From fcf3071cb0d26a96c45be57b6a7862cbd52e13c7 Mon Sep 17 00:00:00 2001 From: Mark <348207283@qq.com> Date: Tue, 14 Apr 2026 16:46:57 +0800 Subject: [PATCH 01/55] [modify] Optimize ES connections and add rerank security checks --- api/app/controllers/chunk_controller.py | 2 +- api/app/core/rag/nlp/search.py | 2 +- api/app/core/rag/utils/es_conn.py | 4 +- .../vdb/elasticsearch/elasticsearch_vector.py | 187 ++++++++---------- 4 files changed, 84 insertions(+), 111 deletions(-) diff --git a/api/app/controllers/chunk_controller.py b/api/app/controllers/chunk_controller.py index b5c0a5ae..b2cc3695 100644 --- a/api/app/controllers/chunk_controller.py +++ b/api/app/controllers/chunk_controller.py @@ -457,7 +457,7 @@ async def retrieve_chunks( if doc.metadata["doc_id"] not in seen_ids: seen_ids.add(doc.metadata["doc_id"]) unique_rs.append(doc) - rs = vector_service.rerank(query=retrieve_data.query, docs=unique_rs, top_k=retrieve_data.top_k) + rs = vector_service.rerank(query=retrieve_data.query, docs=unique_rs, top_k=retrieve_data.top_k) if unique_rs else [] if retrieve_data.retrieve_type == chunk_schema.RetrieveType.Graph: kb_ids = [str(kb_id) for kb_id in private_kb_ids] workspace_ids = [str(workspace_id) for workspace_id in private_workspace_ids] diff --git a/api/app/core/rag/nlp/search.py b/api/app/core/rag/nlp/search.py index 61540ee4..4b99986b 100644 --- a/api/app/core/rag/nlp/search.py +++ b/api/app/core/rag/nlp/search.py @@ -113,7 +113,7 @@ def knowledge_retrieval( continue # Use the specified reranker for re-ranking - if reranker_id: + if reranker_id and all_results: try: all_results = rerank(db=db, reranker_id=reranker_id, query=query, docs=all_results, top_k=reranker_top_k) except Exception as rerank_error: diff --git a/api/app/core/rag/utils/es_conn.py b/api/app/core/rag/utils/es_conn.py index 7fbf0e38..9a0edd24 100644 --- a/api/app/core/rag/utils/es_conn.py +++ b/api/app/core/rag/utils/es_conn.py @@ -68,9 +68,9 @@ class ESConnection(DocStoreConnection): client_config = { "hosts": [hosts], "basic_auth": (os.getenv("ELASTICSEARCH_USERNAME", "elastic"), os.getenv("ELASTICSEARCH_PASSWORD", "elastic")), - "request_timeout": int(os.getenv("ELASTICSEARCH_REQUEST_TIMEOUT", 100000)), + "request_timeout": int(os.getenv("ELASTICSEARCH_REQUEST_TIMEOUT", 30)), "retry_on_timeout": os.getenv("ELASTICSEARCH_RETRY_ON_TIMEOUT", True) == "true", - "max_retries": int(os.getenv("ELASTICSEARCH_MAX_RETRIES", 10000)), + "max_retries": int(os.getenv("ELASTICSEARCH_MAX_RETRIES", 3)), } # Only add SSL settings if using HTTPS diff --git a/api/app/core/rag/vdb/elasticsearch/elasticsearch_vector.py b/api/app/core/rag/vdb/elasticsearch/elasticsearch_vector.py index 386920e0..b388e9be 100644 --- a/api/app/core/rag/vdb/elasticsearch/elasticsearch_vector.py +++ b/api/app/core/rag/vdb/elasticsearch/elasticsearch_vector.py @@ -8,8 +8,6 @@ import requests from elasticsearch import Elasticsearch, helpers from elasticsearch.helpers import BulkIndexError from packaging.version import parse as parse_version -from pydantic import BaseModel, model_validator -from abc import ABC # langchain-community # langchain-xinference # from langchain_community.embeddings import XinferenceEmbeddings @@ -29,37 +27,9 @@ from app.core.rag.models.chunk import DocumentChunk logger = logging.getLogger(__name__) -class ElasticSearchConfig(BaseModel): - # Regular Elasticsearch config - host: str | None = None - port: int | None = None - username: str | None = None - password: str | None = None - - # Common config - ca_certs: str | None = None - verify_certs: bool = False - request_timeout: int = 100000 - retry_on_timeout: bool = True - max_retries: int = 10000 - - @model_validator(mode="before") - @classmethod - def validate_config(cls, values: dict): - # Regular Elasticsearch validation - if not values.get("host"): - raise ValueError("config HOST is required for regular Elasticsearch") - if not values.get("port"): - raise ValueError("config PORT is required for regular Elasticsearch") - if not values.get("username"): - raise ValueError("config USERNAME is required for regular Elasticsearch") - if not values.get("password"): - raise ValueError("config PASSWORD is required for regular Elasticsearch") - return values - - class ElasticSearchVector(BaseVector): - def __init__(self, index_name: str, config: ElasticSearchConfig, embedding_config: ModelApiKey, reranker_config: ModelApiKey): + def __init__(self, index_name: str, client: Elasticsearch, + embedding_config: ModelApiKey, reranker_config: ModelApiKey): super().__init__(index_name.lower()) # 初始化 Embedding 模型（自动支持火山引擎多模态） @@ -77,58 +47,8 @@ class ElasticSearchVector(BaseVector): api_key=reranker_config.api_key, base_url=reranker_config.api_base )) - self._client = self._init_client(config) - self._version = self._get_version() - self._check_version() - - def _init_client(self, config: ElasticSearchConfig) -> Elasticsearch: - """ - Initialize Elasticsearch client for regular Elasticsearch. - """ - try: - # Regular Elasticsearch configuration - parsed_url = urlparse(config.host or "") - if parsed_url.scheme in {"http", "https"}: - hosts = f"{config.host}:{config.port}" - use_https = parsed_url.scheme == "https" - else: - hosts = f"https://{config.host}:{config.port}" - use_https = False - - client_config = { - "hosts": [hosts], - "basic_auth": (config.username, config.password), - "request_timeout": config.request_timeout, - "retry_on_timeout": config.retry_on_timeout, - "max_retries": config.max_retries, - } - - # Only add SSL settings if using HTTPS - if use_https: - client_config["verify_certs"] = config.verify_certs - if config.ca_certs: - client_config["ca_certs"] = config.ca_certs - - client = Elasticsearch(**client_config) - - # Test connection - if not client.ping(): - raise ConnectionError("Failed to connect to Elasticsearch") - - except requests.ConnectionError as e: - raise ConnectionError(f"Vector database connection error: {str(e)}") - except Exception as e: - raise ConnectionError(f"Elasticsearch client initialization failed: {str(e)}") - - return client - - def _get_version(self) -> str: - info = self._client.info() - return cast(str, info["version"]["number"]) - - def _check_version(self): - if parse_version(self._version) < parse_version("8.0.0"): - raise ValueError("Elasticsearch vector database version must be greater than 8.0.0") + # 使用外部传入的共享客户端 + self._client = client def get_type(self) -> str: return "elasticsearch" @@ -744,30 +664,83 @@ class ElasticSearchVector(BaseVector): self._client.indices.create(index=self._collection_name, body=index_mapping) +import threading + + class ElasticSearchVectorFactory: - @staticmethod - def init_vector(knowledge: Knowledge) -> ElasticSearchVector: + """ES 向量服务工厂 - 单例共享连接""" + + _client: Elasticsearch | None = None + _lock = threading.Lock() + _version_checked = False + + @classmethod + def _get_shared_client(cls) -> Elasticsearch: + """获取共享的 ES 客户端（线程安全的懒加载单例）""" + if cls._client is not None: + return cls._client + + with cls._lock: + # 双重检查，防止并发时重复创建 + if cls._client is not None: + return cls._client + + try: + parsed_url = urlparse(os.getenv("ELASTICSEARCH_HOST", "127.0.0.1") or "") + if parsed_url.scheme in {"http", "https"}: + hosts = f'{os.getenv("ELASTICSEARCH_HOST")}:{os.getenv("ELASTICSEARCH_PORT", 9200)}' + use_https = parsed_url.scheme == "https" + else: + hosts = f'https://{os.getenv("ELASTICSEARCH_HOST", "127.0.0.1")}:{os.getenv("ELASTICSEARCH_PORT", 9200)}' + use_https = False + + client_config = { + "hosts": [hosts], + "basic_auth": ( + os.getenv("ELASTICSEARCH_USERNAME", "elastic"), + os.getenv("ELASTICSEARCH_PASSWORD", "elastic"), + ), + "request_timeout": int(os.getenv("ELASTICSEARCH_REQUEST_TIMEOUT", 30)), + "retry_on_timeout": True, + "max_retries": int(os.getenv("ELASTICSEARCH_MAX_RETRIES", 3)), + "connections_per_node": int(os.getenv("ELASTICSEARCH_CONNECTIONS_PER_NODE", 10)), + } + + if use_https: + client_config["verify_certs"] = os.getenv("ELASTICSEARCH_VERIFY_CERTS", "false") == "true" + ca_certs = os.getenv("ELASTICSEARCH_CA_CERTS") + if ca_certs: + client_config["ca_certs"] = str(ca_certs) + + client = Elasticsearch(**client_config) + + if not client.ping(): + raise ConnectionError("Failed to connect to Elasticsearch") + + # 版本检查只做一次 + if not cls._version_checked: + info = client.info() + version = info["version"]["number"] + if parse_version(version) < parse_version("8.0.0"): + raise ValueError(f"Elasticsearch version must be >= 8.0.0, got {version}") + cls._version_checked = True + logger.info(f"Elasticsearch shared client initialized, version: {version}") + + cls._client = client + + except requests.ConnectionError as e: + raise ConnectionError(f"Vector database connection error: {str(e)}") + except Exception as e: + raise ConnectionError(f"Elasticsearch client initialization failed: {str(e)}") + + return cls._client + + @classmethod + def init_vector(cls, knowledge: Knowledge) -> ElasticSearchVector: + """创建向量服务实例（共享 ES 连接）""" + client = cls._get_shared_client() collection_name = f"Vector_index_{knowledge.id}_Node" - # Use regular Elasticsearch with config values - config_dict = { - "host": os.getenv("ELASTICSEARCH_HOST", "127.0.0.1"), - "port": os.getenv("ELASTICSEARCH_PORT", 9200), - "username": os.getenv("ELASTICSEARCH_USERNAME", "elastic"), - "password": os.getenv("ELASTICSEARCH_PASSWORD", "elastic"), - } - - # Common configuration - config_dict.update( - { - "ca_certs": str(os.getenv("ELASTICSEARCH_CA_CERTS")) if os.getenv("ELASTICSEARCH_CA_CERTS") else None, - "verify_certs": os.getenv("ELASTICSEARCH_VERIFY_CERTS", False) == "true", - "request_timeout": int(os.getenv("ELASTICSEARCH_REQUEST_TIMEOUT", 100000)), - "retry_on_timeout": os.getenv("ELASTICSEARCH_RETRY_ON_TIMEOUT", True) == "true", - "max_retries": int(os.getenv("ELASTICSEARCH_MAX_RETRIES", 10000)), - } - ) - if knowledge.embedding is None: raise ValueError(f"embedding_id config error: {str(knowledge.embedding_id)}") if knowledge.reranker is None: @@ -775,9 +748,9 @@ class ElasticSearchVectorFactory: return ElasticSearchVector( index_name=collection_name, - config=ElasticSearchConfig(**config_dict), + client=client, embedding_config=knowledge.embedding.api_keys[0], - reranker_config=knowledge.reranker.api_keys[0] + reranker_config=knowledge.reranker.api_keys[0], ) From 3b359df02fe470546e616a1f6faace74e75500bd Mon Sep 17 00:00:00 2001 From: Mark <348207283@qq.com> Date: Tue, 14 Apr 2026 17:02:11 +0800 Subject: [PATCH 02/55] [modify] fix --- .../rag/vdb/elasticsearch/elasticsearch_vector.py | 12 ++++-------- 1 file changed, 4 insertions(+), 8 deletions(-) diff --git a/api/app/core/rag/vdb/elasticsearch/elasticsearch_vector.py b/api/app/core/rag/vdb/elasticsearch/elasticsearch_vector.py index b388e9be..cc9ec120 100644 --- a/api/app/core/rag/vdb/elasticsearch/elasticsearch_vector.py +++ b/api/app/core/rag/vdb/elasticsearch/elasticsearch_vector.py @@ -1,8 +1,8 @@ import os import logging -from typing import Any, cast +import threading +from typing import Any from urllib.parse import urlparse -import uuid import requests from elasticsearch import Elasticsearch, helpers @@ -14,10 +14,9 @@ from packaging.version import parse as parse_version # from langchain_xinference import XinferenceRerank from langchain_core.documents import Document from app.core.models.base import RedBearModelConfig -from app.core.models import RedBearLLM, RedBearRerank +from app.core.models import RedBearRerank from app.core.models.embedding import RedBearEmbeddings -from app.models.models_model import ModelConfig, ModelApiKey -from app.services.model_service import ModelConfigService +from app.models.models_model import ModelApiKey from app.models.knowledge_model import Knowledge from app.core.rag.vdb.field import Field @@ -664,9 +663,6 @@ class ElasticSearchVector(BaseVector): self._client.indices.create(index=self._collection_name, body=index_mapping) -import threading - - class ElasticSearchVectorFactory: """ES 向量服务工厂 - 单例共享连接""" From a2f85b3d985044fd79dfaeef0ff273d5dc4d6534 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 20 Apr 2026 10:16:31 +0800 Subject: [PATCH 03/55] fix(web): update quotas key --- web/src/components/SiderMenu/SubscriptionDetailModal.tsx | 2 +- web/src/components/SiderMenu/index.tsx | 6 +++--- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/web/src/components/SiderMenu/SubscriptionDetailModal.tsx b/web/src/components/SiderMenu/SubscriptionDetailModal.tsx index ae084fcd..fbb5030c 100644 --- a/web/src/components/SiderMenu/SubscriptionDetailModal.tsx +++ b/web/src/components/SiderMenu/SubscriptionDetailModal.tsx @@ -82,7 +82,7 @@ const SubscriptionDetailModal = forwardRef((_props, {/* Features */} {billingUnits.map(({ key, unit, icon }) => { - const value = detail?.quota[key as keyof Subscription['quota']]; + const value = detail?.quotas[key as keyof Subscription['quotas']]; if (value === undefined || value === null) return null; return ( {['workspace_quota', 'skill_quota', 'app_quota', 'model_quota'].map(key => (
-
{subscription.quota?.[key as keyof typeof subscription.quota]}
+
{subscription.quotas?.[key as keyof typeof subscription.quotas]}
{t(`index.${key}`)}
))} From 4a39fd5f464165299e1b88472ddb6ca194e4f3ed Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 20 Apr 2026 14:31:31 +0800 Subject: [PATCH 04/55] fix(web) if-else port y calculate update --- .../Workflow/components/CheckList/index.tsx | 8 +- .../components/Nodes/ConditionNode.tsx | 32 ++-- .../components/Properties/CaseList/index.tsx | 35 ++-- web/src/views/Workflow/constant.ts | 4 +- web/src/views/Workflow/utils.ts | 160 +++++------------- 5 files changed, 102 insertions(+), 137 deletions(-) diff --git a/web/src/views/Workflow/components/CheckList/index.tsx b/web/src/views/Workflow/components/CheckList/index.tsx index 4afdb863..b2598999 100644 --- a/web/src/views/Workflow/components/CheckList/index.tsx +++ b/web/src/views/Workflow/components/CheckList/index.tsx @@ -1,3 +1,9 @@ +/* + * @Author: ZhaoYing + * @Date: 2026-04-09 18:58:21 + * @Last Modified by: ZhaoYing + * @Last Modified time: 2026-04-20 10:39:17 + */ import { useState, useCallback, useEffect, useRef, type FC } from 'react' import { Popover, Flex } from 'antd' import { WarningFilled } from '@ant-design/icons' @@ -49,7 +55,7 @@ const specialValidators: Record boolean> = { if (expr?.sub_variable_condition?.conditions?.length > 0) return expr.sub_variable_condition?.conditions.every(isSubExprSet) if (!expr.left) return false if (['not_empty', 'empty'].includes(expr.operator)) return true - return !!expr.left && (!!expr.right || typeof expr.right === 'boolean' || typeof expr.right === 'number') + return !!expr.left && (expr?.sub_variable_condition || !!expr.right || typeof expr.right === 'boolean' || typeof expr.right === 'number') } return val.some(c => !c?.expressions?.length || c.expressions.some((expr: any) => !isExprSet(expr))) }, diff --git a/web/src/views/Workflow/components/Nodes/ConditionNode.tsx b/web/src/views/Workflow/components/Nodes/ConditionNode.tsx index 9cd1309e..1bd6559f 100644 --- a/web/src/views/Workflow/components/Nodes/ConditionNode.tsx +++ b/web/src/views/Workflow/components/Nodes/ConditionNode.tsx @@ -9,7 +9,7 @@ import { useVariableList } from '../Properties/hooks/useVariableList' import { isSubExprSet } from '../../utils' import { fileSubFieldOperators } from '../Properties/CaseList' -const caculateIsSet = (item: any, type: string) => { +const calculateIsSet = (item: any, type: string) => { switch (type) { case 'categories': return typeof item?.class_name === 'string' && item?.class_name !== '' @@ -79,7 +79,7 @@ const ConditionNode: ReactShapeConfig['component'] = ({ node }) => {
{t('workflow.config.question-classifier.class_name')} {index + 1} - {caculateIsSet(item, 'categories') ? t(`workflow.config.${data.type}.set`) : t(`workflow.config.${data.type}.unset`)} + {calculateIsSet(item, 'categories') ? t(`workflow.config.${data.type}.set`) : t(`workflow.config.${data.type}.unset`)}
))} @@ -89,17 +89,24 @@ const ConditionNode: ReactShapeConfig['component'] = ({ node }) => { {data.config?.cases?.defaultValue.map((item: any, index: number) => (
0 ? '' : 'rb:mb-1'}> - 0 ? "space-between" : 'end'} className="rb:mb-1"> - {item.expressions.length > 0 && CASE{index + 1}} + 0 ? "space-between" : 'end'} className="rb:mb-1! rb:leading-4"> + {item.expressions.length > 0 && CASE{index + 1}} {index === 0 ? 'IF' : `ELIF`} {item.expressions.length > 0 && {item.expressions.map((expression: any, eIndex: number) => (
- {item.expressions.length > 1 && eIndex > 0 &&
{item.logical_operator?.toLocaleUpperCase()}
} - + {item.expressions.length > 1 && eIndex > 0 && +
{item.logical_operator?.toLocaleUpperCase()}
+ } + 0, + 'rb:py-1!': !expression.sub_variable_condition?.conditions || !expression.sub_variable_condition?.conditions?.length + })} + > - {caculateIsSet(expression, 'cases') + {calculateIsSet(expression, 'cases') ? <> {labelRender(expression.left)} {getLocaleField(expression.operator, typeof expression.right)} @@ -109,11 +116,16 @@ const ConditionNode: ReactShapeConfig['component'] = ({ node }) => { } {expression.sub_variable_condition?.conditions?.length > 0 && expression.sub_variable_condition?.conditions.every(isSubExprSet) - ?
+ ?
