@ const EditorContent = forwardRef(({ root.append(paragraph); }); }, - scrollToBottom, + scrollToBottom: () => { + const editorElement = editor.getRootElement(); + if (editorElement) { + editorElement.scrollTop = editorElement.scrollHeight; + } + } }), [editor]); return ( -
+
{ const { message } = App.useApp() const [loading, setLoading] = useState(false) const [form] = Form.useForm() + const [chatList, setChatList] = useState([]) const [variables, setVariables] = useState([]) const [promptSession, setPromptSession] = useState(null) const aiPromptVariableModalRef = useRef(null) const promptSaveModalRef = useRef(null) - const chatPanelRef = useRef(null) const editorRef = useRef(null) const currentPromptValueRef = useRef(undefined) - const isStreamingRef = useRef(false) - const [hasPrompt, setHasPrompt] = useState(false) const values = Form.useWatch([], form) const [editVo, setEditVo] = useState(null) @@ -57,14 +56,14 @@ const Prompt: FC = () => { useEffect(() => { if (editVo?.id) { form.setFieldValue('current_prompt', editVo.prompt) - setHasPrompt(true) - chatPanelRef.current?.clear() + setChatList([]) } updateSession() }, [editVo]) /** Update session ID */ const updateSession = () => { + console.log('updateSession') createPromptSessions().then(res => { const response = res as { id: string } setPromptSession(response.id) @@ -84,7 +83,9 @@ const Prompt: FC = () => { } const messageContent = values.message setLoading(true) - chatPanelRef.current?.append({ role: 'user', content: messageContent }) + setChatList(prev => { + return [...prev, { role: 'user', content: messageContent}] + }) form.setFieldsValue({ message: undefined, current_prompt: undefined }) const handleStreamMessage = (data: SSEMessage[]) => { @@ -94,35 +95,33 @@ const Prompt: FC = () => { switch (item.event) { case 'start': currentPromptValueRef.current = '' - isStreamingRef.current = true - setHasPrompt(true) if (editorRef.current?.clear) { editorRef.current.clear(); } break; case 'message': - if (content) { + if (typeof content === 'string') { currentPromptValueRef.current += content; if (editorRef.current?.appendText) { editorRef.current.appendText(content); + editorRef.current.scrollToBottom(); } else { form.setFieldsValue({ current_prompt: currentPromptValueRef.current }) } } if (desc) { - chatPanelRef.current?.append({ role: 'assistant', content: desc }) + setChatList(prev => { + return [...prev, { role: 'assistant', content: desc }] + }) } if (variables) { setVariables(variables) } - console.log('currentPromptValueRef.current', currentPromptValueRef.current) break; case 'end': setLoading(false) - isStreamingRef.current = false // Sync form values when stream ends form.setFieldsValue({ current_prompt: currentPromptValueRef.current }) - console.log('currentPromptValueRef.current', currentPromptValueRef.current) break } }) @@ -165,8 +164,7 @@ const Prompt: FC = () => { const handleRefresh = () => { form.setFieldValue('current_prompt', undefined) currentPromptValueRef.current = undefined; - setHasPrompt(false) - chatPanelRef.current?.clear() + setChatList([]) setEditVo(null) updateSession() } @@ -195,11 +193,13 @@ const Prompt: FC = () => { headerType="borderless" bodyClassName="rb:px-4! rb:pt-0! rb:pb-3!" > - } + data={chatList || []} + streamLoading={false} + labelPosition="top" labelFormat={(item) => item.role === 'user' ? t(`prompt.you`) : t(`prompt.ai`)} /> { > } > - {hasPrompt - ? - + {values?.current_prompt + ? { - if (!isStreamingRef.current) { - form.setFieldValue('current_prompt', value) - } - }} + className="rb:h-[calc(100vh-193px)] rb:bg-white! rb:border-none! rb:p-0! rb:text-[#212332] rb:leading-5" + onChange={(value) => form.setFieldValue('current_prompt', value)} /> - - : - } + : + } +
From 643a3fbe094e97d44b6ceb23b8f70b5e67ace039 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 15 Apr 2026 16:09:38 +0800 Subject: [PATCH 065/245] feat(web): node run status --- web/src/components/CodeMirrorEditor/index.tsx | 6 +-- web/src/store/workflow.ts | 8 ++++ .../views/Workflow/components/Chat/Chat.tsx | 17 +++++-- .../views/Workflow/components/NodeLibrary.tsx | 45 +++++++++---------- .../components/Nodes/ConditionNode.tsx | 17 +++++-- .../Workflow/components/Nodes/LoopNode.tsx | 15 ++++++- .../Workflow/components/Nodes/NormalNode.tsx | 17 +++++-- .../views/Workflow/hooks/useWorkflowGraph.ts | 30 ++++++++++++- 8 files changed, 115 insertions(+), 40 deletions(-) diff --git a/web/src/components/CodeMirrorEditor/index.tsx b/web/src/components/CodeMirrorEditor/index.tsx index ec2a6780..8671992a 100644 --- a/web/src/components/CodeMirrorEditor/index.tsx +++ b/web/src/components/CodeMirrorEditor/index.tsx @@ -1,8 +1,8 @@ /* * @Author: ZhaoYing * @Date: 2026-02-04 17:20:52 - * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-02-04 17:20:52 + * @Last Modified by: ZhaoYing + * @Last Modified time: 2026-04-14 18:24:29 */ import { useEffect, useRef, useMemo } from 'react'; import { EditorView, basicSetup } from 'codemirror'; @@ -156,7 +156,7 @@ const CodeMirrorEditor = ({
); }; diff --git a/web/src/store/workflow.ts b/web/src/store/workflow.ts index 0999d35a..382d9255 100644 --- a/web/src/store/workflow.ts +++ b/web/src/store/workflow.ts @@ -6,11 +6,15 @@ */ import { create } from 'zustand' import type { NodeCheckResult } from '@/views/Workflow/components/CheckList' +import type { ChatItem } from '@/components/Chat/types' interface WorkflowState { checkResults: Record setCheckResults: (appId: string, results: NodeCheckResult[]) => void getCheckResults: (appId: string) => NodeCheckResult[] + chatHistoryMap: Record + setChatHistory: (conversationId: string, history: ChatItem[]) => void + getChatHistory: (conversationId: string) => ChatItem[] } export const useWorkflowStore = create((set, get) => ({ @@ -18,4 +22,8 @@ export const useWorkflowStore = create((set, get) => ({ setCheckResults: (appId, results) => set(state => ({ checkResults: { ...state.checkResults, [appId]: results } })), getCheckResults: (appId) => get().checkResults[appId] ?? [], + chatHistoryMap: {}, + setChatHistory: (conversationId, history) => + set(state => ({ chatHistoryMap: { ...state.chatHistoryMap, [conversationId]: history } })), + getChatHistory: (conversationId) => get().chatHistoryMap[conversationId] ?? [], })) diff --git a/web/src/views/Workflow/components/Chat/Chat.tsx b/web/src/views/Workflow/components/Chat/Chat.tsx index e1a0ad95..19b06a0d 100644 --- a/web/src/views/Workflow/components/Chat/Chat.tsx +++ b/web/src/views/Workflow/components/Chat/Chat.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-06 21:10:56 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-07 18:07:38 + * @Last Modified time: 2026-04-15 15:57:35 */ /** * Workflow Chat Component @@ -41,12 +41,15 @@ import type { ChatToolbarRef } from '@/components/Chat/ChatToolbar' import Runtime from './Runtime'; import type { FeaturesConfigForm } from '@/views/ApplicationConfig/types'; import { replaceVariables } from '@/views/ApplicationConfig/Agent'; +import { useWorkflowStore } from '@/store/workflow'; -const Chat = forwardRef(({ +const Chat = forwardRef(({ // eslint-disable-line appId, graphRef, features }, ref) => { const { t } = useTranslation() const { message: messageApi } = App.useApp() + const { setChatHistory } = useWorkflowStore() + const conversationIdRef = useRef('draft') const toolbarRef = useRef(null) const [toolbarReady, setToolbarReady] = useState(false) const toolbarCallbackRef = useCallback((node: ChatToolbarRef | null) => { @@ -118,6 +121,7 @@ const Chat = forwardRef; - status?: 'completed' | 'failed', + status?: 'completed' | 'failed' | 'running', citations?: { document_id: string; file_name: string; @@ -231,6 +235,7 @@ const Chat = forwardRef { + setChatHistory(conversationIdRef.current, chatList) + }, [chatList]) + return ( diff --git a/web/src/views/Workflow/components/NodeLibrary.tsx b/web/src/views/Workflow/components/NodeLibrary.tsx index e6190adb..525c09ae 100644 --- a/web/src/views/Workflow/components/NodeLibrary.tsx +++ b/web/src/views/Workflow/components/NodeLibrary.tsx @@ -34,29 +34,24 @@ const NodeLibrary: FC<{ collapsed: boolean; handleToggle: () => void }> = ({ col > {collapsed - ? <> - {nodeLibrary.map(category => ( - <> - {category.nodes - .filter(node => node.type !== 'cycle-start' && node.type !== 'break') - .map((node, nodeIndex) => ( - -
{ - e.dataTransfer.setData('application/reactflow', node.type); - e.dataTransfer.setData('application/json', JSON.stringify(node)); - }} - > -
-
- - )) - } - - ))} - + ? nodeLibrary.flatMap(category => + category.nodes + .filter(node => node.type !== 'cycle-start' && node.type !== 'break') + .map(node => ( + +
{ + e.dataTransfer.setData('application/reactflow', node.type); + e.dataTransfer.setData('application/json', JSON.stringify(node)); + }} + > +
+
+ + )) + ) : nodeLibrary.map(category => (
void }> = ({ col {category.nodes .filter(node => node.type !== 'cycle-start' && node.type !== 'break') - .map((node, nodeIndex) => ( + .map((node) => ( { return (
-
{data.name ?? t(`workflow.${data.type}`)}
+
{data.name ?? t(`workflow.${data.type}`)}
+ {data.executionStatus === 'completed' + ? + : data.executionStatus === 'failed' + ? + : data.executionStatus === 'running' + ? + : null + } {data.type === 'question-classifier' && diff --git a/web/src/views/Workflow/components/Nodes/LoopNode.tsx b/web/src/views/Workflow/components/Nodes/LoopNode.tsx index ca0eaeff..c540db76 100644 --- a/web/src/views/Workflow/components/Nodes/LoopNode.tsx +++ b/web/src/views/Workflow/components/Nodes/LoopNode.tsx @@ -3,6 +3,7 @@ import { useTranslation } from 'react-i18next' import clsx from 'clsx'; import type { ReactShapeConfig } from '@antv/x6-react-shape'; import { Flex } from 'antd'; +import { CheckCircleFilled, CloseCircleFilled, LoadingOutlined } from '@ant-design/icons'; import { graphNodeLibrary, edgeAttrs } from '../../constant'; import NodeTools from './NodeTools' @@ -131,12 +132,22 @@ const LoopNode: ReactShapeConfig['component'] = ({ node, graph }) => { return (
-
{data.name ?? t(`workflow.${data.type}`)}
+
{data.name ?? t(`workflow.${data.type}`)}
+ {data.executionStatus === 'completed' + ? + : data.executionStatus === 'failed' + ? + : data.executionStatus === 'running' + ? + : null + }
diff --git a/web/src/views/Workflow/components/Nodes/NormalNode.tsx b/web/src/views/Workflow/components/Nodes/NormalNode.tsx index f947d004..ce936be9 100644 --- a/web/src/views/Workflow/components/Nodes/NormalNode.tsx +++ b/web/src/views/Workflow/components/Nodes/NormalNode.tsx @@ -2,6 +2,7 @@ import clsx from 'clsx'; import { useTranslation } from 'react-i18next' import type { ReactShapeConfig } from '@antv/x6-react-shape'; import { Flex } from 'antd'; +import { CheckCircleFilled, CloseCircleFilled, LoadingOutlined } from '@ant-design/icons'; import NodeTools from './NodeTools' @@ -11,13 +12,23 @@ const NormalNode: ReactShapeConfig['component'] = ({ node }) => { return (
-
{data.name ?? t(`workflow.${data.type}`)}
+
{data.name ?? t(`workflow.${data.type}`)}
+ {data.executionStatus === 'completed' + ? + : data.executionStatus === 'failed' + ? + : data.executionStatus === 'running' + ? + : null + }
{t('workflow.clickToConfigure')}
diff --git a/web/src/views/Workflow/hooks/useWorkflowGraph.ts b/web/src/views/Workflow/hooks/useWorkflowGraph.ts index f385acf3..516bc24c 100644 --- a/web/src/views/Workflow/hooks/useWorkflowGraph.ts +++ b/web/src/views/Workflow/hooks/useWorkflowGraph.ts @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 15:17:48 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-07 23:17:50 + * @Last Modified time: 2026-04-15 16:02:49 */ import { Clipboard, Graph, Keyboard, MiniMap, Node, Snapline, type Edge } from '@antv/x6'; import { register } from '@antv/x6-react-shape'; @@ -18,6 +18,7 @@ import type { FeaturesConfigForm } from '@/views/ApplicationConfig/types'; import { conditionNodeHeight, conditionNodeItemHeight, conditionNodePortItemArgsY, defaultAbsolutePortGroups, defaultPortItems, edgeAttrs, edgeHoverTool, edge_color, edge_selected_color, edge_width, graphNodeLibrary, nodeLibrary, nodeRegisterLibrary, nodeWidth, notesConfig, portAttrs, portItemArgsY, portMarkup, portTextAttrs, unknownNode } from '../constant'; import type { ChatVariable, NodeProperties, WorkflowConfig } from '../types'; import { calcConditionNodeTotalHeight, getConditionNodeCasePortY } from '../utils'; +import { useWorkflowStore } from '@/store/workflow'; /** * Props for useWorkflowGraph hook @@ -94,6 +95,8 @@ export const useWorkflowGraph = ({ const { message } = App.useApp(); const { t } = useTranslation() const { user } = useUser(); + const { chatHistoryMap } = useWorkflowStore() + const chatHistory = Object.values(chatHistoryMap).at(-1) ?? [] // Refs const graphRef = useRef(); @@ -1425,6 +1428,31 @@ export const useWorkflowGraph = ({ } } } + useEffect(() => { + if (!graphRef.current) return; + const nodes = graphRef.current.getNodes(); + + const lastWithSub = [...chatHistory].reverse().find(item => item.subContent?.length); + // Reset all node execution status first + nodes.forEach(node => { + const data = node.getData(); + if (typeof data.status === 'string') { + node.setData({ ...data, executionStatus: undefined }); + } + }); + if (!lastWithSub?.subContent) return; + // Build a nodeId -> status map first + const statusMap: Record = {}; + lastWithSub.subContent.forEach(sub => { + if (typeof sub.status === 'string') { + statusMap[sub.node_id] = sub.status; + const node = nodes.find(n => n.getData()?.id === sub.node_id); + if (node) { + node.setData({ ...node.getData(), executionStatus: sub.status }); + } + } + }); + }, [chatHistory, graphRef.current]); return { config, From 466864afe362d784983174bbc10c89cbdbefee91 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 15 Apr 2026 16:46:47 +0800 Subject: [PATCH 066/245] fix(web): Cancel variable snapshot --- .../components/Properties/VariableSelect.tsx | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-) diff --git a/web/src/views/Workflow/components/Properties/VariableSelect.tsx b/web/src/views/Workflow/components/Properties/VariableSelect.tsx index 5523c06e..203dd850 100644 --- a/web/src/views/Workflow/components/Properties/VariableSelect.tsx +++ b/web/src/views/Workflow/components/Properties/VariableSelect.tsx @@ -40,7 +40,7 @@ const VariableSelect: FC = ({ const { t } = useTranslation(); const [open, setOpen] = useState(false); const [search, setSearch] = useState(''); - const [expandedParent, setExpandedParent] = useState(null); + const [expandedParentKey, setExpandedParentKey] = useState(null); const [dropdownPos, setDropdownPos] = useState({ top: 0, left: 0, width: 0 }); const [childPanelPos, setChildPanelPos] = useState({ top: 0, right: 0 }); const containerRef = useRef(null); @@ -84,6 +84,10 @@ const VariableSelect: FC = ({ ? filteredOptions.find(o => o.children?.some(c => `{{${c.value}}}` === value)) : undefined; + const expandedParent = expandedParentKey + ? filteredOptions.find(o => o.key === expandedParentKey) ?? null + : null; + const groupedSuggestions = filteredOptions.reduce((groups: Record, s) => { const nodeId = s.nodeData.id as string; if (!groups[nodeId]) groups[nodeId] = []; @@ -139,7 +143,7 @@ const VariableSelect: FC = ({ ) { setOpen(false); setSearch(''); - setExpandedParent(null); + setExpandedParentKey(null); setChildPanelPos({ top: 0, right: 0 }); } }; @@ -159,7 +163,7 @@ const VariableSelect: FC = ({ onChange?.(`{{${suggestion.value}}}`, suggestion); setOpen(false); setSearch(''); - setExpandedParent(null); + setExpandedParentKey(null); } }; @@ -312,16 +316,16 @@ const VariableSelect: FC = ({ if (s.disabled) return; if (hasChildren) { updateChildPos(s.key); - setExpandedParent(prev => prev?.key === s.key ? null : s); + setExpandedParentKey(prev => prev === s.key ? null : s.key); } handleSelect(s); }} onMouseEnter={() => { if (hasChildren) { updateChildPos(s.key); - setExpandedParent(s); + setExpandedParentKey(s.key); } else { - setExpandedParent(null); + setExpandedParentKey(null); } }} > @@ -358,7 +362,7 @@ const VariableSelect: FC = ({ id="variable-select-child-panel" className="rb:min-w-70 rb:max-h-57.5 rb:overflow-y-auto rb:text-[12px] rb:fixed rb:z-1000 rb:bg-white rb:rounded-lg rb:border-[0.5px] rb:border-[#EBEBEB] rb:shadow-[0px_2px_6px_0px_rgba(0,0,0,0.1)] rb:py-3 rb:px-2" style={{ top: childPanelPos.top, right: childPanelPos.right }} - onMouseEnter={() => setExpandedParent(expandedParent)} + onMouseEnter={() => setExpandedParentKey(expandedParentKey)} >
From cc12ec3fa82b5d8a0acd4bc4ed895d3f9b715004 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Wed, 15 Apr 2026 18:03:39 +0800 Subject: [PATCH 067/245] fix(workflow): support direct variable reference in tool parameters to preserve native types --- api/app/core/workflow/nodes/tool/node.py | 21 +++++++++++++++------ 1 file changed, 15 insertions(+), 6 deletions(-) diff --git a/api/app/core/workflow/nodes/tool/node.py b/api/app/core/workflow/nodes/tool/node.py index 72c5c6a8..410f64c3 100644 --- a/api/app/core/workflow/nodes/tool/node.py +++ b/api/app/core/workflow/nodes/tool/node.py @@ -15,6 +15,7 @@ from app.services.tool_service import ToolService logger = logging.getLogger(__name__) TEMPLATE_PATTERN = re.compile(r"\{\{.*?}}") +PURE_VARIABLE_PATTERN = re.compile(r"^\{\{\s*([\w.]+)\s*}}$") class ToolNode(BaseNode): @@ -52,13 +53,21 @@ class ToolNode(BaseNode): # 渲染工具参数 rendered_parameters = {} for param_name, param_template in self.typed_config.tool_parameters.items(): - if isinstance(param_template, str) and TEMPLATE_PATTERN.search(param_template): - try: - rendered_value = self._render_template(param_template, variable_pool) - except Exception as e: - raise ValueError(f"模板渲染失败：参数 {param_name} 的模板 {param_template} 解析错误") from e + if isinstance(param_template, str): + pure_match = PURE_VARIABLE_PATTERN.match(param_template) + if pure_match: + # 纯单变量引用直接取原始值，保留 int/bool/float 等类型 + rendered_value = self.get_variable(pure_match.group(1), variable_pool, strict=False) + if rendered_value is None: + rendered_value = self._render_template(param_template, variable_pool) + elif TEMPLATE_PATTERN.search(param_template): + try: + rendered_value = self._render_template(param_template, variable_pool) + except Exception as e: + raise ValueError(f"模板渲染失败：参数 {param_name} 的模板 {param_template} 解析错误") from e + else: + rendered_value = param_template else: - # 非模板参数（数字/布尔/普通字符串）直接保留原值 rendered_value = param_template rendered_parameters[param_name] = rendered_value From 1faa258e233eb5712526ac1abe9086ac5a165035 Mon Sep 17 00:00:00 2001 From: wwq Date: Wed, 15 Apr 2026 18:48:09 +0800 Subject: [PATCH 068/245] feat(quota): implement unified quota management system and add community free plan - Add `default_free_plan.py` to define the configuration for the Community Free Plan. - Refactor `quota_stub.py` as a unified entry point, delegating checks to `core/quota_manager`. - Implement core logic in `quota_manager.py` to support retrieving quotas from the premium module or configuration files. - Update `tenant_subscription_controller` to return Community Free Plan information. --- api/app/config/default_free_plan.py | 30 ++ .../tenant_subscription_controller.py | 33 +- api/app/core/quota_manager.py | 473 ++++++++++++++++++ api/app/core/quota_stub.py | 74 ++- 4 files changed, 567 insertions(+), 43 deletions(-) create mode 100644 api/app/config/default_free_plan.py create mode 100644 api/app/core/quota_manager.py diff --git a/api/app/config/default_free_plan.py b/api/app/config/default_free_plan.py new file mode 100644 index 00000000..23a3a10e --- /dev/null +++ b/api/app/config/default_free_plan.py @@ -0,0 +1,30 @@ +""" +社区版默认免费套餐配置 +当无法从 SaaS 版获取 premium 模块时，使用此配置作为兜底 +""" + +DEFAULT_FREE_PLAN = { + "name": "记忆体验版", + "category": "saas_personal", + "tier_level": 0, + "version": "1.0", + "status": True, + "price": 0, + "billing_cycle": "permanent_free", + "core_value": "感受永久记忆", + "tech_support": "社群交流", + "sla_compliance": "无", + "page_customization": "无", + "theme_color": "#64748B", + "quotas": { + "workspace_quota": 1, + "skill_quota": 5, + "app_quota": 2, + "knowledge_capacity_quota": 0.3, + "memory_engine_quota": 1, + "end_user_quota": 1, + "ontology_project_quota": 3, + "model_quota": 1, + "api_ops_rate_limit": 50, + }, +} diff --git a/api/app/controllers/tenant_subscription_controller.py b/api/app/controllers/tenant_subscription_controller.py index 2629f7f1..c3fde572 100644 --- a/api/app/controllers/tenant_subscription_controller.py +++ b/api/app/controllers/tenant_subscription_controller.py @@ -1,6 +1,7 @@ """ 租户套餐查询接口（普通用户可访问） """ +import datetime from typing import Callable from fastapi import APIRouter, Depends @@ -46,8 +47,36 @@ async def get_my_tenant_subscription( return success(data=svc.build_response(sub)) except ModuleNotFoundError: - # 社区版无 premium 模块，返回空 - return success(data=None, msg="套餐功能未启用") + # 社区版无 premium 模块，从配置文件读取免费套餐 + if not current_user.tenant: + return JSONResponse(status_code=404, content=fail(code=404, msg="用户未关联租户")) + + from app.config.default_free_plan import DEFAULT_FREE_PLAN + + plan = DEFAULT_FREE_PLAN + response_data = { + "subscription_id": None, + "tenant_id": str(current_user.tenant.id), + "package_plan_id": None, + "package_version": plan["version"], + "package_plan": { + "id": None, + "name": plan["name"], + "version": plan["version"], + "category": plan["category"], + "tier_level": plan["tier_level"], + "price": float(plan["price"]), + "billing_cycle": plan["billing_cycle"], + }, + "started_at": None, + "expired_at": None, + "status": "active", + "quota": plan["quotas"], + "created_at": int(datetime.datetime.utcnow().timestamp() * 1000), + "updated_at": int(datetime.datetime.utcnow().timestamp() * 1000), + } + return success(data=response_data, msg="社区版免费套餐") + except Exception as e: logger.error(f"获取租户套餐信息失败: {e}", exc_info=True) return JSONResponse(status_code=500, content=fail(code=500, msg="获取套餐信息失败")) diff --git a/api/app/core/quota_manager.py b/api/app/core/quota_manager.py new file mode 100644 index 00000000..6c02ac7a --- /dev/null +++ b/api/app/core/quota_manager.py @@ -0,0 +1,473 @@ +""" +统一配额管理器 - 社区版和 SaaS 版共用 + +配额来源策略： +1. 优先从 premium 模块的 tenant_subscriptions 表读取（SaaS 版） +2. 降级到 default_free_plan.py 配置文件（社区版兜底） +""" +import asyncio +import time +from functools import wraps +from typing import Optional, Callable, Dict, Any +from uuid import UUID + +from sqlalchemy import func +from sqlalchemy.orm import Session + +from app.core.logging_config import get_auth_logger +from app.i18n.exceptions import QuotaExceededError + +logger = get_auth_logger() + + +def _get_user_from_kwargs(kwargs: dict): + """从 kwargs 中获取 user 对象""" + for key in ["user", "current_user"]: + if key in kwargs: + return kwargs[key] + return None + + +def _get_tenant_id_from_kwargs(db: Session, kwargs: dict): + """从 kwargs 中获取 tenant_id""" + user = _get_user_from_kwargs(kwargs) + if user and hasattr(user, 'tenant_id'): + return user.tenant_id + + workspace_id = kwargs.get("workspace_id") + if workspace_id: + from app.models.workspace_model import Workspace + workspace = db.query(Workspace).filter(Workspace.id == workspace_id).first() + if workspace: + return workspace.tenant_id + + api_key_auth = kwargs.get("api_key_auth") + if api_key_auth and hasattr(api_key_auth, 'workspace_id'): + from app.models.workspace_model import Workspace + workspace = db.query(Workspace).filter(Workspace.id == api_key_auth.workspace_id).first() + if workspace: + return workspace.tenant_id + + data = kwargs.get("data") or kwargs.get("body") or kwargs.get("payload") + if data and hasattr(data, "workspace_id"): + from app.models.workspace_model import Workspace + workspace = db.query(Workspace).filter(Workspace.id == data.workspace_id).first() + if workspace: + return workspace.tenant_id + + return None + + +def _get_quota_config(db: Session, tenant_id: UUID) -> Optional[Dict[str, Any]]: + """ + 获取租户的配额配置 + + 优先级： + 1. premium 模块的 tenant_subscriptions（SaaS 版） + 2. default_free_plan.py 配置文件（社区版兜底） + """ + # 尝试从 premium 模块获取 + try: + from premium.platform_admin.package_plan_service import TenantSubscriptionService + quota_config = TenantSubscriptionService(db).get_effective_quota(tenant_id) + if quota_config: + logger.debug(f"从 premium 模块获取租户 {tenant_id} 配额配置") + return quota_config + except (ModuleNotFoundError, ImportError, Exception) as e: + logger.debug(f"无法从 premium 模块获取配额配置: {e}") + + # 降级到配置文件 + try: + from app.config.default_free_plan import DEFAULT_FREE_PLAN + logger.info(f"使用配置文件中的免费套餐配额: tenant={tenant_id}") + return DEFAULT_FREE_PLAN.get("quotas") + except Exception as e: + logger.error(f"无法从配置文件获取配额: {e}") + return None + + +class QuotaUsageRepository: + """配额使用量数据访问层""" + + def __init__(self, db: Session): + self.db = db + + def count_workspaces(self, tenant_id: UUID) -> int: + from app.models.workspace_model import Workspace + return self.db.query(Workspace).filter( + Workspace.tenant_id == tenant_id, + Workspace.is_active.is_(True) + ).count() + + def count_apps(self, tenant_id: UUID) -> int: + from app.models.app_model import App + from app.models.workspace_model import Workspace + return self.db.query(App).join( + Workspace, App.workspace_id == Workspace.id + ).filter( + Workspace.tenant_id == tenant_id, + App.is_active.is_(True) + ).count() + + def count_skills(self, tenant_id: UUID) -> int: + from app.models.skill_model import Skill + return self.db.query(Skill).filter( + Skill.tenant_id == tenant_id, + Skill.is_active.is_(True) + ).count() + + def sum_knowledge_capacity_gb(self, tenant_id: UUID) -> float: + from app.models.document_model import Document + from app.models.knowledge_model import Knowledge + from app.models.workspace_model import Workspace + result = self.db.query(func.coalesce(func.sum(Document.file_size), 0)).join( + Knowledge, Document.kb_id == Knowledge.id + ).join( + Workspace, Knowledge.workspace_id == Workspace.id + ).filter( + Workspace.tenant_id == tenant_id, + Document.status == 1, + ).scalar() + return float(result) / (1024 ** 3) if result else 0.0 + + def count_memory_engines(self, tenant_id: UUID) -> int: + from app.models.memory_config_model import MemoryConfig + from app.models.workspace_model import Workspace + return self.db.query(MemoryConfig).join( + Workspace, MemoryConfig.workspace_id == Workspace.id + ).filter( + Workspace.tenant_id == tenant_id + ).count() + + def count_end_users(self, tenant_id: UUID) -> int: + from app.models.end_user_model import EndUser + from app.models.workspace_model import Workspace + return self.db.query(EndUser).join( + Workspace, EndUser.workspace_id == Workspace.id + ).filter( + Workspace.tenant_id == tenant_id + ).count() + + def count_models(self, tenant_id: UUID) -> int: + from app.models.models_model import ModelConfig + return self.db.query(ModelConfig).filter( + ModelConfig.tenant_id == tenant_id, + ModelConfig.is_active == True + ).count() + + def count_ontology_projects(self, tenant_id: UUID) -> int: + from app.models.ontology_scene import OntologyScene + from app.models.workspace_model import Workspace + return self.db.query(OntologyScene).join( + Workspace, OntologyScene.workspace_id == Workspace.id + ).filter( + Workspace.tenant_id == tenant_id + ).count() + + def get_usage_by_quota_type(self, tenant_id: UUID, quota_type: str): + """按配额类型分发，返回当前使用量""" + dispatch = { + "workspace_quota": self.count_workspaces, + "app_quota": self.count_apps, + "skill_quota": self.count_skills, + "knowledge_capacity_quota": self.sum_knowledge_capacity_gb, + "memory_engine_quota": self.count_memory_engines, + "end_user_quota": self.count_end_users, + "model_quota": self.count_models, + "ontology_project_quota": self.count_ontology_projects, + } + fn = dispatch.get(quota_type) + return fn(tenant_id) if fn else 0 + + +def _check_quota( + db: Session, + tenant_id: UUID, + quota_type: str, + resource_name: str, + usage_func: Optional[Callable] = None, +) -> None: + """核心配额检查逻辑：对比使用量和配额限制""" + try: + quota_config = _get_quota_config(db, tenant_id) + if not quota_config: + logger.warning(f"租户 {tenant_id} 无有效配额配置，跳过配额检查") + return + + quota_limit = quota_config.get(quota_type) + if quota_limit is None: + logger.warning(f"配额配置未包含 {quota_type}，跳过配额检查") + return + + if usage_func: + current_usage = usage_func(db, tenant_id) + else: + current_usage = QuotaUsageRepository(db).get_usage_by_quota_type(tenant_id, quota_type) + + if current_usage >= quota_limit: + logger.warning( + f"配额不足: tenant={tenant_id}, type={quota_type}, " + f"usage={current_usage}, limit={quota_limit}" + ) + raise QuotaExceededError( + resource=resource_name, + current_usage=current_usage, + quota_limit=quota_limit, + ) + + logger.debug( + f"配额检查通过: tenant={tenant_id}, type={quota_type}, " + f"usage={current_usage}, limit={quota_limit}" + ) + + except QuotaExceededError: + raise + except Exception as e: + logger.error( + f"配额检查异常: tenant={tenant_id}, type={quota_type}, " + f"error_type={type(e).__name__}, error={str(e)}", + exc_info=True, + ) + raise + + +# ─── 具名装饰器 ──────────────────────────────────────────────────────────── + +def check_workspace_quota(func: Callable) -> Callable: + @wraps(func) + def wrapper(*args, **kwargs): + db: Session = kwargs.get("db") + user = _get_user_from_kwargs(kwargs) + if not db or not user: + logger.warning("配额检查失败：缺少 db 或 user 参数") + return func(*args, **kwargs) + _check_quota(db, user.tenant_id, "workspace_quota", "workspace") + return func(*args, **kwargs) + return wrapper + + +def check_skill_quota(func: Callable) -> Callable: + @wraps(func) + def wrapper(*args, **kwargs): + db: Session = kwargs.get("db") + user = _get_user_from_kwargs(kwargs) + if not db or not user: + logger.warning("配额检查失败：缺少 db 或 user 参数") + return func(*args, **kwargs) + _check_quota(db, user.tenant_id, "skill_quota", "skill") + return func(*args, **kwargs) + return wrapper + + +def check_app_quota(func: Callable) -> Callable: + @wraps(func) + def wrapper(*args, **kwargs): + db: Session = kwargs.get("db") + user = _get_user_from_kwargs(kwargs) + if not db or not user: + logger.warning("配额检查失败：缺少 db 或 user 参数") + return func(*args, **kwargs) + _check_quota(db, user.tenant_id, "app_quota", "app") + return func(*args, **kwargs) + return wrapper + + +def check_knowledge_capacity_quota(func: Callable) -> Callable: + @wraps(func) + async def async_wrapper(*args, **kwargs): + db: Session = kwargs.get("db") + if not db: + logger.warning("配额检查失败：缺少 db 参数") + return await func(*args, **kwargs) + tenant_id = _get_tenant_id_from_kwargs(db, kwargs) + if not tenant_id: + logger.warning("配额检查失败：无法获取 tenant_id") + return await func(*args, **kwargs) + _check_quota(db, tenant_id, "knowledge_capacity_quota", "knowledge_capacity") + return await func(*args, **kwargs) + + @wraps(func) + def sync_wrapper(*args, **kwargs): + db: Session = kwargs.get("db") + user = _get_user_from_kwargs(kwargs) + if not db or not user: + logger.warning("配额检查失败：缺少 db 或 user 参数") + return func(*args, **kwargs) + _check_quota(db, user.tenant_id, "knowledge_capacity_quota", "knowledge_capacity") + return func(*args, **kwargs) + + return async_wrapper if asyncio.iscoroutinefunction(func) else sync_wrapper + + +def check_memory_engine_quota(func: Callable) -> Callable: + @wraps(func) + def wrapper(*args, **kwargs): + db: Session = kwargs.get("db") + user = _get_user_from_kwargs(kwargs) + if not db or not user: + logger.warning("配额检查失败：缺少 db 或 user 参数") + return func(*args, **kwargs) + _check_quota(db, user.tenant_id, "memory_engine_quota", "memory_engine") + return func(*args, **kwargs) + return wrapper + + +def check_end_user_quota(func: Callable) -> Callable: + @wraps(func) + async def async_wrapper(*args, **kwargs): + db: Session = kwargs.get("db") + if not db: + logger.warning("配额检查失败：缺少 db 参数") + return await func(*args, **kwargs) + tenant_id = _get_tenant_id_from_kwargs(db, kwargs) + if not tenant_id: + logger.warning("配额检查失败：无法获取 tenant_id") + return await func(*args, **kwargs) + _check_quota(db, tenant_id, "end_user_quota", "end_user") + return await func(*args, **kwargs) + + @wraps(func) + def sync_wrapper(*args, **kwargs): + db: Session = kwargs.get("db") + if not db: + logger.warning("配额检查失败：缺少 db 参数") + return func(*args, **kwargs) + tenant_id = _get_tenant_id_from_kwargs(db, kwargs) + if not tenant_id: + logger.warning("配额检查失败：无法获取 tenant_id") + return func(*args, **kwargs) + _check_quota(db, tenant_id, "end_user_quota", "end_user") + return func(*args, **kwargs) + + return async_wrapper if asyncio.iscoroutinefunction(func) else sync_wrapper + + +def check_ontology_project_quota(func: Callable) -> Callable: + @wraps(func) + def wrapper(*args, **kwargs): + db: Session = kwargs.get("db") + user = _get_user_from_kwargs(kwargs) + if not db or not user: + logger.warning("配额检查失败：缺少 db 或 user 参数") + return func(*args, **kwargs) + _check_quota(db, user.tenant_id, "ontology_project_quota", "ontology_project") + return func(*args, **kwargs) + return wrapper + + +def check_model_quota(func: Callable) -> Callable: + @wraps(func) + def wrapper(*args, **kwargs): + db: Session = kwargs.get("db") + user = _get_user_from_kwargs(kwargs) + if not db or not user: + logger.warning("配额检查失败：缺少 db 或 user 参数") + return func(*args, **kwargs) + _check_quota(db, user.tenant_id, "model_quota", "model") + return func(*args, **kwargs) + return wrapper + + +def check_model_activation_quota(func: Callable) -> Callable: + """模型激活时的配额检查装饰器""" + @wraps(func) + def wrapper(*args, **kwargs): + db: Session = kwargs.get("db") + user = _get_user_from_kwargs(kwargs) + if not db or not user: + logger.warning("配额检查失败：缺少 db 或 user 参数") + return func(*args, **kwargs) + + model_id = kwargs.get("model_id") or (args[1] if len(args) > 1 else None) + model_data = kwargs.get("model_data") + + if not model_id or not model_data: + logger.warning("模型激活配额检查失败：缺少 model_id 或 model_data 参数") + return func(*args, **kwargs) + + if model_data.is_active is True: + try: + from app.models.models_model import ModelConfig + from app.services.model_service import ModelConfigService + + existing_model = ModelConfigService.get_model_by_id( + db=db, + model_id=model_id, + tenant_id=user.tenant_id + ) + + if not existing_model.is_active: + logger.info(f"模型激活操作，检查配额: model_id={model_id}, tenant_id={user.tenant_id}") + _check_quota(db, user.tenant_id, "model_quota", "model") + except Exception as e: + logger.error(f"模型激活配额检查异常: model_id={model_id}, error={str(e)}") + raise + + return func(*args, **kwargs) + return wrapper + + +def check_quota(quota_type: str, resource_name: str, usage_func: Optional[Callable] = None): + """通用配额检查装饰器，支持自定义使用量获取函数""" + def decorator(func: Callable) -> Callable: + @wraps(func) + def wrapper(*args, **kwargs): + db: Session = kwargs.get("db") + user = _get_user_from_kwargs(kwargs) + if not db or not user: + logger.warning("配额检查失败：缺少 db 或 user 参数") + return func(*args, **kwargs) + _check_quota(db, user.tenant_id, quota_type, resource_name, usage_func) + return func(*args, **kwargs) + return wrapper + return decorator + + +# ─── 配额使用统计 ──────────────────────────────────────────────────────────── + +def get_quota_usage(db: Session, tenant_id: UUID) -> dict: + """获取租户所有配额的使用情况""" + quota_config = _get_quota_config(db, tenant_id) + if not quota_config: + return {} + + repo = QuotaUsageRepository(db) + + def pct(used, limit): + return round(used / limit * 100, 1) if limit else None + + workspace_count = repo.count_workspaces(tenant_id) + skill_count = repo.count_skills(tenant_id) + app_count = repo.count_apps(tenant_id) + knowledge_gb = repo.sum_knowledge_capacity_gb(tenant_id) + memory_count = repo.count_memory_engines(tenant_id) + end_user_count = repo.count_end_users(tenant_id) + model_count = repo.count_models(tenant_id) + ontology_count = repo.count_ontology_projects(tenant_id) + + api_ops_current = 0 + try: + from app.core.config import settings + import redis + _now = time.time() + _rk = f"rate_limit:tenant_qps:{tenant_id}" + _r = redis.StrictRedis( + host=settings.REDIS_HOST, port=settings.REDIS_PORT, + db=settings.REDIS_DB, password=settings.REDIS_PASSWORD, + decode_responses=True + ) + api_ops_current = int(_r.zcount(_rk, _now - 1, "+inf")) + except Exception: + pass + + return { + "workspace": {"used": workspace_count, "limit": quota_config.get("workspace_quota"), "percentage": pct(workspace_count, quota_config.get("workspace_quota"))}, + "skill": {"used": skill_count, "limit": quota_config.get("skill_quota"), "percentage": pct(skill_count, quota_config.get("skill_quota"))}, + "app": {"used": app_count, "limit": quota_config.get("app_quota"), "percentage": pct(app_count, quota_config.get("app_quota"))}, + "knowledge_capacity": {"used": round(knowledge_gb, 2), "limit": quota_config.get("knowledge_capacity_quota"), "percentage": pct(knowledge_gb, quota_config.get("knowledge_capacity_quota")), "unit": "GB"}, + "memory_engine": {"used": memory_count, "limit": quota_config.get("memory_engine_quota"), "percentage": pct(memory_count, quota_config.get("memory_engine_quota"))}, + "end_user": {"used": end_user_count, "limit": quota_config.get("end_user_quota"), "percentage": pct(end_user_count, quota_config.get("end_user_quota"))}, + "ontology_project": {"used": ontology_count, "limit": quota_config.get("ontology_project_quota"), "percentage": pct(ontology_count, quota_config.get("ontology_project_quota"))}, + "model": {"used": model_count, "limit": quota_config.get("model_quota"), "percentage": pct(model_count, quota_config.get("model_quota"))}, + "api_ops_rate_limit": {"current": api_ops_current, "limit": quota_config.get("api_ops_rate_limit"), "percentage": None, "unit": "次/秒"}, + } diff --git a/api/app/core/quota_stub.py b/api/app/core/quota_stub.py index b8f82e75..577dfadb 100644 --- a/api/app/core/quota_stub.py +++ b/api/app/core/quota_stub.py @@ -1,44 +1,36 @@ """ -配额检查 stub - 社区版使用，所有检查直接放行。 -企业版通过 premium.platform_admin.quota_decorator 提供真实实现。 +配额检查 stub - 社区版和 SaaS 版统一使用 core.quota_manager 实现 + +所有配额检查逻辑统一在 core 层实现，两个版本共用： +- 社区版：从 default_free_plan.py 读取配额限制 +- SaaS 版：优先从 tenant_subscriptions 表读取，降级到配置文件 """ -from functools import wraps -from typing import Callable +from app.core.quota_manager import ( + check_workspace_quota, + check_skill_quota, + check_app_quota, + check_knowledge_capacity_quota, + check_memory_engine_quota, + check_end_user_quota, + check_ontology_project_quota, + check_model_quota, + check_model_activation_quota, + get_quota_usage, + _check_quota, + QuotaUsageRepository, +) - -def _noop_decorator(func: Callable) -> Callable: - """空装饰器，直接放行""" - return func - - -def _noop_check(*args, **kwargs): - """空检查函数，直接放行""" - pass - - -try: - from premium.platform_admin.quota_decorator import ( - check_workspace_quota, - check_skill_quota, - check_app_quota, - check_knowledge_capacity_quota, - check_memory_engine_quota, - check_end_user_quota, - check_ontology_project_quota, - check_model_quota, - check_model_activation_quota, - get_quota_usage, - _check_quota, - ) -except ModuleNotFoundError: - check_workspace_quota = _noop_decorator - check_skill_quota = _noop_decorator - check_app_quota = _noop_decorator - check_knowledge_capacity_quota = _noop_decorator - check_memory_engine_quota = _noop_decorator - check_end_user_quota = _noop_decorator - check_ontology_project_quota = _noop_decorator - check_model_quota = _noop_decorator - check_model_activation_quota = _noop_decorator - get_quota_usage = lambda db, tenant_id: {} - _check_quota = _noop_check +__all__ = [ + "check_workspace_quota", + "check_skill_quota", + "check_app_quota", + "check_knowledge_capacity_quota", + "check_memory_engine_quota", + "check_end_user_quota", + "check_ontology_project_quota", + "check_model_quota", + "check_model_activation_quota", + "get_quota_usage", + "_check_quota", + "QuotaUsageRepository", +] From d5c8d16e642d8797f8ff00605a30dbd57a6cb0f5 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 15 Apr 2026 16:27:45 +0800 Subject: [PATCH 069/245] fix(web): tool methods add cache --- web/src/views/Workflow/components/CheckList/index.tsx | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/web/src/views/Workflow/components/CheckList/index.tsx b/web/src/views/Workflow/components/CheckList/index.tsx index 0256416a..62132a71 100644 --- a/web/src/views/Workflow/components/CheckList/index.tsx +++ b/web/src/views/Workflow/components/CheckList/index.tsx @@ -118,6 +118,7 @@ const CheckList: FC = ({ workflowRef, appId }) => { const { setCheckResults, getCheckResults } = useWorkflowStore() const results = getCheckResults(appId) const timerRef = useRef>() + const toolMethodsCacheRef = useRef }>>>({}) const runCheck = useCallback(async () => { const graph = workflowRef.current?.graphRef?.current @@ -167,7 +168,10 @@ const CheckList: FC = ({ workflowRef, appId }) => { if (typeof toolId === 'string') { try { - const methods = await getToolMethods(toolId) as Array<{ name: string; parameters: Array<{ name: string; required: boolean }> }> + if (!toolMethodsCacheRef.current[toolId]) { + toolMethodsCacheRef.current[toolId] = await getToolMethods(toolId) as Array<{ name: string; parameters: Array<{ name: string; required: boolean }> }> + } + const methods = toolMethodsCacheRef.current[toolId] const operation = toolParameters?.operation const method = operation ? methods.find(m => m.name === operation) : methods[0] if (method) { From 78ba313262ef94d92307408a5414251da128b858 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Thu, 16 Apr 2026 10:10:30 +0800 Subject: [PATCH 070/245] feat(web): Keep the last 4 characters of the API key as original --- web/src/utils/apiKeyReplacer.ts | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/web/src/utils/apiKeyReplacer.ts b/web/src/utils/apiKeyReplacer.ts index 561f146d..cc455fc3 100644 --- a/web/src/utils/apiKeyReplacer.ts +++ b/web/src/utils/apiKeyReplacer.ts @@ -43,7 +43,8 @@ export const maskApiKeys = (text: string): string => { result = result.replace(API_KEY_PATTERNS[key as keyof typeof API_KEY_PREFIX], (match) => { const prefixLength = API_KEY_PREFIX[key].length const prefix = match.substring(0, prefixLength) - return prefix + '*'.repeat(match.length - prefixLength) + const suffix = match.slice(-4) + return prefix + '*'.repeat(match.length - prefixLength - 4) + suffix }) }) From b6aca0b1e7602477686b48547b5b6ab18fc21507 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Thu, 16 Apr 2026 10:28:26 +0800 Subject: [PATCH 071/245] fix(web): userinfo --- web/src/components/Header/index.tsx | 22 +++++++++++----------- 1 file changed, 11 insertions(+), 11 deletions(-) diff --git a/web/src/components/Header/index.tsx b/web/src/components/Header/index.tsx index 23a89894..6819b202 100644 --- a/web/src/components/Header/index.tsx +++ b/web/src/components/Header/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-02 15:07:49 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-07 12:18:58 + * @Last Modified time: 2026-04-16 10:28:18 */ /** * AppHeader Component @@ -31,7 +31,7 @@ const { Header } = Layout; /** * @param source - Breadcrumb source type ('space' or 'manage'), defaults to 'manage' */ -const AppHeader: FC<{source?: 'space' | 'manage';}> = ({source = 'manage'}) => { +const AppHeader: FC<{ source?: 'space' | 'manage'; }> = ({ source = 'manage' }) => { const { t } = useTranslation(); const location = useLocation(); const settingModalRef = useRef(null) @@ -39,7 +39,7 @@ const AppHeader: FC<{source?: 'space' | 'manage';}> = ({source = 'manage'}) => { const { user, logout } = useUser(); const { allBreadcrumbs } = useMenu(); - + /** * Dynamically select breadcrumb source based on current route * - Knowledge base list: uses 'space' breadcrumb @@ -48,24 +48,24 @@ const AppHeader: FC<{source?: 'space' | 'manage';}> = ({source = 'manage'}) => { */ const getBreadcrumbSource = () => { const pathname = location.pathname; - + // Knowledge base list page uses default space breadcrumb if (pathname === '/knowledge-base') { return 'space'; } - + // Knowledge base detail pages use independent breadcrumb if (pathname.includes('/knowledge-base/') && pathname !== '/knowledge-base') { return 'space-detail'; } - + // Other pages use the passed source return source; }; - + const breadcrumbSource = getBreadcrumbSource(); const breadcrumbs = allBreadcrumbs[breadcrumbSource] || []; - + /** Handle user logout */ const handleLogout = () => { @@ -127,7 +127,7 @@ const AppHeader: FC<{source?: 'space' | 'manage';}> = ({source = 'manage'}) => { onClick: handleLogout, }, ]; - + /** * Format breadcrumb items with proper titles, paths, and click handlers * - Translates i18n keys to display text @@ -146,7 +146,7 @@ const AppHeader: FC<{source?: 'space' | 'manage';}> = ({source = 'manage'}) => { ), }; - + if (!isLast) { if ((menu as any).onClick) { item.onClick = (e: React.MouseEvent) => { @@ -162,7 +162,7 @@ const AppHeader: FC<{source?: 'space' | 'manage';}> = ({source = 'manage'}) => { return item; }); } - + const [open, setOpen] = useState(false); const handleOpenChange = (open: boolean) => { setOpen(open); From f04412c455fd21b83283cd6f7adaf0f26bb7f0a0 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Thu, 16 Apr 2026 10:32:34 +0800 Subject: [PATCH 072/245] fix(web): userinfo --- web/src/components/Header/index.tsx | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/web/src/components/Header/index.tsx b/web/src/components/Header/index.tsx index 6819b202..de87dcfc 100644 --- a/web/src/components/Header/index.tsx +++ b/web/src/components/Header/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-02 15:07:49 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-16 10:28:18 + * @Last Modified time: 2026-04-16 10:31:21 */ /** * AppHeader Component @@ -76,9 +76,11 @@ const AppHeader: FC<{ source?: 'space' | 'manage'; }> = ({ source = 'manage' }) const userMenuItems: MenuProps['items'] = [ { key: '1', - icon: - {/[\u4e00-\u9fa5]/.test(user.username) ? user.username.slice(-2) : user.username[0]} - , + icon: user.username + ? + {/[\u4e00-\u9fa5]/.test(user.username) ? user.username.slice(-2) : user.username[0]} + + : null, label: (<>
{user.username}
{user.email}
@@ -181,9 +183,9 @@ const AppHeader: FC<{ source?: 'space' | 'manage'; }> = ({ source = 'manage' }) overlayClassName={styles.userDropdown} > - + {user.username && {/[\u4e00-\u9fa5]/.test(user.username) ? user.username.slice(-2) : user.username[0]} - + } {user.username}
Date: Thu, 16 Apr 2026 13:35:35 +0800 Subject: [PATCH 073/245] fix(model): fix issue where associated model config status was not updated when deleting API Key When deleting an API Key, check if the associated model configuration has other active keys; if not, automatically set it to inactive. Also optimize the model configuration query method to support multi-type queries and add sorting conditions. --- api/app/repositories/model_repository.py | 25 ++++++++++++------------ api/app/services/model_service.py | 13 +++++++++++- 2 files changed, 25 insertions(+), 13 deletions(-) diff --git a/api/app/repositories/model_repository.py b/api/app/repositories/model_repository.py index 8c477d39..03870b4d 100644 --- a/api/app/repositories/model_repository.py +++ b/api/app/repositories/model_repository.py @@ -263,16 +263,15 @@ class ModelConfigRepository: raise @staticmethod - def get_by_type(db: Session, model_type: ModelType, tenant_id: uuid.UUID | None = None, is_active: bool = True) -> List[ModelConfig]: - """根据类型获取模型配置""" - db_logger.debug(f"根据类型查询模型配置: type={model_type}, tenant_id={tenant_id}, is_active={is_active}") - + def get_by_type(db: Session, model_types: List[ModelType], tenant_id: uuid.UUID | None = None, is_active: bool = True) -> List[ModelConfig]: + """根据类型获取模型配置，支持多类型查询""" + db_logger.debug(f"根据类型查询模型配置: types={[t.value for t in model_types]}, tenant_id={tenant_id}, is_active={is_active}") + try: query = db.query(ModelConfig).options( joinedload(ModelConfig.api_keys) - ).filter(ModelConfig.type == model_type) - - # 添加租户过滤 + ).filter(ModelConfig.type.in_([t.value for t in model_types])) + if tenant_id: query = query.filter( or_( @@ -280,16 +279,18 @@ class ModelConfigRepository: ModelConfig.is_public ) ) - + if is_active: query = query.filter(ModelConfig.is_active) - - models = query.order_by(ModelConfig.name).all() + + query = query.filter(ModelConfig.is_composite == False) + + models = query.order_by(ModelConfig.created_at.desc()).all() db_logger.debug(f"根据类型查询模型配置成功: 数量={len(models)}") return models - + except Exception as e: - db_logger.error(f"根据类型查询模型配置失败: type={model_type} - {str(e)}") + db_logger.error(f"根据类型查询模型配置失败: types={model_types} - {str(e)}") raise @staticmethod diff --git a/api/app/services/model_service.py b/api/app/services/model_service.py index 4cbb3509..d202b83a 100644 --- a/api/app/services/model_service.py +++ b/api/app/services/model_service.py @@ -729,10 +729,21 @@ class ModelApiKeyService: @staticmethod def delete_api_key(db: Session, api_key_id: uuid.UUID) -> bool: """删除API Key""" - if not ModelApiKeyRepository.get_by_id(db, api_key_id): + api_key = ModelApiKeyRepository.get_by_id(db, api_key_id) + if not api_key: raise BusinessException("API Key不存在", BizCode.NOT_FOUND) + model_config_ids = [mc.id for mc in api_key.model_configs] + success = ModelApiKeyRepository.delete(db, api_key_id) + + for model_config_id in model_config_ids: + model_config = ModelConfigRepository.get_by_id(db, model_config_id) + if model_config: + has_active_key = any(key.is_active for key in model_config.api_keys) + if not has_active_key and model_config.is_active: + model_config.is_active = False + db.commit() return success From a01525e239b00f3fbeed87c769eb8ada5b08863b Mon Sep 17 00:00:00 2001 From: Eternity <1533512157@qq.com> Date: Thu, 16 Apr 2026 13:27:36 +0800 Subject: [PATCH 074/245] refactor(memory): consolidate memory search services and update model client handling - Consolidate memory search services by removing separate content_search.py and perceptual_search.py - Update model client handling in base_pipeline.py to use ModelApiKeyService for LLM client initialization - Add new prompt files and modify existing services to support consolidated search architecture - Refactor memory read pipeline and related services to use updated model client approach --- api/app/core/memory/memory_service.py | 11 +- api/app/core/memory/models/service_models.py | 17 +- .../core/memory/pipelines/base_pipeline.py | 31 ++- api/app/core/memory/pipelines/memory_read.py | 34 ++- api/app/core/memory/prompt/__init__.py | 85 +++++++ .../core/memory/prompt/problem_split.jinja2 | 212 ++++++++++++++++ .../memory/read_services/content_search.py | 46 ++-- .../read_services/memory_search/__init__.py | 0 .../memory_search/content_search.py | 14 -- .../memory_search/perceptual_search.py | 228 ------------------ .../read_services/query_preprocessor.py | 29 ++- .../memory/read_services/result_builder.py | 2 +- .../memory/read_services/retrieval_summary.py | 11 + api/app/core/memory/utils/llm/llm_utils.py | 48 +++- api/app/repositories/neo4j/graph_search.py | 10 +- .../prompt/prompt_optimizer_system.jinja2 | 2 +- 16 files changed, 471 insertions(+), 309 deletions(-) create mode 100644 api/app/core/memory/prompt/__init__.py create mode 100644 api/app/core/memory/prompt/problem_split.jinja2 delete mode 100644 api/app/core/memory/read_services/memory_search/__init__.py delete mode 100644 api/app/core/memory/read_services/memory_search/content_search.py delete mode 100644 api/app/core/memory/read_services/memory_search/perceptual_search.py create mode 100644 api/app/core/memory/read_services/retrieval_summary.py diff --git a/api/app/core/memory/memory_service.py b/api/app/core/memory/memory_service.py index 67c814b1..15ea14b4 100644 --- a/api/app/core/memory/memory_service.py +++ b/api/app/core/memory/memory_service.py @@ -1,7 +1,7 @@ from sqlalchemy.orm import Session from app.core.memory.enums import StorageType, SearchStrategy -from app.core.memory.models.service_models import Memory, MemoryContext +from app.core.memory.models.service_models import MemoryContext, MemorySearchResult from app.core.memory.pipelines.memory_read import ReadPipeLine from app.db import get_db_context from app.services.memory_config_service import MemoryConfigService @@ -35,9 +35,14 @@ class MemoryService: async def write(self, messages: list[dict]) -> str: raise NotImplementedError - async def read(self, query: str, history: list, search_switch: SearchStrategy) -> list[Memory]: + async def read( + self, + query: str, + search_switch: SearchStrategy, + limit: int = 10, + ) -> MemorySearchResult: with get_db_context() as db: - return await ReadPipeLine(self.ctx, db).run(query, search_switch, limit=10) + return await ReadPipeLine(self.ctx, db).run(query, search_switch, limit) async def forget(self, max_batch: int = 100, min_days: int = 30) -> dict: raise NotImplementedError diff --git a/api/app/core/memory/models/service_models.py b/api/app/core/memory/models/service_models.py index 82a867c7..477c0ba8 100644 --- a/api/app/core/memory/models/service_models.py +++ b/api/app/core/memory/models/service_models.py @@ -1,6 +1,7 @@ -from pydantic import BaseModel, Field, field_serializer, ConfigDict +from pydantic import BaseModel, Field, field_serializer, ConfigDict, model_validator, computed_field from app.core.memory.enums import Neo4jNodeType, StorageType +from app.core.validators import file_validator from app.schemas.memory_config_schema import MemoryConfig @@ -24,3 +25,17 @@ class Memory(BaseModel): @field_serializer("source") def serialize_source(self, v) -> str: return v.value + + +class MemorySearchResult(BaseModel): + memories: list[Memory] + + @computed_field + @property + def content(self) -> str: + return "\n".join([memory.content for memory in self.memories]) + + @computed_field + @property + def count(self) -> int: + return len(self.memories) diff --git a/api/app/core/memory/pipelines/base_pipeline.py b/api/app/core/memory/pipelines/base_pipeline.py index 322f6787..60c48b9d 100644 --- a/api/app/core/memory/pipelines/base_pipeline.py +++ b/api/app/core/memory/pipelines/base_pipeline.py @@ -4,22 +4,32 @@ from typing import Any from sqlalchemy.orm import Session -from app.core.memory.llm_tools import OpenAIEmbedderClient from app.core.memory.models.service_models import MemoryContext -from app.core.models import RedBearModelConfig +from app.core.models import RedBearModelConfig, RedBearLLM, RedBearEmbeddings from app.services.memory_config_service import MemoryConfigService +from app.services.model_service import ModelApiKeyService class ModelClientMixin(ABC): @staticmethod - def get_llm_client(db: Session, model_id: uuid.UUID): - pass + def get_llm_client(db: Session, model_id: uuid.UUID) -> RedBearLLM: + api_config = ModelApiKeyService.get_available_api_key(db, model_id) + return RedBearLLM( + RedBearModelConfig( + model_name=api_config.model_name, + provider=api_config.provider, + api_key=api_config.api_key, + base_url=api_config.api_base, + is_omni=api_config.is_omni, + support_thinking="thinking" in (api_config.capability or []), + ) + ) @staticmethod - def get_embedding_client(db: Session, model_id: uuid.UUID) -> OpenAIEmbedderClient: + def get_embedding_client(db: Session, model_id: uuid.UUID) -> RedBearEmbeddings: config_service = MemoryConfigService(db) embedder_client_config = config_service.get_embedder_config(str(model_id)) - return OpenAIEmbedderClient( + return RedBearEmbeddings( RedBearModelConfig( model_name=embedder_client_config["model_name"], provider=embedder_client_config["provider"], @@ -30,10 +40,15 @@ class ModelClientMixin(ABC): class BasePipeline(ABC): - def __init__(self, ctx: MemoryContext, db: Session): + def __init__(self, ctx: MemoryContext): self.ctx = ctx - self.db = db @abstractmethod async def run(self, *args, **kwargs) -> Any: pass + + +class DBRequiredPipeline(BasePipeline, ABC): + def __init__(self, ctx: MemoryContext, db: Session): + super().__init__(ctx) + self.db = db diff --git a/api/app/core/memory/pipelines/memory_read.py b/api/app/core/memory/pipelines/memory_read.py index 5f5a1a1f..83662d90 100644 --- a/api/app/core/memory/pipelines/memory_read.py +++ b/api/app/core/memory/pipelines/memory_read.py @@ -1,31 +1,41 @@ -from app.core.memory.enums import SearchStrategy -from app.core.memory.pipelines.base_pipeline import BasePipeline, ModelClientMixin -from app.core.memory.read_services.content_search import Neo4jSearchService +from app.core.memory.enums import SearchStrategy, StorageType +from app.core.memory.models.service_models import MemorySearchResult +from app.core.memory.pipelines.base_pipeline import ModelClientMixin, DBRequiredPipeline +from app.core.memory.read_services.content_search import Neo4jSearchService, RAGSearchService from app.core.memory.read_services.query_preprocessor import QueryPreprocessor -class ReadPipeLine(ModelClientMixin, BasePipeline): - async def run(self, query: str, search_switch: SearchStrategy, limit: int = 10): +class ReadPipeLine(ModelClientMixin, DBRequiredPipeline): + async def run(self, query: str, search_switch: SearchStrategy, limit: int = 10, includes=None) -> MemorySearchResult: query = QueryPreprocessor.process(query) + if self.ctx.storage_type == StorageType.RAG: + return await self._rag_read(query, limit) match search_switch: case SearchStrategy.DEEP: - return await self._deep_read() + return await self._deep_read(query, limit, includes) case SearchStrategy.NORMAL: - return await self._normal_read(query) + return await self._normal_read(query, limit, includes) case SearchStrategy.QUICK: - return await self._quick_read(query, limit) + return await self._quick_read(query, limit, includes) case _: raise RuntimeError("Unsupported search strategy") - async def _deep_read(self): + async def _rag_read(self, query: str, limit: int) -> MemorySearchResult: + service = RAGSearchService( + self.ctx + ) + return await service.search() + + async def _deep_read(self, query: str, limit: int, includes=None) -> MemorySearchResult: pass - async def _normal_read(self, query): + async def _normal_read(self, query: str, limit: int, includes=None) -> MemorySearchResult: pass - async def _quick_read(self, query, limit): + async def _quick_read(self, query: str, limit: int, includes=None) -> MemorySearchResult: search_service = Neo4jSearchService( self.ctx, - self.get_embedding_client(self.db, self.ctx.memory_config.embedding_model_id) + self.get_embedding_client(self.db, self.ctx.memory_config.embedding_model_id), + includes=includes, ) return await search_service.search(query, limit) diff --git a/api/app/core/memory/prompt/__init__.py b/api/app/core/memory/prompt/__init__.py new file mode 100644 index 00000000..299470f8 --- /dev/null +++ b/api/app/core/memory/prompt/__init__.py @@ -0,0 +1,85 @@ +import logging +import threading +from pathlib import Path + +from jinja2 import Environment, FileSystemLoader, TemplateNotFound, TemplateSyntaxError + +logger = logging.getLogger(__name__) + +PROMPT_DIR = Path(__file__).parent + + +class PromptRenderError(Exception): + def __init__(self, template_name: str, error: Exception): + self.template_name = template_name + self.error = error + super().__init__(f"Failed to render prompt '{template_name}': {error}") + + +class PromptManager: + _instance = None + _lock = threading.Lock() + + def __new__(cls, *args, **kwargs): + if cls._instance is None: + with cls._lock: + if cls._instance is None: + cls._instance = super().__new__(cls) + cls._instance._init_once() + return cls._instance + + def _init_once(self): + self.env = Environment( + loader=FileSystemLoader(str(PROMPT_DIR)), + autoescape=False, + keep_trailing_newline=True, + ) + logger.info(f"PromptManager initialized: template_dir={PROMPT_DIR}") + + def __repr__(self): + templates = self.list_templates() + return f"" + + def list_templates(self) -> list[str]: + return [ + Path(name).stem + for name in self.env.loader.list_templates() + if name.endswith('.jinja2') + ] + + def get(self, name: str) -> str: + template_name = self._resolve_name(name) + try: + source, _, _ = self.env.loader.get_source(self.env, template_name) + return source + except TemplateNotFound: + raise FileNotFoundError( + f"Prompt '{name}' not found. " + f"Available: {self.list_templates()}" + ) + + def render(self, name: str, **kwargs) -> str: + template_name = self._resolve_name(name) + try: + template = self.env.get_template(template_name) + return template.render(**kwargs) + except TemplateNotFound: + raise FileNotFoundError( + f"Prompt '{name}' not found. " + f"Available: {self.list_templates()}" + ) + except TemplateSyntaxError as e: + logger.error(f"Prompt syntax error in '{name}': {e}", exc_info=True) + raise PromptRenderError(name, e) + except Exception as e: + logger.error(f"Prompt render failed for '{name}': {e}", exc_info=True) + raise PromptRenderError(name, e) + + @staticmethod + def _resolve_name(name: str) -> str: + if not name.endswith('.jinja2'): + return f"{name}.jinja2" + return name + + +prompt_manager = PromptManager() diff --git a/api/app/core/memory/prompt/problem_split.jinja2 b/api/app/core/memory/prompt/problem_split.jinja2 new file mode 100644 index 00000000..ff134ddb --- /dev/null +++ b/api/app/core/memory/prompt/problem_split.jinja2 @@ -0,0 +1,212 @@ + +# 角色：{#InputSlot placeholder="角色名称" mode="input"#}{#/InputSlot#} +你是一个智能数据拆分助手，请根据数据特性判断输入属于哪种类型： +## 目标： +你需要根据以下类型对输入数据进行分类，并生成相应的拆分策略和示例。 +--- + +### 历史信息参考 +在生成扩展问题时，你可以参考以下历史数据（如果提供）： +- 历史对话或任务的主题； +- 历史中出现的关键实体（时间、人物、地点、研究主题等）； +- 历史中已解答的问题（避免重复）； +- 历史推理链（保持逻辑一致性）。 + +> 如果没有提供历史信息，则仅根据当前输入问题进行分析。 +输入历史信息内容:{{history}} + +## User Input +{{ sentence }} + +## 需求： +1:首先判断类型（单跳、多跳、开放域、时间）。 +2:根据类型进行拆分。 +3:拆分后的内容需保证信息完整且可独立处理。 +4:对每个拆分条目，可附加示例或说明。 +5:拆分问题的时候可以考虑输入的历史内容，以保持逻辑连贯。 + 比如：输入历史信息内容:[{'Query': '4月27日，我和你推荐过一本书，书名是什么？', 'ANswer': '张曼玉推荐了《小王子》'}] + 拆分问题：4月27日，我和你推荐过一本书，书名是什么？，可以拆分为：4月27日，张曼玉推荐过一本书，书名是什么？ + +## 指代消歧规则（Coreference Resolution）： +在拆分问题时，必须解析并替换所有指代词和抽象称呼，使问题具体化： + +1. **"用户"的消歧**： + - "用户是谁？" → 分析历史记录，找出对话发起者的姓名 + - 如果历史中有"我叫X"、"我的名字是X"、或多次提到某个人物（如"老李"、"李建国"），则"用户"指的就是这个人 + - 示例：历史中反复出现"老李/李建国/建国哥"，则"用户是谁？"应拆分为"李建国是谁？"或"老李（李建国）是谁？" + +2. **"我"的消歧**： + - "我喜欢什么？" → 从历史中找出对话发起者的姓名，替换为"X喜欢什么？" + - 示例：历史中有"张曼玉推荐了《小王子》"，则"我推荐的书是什么？"应拆分为"张曼玉推荐的书是什么？" + +3. **"他/她/它"的消歧**： + - 从上下文或历史中找出最近提到的同类实体 + - 示例：历史中有"老李的同事叫他建国哥"，则"他的同事怎么称呼他？"应拆分为"老李的同事怎么称呼他？" + +4. **"那个人/这个人"的消歧**： + - 从历史中找出最近提到的人物 + - 示例：历史中有"李建国"，则"那个人的原名是什么？"应拆分为"李建国的原名是什么？" + +5. **优先级**： + - 如果历史记录中反复出现某个人物（如"老李"、"李建国"、"建国哥"），则"用户"很可能指的就是这个人 + - 如果无法从历史中确定指代对象，保留原问题，但在reason中说明"无法确定指代对象" + +## 指令： +你是一个智能数据拆分助手，请根据数据特性判断输入属于哪种类型： +单跳（Single-hop） + 描述：问题或数据只需要通过一步即可得到答案或完成拆分，不依赖其他信息。 + 拆分策略：直接识别核心信息或关键字段，生成可独立处理的片段。 + 示例： + 输入数据："请列出今年诺贝尔物理学奖的得主" + 拆分结果：[ + { + "id": "Q1", + "question": "今年诺贝尔物理学奖得主是谁", + "type": "单跳’", + } + ] + 注意： 当遇到上下文依赖问题时，明确指出缺失的信息类型并且，question可填写输入问题 +多跳（Multi-hop）: + 描述：问题或数据需要通过多步推理或跨多个信息源才能得到答案。 + 拆分策略：将问题拆解为多个子问题，每个子问题对应一个独立处理步骤，需要具备推理链条与逻辑连接数量。 + 示例： + 输入数据："今年诺贝尔物理学奖得主的研究领域及代表性成果" + 拆分结果： + [ + { + "id": "Q1", + "question": 今年诺贝尔物理学奖得主是谁？", + "type": "多跳’", + }, + { + "id": "Q2", + "question": "该得主的研究领域是什么？", + "type": "多跳’", + }, + { + "id": "Q3", + "question": "该得主的代表性成果有哪些？", + "type": "多跳’" + } + ] +开放域（Open-domain）: + 描述：问题或数据不局限于特定知识库，需要从大范围信息中检索和生成答案，而不是从一个已知的小范围数据源中查找。。 + 拆分策略：根据主题或关键实体拆分，同时保留上下文以便检索外部知识，问题涉及一般性、常识性、跨学科内容，可能是开放式回答（描述性、推理性、综合性） + 需要外部知识检索或推理才能确定，比如：“为什么人类需要睡眠？”、“量子计算与经典计算的主要区别是什么？”。 + 示例： + 输入数据："介绍量子计算的最新研究进展" + 拆分结果： + [ + { + "id": "Q1", + "question": 量子计算的基本概念是什么？", + "type": "开放域’", + }, + { + "id": "Q2", + "question": "当前量子计算的主要研究方向有哪些？", + "type": "开放域’", + }, + { + "id": "Q3", + "question": "近期在量子计算领域有哪些重大进展？", + "type": "开放域’", + } + ] + +时间（Temporal）: + 描述：问题或数据涉及时间维度，需要按时间顺序或时间点拆分。 + 拆分策略：根据事件时间或时间段拆分为独立条目或问题。 + 示例： + 输入数据："列出苹果公司过去五年的重大事件" + 拆分结果： + [ + { + "id": "Q1", + "question": 苹果公司2019年的重大事件有哪些？", + "type": "时间’", + }, + { + "id": "Q2", + "question": "苹果公司2020年的重大事件有哪些？", + "type": "时间’", + }, + { + "id": "Q3", + "question": "苹果公司2021年的重大事件有哪些？", + "type": "时间’", + }, + { + "id": "Q3", + "question": "苹果公司2022年的重大事件有哪些？", + "type": "时间’", + } + , + { + "id": "Q4", + "question": "苹果公司2023年的重大事件有哪些？", + "type": "时间’", + } + ] + +输出要求： +- 每个子问题包括： + - `id`: 子问题编号（Q1, Q2...） + - `question`: 子问题内容 + - `type`: 类型（事实检索 / 澄清 / 定义 / 比较 / 行动建议等） + - `reason`: 拆分的理由（为什么要这样拆） +- 格式案例: +[ + { + "id": "Q1", + "question": 量子计算的基本概念是什么？", + "type": "开放域’", + }, + { + "id": "Q2", + "question": "当前量子计算的主要研究方向有哪些？", + "type": "开放域’", + }, + { + "id": "Q3", + "question": "近期在量子计算领域有哪些重大进展？", + "type": "开放域’", + } +] +- 必须通过json.loads()的格式支持的形式输出 +- 必须通过json.loads()的格式支持的形式输出,响应必须是与此确切模式匹配的有效JSON对象。不要在JSON之前或之后包含任何文本。 + +## 指代消歧示例（重要）： +示例1 - "用户"的消歧： +输入历史：[{'Query': '老李的原名叫什么？', 'Answer': '李建国'}, {'Query': '老李的同事叫他什么？', 'Answer': '建国哥'}] +输入问题："用户是谁？" +输出： +[ + { + "id": "Q1", + "question": "李建国是谁？", + "type": "单跳", + "reason": "历史中反复提到'老李/李建国/建国哥'，'用户'指的就是对话发起者李建国" + } +] + +示例2 - "我"的消歧： +输入历史：[{'Query': '张曼玉推荐了什么书？', 'Answer': '《小王子》'}] +输入问题："我推荐的书是什么？" +输出： +[ + { + "id": "Q1", + "question": "张曼玉推荐的书是什么？", + "type": "单跳", + "reason": "历史中提到张曼玉推荐了书，'我'指的就是张曼玉" + } +] + +- 关键的JSON格式要求 +1.JSON结构仅使用标准ASCII双引号（“）-切勿使用中文引号（“”）或其他Unicode引号 +2.如果提取的语句文本包含引号，请使用反斜杠（\“）正确转义它们 +3.确保所有JSON字符串都正确关闭并以逗号分隔 +4.JSON字符串值中不包括换行符 +5.正确转义的例子：“statement”：“Zhang Xinhua said：\”我非常喜欢这本书\"" +6.不允许输出```json```相关符号，如```json```、``````、```python```、```javascript```、```html```、```css```、```sql```、```java```、```c```、```c++```、```c#```、```ruby``` diff --git a/api/app/core/memory/read_services/content_search.py b/api/app/core/memory/read_services/content_search.py index 69ca6b11..58356e84 100644 --- a/api/app/core/memory/read_services/content_search.py +++ b/api/app/core/memory/read_services/content_search.py @@ -1,37 +1,28 @@ import asyncio import logging import math -import time - -from pydantic import BaseModel, Field from app.core.memory.enums import Neo4jNodeType -from app.core.memory.llm_tools import OpenAIEmbedderClient from app.core.memory.memory_service import MemoryContext -from app.core.memory.models.service_models import Memory +from app.core.memory.models.service_models import Memory, MemorySearchResult from app.core.memory.read_services.result_builder import data_builder_factory +from app.core.models import RedBearEmbeddings from app.repositories.neo4j.graph_search import search_graph, search_graph_by_embedding from app.repositories.neo4j.neo4j_connector import Neo4jConnector logger = logging.getLogger(__name__) - -class MemorySearchResult(BaseModel): - memories: dict[str, list[dict]] = Field(default_factory=dict) - content: str = Field(default="") - count: int = Field(default=0) +DEFAULT_ALPHA = 0.7 +DEFAULT_FULLTEXT_SCORE_THRESHOLD = 1 +DEFAULT_COSINE_SCORE_THRESHOLD = 0.5 +DEFAULT_CONTENT_SCORE_THRESHOLD = 0.5 class Neo4jSearchService: - DEFAULT_ALPHA = 0.6 - DEFAULT_FULLTEXT_SCORE_THRESHOLD = 1 - DEFAULT_COSINE_SCORE_THRESHOLD = 0.5 - DEFAULT_CONTENT_SCORE_THRESHOLD = 0.5 - def __init__( self, ctx: MemoryContext, - embedder: OpenAIEmbedderClient, + embedder: RedBearEmbeddings, includes: list[Neo4jNodeType] | None = None, alpha: float = DEFAULT_ALPHA, fulltext_score_threshold: float = DEFAULT_FULLTEXT_SCORE_THRESHOLD, @@ -44,7 +35,7 @@ class Neo4jSearchService: self.cosine_score_threshold = cosine_score_threshold self.content_score_threshold = content_score_threshold - self.embedder: OpenAIEmbedderClient = embedder + self.embedder: RedBearEmbeddings = embedder self.connector: Neo4jConnector | None = None self.includes = includes @@ -121,9 +112,12 @@ class Neo4jSearchService: kw = float(combined[item_id].get("kw_score", 0) or 0) emb = float(combined[item_id].get("embedding_score", 0) or 0) base = self.alpha * emb + (1 - self.alpha) * kw - combined[item_id]["content_score"] = base + min(1 - base, 0.1 * kw * emb) + combined[item_id]["content_score"] = base + min(1 - base, kw * emb) results = sorted(combined.values(), key=lambda x: x["content_score"], reverse=True) - # results = [res for res in results if res["content_score"] > self.content_score_threshold] + # results = [ + # res for res in results + # if res["content_score"] > self.content_score_threshold + # ] results = results[:limit] logger.info( @@ -137,14 +131,14 @@ class Neo4jSearchService: return items scores = [float(it.get("score", 0) or 0) for it in items] for it, s in zip(items, scores): - it[f"normalized_kw_score"] = 1 / (1 + math.exp(-(s - self.fulltext_score_threshold) / 2)) + it[f"normalized_kw_score"] = 1 / (1 + math.exp(-(s - self.fulltext_score_threshold) / 2)) if s else 0 return items async def search( self, query: str, limit: int = 10, - ) -> list[Memory]: + ) -> MemorySearchResult: async with Neo4jConnector() as connector: self.connector = connector kw_task = self._keyword_search(query, limit) @@ -175,4 +169,12 @@ class Neo4jSearchService: query=query )) memories.sort(key=lambda x: x.score, reverse=True) - return memories[:limit] + return MemorySearchResult(memories=memories[:limit]) + + +class RAGSearchService: + def __init__(self, ctx: MemoryContext): + pass + + async def search(self) -> MemorySearchResult: + pass diff --git a/api/app/core/memory/read_services/memory_search/__init__.py b/api/app/core/memory/read_services/memory_search/__init__.py deleted file mode 100644 index e69de29b..00000000 diff --git a/api/app/core/memory/read_services/memory_search/content_search.py b/api/app/core/memory/read_services/memory_search/content_search.py deleted file mode 100644 index f5e58696..00000000 --- a/api/app/core/memory/read_services/memory_search/content_search.py +++ /dev/null @@ -1,14 +0,0 @@ -# -*- coding: UTF-8 -*- -# Author: Eternity -# @Email: 1533512157@qq.com -# @Time : 2026/4/9 16:48 -from app.core.memory.llm_tools import OpenAIEmbedderClient -from app.core.memory.memory_service import MemoryContext - - -class ContentSearch: - def __init__(self, ctx: MemoryContext): - self.ctx = ctx - - async def search(self, query): - pass \ No newline at end of file diff --git a/api/app/core/memory/read_services/memory_search/perceptual_search.py b/api/app/core/memory/read_services/memory_search/perceptual_search.py deleted file mode 100644 index db81e2f8..00000000 --- a/api/app/core/memory/read_services/memory_search/perceptual_search.py +++ /dev/null @@ -1,228 +0,0 @@ -import asyncio -import logging -from typing import Any - -from pydantic import BaseModel - -from app.core.memory.llm_tools import OpenAIEmbedderClient -from app.core.memory.memory_service import MemoryContext -from app.core.memory.utils.data import escape_lucene_query -from app.repositories.neo4j.graph_search import search_perceptual, search_perceptual_by_embedding -from app.repositories.neo4j.neo4j_connector import Neo4jConnector - -logger = logging.getLogger(__name__) - - -class PerceptualResult(BaseModel): - memories: list[dict[str, Any]] = [] - content: str = "" - keyword_raw: int = 0 - embedding_raw: int = 0 - - -class PerceptualRetrieverService: - DEFAULT_ALPHA = 0.6 - DEFAULT_FULLTEXT_SCORE_THRESHOLD = 0.5 - DEFAULT_COSINE_SCORE_THRESHOLD = 0.7 - - def __init__( - self, - ctx: MemoryContext, - embedder: OpenAIEmbedderClient, - alpha: float = DEFAULT_ALPHA, - fulltext_score_threshold: float = DEFAULT_FULLTEXT_SCORE_THRESHOLD, - cosine_score_threshold: float = DEFAULT_COSINE_SCORE_THRESHOLD - ): - self.ctx = ctx - self.alpha = alpha - self.fulltext_score_threshold = fulltext_score_threshold - self.cosine_score_threshold = cosine_score_threshold - - self.embedder: OpenAIEmbedderClient = embedder - self.connector = Neo4jConnector() - - async def search( - self, - query: str, - keywords: list[str] | None = None, - limit: int = 10 - ) -> PerceptualResult: - if keywords is None: - keywords = [query] if query else [] - - try: - kw_task = self._keyword_search(keywords, limit) - emb_task = self._embedding_search(query, limit) - kw_results, emb_results = await asyncio.gather(kw_task, emb_task, return_exceptions=True) - if isinstance(kw_results, Exception): - logger.warning(f"[PerceptualSearch] keyword search error: {kw_results}") - kw_results = [] - if isinstance(emb_results, Exception): - logger.warning(f"[PerceptualSearch] embedding search error: {emb_results}") - emb_results = [] - - reranked = self._rerank(kw_results, emb_results, limit) - - memories = [] - content_parts = [] - for record in reranked: - fmt = self._format_result(record) - fmt["score"] = round(record.get("content_score", 0), 4) - memories.append(fmt) - content_parts.append(self._build_content_text(fmt)) - - logger.info( - f"[PerceptualSearch] {len(memories)} results after rerank " - f"(keyword_raw={len(kw_results)}, embedding_raw={len(emb_results)})" - ) - return PerceptualResult( - memories=memories, - content="\n\n".join(content_parts), - keyword_raw=len(kw_results), - embedding_raw=len(emb_results), - ) - except Exception as e: - logger.error(f"[PerceptualSearch] search failed: {e}", exc_info=True) - return PerceptualResult() - finally: - await self.connector.close() - - async def _keyword_search( - self, - keywords: list[str], - limit: int - ) -> list[dict]: - seen_ids: set = set() - all_results: list[dict] = [] - - async def _one(kw: str): - escaped = escape_lucene_query(kw) - if not escaped.strip(): - return [] - r = await search_perceptual( - connector=self.connector, q=escaped, - end_user_id=self.ctx.end_user_id, limit=limit - ) - perceptuals = r.get("perceptuals", []) - return [perceptual for perceptual in perceptuals if perceptual["score"] > self.fulltext_score_threshold] - - tasks = [_one(kw) for kw in keywords] - batch = await asyncio.gather(*tasks, return_exceptions=True) - - for result in batch: - if isinstance(result, Exception): - logger.warning(f"[PerceptualSearch] keyword sub-query error: {result}") - continue - for rec in result: - rid = rec.get("id", "") - if rid and rid not in seen_ids: - seen_ids.add(rid) - all_results.append(rec) - all_results.sort(key=lambda x: float(x.get("score", 0)), reverse=True) - return all_results[:limit] - - async def _embedding_search( - self, - query: str, - limit: int - ) -> list[dict]: - r = await search_perceptual_by_embedding( - connector=self.connector, - embedder_client=self.embedder, - query_text=query, - end_user_id=self.ctx.end_user_id, - limit=limit - ) - perceptuals = r.get("perceptuals", []) - return [perceptual for perceptual in perceptuals if perceptual["score"] > self.cosine_score_threshold] - - def _rerank( - self, - keyword_results: list[dict], - embedding_results: list[dict], - limit: int, - ) -> list[dict]: - keyword_results = self._normalize_scores(keyword_results) - embedding_results = self._normalize_scores(embedding_results) - - kw_norm_map = {} - for item in keyword_results: - item_id = item["id"] - kw_norm_map[item_id] = float(item.get("normalized_score", 0)) - - emb_norm_map = {} - for item in embedding_results: - item_id = item["id"] - emb_norm_map[item_id] = float(item.get("normalized_score", 0)) - - combined = {} - for item in keyword_results: - item_id = item["id"] - combined[item_id] = item.copy() - combined[item_id]["kw_score"] = kw_norm_map.get(item_id, 0) - combined[item_id]["embedding_score"] = emb_norm_map.get(item_id, 0) - - for item in embedding_results: - item_id = item["id"] - if item_id in combined: - combined[item_id]["embedding_score"] = emb_norm_map.get(item_id, 0) - else: - combined[item_id] = item.copy() - combined[item_id]["kw_score"] = kw_norm_map.get(item_id, 0) - combined[item_id]["embedding_score"] = emb_norm_map.get(item_id, 0) - - for item in combined.values(): - kw = float(item.get("kw_score", 0) or 0) - emb = float(item.get("embedding_score", 0) or 0) - item["content_score"] = self.alpha * emb + (1 - self.alpha) * kw - - results = list(combined.values()) - results.sort(key=lambda x: x["content_score"], reverse=True) - results = results[:limit] - - logger.info( - f"[PerceptualSearch] rerank: merged={len(combined)}, after_threshold={len(results)} " - f"(alpha={self.alpha})" - ) - return results - - @staticmethod - def _normalize_scores(items: list[dict], field: str = "score") -> list[dict]: - """Min-max 归一化，将分数线性映射到 [0, 1]。""" - if not items: - return items - scores = [float(it.get(field, 0) or 0) for it in items] - min_s = min(scores) - max_s = max(scores) - diff = max_s - min_s - for it, s in zip(items, scores): - it[f"normalized_{field}"] = (s - min_s) / diff if diff > 0 else 1.0 - return items - - @staticmethod - def _format_result(record: dict) -> dict: - return { - "id": record.get("id", ""), - "perceptual_type": record.get("perceptual_type", ""), - "file_name": record.get("file_name", ""), - "file_path": record.get("file_path", ""), - "summary": record.get("summary", ""), - "topic": record.get("topic", ""), - "domain": record.get("domain", ""), - "keywords": record.get("keywords", []), - "created_at": str(record.get("created_at", "")), - "file_type": record.get("file_type", ""), - "score": record.get("score", 0), - } - - @staticmethod - def _build_content_text(formatted: dict) -> str: - content_text = (f"" - f"{formatted["file_name"]}" - f"{formatted["file_path"]}" - f"{formatted["file_type"]}" - f"{formatted["topic"]}" - f"{formatted["keywords"]}" - f"{formatted["summary"]}" - f"") - return content_text diff --git a/api/app/core/memory/read_services/query_preprocessor.py b/api/app/core/memory/read_services/query_preprocessor.py index 02d757c9..123cae40 100644 --- a/api/app/core/memory/read_services/query_preprocessor.py +++ b/api/app/core/memory/read_services/query_preprocessor.py @@ -1,11 +1,13 @@ -# -*- coding: UTF-8 -*- -# Author: Eternity -# @Email: 1533512157@qq.com -# @Time : 2026/4/8 18:11 +import logging import re +from app.core.memory.prompt import prompt_manager +from app.core.memory.utils.llm.llm_utils import StructResponse +from app.core.models import RedBearLLM from app.schemas.memory_agent_schema import AgentMemoryDataset +logger = logging.getLogger(__name__) + class QueryPreprocessor: @staticmethod @@ -16,3 +18,22 @@ class QueryPreprocessor: text = re.sub(rf"{"|".join(AgentMemoryDataset.PRONOUN)}", AgentMemoryDataset.NAME, text) return text + + @staticmethod + async def split(query: str, llm_client: RedBearLLM): + system_prompt = prompt_manager.render( + name="problem_split", + history=[], + sentence=query, + ) + messages = [{"role": "system", "content": system_prompt}] + try: + sub_queries = await llm_client.ainvoke(messages) | StructResponse(mode='json') + except Exception as e: + logger.error(f"[QueryPreprocessor] Sub-question segmentation failed - {e}") + sub_queries = None + return sub_queries or query + + @staticmethod + async def extension(query: str, llm_client: RedBearLLM): + pass diff --git a/api/app/core/memory/read_services/result_builder.py b/api/app/core/memory/read_services/result_builder.py index 10ff8c86..949ff3ed 100644 --- a/api/app/core/memory/read_services/result_builder.py +++ b/api/app/core/memory/read_services/result_builder.py @@ -114,7 +114,7 @@ class PerceptualBuilder(BaseBuilder): f"{self.record.get('domain')}" f"{self.record.get('keywords')}" f"{self.record.get('file_type')}" - "") + "") class CommunityBuilder(BaseBuilder): diff --git a/api/app/core/memory/read_services/retrieval_summary.py b/api/app/core/memory/read_services/retrieval_summary.py new file mode 100644 index 00000000..6b166cf2 --- /dev/null +++ b/api/app/core/memory/read_services/retrieval_summary.py @@ -0,0 +1,11 @@ +from app.core.models import RedBearLLM + + +class RetrievalSummaryProcessor: + @staticmethod + def summary(content: str, llm_client: RedBearLLM): + return + + @staticmethod + def verify(content: str, llm_client: RedBearLLM): + return \ No newline at end of file diff --git a/api/app/core/memory/utils/llm/llm_utils.py b/api/app/core/memory/utils/llm/llm_utils.py index 19d76d68..c4eee82f 100644 --- a/api/app/core/memory/utils/llm/llm_utils.py +++ b/api/app/core/memory/utils/llm/llm_utils.py @@ -1,4 +1,7 @@ -from typing import TYPE_CHECKING +from typing import TYPE_CHECKING, Literal, Type + +from json_repair import json_repair +from langchain_core.messages import AIMessage from app.core.memory.llm_tools.openai_client import OpenAIClient from app.core.models.base import RedBearModelConfig @@ -13,6 +16,27 @@ async def handle_response(response: type[BaseModel]) -> dict: return response.model_dump() +class StructResponse: + def __init__(self, mode: Literal["json", "pydantic"], model: Type[BaseModel] = None): + self.mode = mode + if mode == "pydantic" and model is None: + raise ValueError("Pydantic model is required") + + self.model = model + + def __ror__(self, other: AIMessage): + if not isinstance(other, AIMessage): + raise RuntimeError(f"Unsupported struct type {type(other)}") + text = '' + for block in other.content_blocks: + if block.get("type") == "text": + text += block.get("text", "") + fixed_json = json_repair.repair_json(text, return_objects=True) + if self.mode == "json": + return fixed_json + return self.model.model_validate(fixed_json) + + class MemoryClientFactory: """ Factory for creating LLM, embedder, and reranker clients. @@ -24,21 +48,21 @@ class MemoryClientFactory: >>> llm_client = factory.get_llm_client(model_id) >>> embedder_client = factory.get_embedder_client(embedding_id) """ - + def __init__(self, db: Session): from app.services.memory_config_service import MemoryConfigService self._config_service = MemoryConfigService(db) - + def get_llm_client(self, llm_id: str) -> OpenAIClient: """Get LLM client by model ID.""" if not llm_id: raise ValueError("LLM ID is required") - + try: model_config = self._config_service.get_model_config(llm_id) except Exception as e: raise ValueError(f"Invalid LLM ID '{llm_id}': {str(e)}") from e - + try: return OpenAIClient( RedBearModelConfig( @@ -52,19 +76,19 @@ class MemoryClientFactory: except Exception as e: model_name = model_config.get('model_name', 'unknown') raise ValueError(f"Failed to initialize LLM client for model '{model_name}': {str(e)}") from e - + def get_embedder_client(self, embedding_id: str): """Get embedder client by model ID.""" from app.core.memory.llm_tools.openai_embedder import OpenAIEmbedderClient - + if not embedding_id: raise ValueError("Embedding ID is required") - + try: embedder_config = self._config_service.get_embedder_config(embedding_id) except Exception as e: raise ValueError(f"Invalid embedding ID '{embedding_id}': {str(e)}") from e - + try: return OpenAIEmbedderClient( RedBearModelConfig( @@ -77,17 +101,17 @@ class MemoryClientFactory: except Exception as e: model_name = embedder_config.get('model_name', 'unknown') raise ValueError(f"Failed to initialize embedder client for model '{model_name}': {str(e)}") from e - + def get_reranker_client(self, rerank_id: str) -> OpenAIClient: """Get reranker client by model ID.""" if not rerank_id: raise ValueError("Rerank ID is required") - + try: model_config = self._config_service.get_model_config(rerank_id) except Exception as e: raise ValueError(f"Invalid rerank ID '{rerank_id}': {str(e)}") from e - + try: return OpenAIClient( RedBearModelConfig( diff --git a/api/app/repositories/neo4j/graph_search.py b/api/app/repositories/neo4j/graph_search.py index 336f4134..354c0e23 100644 --- a/api/app/repositories/neo4j/graph_search.py +++ b/api/app/repositories/neo4j/graph_search.py @@ -8,6 +8,7 @@ import numpy as np from app.core.memory.enums import Neo4jNodeType from app.core.memory.llm_tools import OpenAIEmbedderClient from app.core.memory.utils.data.text_utils import escape_lucene_query +from app.core.models import RedBearEmbeddings from app.repositories.neo4j.cypher_queries import ( EXPAND_COMMUNITY_STATEMENTS, SEARCH_CHUNK_BY_CHUNK_ID, @@ -358,7 +359,7 @@ async def search_by_embedding( USER_ID_QUERY_CYPHER_MAPPING[node_type], end_user_id=end_user_id, ) - records = [record for record in records if record if record["embedding"] is not None] + records = [record for record in records if record and record.get("embedding") is not None] ids = [item['id'] for item in records] vectors = [item['embedding'] for item in records] sim_res = cosine_similarity_search(query_embedding, vectors, limit=limit) @@ -469,7 +470,7 @@ async def search_graph( async def search_graph_by_embedding( connector: Neo4jConnector, - embedder_client, + embedder_client: RedBearEmbeddings | OpenAIEmbedderClient, query_text: str, end_user_id: str, limit: int = 50, @@ -495,7 +496,10 @@ async def search_graph_by_embedding( Neo4jNodeType.PERCEPTUAL ] - embeddings = await embedder_client.response([query_text]) + if isinstance(embedder_client, RedBearEmbeddings): + embeddings = embedder_client.embed_documents([query_text]) + else: + embeddings = await embedder_client.response([query_text]) if not embeddings or not embeddings[0]: logger.warning(f"search_graph_by_embedding: embedding generation failed for '{query_text[:50]}'") return {search_key: [] for search_key in include} diff --git a/api/app/services/prompt/prompt_optimizer_system.jinja2 b/api/app/services/prompt/prompt_optimizer_system.jinja2 index 39a4ba68..5611ae94 100644 --- a/api/app/services/prompt/prompt_optimizer_system.jinja2 +++ b/api/app/services/prompt/prompt_optimizer_system.jinja2 @@ -34,7 +34,7 @@ Readability Guideline: Ensure optimized prompts have good readability and logica Constraint Handling Guideline: Do not mention variable-related limitations under the [Constraints] label.{% endraw %}{% endif %} Constraints -Output Constraint: Must output in JSON format including the fields "prompt" and "desc". +Output Constraint: Must output in JSON format including the string fields "prompt" and "desc". Content Constraint: Must not include any explanations, analyses, or additional comments. Language Constraint: Must use clear and concise language. {% if skill != true %}Completeness Constraint: Must fully define all missing elements (input details, output format, constraints, etc.).{% endif %} From 8c6b65db12523ce47cebca47d34b5322e0087d5b Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Thu, 16 Apr 2026 16:27:55 +0800 Subject: [PATCH 075/245] feat(llm): add json_output support for structured LLM responses --- api/app/controllers/chunk_controller.py | 4 +- api/app/controllers/ontology_controller.py | 2 +- api/app/core/agent/langchain_agent.py | 27 ++-- api/app/core/models/base.py | 125 ++++++++++++------ .../core/models/scripts/bedrock_models.yaml | 21 ++- .../core/models/scripts/dashscope_models.yaml | 70 +++++++--- .../core/models/scripts/openai_models.yaml | 33 ++++- .../core/models/scripts/volcano_models.yaml | 16 ++- api/app/core/workflow/nodes/llm/config.py | 5 + api/app/core/workflow/nodes/llm/node.py | 21 ++- api/app/schemas/app_schema.py | 1 + api/app/services/app_chat_service.py | 2 + api/app/services/conversation_service.py | 2 +- api/app/services/draft_run_service.py | 2 + api/app/services/llm_router.py | 8 +- api/app/services/master_agent_router.py | 2 +- api/app/services/memory_perceptual_service.py | 2 +- api/app/services/model_parameter_merger.py | 3 +- api/app/services/model_service.py | 8 +- api/app/services/multi_agent_orchestrator.py | 18 ++- api/app/services/prompt_optimizer_service.py | 2 +- api/app/services/shared_chat_service.py | 4 +- api/app/version_info.json | 36 +++++ 23 files changed, 304 insertions(+), 110 deletions(-) diff --git a/api/app/controllers/chunk_controller.py b/api/app/controllers/chunk_controller.py index b5c0a5ae..cc1f8c98 100644 --- a/api/app/controllers/chunk_controller.py +++ b/api/app/controllers/chunk_controller.py @@ -443,10 +443,10 @@ async def retrieve_chunks( match retrieve_data.retrieve_type: case chunk_schema.RetrieveType.PARTICIPLE: rs = vector_service.search_by_full_text(query=retrieve_data.query, top_k=retrieve_data.top_k, indices=indices, score_threshold=retrieve_data.similarity_threshold, file_names_filter=retrieve_data.file_names_filter) - return success(data=rs, msg="retrieval successful") + return success(data=jsonable_encoder(rs), msg="retrieval successful") case chunk_schema.RetrieveType.SEMANTIC: rs = vector_service.search_by_vector(query=retrieve_data.query, top_k=retrieve_data.top_k, indices=indices, score_threshold=retrieve_data.vector_similarity_weight, file_names_filter=retrieve_data.file_names_filter) - return success(data=rs, msg="retrieval successful") + return success(data=jsonable_encoder(rs), msg="retrieval successful") case _: rs1 = vector_service.search_by_vector(query=retrieve_data.query, top_k=retrieve_data.top_k, indices=indices, score_threshold=retrieve_data.vector_similarity_weight, file_names_filter=retrieve_data.file_names_filter) rs2 = vector_service.search_by_full_text(query=retrieve_data.query, top_k=retrieve_data.top_k, indices=indices, score_threshold=retrieve_data.similarity_threshold, file_names_filter=retrieve_data.file_names_filter) diff --git a/api/app/controllers/ontology_controller.py b/api/app/controllers/ontology_controller.py index 83f75888..602ee709 100644 --- a/api/app/controllers/ontology_controller.py +++ b/api/app/controllers/ontology_controller.py @@ -165,7 +165,7 @@ def _get_ontology_service( api_key=api_key_config.api_key, base_url=api_key_config.api_base, is_omni=api_key_config.is_omni, - support_thinking="thinking" in (api_key_config.capability or []), + capability=api_key_config.capability, max_retries=3, timeout=60.0 ) diff --git a/api/app/core/agent/langchain_agent.py b/api/app/core/agent/langchain_agent.py index ca7172e8..927eb734 100644 --- a/api/app/core/agent/langchain_agent.py +++ b/api/app/core/agent/langchain_agent.py @@ -41,6 +41,7 @@ class LangChainAgent: max_tool_consecutive_calls: int = 3, # 单个工具最大连续调用次数 deep_thinking: bool = False, # 是否启用深度思考模式 thinking_budget_tokens: Optional[int] = None, # 深度思考 token 预算 + json_output: bool = False, # 是否强制 JSON 输出 capability: Optional[List[str]] = None # 模型能力列表，用于校验是否支持深度思考 ): """初始化 LangChain Agent @@ -64,7 +65,6 @@ class LangChainAgent: self.streaming = streaming self.is_omni = is_omni self.max_tool_consecutive_calls = max_tool_consecutive_calls - self.deep_thinking = deep_thinking and ("thinking" in (capability or [])) # 工具调用计数器：记录每个工具的连续调用次数 self.tool_call_counter: Dict[str, int] = {} @@ -80,6 +80,12 @@ class LangChainAgent: self.system_prompt = system_prompt or "你是一个专业的AI助手" + # ChatTongyi 要求 messages 含 'json' 字样才能使用 response_format + # 在 system prompt 中注入 JSON 要求 + from app.models.models_model import ModelProvider + if json_output and provider.lower() == ModelProvider.DASHSCOPE and not is_omni: + self.system_prompt += "\n请以JSON格式输出。" + logger.debug( f"Agent 迭代次数配置: max_iterations={self.max_iterations}, " f"tool_count={len(self.tools)}, " @@ -87,23 +93,17 @@ class LangChainAgent: f"auto_calculated={max_iterations is None}" ) - # 根据 capability 校验是否真正支持深度思考 - actual_deep_thinking = self.deep_thinking - if deep_thinking and not actual_deep_thinking: - logger.warning( - f"模型 {model_name} 不支持深度思考（capability 中无 'thinking'），已自动关闭 deep_thinking" - ) - - # 创建 RedBearLLM（支持多提供商） + # 创建 RedBearLLM，capability 校验由 RedBearModelConfig 统一处理 model_config = RedBearModelConfig( model_name=model_name, provider=provider, api_key=api_key, base_url=api_base, is_omni=is_omni, - deep_thinking=actual_deep_thinking, - thinking_budget_tokens=thinking_budget_tokens if actual_deep_thinking else None, - support_thinking="thinking" in (capability or []), + capability=capability, + deep_thinking=deep_thinking, + thinking_budget_tokens=thinking_budget_tokens, + json_output=json_output, extra_params={ "temperature": temperature, "max_tokens": max_tokens, @@ -112,6 +112,9 @@ class LangChainAgent: ) self.llm = RedBearLLM(model_config, type=ModelType.CHAT) + # 从经过校验的 config 读取实际生效的能力开关 + self.deep_thinking = model_config.deep_thinking + self.json_output = model_config.json_output # 获取底层模型用于真正的流式调用 self._underlying_llm = self.llm._model if hasattr(self.llm, '_model') else self.llm diff --git a/api/app/core/models/base.py b/api/app/core/models/base.py index 89a7dcee..90046412 100644 --- a/api/app/core/models/base.py +++ b/api/app/core/models/base.py @@ -1,7 +1,7 @@ from __future__ import annotations import os -from typing import Any, Dict, Optional, TypeVar +from typing import Any, Dict, List, Optional, TypeVar from langchain_aws import ChatBedrock from langchain_community.chat_models import ChatTongyi @@ -9,7 +9,7 @@ from langchain_core.embeddings import Embeddings from langchain_core.language_models import BaseLLM from langchain_ollama import OllamaLLM from langchain_openai import ChatOpenAI, OpenAI -from pydantic import BaseModel, Field +from pydantic import BaseModel, Field, model_validator from app.core.error_codes import BizCode from app.core.exceptions import BusinessException @@ -25,10 +25,11 @@ class RedBearModelConfig(BaseModel): provider: str api_key: str base_url: Optional[str] = None + capability: List[str] = Field(default_factory=list) # 模型能力列表，驱动所有能力开关 is_omni: bool = False # 是否为 Omni 模型 deep_thinking: bool = False # 是否启用深度思考模式 thinking_budget_tokens: Optional[int] = None # 深度思考 token 预算 - support_thinking: bool = False # 模型是否支持 enable_thinking 参数（capability 含 thinking） + json_output: bool = False # 是否强制 JSON 输出 # 请求超时时间（秒）- 默认120秒以支持复杂的LLM调用，可通过环境变量 LLM_TIMEOUT 配置 timeout: float = Field(default_factory=lambda: float(os.getenv("LLM_TIMEOUT", "120.0"))) # 最大重试次数 - 默认2次以避免过长等待，可通过环境变量 LLM_MAX_RETRIES 配置 @@ -36,6 +37,29 @@ class RedBearModelConfig(BaseModel): concurrency: int = 5 # 并发限流 extra_params: Dict[str, Any] = {} + @model_validator(mode="after") + def _resolve_capabilities(self) -> "RedBearModelConfig": + from app.core.logging_config import get_business_logger + logger = get_business_logger() + if self.deep_thinking and "thinking" not in self.capability: + logger.warning( + f"模型 {self.model_name} 不支持深度思考（capability 中无 'thinking'），已自动关闭 deep_thinking" + ) + self.deep_thinking = False + self.thinking_budget_tokens = None + if self.json_output and "json_output" not in self.capability: + logger.warning( + f"模型 {self.model_name} 不支持 JSON 输出（capability 中无 'json_output'），已自动关闭 json_output" + ) + self.json_output = False + if self.json_output and self.deep_thinking: + logger.warning( + f"模型 {self.model_name} json_output 与 deep_thinking 互斥，已自动关闭 deep_thinking" + ) + self.deep_thinking = False + self.thinking_budget_tokens = None + return self + class RedBearModelFactory: """模型工厂类""" @@ -74,17 +98,20 @@ class RedBearModelFactory: is_streaming = bool(config.extra_params.get("streaming")) if is_streaming: params["stream_usage"] = True - # 只有支持 thinking 的模型才传 enable_thinking - if config.support_thinking: - model_kwargs: Dict[str, Any] = config.extra_params.get("model_kwargs", {}) - if is_streaming: - model_kwargs["enable_thinking"] = config.deep_thinking - if config.deep_thinking: - model_kwargs["incremental_output"] = True - if config.thinking_budget_tokens: - model_kwargs["thinking_budget"] = config.thinking_budget_tokens - else: - model_kwargs["enable_thinking"] = False + # 支持 thinking 的模型始终传 enable_thinking，关闭时显式传 False 避免模型默认开启思考 + if "thinking" in config.capability: + extra_body = params.setdefault("extra_body", {}) + if config.deep_thinking: + extra_body["enable_thinking"] = False + if is_streaming: + extra_body["enable_thinking"] = True + if config.thinking_budget_tokens: + extra_body["thinking_budget"] = config.thinking_budget_tokens + params["extra_body"] = extra_body + # JSON 输出模式 + if config.json_output: + model_kwargs = params.setdefault("model_kwargs", {}) + model_kwargs["response_format"] = {"type": "json_object"} params["model_kwargs"] = model_kwargs return params @@ -108,27 +135,30 @@ class RedBearModelFactory: **config.extra_params } # 流式模式下启用 stream_usage 以获取 token 统计 - if config.extra_params.get("streaming"): - params["stream_usage"] = True - # 深度思考模式 is_streaming = bool(config.extra_params.get("streaming")) - if config.support_thinking: - if is_streaming and not config.is_omni: - if provider == ModelProvider.VOLCANO: - # 火山引擎深度思考仅流式调用支持，非流式时不传 thinking 参数 - thinking_config: Dict[str, Any] = { - "type": "enabled" if config.deep_thinking else "disabled" - } - if config.deep_thinking and config.thinking_budget_tokens: - thinking_config["budget_tokens"] = config.thinking_budget_tokens - params["extra_body"] = {"thinking": thinking_config} - else: - # 始终显式传递 enable_thinking，不支持该参数的模型（如 DeepSeek-R1）会直接忽略 - model_kwargs: Dict[str, Any] = config.extra_params.get("model_kwargs", {}) - model_kwargs["enable_thinking"] = config.deep_thinking - if config.deep_thinking and config.thinking_budget_tokens: - model_kwargs["thinking_budget"] = config.thinking_budget_tokens - params["model_kwargs"] = model_kwargs + if is_streaming: + params["stream_usage"] = True + # 支持 thinking 的模型始终传 enable_thinking，关闭时显式传 False 避免模型默认开启思考 + if "thinking" in config.capability: + # VOLCANO 深度思考仅流式支持 + if provider == ModelProvider.VOLCANO: + thinking_config: Dict[str, Any] = {"type": "enabled" if config.deep_thinking else "disabled"} + if config.deep_thinking and config.thinking_budget_tokens: + thinking_config["budget_tokens"] = config.thinking_budget_tokens + params["extra_body"] = {"thinking": thinking_config} + else: + extra_body = params.setdefault("extra_body", {}) + if config.deep_thinking: + extra_body["enable_thinking"] = False + if is_streaming: + extra_body["enable_thinking"] = True + if config.thinking_budget_tokens: + extra_body["thinking_budget"] = config.thinking_budget_tokens + params["extra_body"] = extra_body + # JSON 输出模式 + if config.json_output: + params.setdefault("model_kwargs", {}) + params["model_kwargs"]["response_format"] = {"type": "json_object"} return params elif provider == ModelProvider.DASHSCOPE: params = { @@ -137,18 +167,21 @@ class RedBearModelFactory: "max_retries": config.max_retries, **config.extra_params } - # 只有支持 thinking 的模型才传 enable_thinking - if config.support_thinking: + # 支持 thinking 的模型始终传 enable_thinking，关闭时显式传 False 避免模型默认开启思考 + if "thinking" in config.capability: is_streaming = bool(config.extra_params.get("streaming")) - model_kwargs: Dict[str, Any] = config.extra_params.get("model_kwargs", {}) - if is_streaming: - model_kwargs["enable_thinking"] = config.deep_thinking - if config.deep_thinking: - model_kwargs["incremental_output"] = True - if config.thinking_budget_tokens: - model_kwargs["thinking_budget"] = config.thinking_budget_tokens - else: + model_kwargs = params.setdefault("model_kwargs", {}) + if config.deep_thinking: model_kwargs["enable_thinking"] = False + if is_streaming: + model_kwargs["enable_thinking"] = True + model_kwargs["incremental_output"] = True + if config.thinking_budget_tokens: + model_kwargs["thinking_budget"] = config.thinking_budget_tokens + params["model_kwargs"] = model_kwargs + if config.json_output: + model_kwargs = params.setdefault("model_kwargs", {}) + model_kwargs["response_format"] = {"type": "json_object"} params["model_kwargs"] = model_kwargs return params elif provider == ModelProvider.BEDROCK: @@ -196,6 +229,10 @@ class RedBearModelFactory: params["additional_model_request_fields"] = { "thinking": {"type": "enabled", "budget_tokens": budget} } + # JSON 输出模式 + if config.json_output: + params.setdefault("model_kwargs", {}) + params["model_kwargs"]["response_format"] = {"type": "json_object"} return params else: raise BusinessException(f"不支持的提供商: {provider}", code=BizCode.PROVIDER_NOT_SUPPORTED) diff --git a/api/app/core/models/scripts/bedrock_models.yaml b/api/app/core/models/scripts/bedrock_models.yaml index 5b3a2f64..f96dba15 100644 --- a/api/app/core/models/scripts/bedrock_models.yaml +++ b/api/app/core/models/scripts/bedrock_models.yaml @@ -6,7 +6,8 @@ models: description: AI21 Labs大语言模型，completion生成模式，256000上下文窗口 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -20,6 +21,7 @@ models: is_official: true capability: - vision + - json_output is_omni: false tags: - 大语言模型 @@ -38,6 +40,7 @@ models: capability: - vision - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -54,7 +57,8 @@ models: description: Cohere大语言模型，支持智能体思考、工具调用、流式工具调用，128000上下文窗口，对话模式 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -72,6 +76,7 @@ models: capability: - vision - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -87,7 +92,8 @@ models: description: Meta Llama大语言模型，支持智能体思考、工具调用，128000上下文窗口，对话模式 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -101,7 +107,8 @@ models: description: Mistral AI大语言模型，支持智能体思考、工具调用，32000上下文窗口，对话模式 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -115,7 +122,8 @@ models: description: OpenAI大语言模型，支持智能体思考、工具调用、流式工具调用，32768上下文窗口，对话模式 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -130,7 +138,8 @@ models: description: Qwen大语言模型，支持智能体思考、工具调用、流式工具调用，32768上下文窗口，对话模式 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 diff --git a/api/app/core/models/scripts/dashscope_models.yaml b/api/app/core/models/scripts/dashscope_models.yaml index d9e6a00f..26c3cc2c 100644 --- a/api/app/core/models/scripts/dashscope_models.yaml +++ b/api/app/core/models/scripts/dashscope_models.yaml @@ -8,6 +8,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -22,6 +23,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -36,6 +38,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -48,7 +51,8 @@ models: description: DeepSeek-V3.1大语言模型，支持智能体思考，131072超大上下文窗口，对话模式，支持丰富生成参数调节 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -61,7 +65,8 @@ models: description: DeepSeek-V3.2-exp实验版大语言模型，支持智能体思考，131072超大上下文窗口，对话模式，支持丰富生成参数调节 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -74,7 +79,8 @@ models: description: DeepSeek-V3.2大语言模型，支持智能体思考，131072超大上下文窗口，对话模式，支持丰富生成参数调节 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -87,7 +93,8 @@ models: description: DeepSeek-V3大语言模型，支持智能体思考，64000上下文窗口，对话模式，支持文本与JSON格式输出 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -100,7 +107,8 @@ models: description: farui-plus大语言模型，支持多工具调用、智能体思考、流式工具调用，12288上下文窗口，对话模式 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -115,7 +123,8 @@ models: description: GLM-4.7大语言模型，支持多工具调用、智能体思考、流式工具调用，202752超大上下文窗口，对话模式 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -133,6 +142,7 @@ models: capability: - vision - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -150,6 +160,7 @@ models: capability: - vision - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -180,6 +191,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -210,7 +222,7 @@ models: is_deprecated: false is_official: true capability: - - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -376,6 +388,7 @@ models: capability: - vision - video + - json_output is_omni: false tags: - 大语言模型 @@ -448,6 +461,7 @@ models: capability: - vision - video + - qwen-vl-plus-latest is_omni: false tags: - 大语言模型 @@ -466,6 +480,7 @@ models: capability: - vision - video + - qwen-vl-plus-latest is_omni: false tags: - 大语言模型 @@ -481,7 +496,8 @@ models: description: qwen2.5-0.5b-instruct大语言模型，支持多工具调用、智能体思考、流式工具调用，32768上下文窗口，对话模式，未废弃 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -498,6 +514,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -513,7 +530,7 @@ models: is_deprecated: false is_official: true capability: - - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -530,6 +547,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -546,6 +564,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -561,7 +580,7 @@ models: is_deprecated: false is_official: true capability: - - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -578,6 +597,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -594,6 +614,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -610,6 +631,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -626,6 +648,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -641,7 +664,7 @@ models: is_deprecated: false is_official: true capability: - - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -656,7 +679,7 @@ models: is_deprecated: false is_official: true capability: - - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -672,6 +695,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -687,6 +711,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -702,6 +727,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -719,6 +745,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -736,6 +763,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -752,6 +780,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -768,7 +797,7 @@ models: is_deprecated: false is_official: true capability: - - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -785,6 +814,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -803,6 +833,8 @@ models: - vision - video - audio + - thinking + - json_output is_omni: true tags: - 大语言模型 @@ -822,7 +854,7 @@ models: capability: - vision - video - - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -844,6 +876,7 @@ models: - vision - video - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -864,7 +897,7 @@ models: capability: - vision - video - - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -886,6 +919,7 @@ models: - vision - video - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -907,6 +941,7 @@ models: - vision - video - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -928,6 +963,7 @@ models: - vision - video - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -947,6 +983,7 @@ models: - vision - video - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -964,6 +1001,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -979,6 +1017,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -994,6 +1033,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 diff --git a/api/app/core/models/scripts/openai_models.yaml b/api/app/core/models/scripts/openai_models.yaml index 08b81008..1c0a0b2d 100644 --- a/api/app/core/models/scripts/openai_models.yaml +++ b/api/app/core/models/scripts/openai_models.yaml @@ -10,6 +10,7 @@ models: - vision - audio - video + - json_output is_omni: true tags: - 大语言模型 @@ -27,7 +28,8 @@ models: description: gpt-3.5-turbo-0125大语言模型，支持多工具调用、智能体思考、流式工具调用，16385上下文窗口，对话模式 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -42,7 +44,8 @@ models: description: gpt-3.5-turbo-1106大语言模型，支持多工具调用、智能体思考、流式工具调用，16385上下文窗口，对话模式 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -57,7 +60,8 @@ models: description: gpt-3.5-turbo-16k大语言模型，支持多工具调用、智能体思考、流式工具调用，16385上下文窗口，对话模式 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -84,7 +88,8 @@ models: description: gpt-3.5-turbo大语言模型，支持多工具调用、智能体思考、流式工具调用，16385上下文窗口，对话模式 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -99,7 +104,8 @@ models: description: gpt-4-0125-preview大语言模型，支持多工具调用、智能体思考、流式工具调用，128000上下文窗口，对话模式 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -114,7 +120,8 @@ models: description: gpt-4-1106-preview大语言模型，支持多工具调用、智能体思考、流式工具调用，128000上下文窗口，对话模式 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -131,6 +138,7 @@ models: is_official: true capability: - vision + - json_output is_omni: false tags: - 大语言模型 @@ -146,7 +154,8 @@ models: description: gpt-4-turbo-preview大语言模型，支持多工具调用、智能体思考、流式工具调用，128000上下文窗口，对话模式 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -163,6 +172,7 @@ models: is_official: true capability: - vision + - json_output is_omni: false tags: - 大语言模型 @@ -194,6 +204,7 @@ models: capability: - vision - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -213,6 +224,7 @@ models: capability: - vision - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -231,6 +243,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -248,6 +261,7 @@ models: is_official: true capability: - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -266,6 +280,7 @@ models: capability: - vision - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -284,6 +299,7 @@ models: capability: - vision - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -302,6 +318,7 @@ models: capability: - vision - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -321,6 +338,7 @@ models: capability: - vision - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -340,6 +358,7 @@ models: capability: - vision - thinking + - json_output is_omni: false tags: - 大语言模型 diff --git a/api/app/core/models/scripts/volcano_models.yaml b/api/app/core/models/scripts/volcano_models.yaml index c86d41ac..6658c2f9 100644 --- a/api/app/core/models/scripts/volcano_models.yaml +++ b/api/app/core/models/scripts/volcano_models.yaml @@ -11,6 +11,7 @@ models: - vision - video - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -26,6 +27,7 @@ models: - vision - video - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -41,6 +43,7 @@ models: - vision - video - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -56,6 +59,7 @@ models: - vision - video - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -72,6 +76,7 @@ models: capability: - vision - video + - json_output is_omni: false tags: - 大语言模型 @@ -87,6 +92,7 @@ models: - vision - video - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -102,6 +108,7 @@ models: - vision - video - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -117,6 +124,7 @@ models: - vision - video - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -132,6 +140,7 @@ models: - vision - video - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -148,6 +157,7 @@ models: - vision - video - thinking + - json_output is_omni: false tags: - 大语言模型 @@ -175,7 +185,8 @@ models: description: 全新一代主力模型，性能全面升级，在知识、代码、推理等方面表现卓越。最大支持 128k 上下文窗口，输出长度支持最大 12k tokens。 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 @@ -187,7 +198,8 @@ models: description: 全新一代轻量版模型，极致响应速度，效果与时延均达到全球一流水平。支持 32k 上下文窗口，输出长度支持最大 12k tokens。 is_deprecated: false is_official: true - capability: [] + capability: + - json_output is_omni: false tags: - 大语言模型 diff --git a/api/app/core/workflow/nodes/llm/config.py b/api/app/core/workflow/nodes/llm/config.py index 771262c1..b815c80f 100644 --- a/api/app/core/workflow/nodes/llm/config.py +++ b/api/app/core/workflow/nodes/llm/config.py @@ -116,6 +116,11 @@ class LLMNodeConfig(BaseNodeConfig): description="Top-p 采样参数" ) + json_output: bool = Field( + default=False, + description="是否以 JSON 格式输出" + ) + frequency_penalty: float | None = Field( default=None, ge=-2.0, diff --git a/api/app/core/workflow/nodes/llm/node.py b/api/app/core/workflow/nodes/llm/node.py index bb87c845..664a28fa 100644 --- a/api/app/core/workflow/nodes/llm/node.py +++ b/api/app/core/workflow/nodes/llm/node.py @@ -22,6 +22,7 @@ from app.db import get_db_context from app.models import ModelType from app.schemas.model_schema import ModelInfo from app.services.model_service import ModelConfigService +from app.models.models_model import ModelProvider logger = logging.getLogger(__name__) @@ -126,7 +127,11 @@ class LLMNode(BaseNode): # 4. 创建 LLM 实例（使用已提取的数据） # 注意：对于流式输出，需要在模型初始化时设置 streaming=True - extra_params = {"streaming": stream} if stream else {} + extra_params: dict[str, Any] = {"streaming": stream} if stream else {} + if self.typed_config.temperature is not None: + extra_params["temperature"] = self.typed_config.temperature + if self.typed_config.max_tokens is not None: + extra_params["max_tokens"] = self.typed_config.max_tokens llm = RedBearLLM( RedBearModelConfig( @@ -135,7 +140,9 @@ class LLMNode(BaseNode): api_key=model_info.api_key, base_url=model_info.api_base, extra_params=extra_params, - is_omni=model_info.is_omni + is_omni=model_info.is_omni, + capability=model_info.capability, + json_output=self.typed_config.json_output, ), type=model_info.model_type ) @@ -218,6 +225,16 @@ class LLMNode(BaseNode): rendered = self._render_template(prompt_template, variable_pool) self.messages = [{"role": "user", "content": rendered}] + # ChatTongyi 要求 messages 含 'json' 字样才能使用 response_format，在 system prompt 中注入 + if (self.typed_config.json_output + and model_info.provider.lower() == ModelProvider.DASHSCOPE + and not model_info.is_omni): + system_msg = next((m for m in self.messages if m["role"] == "system"), None) + if system_msg: + system_msg["content"] += "\n请以JSON格式输出。" + else: + self.messages.insert(0, {"role": "system", "content": "请以JSON格式输出。"}) + return llm async def execute(self, state: WorkflowState, variable_pool: VariablePool) -> AIMessage: diff --git a/api/app/schemas/app_schema.py b/api/app/schemas/app_schema.py index 1ed98f68..e93c513d 100644 --- a/api/app/schemas/app_schema.py +++ b/api/app/schemas/app_schema.py @@ -245,6 +245,7 @@ class ModelParameters(BaseModel): stop: Optional[List[str]] = Field(default=None, description="停止序列") deep_thinking: bool = Field(default=False, description="是否启用深度思考模式（需模型支持，如 DeepSeek-R1、QwQ 等）") thinking_budget_tokens: Optional[int] = Field(default=None, ge=1024, le=131072, description="深度思考 token 预算（仅部分模型支持）") + json_output: bool = Field(default=False, description="是否强制 JSON 格式输出（需模型支持 json_output 能力）") class VariableDefinition(BaseModel): diff --git a/api/app/services/app_chat_service.py b/api/app/services/app_chat_service.py index 2d10ed44..56e25713 100644 --- a/api/app/services/app_chat_service.py +++ b/api/app/services/app_chat_service.py @@ -120,6 +120,7 @@ class AppChatService: tools=tools, deep_thinking=model_parameters.get("deep_thinking", False), thinking_budget_tokens=model_parameters.get("thinking_budget_tokens"), + json_output=model_parameters.get("json_output", False), capability=api_key_obj.capability or [], ) @@ -392,6 +393,7 @@ class AppChatService: streaming=True, deep_thinking=model_parameters.get("deep_thinking", False), thinking_budget_tokens=model_parameters.get("thinking_budget_tokens"), + json_output=model_parameters.get("json_output", False), capability=api_key_obj.capability or [], ) diff --git a/api/app/services/conversation_service.py b/api/app/services/conversation_service.py index 6e9f3544..61744ec7 100644 --- a/api/app/services/conversation_service.py +++ b/api/app/services/conversation_service.py @@ -544,7 +544,7 @@ class ConversationService: api_key=api_key, base_url=api_base, is_omni=is_omni, - support_thinking="thinking" in (capability or []), + capability=capability, ), type=ModelType(model_type) ) diff --git a/api/app/services/draft_run_service.py b/api/app/services/draft_run_service.py index b47bd4cd..81457a08 100644 --- a/api/app/services/draft_run_service.py +++ b/api/app/services/draft_run_service.py @@ -597,6 +597,7 @@ class AgentRunService: tools=tools, deep_thinking=effective_params.get("deep_thinking", False), thinking_budget_tokens=effective_params.get("thinking_budget_tokens"), + json_output=effective_params.get("json_output", False), capability=api_key_config.get("capability", []), ) @@ -853,6 +854,7 @@ class AgentRunService: streaming=True, deep_thinking=effective_params.get("deep_thinking", False), thinking_budget_tokens=effective_params.get("thinking_budget_tokens"), + json_output=effective_params.get("json_output", False), capability=api_key_config.get("capability", []), ) diff --git a/api/app/services/llm_router.py b/api/app/services/llm_router.py index 7087415e..bd90eee9 100644 --- a/api/app/services/llm_router.py +++ b/api/app/services/llm_router.py @@ -415,9 +415,11 @@ class LLMRouter: api_key=api_key_config.api_key, base_url=api_key_config.api_base, is_omni=api_key_config.is_omni, - support_thinking="thinking" in (api_key_config.capability or []), - temperature=0.3, - max_tokens=500 + capability=api_key_config.capability, + extra_params={ + "temperature": 0.3, + "max_tokens": 500 + } ) logger.debug(f"创建 LLM 实例 - Provider: {api_key_config.provider}, Model: {api_key_config.model_name}") diff --git a/api/app/services/master_agent_router.py b/api/app/services/master_agent_router.py index 206443bd..dfb3c2da 100644 --- a/api/app/services/master_agent_router.py +++ b/api/app/services/master_agent_router.py @@ -393,7 +393,7 @@ class MasterAgentRouter: api_key=api_key_config.api_key, base_url=api_key_config.api_base, is_omni=api_key_config.is_omni, - support_thinking="thinking" in (api_key_config.capability or []), + capability=api_key_config.capability, extra_params = extra_params ) diff --git a/api/app/services/memory_perceptual_service.py b/api/app/services/memory_perceptual_service.py index 7d6d1092..8fa9c9bf 100644 --- a/api/app/services/memory_perceptual_service.py +++ b/api/app/services/memory_perceptual_service.py @@ -233,7 +233,7 @@ class MemoryPerceptualService: api_key=model_config.api_key, base_url=model_config.api_base, is_omni=model_config.is_omni, - support_thinking="thinking" in (model_config.capability or []), + capability=model_config.capability, ) ) return llm, model_config diff --git a/api/app/services/model_parameter_merger.py b/api/app/services/model_parameter_merger.py index 4be83851..6911a9d5 100644 --- a/api/app/services/model_parameter_merger.py +++ b/api/app/services/model_parameter_merger.py @@ -47,7 +47,8 @@ class ModelParameterMerger: "n": 1, "stop": None, "deep_thinking": False, - "thinking_budget_tokens": None + "thinking_budget_tokens": None, + "json_output": False } # 合并参数：默认值 -> 模型配置 -> Agent 配置 diff --git a/api/app/services/model_service.py b/api/app/services/model_service.py index d202b83a..8807020b 100644 --- a/api/app/services/model_service.py +++ b/api/app/services/model_service.py @@ -125,9 +125,11 @@ class ModelConfigService: api_key=api_key, base_url=api_base, is_omni=is_omni, - support_thinking="thinking" in (capability or []), - temperature=0.7, - max_tokens=100 + capability=capability, + extra_params={ + "temperature": 0.7, + "max_tokens": 100 + } ) # 根据模型类型选择不同的验证方式 diff --git a/api/app/services/multi_agent_orchestrator.py b/api/app/services/multi_agent_orchestrator.py index 216aeb6e..d30dc822 100644 --- a/api/app/services/multi_agent_orchestrator.py +++ b/api/app/services/multi_agent_orchestrator.py @@ -2616,9 +2616,11 @@ class MultiAgentOrchestrator: api_key=api_key_config.api_key, base_url=api_key_config.api_base, is_omni=api_key_config.is_omni, - support_thinking="thinking" in (api_key_config.capability or []), - temperature=0.7, # 整合任务使用中等温度 - max_tokens=2000 + capability=api_key_config.capability, + extra_params={ + "temperature": 0.7, # 整合任务使用中等温度 + "max_tokens": 2000 + } ) # 创建 LLM 实例 @@ -2795,10 +2797,12 @@ class MultiAgentOrchestrator: api_key=api_key_config.api_key, base_url=api_key_config.api_base, is_omni=api_key_config.is_omni, - support_thinking="thinking" in (api_key_config.capability or []), - temperature=0.7, - max_tokens=2000, - extra_params={"streaming": True} # 启用流式输出 + capability=api_key_config.capability, + extra_params={ + "temperature": 0.7, + "max_tokens": 2000, + "streaming": True # 启用流式输出 + } ) # 创建 LLM 实例 diff --git a/api/app/services/prompt_optimizer_service.py b/api/app/services/prompt_optimizer_service.py index 30901111..1686a164 100644 --- a/api/app/services/prompt_optimizer_service.py +++ b/api/app/services/prompt_optimizer_service.py @@ -186,7 +186,7 @@ class PromptOptimizerService: api_key=api_config.api_key, base_url=api_config.api_base, is_omni=api_config.is_omni, - support_thinking="thinking" in (api_config.capability or []), + capability=api_config.capability, ), type=ModelType(model_config.type)) try: prompt_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'prompt') diff --git a/api/app/services/shared_chat_service.py b/api/app/services/shared_chat_service.py index b1e40a2d..37956d77 100644 --- a/api/app/services/shared_chat_service.py +++ b/api/app/services/shared_chat_service.py @@ -250,7 +250,8 @@ class SharedChatService: tools=tools, deep_thinking=model_parameters.get("deep_thinking", False), thinking_budget_tokens=model_parameters.get("thinking_budget_tokens"), - capability=api_key_obj.capability or [], + json_output=model_parameters.get("json_output", False), + capability=api_key_obj.capability, ) # 加载历史消息 @@ -455,6 +456,7 @@ class SharedChatService: streaming=True, deep_thinking=model_parameters.get("deep_thinking", False), thinking_budget_tokens=model_parameters.get("thinking_budget_tokens"), + json_output=model_parameters.get("json_output", False), capability=api_key_obj.capability or [], ) diff --git a/api/app/version_info.json b/api/app/version_info.json index d07035e2..a094b64c 100644 --- a/api/app/version_info.json +++ b/api/app/version_info.json @@ -1,4 +1,40 @@ { + "v0.3.0": { + "introduction": { + "codeName": "破晓", + "releaseDate": "2026-4-15", + "upgradePosition": "🐻 全面升级应用工作流、记忆智能与系统稳健性，引入版本化API、多模态记忆感知及大量工作流增强，打造更可靠、精准的 MemoryBear", + "coreUpgrades": [ + "1. 应用与API增强
* 版本化API调用支持：对外服务API支持指定版本调用
* 工作流检查清单：新增结构化验证步骤
* 深度思考参数精准控制：仅向支持深度推理的模型发送思考参数
* 提示器模型返回优化：优化提示器模型响应处理", + "2. 记忆智能 🧠
* 多模态记忆感知Agent：支持多模态记忆读取与写入
* OpenClaw内置工具：新增内置工具扩展Agent工具集", + "3. 用户体验 🎨
* 流式渲染稳定性优化：解决LLM流式输出页面抖动问题
* 记忆中枢更名：「记忆相关」更名为「记忆中枢」", + "4. 工作流改进 ⚙️
* 三级变量模板转换：支持三级变量解析
* VL模型Token统计：修复模型组合中VL模型Token未统计问题
* 导入工作流功能特性同步：正确同步开场白、引用等属性
* 会话变量名称唯一性校验：防止变量名冲突
* 文件类型提取修复：正确提取file.type信息
* 条件分支显示修复：值为0或会话变量时正确渲染
* Object/Array校验规则：防止JSON序列化错误
* HTTP请求Body字段修正：body字段从name改为key", + "5. 知识库 📚
* Embedding Token截断安全边界：统一添加8000 token截断，优化Excel独立chunk处理", + "6. 稳健性与缺陷修复 🔧
* 原子性更新与批量访问失败修复
* 对话别名提取错误修复
* 工作流别名提取修正（区分用户和AI回复）
* RAG记忆分页数据修复
* 隐式记忆详情显示修复
* 向量查询驱动关闭异常修复
* 用户管理启停异常修复
* 模型列表筛选不一致修复", + "
", + "v0.3.0 标志着 MemoryBear 向生产成熟度迈出坚实一步。后续版本将持续深化工作流表达力、记忆检索精度和跨模态理解能力，强化复杂Agent编排支持，稳固大规模生产部署基础。", + "
", + "MemoryBear — 破晓 🐻✨" + ] + }, + "introduction_en": { + "codeName": "PoXiao", + "releaseDate": "2026-4-15", + "upgradePosition": "🐻 Comprehensive upgrades across application workflows, memory intelligence, and system robustness — introducing versioned APIs, multimodal memory perception, and extensive workflow enhancements for a more reliable MemoryBear", + "coreUpgrades": [ + "1. Application & API Enhancements
* Versioned API Support: External APIs now support version-specific calls
* Workflow Checklist: Structured validation steps before deployment
* Deep Thinking Parameter Control: Only send thinking params to supported models
* Prompt Optimizer Return Optimization: Improved prompt optimizer response handling", + "2. Memory Intelligence 🧠
* Multimodal Memory Perception Agent: Read/write multimodal memory
* OpenClaw Built-in Tool: New built-in tool for agent operations", + "3. User Experience 🎨
* Streaming Render Stabilization: Eliminated page jitter during LLM output
* Memory Hub Renaming: Renamed to better reflect central memory role", + "4. Workflow Improvements ⚙️
* Three-Level Variable Template Conversion: Support for three-level variable resolution
* VL Model Token Tracking: Fixed token tracking for VL models in model groups
* Imported Workflow Feature Sync: Properly sync opening messages, citations, etc.
* Session Variable Name Uniqueness: Prevent variable name conflicts
* File Type Extraction Fix: Correctly extract file.type information
* Condition Branch Display Fix: Correct rendering for value 0 or session variables
* Object/Array Validation Rules: Prevent JSON serialization save errors
* HTTP Request Body Key Fix: Body field uses key instead of name", + "5. Knowledge Base 📚
* Embedding Token Truncation Safety: Unified 8000-token boundary, optimized Excel chunk processing", + "6. Robustness & Bug Fixes 🔧
* Atomic update & batch access failure fixes
* Conversation alias extraction fix
* Workflow alias extraction correction (user vs AI distinction)
* RAG memory pagination fix
* Implicit memory detail display fix
* Vector query driver closed exception fix
* User management enable/disable fix
* Model list filter inconsistency fix", + "
", + "v0.3.0 marks a meaningful step toward production maturity for MemoryBear. Upcoming releases will deepen workflow expressiveness, memory retrieval precision, and cross-modal understanding while strengthening complex agent orchestration and large-scale deployment foundations.", + "
", + "MemoryBear — Daybreak 🐻✨" + ] + } + }, "v0.2.10": { "introduction": { "codeName": "炼剑", From 7ba0726473a0443e482c506514d7ee60b65b4f1c Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Thu, 16 Apr 2026 16:36:15 +0800 Subject: [PATCH 076/245] refactor(model): remove mutual exclusion logic between json_output and deep_thinking --- api/app/core/models/base.py | 6 ------ 1 file changed, 6 deletions(-) diff --git a/api/app/core/models/base.py b/api/app/core/models/base.py index 90046412..57054da1 100644 --- a/api/app/core/models/base.py +++ b/api/app/core/models/base.py @@ -52,12 +52,6 @@ class RedBearModelConfig(BaseModel): f"模型 {self.model_name} 不支持 JSON 输出（capability 中无 'json_output'），已自动关闭 json_output" ) self.json_output = False - if self.json_output and self.deep_thinking: - logger.warning( - f"模型 {self.model_name} json_output 与 deep_thinking 互斥，已自动关闭 deep_thinking" - ) - self.deep_thinking = False - self.thinking_budget_tokens = None return self From 73fbc197471780a750b5298cb351fb3c3307bd01 Mon Sep 17 00:00:00 2001 From: lanceyq <1982376970@qq.com> Date: Thu, 16 Apr 2026 17:14:30 +0800 Subject: [PATCH 077/245] refactor(memory): switch metadata extraction from full-replace to incremental changes - Replace UserMetadata full-object overwrite with incremental MetadataFieldChange operations (set/remove per field path) - Convert profile.role and profile.domain from scalar strings to lists - Remove UserMetadataBehavioralHints and knowledge_tags fields - Update Jinja2 prompt to instruct LLM to output incremental changes - Update extract_user_metadata_task to apply changes via deep-copy and per-field mutation for proper SQLAlchemy change detection - Minor lint: remove unnecessary f-string prefixes in tasks.py --- api/app/core/memory/models/__init__.py | 4 +- api/app/core/memory/models/metadata_models.py | 40 ++++--- .../metadata_extractor.py | 9 +- .../prompts/extract_user_metadata.jinja2 | 103 ++++++++--------- api/app/tasks.py | 104 ++++++++++++------ 5 files changed, 152 insertions(+), 108 deletions(-) diff --git a/api/app/core/memory/models/__init__.py b/api/app/core/memory/models/__init__.py index eed8e8c4..2a34159b 100644 --- a/api/app/core/memory/models/__init__.py +++ b/api/app/core/memory/models/__init__.py @@ -61,9 +61,9 @@ from app.core.memory.models.triplet_models import ( # User metadata models from app.core.memory.models.metadata_models import ( UserMetadata, - UserMetadataBehavioralHints, UserMetadataProfile, MetadataExtractionResponse, + MetadataFieldChange, ) # Ontology scenario models (LLM extracted from scenarios) @@ -133,9 +133,9 @@ __all__ = [ "Triplet", "TripletExtractionResponse", "UserMetadata", - "UserMetadataBehavioralHints", "UserMetadataProfile", "MetadataExtractionResponse", + "MetadataFieldChange", # Ontology models "OntologyClass", "OntologyExtractionResponse", diff --git a/api/app/core/memory/models/metadata_models.py b/api/app/core/memory/models/metadata_models.py index 55c2359e..1b490dbc 100644 --- a/api/app/core/memory/models/metadata_models.py +++ b/api/app/core/memory/models/metadata_models.py @@ -4,7 +4,7 @@ Independent from triplet_models.py - these models are used by the standalone metadata extraction pipeline (post-dedup async Celery task). """ -from typing import List +from typing import List, Optional from pydantic import BaseModel, ConfigDict, Field @@ -13,8 +13,8 @@ class UserMetadataProfile(BaseModel): """用户画像信息""" model_config = ConfigDict(extra="ignore") - role: str = Field(default="", description="用户职业或角色") - domain: str = Field(default="", description="用户所在领域") + role: List[str] = Field(default_factory=list, description="用户职业或角色") + domain: List[str] = Field(default_factory=list, description="用户所在领域") expertise: List[str] = Field( default_factory=list, description="用户擅长的技能或工具" ) @@ -23,31 +23,37 @@ class UserMetadataProfile(BaseModel): ) -class UserMetadataBehavioralHints(BaseModel): - """行为偏好""" - - model_config = ConfigDict(extra="ignore") - learning_stage: str = Field(default="", description="学习阶段") - preferred_depth: str = Field(default="", description="偏好深度") - tone_preference: str = Field(default="", description="语气偏好") - - class UserMetadata(BaseModel): """用户元数据顶层结构""" model_config = ConfigDict(extra="ignore") profile: UserMetadataProfile = Field(default_factory=UserMetadataProfile) - behavioral_hints: UserMetadataBehavioralHints = Field( - default_factory=UserMetadataBehavioralHints + + +class MetadataFieldChange(BaseModel): + """单个元数据字段的变更操作""" + + model_config = ConfigDict(extra="ignore") + field_path: str = Field( + description="字段路径，用点号分隔，如 'profile.role'、'knowledge_tags'、'behavioral_hints.tone_preference'" + ) + action: str = Field( + description="操作类型：'set' 表示新增或修改，'remove' 表示移除" + ) + value: Optional[str] = Field( + default=None, + description="字段的新值（action='set' 时必填）。标量字段直接填值，列表字段填单个要新增的元素" ) - knowledge_tags: List[str] = Field(default_factory=list, description="知识标签") class MetadataExtractionResponse(BaseModel): - """元数据提取 LLM 响应结构""" + """元数据提取 LLM 响应结构（增量模式）""" model_config = ConfigDict(extra="ignore") - user_metadata: UserMetadata = Field(default_factory=UserMetadata) + metadata_changes: List[MetadataFieldChange] = Field( + default_factory=list, + description="元数据的增量变更列表，每项描述一个字段的新增、修改或移除操作", + ) aliases_to_add: List[str] = Field( default_factory=list, description="本次新发现的用户别名（用户自我介绍或他人对用户的称呼）", diff --git a/api/app/core/memory/storage_services/extraction_engine/knowledge_extraction/metadata_extractor.py b/api/app/core/memory/storage_services/extraction_engine/knowledge_extraction/metadata_extractor.py index 19f1e533..29f4e85b 100644 --- a/api/app/core/memory/storage_services/extraction_engine/knowledge_extraction/metadata_extractor.py +++ b/api/app/core/memory/storage_services/extraction_engine/knowledge_extraction/metadata_extractor.py @@ -118,7 +118,7 @@ class MetadataExtractor: existing_aliases: Optional[List[str]] = None, ) -> Optional[tuple]: """ - 对筛选后的 statement 列表调用 LLM 提取元数据和用户别名。 + 对筛选后的 statement 列表调用 LLM 提取元数据增量变更和用户别名。 Args: statements: 用户发言的 statement 文本列表 @@ -126,7 +126,8 @@ class MetadataExtractor: existing_aliases: 数据库已有的用户别名列表（可选） Returns: - (UserMetadata, List[str], List[str]) tuple: (metadata, aliases_to_add, aliases_to_remove) on success, None on failure + (List[MetadataFieldChange], List[str], List[str]) tuple: + (metadata_changes, aliases_to_add, aliases_to_remove) on success, None on failure """ if not statements: return None @@ -160,12 +161,12 @@ class MetadataExtractor: ) if response: - metadata = response.user_metadata if response.user_metadata else None + changes = response.metadata_changes if response.metadata_changes else [] to_add = response.aliases_to_add if response.aliases_to_add else [] to_remove = ( response.aliases_to_remove if response.aliases_to_remove else [] ) - return metadata, to_add, to_remove + return changes, to_add, to_remove logger.warning("LLM 返回的响应为空") return None diff --git a/api/app/core/memory/utils/prompt/prompts/extract_user_metadata.jinja2 b/api/app/core/memory/utils/prompt/prompts/extract_user_metadata.jinja2 index 5d019b12..1c32d369 100644 --- a/api/app/core/memory/utils/prompt/prompts/extract_user_metadata.jinja2 +++ b/api/app/core/memory/utils/prompt/prompts/extract_user_metadata.jinja2 @@ -1,5 +1,5 @@ ===Task=== -Extract user metadata from the following conversation statements spoken by the user. +Extract user metadata changes from the following conversation statements spoken by the user. {% if language == "zh" %} **"三度原则"判断标准：** @@ -10,28 +10,36 @@ Extract user metadata from the following conversation statements spoken by the u **提取规则：** - **只提取关于"用户本人"的画像信息**，忽略用户提到的第三方人物（如朋友、同事、家人）的信息 - 仅提取文本中明确提到的信息，不要推测 -- 如果文本中没有可提取的用户画像信息，返回空的 user_metadata 对象 - **输出语言必须与输入文本的语言一致**（输入中文则输出中文值，输入英文则输出英文值） +**增量模式（重要）：** +你只需要输出**本次对话引起的变更操作**，不要输出完整的元数据。每个变更是一个对象，包含： +- `field_path`：字段路径，用点号分隔（如 `profile.role`、`profile.expertise`） +- `action`：操作类型 + * `set`：新增或修改一个字段的值 + * `remove`：移除一个字段的值 +- `value`：字段的新值（`action="set"` 时必填，`action="remove"` 时填要移除的元素值） + * 所有字段均为列表类型，每个元素一条变更记录 + +**判断规则：** +- 用户提到新信息 → `action="set"`，填入新值 +- 用户明确否定已有信息（如"我不再做老师了"、"我已经不学Python了"）→ `action="remove"`，`value` 填要移除的元素值 +- 如果本次对话没有任何可提取的变更，返回空的 `metadata_changes` 数组 `[]` +- **不要为未被提及的字段生成任何变更操作** + {% if existing_metadata %} -**重要：合并已有元数据** -下方提供了数据库中已有的用户元数据。请结合用户最新发言，输出**合并后的完整元数据**： -- 如果用户明确否定了已有信息（如"我不再教高中物理了"），在输出中**移除**该信息 -- 如果用户提到了新信息，**添加**到对应字段中 -- 如果已有信息未被用户否定，**保留**在输出中 -- 标量字段（如 role、domain）：如果用户提到了新值，用新值替换；否则保留已有值 -- 最终输出应该是完整的、合并后的元数据，不是增量 +**已有元数据（仅供参考，用于判断是否需要变更）：** +请对比已有数据和用户最新发言，只输出差异部分的变更操作。 +- 如果用户说的信息和已有数据一致，不需要输出变更 +- 如果用户否定了已有数据中的某个值，输出 `remove` 操作 +- 如果用户提到了新信息，输出 `set` 操作 {% endif %} **字段说明：** -- profile.role：用户的职业或角色，如 教师、医生、后端工程师 -- profile.domain：用户所在领域，如 教育、医疗、软件开发 -- profile.expertise：用户擅长的技能或工具（通用，不限于编程），如 Python、心理咨询、高中物理 -- profile.interests：用户主动表达兴趣的话题或领域标签 -- behavioral_hints.learning_stage：学习阶段（初学者/中级/高级） -- behavioral_hints.preferred_depth：偏好深度（概览/技术细节/深入探讨） -- behavioral_hints.tone_preference：语气偏好（轻松随意/专业简洁/学术严谨） -- knowledge_tags：用户涉及的知识领域标签 +- profile.role：用户的职业或角色（列表），如 教师、医生、后端工程师，一个人可以有多个角色 +- profile.domain：用户所在领域（列表），如 教育、医疗、软件开发，一个人可以涉及多个领域 +- profile.expertise：用户擅长的技能或工具（列表），如 Python、心理咨询、高中物理 +- profile.interests：用户主动表达兴趣的话题或领域标签（列表） **用户别名变更（增量模式）：** - **aliases_to_add**：本次新发现的用户别名，包括： @@ -43,7 +51,6 @@ Extract user metadata from the following conversation statements spoken by the u - **aliases_to_remove**：用户明确否认的别名，包括： * 用户说"我不叫XX了"、"别叫我XX"、"我改名了，不叫XX" → 将 XX 放入此数组 * **严格限制**：只将用户原文中**逐字提到**的被否认名字放入，不要推断关联的其他别名 - * 例如：用户说"我不叫陈小刀了" → 只移除"陈小刀"，不要移除"陈哥"、"老陈"等未被提及的别名 * 如果没有要移除的别名，返回空数组 `[]` {% if existing_aliases %} - 已有别名：{{ existing_aliases | tojson }}（仅供参考，不需要在输出中重复） @@ -57,28 +64,36 @@ Extract user metadata from the following conversation statements spoken by the u **Extraction rules:** - **Only extract profile information about the user themselves**, ignore information about third parties (friends, colleagues, family) mentioned by the user - Only extract information explicitly mentioned in the text, do not speculate -- If no user profile information can be extracted, return an empty user_metadata object - **Output language must match the input text language** +**Incremental mode (important):** +You should only output **the change operations caused by this conversation**, not the complete metadata. Each change is an object containing: +- `field_path`: Field path separated by dots (e.g. `profile.role`, `profile.expertise`) +- `action`: Operation type + * `set`: Add or update a field value + * `remove`: Remove a field value +- `value`: The new value for the field (required when `action="set"`, for `action="remove"` fill in the element value to remove) + * All fields are list types, one change record per element + +**Decision rules:** +- User mentions new information → `action="set"`, fill in the new value +- User explicitly negates existing info (e.g. "I'm no longer a teacher", "I stopped learning Python") → `action="remove"`, `value` is the element to remove +- If this conversation has no extractable changes, return an empty `metadata_changes` array `[]` +- **Do NOT generate any change operations for fields not mentioned in the conversation** + {% if existing_metadata %} -**Important: Merge with existing metadata** -Existing user metadata from the database is provided below. Combine with the user's latest statements to output the **complete merged metadata**: -- If the user explicitly negates existing info (e.g. "I no longer teach high school physics"), **remove** it from output -- If the user mentions new info, **add** it to the corresponding field -- If existing info is not negated by the user, **keep** it in the output -- Scalar fields (e.g. role, domain): replace with new value if user mentions one; otherwise keep existing -- The final output should be the complete, merged metadata — not an incremental update +**Existing metadata (for reference only, to determine if changes are needed):** +Compare existing data with the user's latest statements, and only output change operations for the differences. +- If the user's statement matches existing data, no change is needed +- If the user negates a value in existing data, output a `remove` operation +- If the user mentions new information, output a `set` operation {% endif %} **Field descriptions:** -- profile.role: User's occupation or role, e.g. teacher, doctor, software engineer -- profile.domain: User's domain, e.g. education, healthcare, software development -- profile.expertise: User's skills or tools (general, not limited to programming) -- profile.interests: Topics or domain tags the user actively expressed interest in -- behavioral_hints.learning_stage: Learning stage (beginner/intermediate/advanced) -- behavioral_hints.preferred_depth: Preferred depth (overview/detailed/deep dive) -- behavioral_hints.tone_preference: Tone preference (casual/professional/academic) -- knowledge_tags: Knowledge domain tags related to the user +- profile.role: User's occupation or role (list), e.g. teacher, doctor, software engineer. A person can have multiple roles +- profile.domain: User's domain (list), e.g. education, healthcare, software development. A person can span multiple domains +- profile.expertise: User's skills or tools (list), e.g. Python, counseling, physics +- profile.interests: Topics or domain tags the user actively expressed interest in (list) **User alias changes (incremental mode):** - **aliases_to_add**: Newly discovered user aliases from this conversation, including: @@ -90,7 +105,6 @@ Existing user metadata from the database is provided below. Combine with the use - **aliases_to_remove**: Aliases the user explicitly denies, including: * User says "Don't call me XX anymore", "I'm not called XX", "I changed my name from XX" → put XX in this array * **Strict rule**: Only include the exact name the user **verbatim mentions** as denied. Do NOT infer or remove related aliases - * Example: User says "I'm not called John anymore" → only remove "John", do NOT remove "Johnny", "J" or other related aliases not mentioned * If no aliases to remove, return empty array `[]` {% if existing_aliases %} - Existing aliases: {{ existing_aliases | tojson }} (for reference only, do not repeat in output) @@ -113,20 +127,11 @@ Existing user metadata from the database is provided below. Combine with the use Return a JSON object with the following structure: ```json { - "user_metadata": { - "profile": { - "role": "", - "domain": "", - "expertise": [], - "interests": [] - }, - "behavioral_hints": { - "learning_stage": "", - "preferred_depth": "", - "tone_preference": "" - }, - "knowledge_tags": [] - }, + "metadata_changes": [ + {"field_path": "profile.role", "action": "set", "value": "后端工程师"}, + {"field_path": "profile.expertise", "action": "set", "value": "Python"}, + {"field_path": "profile.expertise", "action": "remove", "value": "Java"} + ], "aliases_to_add": [], "aliases_to_remove": [] } diff --git a/api/app/tasks.py b/api/app/tasks.py index 5a71066a..2c2f50a9 100644 --- a/api/app/tasks.py +++ b/api/app/tasks.py @@ -455,7 +455,7 @@ def build_graphrag_for_kb(kb_id: uuid.UUID): db_knowledge = db.query(Knowledge).filter(Knowledge.id == kb_id).first() if db_knowledge is None: logger.error(f"[GraphRAG-KB] knowledge={kb_id} not found") - return f"build knowledge graph failed: knowledge not found" + return "build knowledge graph failed: knowledge not found" if not (db_knowledge.parser_config and db_knowledge.parser_config.get("graphrag", {}).get("use_graphrag", False)): @@ -538,7 +538,7 @@ def build_graphrag_for_document(document_id: str, knowledge_id: str): db_knowledge = db.query(Knowledge).filter(Knowledge.id == uuid.UUID(knowledge_id)).first() if db_document is None or db_knowledge is None: logger.error(f"[GraphRAG] document={document_id} or knowledge={knowledge_id} not found") - return f"build_graphrag_for_document failed: record not found" + return "build_graphrag_for_document failed: record not found" graphrag_conf = db_knowledge.parser_config.get("graphrag", {}) with_resolution = graphrag_conf.get("resolution", False) @@ -617,7 +617,7 @@ def sync_knowledge_for_kb(kb_id: uuid.UUID): db_knowledge = db.query(Knowledge).filter(Knowledge.id == kb_id).first() if db_knowledge is None: logger.error(f"[SyncKB] knowledge={kb_id} not found") - return f"sync knowledge failed: knowledge not found" + return "sync knowledge failed: knowledge not found" # 1. get vector_service vector_service = ElasticSearchVectorFactory().init_vector(knowledge=db_knowledge) @@ -3102,29 +3102,11 @@ def extract_user_metadata_task( logger.info(f"[CELERY METADATA] No metadata extracted for end_user_id={end_user_id}") return {"status": "SUCCESS", "result": "no_metadata_extracted"} - user_metadata, aliases_to_add, aliases_to_remove = extract_result - logger.info(f"[CELERY METADATA] LLM 别名新增: {aliases_to_add}, 移除: {aliases_to_remove}") - - # 4. 清洗元数据、覆盖写入元数据和别名 - def clean_metadata(raw: dict) -> dict: - """递归移除空字符串、空列表、空字典。""" - result = {} - for k, v in raw.items(): - if v == "" or v == []: - continue - if isinstance(v, dict): - cleaned = clean_metadata(v) - if cleaned: - result[k] = cleaned - else: - result[k] = v - return result - - raw_dict = user_metadata.model_dump(exclude_none=True) if user_metadata else {} - logger.info(f"[CELERY METADATA] LLM 输出完整元数据: {json.dumps(raw_dict, ensure_ascii=False)}") - - cleaned = clean_metadata(raw_dict) if raw_dict else {} - logger.info(f"[CELERY METADATA] 清洗后元数据: {json.dumps(cleaned, ensure_ascii=False)}") + metadata_changes, aliases_to_add, aliases_to_remove = extract_result + logger.info( + f"[CELERY METADATA] LLM 元数据变更: {[c.model_dump() for c in metadata_changes]}, " + f"别名新增: {aliases_to_add}, 移除: {aliases_to_remove}" + ) from datetime import datetime as dt, timezone as tz now = dt.now(tz.utc).isoformat() @@ -3152,15 +3134,50 @@ def extract_user_metadata_task( end_user = EndUserRepository(db).get_by_id(end_user_uuid) if info: - # 元数据覆盖写入 - if cleaned: - existing_meta = info.meta_data if info.meta_data else {} + # 4. 元数据增量更新（按 LLM 输出的变更操作逐条执行，所有字段均为列表类型） + if metadata_changes: + # 深拷贝，确保 SQLAlchemy 能检测到变更 + import copy + existing_meta = copy.deepcopy(info.meta_data) if info.meta_data else {} updated_at = dict(existing_meta.get("_updated_at", {})) - _update_timestamps(existing_meta, cleaned, updated_at, now) - final = dict(cleaned) - final["_updated_at"] = updated_at - info.meta_data = final - logger.info("[CELERY METADATA] 覆盖写入元数据") + + for change in metadata_changes: + field_path = change.field_path + action = change.action + value = change.value + + if not value or not value.strip(): + continue + + # 定位到目标字段的父级节点 + parts = field_path.split(".") + target = existing_meta + for part in parts[:-1]: + target = target.setdefault(part, {}) + leaf = parts[-1] + + current_list = target.get(leaf, []) + if not isinstance(current_list, list): + current_list = [] + + if action == "set": + if value not in current_list: + current_list.append(value) + target[leaf] = current_list + logger.info(f"[CELERY METADATA] set {field_path} = {value}") + + elif action == "remove": + if value in current_list: + current_list.remove(value) + target[leaf] = current_list + logger.info(f"[CELERY METADATA] remove {value} from {field_path}") + + updated_at[field_path] = now + + existing_meta["_updated_at"] = updated_at + # 赋值深拷贝后的新对象，SQLAlchemy 会检测到字段变更并写入 + info.meta_data = existing_meta + logger.info(f"[CELERY METADATA] 增量更新元数据完成: {json.dumps(existing_meta, ensure_ascii=False)}") # 别名增量增删：(已有 - remove) + add old_aliases = info.aliases if info.aliases else [] @@ -3196,12 +3213,27 @@ def extract_user_metadata_task( from app.models.end_user_info_model import EndUserInfo initial_aliases = filtered_add # 新记录只有 add，没有 remove first_alias = initial_aliases[0] if initial_aliases else "" - if first_alias or cleaned: + + # 从变更操作构建初始元数据（所有字段均为列表类型） + initial_meta = {} + for change in metadata_changes: + if change.action == "set" and change.value is not None and change.value.strip(): + parts = change.field_path.split(".") + target = initial_meta + for part in parts[:-1]: + target = target.setdefault(part, {}) + leaf = parts[-1] + current_list = target.get(leaf, []) + if change.value not in current_list: + current_list.append(change.value) + target[leaf] = current_list + + if first_alias or initial_meta: new_info = EndUserInfo( end_user_id=end_user_uuid, other_name=first_alias or "", aliases=initial_aliases, - meta_data=cleaned if cleaned else None, + meta_data=initial_meta if initial_meta else None, ) db.add(new_info) if end_user and first_alias and ( From 643f69bb907d1dcb80aa0518550c3ede4782235c Mon Sep 17 00:00:00 2001 From: lanceyq <1982376970@qq.com> Date: Thu, 16 Apr 2026 17:29:00 +0800 Subject: [PATCH 078/245] refactor(memory): tighten metadata field types and clean up descriptions - Use Literal['set', 'remove'] for MetadataFieldChange.action instead of str - Simplify field_path description to reflect current schema - Remove redundant isinstance check in extract_user_metadata_task --- api/app/core/memory/models/metadata_models.py | 6 +++--- api/app/tasks.py | 2 -- 2 files changed, 3 insertions(+), 5 deletions(-) diff --git a/api/app/core/memory/models/metadata_models.py b/api/app/core/memory/models/metadata_models.py index 1b490dbc..e12c3d97 100644 --- a/api/app/core/memory/models/metadata_models.py +++ b/api/app/core/memory/models/metadata_models.py @@ -4,7 +4,7 @@ Independent from triplet_models.py - these models are used by the standalone metadata extraction pipeline (post-dedup async Celery task). """ -from typing import List, Optional +from typing import List, Literal, Optional from pydantic import BaseModel, ConfigDict, Field @@ -35,9 +35,9 @@ class MetadataFieldChange(BaseModel): model_config = ConfigDict(extra="ignore") field_path: str = Field( - description="字段路径，用点号分隔，如 'profile.role'、'knowledge_tags'、'behavioral_hints.tone_preference'" + description="字段路径，用点号分隔，如 'profile.role'、'profile.expertise'" ) - action: str = Field( + action: Literal["set", "remove"] = Field( description="操作类型：'set' 表示新增或修改，'remove' 表示移除" ) value: Optional[str] = Field( diff --git a/api/app/tasks.py b/api/app/tasks.py index 2c2f50a9..6d2e07a3 100644 --- a/api/app/tasks.py +++ b/api/app/tasks.py @@ -3157,8 +3157,6 @@ def extract_user_metadata_task( leaf = parts[-1] current_list = target.get(leaf, []) - if not isinstance(current_list, list): - current_list = [] if action == "set": if value not in current_list: From ccdf7ae81da4107fbd5eb2d4686b2ebd2b62fe11 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Thu, 16 Apr 2026 17:40:30 +0800 Subject: [PATCH 079/245] refactor(model): replace VolcanoChatOpenAI with CompatibleChatOpenAI for unified omni model support --- api/app/core/models/base.py | 8 ++++---- .../core/models/{volcano_chat.py => compatible_chat.py} | 4 ++-- 2 files changed, 6 insertions(+), 6 deletions(-) rename api/app/core/models/{volcano_chat.py => compatible_chat.py} (92%) diff --git a/api/app/core/models/base.py b/api/app/core/models/base.py index 57054da1..7b570b47 100644 --- a/api/app/core/models/base.py +++ b/api/app/core/models/base.py @@ -14,7 +14,7 @@ from pydantic import BaseModel, Field, model_validator from app.core.error_codes import BizCode from app.core.exceptions import BusinessException from app.models.models_model import ModelProvider, ModelType -from app.core.models.volcano_chat import VolcanoChatOpenAI +from app.core.models.compatible_chat import CompatibleChatOpenAI T = TypeVar("T") @@ -255,11 +255,11 @@ def get_provider_llm_class(config: RedBearModelConfig, type: ModelType = ModelTy """根据模型提供商获取对应的模型类""" provider = config.provider.lower() - # dashscope 的 omni 模型使用 OpenAI 兼容模式 + # dashscope的omni模型 和 volcano模型使用 if provider == ModelProvider.DASHSCOPE and config.is_omni: - return ChatOpenAI + return CompatibleChatOpenAI if provider == ModelProvider.VOLCANO: - return VolcanoChatOpenAI + return CompatibleChatOpenAI if provider in [ModelProvider.OPENAI, ModelProvider.XINFERENCE, ModelProvider.GPUSTACK]: if type == ModelType.LLM: return OpenAI diff --git a/api/app/core/models/volcano_chat.py b/api/app/core/models/compatible_chat.py similarity index 92% rename from api/app/core/models/volcano_chat.py rename to api/app/core/models/compatible_chat.py index d9a51d13..114a3567 100644 --- a/api/app/core/models/volcano_chat.py +++ b/api/app/core/models/compatible_chat.py @@ -12,8 +12,8 @@ from langchain_core.outputs import ChatGenerationChunk, ChatResult from langchain_openai import ChatOpenAI -class VolcanoChatOpenAI(ChatOpenAI): - """火山引擎 Chat 模型，支持深度思考内容（reasoning_content）的流式和非流式透传。""" +class CompatibleChatOpenAI(ChatOpenAI): + """火山和千问的omni兼容模型，支持深度思考内容（reasoning_content）的流式和非流式透传。""" def _create_chat_result(self, response: Union[dict, Any], generation_info: Optional[dict] = None) -> ChatResult: result = super()._create_chat_result(response, generation_info) From 62edafcebe7e268063596194b3ae71eb1b612781 Mon Sep 17 00:00:00 2001 From: Ke Sun Date: Thu, 16 Apr 2026 17:43:23 +0800 Subject: [PATCH 080/245] ci(workflow): add PR number and merge commit SHA to WeChat release notification - Add PR_NUMBER environment variable to capture pull request number - Add MERGE_SHA environment variable to capture merge commit SHA - Extract short SHA (first 7 characters) from merge commit for display - Update notification content to include PR number with # prefix - Update notification content to include short commit SHA - Improve release notification with additional metadata for better traceability --- .github/workflows/release-notify-wechat.yml | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/.github/workflows/release-notify-wechat.yml b/.github/workflows/release-notify-wechat.yml index bc67518b..935d84d5 100644 --- a/.github/workflows/release-notify-wechat.yml +++ b/.github/workflows/release-notify-wechat.yml @@ -121,6 +121,8 @@ jobs: AUTHOR: ${{ github.event.pull_request.user.login }} PR_TITLE: ${{ github.event.pull_request.title }} PR_URL: ${{ github.event.pull_request.html_url }} + PR_NUMBER: ${{ github.event.pull_request.number }} + MERGE_SHA: ${{ github.event.pull_request.merge_commit_sha }} SOURCERY_FOUND: ${{ steps.sourcery.outputs.found }} SOURCERY_SUMMARY: ${{ steps.sourcery.outputs.summary }} QWEN_SUMMARY: ${{ steps.qwen.outputs.summary }} @@ -135,11 +137,16 @@ jobs: label = "AI变更摘要" summary = os.environ.get("QWEN_SUMMARY", "AI 摘要生成失败") + pr_number = os.environ.get("PR_NUMBER", "") + short_sha = os.environ.get("MERGE_SHA", "")[:7] + content = ( "## 🚀 Release 发布通知\n" - "> 📦 **分支**: " + os.environ["BRANCH"] + "\n" + "> � **分支**: " + os.environ["BRANCH"] + "\n" "> 👤 **提交人**: " + os.environ["AUTHOR"] + "\n" - "> 📝 **标题**: " + os.environ["PR_TITLE"] + "\n\n" + "> 📝 **标题**: " + os.environ["PR_TITLE"] + "\n" + "> 🔢 **PR编号**: #" + pr_number + "\n" + "> 🔖 **Commit**: " + short_sha + "\n\n" "### 🧠 " + label + "\n" + summary + "\n\n" "---\n" From e1f8ad871b4c78b29177cc9df837228ebfbc8504 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Thu, 16 Apr 2026 17:47:47 +0800 Subject: [PATCH 081/245] refactor(model): replace qwen-vl-plus-latest with json_output capability in dashscope_models.yaml --- api/app/core/models/scripts/dashscope_models.yaml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/api/app/core/models/scripts/dashscope_models.yaml b/api/app/core/models/scripts/dashscope_models.yaml index 26c3cc2c..9b45f107 100644 --- a/api/app/core/models/scripts/dashscope_models.yaml +++ b/api/app/core/models/scripts/dashscope_models.yaml @@ -461,7 +461,7 @@ models: capability: - vision - video - - qwen-vl-plus-latest + - json_output is_omni: false tags: - 大语言模型 @@ -480,7 +480,7 @@ models: capability: - vision - video - - qwen-vl-plus-latest + - json_output is_omni: false tags: - 大语言模型 From 915cb54f2130cb8b0479d7508763cda0f42b27b6 Mon Sep 17 00:00:00 2001 From: wxy Date: Thu, 16 Apr 2026 17:54:50 +0800 Subject: [PATCH 082/245] feat(tenant): add public subscription plan list endpoint and enhance plan information Add a public subscription plan list endpoint that can be accessed without authentication. Enhance the returned subscription plan information fields, including multi-language support and default free plan fallback logic. Additionally, implement automatic model binding for the knowledge base service. --- api/app/config/default_free_plan.py | 5 + api/app/controllers/__init__.py | 1 + .../tenant_subscription_controller.py | 95 ++++++++++++++++++- api/app/services/knowledge_service.py | 51 +++++++++- 4 files changed, 148 insertions(+), 4 deletions(-) diff --git a/api/app/config/default_free_plan.py b/api/app/config/default_free_plan.py index 23a3a10e..4b357236 100644 --- a/api/app/config/default_free_plan.py +++ b/api/app/config/default_free_plan.py @@ -5,6 +5,7 @@ DEFAULT_FREE_PLAN = { "name": "记忆体验版", + "name_en": "Memory Experience", "category": "saas_personal", "tier_level": 0, "version": "1.0", @@ -12,9 +13,13 @@ DEFAULT_FREE_PLAN = { "price": 0, "billing_cycle": "permanent_free", "core_value": "感受永久记忆", + "core_value_en": "Experience Permanent Memory", "tech_support": "社群交流", + "tech_support_en": "Community Support", "sla_compliance": "无", + "sla_compliance_en": "None", "page_customization": "无", + "page_customization_en": "None", "theme_color": "#64748B", "quotas": { "workspace_quota": 1, diff --git a/api/app/controllers/__init__.py b/api/app/controllers/__init__.py index 377205c4..e9417d68 100644 --- a/api/app/controllers/__init__.py +++ b/api/app/controllers/__init__.py @@ -100,5 +100,6 @@ manager_router.include_router(ontology_controller.router) manager_router.include_router(skill_controller.router) manager_router.include_router(i18n_controller.router) manager_router.include_router(tenant_subscription_controller.router) +manager_router.include_router(tenant_subscription_controller.public_router) __all__ = ["manager_router"] diff --git a/api/app/controllers/tenant_subscription_controller.py b/api/app/controllers/tenant_subscription_controller.py index c3fde572..141d48a8 100644 --- a/api/app/controllers/tenant_subscription_controller.py +++ b/api/app/controllers/tenant_subscription_controller.py @@ -2,7 +2,7 @@ 租户套餐查询接口（普通用户可访问） """ import datetime -from typing import Callable +from typing import Callable, Optional from fastapi import APIRouter, Depends from fastapi.responses import JSONResponse @@ -19,6 +19,7 @@ from app.schemas.response_schema import ApiResponse logger = get_api_logger() router = APIRouter(prefix="/tenant", tags=["Tenant"]) +public_router = APIRouter(tags=["Tenant"]) @router.get("/subscription", response_model=ApiResponse, summary="获取当前用户所属租户的套餐信息") @@ -42,7 +43,41 @@ async def get_my_tenant_subscription( sub = svc.get_subscription(tenant_id) if not sub: - return success(data=None, msg="暂无有效套餐") + # 无订阅记录时，兜底返回免费套餐信息 + free_plan = svc.plan_repo.get_free_plan() + if not free_plan: + return success(data=None, msg="暂无有效套餐") + return success(data={ + "subscription_id": None, + "tenant_id": str(tenant_id), + "package_plan_id": str(free_plan.id), + "package_version": free_plan.version, + "package_plan": { + "id": str(free_plan.id), + "name": free_plan.name, + "name_en": free_plan.name_en, + "version": free_plan.version, + "category": free_plan.category, + "tier_level": free_plan.tier_level, + "price": float(free_plan.price) if free_plan.price is not None else 0.0, + "billing_cycle": free_plan.billing_cycle, + "core_value": free_plan.core_value, + "core_value_en": free_plan.core_value_en, + "tech_support": free_plan.tech_support, + "tech_support_en": free_plan.tech_support_en, + "sla_compliance": free_plan.sla_compliance, + "sla_compliance_en": free_plan.sla_compliance_en, + "page_customization": free_plan.page_customization, + "page_customization_en": free_plan.page_customization_en, + "theme_color": free_plan.theme_color, + }, + "started_at": None, + "expired_at": None, + "status": "active", + "quota": free_plan.quotas or {}, + "created_at": int(datetime.datetime.utcnow().timestamp() * 1000), + "updated_at": int(datetime.datetime.utcnow().timestamp() * 1000), + }, msg="免费套餐") return success(data=svc.build_response(sub)) @@ -62,11 +97,21 @@ async def get_my_tenant_subscription( "package_plan": { "id": None, "name": plan["name"], + "name_en": plan.get("name_en"), "version": plan["version"], "category": plan["category"], "tier_level": plan["tier_level"], "price": float(plan["price"]), "billing_cycle": plan["billing_cycle"], + "core_value": plan.get("core_value"), + "core_value_en": plan.get("core_value_en"), + "tech_support": plan.get("tech_support"), + "tech_support_en": plan.get("tech_support_en"), + "sla_compliance": plan.get("sla_compliance"), + "sla_compliance_en": plan.get("sla_compliance_en"), + "page_customization": plan.get("page_customization"), + "page_customization_en": plan.get("page_customization_en"), + "theme_color": plan.get("theme_color"), }, "started_at": None, "expired_at": None, @@ -80,3 +125,49 @@ async def get_my_tenant_subscription( except Exception as e: logger.error(f"获取租户套餐信息失败: {e}", exc_info=True) return JSONResponse(status_code=500, content=fail(code=500, msg="获取套餐信息失败")) + + +@public_router.get("/package-plans", response_model=ApiResponse, summary="获取套餐列表（公开）") +async def list_package_plans_public( + category: Optional[str] = None, + status: Optional[bool] = None, + search: Optional[str] = None, + db: Session = Depends(get_db), +): + """ + 公开接口，无需鉴权。 + SaaS 版从数据库读取套餐列表；社区版降级返回 default_free_plan.py 中的免费套餐。 + """ + try: + from premium.platform_admin.package_plan_service import PackagePlanService + from premium.platform_admin.package_plan_schema import PackagePlanResponse + svc = PackagePlanService(db) + result = svc.get_list(page=1, size=9999, category=category, status=status, search=search) + return success(data=[PackagePlanResponse.model_validate(p).model_dump(mode="json") for p in result["items"]]) + except ModuleNotFoundError: + from app.config.default_free_plan import DEFAULT_FREE_PLAN + plan = DEFAULT_FREE_PLAN + return success(data=[{ + "id": None, + "name": plan["name"], + "name_en": plan.get("name_en"), + "version": plan["version"], + "category": plan["category"], + "tier_level": plan["tier_level"], + "price": float(plan["price"]), + "billing_cycle": plan["billing_cycle"], + "core_value": plan.get("core_value"), + "core_value_en": plan.get("core_value_en"), + "tech_support": plan.get("tech_support"), + "tech_support_en": plan.get("tech_support_en"), + "sla_compliance": plan.get("sla_compliance"), + "sla_compliance_en": plan.get("sla_compliance_en"), + "page_customization": plan.get("page_customization"), + "page_customization_en": plan.get("page_customization_en"), + "theme_color": plan.get("theme_color"), + "status": plan.get("status", True), + "quota": plan["quotas"], + }]) + except Exception as e: + logger.error(f"获取套餐列表失败: {e}", exc_info=True) + return JSONResponse(status_code=500, content=fail(code=500, msg="获取套餐列表失败")) diff --git a/api/app/services/knowledge_service.py b/api/app/services/knowledge_service.py index bac02e96..46108efd 100644 --- a/api/app/services/knowledge_service.py +++ b/api/app/services/knowledge_service.py @@ -2,11 +2,14 @@ import uuid from sqlalchemy.orm import Session from app.models.user_model import User from app.models.knowledge_model import Knowledge +from app.models.workspace_model import Workspace +from app.models.models_model import ModelConfig from app.schemas.knowledge_schema import KnowledgeCreate, KnowledgeUpdate from app.repositories import knowledge_repository from app.core.logging_config import get_business_logger +from app.repositories.model_repository import ModelConfigRepository +from app.models.models_model import ModelType -# Obtain a dedicated logger for business logic business_logger = get_business_logger() @@ -60,13 +63,57 @@ def create_knowledge( db: Session, knowledge: KnowledgeCreate, current_user: User ) -> Knowledge: business_logger.info(f"Create a knowledge base: {knowledge.name}, creator: {current_user.username}") - + try: knowledge.created_by = current_user.id if knowledge.workspace_id is None: knowledge.workspace_id = current_user.current_workspace_id if knowledge.parent_id is None: knowledge.parent_id = knowledge.workspace_id + + workspace = db.query(Workspace).filter(Workspace.id == knowledge.workspace_id).first() + if not workspace: + raise Exception(f"Workspace {knowledge.workspace_id} not found") + + tenant_id = workspace.tenant_id + + if not knowledge.embedding_id: + embedding_models = ModelConfigRepository.get_by_type( + db=db, model_types=[ModelType.EMBEDDING], tenant_id=tenant_id, is_active=True + ) + if embedding_models: + knowledge.embedding_id = embedding_models[0].id + business_logger.debug(f"Auto-bind embedding model: {embedding_models[0].id}") + + if not knowledge.reranker_id: + rerank_models = ModelConfigRepository.get_by_type( + db=db, model_types=[ModelType.RERANK], tenant_id=tenant_id, is_active=True + ) + if rerank_models: + knowledge.reranker_id = rerank_models[0].id + business_logger.debug(f"Auto-bind rerank model: {rerank_models[0].id}") + + if not knowledge.llm_id: + llm_models = ModelConfigRepository.get_by_type( + db=db, model_types=[ModelType.LLM, ModelType.CHAT], tenant_id=tenant_id, is_active=True + ) + if llm_models: + knowledge.llm_id = llm_models[0].id + business_logger.debug(f"Auto-bind llm model: {llm_models[0].id}") + + if not knowledge.image2text_id: + image2text_models = db.query(ModelConfig).filter( + ModelConfig.tenant_id == tenant_id, + ModelConfig.type.in_([ModelType.CHAT.value, ModelType.IMAGE.value]), + ModelConfig.capability.contains(["vision"]), + ModelConfig.is_active == True, + ModelConfig.is_composite == False + ).order_by(ModelConfig.created_at.desc()).all() + if not image2text_models: + raise Exception("租户下没有可用的视觉模型，创建知识库失败") + knowledge.image2text_id = image2text_models[0].id + business_logger.debug(f"Auto-bind image2text model: {image2text_models[0].id}") + business_logger.debug(f"Start creating the knowledge base: {knowledge.name}") db_knowledge = knowledge_repository.create_knowledge( db=db, knowledge=knowledge From e015455fb89c2443d3685168ff3752e52bea1382 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Thu, 16 Apr 2026 19:00:58 +0800 Subject: [PATCH 083/245] feat(web): model support json --- web/src/components/ModelSelect/index.tsx | 24 +++++++------- web/src/i18n/en.ts | 3 ++ web/src/i18n/zh.ts | 3 ++ .../components/ModelConfigModal.tsx | 22 +++++++------ web/src/views/ApplicationConfig/types.ts | 15 +++++---- .../components/CustomModelModal.tsx | 17 +++++++--- web/src/views/ModelManagement/types.ts | 5 +-- .../Properties/ModelConfig/index.tsx | 32 ++++++++++++++++--- web/src/views/Workflow/constant.ts | 6 +++- 9 files changed, 88 insertions(+), 39 deletions(-) diff --git a/web/src/components/ModelSelect/index.tsx b/web/src/components/ModelSelect/index.tsx index 8f9152fb..0b4f671e 100644 --- a/web/src/components/ModelSelect/index.tsx +++ b/web/src/components/ModelSelect/index.tsx @@ -2,9 +2,9 @@ * @Author: ZhaoYing * @Date: 2026-03-07 16:49:59 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-25 11:21:59 + * @Last Modified time: 2026-04-16 17:46:02 */ -import { useEffect, useState, type FC } from 'react'; +import { useEffect, useState, forwardRef, useImperativeHandle } from 'react'; import { Select, Flex, Space } from 'antd'; import type { SelectProps } from 'antd/es/select'; import { useTranslation } from 'react-i18next'; @@ -14,6 +14,10 @@ import type { Query, Model } from '@/views/ModelManagement/types'; import { getListLogoUrl } from '@/views/ModelManagement/utils'; import Tag from '@/components/Tag'; +export interface ModelSelectRef { + options: Model[]; +} + /** Extends AntD SelectProps; omits filterOption since it's handled internally */ interface ModelSelectProps extends SelectProps { /** Extra query params passed to getModelList */ @@ -24,17 +28,15 @@ interface ModelSelectProps extends SelectProps { initialData?: Model[]; } -const ModelSelect: FC = ({ - params, - placeholder, - fontClassName, - isAutoFetch = true, - initialData = [], - ...props -}) => { +const ModelSelect = forwardRef(( + { params, placeholder, fontClassName, isAutoFetch = true, initialData = [], ...props }, + ref +) => { const { t } = useTranslation(); const [options, setOptions] = useState([]); + useImperativeHandle(ref, () => ({ options }), [options]); + // Fetch active models whenever params change; stringify for stable deep comparison useEffect(() => { if (!isAutoFetch) return @@ -89,6 +91,6 @@ const ModelSelect: FC = ({ {...props} /> ); -}; +}); export default ModelSelect; diff --git a/web/src/i18n/en.ts b/web/src/i18n/en.ts index 394606a3..d65e4436 100644 --- a/web/src/i18n/en.ts +++ b/web/src/i18n/en.ts @@ -629,6 +629,7 @@ export const en = { video: 'Video', thinking: 'Deep Thinking', is_thinking: 'Deep Thinking Support', + json_output: 'Support JSON formatted output', }, knowledgeBase: { home: 'Home', @@ -1524,6 +1525,7 @@ export const en = { }`, uploadCover: 'Import and Overwrite', refresh: 'Refresh Current Page', + json_output: 'Support JSON formatted output', }, userMemory: { userMemory: 'User Memory', @@ -2287,6 +2289,7 @@ Memory Bear: After the rebellion, regional warlordism intensified for several re messagesPlaceholder: 'Write prompts here, type "{" to insert variables, type "insert" to insert', vision: 'Vision', parameterSettings: 'Parameter Settings', + json_output: 'Support JSON formatted output', }, start: { variables: 'Input Fields', diff --git a/web/src/i18n/zh.ts b/web/src/i18n/zh.ts index 98676ef4..921b3554 100644 --- a/web/src/i18n/zh.ts +++ b/web/src/i18n/zh.ts @@ -859,6 +859,7 @@ export const zh = { }`, uploadCover: '导入并覆盖', refresh: '刷新当前页', + json_output: '支持JSON格式化输出', }, table: { totalRecords: '共 {{total}} 条记录' @@ -1307,6 +1308,7 @@ export const zh = { video: '视频', thinking: '深度思考', is_thinking: '支持深度思考', + json_output: '支持JSON格式化输出', }, timezones: { 'Asia/Shanghai': '中国标准时间 (UTC+8)', @@ -2248,6 +2250,7 @@ export const zh = { messagesPlaceholder: '在此处编写提示，输入“{”插入变量，输入“insert”插入', vision: '视觉', parameterSettings: '参数设置', + json_output: '支持JSON格式化输出', }, start: { variables: '输入字段', diff --git a/web/src/views/ApplicationConfig/components/ModelConfigModal.tsx b/web/src/views/ApplicationConfig/components/ModelConfigModal.tsx index 8e3e3257..30af7a8c 100644 --- a/web/src/views/ApplicationConfig/components/ModelConfigModal.tsx +++ b/web/src/views/ApplicationConfig/components/ModelConfigModal.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 16:28:07 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-31 16:56:57 + * @Last Modified time: 2026-04-16 18:51:01 */ /** * Model Configuration Modal @@ -102,14 +102,15 @@ const ModelConfigModal = forwardRef( } /** Handle model selection change */ const handleChange: SelectProps['onChange'] = (_value, option) => { - if (source === 'chat') { - form.setFieldValue('label', (option as Model).name) - } - - form.setFieldsValue({ + const newValues: ModelConfig = { capability: (option as Model).capability, deep_thinking: false, - }) + json_output: false + } + if (source === 'chat') { + newValues.label = (option as Model).name + } + form.setFieldsValue(newValues) } /** Expose methods to parent component */ @@ -119,12 +120,10 @@ const ModelConfigModal = forwardRef( })); useEffect(() => { - const { deep_thinking: _, ...rest } = data?.model_parameters || {} + const { deep_thinking: _, json_output: __, ...rest } = data?.model_parameters || {} form.setFieldsValue(rest) }, [values?.default_model_config_id]) - - console.log('handleChange values', values) return ( ( + {source === 'chat' &&
{t('userMemory.role')}
-
{data?.profile?.role || '-'}
+
{data?.profile?.role?.join(' | ') || '-'}
{t('userMemory.domain')}
-
{data?.profile?.domain || '-'}
+
{data?.profile?.domain?.join(' | ') || '-'}
{t('userMemory.expertise')}
diff --git a/web/src/views/UserMemoryDetail/types.ts b/web/src/views/UserMemoryDetail/types.ts index 4eb4fe5b..667d8272 100644 --- a/web/src/views/UserMemoryDetail/types.ts +++ b/web/src/views/UserMemoryDetail/types.ts @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 17:57:15 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-14 16:03:16 + * @Last Modified time: 2026-04-17 17:57:00 */ /** * User Memory Detail Types @@ -178,8 +178,8 @@ export interface EndUser { created_at: string; updated_at: string; profile: { - role: string; - domain: string; + role: string[]; + domain: string[]; expertise: string[]; interests: string[]; }; From 48b56a3d88d3a76623cd54eca8e550134e4fc735 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Fri, 17 Apr 2026 19:58:44 +0800 Subject: [PATCH 113/245] fix(web): update interface type --- .../Workflow/components/Properties/CaseList/index.tsx | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/web/src/views/Workflow/components/Properties/CaseList/index.tsx b/web/src/views/Workflow/components/Properties/CaseList/index.tsx index 227b6ffb..6eaa08e1 100644 --- a/web/src/views/Workflow/components/Properties/CaseList/index.tsx +++ b/web/src/views/Workflow/components/Properties/CaseList/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-09 18:24:53 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-17 19:48:59 + * @Last Modified time: 2026-04-17 19:58:13 */ import { useEffect, useMemo, type FC } from 'react' import clsx from 'clsx' @@ -45,7 +45,7 @@ interface CaseItem { } interface CaseListProps { value?: CaseItem[]; - onChange?: (value: Array<{ logical_operator: 'and' | 'or'; expressions: { left: string; operator: string; right: string; }[] }>) => void; + onChange?: (value: CaseItem[]) => void; options: Suggestion[]; name: string; selectedNode?: any; @@ -221,7 +221,7 @@ const ArrayFileSubConditions: FC = ({ conditionFiel onChange={(value: string) => { form.setFieldValue([name, caseIndex, 'expressions', conditionIndex, 'sub_variable_condition', 'conditions', subIndex], { key: value, - input_type: value === 'size' ? 'Constant' : undefined, + input_type: value === 'size' ? 'constant' : undefined, value: undefined, operator: value === 'size' ? 'ge' : 'eq', }); @@ -488,7 +488,7 @@ const CaseList: FC = ({ left: newValue, operator: undefined, right: undefined, - input_type: 'Constant' + input_type: 'constant' }); }; From 7b5d7696cb6c4d55e4f833fcd9f547ffd2e67324 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Fri, 17 Apr 2026 20:26:44 +0800 Subject: [PATCH 114/245] feat(workflow): support variable input type in if-else node conditions --- api/app/core/workflow/nodes/if_else/config.py | 14 +++++- api/app/core/workflow/nodes/if_else/node.py | 2 +- api/app/core/workflow/nodes/operators.py | 50 ++++++++++--------- 3 files changed, 39 insertions(+), 27 deletions(-) diff --git a/api/app/core/workflow/nodes/if_else/config.py b/api/app/core/workflow/nodes/if_else/config.py index 302d469f..4a5b3860 100644 --- a/api/app/core/workflow/nodes/if_else/config.py +++ b/api/app/core/workflow/nodes/if_else/config.py @@ -10,8 +10,18 @@ class SubVariableConditionItem(BaseModel): """A single condition on a file object's field, used inside sub_variable_condition.""" key: str = Field(..., description="Field name of the file object, e.g. type, size, name") operator: ComparisonOperator = Field(..., description="Comparison operator") - value: Any = Field(default=None, description="Value to compare with") - var_type: str = Field(default="string", description="Field value type: string or number") + value: Any = Field(default=None, description="Value to compare with, or variable selector when input_type=variable") + input_type: ValueInputType = Field(default=ValueInputType.CONSTANT, description="constant or variable") + + @field_validator("input_type", mode="before") + @classmethod + def lower_input_type(cls, v): + if isinstance(v, str): + try: + return ValueInputType(v.lower()) + except ValueError: + raise ValueError(f"Invalid input_type: {v}") + return v class SubVariableCondition(BaseModel): diff --git a/api/app/core/workflow/nodes/if_else/node.py b/api/app/core/workflow/nodes/if_else/node.py index faecd87c..c4d3a0e6 100644 --- a/api/app/core/workflow/nodes/if_else/node.py +++ b/api/app/core/workflow/nodes/if_else/node.py @@ -103,7 +103,7 @@ class IfElseNode(BaseNode): left_value = None if expression.sub_variable_condition is not None and isinstance(left_value, list): - evaluator = ArrayFileContainsOperator(left_value, expression.sub_variable_condition) + evaluator = ArrayFileContainsOperator(left_value, expression.sub_variable_condition, variable_pool) else: evaluator = ConditionExpressionResolver.resolve_by_value(left_value)( variable_pool, diff --git a/api/app/core/workflow/nodes/operators.py b/api/app/core/workflow/nodes/operators.py index 30634bbe..62eebbfe 100644 --- a/api/app/core/workflow/nodes/operators.py +++ b/api/app/core/workflow/nodes/operators.py @@ -396,47 +396,49 @@ class NoneObjectComparisonOperator: class ArrayFileContainsOperator: - """Handles contains/not_contains on array[file] with sub_variable_condition. + """Handles contains/not_contains on array[file] with sub_variable_condition.""" - Evaluates whether any (contains) or no (not_contains) file element - in the array satisfies all sub-conditions. - """ - - def __init__(self, left_value: list[dict], sub_variable_condition: Any): + def __init__(self, left_value: list[dict], sub_variable_condition: Any, pool: VariablePool | None = None): self.left_value = left_value self.sub_variable_condition = sub_variable_condition + self.pool = pool + + def _resolve_value(self, cond: Any) -> Any: + if cond.input_type == ValueInputType.VARIABLE and self.pool is not None: + pattern = r"\{\{\s*(.*?)\s*\}\}" + selector = re.sub(pattern, r"\1", str(cond.value)).strip() + return self.pool.get_value(selector, default=None, strict=False) + return cond.value def _match_item(self, file_item: dict) -> bool: - """Check if a single file dict satisfies all sub-conditions.""" results = [] for cond in self.sub_variable_condition.conditions: field_val = file_item.get(cond.key) - result = self._eval_sub(field_val, cond) + expected = self._resolve_value(cond) + result = self._eval_sub(field_val, cond.operator.value, expected) results.append(result) if self.sub_variable_condition.logical_operator.value == "and": return all(results) return any(results) @staticmethod - def _eval_sub(field_val: Any, cond: Any) -> bool: - op = cond.operator.value - expected = cond.value + def _eval_sub(field_val: Any, op: str, expected: Any) -> bool: if field_val is None: - return op in ("empty", "not_empty") and op == "empty" + return op == "empty" match op: - case "eq": return str(field_val) == str(expected) - case "ne": return str(field_val) != str(expected) - case "contains": return isinstance(field_val, str) and str(expected) in field_val + case "eq": return str(field_val) == str(expected) + case "ne": return str(field_val) != str(expected) + case "contains": return isinstance(field_val, str) and str(expected) in field_val case "not_contains": return isinstance(field_val, str) and str(expected) not in field_val - case "in": return field_val in (expected if isinstance(expected, list) else [expected]) - case "not_in": return field_val not in (expected if isinstance(expected, list) else [expected]) - case "gt": return isinstance(field_val, (int, float)) and field_val > float(expected) - case "ge": return isinstance(field_val, (int, float)) and field_val >= float(expected) - case "lt": return isinstance(field_val, (int, float)) and field_val < float(expected) - case "le": return isinstance(field_val, (int, float)) and field_val <= float(expected) - case "empty": return field_val in (None, "", 0) - case "not_empty": return field_val not in (None, "", 0) - case _: return False + case "in": return field_val in (expected if isinstance(expected, list) else [expected]) + case "not_in": return field_val not in (expected if isinstance(expected, list) else [expected]) + case "gt": return isinstance(field_val, (int, float)) and field_val > float(expected) + case "ge": return isinstance(field_val, (int, float)) and field_val >= float(expected) + case "lt": return isinstance(field_val, (int, float)) and field_val < float(expected) + case "le": return isinstance(field_val, (int, float)) and field_val <= float(expected) + case "empty": return field_val in (None, "", 0) + case "not_empty": return field_val not in (None, "", 0) + case _: return False def contains(self) -> bool: return any(self._match_item(f) for f in self.left_value if isinstance(f, dict)) From bbd85733b8b24a9abc00aa2ace9745ba10f21da3 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Fri, 17 Apr 2026 20:41:23 +0800 Subject: [PATCH 115/245] fix(web): if-else port position --- .../Workflow/components/Nodes/ConditionNode.tsx | 2 +- web/src/views/Workflow/utils.ts | 16 +++++++++++----- 2 files changed, 12 insertions(+), 6 deletions(-) diff --git a/web/src/views/Workflow/components/Nodes/ConditionNode.tsx b/web/src/views/Workflow/components/Nodes/ConditionNode.tsx index 5e62775d..9cd1309e 100644 --- a/web/src/views/Workflow/components/Nodes/ConditionNode.tsx +++ b/web/src/views/Workflow/components/Nodes/ConditionNode.tsx @@ -109,7 +109,7 @@ const ConditionNode: ReactShapeConfig['component'] = ({ node }) => { } {expression.sub_variable_condition?.conditions?.length > 0 && expression.sub_variable_condition?.conditions.every(isSubExprSet) - ?