{expression.sub_variable_condition?.conditions.map((sub: any, sIndex: number) => (
{expression.sub_variable_condition?.conditions.length > 1 && sIndex > 0 &&
{expression.sub_variable_condition?.logical_operator?.toLocaleUpperCase()}
} - + {sub.key} {getSubLocaleField(sub.operator, sub.key)} @@ -129,7 +141,7 @@ const ConditionNode: ReactShapeConfig['component'] = ({ node }) => { ))}
: expression.sub_variable_condition?.conditions?.length > 0 - ? + ? {t(`workflow.config.${data.type}.unset`)} : null diff --git a/web/src/views/Workflow/components/Properties/CaseList/index.tsx b/web/src/views/Workflow/components/Properties/CaseList/index.tsx index 2fd24628..a9da1457 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 20:47:49 + * @Last Modified time: 2026-04-20 10:46:05 */ import { useEffect, useMemo, type FC } from 'react' import clsx from 'clsx' @@ -39,7 +39,7 @@ interface Expression { sub_variable_condition?: SubVariableCondition; } -interface CaseItem { +export interface CaseItem { logical_operator: 'and' | 'or'; expressions: Expression[]; } @@ -274,7 +274,9 @@ const ArrayFileSubConditions: FC = ({ conditionFiel className="rb:w-full!" suffix="Byte" size="small" - onChange={(value) => { form.setFieldValue([name, caseIndex, 'expressions', conditionIndex, 'right'], value); }} + onChange={(value) => { + form.setFieldValue([name, caseIndex, 'expressions', conditionIndex, 'sub_variable_condition', 'conditions', subIndex, 'value'], value); + }} /> } @@ -483,13 +485,24 @@ const CaseList: FC = ({ form.setFieldValue([name, index, 'logical_operator'], currentValue === 'and' ? 'or' : 'and'); }; - const handleLeftFieldChange = (caseIndex: number, conditionIndex: number, newValue: string) => { - form.setFieldValue([name, caseIndex, 'expressions', conditionIndex], { - left: newValue, - operator: undefined, - right: undefined, - input_type: 'constant' - }); + const handleLeftFieldChange = (caseIndex: number, conditionIndex: number, newValue: string, option?: Suggestion | undefined) => { + if (option?.dataType === 'array[file]') { + form.setFieldValue([name, caseIndex, 'expressions', conditionIndex], { + left: newValue, + operator: undefined, + sub_variable_condition: { + conditions: [], + logical_operator: 'and' + } + }); + } else { + form.setFieldValue([name, caseIndex, 'expressions', conditionIndex], { + left: newValue, + operator: undefined, + right: undefined, + input_type: 'constant' + }); + } }; const handleAddCase = (addCaseFunc: Function) => { @@ -590,7 +603,7 @@ const CaseList: FC = ({ options={options} size="small" allowClear={false} - onChange={(val) => handleLeftFieldChange(caseIndex, conditionIndex, val as string)} + onChange={(val, option) => handleLeftFieldChange(caseIndex, conditionIndex, val as string, option as unknown as Suggestion)} variant="borderless" className="rb:w-36!" /> diff --git a/web/src/views/Workflow/constant.ts b/web/src/views/Workflow/constant.ts index cae20180..c11eb4c0 100644 --- a/web/src/views/Workflow/constant.ts +++ b/web/src/views/Workflow/constant.ts @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-03 15:06:18 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-16 17:52:30 + * @Last Modified time: 2026-04-20 11:39:40 */ import LoopNode from './components/Nodes/LoopNode'; import NormalNode from './components/Nodes/NormalNode'; @@ -734,7 +734,7 @@ export const portTextAttrs = { fontSize: 12, fill: '#5B6167' } /** * Port position arguments */ -export const portItemArgsY = 26.5; +export const portItemArgsY = 27.5; export const portArgs = { x: nodeWidth, y: portItemArgsY } const defaultPortGroup = { diff --git a/web/src/views/Workflow/utils.ts b/web/src/views/Workflow/utils.ts index bd81b6eb..74dfca2c 100644 --- a/web/src/views/Workflow/utils.ts +++ b/web/src/views/Workflow/utils.ts @@ -2,136 +2,70 @@ * @Author: ZhaoYing * @Date: 2026-03-24 15:07:49 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-17 20:40:47 + * @Last Modified time: 2026-04-20 14:20:34 */ -import { portItemArgsY, conditionNodePortItemArgsY, conditionNodeHeight } from './constant' +import { conditionNodePortItemArgsY, conditionNodeHeight } from './constant' -/** - * Calculate the total height of a condition (if-else) node based on its cases. - * - * The height is composed of: - * - `conditionNodeHeight`: the base height of the node (header + padding). - * - `(cases.length - 1) * 26`: vertical spacing added for each additional case - * beyond the first (each case separator row is 26px). - * - `exprCount * 20`: each individual expression row occupies 20px. - * - `hasMultiExprCount * 3`: a small extra padding (3px per expression) is added - * for cases that contain more than one expression, to account for the logical - * operator indicator (AND/OR) between expressions. - * - * @param cases - Array of case objects, each containing an `expressions` array. - * @returns The total pixel height for the condition node. - */ export const isSubExprSet = (sub: any) => { if (!sub?.key) return false; if (['not_empty', 'empty'].includes(sub?.operator)) return true; return !!sub.value || typeof sub.value === 'boolean' || typeof sub.value === 'number'; }; - -const getEffectiveExprCount = (expr: any): number => { - const subs = expr?.sub_variable_condition?.conditions; - if (subs?.length && subs.every(isSubExprSet)) return 1 + subs.length; - if (subs?.length > 0) { - return 2 - } - return 1; +/** + * Calculate the total height of a condition (if-else) node based on its cases. + * Uses the same per-expression height logic as getConditionNodeCasePortY. + */ +export const calcConditionNodeTotalHeight = (cases: any[]) => { + const casesHeight = cases.reduce((acc: number, c: any) => { + const exprs = c?.expressions ?? []; + const n = exprs.length; + const exprsHeight = n === 0 ? 0 : exprs.reduce((s: number, e: any) => s + calcExpressionHeight(e), 0) + 2 * (n - 1); + return acc + 20 + exprsHeight; + }, 0); + return conditionNodeHeight + casesHeight + (cases.length - 1) * 4 - 27.5; }; -export const calcConditionNodeTotalHeight = (cases: any[]) => { - // Total number of effective expression rows (sub_variable_condition expand height when all set) - const exprCount = cases.reduce((acc: number, c: any) => - acc + (c?.expressions?.reduce((s: number, e: any) => s + getEffectiveExprCount(e), 0) || 0), 0); - // Sum of effective expression counts only for cases that have more than one expression - const hasMultiExprCount = cases.reduce((acc: number, c: any) => { - if (!c?.expressions?.length || c.expressions.length <= 1) return acc; - const effectiveCount = c.expressions.reduce((s: number, e: any) => s + getEffectiveExprCount(e), 0); - return acc + effectiveCount; - }, 0); - - return conditionNodeHeight + (cases.length - 1) * 26 + exprCount * 20 + hasMultiExprCount * 3; +/** + * Height of a single expression block in ConditionNode (px). + * + * expression outer Flex padding: + * - has sub conditions (length > 0): pt-1 (4px top only) + * - no sub conditions: py-1 (4px top + 4px bottom) + * expression main row: leading-4 = 16px + * sub_variable_condition block (mt-1 = 4px gap): + * - all isSet, m subs: sub[0] = leading-3.5(14) + pb-1(4) = 18px; + * sub[k>0] = py-1(8) + leading-3.5(14) = 22px + * total = 18 + 22*(m-1) + * - exists but not all isSet: pb-1(4) + leading-4(16) = 20px + */ +const calcExpressionHeight = (expression: any): number => { + const subs = expression?.sub_variable_condition?.conditions; + if (!subs?.length) return 24; // py-1(8) + leading-4(16) + const subBlockHeight = subs.every(isSubExprSet) + ? 18 + 22 * (subs.length - 1) + : 20; + return 4 + 16 + 4 + subBlockHeight - 2; // pt-1 + main row + mt-1 + sub block (-2 rendering correction) }; /** * Calculate the Y-coordinate of the right-side output port for a specific case - * in a condition (if-else) node. + * in a condition (if-else) node, aligned with the IF/ELIF label in ConditionNode. * - * The port position is determined by iterating through all preceding cases - * (index 0 to caseIndex - 1) and accumulating their visual heights. Several - * pixel-level corrections are applied to align ports with the rendered UI: - * - * 1. **Base offset**: starts at `conditionNodePortItemArgsY`, which is the Y - * position of the first case port relative to the node top. - * - * 2. **Per-case accumulation**: for each preceding case with `n` expressions, - * add `portItemArgsY * (n + 1)` — this accounts for `n` expression rows - * plus one case header/separator row. - * - * 3. **Single-expression correction**: cases with exactly 1 expression render - * slightly shorter than the generic formula predicts. Subtract - * `singleExprCount * 7 + 2` to compensate for the reduced row height when - * no logical operator row is shown. - * - * 4. **Multi-expression correction**: cases with 2+ expressions have a compact - * logical operator row. Subtract `multiExprCount * 9` to offset the - * over-estimated spacing. - * - * 5. **Extra expression correction**: for cases with more than 2 expressions, - * each additional expression beyond the second introduces a minor spacing - * discrepancy. Subtract `(extraExprs + 1) * 2` to fine-tune alignment. - * - * @param cases - Array of case objects, each containing an `expressions` array. - * @param caseIndex - The zero-based index of the target case whose port Y is needed. - * @returns The Y-coordinate (in pixels) for the output port of the given case. + * Layout (from node top): + * - 12px padding-top + 24px header + 12px mt-3 = 48px to cases area + * - Each IF/ELIF label row: leading-4 (16px), center at +8px → first port Y = 56.5 + * - Each case: IF/ELIF row (leading-4=16) + mb-1(4) + expressions (gap={2}=2px between) + * - Gap between cases (Flex gap={4}): 4px */ export const getConditionNodeCasePortY = (cases: any[], caseIndex: number) => { - let y = conditionNodePortItemArgsY; - let singleExprCount = 0; - let multiExprCount = 0; - let extraExprs = 0; - let portItemArgsYNum = 0; - + let y = conditionNodePortItemArgsY; // 56.5, center of first IF label 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 - cases[i]?.expressions?.forEach((expr: any) => { - const subs = expr?.sub_variable_condition?.conditions; - if (subs?.length && subs.every(isSubExprSet)) { - casePortItemArgsYNum += subs.length; - } else if (subs?.length) { - casePortItemArgsYNum += 1 - } - }); - portItemArgsYNum += casePortItemArgsYNum; - if (n === 1 && !cases[i]?.expressions?.some((e: any) => e?.sub_variable_condition?.conditions?.length > 0)) { - singleExprCount++ - } else if (n >= 2 || cases[i]?.expressions?.some((e: any) => e?.sub_variable_condition?.conditions?.length > 0)) { - multiExprCount++; - 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 + 2; - } - }); - - console.log('extraExprs notHasSub', notHasSub) - if (notHasSub > 3) { - extraExprs += n - 2 + notHasSub/4; - } else { - extraExprs += n - 2 + notHasSub/4 - } - } + const exprs = cases[i]?.expressions ?? []; + const n = exprs.length; + // IF/ELIF row (16) + mb-1 (4) = 20px base; expressions: sum of heights + 2px gap between + const exprsHeight = n === 0 ? 