+ ?
{expression.sub_variable_condition?.conditions.map((sub: any, sIndex: number) => (
{expression.sub_variable_condition?.conditions.length > 1 && sIndex > 0 &&
{expression.sub_variable_condition?.logical_operator?.toLocaleUpperCase()}
} diff --git a/web/src/views/Workflow/utils.ts b/web/src/views/Workflow/utils.ts index 429be70d..bd81b6eb 100644 --- a/web/src/views/Workflow/utils.ts +++ b/web/src/views/Workflow/utils.ts @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-03-24 15:07:49 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-17 19:13:22 + * @Last Modified time: 2026-04-17 20:40:47 */ import { portItemArgsY, conditionNodePortItemArgsY, conditionNodeHeight } from './constant' @@ -91,6 +91,7 @@ export const getConditionNodeCasePortY = (cases: any[], caseIndex: number) => { let portItemArgsYNum = 0; for (let i = 0; i < caseIndex; i++) { + const notHasSub = cases[i]?.expressions?.filter((e: any) => !e?.sub_variable_condition?.conditions || e?.sub_variable_condition?.conditions.length <1).length const n = cases[i]?.expressions?.length || 0; let casePortItemArgsYNum = n + 1; // Add extra y for expressions with all sub_variable_condition set @@ -110,11 +111,16 @@ export const getConditionNodeCasePortY = (cases: any[], caseIndex: number) => { cases[i]?.expressions?.forEach((e: any) => { const subs = e?.sub_variable_condition?.conditions; if (subs?.length && subs.every(isSubExprSet) && subs.length > 1) { - extraExprs += subs.length; - } else if (!subs && n > 2) { - extraExprs += n - 2; - } + extraExprs += subs.length + 2; + } }); + + console.log('extraExprs notHasSub', notHasSub) + if (notHasSub > 3) { + extraExprs += n - 2 + notHasSub/4; + } else { + extraExprs += n - 2 + notHasSub/4 + } } } From 86540a81d1779f622101bd4b837e2949656ab716 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Fri, 17 Apr 2026 20:46:03 +0800 Subject: [PATCH 116/245] fix(web): SubCondition interface --- .../views/Workflow/components/Properties/CaseList/index.tsx | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/web/src/views/Workflow/components/Properties/CaseList/index.tsx b/web/src/views/Workflow/components/Properties/CaseList/index.tsx index 6eaa08e1..bdf516bc 100644 --- a/web/src/views/Workflow/components/Properties/CaseList/index.tsx +++ b/web/src/views/Workflow/components/Properties/CaseList/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-09 18:24:53 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-17 19:58:13 + * @Last Modified time: 2026-04-17 20:45:31 */ import { useEffect, useMemo, type FC } from 'react' import clsx from 'clsx' @@ -23,7 +23,7 @@ interface SubCondition { key: string; operator: string; value: string | number; - var_type: string; + input_type: string; } interface SubVariableCondition { @@ -303,7 +303,7 @@ const ArrayFileSubConditions: FC = ({ conditionFiel })}
} + onClick={handleKnowledgeConfig} + >{t('application.globalConfig')} + + + } + headerType="borderless" + headerClassName="rb:h-11.5! rb:py-3! rb:leading-5.5!" + titleClassName="rb:font-[MiSans-Bold] rb:font-bold" + > +
+ {t('application.associatedKnowledgeBase')} +
+ {knowledgeList.length === 0 + ?
+ +
+ : {knowledgeItems} + } + {modals} + + ) + } + + return ( +
+ +
+ * + {t('application.knowledgeBaseAssociation')} +
+
} + className="rb:py-0! rb:px-1! rb:text-[12px]! rb:group rb:gap-0.5!" + size="small" + disabled={knowledgeList.length === 0} + > + {t('application.globalConfig')} + + + + + {knowledgeList.length > 0 && knowledgeItems} + + {modals} +
+ ) +} + +export default Knowledge diff --git a/web/src/components/Knowledge/KnowledgeConfigModal.tsx b/web/src/components/Knowledge/KnowledgeConfigModal.tsx new file mode 100644 index 00000000..c91230ee --- /dev/null +++ b/web/src/components/Knowledge/KnowledgeConfigModal.tsx @@ -0,0 +1,124 @@ +import { forwardRef, useEffect, useImperativeHandle, useState } from 'react'; +import { Form, Select, InputNumber, Flex } from 'antd'; +import { useTranslation } from 'react-i18next'; + +import type { KnowledgeConfigModalRef, KnowledgeBase, KnowledgeConfigForm, RetrieveType } from './types' +import RbModal from '@/components/RbModal' +import RbSlider from '@/components/RbSlider' +import { formatDateTime } from '@/utils/format'; + +const FormItem = Form.Item; + +interface KnowledgeConfigModalProps { + refresh: (values: KnowledgeConfigForm, type: 'knowledgeConfig') => void; +} +const retrieveTypes: RetrieveType[] = ['participle', 'semantic', 'hybrid', 'graph'] + +const KnowledgeConfigModal = forwardRef(({ refresh }, ref) => { + const { t } = useTranslation(); + const [visible, setVisible] = useState(false); + const [form] = Form.useForm(); + const [data, setData] = useState(null); + const values = Form.useWatch([], form); + + const handleClose = () => { + setVisible(false); + form.resetFields(); + setData(null) + }; + + const handleOpen = (data: KnowledgeBase) => { + form.setFieldsValue({ + retrieve_type: data?.config?.retrieve_type || retrieveTypes[0], + kb_id: data.id, + top_k: data?.config?.top_k || 5, + similarity_threshold: data?.config?.similarity_threshold || 0.5, + vector_similarity_weight: data?.config?.vector_similarity_weight || 0.5, + ...(data || {}), + ...(data?.config || {}), + }) + setData({...data}) + setVisible(true); + }; + + const handleSave = () => { + form.validateFields() + .then(() => { + refresh(values, 'knowledgeConfig') + handleClose() + }) + .catch((err) => console.log('err', err)); + } + + useImperativeHandle(ref, () => ({ handleOpen, handleClose })); + + useEffect(() => { + if (values?.retrieve_type) { + const fieldsToReset = Object.keys(values).filter(key => + key !== 'kb_id' && key !== 'retrieve_type' && key !== 'top_k' + ) as (keyof KnowledgeConfigForm)[]; + form.resetFields(fieldsToReset); + } + }, [values?.retrieve_type]) + + return ( + +
+ {data && ( + +
+ {data.name} +
{t('application.contains', {include_count: data.doc_num})}
+
+
{formatDateTime(data.updated_at, 'YYYY-MM-DD HH:mm:ss')}
+ + )} +
- :
+ :
{title}
} diff --git a/web/src/components/Tag/index.tsx b/web/src/components/Tag/index.tsx index d49c46ae..e7307843 100644 --- a/web/src/components/Tag/index.tsx +++ b/web/src/components/Tag/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-02 15:29:57 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-22 11:39:15 + * @Last Modified time: 2026-04-22 13:48:09 */ /** * Tag Component @@ -40,7 +40,7 @@ const colors = { /** Custom tag component with color themes */ const Tag: FC = ({ color = 'processing', children, className, variant = 'outline' }) => { return ( - + {children} ) diff --git a/web/src/main.tsx b/web/src/main.tsx index 3e86feae..d4b69f7a 100644 --- a/web/src/main.tsx +++ b/web/src/main.tsx @@ -11,6 +11,15 @@ import App from '@/App.tsx' // Synchronously import i18n config to ensure initialization before component rendering import './i18n' +// Fix autofill background color on focus +document.addEventListener('animationstart', (e) => { + if (e.animationName === 'onAutoFillStart') { + const input = e.target as HTMLInputElement + input.style.backgroundColor = 'transparent' + input.addEventListener('focus', () => { input.style.backgroundColor = 'transparent' }, { once: false }) + } +}) + // After a new release, old dynamic chunk files are deleted; force a page reload on preload error window.addEventListener('vite:preloadError', () => { console.warn('New version detected, reloading page to load latest assets...') diff --git a/web/src/styles/index.css b/web/src/styles/index.css index 13904435..0bbacb51 100644 --- a/web/src/styles/index.css +++ b/web/src/styles/index.css @@ -457,4 +457,14 @@ body { .pageTabs.ant-segmented .ant-segmented-item-selected { box-shadow: 0px 2px 4px 0px rgba(33, 35, 50, 0.16); -} \ No newline at end of file +} +input:-webkit-autofill, +input:-webkit-autofill:hover, +input:-webkit-autofill:focus, +input:-webkit-autofill:active { + -webkit-box-shadow: 0 0 0 1000px transparent inset !important; + transition: background-color 5000s ease-in-out 0s !important; + animation-name: onAutoFillStart; + animation-duration: 1ms; +} +@keyframes onAutoFillStart { from {} to {} } \ No newline at end of file diff --git a/web/src/views/ApiKeyManagement/index.tsx b/web/src/views/ApiKeyManagement/index.tsx index 211c7745..b14899f5 100644 --- a/web/src/views/ApiKeyManagement/index.tsx +++ b/web/src/views/ApiKeyManagement/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 15:52:50 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-22 11:34:06 + * @Last Modified time: 2026-04-22 12:07:40 */ import React, { useRef } from 'react'; import { useTranslation } from 'react-i18next'; @@ -95,44 +95,39 @@ const ApiKeyManagement: React.FC = () => { renderItem={(apiKeyItem) => { return ( - - {apiKeyItem.name} - - {apiKeyItem.scopes?.includes('memory') && {t('apiKey.memoryEngine')}} - {apiKeyItem.scopes?.includes('rag') && {t('apiKey.knowledgeBase')}} - {!apiKeyItem.scopes?.includes('memory') && !apiKeyItem.scopes?.includes('rag') &&
{t('apiKey.noScopes')}
} - - - , - label: t('common.edit'), - onClick: () => handleEdit(apiKeyItem), - }, - { - key: 'view', - icon:
, - label: t('common.view'), - onClick: () => handleView(apiKeyItem), - }, - { - key: 'delete', - danger: true, - icon:
, - label: t('common.delete'), - onClick: () => handleDelete(apiKeyItem), - }, - ]} - /> - - } - isNeedTooltip={false} - headerClassName="rb:min-h-[78px]!" + title={apiKeyItem.name} + extra={, + label: t('common.edit'), + onClick: () => handleEdit(apiKeyItem), + }, + { + key: 'view', + icon:
, + label: t('common.view'), + onClick: () => handleView(apiKeyItem), + }, + { + key: 'delete', + danger: true, + icon:
, + label: t('common.delete'), + onClick: () => handleDelete(apiKeyItem), + }, + ]} + />} + variant="borderless" + headerClassName="rb:min-h-[42px]!" + titleClassName="rb:line-clamp-1!" > + + {apiKeyItem.scopes?.includes('memory') && {t('apiKey.memoryEngine')}} + {apiKeyItem.scopes?.includes('rag') && {t('apiKey.knowledgeBase')}} + {!apiKeyItem.scopes?.includes('memory') && !apiKeyItem.scopes?.includes('rag') &&
{t('apiKey.noScopes')}
} + ({ key, From b4885905374b1dde1a552de4e0dcb92f37295e86 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 22 Apr 2026 14:34:09 +0800 Subject: [PATCH 178/245] fix(web): agent deep thinking loading --- .../ApplicationConfig/components/Chat.tsx | 52 +++++++++---------- 1 file changed, 26 insertions(+), 26 deletions(-) diff --git a/web/src/views/ApplicationConfig/components/Chat.tsx b/web/src/views/ApplicationConfig/components/Chat.tsx index eb3a9ea0..8961ac79 100644 --- a/web/src/views/ApplicationConfig/components/Chat.tsx +++ b/web/src/views/ApplicationConfig/components/Chat.tsx @@ -68,14 +68,14 @@ const Chat: FC = ({ const [loading, setLoading] = useState(false) const [isCluster, setIsCluster] = useState(source === 'multi_agent') const [conversationId, setConversationId] = useState(null) - const [compareLoading, setCompareLoading] = useState(false) + const compareLoadingRef = useRef(false) const [fileList, setFileList] = useState([]) const [message, setMessage] = useState(undefined) const [features, setFeatures] = useState({} as FeaturesConfigForm) const [audioStatusMap, setAudioStatusMap] = useState>({}) useEffect(() => { - setCompareLoading(false) + compareLoadingRef.current = false setLoading(false) return () => { audioPollingRef.current.forEach(timer => clearInterval(timer)) @@ -254,7 +254,7 @@ const Chat: FC = ({ const handleSend = (msg?: string) => { if (loading || !id) return setLoading(true) - setCompareLoading(true) + compareLoadingRef.current = true const files = (fileList || []).filter(item => !['uploading', 'error'].includes(item.status)) handleSave(false) .then(() => { @@ -280,7 +280,7 @@ const Chat: FC = ({ } if (!isCanSend) { setLoading(false) - setCompareLoading(false) + compareLoadingRef.current = false return } @@ -305,20 +305,20 @@ const Chat: FC = ({ switch (item.event) { case 'model_reasoning': - if (compareLoading) { - setCompareLoading(false) + if (compareLoadingRef.current) { + compareLoadingRef.current = false } updateAssistantReasoningMessage(content, model_config_id, conversation_id) break; case 'model_message': - if (compareLoading) { - setCompareLoading(false) + if (compareLoadingRef.current) { + compareLoadingRef.current = false } updateAssistantMessage(content, model_config_id, conversation_id, audio_url) break; case 'model_end': - if (compareLoading) { - setCompareLoading(false) + if (compareLoadingRef.current) { + compareLoadingRef.current = false } const idToPoll = `${model_config_id}_${audio_url}` if (audio_url && !audioStatusMap[idToPoll]) { @@ -360,8 +360,8 @@ const Chat: FC = ({ updateErrorAssistantMessage(message_length, model_config_id) break; case 'compare_end': - if (compareLoading) { - setCompareLoading(false) + if (compareLoadingRef.current) { + compareLoadingRef.current = false } setLoading(false); break; @@ -396,18 +396,18 @@ const Chat: FC = ({ }, handleStreamMessage) .catch(() => { setLoading(false) - setCompareLoading(false) + compareLoadingRef.current = false updateClusterErrorAssistantMessage(0) }) .finally(() => { setLoading(false) - setCompareLoading(false) + compareLoadingRef.current = false }) }, 0) }) .catch(() => { setLoading(false) - setCompareLoading(false) + compareLoadingRef.current = false }) } @@ -471,7 +471,7 @@ const Chat: FC = ({ const handleClusterSend = (msg?: string) => { if (loading || !id) return setLoading(true) - setCompareLoading(true) + compareLoadingRef.current = true const files = (fileList || []).filter(item => !['uploading', 'error'].includes(item.status)) handleSave(false) .then(() => { @@ -495,8 +495,8 @@ const Chat: FC = ({ } break case 'message': - if (compareLoading) { - setCompareLoading(false) + if (compareLoadingRef.current) { + compareLoadingRef.current = false } updateClusterAssistantMessage(content) if (conversation_id && conversationId !== conversation_id) { @@ -504,14 +504,14 @@ const Chat: FC = ({ } break; case 'model_end': - if (compareLoading) { - setCompareLoading(false) + if (compareLoadingRef.current) { + compareLoadingRef.current = false } updateClusterErrorAssistantMessage(message_length) break; case 'compare_end': - if (compareLoading) { - setCompareLoading(false) + if (compareLoadingRef.current) { + compareLoadingRef.current = false } setLoading(false); break; @@ -541,18 +541,18 @@ const Chat: FC = ({ ) .catch(() => { setLoading(false) - setCompareLoading(false) + compareLoadingRef.current = false updateClusterErrorAssistantMessage(0) }) .finally(() => { setLoading(false) - setCompareLoading(false) + compareLoadingRef.current = false }) }, 0) }) .catch(() => { setLoading(false) - setCompareLoading(false) + compareLoadingRef.current = false }) } @@ -622,7 +622,7 @@ const Chat: FC = ({ />} onSend={isCluster ? handleClusterSend : handleSend} data={chat.list || []} - streamLoading={compareLoading} + streamLoading={compareLoadingRef.current} labelPosition="top" labelFormat={(item) => item.role === 'user' ? t('application.you') : chat.label || t(`application.ai`)} errorDesc={t('application.ReplyException')} From 7cd76444f142ed23bba0783989c16add619bf7ed Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 22 Apr 2026 14:38:18 +0800 Subject: [PATCH 179/245] fix(web): ui --- web/src/components/Knowledge/Knowledge.tsx | 4 ++-- .../views/ApplicationConfig/components/ToolList/ToolList.tsx | 2 +- .../components/VariableList/VariableList.tsx | 2 +- web/src/views/Skills/components/ToolList/ToolList.tsx | 2 +- 4 files changed, 5 insertions(+), 5 deletions(-) diff --git a/web/src/components/Knowledge/Knowledge.tsx b/web/src/components/Knowledge/Knowledge.tsx index 4721883c..b1c9b78f 100644 --- a/web/src/components/Knowledge/Knowledge.tsx +++ b/web/src/components/Knowledge/Knowledge.tsx @@ -161,11 +161,11 @@ const Knowledge: FC = ({ value = { knowledge_bases: [] }, onChan extra={
} onClick={handleKnowledgeConfig} >{t('application.globalConfig')} - + } headerType="borderless" diff --git a/web/src/views/ApplicationConfig/components/ToolList/ToolList.tsx b/web/src/views/ApplicationConfig/components/ToolList/ToolList.tsx index 0feb814b..b9981432 100644 --- a/web/src/views/ApplicationConfig/components/ToolList/ToolList.tsx +++ b/web/src/views/ApplicationConfig/components/ToolList/ToolList.tsx @@ -137,7 +137,7 @@ const ToolList: FC<{ value?: ToolOption[]; onChange?: (config: ToolOption[]) => + {t('application.addTool')} + } >
diff --git a/web/src/views/ApplicationConfig/components/VariableList/VariableList.tsx b/web/src/views/ApplicationConfig/components/VariableList/VariableList.tsx index 9f7f71f5..432d0e3d 100644 --- a/web/src/views/ApplicationConfig/components/VariableList/VariableList.tsx +++ b/web/src/views/ApplicationConfig/components/VariableList/VariableList.tsx @@ -60,7 +60,7 @@ interface VariableListProps { extra={ + } > {toolList.length === 0 From 881d74d29d8c9f49fca6e5464554e594c6405762 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 22 Apr 2026 14:44:02 +0800 Subject: [PATCH 180/245] feat(web): skill keywords not required --- web/src/views/Skills/pages/SkillConfig.tsx | 1 - 1 file changed, 1 deletion(-) diff --git a/web/src/views/Skills/pages/SkillConfig.tsx b/web/src/views/Skills/pages/SkillConfig.tsx index 440356fd..47082538 100644 --- a/web/src/views/Skills/pages/SkillConfig.tsx +++ b/web/src/views/Skills/pages/SkillConfig.tsx @@ -182,7 +182,6 @@ const SkillConfig: FC = () => { ({ + value: type, label: t(`explicitDetail.${type}`) + })) + ]} + placeholder={t('explicitDetail.episodic_type')} + className="rb:w-full!" + /> + + + + + {episodicLoading ? + + : ( + <> + PAGE_SIZE, + 'rb:max-h-[calc(100%-36px)]!': total <= PAGE_SIZE && episodicMemories.length > 0, + 'rb:h-full!': episodicMemories.length === 0 + })}> + {episodicMemories.length > 0 ? episodicMemories.map(item => ( +
handleView(item)} + > + + {item.title} +
{formatDateTime(item.created_at)}
+ +
{item.content}
+
+ )) : } + + {total > PAGE_SIZE && ( + + )} + + ) + } + @@ -229,7 +301,7 @@ const ExplicitDetail: FC = () => { > {loading ? - : data.semantic_memories?.length > 0 + : semanticsMemory?.length > 0 ?
: } From ae770fb13162e965cd2a44ba18fbbdabc6b283bc Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 22 Apr 2026 17:34:32 +0800 Subject: [PATCH 184/245] fix(web): move EpisodicMemoryType type --- web/src/api/memory.ts | 3 ++- web/src/views/UserMemoryDetail/pages/ExplicitDetail.tsx | 4 ++-- 2 files changed, 4 insertions(+), 3 deletions(-) diff --git a/web/src/api/memory.ts b/web/src/api/memory.ts index c5120cb2..90c4e13f 100644 --- a/web/src/api/memory.ts +++ b/web/src/api/memory.ts @@ -21,7 +21,6 @@ import type { import type { ConfigForm as SelfReflectionEngineConfig } from '@/views/SelfReflectionEngine/types' -import type { EpisodicMemoryType } from '@/views/UserMemoryDetail/pages/ExplicitDetail' import type { TestParams } from '@/views/MemoryConversation' import type { EndUser } from '@/views/UserMemoryDetail/types' import { handleSSE, type SSEMessage } from '@/utils/stream' @@ -219,6 +218,8 @@ export const getTimelineMemories = (data: { id: string; label: string; }) => { export const getExplicitMemory = (end_user_id: string) => { return request.post(`/memory/explicit-memory/overview`, { end_user_id }) } + +export type EpisodicMemoryType = "conversation" | "project_work" | "learning" | "decision" | "important_event" export interface EpisodicMemoryQuery { end_user_id?: string; page?: number; diff --git a/web/src/views/UserMemoryDetail/pages/ExplicitDetail.tsx b/web/src/views/UserMemoryDetail/pages/ExplicitDetail.tsx index a8ea9368..7ba161cd 100644 --- a/web/src/views/UserMemoryDetail/pages/ExplicitDetail.tsx +++ b/web/src/views/UserMemoryDetail/pages/ExplicitDetail.tsx @@ -18,6 +18,7 @@ import { getSemanticsMemory, getEpisodicMemory, type EpisodicMemoryQuery, + type EpisodicMemoryType, } from '@/api/memory' import { formatDateTime } from '@/utils/format' import Empty from '@/components/Empty' @@ -25,7 +26,6 @@ import ExplicitDetailModal from '../components/ExplicitDetailModal' /** An episodic (event-based) memory entry with a title and free-text content. */ -export type EpisodicMemoryType = "conversation" | "project_work" | "learning" | "decision" | "important_event" export interface EpisodicMemory { id: string; title: string; @@ -116,7 +116,7 @@ const ExplicitDetail: FC = () => { if (range && range.length === 2) { params.start_date = range[0]!.startOf('day').valueOf() - params.end_date = range[0]!.endOf('day').valueOf() + params.end_date = range[1]!.endOf('day').valueOf() } setEpisodicLoading(true) getEpisodicMemory(params) From f93ec8d6096af6d792fe13877c6e55b37242aade Mon Sep 17 00:00:00 2001 From: Eternity <1533512157@qq.com> Date: Wed, 22 Apr 2026 17:46:38 +0800 Subject: [PATCH 185/245] fix(core): fix end_user_id reference and add task status tracking - Fix write_router to use actual_end_user_id instead of end_user_id - Add task status tracking via Redis in scheduler - Expose task_id in memory write response - Fix logging import path in scheduler --- api/app/celery_task_scheduler.py | 50 +++++++++++++++++-- .../service/memory_api_controller.py | 6 +-- .../langgraph_graph/routing/write_router.py | 4 +- api/app/schemas/memory_api_schema.py | 8 +-- api/app/services/memory_api_service.py | 6 +-- 5 files changed, 58 insertions(+), 16 deletions(-) diff --git a/api/app/celery_task_scheduler.py b/api/app/celery_task_scheduler.py index 88afe341..8cdce658 100644 --- a/api/app/celery_task_scheduler.py +++ b/api/app/celery_task_scheduler.py @@ -6,7 +6,7 @@ import redis from app.core.config import settings from celery_app import celery_app -from core.logging_config import get_named_logger +from app.core.logging_config import get_named_logger logger = get_named_logger("task_scheduler") @@ -62,11 +62,34 @@ class RedisTaskScheduler: "params": json.dumps(params), } ) + self.redis.set( + f"task_tracker:{msg_id}", + json.dumps({"status": "QUEUED", "task_id": None}), + ex=86400 + ) return msg_id except Exception as e: logger.error("Push task exception %s", e, exc_info=True) raise e + def get_task_status(self, msg_id: str) -> dict: + raw = self.redis.get(f"task_tracker:{msg_id}") + if raw is None: + return {"status": "NOT_FOUND"} + + tracker = json.loads(raw) + status = tracker["status"] + task_id = tracker.get("task_id") + result_content = tracker.get("result") or {} + if status == "DISPATCHED" and task_id: + result_raw = self.redis.get(f"celery-task-meta-{task_id}") + if result_raw: + result_data = json.loads(result_raw) + status = result_data.get("status", status) + result_content = result_data.get("result") + + return {"status": status, "task_id": task_id, "result": result_content} + def _cleanup_finished(self): pending = self.redis.hgetall(PENDING_HASH) if not pending: @@ -91,6 +114,7 @@ class RedisTaskScheduler: age = now - dispatched_at should_cleanup = False + result_data = None if raw_result is not None: result_data = json.loads(raw_result) if result_data.get("status") in ("SUCCESS", "FAILURE", "REVOKED"): @@ -104,8 +128,20 @@ class RedisTaskScheduler: ) if should_cleanup: + final_status = result_data.get("status", "UNKNOWN") if result_data else "EXPIRED" cleanup_pipe.delete(lock_key) cleanup_pipe.hdel(PENDING_HASH, task_id) + tracker_msg_id = meta.get("msg_id") + if tracker_msg_id: + cleanup_pipe.set( + f"task_tracker:{tracker_msg_id}", + json.dumps({ + "status": final_status, + "task_id": task_id, + "result": result_data.get("result") or {} + }), + ex=86400, + ) has_cleanup = True except Exception as e: logger.error("Cleanup error for %s: %s", task_id, e, exc_info=True) @@ -125,9 +161,15 @@ class RedisTaskScheduler: pipe.set(lock_key, task.id, ex=3600) pipe.hset(PENDING_HASH, task.id, json.dumps({ "lock_key": lock_key, - "dispatched_at": time.time() + "dispatched_at": time.time(), + "msg_id": msg_id })) pipe.xdel(STREAM_KEY, msg_id) + pipe.set( + f"task_tracker:{msg_id}", + json.dumps({"status": "DISPATCHED", "task_id": task.id}), + ex=86400, + ) pipe.execute() self.dispatched += 1 logger.info("Task dispatched: %s", task.id) @@ -183,8 +225,8 @@ class RedisTaskScheduler: if locked or lock_key in deliver_keys: continue - key = self._dispatch(msg_id, msg_data) - if key: + dispatched_successfully = self._dispatch(msg_id, msg_data) + if dispatched_successfully: deliver_keys.add(lock_key) time.sleep(0.1) diff --git a/api/app/controllers/service/memory_api_controller.py b/api/app/controllers/service/memory_api_controller.py index de56f56e..57664e4e 100644 --- a/api/app/controllers/service/memory_api_controller.py +++ b/api/app/controllers/service/memory_api_controller.py @@ -18,6 +18,7 @@ from app.schemas.memory_api_schema import ( MemoryWriteSyncResponse, ) from app.services.memory_api_service import MemoryAPIService +from celery_task_scheduler import scheduler router = APIRouter(prefix="/memory", tags=["V1 - Memory API"]) logger = get_business_logger() @@ -86,7 +87,7 @@ async def write_memory( user_rag_memory_id=payload.user_rag_memory_id, ) - logger.info(f"Memory write task submitted: end_user_id: {payload.end_user_id}") + logger.info(f"Memory write task submitted: task_id: {result['task_id']} end_user_id: {payload.end_user_id}") return success(data=MemoryWriteResponse(**result).model_dump(), msg="Memory write task submitted") @@ -105,8 +106,7 @@ async def get_write_task_status( """ logger.info(f"Write task status check - task_id: {task_id}") - from app.services.task_service import get_task_memory_write_result - result = get_task_memory_write_result(task_id) + result = scheduler.get_task_status(task_id) return success(data=_sanitize_task_result(result), msg="Task status retrieved") diff --git a/api/app/core/memory/agent/langgraph_graph/routing/write_router.py b/api/app/core/memory/agent/langgraph_graph/routing/write_router.py index 50f7ddb9..7ef4ed12 100644 --- a/api/app/core/memory/agent/langgraph_graph/routing/write_router.py +++ b/api/app/core/memory/agent/langgraph_graph/routing/write_router.py @@ -94,9 +94,9 @@ async def write( # ) scheduler.push_task( "app.core.memory.agent.write_message", - end_user_id, + actual_end_user_id, { - "end_user_id": end_user_id, + "end_user_id": actual_end_user_id, "message": structured_messages, "config_id": str(actual_config_id), "storage_type": storage_type, diff --git a/api/app/schemas/memory_api_schema.py b/api/app/schemas/memory_api_schema.py index 4cc548f3..7e4ca74a 100644 --- a/api/app/schemas/memory_api_schema.py +++ b/api/app/schemas/memory_api_schema.py @@ -112,12 +112,12 @@ class MemoryWriteResponse(BaseModel): """Response schema for memory write operation. Attributes: - task_id: Celery task ID for status polling - status: Initial task status (PENDING) + task_id: task ID for status polling + status: Initial task status (QUEUED) end_user_id: End user ID the write was submitted for """ - task_id: str = Field(..., description="Celery task ID for polling") - status: str = Field(..., description="Task status: PENDING") + task_id: str = Field(..., description="task ID for polling") + status: str = Field(..., description="Task status: QUEUED") end_user_id: str = Field(..., description="End user ID") diff --git a/api/app/services/memory_api_service.py b/api/app/services/memory_api_service.py index 221ca4cf..a1ceef86 100644 --- a/api/app/services/memory_api_service.py +++ b/api/app/services/memory_api_service.py @@ -175,7 +175,7 @@ class MemoryAPIService: # storage_type, # user_rag_memory_id or "", # ) - scheduler.push_task( + task_id = scheduler.push_task( "app.core.memory.agent.write_message", end_user_id, { @@ -190,8 +190,8 @@ class MemoryAPIService: logger.info(f"Memory write task submitted, end_user_id={end_user_id}") return { - # "task_id": task.id, - "status": "PENDING", + "task_id": task_id, + "status": "QUEUED", "end_user_id": end_user_id, } From ec6b08cde2674fbf8287628e4822bce1efdd698b Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 22 Apr 2026 18:39:39 +0800 Subject: [PATCH 186/245] feat(web): support switch space --- web/src/api/workspaces.ts | 3 +- web/src/assets/images/menuNew/return.svg | 19 +++ web/src/assets/images/menuNew/switch.svg | 18 +++ .../components/SiderMenu/SwitchSpaceModal.tsx | 116 ++++++++++++++++++ web/src/components/SiderMenu/index.module.css | 4 + web/src/components/SiderMenu/index.tsx | 53 +++++--- web/src/i18n/en.ts | 1 + web/src/i18n/zh.ts | 1 + 8 files changed, 196 insertions(+), 19 deletions(-) create mode 100644 web/src/assets/images/menuNew/return.svg create mode 100644 web/src/assets/images/menuNew/switch.svg create mode 100644 web/src/components/SiderMenu/SwitchSpaceModal.tsx diff --git a/web/src/api/workspaces.ts b/web/src/api/workspaces.ts index 5c62489d..ee394abc 100644 --- a/web/src/api/workspaces.ts +++ b/web/src/api/workspaces.ts @@ -9,8 +9,9 @@ import type { SpaceModalData } from '@/views/SpaceManagement/types' import type { SpaceConfigData } from '@/views/SpaceConfig/types' // Workspace list +export const getWorkspacesUrl = '/workspaces' export const getWorkspaces = (data?: { include_current?: boolean }) => { - return request.get('/workspaces', data) + return request.get(getWorkspacesUrl, data) } // Create workspace export const createWorkspace = (values: SpaceModalData) => { diff --git a/web/src/assets/images/menuNew/return.svg b/web/src/assets/images/menuNew/return.svg new file mode 100644 index 00000000..7fb038dd --- /dev/null +++ b/web/src/assets/images/menuNew/return.svg @@ -0,0 +1,19 @@ + + + 退出 + + + + + + + + + + + + + + + + \ No newline at end of file diff --git a/web/src/assets/images/menuNew/switch.svg b/web/src/assets/images/menuNew/switch.svg new file mode 100644 index 00000000..8adfd3ee --- /dev/null +++ b/web/src/assets/images/menuNew/switch.svg @@ -0,0 +1,18 @@ + + + 切换 + + + + + + + + + + + + + + + \ No newline at end of file diff --git a/web/src/components/SiderMenu/SwitchSpaceModal.tsx b/web/src/components/SiderMenu/SwitchSpaceModal.tsx new file mode 100644 index 00000000..2e486775 --- /dev/null +++ b/web/src/components/SiderMenu/SwitchSpaceModal.tsx @@ -0,0 +1,116 @@ +/* + * @Author: ZhaoYing + * @Date: 2026-02-25 11:45:07 + * @Last Modified by: ZhaoYing + * @Last Modified time: 2026-02-27 09:59:41 + */ +/** + * SwitchSpaceModal Component + * + * A two-step modal for changing user email address with verification code. + * Step 1: Enter new email and send verification code + * Step 2: Confirm the email change + */ + +import { forwardRef, useImperativeHandle, useState } from 'react'; +import { Form, App, Space } from 'antd'; +import { useTranslation } from 'react-i18next'; + +import RbModal from '@/components/RbModal' +import { switchWorkspace, getWorkspacesUrl } from '@/api/workspaces' +import CustomSelect from '@/components/CustomSelect'; +import Tag from '@/components/Tag' +import { useUser } from '@/store/user'; + +const FormItem = Form.Item; + +export interface SwitchSpaceModalRef { + handleOpen: () => void; +} + +const SwitchSpaceModal = forwardRef((_props, ref) => { + const { t } = useTranslation(); + const { message } = App.useApp(); + const [visible, setVisible] = useState(false); + const [form] = Form.useForm<{ space_id: string }>(); + const [loading, setLoading] = useState(false) + const { user } = useUser() + + /** Close modal and reset form */ + const handleClose = () => { + setVisible(false); + form.resetFields(); + setLoading(false) + }; + + /** Open modal */ + const handleOpen = () => { + form.resetFields(); + setVisible(true); + form.setFieldsValue({ space_id: user?.current_workspace_id }) + }; + /** Handle save/next button click - proceed to next step or submit email change */ + const handleSave = () => { + form + .validateFields() + .then((values) => { + if (user?.current_workspace_id === values.space_id) { + handleClose() + return + } + setLoading(true) + switchWorkspace(values.space_id) + .then(res => { + if (res) { + message.success(t('common.operateSuccess')); + localStorage.removeItem('user') + window.location.reload() + } + }) + .finally(() => setLoading(false)) + }) + .catch((err) => { + console.log('err', err) + }); + } + + /** Expose methods to parent component */ + useImperativeHandle(ref, () => ({ + handleOpen, + handleClose + })); + + return ( + +
+ + list.map(item => ({ + value: item.id, + label: {item.name}{t(`space.${item.storage_type || 'neo4j'}`)} + }))} + /> + +
+ + ); +}); + +export default SwitchSpaceModal; \ No newline at end of file diff --git a/web/src/components/SiderMenu/index.module.css b/web/src/components/SiderMenu/index.module.css index 9cc61665..e07f4c9c 100644 --- a/web/src/components/SiderMenu/index.module.css +++ b/web/src/components/SiderMenu/index.module.css @@ -2,6 +2,10 @@ /* border-right: 1px solid #EAECEE; */ max-height: 100vh; } +.sider :global(.ant-layout-sider-children) { + display: flex; + flex-direction: column; +} .title { height: 64px; padding: 24px 10px 12px 12px; diff --git a/web/src/components/SiderMenu/index.tsx b/web/src/components/SiderMenu/index.tsx index 0363ce45..4acc8b46 100644 --- a/web/src/components/SiderMenu/index.tsx +++ b/web/src/components/SiderMenu/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-02 15:25:31 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-16 17:35:38 + * @Last Modified time: 2026-04-21 17:56:09 */ /** * SiderMenu Component @@ -19,7 +19,7 @@ */ import { useState, useEffect, useRef, type FC } from 'react'; -import { Menu as AntMenu, Layout, Flex } from 'antd'; +import { Menu as AntMenu, Layout, Flex, Divider } from 'antd'; import { UserOutlined } from '@ant-design/icons'; import type { MenuProps } from 'antd'; import { useNavigate, useLocation } from 'react-router-dom'; @@ -32,7 +32,8 @@ import logo from '@/assets/images/logo.png' import { useUser } from '@/store/user'; import { getTenantSubscription } from '@/api/user'; import { useI18n } from '@/store/locale' -import SubscriptionDetailModal, { type SubscriptionDetailModalRef } from './SubscriptionDetailModal' +import SubscriptionDetailModal, { type SubscriptionDetailModalRef } from './SubscriptionDetailModal'; +import SwitchSpaceModal, { type SwitchSpaceModalRef } from './SwitchSpaceModal'; // Import SVG files // space @@ -174,6 +175,7 @@ const Menu: FC<{ const [menus, setMenus] = useState([]) const { user, storageType } = useUser() const subscriptionDetailRef = useRef(null) + const switchSpaceModalRef = useRef(null) /** Filter menus based on user role and source */ useEffect(() => { @@ -346,6 +348,9 @@ const Menu: FC<{ const handleViewDetail = () => { subscriptionDetailRef.current?.handleOpen(subscription) } + const handleSwitchSpace = () => { + switchSpaceModalRef.current?.handleOpen() + } return ( {/* Return to space button for superusers */} {user?.is_superuser && source === 'space' && - -
- {collapsed ? null : t('common.returnToSpace')} + + + +
+ {collapsed ? null : t('common.switchSpace')} + + +
+ {collapsed ? null : t('common.returnToSpace')} + } {source === 'manage' && subscription && !collapsed && -
+
{subscription.package_plan?.[getKeyWithLanguage('name')]}
@@ -434,6 +448,9 @@ const Menu: FC<{ + ); }; diff --git a/web/src/i18n/en.ts b/web/src/i18n/en.ts index cbdf921d..b1fe36b5 100644 --- a/web/src/i18n/en.ts +++ b/web/src/i18n/en.ts @@ -474,6 +474,7 @@ export const en = { view: 'View', updated_at: 'Updated At', callbackUrlInvalid: 'Please enter a valid URL', + switchSpace: 'Switch Space', }, model: { searchPlaceholder: 'search model…', diff --git a/web/src/i18n/zh.ts b/web/src/i18n/zh.ts index 7bd39034..6a5e2b6e 100644 --- a/web/src/i18n/zh.ts +++ b/web/src/i18n/zh.ts @@ -1153,6 +1153,7 @@ export const zh = { view: '查看', updated_at: '更新时间', callbackUrlInvalid: '请输入有效的 URL', + switchSpace: '切换空间', }, model: { searchPlaceholder: '搜索模型…', From 7193eed9e3651d23fe45a4481db5e1db25060fe4 Mon Sep 17 00:00:00 2001 From: wwq Date: Wed, 22 Apr 2026 18:46:22 +0800 Subject: [PATCH 187/245] feat(quota): implement workspace-level quota enforcement and statistics - Refactor quota management logic to support usage checks scoped by workspace. - Update quota statistics API to return granular quota details for each workspace. - Revise default configuration settings for terminal user and model limits. - Remove quota check decorators from the model controller. --- api/app/config/default_free_plan.py | 4 +- api/app/controllers/app_controller.py | 2 +- api/app/controllers/model_controller.py | 2 - .../controllers/public_share_controller.py | 4 +- .../controllers/service/app_api_controller.py | 2 +- api/app/core/quota_manager.py | 275 +++++++++++++++--- 6 files changed, 237 insertions(+), 52 deletions(-) diff --git a/api/app/config/default_free_plan.py b/api/app/config/default_free_plan.py index a9ac3c66..3ecc0498 100644 --- a/api/app/config/default_free_plan.py +++ b/api/app/config/default_free_plan.py @@ -60,9 +60,9 @@ def _build_default_free_plan(): "app_quota": 2, "knowledge_capacity_quota": 0.3, "memory_engine_quota": 1, - "end_user_quota": 1, + "end_user_quota": 10, "ontology_project_quota": 3, - "model_quota": 4, + "model_quota": 1, "api_ops_rate_limit": 50, }, } diff --git a/api/app/controllers/app_controller.py b/api/app/controllers/app_controller.py index f3cbe5ea..eda5e76a 100644 --- a/api/app/controllers/app_controller.py +++ b/api/app/controllers/app_controller.py @@ -1286,7 +1286,7 @@ async def import_app( # 仅新建应用时检查配额，覆盖已有应用时跳过 if target_app_id is None: from app.core.quota_manager import _check_quota - _check_quota(db, current_user.tenant_id, "app_quota", "app") + _check_quota(db, current_user.tenant_id, "app_quota", "app", workspace_id=current_user.current_workspace_id) result_app, warnings = AppDslService(db).import_dsl( dsl=dsl, workspace_id=current_user.current_workspace_id, diff --git a/api/app/controllers/model_controller.py b/api/app/controllers/model_controller.py index 6105c3d8..57c22337 100644 --- a/api/app/controllers/model_controller.py +++ b/api/app/controllers/model_controller.py @@ -237,7 +237,6 @@ def delete_model_base( @router.post("/model_plaza/{model_base_id}/add", response_model=ApiResponse) -@check_model_quota def add_model_from_plaza( model_base_id: uuid.UUID, db: Session = Depends(get_db), @@ -275,7 +274,6 @@ def get_model_by_id( @router.post("", response_model=ApiResponse) -@check_model_quota async def create_model( model_data: model_schema.ModelConfigCreate, db: Session = Depends(get_db), diff --git a/api/app/controllers/public_share_controller.py b/api/app/controllers/public_share_controller.py index 486854ba..97b500fa 100644 --- a/api/app/controllers/public_share_controller.py +++ b/api/app/controllers/public_share_controller.py @@ -228,7 +228,7 @@ def list_conversations( from app.models.workspace_model import Workspace ws = db.query(Workspace).filter(Workspace.id == workspace_id).first() if ws: - _check_quota(db, ws.tenant_id, "end_user_quota", "end_user") + _check_quota(db, ws.tenant_id, "end_user_quota", "end_user", workspace_id=workspace_id) new_end_user = end_user_repo.get_or_create_end_user( app_id=share.app_id, @@ -370,7 +370,7 @@ async def chat( ws = db.query(Workspace).filter(Workspace.id == workspace_id).first() if ws: logger.info(f"新终端用户，执行配额检查: tenant_id={ws.tenant_id}") - _check_quota(db, ws.tenant_id, "end_user_quota", "end_user") + _check_quota(db, ws.tenant_id, "end_user_quota", "end_user", workspace_id=workspace_id) new_end_user = end_user_repo.get_or_create_end_user( app_id=share.app_id, diff --git a/api/app/controllers/service/app_api_controller.py b/api/app/controllers/service/app_api_controller.py index d8aefc72..93e88dc5 100644 --- a/api/app/controllers/service/app_api_controller.py +++ b/api/app/controllers/service/app_api_controller.py @@ -114,7 +114,7 @@ async def chat( from app.models.workspace_model import Workspace ws = db.query(Workspace).filter(Workspace.id == workspace_id).first() if ws: - _check_quota(db, ws.tenant_id, "end_user_quota", "end_user") + _check_quota(db, ws.tenant_id, "end_user_quota", "end_user", workspace_id=workspace_id) new_end_user = end_user_repo.get_or_create_end_user( app_id=app.id, diff --git a/api/app/core/quota_manager.py b/api/app/core/quota_manager.py index ad684eb9..c456c202 100644 --- a/api/app/core/quota_manager.py +++ b/api/app/core/quota_manager.py @@ -30,6 +30,60 @@ def _get_user_from_kwargs(kwargs: dict): return None +def _get_workspace_id_from_kwargs(kwargs: dict): + """从 kwargs 中获取 workspace_id""" + workspace_id = kwargs.get("workspace_id") + if workspace_id: + logger.info(f"_get_workspace_id_from_kwargs: 从 kwargs['workspace_id'] 获取: {workspace_id}") + return workspace_id + + data = kwargs.get("data") or kwargs.get("body") or kwargs.get("payload") + if data and hasattr(data, "workspace_id"): + ws_id = data.workspace_id + if ws_id: + logger.info(f"_get_workspace_id_from_kwargs: 从 payload 获取: {ws_id}") + return ws_id + logger.info(f"_get_workspace_id_from_kwargs: payload.workspace_id 为 None，继续尝试其他方式") + + api_key_auth = kwargs.get("api_key_auth") + if api_key_auth and hasattr(api_key_auth, 'workspace_id'): + ws_id = api_key_auth.workspace_id + logger.info(f"_get_workspace_id_from_kwargs: 从 api_key_auth 获取: {ws_id}") + return ws_id + + # 从 current_user.current_workspace_id 获取（cur_workspace_access_guard 模式） + user = _get_user_from_kwargs(kwargs) + logger.info(f"_get_workspace_id_from_kwargs: user={user}, type={type(user)}") + if user: + ws_id = getattr(user, 'current_workspace_id', None) + logger.info(f"_get_workspace_id_from_kwargs: user.current_workspace_id={ws_id}") + if ws_id: + return ws_id + + # 如果 current_workspace_id 为空，尝试获取用户的第一个工作空间 + if hasattr(user, 'id'): + try: + from app.models.workspace_model import WorkspaceMember + db_session = kwargs.get("db") + logger.info(f"_get_workspace_id_from_kwargs: db_session={db_session is not None}, user.id={user.id}") + if db_session: + first_workspace = ( + db_session.query(WorkspaceMember.workspace_id) + .filter(WorkspaceMember.user_id == user.id) + .filter(WorkspaceMember.is_active.is_(True)) + .first() + ) + logger.info(f"_get_workspace_id_from_kwargs: first_workspace={first_workspace}") + if first_workspace: + logger.info(f"用户 {user.username} 的 current_workspace_id 为空，使用第一个工作空间: {first_workspace.workspace_id}") + return first_workspace.workspace_id + except Exception as e: + logger.error(f"获取用户第一个工作空间失败: {e}", exc_info=True) + + logger.warning(f"_get_workspace_id_from_kwargs: 无法从 kwargs 获取 workspace_id, keys={list(kwargs.keys())}") + return None + + def _get_tenant_id_from_kwargs(db: Session, kwargs: dict): """从 kwargs 中获取 tenant_id""" user = _get_user_from_kwargs(kwargs) @@ -136,15 +190,19 @@ class QuotaUsageRepository: Workspace.is_active.is_(True) ).count() - def count_apps(self, tenant_id: UUID) -> int: + def count_apps(self, tenant_id: UUID, workspace_id: Optional[UUID] = None) -> int: from app.models.app_model import App from app.models.workspace_model import Workspace - return self.db.query(App).join( + query = self.db.query(App).join( Workspace, App.workspace_id == Workspace.id ).filter( - Workspace.tenant_id == tenant_id, App.is_active.is_(True) - ).count() + ) + if workspace_id: + query = query.filter(App.workspace_id == workspace_id) + else: + query = query.filter(Workspace.tenant_id == tenant_id) + return query.count() def count_skills(self, tenant_id: UUID) -> int: from app.models.skill_model import Skill @@ -153,41 +211,50 @@ class QuotaUsageRepository: Skill.is_active.is_(True) ).count() - def sum_knowledge_capacity_gb(self, tenant_id: UUID) -> float: + def sum_knowledge_capacity_gb(self, tenant_id: UUID, workspace_id: Optional[UUID] = None) -> float: from app.models.document_model import Document from app.models.knowledge_model import Knowledge from app.models.workspace_model import Workspace - result = self.db.query(func.coalesce(func.sum(Document.file_size), 0)).join( + query = self.db.query(func.coalesce(func.sum(Document.file_size), 0)).join( Knowledge, Document.kb_id == Knowledge.id ).join( Workspace, Knowledge.workspace_id == Workspace.id ).filter( - Workspace.tenant_id == tenant_id, Document.status == 1, - ).scalar() + ) + if workspace_id: + query = query.filter(Knowledge.workspace_id == workspace_id) + else: + query = query.filter(Workspace.tenant_id == tenant_id) + result = query.scalar() return float(result) / (1024 ** 3) if result else 0.0 - def count_memory_engines(self, tenant_id: UUID) -> int: + def count_memory_engines(self, tenant_id: UUID, workspace_id: Optional[UUID] = None) -> int: from app.models.memory_config_model import MemoryConfig from app.models.workspace_model import Workspace - return self.db.query(MemoryConfig).join( + query = self.db.query(MemoryConfig).join( Workspace, MemoryConfig.workspace_id == Workspace.id - ).filter( - Workspace.tenant_id == tenant_id - ).count() + ) + if workspace_id: + query = query.filter(MemoryConfig.workspace_id == workspace_id) + else: + query = query.filter(Workspace.tenant_id == tenant_id) + return query.count() - def count_end_users(self, tenant_id: UUID) -> int: + def count_end_users(self, tenant_id: UUID, workspace_id: Optional[UUID] = None) -> int: from app.models.end_user_model import EndUser from app.models.workspace_model import Workspace from app.models.user_model import User + query = self.db.query(EndUser).join( + Workspace, EndUser.workspace_id == Workspace.id + ) + if workspace_id: + query = query.filter(EndUser.workspace_id == workspace_id) + else: + query = query.filter(Workspace.tenant_id == tenant_id) trial_user_ids = [ str(u.id) for u in self.db.query(User.id).filter(User.tenant_id == tenant_id).all() ] - query = self.db.query(EndUser).join( - Workspace, EndUser.workspace_id == Workspace.id - ).filter( - Workspace.tenant_id == tenant_id - ) if trial_user_ids: query = query.filter(~EndUser.other_id.in_(trial_user_ids)) return query.count() @@ -196,19 +263,24 @@ class QuotaUsageRepository: from app.models.models_model import ModelConfig return self.db.query(ModelConfig).filter( ModelConfig.tenant_id == tenant_id, - ModelConfig.is_active == True + ModelConfig.is_active == True, + ModelConfig.is_composite == True ).count() - def count_ontology_projects(self, tenant_id: UUID) -> int: + def count_ontology_projects(self, tenant_id: UUID, workspace_id: Optional[UUID] = None) -> int: from app.models.ontology_scene import OntologyScene from app.models.workspace_model import Workspace + if workspace_id: + return self.db.query(OntologyScene).filter( + OntologyScene.workspace_id == workspace_id + ).count() return self.db.query(OntologyScene).join( Workspace, OntologyScene.workspace_id == Workspace.id ).filter( Workspace.tenant_id == tenant_id ).count() - def get_usage_by_quota_type(self, tenant_id: UUID, quota_type: str): + def get_usage_by_quota_type(self, tenant_id: UUID, quota_type: str, workspace_id: Optional[UUID] = None): """按配额类型分发，返回当前使用量""" dispatch = { "workspace_quota": self.count_workspaces, @@ -221,6 +293,8 @@ class QuotaUsageRepository: "ontology_project_quota": self.count_ontology_projects, } fn = dispatch.get(quota_type) + if workspace_id: + return fn(tenant_id, workspace_id) if fn else 0 return fn(tenant_id) if fn else 0 @@ -230,6 +304,7 @@ def _check_quota( quota_type: str, resource_name: str, usage_func: Optional[Callable] = None, + workspace_id: Optional[UUID] = None, ) -> None: """核心配额检查逻辑：对比使用量和配额限制""" try: @@ -244,13 +319,13 @@ def _check_quota( return if usage_func: - current_usage = usage_func(db, tenant_id) + current_usage = usage_func(db, tenant_id, workspace_id) if workspace_id else usage_func(db, tenant_id) else: - current_usage = QuotaUsageRepository(db).get_usage_by_quota_type(tenant_id, quota_type) + current_usage = QuotaUsageRepository(db).get_usage_by_quota_type(tenant_id, quota_type, workspace_id) if current_usage >= quota_limit: logger.warning( - f"配额不足: tenant={tenant_id}, type={quota_type}, " + f"配额不足: tenant={tenant_id}, workspace={workspace_id}, type={quota_type}, " f"usage={current_usage}, limit={quota_limit}" ) raise QuotaExceededError( @@ -260,7 +335,7 @@ def _check_quota( ) logger.debug( - f"配额检查通过: tenant={tenant_id}, type={quota_type}, " + f"配额检查通过: tenant={tenant_id}, workspace={workspace_id}, type={quota_type}, " f"usage={current_usage}, limit={quota_limit}" ) @@ -268,7 +343,7 @@ def _check_quota( raise except Exception as e: logger.error( - f"配额检查异常: tenant={tenant_id}, type={quota_type}, " + f"配额检查异常: tenant={tenant_id}, workspace={workspace_id}, type={quota_type}, " f"error_type={type(e).__name__}, error={str(e)}", exc_info=True, ) @@ -333,7 +408,11 @@ def check_app_quota(func: Callable) -> Callable: if not db or not user: logger.error(f"配额检查失败：{func.__name__} 缺少 db 或 user 参数，拒绝请求") raise InternalServerError() - _check_quota(db, user.tenant_id, "app_quota", "app") + workspace_id = _get_workspace_id_from_kwargs(kwargs) + if not workspace_id: + logger.error(f"配额检查失败：{func.__name__} 无法获取 workspace_id，拒绝请求") + raise InternalServerError() + _check_quota(db, user.tenant_id, "app_quota", "app", workspace_id=workspace_id) return await func(*args, **kwargs) @wraps(func) @@ -343,7 +422,11 @@ def check_app_quota(func: Callable) -> Callable: if not db or not user: logger.error(f"配额检查失败：{func.__name__} 缺少 db 或 user 参数，拒绝请求") raise InternalServerError() - _check_quota(db, user.tenant_id, "app_quota", "app") + workspace_id = _get_workspace_id_from_kwargs(kwargs) + if not workspace_id: + logger.error(f"配额检查失败：{func.__name__} 无法获取 workspace_id，拒绝请求") + raise InternalServerError() + _check_quota(db, user.tenant_id, "app_quota", "app", workspace_id=workspace_id) return func(*args, **kwargs) return async_wrapper if asyncio.iscoroutinefunction(func) else sync_wrapper @@ -360,7 +443,11 @@ def check_knowledge_capacity_quota(func: Callable) -> Callable: if not tenant_id: logger.error(f"配额检查失败：{func.__name__} 无法获取 tenant_id，拒绝请求") raise InternalServerError() - _check_quota(db, tenant_id, "knowledge_capacity_quota", "knowledge_capacity") + workspace_id = _get_workspace_id_from_kwargs(kwargs) + if not workspace_id: + logger.error(f"配额检查失败：{func.__name__} 无法获取 workspace_id，拒绝请求") + raise InternalServerError() + _check_quota(db, tenant_id, "knowledge_capacity_quota", "knowledge_capacity", workspace_id=workspace_id) return await func(*args, **kwargs) @wraps(func) @@ -370,7 +457,11 @@ def check_knowledge_capacity_quota(func: Callable) -> Callable: if not db or not user: logger.error(f"配额检查失败：{func.__name__} 缺少 db 或 user 参数，拒绝请求") raise InternalServerError() - _check_quota(db, user.tenant_id, "knowledge_capacity_quota", "knowledge_capacity") + workspace_id = _get_workspace_id_from_kwargs(kwargs) + if not workspace_id: + logger.error(f"配额检查失败：{func.__name__} 无法获取 workspace_id，拒绝请求") + raise InternalServerError() + _check_quota(db, user.tenant_id, "knowledge_capacity_quota", "knowledge_capacity", workspace_id=workspace_id) return func(*args, **kwargs) return async_wrapper if asyncio.iscoroutinefunction(func) else sync_wrapper @@ -381,20 +472,30 @@ def check_memory_engine_quota(func: Callable) -> Callable: async def async_wrapper(*args, **kwargs): db: Session = kwargs.get("db") user = _get_user_from_kwargs(kwargs) + logger.debug(f"check_memory_engine_quota async_wrapper: db={db is not None}, user={user}, kwargs_keys={list(kwargs.keys())}") if not db or not user: logger.error(f"配额检查失败：{func.__name__} 缺少 db 或 user 参数，拒绝请求") raise InternalServerError() - _check_quota(db, user.tenant_id, "memory_engine_quota", "memory_engine") + workspace_id = _get_workspace_id_from_kwargs(kwargs) + if not workspace_id: + logger.error(f"配额检查失败：{func.__name__} 无法获取 workspace_id，拒绝请求") + raise InternalServerError() + _check_quota(db, user.tenant_id, "memory_engine_quota", "memory_engine", workspace_id=workspace_id) return await func(*args, **kwargs) @wraps(func) def sync_wrapper(*args, **kwargs): db: Session = kwargs.get("db") user = _get_user_from_kwargs(kwargs) + logger.debug(f"check_memory_engine_quota sync_wrapper: db={db is not None}, user={user}, kwargs_keys={list(kwargs.keys())}") if not db or not user: logger.error(f"配额检查失败：{func.__name__} 缺少 db 或 user 参数，拒绝请求") raise InternalServerError() - _check_quota(db, user.tenant_id, "memory_engine_quota", "memory_engine") + workspace_id = _get_workspace_id_from_kwargs(kwargs) + if not workspace_id: + logger.error(f"配额检查失败：{func.__name__} 无法获取 workspace_id，拒绝请求") + raise InternalServerError() + _check_quota(db, user.tenant_id, "memory_engine_quota", "memory_engine", workspace_id=workspace_id) return func(*args, **kwargs) return async_wrapper if asyncio.iscoroutinefunction(func) else sync_wrapper @@ -411,7 +512,11 @@ def check_end_user_quota(func: Callable) -> Callable: if not tenant_id: logger.error(f"配额检查失败：{func.__name__} 无法获取 tenant_id，拒绝请求") raise InternalServerError() - _check_quota(db, tenant_id, "end_user_quota", "end_user") + workspace_id = _get_workspace_id_from_kwargs(kwargs) + if not workspace_id: + logger.error(f"配额检查失败：{func.__name__} 无法获取 workspace_id，拒绝请求") + raise InternalServerError() + _check_quota(db, tenant_id, "end_user_quota", "end_user", workspace_id=workspace_id) return await func(*args, **kwargs) @wraps(func) @@ -424,7 +529,11 @@ def check_end_user_quota(func: Callable) -> Callable: if not tenant_id: logger.error(f"配额检查失败：{func.__name__} 无法获取 tenant_id，拒绝请求") raise InternalServerError() - _check_quota(db, tenant_id, "end_user_quota", "end_user") + workspace_id = _get_workspace_id_from_kwargs(kwargs) + if not workspace_id: + logger.error(f"配额检查失败：{func.__name__} 无法获取 workspace_id，拒绝请求") + raise InternalServerError() + _check_quota(db, tenant_id, "end_user_quota", "end_user", workspace_id=workspace_id) return func(*args, **kwargs) return async_wrapper if asyncio.iscoroutinefunction(func) else sync_wrapper @@ -438,7 +547,11 @@ def check_ontology_project_quota(func: Callable) -> Callable: if not db or not user: logger.error(f"配额检查失败：{func.__name__} 缺少 db 或 user 参数，拒绝请求") raise InternalServerError() - _check_quota(db, user.tenant_id, "ontology_project_quota", "ontology_project") + workspace_id = _get_workspace_id_from_kwargs(kwargs) + if not workspace_id: + logger.error(f"配额检查失败：{func.__name__} 无法获取 workspace_id，拒绝请求") + raise InternalServerError() + _check_quota(db, user.tenant_id, "ontology_project_quota", "ontology_project", workspace_id=workspace_id) return await func(*args, **kwargs) @wraps(func) @@ -448,7 +561,11 @@ def check_ontology_project_quota(func: Callable) -> Callable: if not db or not user: logger.error(f"配额检查失败：{func.__name__} 缺少 db 或 user 参数，拒绝请求") raise InternalServerError() - _check_quota(db, user.tenant_id, "ontology_project_quota", "ontology_project") + workspace_id = _get_workspace_id_from_kwargs(kwargs) + if not workspace_id: + logger.error(f"配额检查失败：{func.__name__} 无法获取 workspace_id，拒绝请求") + raise InternalServerError() + _check_quota(db, user.tenant_id, "ontology_project_quota", "ontology_project", workspace_id=workspace_id) return func(*args, **kwargs) return async_wrapper if asyncio.iscoroutinefunction(func) else sync_wrapper @@ -581,7 +698,14 @@ def check_quota(quota_type: str, resource_name: str, usage_func: Optional[Callab # ─── 配额使用统计 ──────────────────────────────────────────────────────────── async def get_quota_usage(db: Session, tenant_id: UUID) -> dict: - """获取租户所有配额的使用情况""" + """获取租户所有配额的使用情况 + + 对于 workspace 级别的配额（app/knowledge_capacity/memory_engine/end_user）： + - used: 租户汇总（所有空间加总） + - limit: quota × 活跃工作区数（有效总限额，使汇总数据自洽） + - per_workspace: 各空间明细，包含 workspace_id、workspace_name、used、limit、percentage + - 配额检查逻辑不变：仍按单个空间独立检查 + """ quota_config = _get_quota_config(db, tenant_id) if not quota_config: return {} @@ -600,11 +724,48 @@ async def get_quota_usage(db: Session, tenant_id: UUID) -> dict: model_count = repo.count_models(tenant_id) ontology_count = repo.count_ontology_projects(tenant_id) + # 获取租户下所有活跃工作区，用于按空间拆分明细 + from app.models.workspace_model import Workspace + active_workspaces = db.query(Workspace).filter( + Workspace.tenant_id == tenant_id, + Workspace.is_active.is_(True) + ).all() + + # 构建各空间的 workspace 级配额明细 + def _build_per_workspace_detail(count_func, per_unit_limit): + """为 workspace 级配额构建 per_workspace 明细列表""" + if not per_unit_limit or not active_workspaces: + return [] + details = [] + for ws in active_workspaces: + ws_used = count_func(tenant_id, ws.id) + details.append({ + "workspace_id": str(ws.id), + "workspace_name": ws.name, + "used": ws_used, + "limit": per_unit_limit, + "percentage": pct(ws_used, per_unit_limit), + }) + return details + + # workspace 级配额的每空间限额 + app_quota_per_ws = quota_config.get("app_quota") + knowledge_quota_per_ws = quota_config.get("knowledge_capacity_quota") + memory_quota_per_ws = quota_config.get("memory_engine_quota") + end_user_quota_per_ws = quota_config.get("end_user_quota") + ontology_quota_per_ws = quota_config.get("ontology_project_quota") + + # workspace 级配额的有效总限额 = 每空间限额 × 活跃工作区数 + app_effective_limit = app_quota_per_ws * workspace_count if app_quota_per_ws is not None and workspace_count > 0 else app_quota_per_ws + knowledge_effective_limit = knowledge_quota_per_ws * workspace_count if knowledge_quota_per_ws is not None and workspace_count > 0 else knowledge_quota_per_ws + memory_effective_limit = memory_quota_per_ws * workspace_count if memory_quota_per_ws is not None and workspace_count > 0 else memory_quota_per_ws + end_user_effective_limit = end_user_quota_per_ws * workspace_count if end_user_quota_per_ws is not None and workspace_count > 0 else end_user_quota_per_ws + ontology_effective_limit = ontology_quota_per_ws * workspace_count if ontology_quota_per_ws is not None and workspace_count > 0 else ontology_quota_per_ws + api_ops_current = 0 try: from app.aioRedis import aio_redis as _aio_redis from app.models.api_key_model import ApiKey - from app.models.workspace_model import Workspace # api_ops_rate_limit 限的是每个 api_key 每秒最高限额 # 展示当前最接近触发限流的 key 的 QPS（取最大值） api_key_ids = db.query(ApiKey.id).join( @@ -625,11 +786,37 @@ async def get_quota_usage(db: Session, tenant_id: UUID) -> dict: return { "workspace": {"used": workspace_count, "limit": quota_config.get("workspace_quota"), "percentage": pct(workspace_count, quota_config.get("workspace_quota"))}, "skill": {"used": skill_count, "limit": quota_config.get("skill_quota"), "percentage": pct(skill_count, quota_config.get("skill_quota"))}, - "app": {"used": app_count, "limit": quota_config.get("app_quota"), "percentage": pct(app_count, quota_config.get("app_quota"))}, - "knowledge_capacity": {"used": round(knowledge_gb, 2), "limit": quota_config.get("knowledge_capacity_quota"), "percentage": pct(knowledge_gb, quota_config.get("knowledge_capacity_quota")), "unit": "GB"}, - "memory_engine": {"used": memory_count, "limit": quota_config.get("memory_engine_quota"), "percentage": pct(memory_count, quota_config.get("memory_engine_quota"))}, - "end_user": {"used": end_user_count, "limit": quota_config.get("end_user_quota"), "percentage": pct(end_user_count, quota_config.get("end_user_quota"))}, - "ontology_project": {"used": ontology_count, "limit": quota_config.get("ontology_project_quota"), "percentage": pct(ontology_count, quota_config.get("ontology_project_quota"))}, + "app": { + "used": app_count, + "limit": app_effective_limit, + "percentage": pct(app_count, app_effective_limit), + "per_workspace": _build_per_workspace_detail(repo.count_apps, app_quota_per_ws), + }, + "knowledge_capacity": { + "used": round(knowledge_gb, 2), + "limit": knowledge_effective_limit, + "percentage": pct(knowledge_gb, knowledge_effective_limit), + "unit": "GB", + "per_workspace": _build_per_workspace_detail(repo.sum_knowledge_capacity_gb, knowledge_quota_per_ws), + }, + "memory_engine": { + "used": memory_count, + "limit": memory_effective_limit, + "percentage": pct(memory_count, memory_effective_limit), + "per_workspace": _build_per_workspace_detail(repo.count_memory_engines, memory_quota_per_ws), + }, + "end_user": { + "used": end_user_count, + "limit": end_user_effective_limit, + "percentage": pct(end_user_count, end_user_effective_limit), + "per_workspace": _build_per_workspace_detail(repo.count_end_users, end_user_quota_per_ws), + }, + "ontology_project": { + "used": ontology_count, + "limit": ontology_effective_limit, + "percentage": pct(ontology_count, ontology_effective_limit), + "per_workspace": _build_per_workspace_detail(repo.count_ontology_projects, ontology_quota_per_ws), + }, "model": {"used": model_count, "limit": quota_config.get("model_quota"), "percentage": pct(model_count, quota_config.get("model_quota"))}, "api_ops_rate_limit": {"current": api_ops_current, "limit": quota_config.get("api_ops_rate_limit"), "percentage": None, "unit": "次/秒"}, } From f6504068696a04f98f1906c7f468b0ad1485feaa Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 22 Apr 2026 18:50:36 +0800 Subject: [PATCH 188/245] fix(web):switch space --- web/src/components/SiderMenu/SwitchSpaceModal.tsx | 12 +++++------- 1 file changed, 5 insertions(+), 7 deletions(-) diff --git a/web/src/components/SiderMenu/SwitchSpaceModal.tsx b/web/src/components/SiderMenu/SwitchSpaceModal.tsx index 2e486775..e5483f63 100644 --- a/web/src/components/SiderMenu/SwitchSpaceModal.tsx +++ b/web/src/components/SiderMenu/SwitchSpaceModal.tsx @@ -1,15 +1,14 @@ /* * @Author: ZhaoYing - * @Date: 2026-02-25 11:45:07 - * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-02-27 09:59:41 + * @Date: 2026-04-22 18:50:14 + * @Last Modified by: ZhaoYing + * @Last Modified time: 2026-04-22 18:50:14 */ /** * SwitchSpaceModal Component * - * A two-step modal for changing user email address with verification code. - * Step 1: Enter new email and send verification code - * Step 2: Confirm the email change + * A modal for switching the current workspace. + * Displays a dropdown to select a workspace and reloads the page upon confirmation. */ import { forwardRef, useImperativeHandle, useState } from 'react'; @@ -77,7 +76,6 @@ const SwitchSpaceModal = forwardRef((_props, ref) => { /** Expose methods to parent component */ useImperativeHandle(ref, () => ({ handleOpen, - handleClose })); return ( From 4a4391a19c136fb636f7d84987a4f7a24be56c03 Mon Sep 17 00:00:00 2001 From: wwq Date: Wed, 22 Apr 2026 18:52:27 +0800 Subject: [PATCH 189/245] feat(quota): implement workspace-level quota enforcement and statistics - Refactor quota management logic to support usage checks scoped by workspace. - Update quota statistics API to return granular quota details for each workspace. - Revise default configuration settings for terminal user and model limits. - Remove quota check decorators from the model controller. --- api/app/core/quota_manager.py | 52 ++++++++++++----------------------- 1 file changed, 17 insertions(+), 35 deletions(-) diff --git a/api/app/core/quota_manager.py b/api/app/core/quota_manager.py index c456c202..e5d73063 100644 --- a/api/app/core/quota_manager.py +++ b/api/app/core/quota_manager.py @@ -32,26 +32,13 @@ def _get_user_from_kwargs(kwargs: dict): def _get_workspace_id_from_kwargs(kwargs: dict): """从 kwargs 中获取 workspace_id""" + # 优先从 kwargs['workspace_id'] 获取 workspace_id = kwargs.get("workspace_id") if workspace_id: logger.info(f"_get_workspace_id_from_kwargs: 从 kwargs['workspace_id'] 获取: {workspace_id}") return workspace_id - data = kwargs.get("data") or kwargs.get("body") or kwargs.get("payload") - if data and hasattr(data, "workspace_id"): - ws_id = data.workspace_id - if ws_id: - logger.info(f"_get_workspace_id_from_kwargs: 从 payload 获取: {ws_id}") - return ws_id - logger.info(f"_get_workspace_id_from_kwargs: payload.workspace_id 为 None，继续尝试其他方式") - - api_key_auth = kwargs.get("api_key_auth") - if api_key_auth and hasattr(api_key_auth, 'workspace_id'): - ws_id = api_key_auth.workspace_id - logger.info(f"_get_workspace_id_from_kwargs: 从 api_key_auth 获取: {ws_id}") - return ws_id - - # 从 current_user.current_workspace_id 获取（cur_workspace_access_guard 模式） + # 从 current_user.current_workspace_id 获取（管理端接口） user = _get_user_from_kwargs(kwargs) logger.info(f"_get_workspace_id_from_kwargs: user={user}, type={type(user)}") if user: @@ -59,26 +46,21 @@ def _get_workspace_id_from_kwargs(kwargs: dict): logger.info(f"_get_workspace_id_from_kwargs: user.current_workspace_id={ws_id}") if ws_id: return ws_id - - # 如果 current_workspace_id 为空，尝试获取用户的第一个工作空间 - if hasattr(user, 'id'): - try: - from app.models.workspace_model import WorkspaceMember - db_session = kwargs.get("db") - logger.info(f"_get_workspace_id_from_kwargs: db_session={db_session is not None}, user.id={user.id}") - if db_session: - first_workspace = ( - db_session.query(WorkspaceMember.workspace_id) - .filter(WorkspaceMember.user_id == user.id) - .filter(WorkspaceMember.is_active.is_(True)) - .first() - ) - logger.info(f"_get_workspace_id_from_kwargs: first_workspace={first_workspace}") - if first_workspace: - logger.info(f"用户 {user.username} 的 current_workspace_id 为空，使用第一个工作空间: {first_workspace.workspace_id}") - return first_workspace.workspace_id - except Exception as e: - logger.error(f"获取用户第一个工作空间失败: {e}", exc_info=True) + + # 从 payload/data/body 获取（仅当上述方式都失败时） + data = kwargs.get("data") or kwargs.get("body") or kwargs.get("payload") + if data and hasattr(data, "workspace_id"): + ws_id = data.workspace_id + if ws_id: + logger.info(f"_get_workspace_id_from_kwargs: 从 payload 获取: {ws_id}") + return ws_id + + # 从 api_key_auth 获取（API Key 认证方式） + api_key_auth = kwargs.get("api_key_auth") + if api_key_auth and hasattr(api_key_auth, 'workspace_id'): + ws_id = api_key_auth.workspace_id + logger.info(f"_get_workspace_id_from_kwargs: 从 api_key_auth 获取: {ws_id}") + return ws_id logger.warning(f"_get_workspace_id_from_kwargs: 无法从 kwargs 获取 workspace_id, keys={list(kwargs.keys())}") return None From b9cfc47e1ebe8bd53780542e3d32abfd77621df8 Mon Sep 17 00:00:00 2001 From: wwq Date: Wed, 22 Apr 2026 19:54:42 +0800 Subject: [PATCH 190/245] feat(quota): implement workspace-level quota enforcement and statistics - Refactor quota management logic to support usage checks scoped by workspace. - Update quota statistics API to return granular quota details for each workspace. - Revise default configuration settings for terminal user and model limits. - Remove quota check decorators from the model controller. --- api/app/core/quota_manager.py | 22 ++-------------------- 1 file changed, 2 insertions(+), 20 deletions(-) diff --git a/api/app/core/quota_manager.py b/api/app/core/quota_manager.py index e5d73063..6f1be886 100644 --- a/api/app/core/quota_manager.py +++ b/api/app/core/quota_manager.py @@ -35,34 +35,16 @@ def _get_workspace_id_from_kwargs(kwargs: dict): # 优先从 kwargs['workspace_id'] 获取 workspace_id = kwargs.get("workspace_id") if workspace_id: - logger.info(f"_get_workspace_id_from_kwargs: 从 kwargs['workspace_id'] 获取: {workspace_id}") return workspace_id - # 从 current_user.current_workspace_id 获取（管理端接口） + # 从 user.current_workspace_id 获取 user = _get_user_from_kwargs(kwargs) - logger.info(f"_get_workspace_id_from_kwargs: user={user}, type={type(user)}") if user: ws_id = getattr(user, 'current_workspace_id', None) - logger.info(f"_get_workspace_id_from_kwargs: user.current_workspace_id={ws_id}") if ws_id: return ws_id - # 从 payload/data/body 获取（仅当上述方式都失败时） - data = kwargs.get("data") or kwargs.get("body") or kwargs.get("payload") - if data and hasattr(data, "workspace_id"): - ws_id = data.workspace_id - if ws_id: - logger.info(f"_get_workspace_id_from_kwargs: 从 payload 获取: {ws_id}") - return ws_id - - # 从 api_key_auth 获取（API Key 认证方式） - api_key_auth = kwargs.get("api_key_auth") - if api_key_auth and hasattr(api_key_auth, 'workspace_id'): - ws_id = api_key_auth.workspace_id - logger.info(f"_get_workspace_id_from_kwargs: 从 api_key_auth 获取: {ws_id}") - return ws_id - - logger.warning(f"_get_workspace_id_from_kwargs: 无法从 kwargs 获取 workspace_id, keys={list(kwargs.keys())}") + logger.warning(f"无法获取 workspace_id, kwargs keys: {list(kwargs.keys())}") return None From b0bb5e883cbf5aa1fbc963185475992f88f03b9c Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Wed, 22 Apr 2026 20:05:45 +0800 Subject: [PATCH 191/245] refactor(workflow/llm): replace regex substitution with string replace for context rendering --- api/app/core/workflow/nodes/llm/node.py | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/api/app/core/workflow/nodes/llm/node.py b/api/app/core/workflow/nodes/llm/node.py index db7f1009..352e735d 100644 --- a/api/app/core/workflow/nodes/llm/node.py +++ b/api/app/core/workflow/nodes/llm/node.py @@ -5,7 +5,6 @@ LLM 节点实现 """ import logging -import re from typing import Any from langchain_core.messages import AIMessage @@ -81,7 +80,7 @@ class LLMNode(BaseNode): def _render_context(self, message: str, variable_pool: VariablePool): context = f"{self._render_template(self.typed_config.context, variable_pool)}" - return re.sub(r"{{context}}", context, message) + return message.replace("{{context}}", context) async def _prepare_llm( self, From 46dfd92a9f61ca9c5fadfc90708027bdbb8cd7a6 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Wed, 22 Apr 2026 20:23:05 +0800 Subject: [PATCH 192/245] feat(api-key): adjust default rate limit and daily request limit values --- api/app/schemas/api_key_schema.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/api/app/schemas/api_key_schema.py b/api/app/schemas/api_key_schema.py index c7ca1e55..37245aa6 100644 --- a/api/app/schemas/api_key_schema.py +++ b/api/app/schemas/api_key_schema.py @@ -15,8 +15,8 @@ class ApiKeyCreate(BaseModel): type: ApiKeyType = Field(..., description="API Key 类型") scopes: List[str] = Field(default_factory=list, description="权限范围列表") resource_id: Optional[uuid.UUID] = Field(None, description="关联资源ID") - rate_limit: Optional[int] = Field(100, ge=1, le=1000, description="QPS限制（请求/秒）") - daily_request_limit: Optional[int] = Field(10000, description="日请求限制", ge=1) + rate_limit: Optional[int] = Field(50, ge=1, le=1000, description="QPS限制（请求/秒）") + daily_request_limit: Optional[int] = Field(100000, description="日请求限制", ge=1) quota_limit: Optional[int] = Field(None, description="配额限制（总请求数）", ge=1) expires_at: Optional[datetime.datetime] = Field(None, description="过期时间") @@ -55,7 +55,7 @@ class ApiKeyUpdate(BaseModel): description: Optional[str] = Field(None, description="描述") scopes: Optional[List[str]] = Field(None, description="权限范围列表") rate_limit: Optional[int] = Field(None, description="速率限制（请求/分钟）", ge=1) - daily_request_limit: Optional[int] = Field(10000, description="每日请求数限制", ge=1) + daily_request_limit: Optional[int] = Field(100000, description="每日请求数限制", ge=1) quota_limit: Optional[int] = Field(None, description="配额限制（总请求数）", ge=1) is_active: Optional[bool] = Field(None, description="是否激活") expires_at: Optional[datetime.datetime] = Field(None, description="过期时间") From 2a514a9e041e7ae39f08500835203958596b5117 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 22 Apr 2026 22:03:31 +0800 Subject: [PATCH 193/245] feat(web): create api support rate_limit & daily_request_limit config --- web/src/components/RbSlider/index.tsx | 5 +- .../components/ApiKeyDetailModal.tsx | 20 +++- .../components/ApiKeyModal.tsx | 40 +++++++- web/src/views/ApplicationConfig/Api.tsx | 94 ++++++++++--------- .../components/ApiKeyModal.tsx | 35 +++++++ 5 files changed, 140 insertions(+), 54 deletions(-) diff --git a/web/src/components/RbSlider/index.tsx b/web/src/components/RbSlider/index.tsx index 70b9fa84..5e90e419 100644 --- a/web/src/components/RbSlider/index.tsx +++ b/web/src/components/RbSlider/index.tsx @@ -17,6 +17,7 @@ import { type FC, type ReactNode, useEffect, useState } from 'react'; import { Slider, type SliderSingleProps, Flex, InputNumber, type InputNumberProps } from 'antd'; +import { useTranslation } from 'react-i18next'; /** Props interface for RbSlider component */ interface RbSliderProps extends SliderSingleProps { @@ -41,12 +42,13 @@ const RbSlider: FC = ({ step = 0.01, size = 'default' , isInput = false, - className = '', + className = 'rb:pl-1!', prefix, inputClassName, disabled, ...rest }) => { + const { t } = useTranslation() const [curValue, setCurValue] = useState(0) useEffect(() => { setCurValue(value) @@ -102,6 +104,7 @@ const RbSlider: FC = ({ onChange={handleInputChange} prefix={prefix} className={`${inputClassName || '' } rb:w-20!`} + placeholder={t('common.pleaseEnter')} /> :
{curValue || min}
} diff --git a/web/src/views/ApiKeyManagement/components/ApiKeyDetailModal.tsx b/web/src/views/ApiKeyManagement/components/ApiKeyDetailModal.tsx index f9e1df51..f7dd9cbb 100644 --- a/web/src/views/ApiKeyManagement/components/ApiKeyDetailModal.tsx +++ b/web/src/views/ApiKeyManagement/components/ApiKeyDetailModal.tsx @@ -106,16 +106,28 @@ const ApiKeyDetailModal = forwardRef
- {data.expires_at && <> -
{t('apiKey.advancedSettings')}
+
{t('apiKey.advancedSettings')}
+ {data.expires_at &&
{t(`apiKey.expires_at`)} {data.expires_at ? formatDateTime(data.expires_at as number, 'YYYY-MM-DD HH:mm:ss') : '-'} -
- } +
+ } +
+ {t(`application.qpsLimit`)} + + {data.rate_limit} {t('application.qpsLimitUnit')} + +
+
+ {t(`application.dailyUsageLimit`)} + + {data.daily_request_limit} {t('application.dailyUsageLimitUnit')} + +
); }); diff --git a/web/src/views/ApiKeyManagement/components/ApiKeyModal.tsx b/web/src/views/ApiKeyManagement/components/ApiKeyModal.tsx index 05e73992..68f95d0b 100644 --- a/web/src/views/ApiKeyManagement/components/ApiKeyModal.tsx +++ b/web/src/views/ApiKeyManagement/components/ApiKeyModal.tsx @@ -13,6 +13,7 @@ import type { ApiKey, ApiKeyModalRef } from '../types'; import RbModal from '@/components/RbModal' import { createApiKey, updateApiKey } from '@/api/apiKey'; import { stringRegExp } from '@/utils/validator'; +import RbSlider from '@/components/RbSlider' const FormItem = Form.Item; @@ -57,11 +58,10 @@ const ApiKeyModal = forwardRef(({ */ const handleOpen = (apiKey?: ApiKey) => { if (apiKey?.id) { - const { scopes = [], expires_at } = apiKey + const { scopes = [], expires_at, ...rest } = apiKey // Edit mode - populate form with existing data form.setFieldsValue({ - name: apiKey.name, - description: apiKey.description, + ...rest, memory: scopes.includes('memory'), rag: scopes.includes('rag'), expires_at: expires_at ? dayjs(expires_at) : undefined @@ -126,6 +126,10 @@ const ApiKeyModal = forwardRef(({
{t('apiKey.baseInfo')}
(({ disabledDate={(current) => current && current < dayjs().subtract(1, 'day').endOf('day')} /> + {t(`application.qpsLimit`)}({t('application.qpsLimitTip')}, {t('application.qpsLimitUnit')})} + extra={t('application.qpsLimitDesc')} + rules={[ + { required: true, message: t('common.pleaseEnter') }, + ]} + > + + + {t(`application.dailyUsageLimit`)} ({t('application.dailyUsageLimitUnit')})} + extra={t('application.dailyUsageLimitDesc')} + rules={[ + { required: true, message: t('common.pleaseEnter') }, + ]} + > + +
); diff --git a/web/src/views/ApplicationConfig/Api.tsx b/web/src/views/ApplicationConfig/Api.tsx index 4fa19c3e..b871d3bd 100644 --- a/web/src/views/ApplicationConfig/Api.tsx +++ b/web/src/views/ApplicationConfig/Api.tsx @@ -195,53 +195,55 @@ const Api: FC<{ application: Application | null }> = ({ application }) => { {/* API Key List */} - {apiKeyList.sort((a, b) => b.created_at - a.created_at).map(item => ( -
- - -
{item.name}
-
ID: {item.id}
- - -
handleEdit(item)} - >
-
handleDelete(item)} - >
- - - - - - - -
{item.total_requests}
-
{t('application.apiKeyRequestTotal')}
- - -
{item.rate_limit}
-
{t('application.qpsLimit')}
- - - - - - {maskApiKeys(item.api_key)} - - + + {apiKeyList.sort((a, b) => b.created_at - a.created_at).map(item => ( +
+ + +
{item.name}
+
ID: {item.id}
 - - -
- ))} + +
handleEdit(item)} + >
+
handleDelete(item)} + >
+ + + + + + + +
{item.total_requests}
+
{t('application.apiKeyRequestTotal')}
+ + +
{item.rate_limit}
+
{t('application.qpsLimit')}
+ + + + + + {maskApiKeys(item.api_key)} + + + + + +
+ ))} + diff --git a/web/src/views/ApplicationConfig/components/ApiKeyModal.tsx b/web/src/views/ApplicationConfig/components/ApiKeyModal.tsx index b43f0e4a..6de18781 100644 --- a/web/src/views/ApplicationConfig/components/ApiKeyModal.tsx +++ b/web/src/views/ApplicationConfig/components/ApiKeyModal.tsx @@ -17,6 +17,7 @@ import type { Application } from '@/views/ApplicationManagement/types' import type { ApiKeyModalRef } from '../types' import { createApiKey } from '@/api/apiKey'; import RbModal from '@/components/RbModal' +import RbSlider from '@/components/RbSlider' const FormItem = Form.Item; @@ -97,6 +98,10 @@ const ApiKeyModal = forwardRef(({ form={form} layout="vertical" scrollToFirstError={{ behavior: 'instant', block: 'end', focus: true }} + initialValues={{ + rate_limit: 50, + daily_request_limit: 100000 + }} > {/* Key name */} (({ > + {t(`application.qpsLimit`)}({t('application.qpsLimitTip')}, {t('application.qpsLimitUnit')})} + extra={t('application.qpsLimitDesc')} + rules={[ + { required: true, message: t('common.pleaseEnter') }, + ]} + > + + + {t(`application.dailyUsageLimit`)} ({t('application.dailyUsageLimitUnit')})} + extra={t('application.dailyUsageLimitDesc')} + rules={[ + { required: true, message: t('common.pleaseEnter') }, + ]} + > + + ); From 52bccfaedeb3ec0e7e4333d86f5bee9f29421e9b Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 22 Apr 2026 23:14:43 +0800 Subject: [PATCH 194/245] fix(web): i18n update --- web/src/i18n/en.ts | 2 +- web/src/i18n/zh.ts | 2 +- web/src/views/ApplicationConfig/Agent.tsx | 5 ----- web/src/views/Workflow/hooks/useWorkflowGraph.ts | 2 +- 4 files changed, 3 insertions(+), 8 deletions(-) diff --git a/web/src/i18n/en.ts b/web/src/i18n/en.ts index 4d284905..e7f71f5f 100644 --- a/web/src/i18n/en.ts +++ b/web/src/i18n/en.ts @@ -3096,7 +3096,7 @@ Memory Bear: After the rebellion, regional warlordism intensified for several re editPackage: 'Edit Package', viewDetail: 'View full package details', - noLimit: 'Unlimited', + noLimit: 'Infinite', }, }, }; diff --git a/web/src/i18n/zh.ts b/web/src/i18n/zh.ts index aea3b965..2c484846 100644 --- a/web/src/i18n/zh.ts +++ b/web/src/i18n/zh.ts @@ -3060,7 +3060,7 @@ export const zh = { editPackage: '编辑套餐', viewDetail: '查看完整套餐详情', - noLimit: '不限制', + noLimit: '无限', }, }, } \ No newline at end of file diff --git a/web/src/views/ApplicationConfig/Agent.tsx b/web/src/views/ApplicationConfig/Agent.tsx index b694d1eb..4d86fba7 100644 --- a/web/src/views/ApplicationConfig/Agent.tsx +++ b/web/src/views/ApplicationConfig/Agent.tsx @@ -62,7 +62,6 @@ const Agent = forwardRef(null); const modelConfigModalRef = useRef(null) const [modelList, setModelList] = useState([]) @@ -94,7 +93,6 @@ const Agent = forwardRef { - setLoading(true) getApplicationConfig(id as string).then(res => { const response = res as Config const { skills, variables } = response @@ -127,8 +125,6 @@ const Agent = forwardRef { - setLoading(false) }) } @@ -421,7 +417,6 @@ const Agent = forwardRef - {loading && }
diff --git a/web/src/views/Workflow/hooks/useWorkflowGraph.ts b/web/src/views/Workflow/hooks/useWorkflowGraph.ts index 535a4eb0..efc637c2 100644 --- a/web/src/views/Workflow/hooks/useWorkflowGraph.ts +++ b/web/src/views/Workflow/hooks/useWorkflowGraph.ts @@ -451,11 +451,11 @@ export const useWorkflowGraph = ({ graphRef.current.addEdges(edgeList.filter(vo => vo !== null)) } + graphRef.current.centerContent() // Initialize after completion, display nodes in visible area if (nodes.length > 0 || edges.length > 0) { setTimeout(() => { if (graphRef.current) { - graphRef.current.centerContent() graphRef.current.getNodes().forEach(node => { if (!node.getData()?.cycle) node.toFront(); }); From 9480a612298234afd886a99d631f7842f2cc2393 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Thu, 23 Apr 2026 00:07:29 +0800 Subject: [PATCH 195/245] fix(web): user profile Co-authored-by: Copilot --- .../components/EndUserProfile.tsx | 15 +++++++++++---- web/src/views/UserMemoryDetail/types.ts | 2 +- 2 files changed, 12 insertions(+), 5 deletions(-) diff --git a/web/src/views/UserMemoryDetail/components/EndUserProfile.tsx b/web/src/views/UserMemoryDetail/components/EndUserProfile.tsx index 702d152e..e3aac644 100644 --- a/web/src/views/UserMemoryDetail/components/EndUserProfile.tsx +++ b/web/src/views/UserMemoryDetail/components/EndUserProfile.tsx @@ -31,6 +31,13 @@ interface EndUserProfileProps { className?: string; } +const formatValue = (value: string | string[] | null | undefined) => { + if (!value) return '-' + if (Array.isArray(value)) { + return value.length ? value.join(' | ') : '-' + } + return value +} const EndUserProfile = forwardRef(({ className, onDataLoaded }, ref) => { const { t } = useTranslation() const { id } = useParams() @@ -89,19 +96,19 @@ const EndUserProfile = forwardRef(({ cla
{t('userMemory.role')}
-
{data?.profile?.role?.join(' | ') || '-'}
+
{formatValue(data?.profile?.role)}
{t('userMemory.domain')}
-
{data?.profile?.domain?.join(' | ') || '-'}
+
{formatValue(data?.profile?.domain)}
{t('userMemory.expertise')}
-
{data?.profile?.expertise?.join(' | ') || '-'}
+
{formatValue(data?.profile?.expertise)}
{t('userMemory.interests')}
-
{data?.profile?.interests?.join(' | ') || '-'}
+
{formatValue(data?.profile?.interests)}
diff --git a/web/src/views/UserMemoryDetail/types.ts b/web/src/views/UserMemoryDetail/types.ts index 667d8272..8efbd890 100644 --- a/web/src/views/UserMemoryDetail/types.ts +++ b/web/src/views/UserMemoryDetail/types.ts @@ -178,7 +178,7 @@ export interface EndUser { created_at: string; updated_at: string; profile: { - role: string[]; + role: string[] | string; domain: string[]; expertise: string[]; interests: string[]; From eaa66ba71aad5a26ee0929e582468105c55eabae Mon Sep 17 00:00:00 2001 From: wwq Date: Thu, 23 Apr 2026 00:14:29 +0800 Subject: [PATCH 196/245] fix(quota_manager): retrieve workspace_id from api_key_auth context - Add logic to resolve the workspace ID derived from the API key authentication context. --- api/app/core/quota_manager.py | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/api/app/core/quota_manager.py b/api/app/core/quota_manager.py index 6f1be886..d59c42e0 100644 --- a/api/app/core/quota_manager.py +++ b/api/app/core/quota_manager.py @@ -37,6 +37,11 @@ def _get_workspace_id_from_kwargs(kwargs: dict): if workspace_id: return workspace_id + # 从 api_key_auth.workspace_id 获取（API Key 认证场景） + api_key_auth = kwargs.get("api_key_auth") + if api_key_auth and hasattr(api_key_auth, 'workspace_id'): + return api_key_auth.workspace_id + # 从 user.current_workspace_id 获取 user = _get_user_from_kwargs(kwargs) if user: From 17905196c99ce9bf1a9ded0a3ceaba3f84866f82 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Thu, 23 Apr 2026 14:50:05 +0800 Subject: [PATCH 197/245] fix(web): memory ui --- .../images/userMemory/memoryInsight.svg | 33 +++++-------------- .../userMemory/memoryInsight_active.svg | 2 +- web/src/views/UserMemoryDetail/Neo4j.tsx | 16 ++++----- 3 files changed, 17 insertions(+), 34 deletions(-) diff --git a/web/src/assets/images/userMemory/memoryInsight.svg b/web/src/assets/images/userMemory/memoryInsight.svg index 7dfa3dcf..84baf7e0 100644 --- a/web/src/assets/images/userMemory/memoryInsight.svg +++ b/web/src/assets/images/userMemory/memoryInsight.svg @@ -1,29 +1,12 @@ - - 编组 26 - - - - - - - - - - - - - - - - - - - - - - - + + 热点洞察 + + + + + + diff --git a/web/src/assets/images/userMemory/memoryInsight_active.svg b/web/src/assets/images/userMemory/memoryInsight_active.svg index 43c73a4b..94af6953 100644 --- a/web/src/assets/images/userMemory/memoryInsight_active.svg +++ b/web/src/assets/images/userMemory/memoryInsight_active.svg @@ -2,7 +2,7 @@ 热点洞察 - + diff --git a/web/src/views/UserMemoryDetail/Neo4j.tsx b/web/src/views/UserMemoryDetail/Neo4j.tsx index 4de310b3..b3d7fff7 100644 --- a/web/src/views/UserMemoryDetail/Neo4j.tsx +++ b/web/src/views/UserMemoryDetail/Neo4j.tsx @@ -98,8 +98,8 @@ const Neo4j: FC = () => { align="center" justify="center" className={clsx("rb:cursor-pointer rb:size-12 rb:rounded-xl rb:group", { - 'rb:bg-[#155EEF]': selectedKey === 'userProfile', - 'rb:hover:bg-[#F0F3F8]': selectedKey !== 'userProfile', + 'rb:bg-[#171719]': selectedKey === 'userProfile', + 'rb:hover:bg-[#EBEBEB]': selectedKey !== 'userProfile', })} onClick={(e) => onOpenChange(e, 'userProfile')} > @@ -113,8 +113,8 @@ const Neo4j: FC = () => { align="center" justify="center" className={clsx("rb:cursor-pointer rb:size-12 rb:rounded-xl rb:group", { - 'rb:bg-[#155EEF]': selectedKey === 'aboutMe', - 'rb:hover:bg-[#F0F3F8]': selectedKey !== 'aboutMe', + 'rb:bg-[#171719]': selectedKey === 'aboutMe', + 'rb:hover:bg-[#EBEBEB]': selectedKey !== 'aboutMe', })} onClick={(e) => onOpenChange(e, 'aboutMe')} > @@ -128,8 +128,8 @@ const Neo4j: FC = () => { align="center" justify="center" className={clsx("rb:cursor-pointer rb:size-12 rb:rounded-xl rb:group", { - 'rb:bg-[#155EEF]': selectedKey === 'interestDistribution', - 'rb:hover:bg-[#F0F3F8]': selectedKey !== 'interestDistribution', + 'rb:bg-[#171719]': selectedKey === 'interestDistribution', + 'rb:hover:bg-[#EBEBEB]': selectedKey !== 'interestDistribution', })} onClick={(e) => onOpenChange(e, 'interestDistribution')} > @@ -143,8 +143,8 @@ const Neo4j: FC = () => { align="center" justify="center" className={clsx("rb:cursor-pointer rb:size-12 rb:rounded-xl rb:group", { - 'rb:bg-[#155EEF]': selectedKey === 'memoryInsight', - 'rb:hover:bg-[#F0F3F8]': selectedKey !== 'memoryInsight', + 'rb:bg-[#171719]': selectedKey === 'memoryInsight', + 'rb:hover:bg-[#EBEBEB]': selectedKey !== 'memoryInsight', })} onClick={(e) => onOpenChange(e, 'memoryInsight')} > From 5c836c90c94e46dde6bf98fe0c56482b0b716c81 Mon Sep 17 00:00:00 2001 From: miao <1468212639@qq.com> Date: Thu, 23 Apr 2026 12:05:31 +0800 Subject: [PATCH 198/245] feat(memory): add episodic memory pagination and semantic memory list API Split explicit memory overview into two independent endpoints: - GET /memory/explicit-memory/episodics: episodic memory paginated query with date range filter (millisecond timestamp) and episodic type filter using Neo4j datetime() for precise time comparison - GET /memory/explicit-memory/semantics: semantic memory full list query returns data as array directly Modified files: - api/app/controllers/memory_explicit_controller.py - api/app/services/memory_explicit_service.py --- .../controllers/memory_explicit_controller.py | 139 +++++++++++- api/app/services/memory_explicit_service.py | 206 +++++++++++++++++- 2 files changed, 342 insertions(+), 3 deletions(-) diff --git a/api/app/controllers/memory_explicit_controller.py b/api/app/controllers/memory_explicit_controller.py index c52f308c..90758dc7 100644 --- a/api/app/controllers/memory_explicit_controller.py +++ b/api/app/controllers/memory_explicit_controller.py @@ -4,7 +4,10 @@ 处理显性记忆相关的API接口，包括情景记忆和语义记忆的查询。 """ -from fastapi import APIRouter, Depends +from datetime import date +from typing import Optional + +from fastapi import APIRouter, Depends, Query from app.core.logging_config import get_api_logger from app.core.response_utils import success, fail @@ -69,6 +72,140 @@ async def get_explicit_memory_overview_api( return fail(BizCode.INTERNAL_ERROR, "显性记忆总览查询失败", str(e)) +@router.get("/episodics", response_model=ApiResponse) +async def get_episodic_memory_list_api( + end_user_id: str = Query(..., description="end user ID"), + page: int = Query(1, gt=0, description="page number, starting from 1"), + pagesize: int = Query(10, gt=0, le=100, description="number of items per page, max 100"), + start_date: Optional[int] = Query(None, description="start timestamp (ms)"), + end_date: Optional[int] = Query(None, description="end timestamp (ms)"), + episodic_type: str = Query("all", description="episodic type ：all/conversation/project_work/learning/decision/important_event"), + current_user: User = Depends(get_current_user), +) -> dict: + """ + 获取情景记忆分页列表 + + 返回指定用户的情景记忆列表，支持分页、时间范围筛选和情景类型筛选。 + + Args: + end_user_id: 终端用户ID（必填） + page: 页码（从1开始，默认1） + pagesize: 每页数量（默认10，最大100） + start_date: 开始时间戳（可选，毫秒），自动扩展到当天 00:00:00 + end_date: 结束时间戳（可选，毫秒），自动扩展到当天 23:59:59 + episodic_type: 情景类型筛选（可选，默认all） + current_user: 当前用户 + + Returns: + ApiResponse: 包含情景记忆分页列表 + + Examples: + - 基础分页查询：GET /episodic-list?end_user_id=xxx&page=1&pagesize=5 + 返回第1页，每页5条数据 + - 按时间范围筛选：GET /episodic-list?end_user_id=xxx&page=1&pagesize=5&start_date=1738684800000&end_date=1738771199000 + 返回指定时间范围内的数据 + - 按情景类型筛选：GET /episodic-list?end_user_id=xxx&page=1&pagesize=5&episodic_type=important_event + 返回类型为"重要事件"的数据 + + Notes: + - start_date 和 end_date 必须同时提供或同时不提供 + - start_date 不能大于 end_date + - episodic_type 可选值：all, conversation, project_work, learning, decision, important_event + - total 为该用户情景记忆总数（不受筛选条件影响） + - page.total 为筛选后的总条数 + """ + workspace_id = current_user.current_workspace_id + + # 检查用户是否已选择工作空间 + if workspace_id is None: + api_logger.warning(f"用户 {current_user.username} 尝试查询情景记忆列表但未选择工作空间") + return fail(BizCode.INVALID_PARAMETER, "请先切换到一个工作空间", "current_workspace_id is None") + + api_logger.info( + f"情景记忆分页查询: end_user_id={end_user_id}, " + f"start_date={start_date}, end_date={end_date}, episodic_type={episodic_type}, " + f"page={page}, pagesize={pagesize}, username={current_user.username}" + ) + + # 1. 参数校验 + if page < 1 or pagesize < 1: + api_logger.warning(f"分页参数错误: page={page}, pagesize={pagesize}") + return fail(BizCode.INVALID_PARAMETER, "分页参数必须大于0") + + valid_episodic_types = ["all", "conversation", "project_work", "learning", "decision", "important_event"] + if episodic_type not in valid_episodic_types: + api_logger.warning(f"无效的情景类型参数: {episodic_type}") + return fail(BizCode.INVALID_PARAMETER, f"无效的情景类型参数，可选值：{', '.join(valid_episodic_types)}") + + # 时间戳参数校验 + if (start_date is not None and end_date is None) or (end_date is not None and start_date is None): + return fail(BizCode.INVALID_PARAMETER, "start_date和end_date必须同时提供") + + if start_date is not None and end_date is not None and start_date > end_date: + return fail(BizCode.INVALID_PARAMETER, "start_date不能大于end_date") + + # 2. 执行查询 + try: + result = await memory_explicit_service.get_episodic_memory_list( + end_user_id=end_user_id, + page=page, + pagesize=pagesize, + start_date=start_date, + end_date=end_date, + episodic_type=episodic_type, + ) + api_logger.info( + f"情景记忆分页查询成功: end_user_id={end_user_id}, " + f"total={result['total']}, 返回={len(result['items'])}条" + ) + except Exception as e: + api_logger.error(f"情景记忆分页查询失败: end_user_id={end_user_id}, error={str(e)}") + return fail(BizCode.INTERNAL_ERROR, "情景记忆分页查询失败", str(e)) + + # 3. 返回结构化响应 + return success(data=result, msg="查询成功") + +@router.get("/semantics", response_model=ApiResponse) +async def get_semantic_memory_list_api( + end_user_id: str = Query(..., description="终端用户ID"), + current_user: User = Depends(get_current_user), +) -> dict: + """ + 获取语义记忆列表 + + 返回指定用户的全量语义记忆列表。 + + Args: + end_user_id: 终端用户ID（必填） + current_user: 当前用户 + + Returns: + ApiResponse: 包含语义记忆全量列表 + """ + workspace_id = current_user.current_workspace_id + + if workspace_id is None: + api_logger.warning(f"用户 {current_user.username} 尝试查询语义记忆列表但未选择工作空间") + return fail(BizCode.INVALID_PARAMETER, "请先切换到一个工作空间", "current_workspace_id is None") + + api_logger.info( + f"语义记忆列表查询: end_user_id={end_user_id}, username={current_user.username}" + ) + + try: + result = await memory_explicit_service.get_semantic_memory_list( + end_user_id=end_user_id + ) + api_logger.info( + f"语义记忆列表查询成功: end_user_id={end_user_id}, total={len(result)}" + ) + except Exception as e: + api_logger.error(f"语义记忆列表查询失败: end_user_id={end_user_id}, error={str(e)}") + return fail(BizCode.INTERNAL_ERROR, "语义记忆列表查询失败", str(e)) + + return success(data=result, msg="查询成功") + + @router.post("/details", response_model=ApiResponse) async def get_explicit_memory_details_api( request: ExplicitMemoryDetailsRequest, diff --git a/api/app/services/memory_explicit_service.py b/api/app/services/memory_explicit_service.py index f8d39ae8..c5ab2e38 100644 --- a/api/app/services/memory_explicit_service.py +++ b/api/app/services/memory_explicit_service.py @@ -4,7 +4,8 @@ 处理显性记忆相关的业务逻辑，包括情景记忆和语义记忆的查询。 """ -from typing import Any, Dict +from datetime import date +from typing import Any, Dict, Optional from app.core.logging_config import get_logger from app.services.memory_base_service import MemoryBaseService @@ -104,7 +105,7 @@ class MemoryExplicitService(MemoryBaseService): e.description AS core_definition ORDER BY e.name ASC """ - + semantic_result = await self.neo4j_connector.execute_query( semantic_query, end_user_id=end_user_id @@ -146,6 +147,207 @@ class MemoryExplicitService(MemoryBaseService): logger.error(f"获取显性记忆总览时出错: {str(e)}", exc_info=True) raise + + async def get_episodic_memory_list( + self, + end_user_id: str, + page: int, + pagesize: int, + start_date: Optional[int] = None, + end_date: Optional[int] = None, + episodic_type: str = "all", + ) -> Dict[str, Any]: + """ + 获取情景记忆分页列表 + + Args: + end_user_id: 终端用户ID + page: 页码 + pagesize: 每页数量 + start_date: 开始时间戳（毫秒），可选 + end_date: 结束时间戳（毫秒），可选 + episodic_type: 情景类型筛选 + + Returns: + { + "total": int, # 该用户情景记忆总数（不受筛选影响） + "items": [...], # 当前页数据 + "page": { + "page": int, + "pagesize": int, + "total": int, # 筛选后总数 + "hasnext": bool + } + } + """ + try: + logger.info( + f"情景记忆分页查询: end_user_id={end_user_id}, " + f"start_date={start_date}, end_date={end_date}, " + f"episodic_type={episodic_type}, page={page}, pagesize={pagesize}" + ) + + # 1. 查询情景记忆总数（不受筛选条件限制） + total_all_query = """ + MATCH (s:MemorySummary) + WHERE s.end_user_id = $end_user_id + RETURN count(s) AS total + """ + total_all_result = await self.neo4j_connector.execute_query( + total_all_query, end_user_id=end_user_id + ) + total_all = total_all_result[0]["total"] if total_all_result else 0 + + # 2. 构建筛选条件 + where_clauses = ["s.end_user_id = $end_user_id"] + params = {"end_user_id": end_user_id} + + # 时间戳筛选（毫秒时间戳转为 ISO 字符串，使用 Neo4j datetime() 精确比较） + if start_date is not None and end_date is not None: + from datetime import datetime + start_dt = datetime.fromtimestamp(start_date / 1000) + end_dt = datetime.fromtimestamp(end_date / 1000) + # 开始时间取当天 00:00:00，结束时间取当天 23:59:59.999999 + start_iso = start_dt.strftime("%Y-%m-%dT") + "00:00:00.000000" + end_iso = end_dt.strftime("%Y-%m-%dT") + "23:59:59.999999" + + where_clauses.append("datetime(s.created_at) >= datetime($start_iso) AND datetime(s.created_at) <= datetime($end_iso)") + params["start_iso"] = start_iso + params["end_iso"] = end_iso + + # 类型筛选下推到 Cypher（兼容中英文） + if episodic_type != "all": + type_mapping = { + "conversation": "对话", + "project_work": "项目/工作", + "learning": "学习", + "decision": "决策", + "important_event": "重要事件" + } + chinese_type = type_mapping.get(episodic_type) + if chinese_type: + where_clauses.append( + "(s.memory_type = $episodic_type OR s.memory_type = $chinese_type)" + ) + params["episodic_type"] = episodic_type + params["chinese_type"] = chinese_type + else: + where_clauses.append("s.memory_type = $episodic_type") + params["episodic_type"] = episodic_type + + where_str = " AND ".join(where_clauses) + + # 3. 查询筛选后的总数 + count_query = f""" + MATCH (s:MemorySummary) + WHERE {where_str} + RETURN count(s) AS total + """ + count_result = await self.neo4j_connector.execute_query(count_query, **params) + filtered_total = count_result[0]["total"] if count_result else 0 + + # 4. 查询分页数据 + skip = (page - 1) * pagesize + data_query = f""" + MATCH (s:MemorySummary) + WHERE {where_str} + RETURN elementId(s) AS id, + s.name AS title, + s.memory_type AS memory_type, + s.content AS content, + s.created_at AS created_at + ORDER BY s.created_at DESC + SKIP {skip} LIMIT {pagesize} + """ + + result = await self.neo4j_connector.execute_query(data_query, **params) + + # 5. 处理结果 + items = [] + if result: + for record in result: + raw_created_at = record.get("created_at") + created_at_timestamp = self.parse_timestamp(raw_created_at) + items.append({ + "id": record["id"], + "title": record.get("title") or "未命名", + "memory_type": record.get("memory_type") or "其他", + "content": record.get("content") or "", + "created_at": created_at_timestamp + }) + + # 6. 构建返回结果 + return { + "total": total_all, + "items": items, + "page": { + "page": page, + "pagesize": pagesize, + "total": filtered_total, + "hasnext": (page * pagesize) < filtered_total + } + } + + except Exception as e: + logger.error(f"情景记忆分页查询出错: {str(e)}", exc_info=True) + raise + + async def get_semantic_memory_list( + self, + end_user_id: str + ) -> list: + """ + 获取语义记忆全量列表 + + Args: + end_user_id: 终端用户ID + + Returns: + [ + { + "id": str, + "name": str, + "entity_type": str, + "core_definition": str + } + ] + """ + try: + logger.info(f"语义记忆列表查询: end_user_id={end_user_id}") + + semantic_query = """ + MATCH (e:ExtractedEntity) + WHERE e.end_user_id = $end_user_id + AND e.is_explicit_memory = true + RETURN elementId(e) AS id, + e.name AS name, + e.entity_type AS entity_type, + e.description AS core_definition + ORDER BY e.name ASC + """ + + result = await self.neo4j_connector.execute_query( + semantic_query, end_user_id=end_user_id + ) + + items = [] + if result: + for record in result: + items.append({ + "id": record["id"], + "name": record.get("name") or "未命名", + "entity_type": record.get("entity_type") or "未分类", + "core_definition": record.get("core_definition") or "" + }) + + logger.info(f"语义记忆列表查询成功: end_user_id={end_user_id}, total={len(items)}") + + return items + + except Exception as e: + logger.error(f"语义记忆列表查询出错: {str(e)}", exc_info=True) + raise + async def get_explicit_memory_details( self, end_user_id: str, From bf9a3503de19ec3291b423cbf4d6d80648eff6cd Mon Sep 17 00:00:00 2001 From: miao <1468212639@qq.com> Date: Thu, 23 Apr 2026 15:23:39 +0800 Subject: [PATCH 199/245] feat(memory-api): add memory detail external service APIs Add external service APIs for memory detail queries Provides memory data access endpoints for external service integration Add utility functions for API key user resolution and end_user validation Modified files: - api/app/controllers/service/user_memory_api_controller.py - api/app/core/api_key_utils.py - api/app/controllers/service/__init__.py --- api/app/controllers/service/__init__.py | 2 + .../service/user_memory_api_controller.py | 197 ++++++++++++++++++ api/app/core/api_key_utils.py | 66 ++++++ 3 files changed, 265 insertions(+) create mode 100644 api/app/controllers/service/user_memory_api_controller.py diff --git a/api/app/controllers/service/__init__.py b/api/app/controllers/service/__init__.py index 52d4b732..850b496d 100644 --- a/api/app/controllers/service/__init__.py +++ b/api/app/controllers/service/__init__.py @@ -14,6 +14,7 @@ from . import ( rag_api_document_controller, rag_api_file_controller, rag_api_knowledge_controller, + user_memory_api_controller, ) # 创建 V1 API 路由器 @@ -28,5 +29,6 @@ service_router.include_router(rag_api_chunk_controller.router) service_router.include_router(memory_api_controller.router) service_router.include_router(end_user_api_controller.router) service_router.include_router(memory_config_api_controller.router) +service_router.include_router(user_memory_api_controller.router) __all__ = ["service_router"] diff --git a/api/app/controllers/service/user_memory_api_controller.py b/api/app/controllers/service/user_memory_api_controller.py new file mode 100644 index 00000000..6d2498c0 --- /dev/null +++ b/api/app/controllers/service/user_memory_api_controller.py @@ -0,0 +1,197 @@ +"""User Memory 服务接口 — 基于 API Key 认证 + +包装 user_memory_controllers.py 和 memory_agent_controller.py 中的内部接口， +提供基于 API Key 认证的对外服务: +1./analytics/graph_data - 知识图谱数据接口 +2./analytics/community_graph - 社区图谱接口 +3./analytics/node_statistics - 记忆节点统计接口 +4./analytics/user_summary - 用户摘要接口 +5./analytics/memory_insight - 记忆洞察接口 +6./analytics/interest_distribution - 兴趣分布接口 +7./analytics/end_user_info - 终端用户信息接口 + + +路由前缀: /memory +子路径: /analytics/... +最终路径: /v1/memory/analytics/... +认证方式: API Key (@require_api_key) +""" + +from typing import Optional + +from fastapi import APIRouter, Depends, Header, Query, Request +from sqlalchemy.orm import Session + +from app.core.api_key_auth import require_api_key +from app.core.api_key_utils import get_current_user_from_api_key, validate_end_user_in_workspace +from app.core.logging_config import get_business_logger +from app.db import get_db +from app.schemas.api_key_schema import ApiKeyAuth + +# 包装内部服务 controller +from app.controllers import user_memory_controllers, memory_agent_controller + +router = APIRouter(prefix="/memory", tags=["V1 - User Memory API"]) +logger = get_business_logger() + + +# ==================== 知识图谱 ==================== + + +@router.get("/analytics/graph_data") +@require_api_key(scopes=["memory"]) +async def get_graph_data( + request: Request, + end_user_id: str = Query(..., description="End user ID"), + node_types: Optional[str] = Query(None, description="Comma-separated node types filter"), + limit: int = Query(100, description="Max nodes to return, capped at 1000"), + depth: int = Query(1, description="Graph traversal depth, capped at 3"), + center_node_id: Optional[str] = Query(None, description="Center node for subgraph"), + api_key_auth: ApiKeyAuth = None, + db: Session = Depends(get_db), +): + """Get knowledge graph data (nodes + edges) for an end user.""" + current_user = get_current_user_from_api_key(db, api_key_auth) + validate_end_user_in_workspace(db, end_user_id, api_key_auth.workspace_id) + + return await user_memory_controllers.get_graph_data_api( + end_user_id=end_user_id, + node_types=node_types, + limit=limit, + depth=depth, + center_node_id=center_node_id, + current_user=current_user, + db=db, + ) + + +@router.get("/analytics/community_graph") +@require_api_key(scopes=["memory"]) +async def get_community_graph( + request: Request, + end_user_id: str = Query(..., description="End user ID"), + api_key_auth: ApiKeyAuth = None, + db: Session = Depends(get_db), +): + """Get community clustering graph for an end user.""" + current_user = get_current_user_from_api_key(db, api_key_auth) + validate_end_user_in_workspace(db, end_user_id, api_key_auth.workspace_id) + + return await user_memory_controllers.get_community_graph_data_api( + end_user_id=end_user_id, + current_user=current_user, + db=db, + ) + + +# ==================== 节点统计 ==================== + + +@router.get("/analytics/node_statistics") +@require_api_key(scopes=["memory"]) +async def get_node_statistics( + request: Request, + end_user_id: str = Query(..., description="End user ID"), + api_key_auth: ApiKeyAuth = None, + db: Session = Depends(get_db), +): + """Get memory node type statistics for an end user.""" + current_user = get_current_user_from_api_key(db, api_key_auth) + validate_end_user_in_workspace(db, end_user_id, api_key_auth.workspace_id) + + return await user_memory_controllers.get_node_statistics_api( + end_user_id=end_user_id, + current_user=current_user, + db=db, + ) + + +# ==================== 用户摘要 & 洞察 ==================== + + +@router.get("/analytics/user_summary") +@require_api_key(scopes=["memory"]) +async def get_user_summary( + request: Request, + end_user_id: str = Query(..., description="End user ID"), + language_type: str = Header(default=None, alias="X-Language-Type"), + api_key_auth: ApiKeyAuth = None, + db: Session = Depends(get_db), +): + """Get cached user summary for an end user.""" + current_user = get_current_user_from_api_key(db, api_key_auth) + validate_end_user_in_workspace(db, end_user_id, api_key_auth.workspace_id) + + return await user_memory_controllers.get_user_summary_api( + end_user_id=end_user_id, + language_type=language_type, + current_user=current_user, + db=db, + ) + + +@router.get("/analytics/memory_insight") +@require_api_key(scopes=["memory"]) +async def get_memory_insight( + request: Request, + end_user_id: str = Query(..., description="End user ID"), + api_key_auth: ApiKeyAuth = None, + db: Session = Depends(get_db), +): + """Get cached memory insight report for an end user.""" + current_user = get_current_user_from_api_key(db, api_key_auth) + validate_end_user_in_workspace(db, end_user_id, api_key_auth.workspace_id) + + return await user_memory_controllers.get_memory_insight_report_api( + end_user_id=end_user_id, + current_user=current_user, + db=db, + ) + + +# ==================== 兴趣分布 ==================== + + +@router.get("/analytics/interest_distribution") +@require_api_key(scopes=["memory"]) +async def get_interest_distribution( + request: Request, + end_user_id: str = Query(..., description="End user ID"), + limit: int = Query(5, le=5, description="Max interest tags to return"), + language_type: str = Header(default=None, alias="X-Language-Type"), + api_key_auth: ApiKeyAuth = None, + db: Session = Depends(get_db), +): + """Get interest distribution tags for an end user.""" + current_user = get_current_user_from_api_key(db, api_key_auth) + validate_end_user_in_workspace(db, end_user_id, api_key_auth.workspace_id) + + return await memory_agent_controller.get_interest_distribution_by_user_api( + end_user_id=end_user_id, + limit=limit, + language_type=language_type, + current_user=current_user, + db=db, + ) + + +# ==================== 终端用户信息 ==================== + + +@router.get("/analytics/end_user_info") +@require_api_key(scopes=["memory"]) +async def get_end_user_info( + request: Request, + end_user_id: str = Query(..., description="End user ID"), + api_key_auth: ApiKeyAuth = None, + db: Session = Depends(get_db), +): + """Get end user basic information (name, aliases, metadata).""" + current_user = get_current_user_from_api_key(db, api_key_auth) + validate_end_user_in_workspace(db, end_user_id, api_key_auth.workspace_id) + + return await user_memory_controllers.get_end_user_info( + end_user_id=end_user_id, + current_user=current_user, + db=db, + ) \ No newline at end of file diff --git a/api/app/core/api_key_utils.py b/api/app/core/api_key_utils.py index fb6b9552..290630ce 100644 --- a/api/app/core/api_key_utils.py +++ b/api/app/core/api_key_utils.py @@ -3,6 +3,12 @@ import secrets from typing import Optional, Union from datetime import datetime +from sqlalchemy.orm import Session as _Session +from app.core.error_codes import BizCode as _BizCode +from app.core.exceptions import BusinessException as _BusinessException +from app.models.end_user_model import EndUser as _EndUser +from app.repositories.end_user_repository import EndUserRepository as _EndUserRepository + from app.models.api_key_model import ApiKeyType from fastapi import Response from fastapi.responses import JSONResponse @@ -65,3 +71,63 @@ def datetime_to_timestamp(dt: Optional[datetime]) -> Optional[int]: return None return int(dt.timestamp() * 1000) + + +def get_current_user_from_api_key(db: _Session, api_key_auth): + """通过 API Key 构造 current_user 对象。 + + 从 API Key 反查创建者（管理员用户），并设置其 workspace 上下文。 + 与内部接口的 Depends(get_current_user) (JWT) 等价。 + + Args: + db: 数据库会话 + api_key_auth: API Key 认证信息（ApiKeyAuth） + + Returns: + User ORM 对象，已设置 current_workspace_id + """ + from app.services import api_key_service + + api_key = api_key_service.ApiKeyService.get_api_key( + db, api_key_auth.api_key_id, api_key_auth.workspace_id + ) + current_user = api_key.creator + current_user.current_workspace_id = api_key_auth.workspace_id + return current_user + + +def validate_end_user_in_workspace( + db: _Session, + end_user_id: str, + workspace_id, +) -> _EndUser: + """校验 end_user 是否存在且属于指定 workspace。 + + Args: + db: 数据库会话 + end_user_id: 终端用户 ID + workspace_id: 工作空间 ID（UUID 或字符串均可） + + Returns: + EndUser ORM 对象（校验通过时） + + Raises: + BusinessException(USER_NOT_FOUND): end_user 不存在 + BusinessException(PERMISSION_DENIED): end_user 不属于该 workspace + """ + end_user_repo = _EndUserRepository(db) + end_user = end_user_repo.get_end_user_by_id(end_user_id) + + if end_user is None: + raise _BusinessException( + "End user not found", + _BizCode.USER_NOT_FOUND, + ) + + if str(end_user.workspace_id) != str(workspace_id): + raise _BusinessException( + "End user does not belong to this workspace", + _BizCode.PERMISSION_DENIED, + ) + + return end_user \ No newline at end of file From aac89b172fe0aa7ba66ccb449ba8ba7cc67f060b Mon Sep 17 00:00:00 2001 From: miao <1468212639@qq.com> Date: Thu, 23 Apr 2026 15:26:49 +0800 Subject: [PATCH 200/245] fix(memory): remove unused date import and fix docstring route paths Remove unused rom datetime import date in controller and service Fix Examples route paths from /episodic-list to /episodics to match actual router --- api/app/controllers/memory_explicit_controller.py | 7 +++---- api/app/services/memory_explicit_service.py | 1 - 2 files changed, 3 insertions(+), 5 deletions(-) diff --git a/api/app/controllers/memory_explicit_controller.py b/api/app/controllers/memory_explicit_controller.py index 90758dc7..88877de3 100644 --- a/api/app/controllers/memory_explicit_controller.py +++ b/api/app/controllers/memory_explicit_controller.py @@ -4,7 +4,6 @@ 处理显性记忆相关的API接口，包括情景记忆和语义记忆的查询。 """ -from datetime import date from typing import Optional from fastapi import APIRouter, Depends, Query @@ -100,11 +99,11 @@ async def get_episodic_memory_list_api( ApiResponse: 包含情景记忆分页列表 Examples: - - 基础分页查询：GET /episodic-list?end_user_id=xxx&page=1&pagesize=5 + - 基础分页查询：GET /episodics?end_user_id=xxx&page=1&pagesize=5 返回第1页，每页5条数据 - - 按时间范围筛选：GET /episodic-list?end_user_id=xxx&page=1&pagesize=5&start_date=1738684800000&end_date=1738771199000 + - 按时间范围筛选：GET /episodics?end_user_id=xxx&page=1&pagesize=5&start_date=1738684800000&end_date=1738771199000 返回指定时间范围内的数据 - - 按情景类型筛选：GET /episodic-list?end_user_id=xxx&page=1&pagesize=5&episodic_type=important_event + - 按情景类型筛选：GET /episodics?end_user_id=xxx&page=1&pagesize=5&episodic_type=important_event 返回类型为"重要事件"的数据 Notes: diff --git a/api/app/services/memory_explicit_service.py b/api/app/services/memory_explicit_service.py index c5ab2e38..8da3b167 100644 --- a/api/app/services/memory_explicit_service.py +++ b/api/app/services/memory_explicit_service.py @@ -4,7 +4,6 @@ 处理显性记忆相关的业务逻辑，包括情景记忆和语义记忆的查询。 """ -from datetime import date from typing import Any, Dict, Optional from app.core.logging_config import get_logger From 404ce9f9ba2dd04029126d6c51902812787bbace Mon Sep 17 00:00:00 2001 From: wwq Date: Thu, 23 Apr 2026 15:46:12 +0800 Subject: [PATCH 201/245] feat(workflow): enhance HTTP request node with curl debugging support - Augment HTTP request node capabilities and add generated curl commands for easier debugging. feat(log): implement workflow execution logs and search functionality - Add detailed logging for workflow node execution and enable search capabilities within application logs. feat(auth): introduce middleware to verify application publication status - Add a check to ensure the application is published before allowing access. fix(converter): rectify variable handling logic in Dify converter - Correct issues related to processing variables within the Dify converter module. refactor(model): remove quota check decorator from model update operations - Decouple quota validation from the model update process to streamline the logic. --- api/app/controllers/app_log_controller.py | 10 +- api/app/controllers/model_controller.py | 1 - api/app/core/api_key_auth.py | 2 + api/app/core/error_codes.py | 1 + .../core/workflow/adapters/dify/converter.py | 20 ++- .../workflow/nodes/http_request/config.py | 15 +++ .../core/workflow/nodes/http_request/node.py | 106 +++++++++++++-- api/app/dependencies.py | 1 + .../repositories/conversation_repository.py | 27 ++-- api/app/schemas/app_log_schema.py | 14 ++ api/app/services/api_key_service.py | 12 ++ api/app/services/app_log_service.py | 127 ++++++++++++++++-- api/app/services/workflow_import_service.py | 4 +- 13 files changed, 305 insertions(+), 35 deletions(-) diff --git a/api/app/controllers/app_log_controller.py b/api/app/controllers/app_log_controller.py index 92b5becd..7732dd37 100644 --- a/api/app/controllers/app_log_controller.py +++ b/api/app/controllers/app_log_controller.py @@ -24,13 +24,15 @@ def list_app_logs( app_id: uuid.UUID, page: int = Query(1, ge=1), pagesize: int = Query(20, ge=1, le=100), - is_draft: Optional[bool] = None, + is_draft: bool = Query(False, description="是否草稿会话（默认false，即发布会话）"), + keyword: Optional[str] = Query(None, description="搜索关键词（匹配消息内容）"), db: Session = Depends(get_db), current_user=Depends(get_current_user), ): """查看应用下所有会话记录（分页） - 支持按 is_draft 筛选（草稿会话 / 发布会话） + - 支持按 keyword 搜索（匹配消息内容） - 按最新更新时间倒序排列 - 所有人（包括共享者和被共享者）都只能查看自己的会话记录 """ @@ -47,7 +49,8 @@ def list_app_logs( workspace_id=workspace_id, page=page, pagesize=pagesize, - is_draft=is_draft + is_draft=is_draft, + keyword=keyword ) items = [AppLogConversation.model_validate(c) for c in conversations] @@ -78,12 +81,13 @@ def get_app_log_detail( # 使用 Service 层查询 log_service = AppLogService(db) - conversation = log_service.get_conversation_detail( + conversation, node_executions_map = log_service.get_conversation_detail( app_id=app_id, conversation_id=conversation_id, workspace_id=workspace_id ) detail = AppLogConversationDetail.model_validate(conversation) + detail.node_executions_map = node_executions_map return success(data=detail) diff --git a/api/app/controllers/model_controller.py b/api/app/controllers/model_controller.py index 57c22337..4958152b 100644 --- a/api/app/controllers/model_controller.py +++ b/api/app/controllers/model_controller.py @@ -373,7 +373,6 @@ def delete_composite_model( @router.put("/{model_id}", response_model=ApiResponse) -@check_model_activation_quota def update_model( model_id: uuid.UUID, model_data: model_schema.ModelConfigUpdate, diff --git a/api/app/core/api_key_auth.py b/api/app/core/api_key_auth.py index 05bca945..448a0f26 100644 --- a/api/app/core/api_key_auth.py +++ b/api/app/core/api_key_auth.py @@ -70,6 +70,8 @@ def require_api_key( }) raise BusinessException("API Key 无效或已过期", BizCode.API_KEY_INVALID) + ApiKeyAuthService.check_app_published(db, api_key_obj) + if scopes: missing_scopes = [] for scope in scopes: diff --git a/api/app/core/error_codes.py b/api/app/core/error_codes.py index 77bce6b4..2917a203 100644 --- a/api/app/core/error_codes.py +++ b/api/app/core/error_codes.py @@ -66,6 +66,7 @@ class BizCode(IntEnum): PERMISSION_DENIED = 6010 INVALID_CONVERSATION = 6011 CONFIG_MISSING = 6012 + APP_NOT_PUBLISHED = 6013 # 模型（7xxx） MODEL_CONFIG_INVALID = 7001 diff --git a/api/app/core/workflow/adapters/dify/converter.py b/api/app/core/workflow/adapters/dify/converter.py index ad9312e1..b61c2118 100644 --- a/api/app/core/workflow/adapters/dify/converter.py +++ b/api/app/core/workflow/adapters/dify/converter.py @@ -155,8 +155,13 @@ class DifyConverter(BaseConverter): def replacer(match: re.Match) -> str: raw_name = match.group(1) - new_name = self.process_var_selector(raw_name) - return f"{{{{{new_name}}}}}" + try: + new_name = self.process_var_selector(raw_name) + if not new_name: + return match.group(0) + return f"{{{{{new_name}}}}}" + except Exception: + return match.group(0) return pattern.sub(replacer, content) @@ -600,8 +605,15 @@ class DifyConverter(BaseConverter): ] = self.trans_variable_format(content["value"]) else: if node_data["body"]["data"]: - body_content = (node_data["body"]["data"][0].get("value") or - self._process_list_variable_literal(node_data["body"]["data"][0].get("file"))) + data_entry = node_data["body"]["data"][0] + body_content = data_entry.get("value") + if not body_content and data_entry.get("file"): + body_content = self._process_list_variable_literal(data_entry.get("file")) + if not body_content: + body_content = "" + elif isinstance(body_content, str): + # Convert session variable format for JSON body + body_content = self.trans_variable_format(body_content) else: body_content = "" diff --git a/api/app/core/workflow/nodes/http_request/config.py b/api/app/core/workflow/nodes/http_request/config.py index 72474436..527a80ff 100644 --- a/api/app/core/workflow/nodes/http_request/config.py +++ b/api/app/core/workflow/nodes/http_request/config.py @@ -132,11 +132,21 @@ class HttpErrorDefaultTemplate(BaseModel): description="Default HTTP headers returned on error", ) + files: list = Field( + default_factory=list, + description="Default files list returned on error", + ) + output: str = Field( default="SUCCESS", description="HTTP response body", ) + curl: str = Field( + default="", + description="Default curl command returned on error", + ) + class HttpErrorHandleConfig(BaseModel): method: HttpErrorHandle = Field( @@ -272,6 +282,11 @@ class HttpRequestNodeOutput(BaseModel): description="HTTP response body", ) + curl: str = Field( + default="", + description="Equivalent curl command for the HTTP request", + ) + # files: list[File] = Field( # ... # ) diff --git a/api/app/core/workflow/nodes/http_request/node.py b/api/app/core/workflow/nodes/http_request/node.py index 783c230b..f08e0bc1 100644 --- a/api/app/core/workflow/nodes/http_request/node.py +++ b/api/app/core/workflow/nodes/http_request/node.py @@ -167,7 +167,8 @@ class HttpRequestNode(BaseNode): "status_code": VariableType.NUMBER, "headers": VariableType.OBJECT, "files": VariableType.ARRAY_FILE, - "output": VariableType.STRING + "output": VariableType.STRING, + "curl": VariableType.STRING } def _build_timeout(self) -> Timeout: @@ -255,9 +256,13 @@ class HttpRequestNode(BaseNode): case HttpContentType.NONE: return {} case HttpContentType.JSON: - content["json"] = json.loads(self._render_template( + rendered_body = self._render_template( self.typed_config.body.data, variable_pool - )) + ).strip() + if not rendered_body: + content["json"] = {} + else: + content["json"] = json.loads(rendered_body) case HttpContentType.FROM_DATA: data = {} files = [] @@ -325,6 +330,65 @@ class HttpRequestNode(BaseNode): case _: raise RuntimeError(f"HttpRequest method not supported: {self.typed_config.method}") + def _generate_curl_command( + self, + variable_pool: VariablePool, + url: str, + headers: dict[str, str], + params: dict[str, str], + content: dict[str, Any] + ) -> str: + """ + Generate equivalent curl command for debugging. + + Args: + variable_pool: Variable Pool + url: Rendered URL + headers: Request headers + params: Query parameters + content: Request body content + + Returns: + Curl command string + """ + # Start with curl command + curl_parts = ["curl"] + + # Add HTTP method + method = self.typed_config.method.value + if method != "GET": + curl_parts.append(f"-X {method}") + + # Add URL with query parameters + if params: + param_str = "&".join([f"{k}={v}" for k, v in params.items()]) + full_url = f"{url}?