0 : exprs.reduce((acc: number, e: any) => acc + calcExpressionHeight(e), 0) + 2 * (n - 1); + y += 20 + exprsHeight + 4; // case height + Flex gap between cases } - - console.log('singleExprCount', singleExprCount, 'multiExprCount', multiExprCount, 'extraExprs', extraExprs) - y += portItemArgsY * portItemArgsYNum - // Correction for single-expression cases (slightly shorter rendered height) - if (singleExprCount > 0) y -= singleExprCount * 7 + 2; - // Correction for multi-expression cases (compact logical operator row) - y -= multiExprCount * 9; - // Correction for cases with more than 2 expressions (minor spacing drift) - if (extraExprs > 0) y -= (extraExprs + 1) * 2; - return y; -}; +}; \ No newline at end of file From 559b4bef6b362cd6bdfa82efa60e48d0a11145ba Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 20 Apr 2026 14:47:16 +0800 Subject: [PATCH 05/55] fix(web): add tool_id required check list --- web/src/i18n/en.ts | 1 + web/src/i18n/zh.ts | 1 + .../Workflow/components/Properties/ToolConfig/index.tsx | 5 +++++ web/src/views/Workflow/constant.ts | 5 +++-- 4 files changed, 10 insertions(+), 2 deletions(-) diff --git a/web/src/i18n/en.ts b/web/src/i18n/en.ts index dfc42973..9470e2e9 100644 --- a/web/src/i18n/en.ts +++ b/web/src/i18n/en.ts @@ -2564,6 +2564,7 @@ Memory Bear: After the rebellion, regional warlordism intensified for several re 'jinja-render.template': 'Template', 'document-extractor.file_selector': 'File variable', 'list-operator.input_list': 'Input list', + 'tool.tool_id': 'Tool', }, checkListHasErrors: 'Please resolve all issues in the checklist before publishing', variableSelect: { diff --git a/web/src/i18n/zh.ts b/web/src/i18n/zh.ts index ae0181c9..8a134a06 100644 --- a/web/src/i18n/zh.ts +++ b/web/src/i18n/zh.ts @@ -2528,6 +2528,7 @@ export const zh = { 'jinja-render.template': '模板', 'document-extractor.file_selector': '文件变量', 'list-operator.input_list': '输入变量', + 'tool.tool_id': '工具', }, checkListHasErrors: '发布前确认检查清单中所有问题均已解决', variableSelect: { diff --git a/web/src/views/Workflow/components/Properties/ToolConfig/index.tsx b/web/src/views/Workflow/components/Properties/ToolConfig/index.tsx index ce30ee8f..d38265da 100644 --- a/web/src/views/Workflow/components/Properties/ToolConfig/index.tsx +++ b/web/src/views/Workflow/components/Properties/ToolConfig/index.tsx @@ -147,6 +147,11 @@ const ToolConfig: FC<{ options: Suggestion[]; }> = ({ }; const handleChange: CascaderProps - : log.type === 'search_result' && log.raw_results && typeof log.raw_results !== 'string' + : log.type === 'search_result' && log.result ? -
{log.query}
- {(log.raw_results.reranked_results as AnyObject)?.communities?.length > 0 && <> -
{t('memoryConversation.communities')}
-
    - {((log.raw_results.reranked_results as AnyObject)?.communities as { content: string }[]).map((item, index: number) => ( -
  • {item.content}
    • - ))} -
- } - {(log.raw_results.reranked_results as AnyObject)?.summaries?.length > 0 && <> -
{t('memoryConversation.summaries')}
-
    - {((log.raw_results.reranked_results as AnyObject)?.summaries as { content: string }[]).map((item, index: number) => ( -
  • {item.content}
    • - ))} -
- } + : log.type === 'retrieval_summary' && log.summary ? From 10a655772fa35ba8e722e53dd4594b73c1bad358 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 20 Apr 2026 17:04:00 +0800 Subject: [PATCH 10/55] fix(web): jump list --- .../ApplicationManagement/components/UploadModal.tsx | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/web/src/views/ApplicationManagement/components/UploadModal.tsx b/web/src/views/ApplicationManagement/components/UploadModal.tsx index 4211e72b..623b1b59 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: 2026-04-13 18:17:32 + * @Last Modified time: 2026-04-20 16:52:32 */ /** * UploadModal Component @@ -16,6 +16,7 @@ import { forwardRef, useImperativeHandle, useState, useMemo } from 'react'; import { Form, Steps, Flex, Alert, Button, Result, message } from 'antd'; import { useTranslation } from 'react-i18next'; +import { useNavigate } from 'react-router-dom'; import type { Application, UploadModalRef } from '../types' import RbModal from '@/components/RbModal' @@ -51,6 +52,7 @@ const UploadModal = forwardRef(({ id }, ref) => { const { t } = useTranslation(); + const navigate = useNavigate(); // State management const [visible, setVisible] = useState(false); // Modal visibility @@ -146,6 +148,10 @@ const UploadModal = forwardRef(({ window.open(`/#/application/config/${appId}`, '_blank'); } break; + case 'list': + if (id) { + navigate('/application') + } } }, 100) }; From b03300c804a8f60a5020468399b763bc041a77fb Mon Sep 17 00:00:00 2001 From: wwq Date: Mon, 20 Apr 2026 17:18:05 +0800 Subject: [PATCH 11/55] refactor(rate_limit): refactor API Key rate limiting and remove tenant-level QPS check - Streamline rate limit check flow by removing redundant tenant-level QPS checks. - Restrict checks to API Key QPS and plan degradation protection only. - Update constant naming and error message handling for consistency. --- api/app/core/api_key_auth.py | 3 +- api/app/core/quota_stub.py | 4 +- api/app/services/api_key_service.py | 83 ++++++----------------------- 3 files changed, 18 insertions(+), 72 deletions(-) diff --git a/api/app/core/api_key_auth.py b/api/app/core/api_key_auth.py index b7cacd21..1ded6f81 100644 --- a/api/app/core/api_key_auth.py +++ b/api/app/core/api_key_auth.py @@ -108,8 +108,7 @@ def require_api_key( # 根据错误消息判断限流类型 if "Daily" in error_msg: code = BizCode.API_KEY_DAILY_LIMIT_EXCEEDED - elif "QPS" in error_msg or "ops rate limit" in error_msg: - # "QPS limit exceeded" 和 "API ops rate limit exceeded" 同属 QPS 类 + elif "QPS" in error_msg: code = BizCode.API_KEY_QPS_LIMIT_EXCEEDED else: code = BizCode.API_KEY_QUOTA_EXCEEDED diff --git a/api/app/core/quota_stub.py b/api/app/core/quota_stub.py index 4a8f7cee..248d0875 100644 --- a/api/app/core/quota_stub.py +++ b/api/app/core/quota_stub.py @@ -18,7 +18,7 @@ from app.core.quota_manager import ( get_quota_usage, _check_quota, QuotaUsageRepository, - TENANT_QPS_REDIS_KEY, + API_KEY_QPS_REDIS_KEY, ) __all__ = [ @@ -34,5 +34,5 @@ __all__ = [ "get_quota_usage", "_check_quota", "QuotaUsageRepository", - "TENANT_QPS_REDIS_KEY", + "API_KEY_QPS_REDIS_KEY", ] diff --git a/api/app/services/api_key_service.py b/api/app/services/api_key_service.py index 4fe9d8b5..d710d4ee 100644 --- a/api/app/services/api_key_service.py +++ b/api/app/services/api_key_service.py @@ -282,42 +282,6 @@ class RateLimiterService: def __init__(self): self.redis = aio_redis - async def check_tenant_rate_limit(self, window_id: uuid.UUID, limit: int) -> Tuple[bool, dict]: - """ - 按 window_id（api_key_id）做 1 秒滑动窗口限速。 - 限制值来自套餐配额 api_ops_rate_limit，每个 API Key 独立受此上限约束。 - 只有请求被允许时才计入窗口，超限请求不污染计数。 - """ - now = time.time() - window_start = now - 1 # 1 秒窗口 - key = f"rate_limit:tenant_qps:{window_id}" - - async with self.redis.pipeline() as pipe: - pipe.zremrangebyscore(key, 0, window_start) - pipe.zcard(key) - results = await pipe.execute() - - current = results[1] - - if current >= limit: - return False, { - "limit": limit, - "remaining": 0, - "reset": int(now) + 1, - } - - member = f"{now}:{uuid.uuid4().hex}" - async with self.redis.pipeline() as pipe: - pipe.zadd(key, {member: now}) - pipe.expire(key, 2) - await pipe.execute() - - return True, { - "limit": limit, - "remaining": max(0, limit - current - 1), - "reset": int(now) + 1, - } - async def check_qps(self, api_key_id: uuid.UUID, limit: int) -> Tuple[bool, dict]: """ 检查 API Key 自身 QPS 限制（1 秒滑动窗口）。 @@ -337,6 +301,7 @@ class RateLimiterService: if current >= limit: return False, { "limit": limit, + "current": current, "remaining": 0, "reset": int(now) + 1, } @@ -349,6 +314,7 @@ class RateLimiterService: return True, { "limit": limit, + "current": current + 1, "remaining": max(0, limit - current - 1), "reset": int(now) + 1, } @@ -398,14 +364,9 @@ class RateLimiterService: ) -> Tuple[bool, str, dict]: """ 检查所有限制，按以下顺序： - 1. API Key 自身 QPS（每个 key 独立，不超过其配置的 rate_limit） - 2. 租户套餐 api_ops_rate_limit（每个 key 独立受套餐限额约束） + 1. API Key 自身 QPS（rate_limit 在创建时已保证不超过套餐 api_ops_rate_limit） + 2. 