{param_str}" if "?" not in url else f"{url}&{param_str}" + else: + full_url = url + + curl_parts.append(f"'{full_url}'") + + # Add headers + for key, value in headers.items(): + curl_parts.append(f"-H '{key}: {value}'") + + # Add body based on content type + if "json" in content: + json_body = json.dumps(content["json"], ensure_ascii=False) + curl_parts.append(f"-d '{json_body}'") + elif "data" in content and "files" not in content: + # Form data + if isinstance(content["data"], dict): + for key, value in content["data"].items(): + curl_parts.append(f"-F '{key}={value}'") + elif "content" in content: + # Raw content + curl_parts.append(f"-d '{content['content']}'") + elif "files" in content: + curl_parts.append("# Note: This request includes file uploads") + + return " \\\n ".join(curl_parts) + async def execute(self, state: WorkflowState, variable_pool: VariablePool) -> dict | str: """ Execute the HTTP request node. @@ -343,11 +407,22 @@ class HttpRequestNode(BaseNode): - str: Branch identifier (e.g. "ERROR") when branching is enabled """ self.typed_config = HttpRequestNodeConfig(**self.config) + + # Build request components + headers = self._build_header(variable_pool) | self._build_auth(variable_pool) + params = self._build_params(variable_pool) + content = await self._build_content(variable_pool) + url = self._render_template(self.typed_config.url, variable_pool) + + # Generate curl command for debugging + curl_command = self._generate_curl_command(variable_pool, url, headers, params, content) + logger.info(f"Node {self.node_id}: Generated curl command:\n{curl_command}") + async with httpx.AsyncClient( verify=self.typed_config.verify_ssl, timeout=self._build_timeout(), - headers=self._build_header(variable_pool) | self._build_auth(variable_pool), - params=self._build_params(variable_pool), + headers=headers, + params=params, follow_redirects=True ) as client: retries = self.typed_config.retry.max_attempts @@ -355,8 +430,8 @@ class HttpRequestNode(BaseNode): try: request_func = self._get_client_method(client) resp = await request_func( - url=self._render_template(self.typed_config.url, variable_pool), - **(await self._build_content(variable_pool)) + url=url, + **content ) resp.raise_for_status() logger.info(f"Node {self.node_id}: HTTP request succeeded") @@ -365,7 +440,8 @@ class HttpRequestNode(BaseNode): body=response.body, status_code=resp.status_code, headers=resp.headers, - files=response.files + files=response.files, + curl=curl_command ).model_dump() except (httpx.HTTPStatusError, httpx.RequestError) as e: logger.error(f"HTTP request node exception: {e}") @@ -382,10 +458,20 @@ class HttpRequestNode(BaseNode): logger.warning( f"Node {self.node_id}: HTTP request failed, returning default result" ) - return self.typed_config.error_handle.default.model_dump() + # Update curl command in default error template + error_result = self.typed_config.error_handle.default.model_dump() + error_result["curl"] = curl_command + return error_result case HttpErrorHandle.BRANCH: logger.warning( f"Node {self.node_id}: HTTP request failed, switching to error handling branch" ) - return {"output": "ERROR"} + return { + "output": "ERROR", + "body": "", + "status_code": 500, + "headers": {}, + "files": [], + "curl": curl_command + } raise RuntimeError("http request failed") diff --git a/api/app/dependencies.py b/api/app/dependencies.py index 10684788..e5b656a5 100644 --- a/api/app/dependencies.py +++ b/api/app/dependencies.py @@ -564,6 +564,7 @@ async def get_app_or_workspace( if not app: auth_logger.warning(f"App not found for API Key: {api_key_obj.resource_id}") raise credentials_exception + ApiKeyAuthService.check_app_published(db, api_key_obj) auth_logger.info(f"App access granted: {app.id}") return app diff --git a/api/app/repositories/conversation_repository.py b/api/app/repositories/conversation_repository.py index 0676a255..2b4e7ab7 100644 --- a/api/app/repositories/conversation_repository.py +++ b/api/app/repositories/conversation_repository.py @@ -203,7 +203,8 @@ class ConversationRepository: self, app_id: uuid.UUID, workspace_id: uuid.UUID, - is_draft: Optional[bool] = None, + is_draft: bool = False, + keyword: Optional[str] = None, page: int = 1, pagesize: int = 20 ) -> tuple[list[Conversation], int]: @@ -213,29 +214,38 @@ class ConversationRepository: Args: app_id: 应用 ID workspace_id: 工作空间 ID - is_draft: 是否草稿会话（None 表示不过滤） + is_draft: 是否草稿会话（默认False，即发布会话） + keyword: 搜索关键词（匹配消息内容） page: 页码（从 1 开始） pagesize: 每页数量 Returns: Tuple[List[Conversation], int]: (会话列表，总数) """ - stmt = select(Conversation).where( + base_stmt = select(Conversation).where( Conversation.app_id == app_id, Conversation.workspace_id == workspace_id, - Conversation.is_active.is_(True) + Conversation.is_active.is_(True), + Conversation.is_draft == is_draft ) - if is_draft is not None: - stmt = stmt.where(Conversation.is_draft == is_draft) + # 如果有关键词搜索，通过子查询过滤包含该关键词的 conversation + if keyword: + # 查找包含关键词的 conversation_id 列表 + keyword_stmt = ( + select(Message.conversation_id) + .where(Message.content.ilike(f"%{keyword}%")) + .distinct() + ) + base_stmt = base_stmt.where(Conversation.id.in_(keyword_stmt)) # Calculate total number of records total = int(self.db.execute( - select(func.count()).select_from(stmt.subquery()) + select(func.count()).select_from(base_stmt.subquery()) ).scalar_one()) # Apply pagination - stmt = stmt.order_by(desc(Conversation.updated_at)) + stmt = base_stmt.order_by(desc(Conversation.updated_at)) stmt = stmt.offset((page - 1) * pagesize).limit(pagesize) conversations = list(self.db.scalars(stmt).all()) @@ -245,6 +255,7 @@ class ConversationRepository: extra={ "app_id": str(app_id), "workspace_id": str(workspace_id), + "keyword": keyword, "returned": len(conversations), "total": total } diff --git a/api/app/schemas/app_log_schema.py b/api/app/schemas/app_log_schema.py index bda78138..0a1855f7 100644 --- a/api/app/schemas/app_log_schema.py +++ b/api/app/schemas/app_log_schema.py @@ -48,6 +48,20 @@ class AppLogConversation(BaseModel): return int(dt.timestamp() * 1000) if dt else None +class AppLogNodeExecution(BaseModel): + """工作流节点执行记录""" + node_id: str + node_type: str + node_name: Optional[str] = None + status: str = "pending" + error: Optional[str] = None + input: Optional[Any] = None + output: Optional[Any] = None + elapsed_time: Optional[float] = None + token_usage: Optional[Dict[str, Any]] = None + + class AppLogConversationDetail(AppLogConversation): """会话详情（包含消息列表）""" messages: List[AppLogMessage] = Field(default_factory=list) + node_executions_map: Dict[str, List[AppLogNodeExecution]] = Field(default_factory=dict, description="按消息ID分组的节点执行记录") diff --git a/api/app/services/api_key_service.py b/api/app/services/api_key_service.py index 4856365a..49b07121 100644 --- a/api/app/services/api_key_service.py +++ b/api/app/services/api_key_service.py @@ -19,6 +19,7 @@ from app.core.exceptions import ( ) from app.core.error_codes import BizCode from app.core.logging_config import get_business_logger +from app.models.app_model import App logger = get_business_logger() @@ -442,6 +443,17 @@ class ApiKeyAuthService: return api_key_obj + @staticmethod + def check_app_published(db: Session, api_key_obj: ApiKey) -> None: + """ + 检查应用是否已发布，未发布则抛出异常 + """ + if not api_key_obj.resource_id: + return + app = db.get(App, api_key_obj.resource_id) + if not app or not app.current_release_id: + raise BusinessException("应用未发布，不可用", BizCode.APP_NOT_PUBLISHED) + @staticmethod def check_scope(api_key: ApiKey, required_scope: str) -> bool: """检查权限范围""" diff --git a/api/app/services/app_log_service.py b/api/app/services/app_log_service.py index 856045d1..8f6752ef 100644 --- a/api/app/services/app_log_service.py +++ b/api/app/services/app_log_service.py @@ -3,11 +3,14 @@ import uuid from typing import Optional, Tuple from datetime import datetime +from sqlalchemy import select from sqlalchemy.orm import Session from app.core.logging_config import get_business_logger from app.models.conversation_model import Conversation, Message +from app.models.workflow_model import WorkflowExecution from app.repositories.conversation_repository import ConversationRepository, MessageRepository +from app.schemas.app_log_schema import AppLogNodeExecution logger = get_business_logger() @@ -26,7 +29,8 @@ class AppLogService: workspace_id: uuid.UUID, page: int = 1, pagesize: int = 20, - is_draft: Optional[bool] = None, + is_draft: bool = False, + keyword: Optional[str] = None, ) -> Tuple[list[Conversation], int]: """ 查询应用日志会话列表 @@ -36,7 +40,8 @@ class AppLogService: workspace_id: 工作空间 ID page: 页码（从 1 开始） pagesize: 每页数量 - is_draft: 是否草稿会话（None 表示不过滤） + is_draft: 是否草稿会话（默认False，即发布会话） + keyword: 搜索关键词（匹配消息内容） Returns: Tuple[list[Conversation], int]: (会话列表，总数) @@ -48,7 +53,8 @@ class AppLogService: "workspace_id": str(workspace_id), "page": page, "pagesize": pagesize, - "is_draft": is_draft + "is_draft": is_draft, + "keyword": keyword } ) @@ -57,6 +63,7 @@ class AppLogService: app_id=app_id, workspace_id=workspace_id, is_draft=is_draft, + keyword=keyword, page=page, pagesize=pagesize ) @@ -77,9 +84,9 @@ class AppLogService: app_id: uuid.UUID, conversation_id: uuid.UUID, workspace_id: uuid.UUID - ) -> Conversation: + ) -> Tuple[Conversation, dict[str, list[AppLogNodeExecution]]]: """ - 查询会话详情（包含消息） + 查询会话详情（包含消息和工作流节点执行记录） Args: app_id: 应用 ID @@ -87,7 +94,8 @@ class AppLogService: workspace_id: 工作空间 ID Returns: - Conversation: 包含消息的会话对象 + Tuple[Conversation, dict[str, list[AppLogNodeExecution]]]: + (包含消息的会话对象, 按消息ID分组的节点执行记录) Raises: ResourceNotFoundException: 当会话不存在时 @@ -116,13 +124,116 @@ class AppLogService: # 将消息附加到会话对象 conversation.messages = messages + # 查询工作流节点执行记录（按消息分组） + _, node_executions_map = self._get_workflow_node_executions_with_map( + conversation_id, messages + ) + logger.info( "查询应用日志会话详情成功", extra={ "app_id": str(app_id), "conversation_id": str(conversation_id), - "message_count": len(messages) + "message_count": len(messages), + "message_with_nodes_count": len(node_executions_map) } ) - return conversation + return conversation, node_executions_map + + def _get_workflow_node_executions_with_map( + self, + conversation_id: uuid.UUID, + messages: list[Message] + ) -> Tuple[list[AppLogNodeExecution], dict[str, list[AppLogNodeExecution]]]: + """ + 从 workflow_executions 表中提取节点执行记录，并按 assistant message 分组 + + Args: + conversation_id: 会话 ID + messages: 消息列表 + + Returns: + Tuple[list[AppLogNodeExecution], dict[str, list[AppLogNodeExecution]]]: + (所有节点执行记录列表, 按 message_id 分组的节点执行记录字典) + """ + node_executions = [] + node_executions_map: dict[str, list[AppLogNodeExecution]] = {} + + # 查询该会话关联的所有工作流执行记录（按时间正序） + stmt = select(WorkflowExecution).where( + WorkflowExecution.conversation_id == conversation_id, + WorkflowExecution.status == "completed" + ).order_by(WorkflowExecution.started_at.asc()) + + executions = self.db.scalars(stmt).all() + + logger.info( + f"查询到 {len(executions)} 条工作流执行记录", + extra={ + "conversation_id": str(conversation_id), + "execution_count": len(executions), + "execution_ids": [str(e.id) for e in executions] + } + ) + + # 筛选出 workflow 执行产生的 assistant 消息（排除开场白） + # workflow 结果的 meta_data 包含 usage，而开场白包含 suggested_questions + assistant_messages = [ + m for m in messages + if m.role == "assistant" and m.meta_data and "usage" in m.meta_data + ] + + # 通过时序匹配，将 execution 和 assistant message 关联 + used_message_ids: set[str] = set() + + for execution in executions: + if not execution.output_data: + continue + + # 找到该 execution 对应的 assistant message + # 逻辑：找 execution.started_at 之后最近的、未使用的 assistant message + best_msg = None + best_dt = None + for msg in assistant_messages: + msg_id_str = str(msg.id) + if msg_id_str in used_message_ids: + continue + if msg.created_at and msg.created_at >= execution.started_at: + dt = (msg.created_at - execution.started_at).total_seconds() + if best_dt is None or dt < best_dt: + best_dt = dt + best_msg = msg + + if not best_msg: + continue + + msg_id_str = str(best_msg.id) + used_message_ids.add(msg_id_str) + + # 提取节点输出 + output_data = execution.output_data + if isinstance(output_data, dict): + node_outputs = output_data.get("node_outputs", {}) + execution_nodes = [] + for node_id, node_data in node_outputs.items(): + if not isinstance(node_data, dict): + continue + node_execution = AppLogNodeExecution( + node_id=node_data.get("node_id", node_id), + node_type=node_data.get("node_type", "unknown"), + node_name=node_data.get("node_name"), + status=node_data.get("status", "unknown"), + error=node_data.get("error"), + input=node_data.get("input"), + output=node_data.get("output"), + elapsed_time=node_data.get("elapsed_time"), + token_usage=node_data.get("token_usage"), + ) + node_executions.append(node_execution) + execution_nodes.append(node_execution) + + # 将节点记录关联到 message_id + node_executions_map[msg_id_str] = execution_nodes + + return node_executions, node_executions_map diff --git a/api/app/services/workflow_import_service.py b/api/app/services/workflow_import_service.py index 5a766a72..0c543d1f 100644 --- a/api/app/services/workflow_import_service.py +++ b/api/app/services/workflow_import_service.py @@ -14,6 +14,7 @@ from app.core.exceptions import BusinessException from app.core.workflow.adapters.base_adapter import WorkflowImportResult, WorkflowParserResult from app.core.workflow.adapters.errors import UnsupportedPlatform, InvalidConfiguration from app.core.workflow.adapters.registry import PlatformAdapterRegistry +from app.models.app_model import AppType from app.schemas import AppCreate from app.schemas.workflow_schema import WorkflowConfigCreate from app.services.app_service import AppService @@ -86,11 +87,12 @@ class WorkflowImportService: if config is None: raise BusinessException("Configuration import timed out. Please try again.") config = json.loads(config) + unique_name = self.app_service._unique_app_name(name, workspace_id, AppType.WORKFLOW) app = self.app_service.create_app( user_id=user_id, workspace_id=workspace_id, data=AppCreate( - name=name, + name=unique_name, description=description, type="workflow", workflow_config=WorkflowConfigCreate( From 8deb69b595d2b5918a18ddfe61c9665e73142f8a Mon Sep 17 00:00:00 2001 From: zhaoying Date: Thu, 23 Apr 2026 15:47:43 +0800 Subject: [PATCH 202/245] feat(web): agent model config add thinking_budget_tokens --- web/src/i18n/en.ts | 2 ++ web/src/i18n/zh.ts | 2 ++ .../ApplicationConfig/components/Chat.tsx | 25 +++++++------ .../components/ModelConfigModal.tsx | 36 ++++++++++++++++--- web/src/views/ApplicationConfig/types.ts | 1 + 5 files changed, 51 insertions(+), 15 deletions(-) diff --git a/web/src/i18n/en.ts b/web/src/i18n/en.ts index 72dd701d..31354817 100644 --- a/web/src/i18n/en.ts +++ b/web/src/i18n/en.ts @@ -1534,6 +1534,8 @@ export const en = { uploadCover: 'Import and Overwrite', refresh: 'Refresh Current Page', json_output: 'Support JSON formatted output', + thinking_budget_tokens: 'thinking budget tokens', + thinking_budget_tokens_max_error: "Cannot exceed the max tokens limit ({{max}})", }, userMemory: { userMemory: 'User Memory', diff --git a/web/src/i18n/zh.ts b/web/src/i18n/zh.ts index 88b919ba..d802b731 100644 --- a/web/src/i18n/zh.ts +++ b/web/src/i18n/zh.ts @@ -864,6 +864,8 @@ export const zh = { uploadCover: '导入并覆盖', refresh: '刷新当前页', json_output: '支持JSON格式化输出', + thinking_budget_tokens: '深度思考预算Token数', + thinking_budget_tokens_max_error: "不能超过 最大令牌数 ({{max}})", }, table: { totalRecords: '共 {{total}} 条记录' diff --git a/web/src/views/ApplicationConfig/components/Chat.tsx b/web/src/views/ApplicationConfig/components/Chat.tsx index 04391532..b4fefdc9 100644 --- a/web/src/views/ApplicationConfig/components/Chat.tsx +++ b/web/src/views/ApplicationConfig/components/Chat.tsx @@ -218,17 +218,22 @@ const Chat: FC = ({ const modelChatList = [...prev] const curModelChat = modelChatList[targetIndex] const curChatMsgList = curModelChat.list || [] - const lastMsg = curChatMsgList[curChatMsgList.length - 2] - modelChatList[targetIndex] = { - ...modelChatList[targetIndex], - list: [ - ...curChatMsgList.slice(0, curChatMsgList.length - 2), - { - ...lastMsg, - ...(lastMsg.role === 'user' ? { status: 'error' } : { content: null }) - } - ] + const lastUserMsg = curChatMsgList[curChatMsgList.length - 2] + const lastAssistantMsg = curChatMsgList[curChatMsgList.length - 1] + + if (!lastAssistantMsg.meta_data?.reasoning_content || lastAssistantMsg.meta_data?.reasoning_content.length === 0) { + modelChatList[targetIndex] = { + ...modelChatList[targetIndex], + list: [ + ...curChatMsgList.slice(0, curChatMsgList.length - 2), + { + ...lastUserMsg, + ...(lastUserMsg.role === 'user' ? { status: 'error' } : { content: null }) + } + ] + } } + return [...modelChatList] } diff --git a/web/src/views/ApplicationConfig/components/ModelConfigModal.tsx b/web/src/views/ApplicationConfig/components/ModelConfigModal.tsx index 2d330e9f..9c5935c5 100644 --- a/web/src/views/ApplicationConfig/components/ModelConfigModal.tsx +++ b/web/src/views/ApplicationConfig/components/ModelConfigModal.tsx @@ -108,7 +108,8 @@ const ModelConfigModal = forwardRef( const newValues: ModelConfig = { capability: (option as Model).capability, deep_thinking: false, - json_output: false + thinking_budget_tokens: undefined, + json_output: false, } if (source === 'chat') { newValues.label = (option as Model).name @@ -124,8 +125,8 @@ const ModelConfigModal = forwardRef( useEffect(() => { const { deep_thinking: _, json_output: __, ...rest } = data?.model_parameters || {} - form.setFieldsValue(rest) - }, [values?.default_model_config_id]) + form.setFieldsValue({ ...rest }) + }, [data?.default_model_config_id]) const handleReset = () => { if (!id) return @@ -167,11 +168,36 @@ const ModelConfigModal = forwardRef( {['model', 'chat'].includes(source) && <> } + -
* 模型 Key 全删后自动关闭：避免无 Key 运行时错误
* 模型 JSON 格式化输出开关：支持旧工作流迁移的稳定 JSON 输出
* 配置导入覆盖：支持完整替换当前配置
* 导入时缺失资源清理：自动清空不存在的工具和知识库引用", + "2. 记忆 API 与智能 📚
* 记忆读写 API 与 End-User Key 供给：支持第三方直接交互记忆层
* 记忆库 API 与配置更新：程序化控制记忆设置（提供顺序接口）
* End-User 元数据存储：丰富用户上下文持久化", + "3. 工作流与体验优化 ⚙️
* 会话历史文件元数据：增加文件大小、名称和类型
* 迭代节点并行输入修复：恢复并发执行行为
* API Key 后四位展示：便于密钥识别
* 条件分支多文件子变量：更精细的条件逻辑
* Agent 模型配置重置接口：完善前后端契约
* 三级变量键盘导航：提升变量选择体验
* 应用标签页动态标题：动态显示应用名称
* 变量聚合三级勾选修复：修复勾选行为
* 工作流检查清单校验增强：工具必填和视觉变量必填
* 变量聚合器到参数提取器输出：修复输出变量获取", + "4. 知识库与性能 ⚡
* 文档解析与 Graph 异步执行：提升文档摄入吞吐量", + "5. 稳健性与缺陷修复 🔧
* 工具节点原始参数类型：修复类型不匹配问题
* 前端部署后资源过期导入错误：解决缓存资源导入失败
* 工作流工具节点必填校验：防止不完整配置发布", + "
", + "v0.3.1 是平台哲学演进中的关键时刻——边界的打破。记忆 API 开放和应用体验优化为社区用户提供更强大的集成能力。展望未来，我们将持续提升记忆智能管线的萃取精度与自适应遗忘策略，深化工作流引擎能力。破界而行，臻于无境。", + "MemoryBear — 无境 🐻✨" + ] + }, + "introduction_en": { + "codeName": "WuJing", + "releaseDate": "2026-4-22", + "upgradePosition": "🐻 Focused application improvements, memory API openness, and workflow reliability — dissolving boundaries, flowing freely", + "coreUpgrades": [ + "1. Application & Model Enhancements
* Model Auto-Disable on Key Deletion: Prevents keyless runtime errors
* Model JSON Formatted Output Toggle: Stable JSON output for legacy workflow migration
* Configuration Import with Override: Full configuration replacement support
* Import Cleanup for Missing Resources: Auto-clears missing tool and knowledge base references", + "2. Memory API & Intelligence 📚
* Memory Read/Write API with End-User Key Provisioning: Third-party memory layer interaction
* Memory Store API & Configuration Update: Programmatic memory settings control with sequential interface
* End-User Metadata Storage: Richer user context persistence", + "3. Workflow & UX Improvements ⚙️
* Conversation History File Metadata: File size, name, and type labels
* Iteration Node Parallel Input Fix: Restored concurrent execution
* API Key Last Four Digits Display: Key identification without exposure
* Condition Branch Multi-File Sub-Variables: Granular conditional logic
* Agent Model Config Reset Endpoint: Completed frontend-backend contract
* Three-Level Variable Keyboard Navigation: Improved selection experience
* Dynamic Tab Title for Applications: Dynamic app name in browser tab
* Variable Aggregator Three-Level Checkbox Fix: Corrected checkbox behavior
* Workflow Checklist Validation Enhancements: Tool required and vision variable validation
* Variable Aggregator to Parameter Extractor Output: Fixed output variable access", + "4. Knowledge Base & Performance ⚡
* Async Document Parsing & Graph Execution: Improved document ingestion throughput", + "5. Robustness & Bug Fixes 🔧
* Tool Node Raw Parameter Types: Fixed type mismatch issues
* Stale Frontend Resource Import Error: Resolved cached resource import failure
* Workflow Tool Node Required Validation: Prevents incomplete configuration publishing", + "
", + "v0.3.1 marks a pivotal moment in the platform's evolution — the dissolution of boundaries. Memory API openness and application experience improvements provide community users with stronger integration capabilities. Looking ahead, we will continue improving extraction accuracy, adaptive forgetting strategies, and deepening workflow engine capabilities. Beyond boundaries — the boundless awaits.", + "MemoryBear — The Boundless 🐻✨" + ] + } + }, "v0.3.0": { "introduction": { "codeName": "破晓", diff --git a/api/pyproject.toml b/api/pyproject.toml index 8ced574c..6d4a83c5 100644 --- a/api/pyproject.toml +++ b/api/pyproject.toml @@ -147,7 +147,8 @@ dependencies = [ "modelscope>=1.34.0", "python-magic>=0.4.14; sys_platform == 'linux' or sys_platform == 'darwin'", "python-magic-bin>=0.4.14; sys_platform=='win32'", - "volcengine-python-sdk[ark]==5.0.19" + "volcengine-python-sdk[ark]==5.0.19", + "pymupdf>=1.27.2.2", ] [tool.pytest.ini_options] From 0f7a7263ebad9f7b91267244ea4a837b5d58c5b5 Mon Sep 17 00:00:00 2001 From: wwq Date: Fri, 24 Apr 2026 11:39:33 +0800 Subject: [PATCH 213/245] fix(workflow): rectify error handling and bolster execution logging - Rectify exception propagation during node execution failures to ensure errors are correctly raised. - Bolster workflow logging to support failed status records and persist node execution data, including loop nodes. --- api/app/core/workflow/nodes/base_node.py | 29 ++++----- api/app/services/app_log_service.py | 56 ++++++++--------- api/app/services/workflow_service.py | 78 +++++++++++++++++++++++- 3 files changed, 117 insertions(+), 46 deletions(-) diff --git a/api/app/core/workflow/nodes/base_node.py b/api/app/core/workflow/nodes/base_node.py index 5458a80c..85f4bc5f 100644 --- a/api/app/core/workflow/nodes/base_node.py +++ b/api/app/core/workflow/nodes/base_node.py @@ -228,19 +228,21 @@ class BaseNode(ABC): logger.error( f"Node {self.node_id} execution timed out ({timeout} seconds)." ) - return self._wrap_error( + self._wrap_error( f"Node execution timed out ({timeout} seconds).", elapsed_time, state, variable_pool, ) + raise except Exception as e: elapsed_time = (time.time() - start_time) * 1000 logger.error( f"Node {self.node_id} execution failed: {e}", exc_info=True, ) - return self._wrap_error(str(e), elapsed_time, state, variable_pool) + self._wrap_error(str(e), elapsed_time, state, variable_pool) + raise async def run_stream( self, state: WorkflowState, @@ -351,11 +353,13 @@ class BaseNode(ABC): variable_pool ) yield error_output + raise except Exception as e: elapsed_time = (time.time() - start_time) * 1000 logger.error(f"Node {self.node_id} execution failed: {e}", exc_info=True) error_output = self._wrap_error(str(e), elapsed_time, state, variable_pool) yield error_output + raise def _wrap_output( self, @@ -447,26 +451,19 @@ class BaseNode(ABC): "error": error_message } - # if error_edge: - # # If an error edge exists, log a warning and continue to error node - # logger.warning( - # f"Node {self.node_id} execution failed, redirecting to error node: {error_edge['target']}" - # ) - # return { - # "node_outputs": { - # self.node_id: node_output - # }, - # "error": error_message, - # "error_node": self.node_id - # } - # else: writer = get_stream_writer() writer({ "type": "node_error", **node_output }) logger.error(f"Node {self.node_id} execution failed, stopping workflow: {error_message}") - raise Exception(f"Node {self.node_id} execution failed: {error_message}") + return { + "node_outputs": { + self.node_id: node_output + }, + "error": error_message, + "error_node": self.node_id + } def _extract_input(self, state: WorkflowState, variable_pool: VariablePool) -> dict[str, Any]: """Extracts the input data for this node (used for logging or audit). diff --git a/api/app/services/app_log_service.py b/api/app/services/app_log_service.py index 8f5052e6..b2801df3 100644 --- a/api/app/services/app_log_service.py +++ b/api/app/services/app_log_service.py @@ -163,7 +163,7 @@ class AppLogService: # 查询该会话关联的所有工作流执行记录（按时间正序） stmt = select(WorkflowExecution).where( WorkflowExecution.conversation_id == conversation_id, - WorkflowExecution.status == "completed" + WorkflowExecution.status.in_(["completed", "failed"]) ).order_by(WorkflowExecution.started_at.asc()) executions = self.db.scalars(stmt).all() @@ -188,10 +188,33 @@ class AppLogService: used_message_ids: set[str] = set() for execution in executions: - if not execution.output_data: + # 构建节点执行记录列表 + execution_nodes = [] + for node_exec in execution.node_executions: + node_execution = AppLogNodeExecution( + node_id=node_exec.node_id, + node_type=node_exec.node_type, + node_name=node_exec.node_name, + status=node_exec.status, + error=node_exec.error_message, + input=node_exec.input_data, + process=None, + output=node_exec.output_data, + elapsed_time=node_exec.elapsed_time, + token_usage=node_exec.token_usage, + ) + node_executions.append(node_execution) + execution_nodes.append(node_execution) + + if not execution_nodes: continue - # 找到该 execution 对应的 assistant message + # 失败的执行没有 assistant message，直接用 execution id 作为 key + if execution.status == "failed": + node_executions_map[f"execution_{str(execution.id)}"] = execution_nodes + continue + + # completed：通过时序匹配关联到对应的 assistant message # 逻辑：找 execution.started_at 之后最近的、未使用的 assistant message best_msg = None best_dt = None @@ -210,31 +233,6 @@ class AppLogService: msg_id_str = str(best_msg.id) used_message_ids.add(msg_id_str) - - # 提取节点输出 - output_data = execution.output_data - if isinstance(output_data, dict): - node_outputs = output_data.get("node_outputs", {}) - execution_nodes = [] - for node_id, node_data in node_outputs.items(): - if not isinstance(node_data, dict): - continue - node_execution = AppLogNodeExecution( - node_id=node_data.get("node_id", node_id), - node_type=node_data.get("node_type", "unknown"), - node_name=node_data.get("node_name"), - status=node_data.get("status", "unknown"), - error=node_data.get("error"), - input=node_data.get("input"), - process=node_data.get("process"), - output=node_data.get("output"), - elapsed_time=node_data.get("elapsed_time"), - token_usage=node_data.get("token_usage"), - ) - node_executions.append(node_execution) - execution_nodes.append(node_execution) - - # 将节点记录关联到 message_id - node_executions_map[msg_id_str] = execution_nodes + node_executions_map[msg_id_str] = execution_nodes return node_executions, node_executions_map diff --git a/api/app/services/workflow_service.py b/api/app/services/workflow_service.py index 0d282d78..6c93893a 100644 --- a/api/app/services/workflow_service.py +++ b/api/app/services/workflow_service.py @@ -17,8 +17,9 @@ from app.core.workflow.executor import execute_workflow, execute_workflow_stream from app.core.workflow.nodes.enums import NodeType from app.core.workflow.validator import validate_workflow_config from app.db import get_db +from sqlalchemy import select from app.models import App -from app.models.workflow_model import WorkflowConfig, WorkflowExecution +from app.models.workflow_model import WorkflowConfig, WorkflowExecution, WorkflowNodeExecution from app.repositories import knowledge_repository from app.repositories.workflow_repository import ( WorkflowConfigRepository, @@ -909,6 +910,8 @@ class WorkflowService: input_data["conv_messages"] = conv_messages init_message_length = len(input_data.get("conv_messages", [])) message_id = uuid.uuid4() + _node_order_counter = 0 + _cycle_items: dict[str, list] = {} # 新会话时写入开场白 is_new_conversation = init_message_length == 0 @@ -939,6 +942,52 @@ class WorkflowService: memory_storage_type=storage_type, user_rag_memory_id=user_rag_memory_id ): + event_type = event.get("event") + event_data = event.get("data", {}) + + # 持久化节点执行记录 + if event_type == "node_end": + node_id = event_data.get("node_id") + node_cfg = next((n for n in config.nodes if n.get("id") == node_id), {}) + self.db.add(WorkflowNodeExecution( + execution_id=execution.id, + node_id=node_id, + node_type=node_cfg.get("type", "unknown"), + node_name=node_cfg.get("data", {}).get("label") or node_id, + execution_order=_node_order_counter, + status="completed", + input_data=event_data.get("input"), + output_data=event_data.get("output"), + elapsed_time=event_data.get("elapsed_time"), + token_usage=event_data.get("token_usage"), + )) + self.db.commit() + _node_order_counter += 1 + + elif event_type == "node_error": + node_id = event_data.get("node_id") + node_cfg = next((n for n in config.nodes if n.get("id") == node_id), {}) + self.db.add(WorkflowNodeExecution( + execution_id=execution.id, + node_id=node_id, + node_type=node_cfg.get("type", "unknown"), + node_name=node_cfg.get("data", {}).get("label") or node_id, + execution_order=_node_order_counter, + status="failed", + input_data=event_data.get("input"), + output_data=None, + error_message=event_data.get("error"), + elapsed_time=event_data.get("elapsed_time"), + )) + self.db.commit() + _node_order_counter += 1 + + elif event_type == "cycle_item": + cycle_id = event_data.get("cycle_id") + if cycle_id not in _cycle_items: + _cycle_items[cycle_id] = [] + _cycle_items[cycle_id].append(event_data) + if event.get("event") == "workflow_end": status = event.get("data", {}).get("status") token_usage = event.get("data", {}).get("token_usage", {}) or {} @@ -1003,6 +1052,33 @@ class WorkflowService: ) else: logger.error(f"unexpect workflow run status, status: {status}") + # 把积累的 cycle_item 写入对应循环节点的 output_data + if _cycle_items: + for cycle_node_id, items in _cycle_items.items(): + node_exec = self.db.execute( + select(WorkflowNodeExecution).where( + WorkflowNodeExecution.execution_id == execution.id, + WorkflowNodeExecution.node_id == cycle_node_id + ) + ).scalar_one_or_none() + if node_exec: + node_exec.output_data = { + **(node_exec.output_data or {}), + "cycle_items": items + } + else: + node_cfg = next((n for n in config.nodes if n.get("id") == cycle_node_id), {}) + self.db.add(WorkflowNodeExecution( + execution_id=execution.id, + node_id=cycle_node_id, + node_type=node_cfg.get("type", "cycle"), + node_name=node_cfg.get("data", {}).get("label") or cycle_node_id, + execution_order=_node_order_counter, + status="completed", + output_data={"cycle_items": items}, + )) + _node_order_counter += 1 + self.db.commit() elif event.get("event") == "workflow_start": event["data"]["message_id"] = str(message_id) event = self._emit(public, event) From 2d120a64b1d9975f05dd5df4482ddce5723e1950 Mon Sep 17 00:00:00 2001 From: wwq Date: Fri, 24 Apr 2026 11:50:48 +0800 Subject: [PATCH 214/245] fix(workflow): rectify error handling and bolster execution logging - Rectify exception propagation during node execution failures to ensure errors are correctly raised. - Bolster workflow logging to support failed status records and persist node execution data, including loop nodes. --- api/app/core/workflow/nodes/base_node.py | 16 +++------------- 1 file changed, 3 insertions(+), 13 deletions(-) diff --git a/api/app/core/workflow/nodes/base_node.py b/api/app/core/workflow/nodes/base_node.py index 85f4bc5f..a0f1f01e 100644 --- a/api/app/core/workflow/nodes/base_node.py +++ b/api/app/core/workflow/nodes/base_node.py @@ -228,21 +228,19 @@ class BaseNode(ABC): logger.error( f"Node {self.node_id} execution timed out ({timeout} seconds)." ) - self._wrap_error( + return self._wrap_error( f"Node execution timed out ({timeout} seconds).", elapsed_time, state, variable_pool, ) - raise except Exception as e: elapsed_time = (time.time() - start_time) * 1000 logger.error( f"Node {self.node_id} execution failed: {e}", exc_info=True, ) - self._wrap_error(str(e), elapsed_time, state, variable_pool) - raise + return self._wrap_error(str(e), elapsed_time, state, variable_pool) async def run_stream( self, state: WorkflowState, @@ -353,13 +351,11 @@ class BaseNode(ABC): variable_pool ) yield error_output - raise except Exception as e: elapsed_time = (time.time() - start_time) * 1000 logger.error(f"Node {self.node_id} execution failed: {e}", exc_info=True) error_output = self._wrap_error(str(e), elapsed_time, state, variable_pool) yield error_output - raise def _wrap_output( self, @@ -457,13 +453,7 @@ class BaseNode(ABC): **node_output }) logger.error(f"Node {self.node_id} execution failed, stopping workflow: {error_message}") - return { - "node_outputs": { - self.node_id: node_output - }, - "error": error_message, - "error_node": self.node_id - } + raise Exception(f"Node {self.node_id} execution failed: {error_message}") def _extract_input(self, state: WorkflowState, variable_pool: VariablePool) -> dict[str, Any]: """Extracts the input data for this node (used for logging or audit). From 279353e1ce51da15aa23cc9bcc33b2a7b8ea8e94 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Fri, 24 Apr 2026 11:52:11 +0800 Subject: [PATCH 215/245] feat(web): file upload add document_image_recognition config --- web/src/i18n/en.ts | 3 +++ web/src/i18n/zh.ts | 3 +++ .../FeaturesConfig/FileUploadSettingModal.tsx | 22 ++++++++++++++----- .../OpenStatementSettingModal.tsx | 1 + .../components/ModelConfigModal.tsx | 3 ++- web/src/views/ApplicationConfig/types.ts | 2 ++ 6 files changed, 28 insertions(+), 6 deletions(-) diff --git a/web/src/i18n/en.ts b/web/src/i18n/en.ts index 31354817..e5a80431 100644 --- a/web/src/i18n/en.ts +++ b/web/src/i18n/en.ts @@ -1460,6 +1460,7 @@ export const en = { maxCount: 'Max Files', singleMaxSize: 'Max Size', unix: 'items', + document_image_recognition: 'Enable image recognition in documents', text_to_speech: 'Text to Speech', text_to_speech_desc: 'Text can be converted to speech', opening_statement: 'Conversation Opening', @@ -1536,6 +1537,7 @@ export const en = { json_output: 'Support JSON formatted output', thinking_budget_tokens: 'thinking budget tokens', thinking_budget_tokens_max_error: "Cannot exceed the max tokens limit ({{max}})", + logSearchPlaceholder: 'Search log content', }, userMemory: { userMemory: 'User Memory', @@ -2529,6 +2531,7 @@ Memory Bear: After the rebellion, regional warlordism intensified for several re input_result: 'Input', output_result: 'Output', + process_result: 'Data Processing', error: 'Error Message', loopNum: ' loops', iterationNum: ' iterations', diff --git a/web/src/i18n/zh.ts b/web/src/i18n/zh.ts index d802b731..8d71876e 100644 --- a/web/src/i18n/zh.ts +++ b/web/src/i18n/zh.ts @@ -790,6 +790,7 @@ export const zh = { maxCount: '最大文件数', singleMaxSize: '单文件最大大小', unix: '个', + document_image_recognition: '是否识别文档中的图片', text_to_speech: '文字转语音', text_to_speech_desc: '文本可以转换成语音', opening_statement: '对话开场白', @@ -866,6 +867,7 @@ export const zh = { json_output: '支持JSON格式化输出', thinking_budget_tokens: '深度思考预算Token数', thinking_budget_tokens_max_error: "不能超过 最大令牌数 ({{max}})", + logSearchPlaceholder: '搜索日志内容', }, table: { totalRecords: '共 {{total}} 条记录' @@ -2493,6 +2495,7 @@ export const zh = { input_result: '输入', output_result: '输出', + process_result: '数据处理', error: '错误信息', loopNum: '个循环', iterationNum: '个迭代', diff --git a/web/src/views/ApplicationConfig/components/FeaturesConfig/FileUploadSettingModal.tsx b/web/src/views/ApplicationConfig/components/FeaturesConfig/FileUploadSettingModal.tsx index 5c17aa53..2ae09a5e 100644 --- a/web/src/views/ApplicationConfig/components/FeaturesConfig/FileUploadSettingModal.tsx +++ b/web/src/views/ApplicationConfig/components/FeaturesConfig/FileUploadSettingModal.tsx @@ -97,6 +97,7 @@ export const defaultValues: FileUpload = { "json", "md", ], + document_image_recognition: false, video_enabled: false, video_max_size_mb: 100, video_allowed_extensions: [ @@ -219,11 +220,22 @@ const FileUploadSettingModal = forwardRef {isEnabled && ( - -
{t('application.singleMaxSize')}:
- - - + +
+
{t('application.singleMaxSize')}
+ + + +
+ {option.type === 'document' && +
+
{t('application.document_image_recognition')}
+ + + +
+ } + )} diff --git a/web/src/views/ApplicationConfig/components/FeaturesConfig/OpenStatementSettingModal.tsx b/web/src/views/ApplicationConfig/components/FeaturesConfig/OpenStatementSettingModal.tsx index a46d973a..ed9204da 100644 --- a/web/src/views/ApplicationConfig/components/FeaturesConfig/OpenStatementSettingModal.tsx +++ b/web/src/views/ApplicationConfig/components/FeaturesConfig/OpenStatementSettingModal.tsx @@ -104,6 +104,7 @@ const OpenStatementSettingModal = forwardRef {source === 'workflow' ? diff --git a/web/src/views/ApplicationConfig/components/ModelConfigModal.tsx b/web/src/views/ApplicationConfig/components/ModelConfigModal.tsx index a7f9eb57..bda18571 100644 --- a/web/src/views/ApplicationConfig/components/ModelConfigModal.tsx +++ b/web/src/views/ApplicationConfig/components/ModelConfigModal.tsx @@ -184,7 +184,8 @@ const ModelConfigModal = forwardRef( { validator: (_, value) => { const maxTokens = values?.max_tokens - if (value !== undefined && maxTokens !== undefined && value > maxTokens) { + const deep_thinking = values?.deep_thinking; + if (deep_thinking && value !== undefined && maxTokens !== undefined && value > maxTokens) { return Promise.reject(t('application.thinking_budget_tokens_max_error', { max: maxTokens })) } return Promise.resolve() diff --git a/web/src/views/ApplicationConfig/types.ts b/web/src/views/ApplicationConfig/types.ts index 1b54d3aa..c34f882f 100644 --- a/web/src/views/ApplicationConfig/types.ts +++ b/web/src/views/ApplicationConfig/types.ts @@ -438,6 +438,7 @@ interface FileSetttings { document_enabled: boolean; document_max_size_mb: number; document_allowed_extensions: string[]; + document_image_recognition: boolean; video_enabled: boolean; video_max_size_mb: number; video_allowed_extensions: string[]; @@ -499,6 +500,7 @@ export interface LogItem { is_draft: boolean; created_at: number; updated_at: number; + node_executions_map?: Record } export interface LogDetailModalRef { handleOpen: (vo: LogItem) => void; From b33f5951d854d60ee635cc47b562a720a72e3e62 Mon Sep 17 00:00:00 2001 From: wwq Date: Fri, 24 Apr 2026 11:52:15 +0800 Subject: [PATCH 216/245] fix(workflow): rectify error handling and bolster execution logging - Rectify exception propagation during node execution failures to ensure errors are correctly raised. - Bolster workflow logging to support failed status records and persist node execution data, including loop nodes. --- api/app/core/workflow/nodes/base_node.py | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/api/app/core/workflow/nodes/base_node.py b/api/app/core/workflow/nodes/base_node.py index a0f1f01e..5458a80c 100644 --- a/api/app/core/workflow/nodes/base_node.py +++ b/api/app/core/workflow/nodes/base_node.py @@ -447,6 +447,19 @@ class BaseNode(ABC): "error": error_message } + # if error_edge: + # # If an error edge exists, log a warning and continue to error node + # logger.warning( + # f"Node {self.node_id} execution failed, redirecting to error node: {error_edge['target']}" + # ) + # return { + # "node_outputs": { + # self.node_id: node_output + # }, + # "error": error_message, + # "error_node": self.node_id + # } + # else: writer = get_stream_writer() writer({ "type": "node_error", From 1191f0f54e3f14fe5535dbba50b3d2793b49b4d8 Mon Sep 17 00:00:00 2001 From: Eternity <1533512157@qq.com> Date: Fri, 24 Apr 2026 13:13:38 +0800 Subject: [PATCH 217/245] fix(neo4j): correct community property name in search queries --- api/app/repositories/neo4j/cypher_queries.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/api/app/repositories/neo4j/cypher_queries.py b/api/app/repositories/neo4j/cypher_queries.py index 05a8c4b0..a8c36e34 100644 --- a/api/app/repositories/neo4j/cypher_queries.py +++ b/api/app/repositories/neo4j/cypher_queries.py @@ -1236,7 +1236,7 @@ RETURN s.id AS id, SEARCH_COMMUNITIES_BY_USER_ID = """ MATCH (c:Community) WHERE c.end_user_id = $end_user_id -RETURN c.id AS id, +RETURN c.community_id AS id, c.summary_embedding AS embedding """ @@ -1467,7 +1467,7 @@ LIMIT $limit SEARCH_COMMUNITIES_BY_KEYWORD = """ CALL db.index.fulltext.queryNodes("communitiesFulltext", $query) YIELD node AS c, score WHERE ($end_user_id IS NULL OR c.end_user_id = $end_user_id) -RETURN c.id AS id, +RETURN c.community_id AS id, c.name AS name, c.summary AS content, c.core_entities AS core_entities, From 89f2f9a0450ed029dcc20e7cdfb37841ca5583fc Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Fri, 24 Apr 2026 14:18:25 +0800 Subject: [PATCH 218/245] feat(citation): support downloading cited documents with allow_download toggle Added `allow_download` flag to citation config and `download_url` field to citation output. Implemented `/citations/{document_id}/download` endpoint to serve original files when enabled. Removed unused `files` field and `HttpRequestDataProcessing` model from HTTP request node config. --- api/app/controllers/app_controller.py | 43 +++++++++++++++++++ .../workflow/nodes/http_request/config.py | 12 ------ api/app/schemas/app_schema.py | 4 +- api/app/services/draft_run_service.py | 16 +++++-- 4 files changed, 58 insertions(+), 17 deletions(-) diff --git a/api/app/controllers/app_controller.py b/api/app/controllers/app_controller.py index eda5e76a..41422bd4 100644 --- a/api/app/controllers/app_controller.py +++ b/api/app/controllers/app_controller.py @@ -1298,3 +1298,46 @@ async def import_app( data={"app": app_schema.App.model_validate(result_app), "warnings": warnings}, msg="应用导入成功" + ("，但部分资源需手动配置" if warnings else "") ) + + +@router.get("/citations/{document_id}/download", summary="下载引用文档原始文件") +async def download_citation_file( + document_id: uuid.UUID = Path(..., description="引用文档ID"), + db: Session = Depends(get_db), +): + """ + 下载引用文档的原始文件。 + 仅当应用功能特性 citation.allow_download=true 时，前端才会展示此下载链接。 + 路由本身不做权限校验，由业务层通过 allow_download 开关控制入口。 + """ + import os + from fastapi import HTTPException, status as http_status + from fastapi.responses import FileResponse + from app.core.config import settings + from app.models.document_model import Document + from app.models.file_model import File as FileModel + + doc = db.query(Document).filter(Document.id == document_id).first() + if not doc: + raise HTTPException(status_code=http_status.HTTP_404_NOT_FOUND, detail="文档不存在") + + file_record = db.query(FileModel).filter(FileModel.id == doc.file_id).first() + if not file_record: + raise HTTPException(status_code=http_status.HTTP_404_NOT_FOUND, detail="原始文件不存在") + + file_path = os.path.join( + settings.FILE_PATH, + str(file_record.kb_id), + str(file_record.parent_id), + f"{file_record.id}{file_record.file_ext}" + ) + if not os.path.exists(file_path): + raise HTTPException(status_code=http_status.HTTP_404_NOT_FOUND, detail="文件未找到") + + encoded_name = quote(doc.file_name) + return FileResponse( + path=file_path, + filename=doc.file_name, + media_type="application/octet-stream", + headers={"Content-Disposition": f"attachment; filename*=UTF-8''{encoded_name}"} + ) diff --git a/api/app/core/workflow/nodes/http_request/config.py b/api/app/core/workflow/nodes/http_request/config.py index a6d16912..72474436 100644 --- a/api/app/core/workflow/nodes/http_request/config.py +++ b/api/app/core/workflow/nodes/http_request/config.py @@ -132,11 +132,6 @@ class HttpErrorDefaultTemplate(BaseModel): description="Default HTTP headers returned on error", ) - files: list = Field( - default_factory=list, - description="Default files list returned on error", - ) - output: str = Field( default="SUCCESS", description="HTTP response body", @@ -251,13 +246,6 @@ class HttpRequestNodeConfig(BaseNodeConfig): } -class HttpRequestDataProcessing(BaseModel): - request: str = Field( - default="", - description="Raw HTTP request format for debugging", - ) - - class HttpRequestNodeOutput(BaseModel): body: str = Field( ..., diff --git a/api/app/schemas/app_schema.py b/api/app/schemas/app_schema.py index f64d2ac4..11c27b56 100644 --- a/api/app/schemas/app_schema.py +++ b/api/app/schemas/app_schema.py @@ -200,6 +200,7 @@ class TextToSpeechConfig(BaseModel): class CitationConfig(BaseModel): """引用和归属配置""" enabled: bool = Field(default=False) + allow_download: bool = Field(default=False, description="是否允许下载引用文档") class Citation(BaseModel): @@ -207,6 +208,7 @@ class Citation(BaseModel): file_name: str knowledge_id: str score: float + download_url: Optional[str] = Field(default=None, description="引用文档下载链接（allow_download 开启时返回）") class WebSearchConfig(BaseModel): @@ -657,7 +659,7 @@ class DraftRunResponse(BaseModel): usage: Optional[Dict[str, Any]] = Field(default=None, description="Token 使用情况") elapsed_time: Optional[float] = Field(default=None, description="耗时（秒）") suggested_questions: List[str] = Field(default_factory=list, description="下一步建议问题") - citations: List[CitationSource] = Field(default_factory=list, description="引用来源") + citations: List[Dict[str, Any]] = Field(default_factory=list, description="引用来源") audio_url: Optional[str] = Field(default=None, description="TTS 语音URL") def model_dump(self, **kwargs): diff --git a/api/app/services/draft_run_service.py b/api/app/services/draft_run_service.py index f21dd8da..2869326f 100644 --- a/api/app/services/draft_run_service.py +++ b/api/app/services/draft_run_service.py @@ -475,11 +475,19 @@ class AgentRunService: features_config: Dict[str, Any], citations: List[Citation] ) -> List[Any]: - """根据 citation 开关决定是否返回引用来源""" + """根据 citation 开关决定是否返回引用来源，并根据 allow_download 附加下载链接""" citation_cfg = features_config.get("citation", {}) - if isinstance(citation_cfg, dict) and citation_cfg.get("enabled"): - return [cit.model_dump() for cit in citations] - return [] + if not (isinstance(citation_cfg, dict) and citation_cfg.get("enabled")): + return [] + allow_download = citation_cfg.get("allow_download", False) + result = [] + for cit in citations: + item = cit.model_dump() if hasattr(cit, "model_dump") else dict(cit) + if allow_download and item.get("document_id"): + from app.core.config import settings + item["download_url"] = f"{settings.FILE_LOCAL_SERVER_URL}/apps/citations/{item['document_id']}/download" + result.append(item) + return result async def run( self, From be10bab763e55b492a77afa9a6124bf8df1878fb Mon Sep 17 00:00:00 2001 From: Eternity <1533512157@qq.com> Date: Fri, 24 Apr 2026 13:56:32 +0800 Subject: [PATCH 219/245] refactor(core): migrate task scheduler to per-user queue with dynamic sharding --- api/app/celery_task_scheduler.py | 431 +++++++++++++----- .../generate_engine/retrieval_summary.py | 2 +- api/app/services/memory_api_service.py | 2 +- 3 files changed, 326 insertions(+), 109 deletions(-) diff --git a/api/app/celery_task_scheduler.py b/api/app/celery_task_scheduler.py index 8cdce658..e7f946b6 100644 --- a/api/app/celery_task_scheduler.py +++ b/api/app/celery_task_scheduler.py @@ -1,21 +1,70 @@ +import hashlib import json +import os +import socket import threading import time +import uuid import redis from app.core.config import settings -from celery_app import celery_app from app.core.logging_config import get_named_logger +from app.celery_app import celery_app logger = get_named_logger("task_scheduler") -STREAM_KEY = "celery_task_stream" +# per-user queue scheduler:uq:{user_id} +USER_QUEUE_PREFIX = "scheduler:uq:" +# User Collection of Pending Messages +ACTIVE_USERS = "scheduler:active_users" +# Set of users that can dispatch (ready signal) +READY_SET = "scheduler:ready_users" +# Metadata of tasks that have been dispatched and are pending completion PENDING_HASH = "scheduler:pending_tasks" -TASK_TIMEOUT = 7800 +# Dynamic Sharding: Instance Registry +REGISTRY_KEY = "scheduler:instances" + +TASK_TIMEOUT = 7800 # Task timeout (seconds), considered lost if exceeded +HEARTBEAT_INTERVAL = 10 # Heartbeat interval (seconds) +INSTANCE_TTL = 30 # Instance timeout (seconds) + +LUA_ATOMIC_LOCK = """ +local dispatch_lock = KEYS[1] +local lock_key = KEYS[2] +local instance_id = ARGV[1] +local dispatch_ttl = tonumber(ARGV[2]) +local lock_ttl = tonumber(ARGV[3]) + +if redis.call('SET', dispatch_lock, instance_id, 'NX', 'EX', dispatch_ttl) == false then + return 0 +end + +if redis.call('EXISTS', lock_key) == 1 then + redis.call('DEL', dispatch_lock) + return -1 +end + +redis.call('SET', lock_key, 'dispatching', 'EX', lock_ttl) +return 1 +""" + +LUA_SAFE_DELETE = """ +if redis.call('GET', KEYS[1]) == ARGV[1] then + return redis.call('DEL', KEYS[1]) +end +return 0 +""" -def health_check_server(): +def stable_hash(value: str) -> int: + return int.from_bytes( + hashlib.md5(value.encode("utf-8")).digest(), + "big" + ) + + +def health_check_server(scheduler_ref): import uvicorn from fastapi import FastAPI @@ -23,19 +72,20 @@ def health_check_server(): @health_app.get("/") def health(): - return scheduler.health() + return scheduler_ref.health() + port = int(os.environ.get("SCHEDULER_HEALTH_PORT", "8001")) threading.Thread( target=uvicorn.run, kwargs={ "app": health_app, "host": "0.0.0.0", - "port": 8001, - "log_config": None + "port": port, + "log_config": None, }, - daemon=True + daemon=True, ).start() - logger.info(f"[Health] Server started at http://0.0.0.0:8001") + logger.info("[Health] Server started at http://0.0.0.0:%s", port) class RedisTaskScheduler: @@ -50,27 +100,43 @@ class RedisTaskScheduler: self.running = False self.dispatched = 0 self.errors = 0 - self._leader = False + + self.instance_id = f"{socket.gethostname()}-{os.getpid()}" + self._shard_index = 0 + self._shard_count = 1 + self._last_heartbeat = 0.0 def push_task(self, task_name, user_id, params): try: - msg_id = self.redis.xadd( - STREAM_KEY, - fields={ - "task_name": task_name, - "user_id": user_id, - "params": json.dumps(params), - } - ) - self.redis.set( + msg_id = str(uuid.uuid4()) + msg = json.dumps({ + "msg_id": msg_id, + "task_name": task_name, + "user_id": user_id, + "params": json.dumps(params), + }) + + lock_key = f"{task_name}:{user_id}" + queue_key = f"{USER_QUEUE_PREFIX}{user_id}" + + pipe = self.redis.pipeline() + pipe.rpush(queue_key, msg) + pipe.sadd(ACTIVE_USERS, user_id) + pipe.set( f"task_tracker:{msg_id}", json.dumps({"status": "QUEUED", "task_id": None}), - ex=86400 + ex=86400, ) + pipe.execute() + + if not self.redis.exists(lock_key): + self.redis.sadd(READY_SET, user_id) + + logger.info("Task pushed: msg_id=%s task=%s user=%s", msg_id, task_name, user_id) return msg_id except Exception as e: logger.error("Push task exception %s", e, exc_info=True) - raise e + raise def get_task_status(self, msg_id: str) -> dict: raw = self.redis.get(f"task_tracker:{msg_id}") @@ -81,6 +147,7 @@ class RedisTaskScheduler: status = tracker["status"] task_id = tracker.get("task_id") result_content = tracker.get("result") or {} + if status == "DISPATCHED" and task_id: result_raw = self.redis.get(f"celery-task-meta-{task_id}") if result_raw: @@ -105,6 +172,7 @@ class RedisTaskScheduler: cleanup_pipe = self.redis.pipeline() has_cleanup = False + ready_user_ids = set() for task_id, raw_result in zip(task_ids, results): try: @@ -114,12 +182,16 @@ class RedisTaskScheduler: age = now - dispatched_at should_cleanup = False - result_data = None + result_data = {} + if raw_result is not None: result_data = json.loads(raw_result) if result_data.get("status") in ("SUCCESS", "FAILURE", "REVOKED"): should_cleanup = True - logger.info("Task finished: %s state=%s", task_id, result_data.get("status")) + logger.info( + "Task finished: %s state=%s", task_id, + result_data.get("status"), + ) elif age > TASK_TIMEOUT: should_cleanup = True logger.warning( @@ -128,9 +200,14 @@ class RedisTaskScheduler: ) if should_cleanup: - final_status = result_data.get("status", "UNKNOWN") if result_data else "EXPIRED" - cleanup_pipe.delete(lock_key) + final_status = ( + result_data.get("status", "UNKNOWN") if result_data else "EXPIRED" + ) + + self.redis.eval(LUA_SAFE_DELETE, 1, lock_key, task_id) + cleanup_pipe.hdel(PENDING_HASH, task_id) + tracker_msg_id = meta.get("msg_id") if tracker_msg_id: cleanup_pipe.set( @@ -138,146 +215,286 @@ class RedisTaskScheduler: json.dumps({ "status": final_status, "task_id": task_id, - "result": result_data.get("result") or {} + "result": result_data.get("result") or {}, }), ex=86400, ) has_cleanup = True + + parts = lock_key.split(":", 1) + if len(parts) == 2: + ready_user_ids.add(parts[1]) + except Exception as e: logger.error("Cleanup error for %s: %s", task_id, e, exc_info=True) self.errors += 1 + if has_cleanup: cleanup_pipe.execute() + if ready_user_ids: + self.redis.sadd(READY_SET, *ready_user_ids) + + def _heartbeat(self): + now = time.time() + if now - self._last_heartbeat < HEARTBEAT_INTERVAL: + return + self._last_heartbeat = now + + self.redis.hset(REGISTRY_KEY, self.instance_id, str(now)) + + all_instances = self.redis.hgetall(REGISTRY_KEY) + + alive = [] + dead = [] + for iid, ts in all_instances.items(): + if now - float(ts) < INSTANCE_TTL: + alive.append(iid) + else: + dead.append(iid) + + if dead: + pipe = self.redis.pipeline() + for iid in dead: + pipe.hdel(REGISTRY_KEY, iid) + pipe.execute() + logger.info("Cleaned dead instances: %s", dead) + + alive.sort() + self._shard_count = max(len(alive), 1) + self._shard_index = ( + alive.index(self.instance_id) if self.instance_id in alive else 0 + ) + logger.debug( + "Shard: %s/%s (instance=%s, alive=%d)", + self._shard_index, self._shard_count, + self.instance_id, len(alive), + ) + + def _is_mine(self, user_id: str) -> bool: + if self._shard_count <= 1: + return True + return stable_hash(user_id) % self._shard_count == self._shard_index + def _dispatch(self, msg_id, msg_data) -> bool: - user_id = msg_data['user_id'] - task_name = msg_data['task_name'] - params = json.loads(msg_data.get('params', "{}")) + user_id = msg_data["user_id"] + task_name = msg_data["task_name"] + params = json.loads(msg_data.get("params", "{}")) lock_key = f"{task_name}:{user_id}" + dispatch_lock = f"dispatch:{msg_id}" + + result = self.redis.eval( + LUA_ATOMIC_LOCK, 2, + dispatch_lock, lock_key, + self.instance_id, str(300), str(3600), + ) + + if result == 0: + return False + if result == -1: + return False + try: task = celery_app.send_task(task_name, kwargs=params) + except Exception as e: + pipe = self.redis.pipeline() + pipe.delete(dispatch_lock) + pipe.delete(lock_key) + pipe.execute() + self.errors += 1 + logger.error( + "send_task failed for %s:%s msg=%s: %s", + task_name, user_id, msg_id, e, exc_info=True, + ) + return False + + try: pipe = self.redis.pipeline() pipe.set(lock_key, task.id, ex=3600) pipe.hset(PENDING_HASH, task.id, json.dumps({ "lock_key": lock_key, "dispatched_at": time.time(), - "msg_id": msg_id + "msg_id": msg_id, })) - pipe.xdel(STREAM_KEY, msg_id) + pipe.delete(dispatch_lock) pipe.set( f"task_tracker:{msg_id}", json.dumps({"status": "DISPATCHED", "task_id": task.id}), ex=86400, ) pipe.execute() - self.dispatched += 1 - logger.info("Task dispatched: %s", task.id) - return True except Exception as e: + logger.error( + "Post-dispatch state update failed for %s: %s", + task.id, e, exc_info=True, + ) self.errors += 1 - logger.error("Task dispatch error for %s: %s", task_name, e, exc_info=True) - return False - def _leader_lock_extend(self, lock, interval=20): - while self._leader: - try: - lock.extend(60) - except redis.exceptions.LockNotOwnedError: - logger.warning("Lost leader lock during extend") - self._leader = False - except Exception as e: - logger.error("Lock extend error: %s", e) - for _ in range(interval): - if not self._leader: - break - time.sleep(1) + self.dispatched += 1 + logger.info("Task dispatched: %s (msg=%s)", task.id, msg_id) + return True - def schedule_loop(self): - self.running = True - self._cleanup_finished() - resp = self.redis.xread( - streams={STREAM_KEY: '0-0'}, - count=500, - block=5000, - ) - if not resp: + def _process_batch(self, user_ids): + if not user_ids: return - messages = [] - for stream_key, msgs in resp: - messages.extend(msgs) + pipe = self.redis.pipeline() + for uid in user_ids: + pipe.lindex(f"{USER_QUEUE_PREFIX}{uid}", 0) + heads = pipe.execute() - lock_keys = [] - for msg_id, msg_data in messages: - lock_keys.append(f"{msg_data['task_name']}:{msg_data['user_id']}") + candidates = [] # (user_id, msg_dict) + empty_users = [] + + for uid, head in zip(user_ids, heads): + if head is None: + empty_users.append(uid) + else: + try: + candidates.append((uid, json.loads(head))) + except (json.JSONDecodeError, TypeError) as e: + logger.error("Bad message in queue for user %s: %s", uid, e) + self.redis.lpop(f"{USER_QUEUE_PREFIX}{uid}") + + if empty_users: + pipe = self.redis.pipeline() + for uid in empty_users: + pipe.srem(ACTIVE_USERS, uid) + pipe.execute() + + if not candidates: + return + + for uid, msg in candidates: + if self._dispatch(msg["msg_id"], msg): + self.redis.lpop(f"{USER_QUEUE_PREFIX}{uid}") + + def schedule_loop(self): + self._heartbeat() + self._cleanup_finished() pipe = self.redis.pipeline() - for key in lock_keys: - pipe.exists(key) - lock_exists = pipe.execute() + pipe.smembers(READY_SET) + pipe.delete(READY_SET) + results = pipe.execute() + ready_users = results[0] or set() - deliver_keys = set() - for (msg_id, msg_data), locked in zip(messages, lock_exists): - user_id = msg_data['user_id'] - lock_key = f"{msg_data['task_name']}:{user_id}" + my_users = [uid for uid in ready_users if self._is_mine(uid)] - if locked or lock_key in deliver_keys: - continue + if not my_users: + time.sleep(0.5) + return - dispatched_successfully = self._dispatch(msg_id, msg_data) - if dispatched_successfully: - deliver_keys.add(lock_key) + self._process_batch(my_users) time.sleep(0.1) - def run_server(self): - health_check_server() + def _full_scan(self): + cursor = 0 + ready_batch = [] + while True: + cursor, user_ids = self.redis.sscan( + ACTIVE_USERS, cursor=cursor, count=1000, + ) + if user_ids: + my_users = [uid for uid in user_ids if self._is_mine(uid)] + if my_users: + pipe = self.redis.pipeline() + for uid in my_users: + pipe.lindex(f"{USER_QUEUE_PREFIX}{uid}", 0) + heads = pipe.execute() - lock = self.redis.lock( - "scheduler:leader", - timeout=60, - blocking_timeout=10, - thread_local=False + for uid, head in zip(my_users, heads): + if head is None: + continue + try: + msg = json.loads(head) + lock_key = f"{msg['task_name']}:{uid}" + ready_batch.append((uid, lock_key)) + except (json.JSONDecodeError, TypeError): + continue + + if cursor == 0: + break + + if not ready_batch: + return + + pipe = self.redis.pipeline() + for _, lock_key in ready_batch: + pipe.exists(lock_key) + lock_exists = pipe.execute() + + ready_uids = [ + uid for (uid, _), locked in zip(ready_batch, lock_exists) + if not locked + ] + + if ready_uids: + self.redis.sadd(READY_SET, *ready_uids) + logger.info("Full scan found %d ready users", len(ready_uids)) + + def run_server(self): + health_check_server(self) + self.running = True + + last_full_scan = 0.0 + full_scan_interval = 30.0 + + logger.info( + "Scheduler started: instance=%s", self.instance_id, ) + while True: try: - if lock.acquire(blocking=True): - self._leader = True - t = threading.Thread( - target=self._leader_lock_extend, - args=(lock, 20), - daemon=True - ) - t.start() - try: - while self._leader: - self.schedule_loop() - finally: - self._leader = False - t.join(timeout=30) - try: - lock.release() - except redis.exceptions.LockNotOwnedError: - pass - self.running = False - else: - time.sleep(5) + self.schedule_loop() + + now = time.time() + if now - last_full_scan > full_scan_interval: + self._full_scan() + last_full_scan = now + except Exception as e: logger.error("Scheduler exception %s", e, exc_info=True) + self.errors += 1 time.sleep(5) def health(self) -> dict: return { "running": self.running, - "pending": self.redis.xlen(STREAM_KEY), + "active_users": self.redis.scard(ACTIVE_USERS), + "ready_users": self.redis.scard(READY_SET), + "pending_tasks": self.redis.hlen(PENDING_HASH), "dispatched": self.dispatched, - "errors": self.errors + "errors": self.errors, + "shard": f"{self._shard_index}/{self._shard_count}", + "instance": self.instance_id, } + def shutdown(self): + logger.info("Scheduler shutting down: instance=%s", self.instance_id) + self.running = False + try: + self.redis.hdel(REGISTRY_KEY, self.instance_id) + except Exception as e: + logger.error("Shutdown cleanup error: %s", e) + scheduler: RedisTaskScheduler | None = None if scheduler is None: scheduler = RedisTaskScheduler() -if __name__ == '__main__': +if __name__ == "__main__": + import signal + import sys + + + def _signal_handler(signum, frame): + scheduler.shutdown() + sys.exit(0) + + + signal.signal(signal.SIGTERM, _signal_handler) + signal.signal(signal.SIGINT, _signal_handler) + scheduler.run_server() diff --git a/api/app/core/memory/read_services/generate_engine/retrieval_summary.py b/api/app/core/memory/read_services/generate_engine/retrieval_summary.py index 6b166cf2..c46e93f0 100644 --- a/api/app/core/memory/read_services/generate_engine/retrieval_summary.py +++ b/api/app/core/memory/read_services/generate_engine/retrieval_summary.py @@ -8,4 +8,4 @@ class RetrievalSummaryProcessor: @staticmethod def verify(content: str, llm_client: RedBearLLM): - return \ No newline at end of file + return diff --git a/api/app/services/memory_api_service.py b/api/app/services/memory_api_service.py index a1ceef86..82d1c463 100644 --- a/api/app/services/memory_api_service.py +++ b/api/app/services/memory_api_service.py @@ -187,7 +187,7 @@ class MemoryAPIService: } ) - logger.info(f"Memory write task submitted, end_user_id={end_user_id}") + logger.info(f"Memory write task submitted, task_id={task_id} end_user_id={end_user_id}") return { "task_id": task_id, From 2c2551e15c25373a54f43f95d6b746989aa44480 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Fri, 24 Apr 2026 14:44:27 +0800 Subject: [PATCH 220/245] feat(citation): add download_url to citations when allow_download is enabled --- api/app/services/workflow_service.py | 26 ++++++++++++++++++++------ 1 file changed, 20 insertions(+), 6 deletions(-) diff --git a/api/app/services/workflow_service.py b/api/app/services/workflow_service.py index b8f5174c..63b7073e 100644 --- a/api/app/services/workflow_service.py +++ b/api/app/services/workflow_service.py @@ -773,9 +773,16 @@ class WorkflowService: # 过滤 citations citations = result.get("citations", []) citation_cfg = feature_configs.get("citation", {}) - filtered_citations = ( - citations if isinstance(citation_cfg, dict) and citation_cfg.get("enabled") else [] - ) + if isinstance(citation_cfg, dict) and citation_cfg.get("enabled"): + allow_download = citation_cfg.get("allow_download", False) + if allow_download: + from app.core.config import settings + for c in citations: + if c.get("document_id"): + c["download_url"] = f"{settings.FILE_LOCAL_SERVER_URL}/apps/citations/{c['document_id']}/download" + filtered_citations = citations + else: + filtered_citations = [] assistant_meta = {"usage": token_usage, "audio_url": None} if filtered_citations: assistant_meta["citations"] = filtered_citations @@ -975,9 +982,16 @@ class WorkflowService: # 过滤 citations citations = event.get("data", {}).get("citations", []) citation_cfg = feature_configs.get("citation", {}) - filtered_citations = ( - citations if isinstance(citation_cfg, dict) and citation_cfg.get("enabled") else [] - ) + if isinstance(citation_cfg, dict) and citation_cfg.get("enabled"): + allow_download = citation_cfg.get("allow_download", False) + if allow_download: + from app.core.config import settings + for c in citations: + if c.get("document_id"): + c["download_url"] = f"{settings.FILE_LOCAL_SERVER_URL}/apps/citations/{c['document_id']}/download" + filtered_citations = citations + else: + filtered_citations = [] assistant_meta = {"usage": token_usage, "audio_url": None} if filtered_citations: assistant_meta["citations"] = filtered_citations From c556995f3a65e85445273a105d1ce4cdbdec131a Mon Sep 17 00:00:00 2001 From: zhaoying Date: Fri, 24 Apr 2026 15:10:32 +0800 Subject: [PATCH 221/245] feat(web): app citation features add allow_download --- web/src/assets/images/application/export.svg | 8 ++--- web/src/assets/images/application/import.svg | 8 ++--- web/src/components/Chat/ChatContent.tsx | 24 +++++++++----- web/src/components/Chat/types.ts | 3 +- web/src/i18n/en.ts | 1 + web/src/i18n/zh.ts | 1 + web/src/views/ApplicationConfig/Agent.tsx | 32 ++++++++++--------- .../FeaturesConfig/FeaturesConfigModal.tsx | 6 ++++ 8 files changed, 51 insertions(+), 32 deletions(-) diff --git a/web/src/assets/images/application/export.svg b/web/src/assets/images/application/export.svg index c07a346d..6dde8f3c 100644 --- a/web/src/assets/images/application/export.svg +++ b/web/src/assets/images/application/export.svg @@ -1,12 +1,12 @@ - 导出 + 导入 - - + + - + diff --git a/web/src/assets/images/application/import.svg b/web/src/assets/images/application/import.svg index 6dde8f3c..c07a346d 100644 --- a/web/src/assets/images/application/import.svg +++ b/web/src/assets/images/application/import.svg @@ -1,12 +1,12 @@ - 导入 + 导出 - - + + - + diff --git a/web/src/components/Chat/ChatContent.tsx b/web/src/components/Chat/ChatContent.tsx index f28b5dce..f4c89578 100644 --- a/web/src/components/Chat/ChatContent.tsx +++ b/web/src/components/Chat/ChatContent.tsx @@ -272,14 +272,22 @@ const ChatContent: FC = ({
{t('memoryConversation.citations')}
{item.meta_data?.citations?.map((citation, idx) => ( -
{ - const params = new URLSearchParams({ documentId: citation.document_id, parentId: citation.knowledge_id }); - window.open(`/#/knowledge-base/${citation.knowledge_id}/DocumentDetails?${params}`, '_blank'); - }} - >{citation.file_name}
+ +
{ + const params = new URLSearchParams({ documentId: citation.document_id, parentId: citation.knowledge_id }); + window.open(`/#/knowledge-base/${citation.knowledge_id}/DocumentDetails?${params}`, '_blank'); + }} + >{citation.file_name}
+ + {citation.download_url && +
handleDownload({ url: citation.download_url })} + >
+ } + ))} } diff --git a/web/src/components/Chat/types.ts b/web/src/components/Chat/types.ts index e7967bad..f251db3a 100644 --- a/web/src/components/Chat/types.ts +++ b/web/src/components/Chat/types.ts @@ -24,7 +24,7 @@ export interface ChatItem { subContent?: Record[]; error?: string; meta_data?: { - audio_url?: string; + audio_url?: string | null; audio_status?: string; files?: any[]; suggested_questions?: string[]; @@ -33,6 +33,7 @@ export interface ChatItem { file_name: string; knowledge_id: string; score: string; + download_url?: string; }[]; reasoning_content?: string; }, diff --git a/web/src/i18n/en.ts b/web/src/i18n/en.ts index e5a80431..2a7534c4 100644 --- a/web/src/i18n/en.ts +++ b/web/src/i18n/en.ts @@ -1470,6 +1470,7 @@ export const en = { add_questions: 'Add Option', citation: 'Citation and Attribution', citation_desc: 'Display the attribution of source documents and generated content', + allow_download: 'Allow downloading cited source text', invalidVariablesTitle: "The following undefined variables are referenced in the conversation opening. Do you want to save the opening configuration?", deep_thinking: 'Enable Deep Thinking', diff --git a/web/src/i18n/zh.ts b/web/src/i18n/zh.ts index 8d71876e..6989cf3f 100644 --- a/web/src/i18n/zh.ts +++ b/web/src/i18n/zh.ts @@ -800,6 +800,7 @@ export const zh = { add_questions: '添加选项', citation: '引用和归属', citation_desc: '显示源文档和生成内容的归属部分', + allow_download: '允许下载引用原文', invalidVariablesTitle: "对话开场白中引用了以下未定义的变量，是否保存开场白配置？", deep_thinking: '开启深度思考', diff --git a/web/src/views/ApplicationConfig/Agent.tsx b/web/src/views/ApplicationConfig/Agent.tsx index 4d86fba7..338f50f9 100644 --- a/web/src/views/ApplicationConfig/Agent.tsx +++ b/web/src/views/ApplicationConfig/Agent.tsx @@ -7,7 +7,7 @@ import { useEffect, useRef, useState, forwardRef, useImperativeHandle, useMemo } from 'react'; import { useTranslation } from 'react-i18next' import { useParams } from 'react-router-dom'; -import { Row, Col, Space, Form, Input, Button, App, Spin, Flex } from 'antd' +import { Row, Col, Space, Form, Input, Button, App, Flex } from 'antd' import Chat from './components/Chat' import RbCard from '@/components/RbCard/Card' @@ -357,21 +357,23 @@ const Agent = forwardRef m[1]))] - const variables = values?.variables - const validNames = new Set(variables.map(v => v.name)) - const invalid = usedVars.filter(v => !validNames.has(v)) - if (invalid.length > 0) { - const newVars = invalid.map((name, i) => ({ - index: variables.length + i, - name, - display_name: name, - type: 'text', - required: true, - max_length: 48, - })) + if (value.opening_statement.enabled) { + const usedVars = [...new Set([...(statement?.matchAll(/\{\{(\w+)\}\}/g) ?? [])].map(m => m[1]))] + const variables = values?.variables + const validNames = new Set(variables.map(v => v.name)) + const invalid = usedVars.filter(v => !validNames.has(v)) + if (invalid.length > 0) { + const newVars = invalid.map((name, i) => ({ + index: variables.length + i, + name, + display_name: name, + type: 'text', + required: true, + max_length: 48, + })) - form.setFieldValue('variables', [...variables, ...newVars]) + form.setFieldValue('variables', [...variables, ...newVars]) + } } } const modelLogo = useMemo(() => { diff --git a/web/src/views/ApplicationConfig/components/FeaturesConfig/FeaturesConfigModal.tsx b/web/src/views/ApplicationConfig/components/FeaturesConfig/FeaturesConfigModal.tsx index 57d11295..e3664a03 100644 --- a/web/src/views/ApplicationConfig/components/FeaturesConfig/FeaturesConfigModal.tsx +++ b/web/src/views/ApplicationConfig/components/FeaturesConfig/FeaturesConfigModal.tsx @@ -155,6 +155,12 @@ const FeaturesConfigModal = forwardRef +
From c7c1570d406472d82dcb2c17157ee086664df01d Mon Sep 17 00:00:00 2001 From: zhaoying Date: Fri, 24 Apr 2026 15:18:14 +0800 Subject: [PATCH 222/245] feat(web): app citations --- web/src/components/Chat/ChatContent.tsx | 3 +-- web/src/views/ApplicationConfig/Agent.tsx | 2 +- 2 files changed, 2 insertions(+), 3 deletions(-) diff --git a/web/src/components/Chat/ChatContent.tsx b/web/src/components/Chat/ChatContent.tsx index f4c89578..a785ea49 100644 --- a/web/src/components/Chat/ChatContent.tsx +++ b/web/src/components/Chat/ChatContent.tsx @@ -272,9 +272,8 @@ const ChatContent: FC = ({
{t('memoryConversation.citations')}
{item.meta_data?.citations?.map((citation, idx) => ( - +
{ const params = new URLSearchParams({ documentId: citation.document_id, parentId: citation.knowledge_id }); diff --git a/web/src/views/ApplicationConfig/Agent.tsx b/web/src/views/ApplicationConfig/Agent.tsx index 338f50f9..d7455793 100644 --- a/web/src/views/ApplicationConfig/Agent.tsx +++ b/web/src/views/ApplicationConfig/Agent.tsx @@ -357,7 +357,7 @@ const Agent = forwardRef m[1]))] const variables = values?.variables const validNames = new Set(variables.map(v => v.name)) From cedf47b3bca5bf999b35f7b1bd74e6028e63d1f4 Mon Sep 17 00:00:00 2001 From: wwq Date: Fri, 24 Apr 2026 15:29:33 +0800 Subject: [PATCH 223/245] fix(workflow): rectify error handling and bolster execution logging --- api/app/services/workflow_service.py | 39 +--------------------------- 1 file changed, 1 insertion(+), 38 deletions(-) diff --git a/api/app/services/workflow_service.py b/api/app/services/workflow_service.py index 6c93893a..ea302a9c 100644 --- a/api/app/services/workflow_service.py +++ b/api/app/services/workflow_service.py @@ -945,44 +945,7 @@ class WorkflowService: event_type = event.get("event") event_data = event.get("data", {}) - # 持久化节点执行记录 - if event_type == "node_end": - node_id = event_data.get("node_id") - node_cfg = next((n for n in config.nodes if n.get("id") == node_id), {}) - self.db.add(WorkflowNodeExecution( - execution_id=execution.id, - node_id=node_id, - node_type=node_cfg.get("type", "unknown"), - node_name=node_cfg.get("data", {}).get("label") or node_id, - execution_order=_node_order_counter, - status="completed", - input_data=event_data.get("input"), - output_data=event_data.get("output"), - elapsed_time=event_data.get("elapsed_time"), - token_usage=event_data.get("token_usage"), - )) - self.db.commit() - _node_order_counter += 1 - - elif event_type == "node_error": - node_id = event_data.get("node_id") - node_cfg = next((n for n in config.nodes if n.get("id") == node_id), {}) - self.db.add(WorkflowNodeExecution( - execution_id=execution.id, - node_id=node_id, - node_type=node_cfg.get("type", "unknown"), - node_name=node_cfg.get("data", {}).get("label") or node_id, - execution_order=_node_order_counter, - status="failed", - input_data=event_data.get("input"), - output_data=None, - error_message=event_data.get("error"), - elapsed_time=event_data.get("elapsed_time"), - )) - self.db.commit() - _node_order_counter += 1 - - elif event_type == "cycle_item": + if event_type == "cycle_item": cycle_id = event_data.get("cycle_id") if cycle_id not in _cycle_items: _cycle_items[cycle_id] = [] From 74be09340cd697ef17d067d66ae2c803faf8c609 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Fri, 24 Apr 2026 15:56:06 +0800 Subject: [PATCH 224/245] feat(multimodal): support tenant-aware document image storage and improve image placeholder labeling MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit - Pass workspace_id to multimodal_service.process_files across app_chat_service, draft_run_service - Fetch tenant_id from workspace in multimodal_service for proper file storage scoping - Update image placeholder format from "[第N页 第M张图片]" to "[图片 第N页 第M张图片]" for clarity - Add strict URL preservation rules to system prompt for agents handling document images - Refactor _save_doc_image_to_storage to accept explicit tenant_id and workspace_id instead of inferring from FileMetadata --- api/app/services/app_chat_service.py | 6 ++++-- api/app/services/draft_run_service.py | 11 ++++++++--- api/app/services/multimodal_service.py | 23 +++++++++++------------ 3 files changed, 23 insertions(+), 17 deletions(-) diff --git a/api/app/services/app_chat_service.py b/api/app/services/app_chat_service.py index bc5ae964..407a1cfc 100644 --- a/api/app/services/app_chat_service.py +++ b/api/app/services/app_chat_service.py @@ -170,7 +170,8 @@ class AppChatService: fu_config = fu_config.model_dump() doc_img_recognition = isinstance(fu_config, dict) and fu_config.get("document_image_recognition", False) processed_files = await multimodal_service.process_files( - files, document_image_recognition=doc_img_recognition + files, document_image_recognition=doc_img_recognition, + workspace_id=workspace_id ) logger.info(f"处理了 {len(processed_files)} 个文件") if doc_img_recognition and "vision" in (api_key_obj.capability or []) and any( @@ -462,7 +463,8 @@ class AppChatService: fu_config = fu_config.model_dump() doc_img_recognition = isinstance(fu_config, dict) and fu_config.get("document_image_recognition", False) processed_files = await multimodal_service.process_files( - files, document_image_recognition=doc_img_recognition + files, document_image_recognition=doc_img_recognition, + workspace_id=workspace_id ) logger.info(f"处理了 {len(processed_files)} 个文件") if doc_img_recognition and "vision" in (api_key_obj.capability or []) and any( diff --git a/api/app/services/draft_run_service.py b/api/app/services/draft_run_service.py index 2869326f..9d04369b 100644 --- a/api/app/services/draft_run_service.py +++ b/api/app/services/draft_run_service.py @@ -655,7 +655,8 @@ class AgentRunService: fu_config = fu_config.model_dump() doc_img_recognition = isinstance(fu_config, dict) and fu_config.get("document_image_recognition", False) processed_files = await multimodal_service.process_files( - files, document_image_recognition=doc_img_recognition + files, document_image_recognition=doc_img_recognition, + workspace_id=workspace_id ) logger.info(f"处理了 {len(processed_files)} 个文件，provider={provider}") capability = api_key_config.get("capability", []) @@ -936,7 +937,8 @@ class AgentRunService: fu_config = fu_config.model_dump() doc_img_recognition = isinstance(fu_config, dict) and fu_config.get("document_image_recognition", False) processed_files = await multimodal_service.process_files( - files, document_image_recognition=doc_img_recognition + files, document_image_recognition=doc_img_recognition, + workspace_id=workspace_id ) logger.info(f"处理了 {len(processed_files)} 个文件，provider={provider}") capability = api_key_config.get("capability", []) @@ -947,8 +949,11 @@ class AgentRunService: ) if has_doc_with_images: agent.system_prompt += ( - "\n\n文档中包含图片，图片位置已在文本中以 [第N页 第M张图片]: URL 标记。" + "\n\n文档中包含图片，图片位置已在文本中以 [图片 第N页 第M张图片]: URL 标记。" "请在回答中用 Markdown 格式 ![描述](URL) 展示相关图片，做到图文并茂。" + "**规则1：图片URL必须原封不动、一字不差地复制，禁止修改、禁止省略任何字符**" + "**规则2：禁止修改URL中UUID里的任何数字和字母**" + "**规则3：直接使用 ![描述](完整URL) 格式输出**" ) agent.agent = create_agent( model=agent.llm, diff --git a/api/app/services/multimodal_service.py b/api/app/services/multimodal_service.py index 946649c3..960bf6bb 100644 --- a/api/app/services/multimodal_service.py +++ b/api/app/services/multimodal_service.py @@ -24,6 +24,7 @@ import chardet import httpx import magic import openpyxl +import uuid from docx import Document from sqlalchemy.orm import Session @@ -344,6 +345,7 @@ class MultimodalService: async def process_files( self, files: Optional[List[FileInput]], + workspace_id: uuid.UUID = None, document_image_recognition: bool = False, ) -> List[Dict[str, Any]]: """ @@ -383,17 +385,20 @@ class MultimodalService: # 仅当开关开启且模型支持视觉时，才提取文档内嵌图片 if document_image_recognition and "vision" in self.capability: img_infos = await self.extract_document_images(file) + from app.models.workspace_model import Workspace as WorkspaceModel + ws = self.db.query(WorkspaceModel).filter(WorkspaceModel.id == workspace_id).first() + tenant_id = ws.tenant_id if ws else None for img_info in img_infos: page = img_info["page"] index = img_info["index"] ext = img_info.get("ext", "png") try: - _, img_url = await self._save_doc_image_to_storage(img_info["bytes"], ext) + _, img_url = await self._save_doc_image_to_storage(img_info["bytes"], ext, tenant_id, workspace_id) placeholder = f"第{page}页 第{index + 1}张图片" if page > 0 else f"第{index + 1}张图片" # 在文本内容中追加图片位置标记 if result and result[-1].get("type") in ("text", "document"): key = "text" if "text" in result[-1] else list(result[-1].keys())[-1] - result[-1][key] = result[-1].get(key, "") + f"\n[{placeholder}]: {img_url}" + result[-1][key] = result[-1].get(key, "") + f"\n[图片 {placeholder}]: {img_url}" # 将图片以视觉格式追加到消息内容中 img_file = FileInput( type=FileType.IMAGE, @@ -475,32 +480,26 @@ class MultimodalService: file_name = file_metadata.file_name if file_metadata else "unknown" return await strategy.format_document(file_name, text) + @staticmethod async def _save_doc_image_to_storage( - self, img_bytes: bytes, ext: str, + tenant_id: uuid.UUID, + workspace_id: uuid.UUID, ) -> tuple[str, str]: """ 将文档内嵌图片保存到存储后端，写入 FileMetadata。 - tenant_id / workspace_id 从 api_config 所在的 FileMetadata 上下文获取， - 无法获取时使用占位 UUID（图片仍可通过 permanent URL 访问）。 Returns: (file_id_str, permanent_url) """ - import uuid as _uuid from app.services.file_storage_service import FileStorageService, generate_file_key from app.db import get_db_context - file_id = _uuid.uuid4() + file_id = uuid.uuid4() file_ext = f".{ext}" if not ext.startswith(".") else ext content_type = f"image/{ext}" - # tenant_id / workspace_id 尽量从已有 FileMetadata 推断，否则用占位值 - placeholder = _uuid.UUID(int=0) - tenant_id = placeholder - workspace_id = placeholder - file_key = generate_file_key(tenant_id, workspace_id, file_id, file_ext) storage_svc = FileStorageService() await storage_svc.storage.upload(file_key, img_bytes, content_type) From cf8db47389f3d0db856fc2a9be6ab26f27b1e78c Mon Sep 17 00:00:00 2001 From: wwq Date: Fri, 24 Apr 2026 17:02:03 +0800 Subject: [PATCH 225/245] feat(workflow): augment logging capabilities with execution status and loop support - Augment workflow logs with execution status fields and loop node information. - Refactor log service to handle distinct processing logic for workflows and agents. - Construct message and node logs derived from workflow_executions data. --- api/app/controllers/app_log_controller.py | 27 ++- .../workflow/nodes/cycle_graph/iteration.py | 2 + .../core/workflow/nodes/cycle_graph/loop.py | 2 + api/app/schemas/app_log_schema.py | 2 + api/app/services/app_log_service.py | 165 ++++++++++++++---- 5 files changed, 158 insertions(+), 40 deletions(-) diff --git a/api/app/controllers/app_log_controller.py b/api/app/controllers/app_log_controller.py index dea555b9..ea7962c1 100644 --- a/api/app/controllers/app_log_controller.py +++ b/api/app/controllers/app_log_controller.py @@ -9,7 +9,7 @@ from app.core.logging_config import get_business_logger from app.core.response_utils import success from app.db import get_db from app.dependencies import get_current_user, cur_workspace_access_guard -from app.schemas.app_log_schema import AppLogConversation, AppLogConversationDetail +from app.schemas.app_log_schema import AppLogConversation, AppLogConversationDetail, AppLogMessage from app.schemas.response_schema import PageData, PageMeta from app.services.app_service import AppService from app.services.app_log_service import AppLogService @@ -78,17 +78,32 @@ def get_app_log_detail( # 验证应用访问权限 app_service = AppService(db) - app_service.get_app(app_id, workspace_id) + app = app_service.get_app(app_id, workspace_id) # 使用 Service 层查询 log_service = AppLogService(db) - conversation, node_executions_map = log_service.get_conversation_detail( + conversation, messages, node_executions_map = log_service.get_conversation_detail( app_id=app_id, conversation_id=conversation_id, - workspace_id=workspace_id + workspace_id=workspace_id, + app_type=app.type ) - detail = AppLogConversationDetail.model_validate(conversation) - detail.node_executions_map = node_executions_map + # 构建基础会话信息（不经过 ORM relationship） + base = AppLogConversation.model_validate(conversation) + + # 单独处理 messages，避免触发 SQLAlchemy relationship 校验 + if messages and isinstance(messages[0], AppLogMessage): + # 工作流：已经是 AppLogMessage 实例 + msg_list = messages + else: + # Agent：ORM Message 对象逐个转换 + msg_list = [AppLogMessage.model_validate(m) for m in messages] + + detail = AppLogConversationDetail( + **base.model_dump(), + messages=msg_list, + node_executions_map=node_executions_map, + ) return success(data=detail) diff --git a/api/app/core/workflow/nodes/cycle_graph/iteration.py b/api/app/core/workflow/nodes/cycle_graph/iteration.py index 1633b9c7..1ac46180 100644 --- a/api/app/core/workflow/nodes/cycle_graph/iteration.py +++ b/api/app/core/workflow/nodes/cycle_graph/iteration.py @@ -180,6 +180,8 @@ class IterationRuntime: "cycle_id": self.node_id, "cycle_idx": idx, "node_id": node_name, + "node_type": node_type, + "node_name": node_cfg.get("data", {}).get("label") if node_cfg else node_name, "input": result.get("node_outputs", {}).get(node_name, {}).get("input") if not cycle_variable else cycle_variable, "output": result.get("node_outputs", {}).get(node_name, {}).get("output") diff --git a/api/app/core/workflow/nodes/cycle_graph/loop.py b/api/app/core/workflow/nodes/cycle_graph/loop.py index e555a228..ca432a04 100644 --- a/api/app/core/workflow/nodes/cycle_graph/loop.py +++ b/api/app/core/workflow/nodes/cycle_graph/loop.py @@ -210,6 +210,8 @@ class LoopRuntime: "cycle_id": self.node_id, "cycle_idx": idx, "node_id": node_name, + "node_type": node_type, + "node_name": node_name, "input": result.get("node_outputs", {}).get(node_name, {}).get("input") if not cycle_variable else cycle_variable, "output": result.get("node_outputs", {}).get(node_name, {}).get("output") diff --git a/api/app/schemas/app_log_schema.py b/api/app/schemas/app_log_schema.py index a60a8428..ce9ddd44 100644 --- a/api/app/schemas/app_log_schema.py +++ b/api/app/schemas/app_log_schema.py @@ -14,6 +14,7 @@ class AppLogMessage(BaseModel): conversation_id: uuid.UUID role: str = Field(description="角色: user / assistant / system") content: str + status: Optional[str] = Field(default=None, description="执行状态（工作流专用）: completed / failed") meta_data: Optional[Dict[str, Any]] = None created_at: datetime.datetime @@ -58,6 +59,7 @@ class AppLogNodeExecution(BaseModel): input: Optional[Any] = None process: Optional[Any] = None output: Optional[Any] = None + cycle_items: Optional[List[Any]] = None elapsed_time: Optional[float] = None token_usage: Optional[Dict[str, Any]] = None diff --git a/api/app/services/app_log_service.py b/api/app/services/app_log_service.py index b2801df3..e8a80d40 100644 --- a/api/app/services/app_log_service.py +++ b/api/app/services/app_log_service.py @@ -1,16 +1,17 @@ """应用日志服务层""" import uuid +import datetime as dt from typing import Optional, Tuple -from datetime import datetime from sqlalchemy import select from sqlalchemy.orm import Session from app.core.logging_config import get_business_logger +from app.models.app_model import AppType from app.models.conversation_model import Conversation, Message from app.models.workflow_model import WorkflowExecution from app.repositories.conversation_repository import ConversationRepository, MessageRepository -from app.schemas.app_log_schema import AppLogNodeExecution +from app.schemas.app_log_schema import AppLogMessage, AppLogNodeExecution logger = get_business_logger() @@ -83,51 +84,40 @@ class AppLogService: self, app_id: uuid.UUID, conversation_id: uuid.UUID, - workspace_id: uuid.UUID - ) -> Tuple[Conversation, dict[str, list[AppLogNodeExecution]]]: + workspace_id: uuid.UUID, + app_type: str = AppType.AGENT + ) -> Tuple[Conversation, list, dict[str, list[AppLogNodeExecution]]]: """ - 查询会话详情（包含消息和工作流节点执行记录） - - Args: - app_id: 应用 ID - conversation_id: 会话 ID - workspace_id: 工作空间 ID + 查询会话详情 Returns: - Tuple[Conversation, dict[str, list[AppLogNodeExecution]]]: - (包含消息的会话对象, 按消息ID分组的节点执行记录) - - Raises: - ResourceNotFoundException: 当会话不存在时 + Tuple[Conversation, list[AppLogMessage|Message], dict[str, list[AppLogNodeExecution]]] """ logger.info( "查询应用日志会话详情", extra={ "app_id": str(app_id), "conversation_id": str(conversation_id), - "workspace_id": str(workspace_id) + "workspace_id": str(workspace_id), + "app_type": app_type } ) - # 查询会话 conversation = self.conversation_repository.get_conversation_for_app_log( conversation_id=conversation_id, app_id=app_id, workspace_id=workspace_id ) - # 查询消息（按时间正序） - messages = self.message_repository.get_messages_by_conversation( - conversation_id=conversation_id - ) - - # 将消息附加到会话对象 - conversation.messages = messages - - # 查询工作流节点执行记录（按消息分组） - _, node_executions_map = self._get_workflow_node_executions_with_map( - conversation_id, messages - ) + if app_type == AppType.WORKFLOW: + messages, node_executions_map = self._get_workflow_messages_and_nodes(conversation_id) + else: + messages = self.message_repository.get_messages_by_conversation( + conversation_id=conversation_id + ) + _, node_executions_map = self._get_workflow_node_executions_with_map( + conversation_id, messages + ) logger.info( "查询应用日志会话详情成功", @@ -139,7 +129,97 @@ class AppLogService: } ) - return conversation, node_executions_map + return conversation, messages, node_executions_map + + def _get_workflow_messages_and_nodes( + self, + conversation_id: uuid.UUID, + ) -> Tuple[list[AppLogMessage], dict[str, list[AppLogNodeExecution]]]: + """ + 工作流应用专用：从 workflow_executions 构建 messages 和节点日志。 + + 每条 WorkflowExecution 对应一轮对话： + - user message：来自 execution.input_data + - assistant message：来自 execution.output_data（失败时内容为错误信息） + 节点日志以 execution id 为 key 分组。 + + Returns: + (messages 列表, node_executions_map) + """ + stmt = ( + select(WorkflowExecution) + .where( + WorkflowExecution.conversation_id == conversation_id, + WorkflowExecution.status.in_(["completed", "failed"]) + ) + .order_by(WorkflowExecution.started_at.asc()) + ) + executions = list(self.db.scalars(stmt).all()) + + messages: list[AppLogMessage] = [] + node_executions_map: dict[str, list[AppLogNodeExecution]] = {} + + for execution in executions: + started_at = execution.started_at or dt.datetime.now() + completed_at = execution.completed_at or started_at + + # assistant message 的 id，同时作为 node_executions_map 的 key + assistant_msg_id = uuid.uuid5(execution.id, "assistant") + + # --- user message（输入）--- + input_content = _extract_text(execution.input_data) + user_msg = AppLogMessage( + id=uuid.uuid5(execution.id, "user"), + conversation_id=conversation_id, + role="user", + content=input_content, + meta_data=None, + created_at=started_at, + ) + messages.append(user_msg) + + # --- assistant message（输出）--- + if execution.status == "completed": + output_content = _extract_text(execution.output_data) + meta = {"usage": execution.token_usage or {}, "elapsed_time": execution.elapsed_time} + else: + output_content = _extract_text(execution.output_data) or "" + meta = {"error": execution.error_message, "error_node_id": execution.error_node_id} + + assistant_msg = AppLogMessage( + id=assistant_msg_id, + conversation_id=conversation_id, + role="assistant", + content=output_content, + status=execution.status, + meta_data=meta, + created_at=completed_at, + ) + messages.append(assistant_msg) + + # --- 节点执行记录，key 与 assistant message id 一致 --- + execution_nodes = [] + for node_exec in execution.node_executions: + output_data = dict(node_exec.output_data or {}) + cycle_items = output_data.pop("cycle_items", None) + execution_nodes.append(AppLogNodeExecution( + node_id=node_exec.node_id, + node_type=node_exec.node_type, + node_name=node_exec.node_name, + status=node_exec.status, + error=node_exec.error_message, + input=node_exec.input_data, + process=None, + output=output_data, + cycle_items=cycle_items, + elapsed_time=node_exec.elapsed_time, + token_usage=node_exec.token_usage, + )) + + if execution_nodes: + node_executions_map[str(assistant_msg_id)] = execution_nodes + + return messages, node_executions_map def _get_workflow_node_executions_with_map( self, @@ -191,6 +271,8 @@ class AppLogService: # 构建节点执行记录列表 execution_nodes = [] for node_exec in execution.node_executions: + output_data = dict(node_exec.output_data or {}) + cycle_items = output_data.pop("cycle_items", None) node_execution = AppLogNodeExecution( node_id=node_exec.node_id, node_type=node_exec.node_type, @@ -199,7 +281,8 @@ class AppLogService: error=node_exec.error_message, input=node_exec.input_data, process=None, - output=node_exec.output_data, + output=output_data, + cycle_items=cycle_items, elapsed_time=node_exec.elapsed_time, token_usage=node_exec.token_usage, ) @@ -223,9 +306,9 @@ class AppLogService: if msg_id_str in used_message_ids: continue if msg.created_at and msg.created_at >= execution.started_at: - dt = (msg.created_at - execution.started_at).total_seconds() - if best_dt is None or dt < best_dt: - best_dt = dt + delta = (msg.created_at - execution.started_at).total_seconds() + if best_dt is None or delta < best_dt: + best_dt = delta best_msg = msg if not best_msg: @@ -236,3 +319,17 @@ class AppLogService: node_executions_map[msg_id_str] = execution_nodes return node_executions, node_executions_map + + +def _extract_text(data: Optional[dict]) -> str: + """从 workflow execution 的 input_data / output_data 中提取可读文本。 + + 优先取 'text'、'content'、'output' 字段；若都没有则 JSON 序列化整个 dict。 + """ + if not data: + return "" + for key in ("text", "content", "output", "result", "answer"): + if key in data and isinstance(data[key], str): + return data[key] + import json + return json.dumps(data, ensure_ascii=False) From f2aedd29bcc3ba0c5e7542aa6d4dc266805e5ce0 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Fri, 24 Apr 2026 17:09:01 +0800 Subject: [PATCH 226/245] refactor(http_request): simplify request handling and remove unused fields - Removed `last_request` field and related logic for storing raw request string - Replaced `_extract_output` and `_extract_extra_fields` to use `process_data` instead of `request` - Updated `_build_content` to directly parse JSON body without intermediate rendering step - Modified `execute` to generate `process_data` from actual HTTP request object instead of manual string building - Added `process_data` field to `HttpRequestNodeOutput` model for consistent debugging info --- .../workflow/nodes/http_request/config.py | 5 + .../core/workflow/nodes/http_request/node.py | 174 ++++-------------- 2 files changed, 39 insertions(+), 140 deletions(-) diff --git a/api/app/core/workflow/nodes/http_request/config.py b/api/app/core/workflow/nodes/http_request/config.py index 72474436..66079ada 100644 --- a/api/app/core/workflow/nodes/http_request/config.py +++ b/api/app/core/workflow/nodes/http_request/config.py @@ -272,6 +272,11 @@ class HttpRequestNodeOutput(BaseModel): description="HTTP response body", ) + process_data: dict = Field( + default_factory=dict, + description="Raw HTTP request details for debugging", + ) + # files: list[File] = Field( # ... # ) diff --git a/api/app/core/workflow/nodes/http_request/node.py b/api/app/core/workflow/nodes/http_request/node.py index cfdb95b0..da13ceec 100644 --- a/api/app/core/workflow/nodes/http_request/node.py +++ b/api/app/core/workflow/nodes/http_request/node.py @@ -160,7 +160,6 @@ class HttpRequestNode(BaseNode): def __init__(self, node_config: dict[str, Any], workflow_config: dict[str, Any], down_stream_nodes: list[str]): super().__init__(node_config, workflow_config, down_stream_nodes) self.typed_config: HttpRequestNodeConfig | None = None - self.last_request: str = "" def _output_types(self) -> dict[str, VariableType]: return { @@ -171,47 +170,6 @@ class HttpRequestNode(BaseNode): "output": VariableType.STRING } - def _extract_output(self, business_result: Any) -> Any: - if isinstance(business_result, dict): - result = {k: v for k, v in business_result.items() if k != "request"} - return result - return business_result - - def _extract_extra_fields(self, business_result: Any) -> dict[str, Any]: - if isinstance(business_result, dict) and "request" in business_result: - return { - "process": { - "request": business_result.get("request", "") - } - } - return {} - - def _wrap_error( - self, - error_message: str, - elapsed_time: float, - state: WorkflowState, - variable_pool: VariablePool - ) -> dict[str, Any]: - input_data = self._extract_input(state, variable_pool) - node_output = { - "node_id": self.node_id, - "node_type": self.node_type, - "node_name": self.node_name, - "status": "failed", - "input": input_data, - "output": None, - "process": {"request": self.last_request} if self.last_request else None, - "elapsed_time": elapsed_time, - "token_usage": None, - "error": error_message - } - return { - "node_outputs": {self.node_id: node_output}, - "error": error_message, - "error_node": self.node_id - } - def _build_timeout(self) -> Timeout: """ Build httpx Timeout configuration. @@ -297,13 +255,9 @@ class HttpRequestNode(BaseNode): case HttpContentType.NONE: return {} case HttpContentType.JSON: - rendered_body = self._render_template( + content["json"] = json.loads(self._render_template( self.typed_config.body.data, variable_pool - ).strip() - if not rendered_body: - content["json"] = {} - else: - content["json"] = json.loads(rendered_body) + )) case HttpContentType.FROM_DATA: data = {} files = [] @@ -371,61 +325,15 @@ class HttpRequestNode(BaseNode): case _: raise RuntimeError(f"HttpRequest method not supported: {self.typed_config.method}") - def _generate_raw_request( - self, - variable_pool: VariablePool, - url: str, - headers: dict[str, str], - params: dict[str, str], - content: dict[str, Any] - ) -> str: - """ - Generate raw HTTP request format for debugging. + def _extract_output(self, business_result: Any) -> Any: + if isinstance(business_result, dict): + return {k: v for k, v in business_result.items() if k != "process_data"} + return business_result - Args: - variable_pool: Variable Pool - url: Rendered URL - headers: Request headers - params: Query parameters - content: Request body content - - Returns: - Raw HTTP request string - """ - method = self.typed_config.method.value - - if params: - param_str = "&".join([f"{k}={v}" for k, v in params.items()]) - full_url = f"{url}?{param_str}" if "?" not in url else f"{url}&{param_str}" - else: - full_url = url - - lines = [f"{method} {full_url} HTTP/1.1"] - - for key, value in headers.items(): - lines.append(f"{key}: {value}") - - if "json" in content and content["json"]: - json_body = json.dumps(content["json"], ensure_ascii=False) - lines.append(f"Content-Length: {len(json_body)}") - lines.append("") - lines.append(json_body) - elif "data" in content and "files" not in content: - if isinstance(content["data"], dict): - body_str = "&".join([f"{k}={v}" for k, v in content["data"].items()]) - lines.append(f"Content-Length: {len(body_str)}") - lines.append("") - lines.append(body_str) - elif "content" in content: - lines.append(f"Content-Length: {len(content['content'])}") - lines.append("") - lines.append(content["content"]) - elif "files" in content: - lines.append("Content-Length: 0") - lines.append("") - lines.append("# Note: This request includes file uploads") - - return "\r\n".join(lines) + def _extract_extra_fields(self, business_result: Any) -> dict: + if isinstance(business_result, dict) and "process_data" in business_result: + return {"process": business_result["process_data"]} + return {} async def execute(self, state: WorkflowState, variable_pool: VariablePool) -> dict | str: """ @@ -445,47 +353,42 @@ class HttpRequestNode(BaseNode): - str: Branch identifier (e.g. "ERROR") when branching is enabled """ self.typed_config = HttpRequestNodeConfig(**self.config) - - # Build request components - headers = self._build_header(variable_pool) | self._build_auth(variable_pool) - params = self._build_params(variable_pool) - content = await self._build_content(variable_pool) - url = self._render_template(self.typed_config.url, variable_pool) - - logger.info(f"Node {self.node_id}: headers={headers}, params={params}, content keys={list(content.keys())}") - - # Generate raw HTTP request for debugging - raw_request = self._generate_raw_request(variable_pool, url, headers, params, content) - self.last_request = raw_request - logger.info(f"Node {self.node_id}: Generated HTTP request:\n{raw_request}") - + rendered_url = self._render_template(self.typed_config.url, variable_pool) + built_headers = self._build_header(variable_pool) | self._build_auth(variable_pool) + built_params = self._build_params(variable_pool) async with httpx.AsyncClient( verify=self.typed_config.verify_ssl, timeout=self._build_timeout(), - headers=headers, - params=params, + headers=built_headers, + params=built_params, follow_redirects=True ) as client: retries = self.typed_config.retry.max_attempts while retries > 0: try: request_func = self._get_client_method(client) + built_content = await self._build_content(variable_pool) resp = await request_func( - url=url, - **content + url=rendered_url, + **built_content ) resp.raise_for_status() logger.info(f"Node {self.node_id}: HTTP request succeeded") response = HttpResponse(resp) - return { - **HttpRequestNodeOutput( - body=response.body, - status_code=resp.status_code, - headers=resp.headers, - files=response.files - ).model_dump(), - "request": raw_request - } + # Build raw request summary for process_data + raw_request = ( + f"{self.typed_config.method.upper()} {resp.request.url} HTTP/1.1\r\n" + + "".join(f"{k}: {v}\r\n" for k, v in resp.request.headers.items()) + + "\r\n" + + (resp.request.content.decode(errors="replace") if resp.request.content else "") + ) + return HttpRequestNodeOutput( + body=response.body, + status_code=resp.status_code, + headers=resp.headers, + files=response.files, + process_data={"request": raw_request}, + ).model_dump() except (httpx.HTTPStatusError, httpx.RequestError) as e: logger.error(f"HTTP request node exception: {e}") retries -= 1 @@ -501,19 +404,10 @@ class HttpRequestNode(BaseNode): logger.warning( f"Node {self.node_id}: HTTP request failed, returning default result" ) - error_result = self.typed_config.error_handle.default.model_dump() - error_result["request"] = raw_request - return error_result + return self.typed_config.error_handle.default.model_dump() case HttpErrorHandle.BRANCH: logger.warning( f"Node {self.node_id}: HTTP request failed, switching to error handling branch" ) - return { - "output": "ERROR", - "body": "", - "status_code": 500, - "headers": {}, - "files": [], - "request": raw_request - } + return {"output": "ERROR"} raise RuntimeError("http request failed") From 8f31236303c8c33ea7d6478d1ef47843b3e66ba5 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Fri, 24 Apr 2026 17:48:15 +0800 Subject: [PATCH 227/245] fix(app_chat_service,draft_run_service): move system_prompt augmentation before LangChainAgent instantiation --- api/app/services/app_chat_service.py | 46 +++++++++++------------- api/app/services/draft_run_service.py | 50 +++++++++++---------------- 2 files changed, 41 insertions(+), 55 deletions(-) diff --git a/api/app/services/app_chat_service.py b/api/app/services/app_chat_service.py index 407a1cfc..39a3d4d4 100644 --- a/api/app/services/app_chat_service.py +++ b/api/app/services/app_chat_service.py @@ -399,24 +399,6 @@ class AppChatService: # 获取模型参数 model_parameters = config.model_parameters - # 创建 LangChain Agent - agent = LangChainAgent( - model_name=api_key_obj.model_name, - api_key=api_key_obj.api_key, - provider=api_key_obj.provider, - api_base=api_key_obj.api_base, - is_omni=api_key_obj.is_omni, - temperature=model_parameters.get("temperature", 0.7), - max_tokens=model_parameters.get("max_tokens", 2000), - system_prompt=system_prompt, - tools=tools, - streaming=True, - deep_thinking=model_parameters.get("deep_thinking", False), - thinking_budget_tokens=model_parameters.get("thinking_budget_tokens"), - json_output=model_parameters.get("json_output", False), - capability=api_key_obj.capability or [], - ) - model_info = ModelInfo( model_name=api_key_obj.model_name, provider=api_key_obj.provider, @@ -471,16 +453,28 @@ class AppChatService: f.type == FileType.DOCUMENT for f in files ): from langchain.agents import create_agent - agent.system_prompt += ( - "\n\n文档中包含图片，图片位置已在文本中以 [第N页 第M张图片]: URL 标记。" - "请在回答中用 Markdown 格式 ![描述](URL) 展示相关图片，做到图文并茂。" - ) - agent.agent = create_agent( - model=agent.llm, - tools=agent._wrap_tools_with_tracking(agent.tools) if agent.tools else None, - system_prompt=agent.system_prompt + system_prompt += ( + "\n\n文档文字中包含图片位置标记如 [图片 第2页 第1张]: http://...，请在回答中用 Markdown 格式 ![图片描述](url) 展示对应图片。" ) + # 创建 LangChain Agent + agent = LangChainAgent( + model_name=api_key_obj.model_name, + api_key=api_key_obj.api_key, + provider=api_key_obj.provider, + api_base=api_key_obj.api_base, + is_omni=api_key_obj.is_omni, + temperature=model_parameters.get("temperature", 0.7), + max_tokens=model_parameters.get("max_tokens", 2000), + system_prompt=system_prompt, + tools=tools, + streaming=True, + deep_thinking=model_parameters.get("deep_thinking", False), + thinking_budget_tokens=model_parameters.get("thinking_budget_tokens"), + json_output=model_parameters.get("json_output", False), + capability=api_key_obj.capability or [], + ) + # 为需要运行时上下文的工具注入上下文 for t in tools: if hasattr(t, 'tool_instance') and hasattr(t.tool_instance, 'set_runtime_context'): diff --git a/api/app/services/draft_run_service.py b/api/app/services/draft_run_service.py index 9d04369b..d3cb12db 100644 --- a/api/app/services/draft_run_service.py +++ b/api/app/services/draft_run_service.py @@ -875,24 +875,6 @@ class AgentRunService: user_rag_memory_id) tools.extend(memory_tools) - # 4. 创建 LangChain Agent - agent = LangChainAgent( - model_name=api_key_config["model_name"], - api_key=api_key_config["api_key"], - provider=api_key_config.get("provider", "openai"), - api_base=api_key_config.get("api_base"), - is_omni=api_key_config.get("is_omni", False), - temperature=effective_params.get("temperature", 0.7), - max_tokens=effective_params.get("max_tokens", 2000), - system_prompt=system_prompt, - tools=tools, - streaming=True, - deep_thinking=effective_params.get("deep_thinking", False), - thinking_budget_tokens=effective_params.get("thinking_budget_tokens"), - json_output=effective_params.get("json_output", False), - capability=api_key_config.get("capability", []), - ) - # 5. 处理会话ID（创建或验证），新会话时写入开场白 is_new_conversation = not conversation_id opening, suggested_questions = None, None @@ -948,18 +930,28 @@ class AgentRunService: and any(f.type == FileType.DOCUMENT for f in files) ) if has_doc_with_images: - agent.system_prompt += ( - "\n\n文档中包含图片，图片位置已在文本中以 [图片 第N页 第M张图片]: URL 标记。" - "请在回答中用 Markdown 格式 ![描述](URL) 展示相关图片，做到图文并茂。" - "**规则1：图片URL必须原封不动、一字不差地复制，禁止修改、禁止省略任何字符**" - "**规则2：禁止修改URL中UUID里的任何数字和字母**" - "**规则3：直接使用 ![描述](完整URL) 格式输出**" - ) - agent.agent = create_agent( - model=agent.llm, - tools=agent._wrap_tools_with_tracking(agent.tools) if agent.tools else None, - system_prompt=agent.system_prompt + system_prompt += ( + "\n\n文档文字中包含图片位置标记如 [图片 第2页 第1张]: http://...，请在回答中用 Markdown 格式 ![图片描述](url) 展示对应图片。" ) + + # 创建 LangChain Agent + agent = LangChainAgent( + model_name=api_key_config["model_name"], + api_key=api_key_config["api_key"], + provider=api_key_config.get("provider", "openai"), + api_base=api_key_config.get("api_base"), + is_omni=api_key_config.get("is_omni", False), + temperature=effective_params.get("temperature", 0.7), + max_tokens=effective_params.get("max_tokens", 2000), + system_prompt=system_prompt, + tools=tools, + streaming=True, + deep_thinking=effective_params.get("deep_thinking", False), + thinking_budget_tokens=effective_params.get("thinking_budget_tokens"), + json_output=effective_params.get("json_output", False), + capability=api_key_config.get("capability", []), + ) + # 为需要运行时上下文的工具注入上下文 for t in tools: if hasattr(t, 'tool_instance') and hasattr(t.tool_instance, 'set_runtime_context'): From 4b0afe867a0d490d3b7ff5c8af37340f74c60e52 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Fri, 24 Apr 2026 18:00:44 +0800 Subject: [PATCH 228/245] fix(app_chat_service,draft_run_service): move system_prompt augmentation before LangChainAgent instantiation --- api/app/services/app_chat_service.py | 46 ++++++++++++--------------- api/app/services/draft_run_service.py | 45 +++++++++++--------------- 2 files changed, 39 insertions(+), 52 deletions(-) diff --git a/api/app/services/app_chat_service.py b/api/app/services/app_chat_service.py index 39a3d4d4..7c52c2d6 100644 --- a/api/app/services/app_chat_service.py +++ b/api/app/services/app_chat_service.py @@ -107,23 +107,6 @@ class AppChatService: # 获取模型参数 model_parameters = config.model_parameters - # 创建 LangChain Agent - agent = LangChainAgent( - model_name=api_key_obj.model_name, - api_key=api_key_obj.api_key, - provider=api_key_obj.provider, - api_base=api_key_obj.api_base, - is_omni=api_key_obj.is_omni, - temperature=model_parameters.get("temperature", 0.7), - max_tokens=model_parameters.get("max_tokens", 2000), - system_prompt=system_prompt, - tools=tools, - deep_thinking=model_parameters.get("deep_thinking", False), - thinking_budget_tokens=model_parameters.get("thinking_budget_tokens"), - json_output=model_parameters.get("json_output", False), - capability=api_key_obj.capability or [], - ) - model_info = ModelInfo( model_name=api_key_obj.model_name, provider=api_key_obj.provider, @@ -177,16 +160,27 @@ class AppChatService: if doc_img_recognition and "vision" in (api_key_obj.capability or []) and any( f.type == FileType.DOCUMENT for f in files ): - from langchain.agents import create_agent - agent.system_prompt += ( - "\n\n文档中包含图片，图片位置已在文本中以 [第N页 第M张图片]: URL 标记。" - "请在回答中用 Markdown 格式 ![描述](URL) 展示相关图片，做到图文并茂。" - ) - agent.agent = create_agent( - model=agent.llm, - tools=agent._wrap_tools_with_tracking(agent.tools) if agent.tools else None, - system_prompt=agent.system_prompt + system_prompt += ( + "\n\n文档文字中包含图片位置标记如 [图片 第2页 第1张]: http://...，请在回答中用 Markdown 格式 ![图片描述](url) 展示对应图片。" ) + + # 创建 LangChain Agent + agent = LangChainAgent( + model_name=api_key_obj.model_name, + api_key=api_key_obj.api_key, + provider=api_key_obj.provider, + api_base=api_key_obj.api_base, + is_omni=api_key_obj.is_omni, + temperature=model_parameters.get("temperature", 0.7), + max_tokens=model_parameters.get("max_tokens", 2000), + system_prompt=system_prompt, + tools=tools, + deep_thinking=model_parameters.get("deep_thinking", False), + thinking_budget_tokens=model_parameters.get("thinking_budget_tokens"), + json_output=model_parameters.get("json_output", False), + capability=api_key_obj.capability or [], + ) + # 为需要运行时上下文的工具注入上下文 for t in tools: if hasattr(t, 'tool_instance') and hasattr(t.tool_instance, 'set_runtime_context'): diff --git a/api/app/services/draft_run_service.py b/api/app/services/draft_run_service.py index d3cb12db..10d50ece 100644 --- a/api/app/services/draft_run_service.py +++ b/api/app/services/draft_run_service.py @@ -595,23 +595,6 @@ class AgentRunService: ) tools.extend(memory_tools) - # 4. 创建 LangChain Agent - agent = LangChainAgent( - model_name=api_key_config["model_name"], - api_key=api_key_config["api_key"], - provider=api_key_config.get("provider", "openai"), - api_base=api_key_config.get("api_base"), - is_omni=api_key_config.get("is_omni", False), - temperature=effective_params.get("temperature", 0.7), - max_tokens=effective_params.get("max_tokens", 2000), - system_prompt=system_prompt, - tools=tools, - deep_thinking=effective_params.get("deep_thinking", False), - thinking_budget_tokens=effective_params.get("thinking_budget_tokens"), - json_output=effective_params.get("json_output", False), - capability=api_key_config.get("capability", []), - ) - # 5. 处理会话ID（创建或验证），新会话时写入开场白 is_new_conversation = not conversation_id opening, suggested_questions = None, None @@ -666,16 +649,26 @@ class AgentRunService: and any(f.type == FileType.DOCUMENT for f in files) ) if has_doc_with_images: - agent.system_prompt += ( - "\n\n文档中包含图片，图片位置已在文本中以 [第N页 第M张图片]: URL 标记。" - "请在回答中用 Markdown 格式 ![描述](URL) 展示相关图片，做到图文并茂。" - ) - # 重建 agent graph 以使新 system_prompt 生效 - agent.agent = create_agent( - model=agent.llm, - tools=agent._wrap_tools_with_tracking(agent.tools) if agent.tools else None, - system_prompt=agent.system_prompt + system_prompt += ( + "\n\n文档文字中包含图片位置标记如 [图片 第2页 第1张]: http://...，请在回答中用 Markdown 格式 ![图片描述](url) 展示对应图片。" ) + + agent = LangChainAgent( + model_name=api_key_config["model_name"], + api_key=api_key_config["api_key"], + provider=api_key_config.get("provider", "openai"), + api_base=api_key_config.get("api_base"), + is_omni=api_key_config.get("is_omni", False), + temperature=effective_params.get("temperature", 0.7), + max_tokens=effective_params.get("max_tokens", 2000), + system_prompt=system_prompt, + tools=tools, + deep_thinking=effective_params.get("deep_thinking", False), + thinking_budget_tokens=effective_params.get("thinking_budget_tokens"), + json_output=effective_params.get("json_output", False), + capability=api_key_config.get("capability", []), + ) + # 为需要运行时上下文的工具注入上下文 for t in tools: if hasattr(t, 'tool_instance') and hasattr(t.tool_instance, 'set_runtime_context'): From 32dfee803a8f69b6bfa8f39ede9e725272754195 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Fri, 24 Apr 2026 18:05:01 +0800 Subject: [PATCH 229/245] feat(web): workflow app logs --- web/src/views/ApplicationConfig/Logs.tsx | 18 +++++++- .../components/LogDetailModal.tsx | 38 ++++++++++++++-- .../views/Workflow/components/Chat/Chat.tsx | 6 +-- .../Workflow/components/Chat/Runtime.tsx | 45 ++++++++++--------- .../views/Workflow/hooks/useWorkflowGraph.ts | 4 +- 5 files changed, 80 insertions(+), 31 deletions(-) diff --git a/web/src/views/ApplicationConfig/Logs.tsx b/web/src/views/ApplicationConfig/Logs.tsx index cf56059c..75a5bdec 100644 --- a/web/src/views/ApplicationConfig/Logs.tsx +++ b/web/src/views/ApplicationConfig/Logs.tsx @@ -7,7 +7,7 @@ import { type FC, useRef } from 'react'; import { useTranslation } from 'react-i18next'; import { useParams } from 'react-router-dom'; -import { Flex, Button } from 'antd'; +import { Flex, Button, Form } from 'antd'; import type { ColumnsType } from 'antd/es/table'; import { getAppLogsUrl } from '@/api/application'; @@ -15,11 +15,14 @@ import Table from '@/components/Table' import { formatDateTime } from '@/utils/format'; import type { LogItem, LogDetailModalRef } from './types' import LogDetailModal from './components/LogDetailModal' +import SearchInput from '@/components/SearchInput' const Statistics: FC = () => { const { t } = useTranslation(); const { id } = useParams(); const logDetailRef = useRef(null); + const [form] = Form.useForm(); + const values = Form.useWatch([], form); const handleViewDetail = (item: LogItem) => { logDetailRef.current?.handleOpen(item); @@ -62,15 +65,26 @@ const Statistics: FC = () => { ]; return (