套餐降级保护：若套餐已降级导致 rate_limit > 套餐上限，用套餐上限二次卡控 3. API Key 日调用量 - - 设计语义： - - 租户下 N 个 API Key 可以并发调用，互不影响 - - 每个 API Key 独立受自身 rate_limit 约束 - - 每个 API Key 独立受租户套餐 api_ops_rate_limit 约束（per-key 限额） """ # 1. 检查 API Key 自身 QPS qps_ok, qps_info = await self.check_qps( @@ -419,24 +380,20 @@ class RateLimiterService: "X-RateLimit-Reset": str(qps_info["reset"]) } - # 2. 检查租户套餐 api_ops_rate_limit（运行时流量控制） - # 每个 API Key 的速率都不能超过租户套餐上限，无论 key 自身配置多少 - tenant_info = None + # 2. 套餐降级保护 + # 套餐降级后已有 api_key 的 rate_limit 可能高于新套餐上限 + # 复用第1关已计好的 current，不重复写 Redis if db is not None: try: from app.models.workspace_model import Workspace from app.core.quota_manager import get_api_ops_rate_limit - # 从 Redis 缓存取 tenant_limit，避免每次请求都查数据库 - # TTL 60 秒，套餐变更后最多 1 分钟生效 - # 用 workspace_id 作为滑动窗口维度（一个 workspace 只属于一个 tenant，语义等价） cache_key = f"tenant_api_ops_limit:{api_key.workspace_id}" cached = await self.redis.get(cache_key) if cached is not None: try: tenant_limit = int(cached) if cached != "0" else None except (ValueError, TypeError): - # 缓存数据损坏，回源查数据库 cached = None tenant_limit = None @@ -448,19 +405,14 @@ class RateLimiterService: else: tenant_limit = None - if tenant_limit: - # 用 api_key.id 作为滑动窗口 key，每个 API Key 独立受 api_ops_rate_limit 约束 - tenant_ok, tenant_info = await self.check_tenant_rate_limit( - api_key.id, tenant_limit - ) - if not tenant_ok: - return False, "API ops rate limit exceeded", { - "X-RateLimit-Limit-QPS": str(tenant_limit), - "X-RateLimit-Remaining-QPS": str(tenant_info["remaining"]), - "X-RateLimit-Reset": str(tenant_info["reset"]), - } + if tenant_limit and qps_info["current"] > tenant_limit: + return False, "QPS limit exceeded", { + "X-RateLimit-Limit-QPS": str(tenant_limit), + "X-RateLimit-Remaining-QPS": "0", + "X-RateLimit-Reset": str(qps_info["reset"]), + } except Exception as e: - logger.warning(f"租户 api_ops_rate_limit 运行时检查失败，跳过: {e}") + logger.warning(f"套餐降级保护检查失败，跳过: {e}") # 3. 检查日调用量 daily_ok, daily_info = await self.check_daily_requests( @@ -474,18 +426,13 @@ class RateLimiterService: "X-RateLimit-Reset": str(daily_info["reset"]) } - headers = { + return True, "", { "X-RateLimit-Limit-QPS": str(qps_info["limit"]), "X-RateLimit-Remaining-QPS": str(qps_info["remaining"]), "X-RateLimit-Limit-Day": str(daily_info["limit"]), "X-RateLimit-Remaining-Day": str(daily_info["remaining"]), "X-RateLimit-Reset": str(daily_info["reset"]), } - # 如果租户限速信息存在，补充到响应头 - if tenant_info: - headers["X-RateLimit-Limit-Tenant-QPS"] = str(tenant_info["limit"]) - headers["X-RateLimit-Remaining-Tenant-QPS"] = str(tenant_info["remaining"]) - return True, "", headers class ApiKeyAuthService: From 08b5c7bc8a09ac36fc20427cf795ece38dcdd99d Mon Sep 17 00:00:00 2001 From: wwq Date: Mon, 20 Apr 2026 17:46:05 +0800 Subject: [PATCH 12/55] =?UTF-8?q?perf(=E9=99=90=E6=B5=81=E6=9C=8D=E5=8A=A1?= =?UTF-8?q?):=20=E4=BC=98=E5=8C=96Redis=E6=9F=A5=E8=AF=A2=E4=BB=A5?= =?UTF-8?q?=E5=87=8F=E5=B0=91=E5=91=BD=E4=BB=A4=E6=95=B0=E9=87=8F?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit 使用zcount替代zremrangebyscore和zcard组合查询，减少一次Redis操作 --- api/app/services/api_key_service.py | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/api/app/services/api_key_service.py b/api/app/services/api_key_service.py index d710d4ee..c1bbbdc8 100644 --- a/api/app/services/api_key_service.py +++ b/api/app/services/api_key_service.py @@ -292,11 +292,10 @@ class RateLimiterService: key = f"rate_limit:qps:{api_key_id}" async with self.redis.pipeline() as pipe: - pipe.zremrangebyscore(key, 0, window_start) - pipe.zcard(key) + pipe.zcount(key, window_start, "+inf") results = await pipe.execute() - current = results[1] + current = results[0] if current >= limit: return False, { From 3227c25b07a805caab6d8c4c5fc71ddd7278e33c Mon Sep 17 00:00:00 2001 From: wwq Date: Mon, 20 Apr 2026 18:10:28 +0800 Subject: [PATCH 13/55] fix(quota): fix tenant ID retrieval and QPS counting logic - Fix issue where tenant ID lookup from shared records failed to query the workspace correctly. - Switch QPS counting from sliding window to simple counter to improve performance and simplify logic. - Remove unnecessary `time` module import. --- api/app/core/quota_manager.py | 9 ++-- api/app/services/api_key_service.py | 79 +++++++++++------------------ 2 files changed, 34 insertions(+), 54 deletions(-) diff --git a/api/app/core/quota_manager.py b/api/app/core/quota_manager.py index 28130dee..bf04059e 100644 --- a/api/app/core/quota_manager.py +++ b/api/app/core/quota_manager.py @@ -6,7 +6,6 @@ 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 @@ -68,7 +67,9 @@ def _get_tenant_id_from_kwargs(db: Session, kwargs: dict): if share_record: app = db.query(App).filter(App.id == share_record.app_id, App.is_active.is_(True)).first() if app: - return app.workspace.tenant_id + workspace = db.query(Workspace).filter(Workspace.id == app.workspace_id).first() + if workspace: + return workspace.tenant_id return None @@ -597,7 +598,6 @@ async def get_quota_usage(db: Session, tenant_id: UUID) -> dict: from app.aioRedis import aio_redis as _aio_redis from app.models.api_key_model import ApiKey from app.models.workspace_model import Workspace - _now = time.time() # 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( @@ -608,7 +608,8 @@ async def get_quota_usage(db: Session, tenant_id: UUID) -> dict: ).all() for (key_id,) in api_key_ids: _rk = API_KEY_QPS_REDIS_KEY.format(api_key_id=key_id) - count = int(await _aio_redis.zcount(_rk, _now - 1, "+inf") or 0) + val = await _aio_redis.get(_rk) + count = int(val) if val else 0 if count > api_ops_current: api_ops_current = count except Exception as e: diff --git a/api/app/services/api_key_service.py b/api/app/services/api_key_service.py index c1bbbdc8..e67d623e 100644 --- a/api/app/services/api_key_service.py +++ b/api/app/services/api_key_service.py @@ -283,39 +283,27 @@ class RateLimiterService: self.redis = aio_redis async def check_qps(self, api_key_id: uuid.UUID, limit: int) -> Tuple[bool, dict]: + """检查QPS限制 + + Returns: + (is_allowed, rate_limit_info) """ - 检查 API Key 自身 QPS 限制（1 秒滑动窗口）。 - 只有请求被允许时才计入窗口，超限请求不污染计数。 - """ - now = time.time() - window_start = now - 1 key = f"rate_limit:qps:{api_key_id}" async with self.redis.pipeline() as pipe: - pipe.zcount(key, window_start, "+inf") + pipe.incr(key) + pipe.expire(key, 1, nx=True) # 1 秒过期 results = await pipe.execute() current = results[0] + remaining = max(0, limit - current) + reset_time = int(time.time()) + 1 - if current >= limit: - return False, { - "limit": limit, - "current": current, - "remaining": 0, - "reset": int(now) + 1, - } - - member = f"{now}:{uuid.uuid4().hex}" - async with self.redis.pipeline() as pipe: - pipe.zadd(key, {member: now}) - pipe.expire(key, 2) - await pipe.execute() - - return True, { + return current <= limit, { "limit": limit, - "current": current + 1, - "remaining": max(0, limit - current - 1), - "reset": int(now) + 1, + "current": current, + "remaining": remaining, + "reset": reset_time, } async def check_daily_requests( @@ -363,25 +351,11 @@ class RateLimiterService: ) -> Tuple[bool, str, dict]: """ 检查所有限制，按以下顺序： - 1. API Key 自身 QPS（rate_limit 在创建时已保证不超过套餐 api_ops_rate_limit） - 2. 套餐降级保护：若套餐已降级导致 rate_limit > 套餐上限，用套餐上限二次卡控 - 3. API Key 日调用量 + 1. API Key QPS：取 api_key.rate_limit 与套餐 api_ops_rate_limit 的最小值作为限额 + 2. API Key 日调用量 """ - # 1. 检查 API Key 自身 QPS - qps_ok, qps_info = await self.check_qps( - api_key.id, - api_key.rate_limit - ) - if not qps_ok: - return False, "QPS limit exceeded", { - "X-RateLimit-Limit-QPS": str(qps_info["limit"]), - "X-RateLimit-Remaining-QPS": str(qps_info["remaining"]), - "X-RateLimit-Reset": str(qps_info["reset"]) - } - - # 2. 套餐降级保护 - # 套餐降级后已有 api_key 的 rate_limit 可能高于新套餐上限 - # 复用第1关已计好的 current，不重复写 Redis + # 1. 取套餐限额与 api_key 自身限额的最小值 + effective_limit = api_key.rate_limit if db is not None: try: from app.models.workspace_model import Workspace @@ -404,16 +378,21 @@ class RateLimiterService: else: tenant_limit = None - if tenant_limit and qps_info["current"] > tenant_limit: - return False, "QPS limit exceeded", { - "X-RateLimit-Limit-QPS": str(tenant_limit), - "X-RateLimit-Remaining-QPS": "0", - "X-RateLimit-Reset": str(qps_info["reset"]), - } + if tenant_limit: + effective_limit = min(api_key.rate_limit, tenant_limit) except Exception as e: - logger.warning(f"套餐降级保护检查失败，跳过: {e}") + logger.warning(f"获取套餐限额失败，使用 api_key 自身限额: {e}") - # 3. 检查日调用量 + # 用最终有效限额做 QPS 检查 + qps_ok, qps_info = await self.check_qps(api_key.id, effective_limit) + if not qps_ok: + return False, "QPS limit exceeded", { + "X-RateLimit-Limit-QPS": str(qps_info["limit"]), + "X-RateLimit-Remaining-QPS": str(qps_info["remaining"]), + "X-RateLimit-Reset": str(qps_info["reset"]) + } + + # 2. 检查日调用量 daily_ok, daily_info = await self.check_daily_requests( api_key.id, api_key.daily_request_limit From d59990d3260652a88d55773b47ae2e26e8a713cf Mon Sep 17 00:00:00 2001 From: wwq Date: Mon, 20 Apr 2026 18:25:39 +0800 Subject: [PATCH 14/55] fix(rate_limit): differentiate between tenant plan and API Key QPS limit errors - Add logic to detect tenant plan QPS limits and return a specific error message when triggered. - Simplify boolean check in model activation quota validation. --- api/app/core/api_key_auth.py | 2 ++ api/app/core/quota_manager.py | 4 ++-- api/app/services/api_key_service.py | 7 ++++++- 3 files changed, 10 insertions(+), 3 deletions(-) diff --git a/api/app/core/api_key_auth.py b/api/app/core/api_key_auth.py index 1ded6f81..05bca945 100644 --- a/api/app/core/api_key_auth.py +++ b/api/app/core/api_key_auth.py @@ -108,6 +108,8 @@ def require_api_key( # 根据错误消息判断限流类型 if "Daily" in error_msg: code = BizCode.API_KEY_DAILY_LIMIT_EXCEEDED + elif "Tenant" in error_msg: + code = BizCode.API_KEY_QPS_LIMIT_EXCEEDED # 租户套餐速率超限，同属 QPS 类 elif "QPS" in error_msg: code = BizCode.API_KEY_QPS_LIMIT_EXCEEDED else: diff --git a/api/app/core/quota_manager.py b/api/app/core/quota_manager.py index bf04059e..534e1940 100644 --- a/api/app/core/quota_manager.py +++ b/api/app/core/quota_manager.py @@ -488,7 +488,7 @@ def check_model_activation_quota(func: Callable) -> Callable: logger.warning("模型激活配额检查失败：缺少 model_id 或 model_data 参数") return await func(*args, **kwargs) - if model_data.is_active is True: + if model_data.is_active: try: from app.services.model_service import ModelConfigService @@ -522,7 +522,7 @@ def check_model_activation_quota(func: Callable) -> Callable: 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: + if model_data.is_active: try: from app.services.model_service import ModelConfigService diff --git a/api/app/services/api_key_service.py b/api/app/services/api_key_service.py index e67d623e..5595e93f 100644 --- a/api/app/services/api_key_service.py +++ b/api/app/services/api_key_service.py @@ -386,7 +386,12 @@ class RateLimiterService: # 用最终有效限额做 QPS 检查 qps_ok, qps_info = await self.check_qps(api_key.id, effective_limit) if not qps_ok: - return False, "QPS limit exceeded", { + # 判断是套餐限额触发还是 api_key 自身限额触发 + if tenant_limit and effective_limit == tenant_limit and api_key.rate_limit > tenant_limit: + error_msg = "Tenant QPS limit exceeded" + else: + error_msg = "QPS limit exceeded" + return False, error_msg, { "X-RateLimit-Limit-QPS": str(qps_info["limit"]), "X-RateLimit-Remaining-QPS": str(qps_info["remaining"]), "X-RateLimit-Reset": str(qps_info["reset"]) From 94d2d975ee21ac2c3dc2fe8932a0b58a357e6809 Mon Sep 17 00:00:00 2001 From: lanceyq <1982376970@qq.com> Date: Mon, 20 Apr 2026 18:26:20 +0800 Subject: [PATCH 15/55] fix(memory): use end_user.workspace_id instead of app.workspace_id in log message Corrected variable reference in get_end_user_connected_config log statement. The previous code referenced app.workspace_id which could be incorrect or undefined in this context. --- api/app/services/memory_agent_service.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/api/app/services/memory_agent_service.py b/api/app/services/memory_agent_service.py index b12bb48a..335a0f8b 100644 --- a/api/app/services/memory_agent_service.py +++ b/api/app/services/memory_agent_service.py @@ -1280,7 +1280,7 @@ def get_end_user_connected_config(end_user_id: str, db: Session) -> Dict[str, An } logger.info( - f"Successfully retrieved connected config: memory_config_id={memory_config_id}, workspace_id={app.workspace_id}") + f"Successfully retrieved connected config: memory_config_id={memory_config_id}, workspace_id={end_user.workspace_id}") return result From 817aa78d0394d062e10423e0910acd063f3b524d Mon Sep 17 00:00:00 2001 From: wwq Date: Mon, 20 Apr 2026 18:34:18 +0800 Subject: [PATCH 16/55] fix(rate_limit): differentiate between tenant plan and API Key QPS limit errors - Add logic to detect tenant plan QPS limits and return a specific error message when triggered. - Simplify boolean check in model activation quota validation. --- api/app/services/api_key_service.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/api/app/services/api_key_service.py b/api/app/services/api_key_service.py index 5595e93f..53aad5ce 100644 --- a/api/app/services/api_key_service.py +++ b/api/app/services/api_key_service.py @@ -388,7 +388,7 @@ class RateLimiterService: if not qps_ok: # 判断是套餐限额触发还是 api_key 自身限额触发 if tenant_limit and effective_limit == tenant_limit and api_key.rate_limit > tenant_limit: - error_msg = "Tenant QPS limit exceeded" + error_msg = "Tenant limit exceeded" else: error_msg = "QPS limit exceeded" return False, error_msg, { From 2c2ed0ebf3703481a7bdd8a0d975ff60c988c7a0 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Mon, 20 Apr 2026 18:39:59 +0800 Subject: [PATCH 17/55] fix(web): check list add vision_input --- web/src/components/ModelSelect/index.tsx | 4 ++-- web/src/i18n/en.ts | 1 + web/src/i18n/zh.ts | 1 + .../views/Workflow/components/CheckList/index.tsx | 12 ++++++++++++ 4 files changed, 16 insertions(+), 2 deletions(-) diff --git a/web/src/components/ModelSelect/index.tsx b/web/src/components/ModelSelect/index.tsx index 85977376..4c59c87b 100644 --- a/web/src/components/ModelSelect/index.tsx +++ b/web/src/components/ModelSelect/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-03-07 16:49:59 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-17 10:11:54 + * @Last Modified time: 2026-04-20 18:14:34 */ import { type FC, useEffect, useState } from 'react'; import { Select, Flex, Space } from 'antd'; @@ -56,7 +56,7 @@ const ModelSelect: FC = ({ params, placeholder, fontClassName, useEffect(() => { if (updateOptions) updateOptions([...options, ...initialData]); - }, [options, initialData]) + }, [JSON.stringify(options), JSON.stringify(initialData)]) return ( - + + + + } {['llm', 'chat'].includes(modelType as string) && From 30aed72b7430629c03cb43a7d096885b88a83176 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Tue, 21 Apr 2026 15:48:24 +0800 Subject: [PATCH 28/55] fix(web): package support unlimited --- .../components/SiderMenu/SubscriptionDetailModal.tsx | 11 +++++------ web/src/components/SiderMenu/index.tsx | 4 ++-- web/src/i18n/en.ts | 1 + web/src/i18n/zh.ts | 1 + web/src/views/Package/constant.ts | 4 ++-- web/src/views/Package/index.tsx | 11 +++++------ 6 files changed, 16 insertions(+), 16 deletions(-) diff --git a/web/src/components/SiderMenu/SubscriptionDetailModal.tsx b/web/src/components/SiderMenu/SubscriptionDetailModal.tsx index fbb5030c..699328e2 100644 --- a/web/src/components/SiderMenu/SubscriptionDetailModal.tsx +++ b/web/src/components/SiderMenu/SubscriptionDetailModal.tsx @@ -2,11 +2,11 @@ * @Author: ZhaoYing * @Date: 2026-04-14 12:28:23 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-16 17:34:02 + * @Last Modified time: 2026-04-21 15:46:35 */ import { useState, forwardRef, useImperativeHandle } from 'react'; -import { Flex, Tooltip, Divider } from 'antd'; +import { Flex, Divider } from 'antd'; import { useTranslation } from 'react-i18next'; import clsx from 'clsx'; @@ -82,8 +82,7 @@ const SubscriptionDetailModal = forwardRef((_props, {/* Features */} {billingUnits.map(({ key, unit, icon }) => { - const value = detail?.quotas[key as keyof Subscription['quotas']]; - if (value === undefined || value === null) return null; + const value = detail?.quotas?.[key as keyof Subscription['quotas']]; return ( ((_props, /> ) })} - {detail?.package_plan?.tech_support && ( + {detail?.package_plan?.tech_support && detail?.package_plan?.[getKeyWithLanguage('tech_support')] && ( ((_props, theme_color={detail?.package_plan?.theme_color} /> )} - {detail?.package_plan?.sla_compliance && ( + {detail?.package_plan?.sla_compliance && detail?.package_plan?.[getKeyWithLanguage('sla_compliance')] && ( {['workspace_quota', 'skill_quota', 'app_quota', 'model_quota'].map(key => (
-
{subscription.quotas?.[key as keyof typeof subscription.quotas]}
+
{subscription.quotas?.[key as keyof typeof subscription.quotas] ?? t('package.noLimit')}
{t(`index.${key}`)}
))} diff --git a/web/src/i18n/en.ts b/web/src/i18n/en.ts index 64b600ba..13634544 100644 --- a/web/src/i18n/en.ts +++ b/web/src/i18n/en.ts @@ -3093,6 +3093,7 @@ Memory Bear: After the rebellion, regional warlordism intensified for several re editPackage: 'Edit Package', viewDetail: 'View full package details', + noLimit: 'Unlimited', }, }, }; diff --git a/web/src/i18n/zh.ts b/web/src/i18n/zh.ts index cabb1eca..64689bc2 100644 --- a/web/src/i18n/zh.ts +++ b/web/src/i18n/zh.ts @@ -3057,6 +3057,7 @@ export const zh = { editPackage: '编辑套餐', viewDetail: '查看完整套餐详情', + noLimit: '不限制', }, }, } \ No newline at end of file diff --git a/web/src/views/Package/constant.ts b/web/src/views/Package/constant.ts index 7fc69969..168b65f8 100644 --- a/web/src/views/Package/constant.ts +++ b/web/src/views/Package/constant.ts @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-04-14 11:43:57 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-14 11:44:40 + * @Last Modified time: 2026-04-21 15:44:13 */ export const billingUnits = [ { @@ -42,7 +42,7 @@ export const billingUnits = [ }, { key: 'model_quota', - unit: 'ops', placeholder: 'numberPlaceholder', + unit: 'pcs', placeholder: 'numberPlaceholder', icon: 'model', }, { diff --git a/web/src/views/Package/index.tsx b/web/src/views/Package/index.tsx index b3aed9d5..ea9f1ec5 100644 --- a/web/src/views/Package/index.tsx +++ b/web/src/views/Package/index.tsx @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-04-14 11:34:42 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-04-16 17:23:49 + * @Last Modified time: 2026-04-21 15:45:30 */ /** * Package Component @@ -60,7 +60,7 @@ const btnClassNames = { default: 'rb:h-10! rb:rounded-[8px]! rb:bg-[#212332]! rb:text-white! rb:border-0! rb:hover:border-0! rb:hover:opacity-[0.8]', } -export const UnitWrapper = ({ titleKey, value, icon, unit, theme_color = '#171719' }: { titleKey: string; value: number | string; icon: string; unit?: string; theme_color?: string; }) => { +export const UnitWrapper = ({ titleKey, value, icon, unit, theme_color = '#171719' }: { titleKey: string; value?: number | string | null; icon: string; unit?: string; theme_color?: string; }) => { const { t } = useTranslation(); const renderFeatureIcon = (iconKey: string, color: string) => { @@ -78,7 +78,7 @@ export const UnitWrapper = ({ titleKey, value, icon, unit, theme_color = '#17171 >{renderFeatureIcon(icon, theme_color)}
{t(`package.${titleKey}`)}
-
{value} {unit ? t(`package.${unit}`) : ''}
+ {value ?