+ + + + + + + apiUrl={getAppLogsUrl(id || '')} apiParams={{ is_draft: false, + ...(values ?? {}) }} columns={columns} rowKey="id" isScroll={true} - scrollY="calc(100vh - 214px)" + scrollY="calc(100vh - 242px)" />
diff --git a/web/src/views/ApplicationConfig/components/LogDetailModal.tsx b/web/src/views/ApplicationConfig/components/LogDetailModal.tsx index 26d8741b..b37c3ae2 100644 --- a/web/src/views/ApplicationConfig/components/LogDetailModal.tsx +++ b/web/src/views/ApplicationConfig/components/LogDetailModal.tsx @@ -1,8 +1,8 @@ /* * @Author: ZhaoYing * @Date: 2026-03-24 16:31:24 - * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-24 16:31:24 + * @Last Modified by: ZhaoYing + * @Last Modified time: 2026-04-24 17:49:58 */ import { forwardRef, useImperativeHandle, useState, useEffect } from 'react'; import { Flex, Button, Empty, Skeleton } from 'antd'; @@ -14,6 +14,12 @@ import { getAppLogDetail } from '@/api/application' import ChatContent from '@/components/Chat/ChatContent' import { formatDateTime } from '@/utils/format' import type { ChatItem } from '@/components/Chat/types' +import Runtime from '@/views/Workflow/components/Chat/Runtime' +import { nodeLibrary } from '@/views/Workflow/constant' + +const nodeIconMap = Object.fromEntries( + nodeLibrary.flatMap(c => c.nodes.map(n => [n.type, n.icon])) +) /** Log detail data with conversation messages */ type Data = LogItem & { @@ -54,7 +60,30 @@ const LogDetailModal = forwardRef((_props, ref) => { if (!vo) return setLoading(true) getAppLogDetail(vo.app_id, vo.id).then(res => { - setData(res as Data) + const { node_executions_map, messages, ...rest } = res as Data; + let hasSubContentMessages = messages + if (messages && messages.length > 0 && node_executions_map && Object.keys(node_executions_map).length > 0) { + hasSubContentMessages = messages.map(item => { + if (item.id && node_executions_map[item.id]) { + item.subContent = node_executions_map[item.id]?.map(({ input, output, cycle_items = [], error, process, ...node }: any) => { + const converted: any = { ...node, icon: nodeIconMap[node.node_type], content: { input, output, process, error } } + if (node.node_type === 'loop' && Array.isArray(cycle_items) && cycle_items.length > 0) { + converted.subContent = cycle_items.map(({ input: cInput, output: cOutput, error: cError, process: cProcess, ...cNode }: any) => ({ + ...cNode, + icon: nodeIconMap[cNode.node_type], + content: { input: cInput, output: cOutput, process: cProcess, error: cError } + })) + } + return converted + }) + } + return { ...item } + }) + } + setData({ + ...rest, + messages: hasSubContentMessages + }) }) .finally(() => { setLoading(false) @@ -66,6 +95,8 @@ const LogDetailModal = forwardRef((_props, ref) => { handleClose })); + console.log('data', data) + return ( @@ -92,6 +123,7 @@ const LogDetailModal = forwardRef((_props, ref) => { data={data.messages || []} streamLoading={false} labelFormat={(item) => formatDateTime(item.created_at)} + renderRuntime={(item, index) => } /> ) } diff --git a/web/src/views/Workflow/components/Chat/Chat.tsx b/web/src/views/Workflow/components/Chat/Chat.tsx index a6b4a2a8..820fe8c7 100644 --- a/web/src/views/Workflow/components/Chat/Chat.tsx +++ b/web/src/views/Workflow/components/Chat/Chat.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-06 21:10:56 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-21 14:59:13 + * @Last Modified time: 2026-04-24 17:34:51 */ /** * Workflow Chat Component @@ -66,7 +66,7 @@ const Chat = forwardRef([]) const [message, setMessage] = useState(undefined) - console.log('abortRef', abortRef) + console.log('abortRef', abortRef, chatList) /** * Opens the chat drawer and loads workflow variables from the start node @@ -305,7 +305,7 @@ const Chat = forwardRef = ({ )} {/* Display input and output data as JSON code blocks */} - {['input', 'output'].map(key => ( -
-
- {isLoop ? t(`workflow.runtime.${key}_cycle_vars`) : t(`workflow.${key}_result`)} - + {['input', 'process', 'output'].map(key => { + if (vo.node_type !== 'http-request' && key === 'process') return null + return ( +
+
+ {isLoop ? t(`workflow.runtime.${key}_cycle_vars`) : t(`workflow.${key}_result`)} + +
+
+ +
-
- -
-
- ))} + ) + })} ) }]} diff --git a/web/src/views/Workflow/hooks/useWorkflowGraph.ts b/web/src/views/Workflow/hooks/useWorkflowGraph.ts index f30db10f..a22ee6c0 100644 --- a/web/src/views/Workflow/hooks/useWorkflowGraph.ts +++ b/web/src/views/Workflow/hooks/useWorkflowGraph.ts @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 15:17:48 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-20 16:00:26 + * @Last Modified time: 2026-04-24 17:21:09 */ import { Clipboard, Graph, Keyboard, MiniMap, Node, Snapline, History, type Edge } from '@antv/x6'; import type { HistoryCommand as Command } from '@antv/x6/lib/plugin/history/type'; @@ -1492,7 +1492,7 @@ export const useWorkflowGraph = ({ // Reset all node execution status first nodes.forEach(node => { const data = node.getData(); - if (typeof data.status === 'string') { + if (typeof data.executionStatus === 'string') { node.setData({ ...data, executionStatus: undefined }); } }); From 2c864f63375fac9a4a7d9dfa5d74d695ac368970 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Fri, 24 Apr 2026 18:15:01 +0800 Subject: [PATCH 230/245] feat(web): http request add process --- web/src/views/ApplicationConfig/TestChat/index.tsx | 9 ++++++--- web/src/views/Workflow/components/Chat/Chat.tsx | 7 +++++-- 2 files changed, 11 insertions(+), 5 deletions(-) diff --git a/web/src/views/ApplicationConfig/TestChat/index.tsx b/web/src/views/ApplicationConfig/TestChat/index.tsx index b62efc6b..2fc66aa6 100644 --- a/web/src/views/ApplicationConfig/TestChat/index.tsx +++ b/web/src/views/ApplicationConfig/TestChat/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-03-13 17:27:52 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-07 21:48:30 + * @Last Modified time: 2026-04-24 18:14:25 */ import { type FC, useState, useRef, useEffect } from 'react' import { useTranslation } from 'react-i18next' @@ -59,6 +59,7 @@ interface NodeData { node_type?: string; input?: any; output?: any; + process?: any; elapsed_time?: string; error?: any; state: Record; @@ -485,7 +486,7 @@ const TestChat: FC = ({ } const updateWorkflowNodeEndMessage = (data: NodeData) => { - const { node_id, input, output, error, elapsed_time, status } = data; + const { node_id, input, output, process, error, elapsed_time, status } = data; setChatList(prev => { const newList = [...prev] const lastIndex = newList.length - 1 @@ -498,6 +499,7 @@ const TestChat: FC = ({ content: { input, output, + process, error, }, status: status || 'completed', @@ -514,7 +516,7 @@ const TestChat: FC = ({ } const updateWorkflowCycleMessage = (data: NodeData) => { - const { node_id, cycle_id, cycle_idx, input, output, error, elapsed_time, status } = data; + const { node_id, cycle_id, cycle_idx, input, output, process, error, elapsed_time, status } = data; const { nodes } = config as WorkflowConfig const node = nodes.find(n => n.id === node_id); const { name, type } = node || {} @@ -538,6 +540,7 @@ const TestChat: FC = ({ cycle_idx, input, output, + process, error, }, status: status || 'completed', diff --git a/web/src/views/Workflow/components/Chat/Chat.tsx b/web/src/views/Workflow/components/Chat/Chat.tsx index 820fe8c7..863825ba 100644 --- a/web/src/views/Workflow/components/Chat/Chat.tsx +++ b/web/src/views/Workflow/components/Chat/Chat.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-06 21:10:56 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-24 17:34:51 + * @Last Modified time: 2026-04-24 18:13:22 */ /** * Workflow Chat Component @@ -185,7 +185,7 @@ const Chat = forwardRef { data.forEach(item => { - const { content, conversation_id, node_id, cycle_id, cycle_idx, input, output, error, elapsed_time, status, citations } = item.data as { + const { content, conversation_id, node_id, cycle_id, cycle_idx, input, output, process, error, elapsed_time, status, citations } = item.data as { content: string; conversation_id: string | null; cycle_id: string; @@ -193,6 +193,7 @@ const Chat = forwardRef Date: Fri, 24 Apr 2026 18:20:14 +0800 Subject: [PATCH 231/245] refactor(workflow): streamline node execution handling and log service logic - Consolidate node data retrieval from workflow_executions.output_data to unify storage access. - Optimize the construction of messages and execution records to support opening suggestions. - Eliminate redundant queries and storage logic to simplify the overall codebase structure. --- .../workflow/nodes/cycle_graph/iteration.py | 1 + .../core/workflow/nodes/cycle_graph/loop.py | 1 + api/app/services/app_log_service.py | 157 ++++++++++++------ api/app/services/workflow_service.py | 34 +--- 4 files changed, 121 insertions(+), 72 deletions(-) diff --git a/api/app/core/workflow/nodes/cycle_graph/iteration.py b/api/app/core/workflow/nodes/cycle_graph/iteration.py index 1ac46180..3ee7774f 100644 --- a/api/app/core/workflow/nodes/cycle_graph/iteration.py +++ b/api/app/core/workflow/nodes/cycle_graph/iteration.py @@ -182,6 +182,7 @@ class IterationRuntime: "node_id": node_name, "node_type": node_type, "node_name": node_cfg.get("data", {}).get("label") if node_cfg else node_name, + "status": result.get("node_outputs", {}).get(node_name, {}).get("status", "completed"), "input": result.get("node_outputs", {}).get(node_name, {}).get("input") if not cycle_variable else cycle_variable, "output": result.get("node_outputs", {}).get(node_name, {}).get("output") diff --git a/api/app/core/workflow/nodes/cycle_graph/loop.py b/api/app/core/workflow/nodes/cycle_graph/loop.py index ca432a04..93f1a1e4 100644 --- a/api/app/core/workflow/nodes/cycle_graph/loop.py +++ b/api/app/core/workflow/nodes/cycle_graph/loop.py @@ -212,6 +212,7 @@ class LoopRuntime: "node_id": node_name, "node_type": node_type, "node_name": node_name, + "status": result.get("node_outputs", {}).get(node_name, {}).get("status", "completed"), "input": result.get("node_outputs", {}).get(node_name, {}).get("input") if not cycle_variable else cycle_variable, "output": result.get("node_outputs", {}).get(node_name, {}).get("output") diff --git a/api/app/services/app_log_service.py b/api/app/services/app_log_service.py index e8a80d40..7ca05d42 100644 --- a/api/app/services/app_log_service.py +++ b/api/app/services/app_log_service.py @@ -115,7 +115,7 @@ class AppLogService: messages = self.message_repository.get_messages_by_conversation( conversation_id=conversation_id ) - _, node_executions_map = self._get_workflow_node_executions_with_map( + node_executions_map = self._get_workflow_node_executions_with_map( conversation_id, messages ) @@ -139,9 +139,9 @@ class AppLogService: 工作流应用专用：从 workflow_executions 构建 messages 和节点日志。 每条 WorkflowExecution 对应一轮对话： - - user message：来自 execution.input_data + - user message：来自 execution.input_data（content 取 message 字段，files 放 meta_data） - assistant message：来自 execution.output_data（失败时内容为错误信息） - 节点日志以 execution id 为 key 分组。 + 开场白的 suggested_questions 合并到第一条 assistant message 的 meta_data 里。 Returns: (messages 列表, node_executions_map) @@ -156,9 +156,44 @@ class AppLogService: ) executions = list(self.db.scalars(stmt).all()) + # 查开场白：Message 表里 meta_data 含 suggested_questions 的第一条 assistant 消息 + opening_stmt = ( + select(Message) + .where( + Message.conversation_id == conversation_id, + Message.role == "assistant", + ) + .order_by(Message.created_at.asc()) + .limit(10) + ) + early_messages = list(self.db.scalars(opening_stmt).all()) + suggested_questions: list = [] + for m in early_messages: + if isinstance(m.meta_data, dict) and "suggested_questions" in m.meta_data: + suggested_questions = m.meta_data.get("suggested_questions") or [] + break + messages: list[AppLogMessage] = [] node_executions_map: dict[str, list[AppLogNodeExecution]] = {} + # 如果有开场白，作为第一条 assistant 消息插入 + if suggested_questions or early_messages: + opening_msg = next( + (m for m in early_messages + if isinstance(m.meta_data, dict) and "suggested_questions" in m.meta_data), + None + ) + if opening_msg: + messages.append(AppLogMessage( + id=opening_msg.id, + conversation_id=conversation_id, + role="assistant", + content=opening_msg.content, + status=None, + meta_data={"suggested_questions": suggested_questions}, + created_at=opening_msg.created_at, + )) + for execution in executions: started_at = execution.started_at or dt.datetime.now() completed_at = execution.completed_at or started_at @@ -167,13 +202,20 @@ class AppLogService: assistant_msg_id = uuid.uuid5(execution.id, "assistant") # --- user message（输入）--- - input_content = _extract_text(execution.input_data) + input_data = execution.input_data or {} + input_content = input_data.get("message") or _extract_text(input_data) + + # 跳过没有用户输入的 execution（如开场白触发的记录） + if not input_content or not input_content.strip(): + continue + + files = input_data.get("files") or [] user_msg = AppLogMessage( id=uuid.uuid5(execution.id, "user"), conversation_id=conversation_id, role="user", content=input_content, - meta_data=None, + meta_data={"files": files} if files else None, created_at=started_at, ) messages.append(user_msg) @@ -197,24 +239,8 @@ class AppLogService: ) messages.append(assistant_msg) - # --- 节点执行记录，key 与 assistant message id 一致 --- - execution_nodes = [] - for node_exec in execution.node_executions: - output_data = dict(node_exec.output_data or {}) - cycle_items = output_data.pop("cycle_items", None) - execution_nodes.append(AppLogNodeExecution( - node_id=node_exec.node_id, - node_type=node_exec.node_type, - node_name=node_exec.node_name, - status=node_exec.status, - error=node_exec.error_message, - input=node_exec.input_data, - process=None, - output=output_data, - cycle_items=cycle_items, - elapsed_time=node_exec.elapsed_time, - token_usage=node_exec.token_usage, - )) + # --- 节点执行记录，从 workflow_executions.output_data["node_outputs"] 读取 --- + execution_nodes = _build_nodes_from_output_data(execution.output_data) if execution_nodes: node_executions_map[str(assistant_msg_id)] = execution_nodes @@ -225,7 +251,7 @@ class AppLogService: self, conversation_id: uuid.UUID, messages: list[Message] - ) -> Tuple[list[AppLogNodeExecution], dict[str, list[AppLogNodeExecution]]]: + ) -> dict[str, list[AppLogNodeExecution]]: """ 从 workflow_executions 表中提取节点执行记录，并按 assistant message 分组 @@ -237,7 +263,6 @@ class AppLogService: Tuple[list[AppLogNodeExecution], dict[str, list[AppLogNodeExecution]]]: (所有节点执行记录列表, 按 message_id 分组的节点执行记录字典) """ - node_executions = [] node_executions_map: dict[str, list[AppLogNodeExecution]] = {} # 查询该会话关联的所有工作流执行记录（按时间正序） @@ -268,26 +293,8 @@ class AppLogService: used_message_ids: set[str] = set() for execution in executions: - # 构建节点执行记录列表 - execution_nodes = [] - for node_exec in execution.node_executions: - output_data = dict(node_exec.output_data or {}) - cycle_items = output_data.pop("cycle_items", None) - node_execution = AppLogNodeExecution( - node_id=node_exec.node_id, - node_type=node_exec.node_type, - node_name=node_exec.node_name, - status=node_exec.status, - error=node_exec.error_message, - input=node_exec.input_data, - process=None, - output=output_data, - cycle_items=cycle_items, - elapsed_time=node_exec.elapsed_time, - token_usage=node_exec.token_usage, - ) - node_executions.append(node_execution) - execution_nodes.append(node_execution) + # 构建节点执行记录列表，从 workflow_executions.output_data["node_outputs"] 读取 + execution_nodes = _build_nodes_from_output_data(execution.output_data) if not execution_nodes: continue @@ -318,7 +325,7 @@ class AppLogService: used_message_ids.add(msg_id_str) node_executions_map[msg_id_str] = execution_nodes - return node_executions, node_executions_map + return node_executions_map def _extract_text(data: Optional[dict]) -> str: @@ -328,8 +335,64 @@ def _extract_text(data: Optional[dict]) -> str: """ if not data: return "" - for key in ("text", "content", "output", "result", "answer"): + for key in ("message", "text", "content", "output", "result", "answer"): if key in data and isinstance(data[key], str): return data[key] import json return json.dumps(data, ensure_ascii=False) + + +def _build_nodes_from_output_data(output_data: Optional[dict]) -> list[AppLogNodeExecution]: + """从 workflow_executions.output_data["node_outputs"] 构建节点执行记录列表。 + + output_data 结构： + { + "node_outputs": { + "": { + "node_type": ..., + "node_name": ..., + "status": ..., + "input": ..., + "output": ..., + "elapsed_time": ..., + "token_usage": ..., + "error": ..., + "cycle_items": [...], + ... + } + }, + "error": ..., + ... + } + """ + if not output_data: + return [] + node_outputs: dict = output_data.get("node_outputs") or {} + result = [] + for node_id, node_data in node_outputs.items(): + if not isinstance(node_data, dict): + continue + output = dict(node_data) + cycle_items = output.pop("cycle_items", None) + # 把已知的顶层字段剥离，剩余的作为 output + node_type = output.pop("node_type", "unknown") + node_name = output.pop("node_name", None) + status = output.pop("status", "completed") + error = output.pop("error", None) + inp = output.pop("input", None) + elapsed_time = output.pop("elapsed_time", None) + token_usage = output.pop("token_usage", None) + result.append(AppLogNodeExecution( + node_id=node_id, + node_type=node_type, + node_name=node_name, + status=status, + error=error, + input=inp, + process=None, + output=output if output else None, + cycle_items=cycle_items, + elapsed_time=elapsed_time, + token_usage=token_usage, + )) + return result diff --git a/api/app/services/workflow_service.py b/api/app/services/workflow_service.py index ea302a9c..aa91de89 100644 --- a/api/app/services/workflow_service.py +++ b/api/app/services/workflow_service.py @@ -910,7 +910,6 @@ class WorkflowService: input_data["conv_messages"] = conv_messages init_message_length = len(input_data.get("conv_messages", [])) message_id = uuid.uuid4() - _node_order_counter = 0 _cycle_items: dict[str, list] = {} # 新会话时写入开场白 @@ -1015,32 +1014,17 @@ class WorkflowService: ) else: logger.error(f"unexpect workflow run status, status: {status}") - # 把积累的 cycle_item 写入对应循环节点的 output_data - if _cycle_items: + # 把积累的 cycle_item 写入 workflow_executions.output_data["node_outputs"] + if _cycle_items and execution.output_data: + import copy + new_output_data = copy.deepcopy(execution.output_data) + node_outputs = new_output_data.setdefault("node_outputs", {}) for cycle_node_id, items in _cycle_items.items(): - node_exec = self.db.execute( - select(WorkflowNodeExecution).where( - WorkflowNodeExecution.execution_id == execution.id, - WorkflowNodeExecution.node_id == cycle_node_id - ) - ).scalar_one_or_none() - if node_exec: - node_exec.output_data = { - **(node_exec.output_data or {}), - "cycle_items": items - } + if cycle_node_id in node_outputs: + node_outputs[cycle_node_id]["cycle_items"] = items else: - node_cfg = next((n for n in config.nodes if n.get("id") == cycle_node_id), {}) - self.db.add(WorkflowNodeExecution( - execution_id=execution.id, - node_id=cycle_node_id, - node_type=node_cfg.get("type", "cycle"), - node_name=node_cfg.get("data", {}).get("label") or cycle_node_id, - execution_order=_node_order_counter, - status="completed", - output_data={"cycle_items": items}, - )) - _node_order_counter += 1 + node_outputs[cycle_node_id] = {"cycle_items": items} + execution.output_data = new_output_data self.db.commit() elif event.get("event") == "workflow_start": event["data"]["message_id"] = str(message_id) From caef0fe44e37c2c9b8345518248615e827fa0ba9 Mon Sep 17 00:00:00 2001 From: Eternity <1533512157@qq.com> Date: Fri, 24 Apr 2026 18:36:27 +0800 Subject: [PATCH 232/245] fix(api): correct import paths in memory_read and celery task command - Fix relative imports in memory_read.py to use absolute app paths - Change celery scheduler command from `python app/celery_task_scheduler.py` to `python -m app.celery_task_scheduler` --- api/app/core/memory/pipelines/memory_read.py | 4 ++-- api/docker-compose.yml | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/api/app/core/memory/pipelines/memory_read.py b/api/app/core/memory/pipelines/memory_read.py index 35ace00d..0bd57b08 100644 --- a/api/app/core/memory/pipelines/memory_read.py +++ b/api/app/core/memory/pipelines/memory_read.py @@ -1,8 +1,8 @@ from app.core.memory.enums import SearchStrategy, StorageType from app.core.memory.models.service_models import MemorySearchResult from app.core.memory.pipelines.base_pipeline import ModelClientMixin, DBRequiredPipeline -from core.memory.read_services.search_engine.content_search import Neo4jSearchService, RAGSearchService -from core.memory.read_services.generate_engine.query_preprocessor import QueryPreprocessor +from app.core.memory.read_services.search_engine.content_search import Neo4jSearchService, RAGSearchService +from app.core.memory.read_services.generate_engine.query_preprocessor import QueryPreprocessor class ReadPipeLine(ModelClientMixin, DBRequiredPipeline): diff --git a/api/docker-compose.yml b/api/docker-compose.yml index a0fd4791..a3937add 100644 --- a/api/docker-compose.yml +++ b/api/docker-compose.yml @@ -70,7 +70,7 @@ services: - .env volumes: - /etc/localtime:/etc/localtime:ro - command: python app/celery_task_scheduler.py + command: python -m app.celery_task_scheduler restart: unless-stopped healthcheck: test: CMD curl -f 127.0.0.1:8001 || exit 1 From f63bcd632105ca47eb9a23b432ce29303fc8fdf1 Mon Sep 17 00:00:00 2001 From: wxy Date: Fri, 24 Apr 2026 18:49:55 +0800 Subject: [PATCH 233/245] refactor(tool): flatten request body parameters for model exposure - Refactor the extraction logic in tool service to flatten request body parameters into independent arguments exposed to the model. --- api/app/core/tools/custom/base.py | 18 ++++++++++++++++++ api/app/services/tool_service.py | 7 ++++--- 2 files changed, 22 insertions(+), 3 deletions(-) diff --git a/api/app/core/tools/custom/base.py b/api/app/core/tools/custom/base.py index c03fe206..06237d32 100644 --- a/api/app/core/tools/custom/base.py +++ b/api/app/core/tools/custom/base.py @@ -73,6 +73,7 @@ class CustomTool(BaseTool): # 添加通用参数（基于第一个操作的参数） if self._parsed_operations: first_operation = next(iter(self._parsed_operations.values())) + # path/query 参数 for param_name, param_info in first_operation.get("parameters", {}).items(): params.append(ToolParameter( name=param_name, @@ -85,6 +86,23 @@ class CustomTool(BaseTool): maximum=param_info.get("maximum"), pattern=param_info.get("pattern") )) + # requestBody 参数 — 将 body 字段平铺为独立参数暴露给模型 + request_body = first_operation.get("request_body") + if request_body: + body_schema = request_body.get("properties", {}) + required_fields = request_body.get("required", []) + for prop_name, prop_schema in body_schema.items(): + params.append(ToolParameter( + name=prop_name, + type=self._convert_openapi_type(prop_schema.get("type", "string")), + description=prop_schema.get("description", ""), + required=prop_name in required_fields, + default=prop_schema.get("default"), + enum=prop_schema.get("enum"), + minimum=prop_schema.get("minimum"), + maximum=prop_schema.get("maximum"), + pattern=prop_schema.get("pattern") + )) return params diff --git a/api/app/services/tool_service.py b/api/app/services/tool_service.py index 9a59cd81..ff734c9d 100644 --- a/api/app/services/tool_service.py +++ b/api/app/services/tool_service.py @@ -815,11 +815,12 @@ class ToolService: "default": param_info.get("default") }) - # 请求体参数 + # 请求体参数 — _extract_request_body 返回 {"schema": {...}, "required": bool, ...} request_body = operation.get("request_body") if request_body: - schema_props = request_body.get("schema", {}).get("properties", {}) - required_props = request_body.get("schema", {}).get("required", []) + body_schema = request_body.get("schema", {}) + schema_props = body_schema.get("properties", {}) + required_props = body_schema.get("required", []) for prop_name, prop_schema in schema_props.items(): parameters.append({ From 8dee2eae6ad66a9f61c162b7e9f09975e7d31d1e Mon Sep 17 00:00:00 2001 From: Eternity <1533512157@qq.com> Date: Fri, 24 Apr 2026 18:50:58 +0800 Subject: [PATCH 234/245] fix(api): correct import paths in memory_read and celery task command - Fix relative imports in memory_read.py to use absolute app paths - Change celery scheduler command from `python app/celery_task_scheduler.py` to `python -m app.celery_task_scheduler` --- .../core/memory/read_services/search_engine/content_search.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/api/app/core/memory/read_services/search_engine/content_search.py b/api/app/core/memory/read_services/search_engine/content_search.py index 16c23f91..4ba4dce7 100644 --- a/api/app/core/memory/read_services/search_engine/content_search.py +++ b/api/app/core/memory/read_services/search_engine/content_search.py @@ -8,7 +8,7 @@ from neo4j import Session from app.core.memory.enums import Neo4jNodeType from app.core.memory.memory_service import MemoryContext from app.core.memory.models.service_models import Memory, MemorySearchResult -from core.memory.read_services.search_engine.result_builder import data_builder_factory +from app.core.memory.read_services.search_engine.result_builder import data_builder_factory from app.core.models import RedBearEmbeddings from app.core.rag.nlp.search import knowledge_retrieval from app.repositories import knowledge_repository From 422af699048e524175f1833901f784618b8d08f3 Mon Sep 17 00:00:00 2001 From: Eternity <1533512157@qq.com> Date: Fri, 24 Apr 2026 19:09:18 +0800 Subject: [PATCH 235/245] fix(api): correct import paths in memory_read and celery task command - Fix relative imports in memory_read.py to use absolute app paths - Change celery scheduler command from `python app/celery_task_scheduler.py` to `python -m app.celery_task_scheduler` --- api/app/controllers/service/memory_api_controller.py | 2 +- .../core/memory/agent/langgraph_graph/routing/write_router.py | 2 +- api/app/core/workflow/nodes/memory/node.py | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-) diff --git a/api/app/controllers/service/memory_api_controller.py b/api/app/controllers/service/memory_api_controller.py index 57664e4e..43a8824a 100644 --- a/api/app/controllers/service/memory_api_controller.py +++ b/api/app/controllers/service/memory_api_controller.py @@ -3,6 +3,7 @@ from fastapi import APIRouter, Body, Depends, Query, Request from sqlalchemy.orm import Session +from app.celery_task_scheduler import scheduler from app.core.api_key_auth import require_api_key from app.core.logging_config import get_business_logger from app.core.quota_stub import check_end_user_quota @@ -18,7 +19,6 @@ from app.schemas.memory_api_schema import ( MemoryWriteSyncResponse, ) from app.services.memory_api_service import MemoryAPIService -from celery_task_scheduler import scheduler router = APIRouter(prefix="/memory", tags=["V1 - Memory API"]) logger = get_business_logger() diff --git a/api/app/core/memory/agent/langgraph_graph/routing/write_router.py b/api/app/core/memory/agent/langgraph_graph/routing/write_router.py index 7ef4ed12..d016f2e0 100644 --- a/api/app/core/memory/agent/langgraph_graph/routing/write_router.py +++ b/api/app/core/memory/agent/langgraph_graph/routing/write_router.py @@ -1,6 +1,7 @@ import json import os +from app.celery_task_scheduler import scheduler from app.core.logging_config import get_agent_logger from app.core.memory.agent.langgraph_graph.tools.write_tool import format_parsing, messages_parse from app.core.memory.agent.models.write_aggregate_model import WriteAggregateModel @@ -13,7 +14,6 @@ from app.db import get_db_context from app.repositories.memory_short_repository import LongTermMemoryRepository from app.schemas.memory_agent_schema import AgentMemory_Long_Term from app.utils.config_utils import resolve_config_id -from celery_task_scheduler import scheduler logger = get_agent_logger(__name__) template_root = os.path.join(PROJECT_ROOT_, 'memory', 'agent', 'utils', 'prompt') diff --git a/api/app/core/workflow/nodes/memory/node.py b/api/app/core/workflow/nodes/memory/node.py index b74af6b9..6d9fcdad 100644 --- a/api/app/core/workflow/nodes/memory/node.py +++ b/api/app/core/workflow/nodes/memory/node.py @@ -1,6 +1,7 @@ import re from typing import Any +from app.celery_task_scheduler import scheduler from app.core.memory.enums import SearchStrategy from app.core.memory.memory_service import MemoryService from app.core.workflow.engine.state_manager import WorkflowState @@ -11,7 +12,6 @@ from app.core.workflow.variable.base_variable import VariableType from app.core.workflow.variable.variable_objects import FileVariable, ArrayVariable from app.db import get_db_read from app.schemas import FileInput -from celery_task_scheduler import scheduler class MemoryReadNode(BaseNode): From 0fc463036ef92a03ab1b04275a11d5db1baa0693 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Fri, 24 Apr 2026 19:12:35 +0800 Subject: [PATCH 236/245] fix(web): node status ui --- web/src/views/Workflow/components/Nodes/ConditionNode.tsx | 4 ++-- web/src/views/Workflow/components/Nodes/LoopNode.tsx | 4 ++-- web/src/views/Workflow/components/Nodes/NormalNode.tsx | 4 ++-- 3 files changed, 6 insertions(+), 6 deletions(-) diff --git a/web/src/views/Workflow/components/Nodes/ConditionNode.tsx b/web/src/views/Workflow/components/Nodes/ConditionNode.tsx index 19966823..b431ddd4 100644 --- a/web/src/views/Workflow/components/Nodes/ConditionNode.tsx +++ b/web/src/views/Workflow/components/Nodes/ConditionNode.tsx @@ -65,8 +65,8 @@ const ConditionNode: ReactShapeConfig['component'] = ({ node }) => { return (
diff --git a/web/src/views/Workflow/components/Nodes/LoopNode.tsx b/web/src/views/Workflow/components/Nodes/LoopNode.tsx index c540db76..cffb62dd 100644 --- a/web/src/views/Workflow/components/Nodes/LoopNode.tsx +++ b/web/src/views/Workflow/components/Nodes/LoopNode.tsx @@ -131,8 +131,8 @@ const LoopNode: ReactShapeConfig['component'] = ({ node, graph }) => { return (
diff --git a/web/src/views/Workflow/components/Nodes/NormalNode.tsx b/web/src/views/Workflow/components/Nodes/NormalNode.tsx index ce936be9..cb3ef134 100644 --- a/web/src/views/Workflow/components/Nodes/NormalNode.tsx +++ b/web/src/views/Workflow/components/Nodes/NormalNode.tsx @@ -12,8 +12,8 @@ const NormalNode: ReactShapeConfig['component'] = ({ node }) => { return (
From 141fd9451367e7c08f2b922fb6a580e303c16be2 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Fri, 24 Apr 2026 19:40:57 +0800 Subject: [PATCH 237/245] fix(multimodal_service): refactor image processing to use intermediate list before extending result --- api/app/services/multimodal_service.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/api/app/services/multimodal_service.py b/api/app/services/multimodal_service.py index 960bf6bb..08a33a48 100644 --- a/api/app/services/multimodal_service.py +++ b/api/app/services/multimodal_service.py @@ -388,6 +388,7 @@ class MultimodalService: from app.models.workspace_model import Workspace as WorkspaceModel ws = self.db.query(WorkspaceModel).filter(WorkspaceModel.id == workspace_id).first() tenant_id = ws.tenant_id if ws else None + img_result = [] for img_info in img_infos: page = img_info["page"] index = img_info["index"] @@ -407,9 +408,10 @@ class MultimodalService: file_type="image/png", ) _, img_content = await self._process_image(img_file, strategy_class(img_file)) - result.append(img_content) + img_result.append(img_content) except Exception as img_err: logger.warning(f"文档图片处理失败: {img_err}") + result.extend(img_result) elif file.type == FileType.AUDIO and "audio" in self.capability: is_support, content = await self._process_audio(file, strategy) result.append(content) From b6e27da7b0d63d93a437e94c9813f666258ea914 Mon Sep 17 00:00:00 2001 From: Eternity <1533512157@qq.com> Date: Fri, 24 Apr 2026 19:56:55 +0800 Subject: [PATCH 238/245] fix(api): convert end_user_id to string in write_router --- .../memory/agent/langgraph_graph/routing/write_router.py | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/api/app/core/memory/agent/langgraph_graph/routing/write_router.py b/api/app/core/memory/agent/langgraph_graph/routing/write_router.py index d016f2e0..3a225319 100644 --- a/api/app/core/memory/agent/langgraph_graph/routing/write_router.py +++ b/api/app/core/memory/agent/langgraph_graph/routing/write_router.py @@ -94,9 +94,9 @@ async def write( # ) scheduler.push_task( "app.core.memory.agent.write_message", - actual_end_user_id, + str(actual_end_user_id), { - "end_user_id": actual_end_user_id, + "end_user_id": str(actual_end_user_id), "message": structured_messages, "config_id": str(actual_config_id), "storage_type": storage_type, @@ -177,9 +177,9 @@ async def window_dialogue(end_user_id, langchain_messages, memory_config, scope) scheduler.push_task( "app.core.memory.agent.write_message", - end_user_id, + str(end_user_id), { - "end_user_id": end_user_id, + "end_user_id": str(end_user_id), "message": redis_messages, "config_id": config_id, "storage_type": AgentMemory_Long_Term.STORAGE_NEO4J, From cd34d5f5ce084130785f8978260ad01bfd9fe7ff Mon Sep 17 00:00:00 2001 From: Eternity <1533512157@qq.com> Date: Fri, 24 Apr 2026 20:13:46 +0800 Subject: [PATCH 239/245] fix(api): convert config_id to string in write_router --- .../core/memory/agent/langgraph_graph/routing/write_router.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/api/app/core/memory/agent/langgraph_graph/routing/write_router.py b/api/app/core/memory/agent/langgraph_graph/routing/write_router.py index 3a225319..a896130f 100644 --- a/api/app/core/memory/agent/langgraph_graph/routing/write_router.py +++ b/api/app/core/memory/agent/langgraph_graph/routing/write_router.py @@ -181,7 +181,7 @@ async def window_dialogue(end_user_id, langchain_messages, memory_config, scope) { "end_user_id": str(end_user_id), "message": redis_messages, - "config_id": config_id, + "config_id": str(config_id), "storage_type": AgentMemory_Long_Term.STORAGE_NEO4J, "user_rag_memory_id": "" } From a268d0f7f187a4e05df973d0ead15a97ecef9e3f Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Mon, 27 Apr 2026 12:25:27 +0800 Subject: [PATCH 240/245] =?UTF-8?q?fix(multimodal=5Fservice):=20add=20'?= =?UTF-8?q?=E6=96=87=E6=A1=A3=E5=86=85=E5=AE=B9=EF=BC=9A'=20prefix=20to=20?= =?UTF-8?q?document=20text=20and=20simplify=20image=20placeholder=20text?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- api/app/services/multimodal_service.py | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/api/app/services/multimodal_service.py b/api/app/services/multimodal_service.py index 08a33a48..c362158c 100644 --- a/api/app/services/multimodal_service.py +++ b/api/app/services/multimodal_service.py @@ -95,7 +95,7 @@ class DashScopeFormatStrategy(MultimodalFormatStrategy): """通义千问文档格式""" return True, { "type": "text", - "text": f"\n{text}\n" + "text": f"\n文档内容：\n{text}\n" } async def format_audio( @@ -167,6 +167,7 @@ class BedrockFormatStrategy(MultimodalFormatStrategy): async def format_document(self, file_name: str, text: str) -> tuple[bool, Dict[str, Any]]: """Bedrock/Anthropic 文档格式（需要 base64 编码）""" # Bedrock 文档需要 base64 编码 + text = f"文档内容：\n{text}\n" text_bytes = text.encode('utf-8') base64_text = base64.b64encode(text_bytes).decode('utf-8') @@ -223,7 +224,7 @@ class OpenAIFormatStrategy(MultimodalFormatStrategy): """OpenAI 文档格式""" return True, { "type": "text", - "text": f"\n{text}\n" + "text": f"\n文档内容：\n{text}\n" } async def format_audio( @@ -395,7 +396,7 @@ class MultimodalService: ext = img_info.get("ext", "png") try: _, img_url = await self._save_doc_image_to_storage(img_info["bytes"], ext, tenant_id, workspace_id) - placeholder = f"第{page}页 第{index + 1}张图片" if page > 0 else f"第{index + 1}张图片" + placeholder = f"第{page}页 第{index + 1}张" if page > 0 else f"第{index + 1}张" # 在文本内容中追加图片位置标记 if result and result[-1].get("type") in ("text", "document"): key = "text" if "text" in result[-1] else list(result[-1].keys())[-1] From 546bfb96276258eff37466337d81ac94d4135319 Mon Sep 17 00:00:00 2001 From: wxy Date: Mon, 27 Apr 2026 14:05:06 +0800 Subject: [PATCH 241/245] fix(api_key): bypass publication check for SERVICE type API keys - Exclude SERVICE type keys from application publication validation since their resource_id targets the workspace instead of an application. --- api/app/services/api_key_service.py | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/api/app/services/api_key_service.py b/api/app/services/api_key_service.py index e4367e98..9044af37 100644 --- a/api/app/services/api_key_service.py +++ b/api/app/services/api_key_service.py @@ -9,7 +9,7 @@ from sqlalchemy.orm import Session from sqlalchemy import select from app.aioRedis import aio_redis -from app.models.api_key_model import ApiKey +from app.models.api_key_model import ApiKey, ApiKeyType from app.repositories.api_key_repository import ApiKeyRepository, ApiKeyLogRepository from app.schemas import api_key_schema from app.schemas.response_schema import PageData, PageMeta @@ -65,7 +65,8 @@ class ApiKeyService: BizCode.BAD_REQUEST ) - if data.resource_id: + # SERVICE 类型的 resource_id 指向 workspace，非应用，跳过应用发布校验 + if data.resource_id and data.type != ApiKeyType.SERVICE.value: app = db.get(App, data.resource_id) if not app or not app.current_release_id: raise BusinessException("该应用未发布", BizCode.APP_NOT_PUBLISHED) @@ -452,9 +453,12 @@ class ApiKeyAuthService: def check_app_published(db: Session, api_key_obj: ApiKey) -> None: """ 检查应用是否已发布，未发布则抛出异常 + SERVICE 类型的 api_key 不绑定应用（resource_id 指向 workspace），跳过校验 """ if not api_key_obj.resource_id: return + if api_key_obj.type == ApiKeyType.SERVICE.value: + return app = db.get(App, api_key_obj.resource_id) if not app or not app.current_release_id: raise BusinessException("应用未发布，不可用", BizCode.APP_NOT_PUBLISHED) From b64bcc2c50dc9bc38354acbf219cfc03a6df7746 Mon Sep 17 00:00:00 2001 From: wxy Date: Mon, 27 Apr 2026 15:20:25 +0800 Subject: [PATCH 242/245] feat(workflow): augment logging queries and ameliorate error handling - Augment log search with app type filtering to enable keyword searching within workflow_executions. - Introduce execution sequence markers to ensure logs are displayed in the correct chronological order. - Ameliorate error handling to capture successful node outputs alongside failure details. - Rectify the processing of empty JSON bodies in HTTP request nodes. --- api/app/controllers/app_log_controller.py | 5 ++- api/app/core/workflow/executor.py | 38 ++++++++++++++++++- api/app/core/workflow/nodes/base_node.py | 28 +++++++++++++- .../workflow/nodes/cycle_graph/iteration.py | 3 ++ .../core/workflow/nodes/http_request/node.py | 13 ++++++- .../repositories/conversation_repository.py | 38 +++++++++++++++---- api/app/services/app_log_service.py | 20 ++++++++-- 7 files changed, 126 insertions(+), 19 deletions(-) diff --git a/api/app/controllers/app_log_controller.py b/api/app/controllers/app_log_controller.py index ea7962c1..90fbd4ea 100644 --- a/api/app/controllers/app_log_controller.py +++ b/api/app/controllers/app_log_controller.py @@ -41,7 +41,7 @@ def list_app_logs( # 验证应用访问权限 app_service = AppService(db) - app_service.get_app(app_id, workspace_id) + app = app_service.get_app(app_id, workspace_id) # 使用 Service 层查询 log_service = AppLogService(db) @@ -51,7 +51,8 @@ def list_app_logs( page=page, pagesize=pagesize, is_draft=is_draft, - keyword=keyword + keyword=keyword, + app_type=app.type, ) items = [AppLogConversation.model_validate(c) for c in conversations] diff --git a/api/app/core/workflow/executor.py b/api/app/core/workflow/executor.py index 6ac48ede..ea05db87 100644 --- a/api/app/core/workflow/executor.py +++ b/api/app/core/workflow/executor.py @@ -16,6 +16,7 @@ from app.core.workflow.engine.runtime_schema import ExecutionContext from app.core.workflow.engine.state_manager import WorkflowStateManager from app.core.workflow.engine.stream_output_coordinator import StreamOutputCoordinator from app.core.workflow.engine.variable_pool import VariablePool, VariablePoolInitializer +from app.core.workflow.nodes.base_node import NodeExecutionError logger = logging.getLogger(__name__) @@ -326,10 +327,43 @@ class WorkflowExecutor: logger.error(f"Workflow execution failed: execution_id={self.execution_context.execution_id}, error={e}", exc_info=True) + + # 1) 尝试从 checkpoint 回补已成功节点的 node_outputs + recovered: dict[str, Any] = {} + try: + if self.graph is not None: + recovered = self.graph.get_state( + self.execution_context.checkpoint_config + ).values or {} + except Exception as recover_err: + logger.warning( + f"Recover state on failure failed: {recover_err}, " + f"execution_id={self.execution_context.execution_id}" + ) + if result is None: - result = {"error": str(e)} + result = dict(recovered) if recovered else {} else: - result["error"] = str(e) + # 已有 result 与 recovered 合并，node_outputs 深度合并 + for k, v in recovered.items(): + if k == "node_outputs" and isinstance(v, dict): + existing = result.get("node_outputs") or {} + result["node_outputs"] = {**v, **existing} + else: + result.setdefault(k, v) + + # 2) 如果是节点抛出的 NodeExecutionError，把失败节点的 node_output 注入 node_outputs + failed_node_id: str | None = None + if isinstance(e, NodeExecutionError): + failed_node_id = e.node_id + node_outputs = result.setdefault("node_outputs", {}) + # 不覆盖已有（理论上不会有），保底写入失败节点记录 + node_outputs.setdefault(e.node_id, e.node_output) + + result["error"] = str(e) + if failed_node_id: + result["error_node"] = failed_node_id + yield { "event": "workflow_end", "data": self.result_builder.build_final_output( diff --git a/api/app/core/workflow/nodes/base_node.py b/api/app/core/workflow/nodes/base_node.py index 5458a80c..5d08670a 100644 --- a/api/app/core/workflow/nodes/base_node.py +++ b/api/app/core/workflow/nodes/base_node.py @@ -1,5 +1,6 @@ import asyncio import logging +import time import uuid from abc import ABC, abstractmethod from datetime import datetime @@ -22,6 +23,20 @@ from app.services.multimodal_service import MultimodalService logger = logging.getLogger(__name__) +class NodeExecutionError(Exception): + """节点执行失败异常。 + + 携带失败节点的完整 node_output，供 executor 兜底注入 node_outputs， + 保证 workflow_executions.output_data 里能看到失败节点的日志记录。 + """ + + def __init__(self, node_id: str, node_output: dict[str, Any], error_message: str): + super().__init__(f"Node {node_id} execution failed: {error_message}") + self.node_id = node_id + self.node_output = node_output + self.error_message = error_message + + class BaseNode(ABC): """Base class for workflow nodes. @@ -396,6 +411,8 @@ class BaseNode(ABC): "elapsed_time": elapsed_time, "token_usage": token_usage, "error": None, + # 单调递增序号，用于日志按执行顺序排序（JSONB 不保证 key 顺序） + "execution_order": time.monotonic_ns(), **self._extract_extra_fields(business_result), } final_output = { @@ -444,7 +461,9 @@ class BaseNode(ABC): "output": None, "elapsed_time": elapsed_time, "token_usage": None, - "error": error_message + "error": error_message, + # 单调递增序号，用于日志按执行顺序排序 + "execution_order": time.monotonic_ns(), } # if error_edge: @@ -466,7 +485,12 @@ class BaseNode(ABC): **node_output }) logger.error(f"Node {self.node_id} execution failed, stopping workflow: {error_message}") - raise Exception(f"Node {self.node_id} execution failed: {error_message}") + # 抛出自定义异常，把 node_output 带给 executor，供其写入 node_outputs + raise NodeExecutionError( + node_id=self.node_id, + node_output=node_output, + error_message=error_message, + ) def _extract_input(self, state: WorkflowState, variable_pool: VariablePool) -> dict[str, Any]: """Extracts the input data for this node (used for logging or audit). diff --git a/api/app/core/workflow/nodes/cycle_graph/iteration.py b/api/app/core/workflow/nodes/cycle_graph/iteration.py index 3ee7774f..3ce22ab2 100644 --- a/api/app/core/workflow/nodes/cycle_graph/iteration.py +++ b/api/app/core/workflow/nodes/cycle_graph/iteration.py @@ -174,6 +174,9 @@ class IterationRuntime: continue node_type = result.get("node_outputs", {}).get(node_name, {}).get("node_type") cycle_variable = {"item": item} if node_type == NodeType.CYCLE_START else None + node_cfg = next( + (n for n in self.cycle_nodes if n.get("id") == node_name), None + ) self.event_write({ "type": "cycle_item", "data": { diff --git a/api/app/core/workflow/nodes/http_request/node.py b/api/app/core/workflow/nodes/http_request/node.py index da13ceec..6b117368 100644 --- a/api/app/core/workflow/nodes/http_request/node.py +++ b/api/app/core/workflow/nodes/http_request/node.py @@ -255,9 +255,18 @@ class HttpRequestNode(BaseNode): case HttpContentType.NONE: return {} case HttpContentType.JSON: - content["json"] = json.loads(self._render_template( + rendered = self._render_template( self.typed_config.body.data, variable_pool - )) + ) + if not rendered or not rendered.strip(): + # 第三方导入的工作流可能出现 content_type=json 但 data 为空的情况，视为无 body + return {} + try: + content["json"] = json.loads(rendered) + except json.JSONDecodeError as e: + raise RuntimeError( + f"Invalid JSON body for HTTP request node: {e.msg} (data={rendered!r})" + ) case HttpContentType.FROM_DATA: data = {} files = [] diff --git a/api/app/repositories/conversation_repository.py b/api/app/repositories/conversation_repository.py index 129e1f02..e3447dbd 100644 --- a/api/app/repositories/conversation_repository.py +++ b/api/app/repositories/conversation_repository.py @@ -1,13 +1,15 @@ import uuid from typing import Optional -from sqlalchemy import select, desc, func +from sqlalchemy import select, desc, func, or_, cast, Text from sqlalchemy.orm import Session from app.core.exceptions import ResourceNotFoundException from app.core.logging_config import get_db_logger from app.models import Conversation, Message +from app.models.app_model import AppType from app.models.conversation_model import ConversationDetail +from app.models.workflow_model import WorkflowExecution logger = get_db_logger() @@ -206,7 +208,8 @@ class ConversationRepository: is_draft: Optional[bool] = None, keyword: Optional[str] = None, page: int = 1, - pagesize: int = 20 + pagesize: int = 20, + app_type: Optional[str] = None, ) -> tuple[list[Conversation], int]: """ 查询应用日志会话列表（带分页和过滤） @@ -218,6 +221,9 @@ class ConversationRepository: keyword: 搜索关键词（匹配消息内容） page: 页码（从 1 开始） pagesize: 每页数量 + app_type: 应用类型。WORKFLOW 类型改用 workflow_executions 的 + input_data/output_data 做关键词过滤（因为失败的工作流不会写入 messages 表）； + 其他类型仍走 messages 表。 Returns: Tuple[List[Conversation], int]: (会话列表，总数) @@ -234,12 +240,28 @@ class ConversationRepository: # 如果有关键词搜索，通过子查询过滤包含该关键词的 conversation if keyword: - # 查找包含关键词的 conversation_id 列表 - keyword_stmt = ( - select(Message.conversation_id) - .where(Message.content.ilike(f"%{keyword}%")) - .distinct() - ) + kw_pattern = f"%{keyword}%" + if app_type == AppType.WORKFLOW: + # 工作流：从 workflow_executions 的 input_data / output_data 匹配 + # （messages 表只存开场白 assistant 消息，失败的工作流也不会写入） + keyword_stmt = ( + select(WorkflowExecution.conversation_id) + .where( + WorkflowExecution.conversation_id.is_not(None), + or_( + cast(WorkflowExecution.input_data, Text).ilike(kw_pattern), + cast(WorkflowExecution.output_data, Text).ilike(kw_pattern), + ), + ) + .distinct() + ) + else: + # Agent 等其他类型：仍走 messages 表（user + assistant 内容) + keyword_stmt = ( + select(Message.conversation_id) + .where(Message.content.ilike(kw_pattern)) + .distinct() + ) base_stmt = base_stmt.where(Conversation.id.in_(keyword_stmt)) # Calculate total number of records diff --git a/api/app/services/app_log_service.py b/api/app/services/app_log_service.py index 7ca05d42..c2cff2a6 100644 --- a/api/app/services/app_log_service.py +++ b/api/app/services/app_log_service.py @@ -32,6 +32,7 @@ class AppLogService: pagesize: int = 20, is_draft: Optional[bool] = None, keyword: Optional[str] = None, + app_type: Optional[str] = None, ) -> Tuple[list[Conversation], int]: """ 查询应用日志会话列表 @@ -43,6 +44,7 @@ class AppLogService: pagesize: 每页数量 is_draft: 是否草稿会话（None表示返回全部） keyword: 搜索关键词（匹配消息内容） + app_type: 应用类型（WORKFLOW 时关键词将从 workflow_executions 搜索） Returns: Tuple[list[Conversation], int]: (会话列表，总数) @@ -55,7 +57,8 @@ class AppLogService: "page": page, "pagesize": pagesize, "is_draft": is_draft, - "keyword": keyword + "keyword": keyword, + "app_type": app_type, } ) @@ -66,7 +69,8 @@ class AppLogService: is_draft=is_draft, keyword=keyword, page=page, - pagesize=pagesize + pagesize=pagesize, + app_type=app_type, ) logger.info( @@ -368,8 +372,16 @@ def _build_nodes_from_output_data(output_data: Optional[dict]) -> list[AppLogNod if not output_data: return [] node_outputs: dict = output_data.get("node_outputs") or {} + # 按 execution_order（节点执行时写入的单调递增序号）排序。 + # PostgreSQL JSONB 不保证 key 顺序，不能依赖 dict 插入顺序； + # 缺失 execution_order 的历史数据退化到 0，保持在最前。 + ordered_items = sorted( + node_outputs.items(), + key=lambda kv: (kv[1] or {}).get("execution_order", 0) + if isinstance(kv[1], dict) else 0 + ) result = [] - for node_id, node_data in node_outputs.items(): + for node_id, node_data in ordered_items: if not isinstance(node_data, dict): continue output = dict(node_data) @@ -382,6 +394,8 @@ def _build_nodes_from_output_data(output_data: Optional[dict]) -> list[AppLogNod inp = output.pop("input", None) elapsed_time = output.pop("elapsed_time", None) token_usage = output.pop("token_usage", None) + # execution_order 仅用于排序，不返回给前端 + output.pop("execution_order", None) result.append(AppLogNodeExecution( node_id=node_id, node_type=node_type, From faf8d1a51a8bedb72a1f0109979ceb0e76911374 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Mon, 27 Apr 2026 15:35:26 +0800 Subject: [PATCH 243/245] fix(workflow): add reasoning content, suggested questions, citations and audio status support - Introduce `reasoning_content`, `suggested_questions`, `citations`, and `audio_status` fields in conversation and app response schemas - Conditionally set `audio_status` to `"pending"` only when `audio_url` is present - Replace `model_dump` override with `@model_serializer(mode="wrap")` for cleaner serialization logic - Change knowledge base validation failure from `RuntimeError` to warning + `continue` to avoid halting retrieval on invalid KB --- api/app/core/workflow/nodes/knowledge/node.py | 3 ++- api/app/schemas/app_schema.py | 8 +++++--- api/app/schemas/conversation_schema.py | 14 +++++++++++++- api/app/services/app_chat_service.py | 2 +- api/app/services/draft_run_service.py | 2 +- 5 files changed, 22 insertions(+), 7 deletions(-) diff --git a/api/app/core/workflow/nodes/knowledge/node.py b/api/app/core/workflow/nodes/knowledge/node.py index bd1ba998..c3fda4e2 100644 --- a/api/app/core/workflow/nodes/knowledge/node.py +++ b/api/app/core/workflow/nodes/knowledge/node.py @@ -334,7 +334,8 @@ class KnowledgeRetrievalNode(BaseNode): for kb_config in knowledge_bases: db_knowledge = knowledge_repository.get_knowledge_by_id(db=db, knowledge_id=kb_config.kb_id) if not (db_knowledge and db_knowledge.chunk_num > 0 and db_knowledge.status == 1): - raise RuntimeError("The knowledge base does not exist or access is denied.") + logger.warning("The knowledge base does not exist or access is denied.") + continue tasks.append(self.knowledge_retrieval(db, query, db_knowledge, kb_config)) if tasks: result = await asyncio.gather(*tasks) diff --git a/api/app/schemas/app_schema.py b/api/app/schemas/app_schema.py index 11c27b56..89603322 100644 --- a/api/app/schemas/app_schema.py +++ b/api/app/schemas/app_schema.py @@ -3,7 +3,7 @@ import uuid from typing import Optional, Any, List, Dict, Union from enum import Enum, StrEnum -from pydantic import BaseModel, Field, ConfigDict, field_serializer, field_validator +from pydantic import BaseModel, Field, ConfigDict, field_serializer, field_validator, model_serializer from app.schemas.workflow_schema import WorkflowConfigCreate @@ -661,9 +661,11 @@ class DraftRunResponse(BaseModel): suggested_questions: List[str] = Field(default_factory=list, description="下一步建议问题") citations: List[Dict[str, Any]] = Field(default_factory=list, description="引用来源") audio_url: Optional[str] = Field(default=None, description="TTS 语音URL") + audio_status: Optional[str] = Field(default=None, description="TTS 语音状态") - def model_dump(self, **kwargs): - data = super().model_dump(**kwargs) + @model_serializer(mode="wrap") + def _serialize(self, handler): + data = handler(self) if not data.get("reasoning_content"): data.pop("reasoning_content", None) return data diff --git a/api/app/schemas/conversation_schema.py b/api/app/schemas/conversation_schema.py index fd1be5d9..20782fd9 100644 --- a/api/app/schemas/conversation_schema.py +++ b/api/app/schemas/conversation_schema.py @@ -2,7 +2,7 @@ import uuid import datetime from typing import Optional, Dict, Any, List -from pydantic import BaseModel, Field, ConfigDict, field_serializer +from pydantic import BaseModel, Field, ConfigDict, field_serializer, model_serializer # 导入 FileInput（用于体验运行） from app.schemas.app_schema import FileInput @@ -94,6 +94,18 @@ class ChatResponse(BaseModel): message_id: str usage: Optional[Dict[str, Any]] = None elapsed_time: Optional[float] = None + reasoning_content: Optional[str] = None + suggested_questions: Optional[List[str]] = None + citations: Optional[List[Any]] = None + audio_url: Optional[str] = None + audio_status: Optional[str] = None + + @model_serializer(mode="wrap") + def _serialize(self, handler): + data = handler(self) + if not data.get("reasoning_content"): + data.pop("reasoning_content", None) + return data # ---------- Conversation Summary Schemas ---------- diff --git a/api/app/services/app_chat_service.py b/api/app/services/app_chat_service.py index 7c52c2d6..12f54c03 100644 --- a/api/app/services/app_chat_service.py +++ b/api/app/services/app_chat_service.py @@ -317,7 +317,7 @@ class AppChatService: "suggested_questions": suggested_questions, "citations": filtered_citations, "audio_url": audio_url, - "audio_status": "pending" + "audio_status": "pending" if audio_url else None } async def agnet_chat_stream( diff --git a/api/app/services/draft_run_service.py b/api/app/services/draft_run_service.py index 10d50ece..2566a50f 100644 --- a/api/app/services/draft_run_service.py +++ b/api/app/services/draft_run_service.py @@ -754,7 +754,7 @@ class AgentRunService: ) if not sub_agent else [], "citations": filtered_citations, "audio_url": audio_url, - "audio_status": "pending" + "audio_status": "pending" if audio_url else None } logger.info( From 12a08a487d02f720ede1a666e6a48d0912f443e0 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Mon, 27 Apr 2026 15:47:34 +0800 Subject: [PATCH 244/245] fix(tool_controller): re-raise HTTPException to preserve original status codes --- api/app/controllers/tool_controller.py | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/api/app/controllers/tool_controller.py b/api/app/controllers/tool_controller.py index 74b8d88e..688ab518 100644 --- a/api/app/controllers/tool_controller.py +++ b/api/app/controllers/tool_controller.py @@ -173,6 +173,8 @@ async def delete_tool( return success(msg="工具删除成功") except ValueError as e: raise HTTPException(status_code=400, detail=str(e)) + except HTTPException: + raise except Exception as e: raise HTTPException(status_code=500, detail=str(e)) @@ -249,6 +251,8 @@ async def parse_openapi_schema( if result["success"] is False: raise HTTPException(status_code=400, detail=result["message"]) return success(data=result, msg="Schema解析完成") + except HTTPException: + raise except Exception as e: raise HTTPException(status_code=500, detail=str(e)) From 98d8d7b26138630a622344a235b06a4ed69ad857 Mon Sep 17 00:00:00 2001 From: Timebomb2018 <18868801967@163.com> Date: Mon, 27 Apr 2026 15:49:21 +0800 Subject: [PATCH 245/245] fix(conversation_schema): refine citations field type to Dict[str, Any] --- api/app/schemas/conversation_schema.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/api/app/schemas/conversation_schema.py b/api/app/schemas/conversation_schema.py index 20782fd9..7c3a0f03 100644 --- a/api/app/schemas/conversation_schema.py +++ b/api/app/schemas/conversation_schema.py @@ -96,7 +96,7 @@ class ChatResponse(BaseModel): elapsed_time: Optional[float] = None reasoning_content: Optional[str] = None suggested_questions: Optional[List[str]] = None - citations: Optional[List[Any]] = None + citations: Optional[List[Dict[str, Any]]] = None audio_url: Optional[str] = None audio_status: Optional[str] = None