{value} {unit ? t(`package.${unit}`) : ''}
:
{t('package.noLimit')}
}
) @@ -252,7 +252,6 @@ const Package: FC = () => { > {billingUnits.map(({ key, unit, icon }) => { const value = pkg?.quotas?.[key as keyof Package['quotas']]; - if (value === undefined || value === null) return null; return ( { /> ) })} - {pkg.tech_support && ( + {pkg.tech_support && pkg[getKeyWithLanguage('tech_support')] && ( { theme_color={pkg.theme_color} /> )} - {pkg.sla_compliance && ( + {pkg.sla_compliance && pkg[getKeyWithLanguage('sla_compliance')] && ( Date: Tue, 21 Apr 2026 16:45:14 +0800 Subject: [PATCH 29/55] =?UTF-8?q?refactor(knowledge=5Fservice):=20?= =?UTF-8?q?=E4=BC=98=E5=8C=96=E6=A8=A1=E5=9E=8B=E7=BB=91=E5=AE=9A=E9=80=BB?= =?UTF-8?q?=E8=BE=91=EF=BC=8C=E4=BD=BF=E7=94=A8ID=E6=9F=A5=E8=AF=A2?= =?UTF-8?q?=E5=B9=B6=E7=AE=80=E5=8C=96=E5=9B=9E=E9=80=80=E6=9C=BA=E5=88=B6?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit 将模型绑定逻辑从按名称查询改为按ID查询，提高准确性 简化回退机制，直接查询租户下最新创建的模型 统一处理图像转文本模型的查询方式 --- api/app/services/knowledge_service.py | 45 ++++++++++++++------------- 1 file changed, 23 insertions(+), 22 deletions(-) diff --git a/api/app/services/knowledge_service.py b/api/app/services/knowledge_service.py index b1d0d77b..56b630bb 100644 --- a/api/app/services/knowledge_service.py +++ b/api/app/services/knowledge_service.py @@ -7,7 +7,6 @@ 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 business_logger = get_business_logger() @@ -77,53 +76,55 @@ def create_knowledge( tenant_id = workspace.tenant_id - def _get_model_by_name_or_fallback(model_name: str | None, model_types: list, label: str): - """优先按 workspace 指定的 model name 查，找不到再 fallback 到 tenant 下第一个""" - if model_name: + def _get_model_by_id_or_fallback(model_id: str | None, model_types: list, label: str): + """优先按 workspace 绑定的 model_config id 查，找不到再 fallback 到 tenant 下最新创建的一个""" + if model_id: model = db.query(ModelConfig).filter( + ModelConfig.id == model_id, ModelConfig.tenant_id == tenant_id, - ModelConfig.name == model_name, - ModelConfig.type.in_([t.value for t in model_types]), ModelConfig.is_active == True, ModelConfig.is_composite == False ).first() if model: - business_logger.debug(f"Auto-bind {label} model from workspace default: {model.id} ({model_name})") + business_logger.debug(f"Auto-bind {label} model from workspace default: {model.id}") return model - business_logger.debug(f"Workspace default {label} model '{model_name}' not found, falling back to tenant") - models = ModelConfigRepository.get_by_type(db=db, model_types=model_types, tenant_id=tenant_id, is_active=True) - if models: - business_logger.debug(f"Auto-bind {label} model from tenant fallback: {models[0].id}") - return models[0] - return None + business_logger.debug(f"Workspace default {label} model id '{model_id}' not found, falling back to tenant latest") + model = db.query(ModelConfig).filter( + ModelConfig.tenant_id == tenant_id, + ModelConfig.type.in_([t.value for t in model_types]), + ModelConfig.is_active == True, + ModelConfig.is_composite == False + ).order_by(ModelConfig.created_at.desc()).first() + if model: + business_logger.debug(f"Auto-bind {label} model from tenant fallback (latest): {model.id}") + return model if not knowledge.embedding_id: - model = _get_model_by_name_or_fallback(workspace.embedding, [ModelType.EMBEDDING], "embedding") + model = _get_model_by_id_or_fallback(workspace.embedding, [ModelType.EMBEDDING], "embedding") if model: knowledge.embedding_id = model.id if not knowledge.reranker_id: - model = _get_model_by_name_or_fallback(workspace.rerank, [ModelType.RERANK], "rerank") + model = _get_model_by_id_or_fallback(workspace.rerank, [ModelType.RERANK], "rerank") if model: knowledge.reranker_id = model.id if not knowledge.llm_id: - model = _get_model_by_name_or_fallback(workspace.llm, [ModelType.LLM, ModelType.CHAT], "llm") + model = _get_model_by_id_or_fallback(workspace.llm, [ModelType.LLM, ModelType.CHAT], "llm") if model: knowledge.llm_id = model.id if not knowledge.image2text_id: - image2text_models = db.query(ModelConfig).filter( + model = db.query(ModelConfig).filter( ModelConfig.tenant_id == tenant_id, ModelConfig.type.in_([ModelType.CHAT.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}") + ).order_by(ModelConfig.created_at.desc()).first() + if model: + knowledge.image2text_id = model.id + business_logger.debug(f"Auto-bind image2text model: {model.id}") business_logger.debug(f"Start creating the knowledge base: {knowledge.name}") db_knowledge = knowledge_repository.create_knowledge( From eb98a69a845879721c92d917d6bae65b26ad1516 Mon Sep 17 00:00:00 2001 From: wwq Date: Tue, 21 Apr 2026 16:50:43 +0800 Subject: [PATCH 30/55] =?UTF-8?q?fix(=E7=9F=A5=E8=AF=86=E6=9C=8D=E5=8A=A1)?= =?UTF-8?q?:=20=E4=BF=AE=E5=A4=8D=E5=88=9B=E5=BB=BA=E7=9F=A5=E8=AF=86?= =?UTF-8?q?=E5=BA=93=E6=97=B6=E6=9C=AA=E6=A3=80=E6=9F=A5=E8=A7=86=E8=A7=89?= =?UTF-8?q?=E6=A8=A1=E5=9E=8B=E5=AD=98=E5=9C=A8=E7=9A=84=E9=94=99=E8=AF=AF?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit 当租户下没有可用的视觉模型时，抛出明确异常提示 --- api/app/services/knowledge_service.py | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/api/app/services/knowledge_service.py b/api/app/services/knowledge_service.py index 56b630bb..8937dd1e 100644 --- a/api/app/services/knowledge_service.py +++ b/api/app/services/knowledge_service.py @@ -122,9 +122,10 @@ def create_knowledge( ModelConfig.is_active == True, ModelConfig.is_composite == False ).order_by(ModelConfig.created_at.desc()).first() - if model: - knowledge.image2text_id = model.id - business_logger.debug(f"Auto-bind image2text model: {model.id}") + if not model: + raise Exception("租户下没有可用的视觉模型，创建知识库失败") + knowledge.image2text_id = model.id + business_logger.debug(f"Auto-bind image2text model: {model.id}") business_logger.debug(f"Start creating the knowledge base: {knowledge.name}") db_knowledge = knowledge_repository.create_knowledge( From 402c8aef5d745cdf661df69ed105886bf5913a46 Mon Sep 17 00:00:00 2001 From: wwq Date: Tue, 21 Apr 2026 17:04:42 +0800 Subject: [PATCH 31/55] =?UTF-8?q?refactor(knowledge=5Fservice):=20?= =?UTF-8?q?=E7=AE=80=E5=8C=96=E6=A8=A1=E5=9E=8B=E7=BB=91=E5=AE=9A=E9=80=BB?= =?UTF-8?q?=E8=BE=91=EF=BC=8C=E7=9B=B4=E6=8E=A5=E4=BD=BF=E7=94=A8=E5=B7=A5?= =?UTF-8?q?=E4=BD=9C=E5=8C=BA=E9=85=8D=E7=BD=AE?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit 移除_get_model_by_id_or_fallback方法，直接使用工作区配置的模型ID 对于image2text模型，放宽类型限制并移除composite检查 --- api/app/services/knowledge_service.py | 38 +++------------------------ 1 file changed, 4 insertions(+), 34 deletions(-) diff --git a/api/app/services/knowledge_service.py b/api/app/services/knowledge_service.py index 8937dd1e..7f141b76 100644 --- a/api/app/services/knowledge_service.py +++ b/api/app/services/knowledge_service.py @@ -76,51 +76,21 @@ def create_knowledge( tenant_id = workspace.tenant_id - def _get_model_by_id_or_fallback(model_id: str | None, model_types: list, label: str): - """优先按 workspace 绑定的 model_config id 查，找不到再 fallback 到 tenant 下最新创建的一个""" - if model_id: - model = db.query(ModelConfig).filter( - ModelConfig.id == model_id, - ModelConfig.tenant_id == tenant_id, - ModelConfig.is_active == True, - ModelConfig.is_composite == False - ).first() - if model: - business_logger.debug(f"Auto-bind {label} model from workspace default: {model.id}") - return model - business_logger.debug(f"Workspace default {label} model id '{model_id}' not found, falling back to tenant latest") - model = db.query(ModelConfig).filter( - ModelConfig.tenant_id == tenant_id, - ModelConfig.type.in_([t.value for t in model_types]), - ModelConfig.is_active == True, - ModelConfig.is_composite == False - ).order_by(ModelConfig.created_at.desc()).first() - if model: - business_logger.debug(f"Auto-bind {label} model from tenant fallback (latest): {model.id}") - return model - if not knowledge.embedding_id: - model = _get_model_by_id_or_fallback(workspace.embedding, [ModelType.EMBEDDING], "embedding") - if model: - knowledge.embedding_id = model.id + knowledge.embedding_id = workspace.embedding if not knowledge.reranker_id: - model = _get_model_by_id_or_fallback(workspace.rerank, [ModelType.RERANK], "rerank") - if model: - knowledge.reranker_id = model.id + knowledge.reranker_id = workspace.rerank if not knowledge.llm_id: - model = _get_model_by_id_or_fallback(workspace.llm, [ModelType.LLM, ModelType.CHAT], "llm") - if model: - knowledge.llm_id = model.id + knowledge.llm_id = workspace.llm if not knowledge.image2text_id: model = db.query(ModelConfig).filter( ModelConfig.tenant_id == tenant_id, - ModelConfig.type.in_([ModelType.CHAT.value]), + ModelConfig.type.in_([ModelType.CHAT.value, ModelType.LLM.value]), ModelConfig.capability.contains(["vision"]), ModelConfig.is_active == True, - ModelConfig.is_composite == False ).order_by(ModelConfig.created_at.desc()).first() if not model: raise Exception("租户下没有可用的视觉模型，创建知识库失败") From cb2a7aa60a6c805ccf2e88b24816bd1f645f5762 Mon Sep 17 00:00:00 2001 From: wwq Date: Tue, 21 Apr 2026 17:18:11 +0800 Subject: [PATCH 32/55] =?UTF-8?q?fix(=E7=9F=A5=E8=AF=86=E6=9C=8D=E5=8A=A1)?= =?UTF-8?q?:=20=E6=B7=BB=E5=8A=A0=E5=B7=A5=E4=BD=9C=E7=A9=BA=E9=97=B4?= =?UTF-8?q?=E6=A8=A1=E5=9E=8B=E9=85=8D=E7=BD=AE=E7=9A=84=E6=A0=A1=E9=AA=8C?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit 在创建知识时检查工作空间是否配置了必要的模型，未配置时抛出异常提示用户 --- api/app/services/knowledge_service.py | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/api/app/services/knowledge_service.py b/api/app/services/knowledge_service.py index 7f141b76..20757307 100644 --- a/api/app/services/knowledge_service.py +++ b/api/app/services/knowledge_service.py @@ -77,12 +77,18 @@ def create_knowledge( tenant_id = workspace.tenant_id if not knowledge.embedding_id: + if not workspace.embedding: + raise Exception("工作空间未配置 Embedding 模型，请先完善工作空间配置后重试") knowledge.embedding_id = workspace.embedding if not knowledge.reranker_id: + if not workspace.rerank: + raise Exception("工作空间未配置 Rerank 模型，请先完善工作空间配置后重试") knowledge.reranker_id = workspace.rerank if not knowledge.llm_id: + if not workspace.llm: + raise Exception("工作空间未配置 LLM 模型，请先完善工作空间配置后重试") knowledge.llm_id = workspace.llm if not knowledge.image2text_id: From ea0e65f1ecfccac7532e75f0bc56e6358893dfc9 Mon Sep 17 00:00:00 2001 From: Mark <348207283@qq.com> Date: Tue, 21 Apr 2026 17:29:35 +0800 Subject: [PATCH 33/55] [modify] fix tasks --- api/app/celery_worker.py | 17 +++++++++++++++++ api/app/tasks.py | 10 ++++++++++ 2 files changed, 27 insertions(+) diff --git a/api/app/celery_worker.py b/api/app/celery_worker.py index 4ea4fee1..e6c041f5 100644 --- a/api/app/celery_worker.py +++ b/api/app/celery_worker.py @@ -2,6 +2,8 @@ Celery Worker 入口点 用于启动 Celery Worker: celery -A app.celery_worker worker --loglevel=info """ +from celery.signals import worker_process_init + from app.celery_app import celery_app from app.core.logging_config import LoggingConfig, get_logger @@ -13,4 +15,19 @@ logger.info("Celery worker logging initialized") # 导入任务模块以注册任务 import app.tasks + +@worker_process_init.connect +def _reinit_db_pool(**kwargs): + """ + prefork 子进程启动时重建 SQLAlchemy 连接池。 + + fork() 后子进程继承了父进程的连接池和底层 TCP socket， + 多个子进程共享同一个 socket 会导致 PostgreSQL 连接损坏。 + dispose() 会关闭继承来的连接，后续操作会自动创建新连接。 + """ + from app.db import engine + engine.dispose() + logger.info("DB connection pool disposed for forked worker process") + + __all__ = ['celery_app'] diff --git a/api/app/tasks.py b/api/app/tasks.py index 8bbbdc6e..87bcaec9 100644 --- a/api/app/tasks.py +++ b/api/app/tasks.py @@ -251,8 +251,18 @@ def parse_document(file_path: str, document_id: uuid.UUID): # Prepare vision_model for parsing vision_model = _build_vision_model(file_path, db_knowledge) + # 先将文件读入内存，避免解析过程中依赖 NFS 文件持续可访问 + # python-docx 等库在 binary=None 时会用路径直接打开文件， + # 在 NFS/共享存储上可能因缓存失效导致 "Package not found" + try: + with open(file_path, "rb") as f: + file_binary = f.read() + except FileNotFoundError: + raise FileNotFoundError(f"File not found at '{file_path}'") + from app.core.rag.app.naive import chunk res = chunk(filename=file_path, + binary=file_binary, from_page=0, to_page=DEFAULT_PARSE_TO_PAGE, callback=progress_callback, From 2b5bece9b6162d75e3746dab04decdd760522319 Mon Sep 17 00:00:00 2001 From: Mark <348207283@qq.com> Date: Tue, 21 Apr 2026 17:34:03 +0800 Subject: [PATCH 34/55] [modify] nfs read error --- api/app/tasks.py | 29 ++++++++++++++++++++++++----- 1 file changed, 24 insertions(+), 5 deletions(-) diff --git a/api/app/tasks.py b/api/app/tasks.py index 87bcaec9..e965a281 100644 --- a/api/app/tasks.py +++ b/api/app/tasks.py @@ -254,11 +254,30 @@ def parse_document(file_path: str, document_id: uuid.UUID): # 先将文件读入内存，避免解析过程中依赖 NFS 文件持续可访问 # python-docx 等库在 binary=None 时会用路径直接打开文件， # 在 NFS/共享存储上可能因缓存失效导致 "Package not found" - try: - with open(file_path, "rb") as f: - file_binary = f.read() - except FileNotFoundError: - raise FileNotFoundError(f"File not found at '{file_path}'") + max_wait_seconds = 30 + wait_interval = 2 + waited = 0 + file_binary = None + while waited <= max_wait_seconds: + # os.listdir 强制 NFS 客户端刷新目录缓存 + parent_dir = os.path.dirname(file_path) + try: + os.listdir(parent_dir) + except OSError: + pass + try: + with open(file_path, "rb") as f: + file_binary = f.read() + break + except FileNotFoundError: + if waited >= max_wait_seconds: + raise FileNotFoundError( + f"File not found at '{file_path}' after waiting {max_wait_seconds}s " + f"(NFS cache may be stale)" + ) + logger.warning(f"File not visible yet on this node, retrying in {wait_interval}s: {file_path}") + time.sleep(wait_interval) + waited += wait_interval from app.core.rag.app.naive import chunk res = chunk(filename=file_path, From 4c9f327833742ff3f6ecf98d173e4917bd070116 Mon Sep 17 00:00:00 2001 From: wwq Date: Tue, 21 Apr 2026 18:15:31 +0800 Subject: [PATCH 35/55] feat(quota): add quota checks during app duplication and import operations - Integrate quota check decorators into app duplication, workflow import save, and app import actions. - Explicitly validate application quotas for new app imports. --- api/app/controllers/app_controller.py | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/api/app/controllers/app_controller.py b/api/app/controllers/app_controller.py index 3d97f2a2..f3cbe5ea 100644 --- a/api/app/controllers/app_controller.py +++ b/api/app/controllers/app_controller.py @@ -219,6 +219,7 @@ def delete_app( @router.post("/{app_id}/copy", summary="复制应用") @cur_workspace_access_guard() +@check_app_quota def copy_app( app_id: uuid.UUID, new_name: Optional[str] = None, @@ -1144,6 +1145,7 @@ async def import_workflow_config( @router.post("/workflow/import/save") @cur_workspace_access_guard() +@check_app_quota async def save_workflow_import( data: WorkflowImportSave, db: Session = Depends(get_db), @@ -1281,6 +1283,10 @@ async def import_app( return fail(msg="YAML 格式无效，缺少 app 字段", code=BizCode.BAD_REQUEST) target_app_id = uuid.UUID(app_id) if app_id else None + # 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") result_app, warnings = AppDslService(db).import_dsl( dsl=dsl, workspace_id=current_user.current_workspace_id, From 8bb5a664019448a1885a094f7086961317b6ce6c Mon Sep 17 00:00:00 2001 From: wwq Date: Tue, 21 Apr 2026 18:16:38 +0800 Subject: [PATCH 36/55] feat(exception): enhance I18nException response format and add error code mapping - Standardize error response format to include business error codes, timestamps, and other fields. - Add ERROR_CODE_TO_BIZ_CODE mapping table for error code conversion. - Introduce QUOTA_EXCEEDED and RATE_LIMIT_EXCEEDED business error codes. --- api/app/core/error_codes.py | 22 +++++++++++++++++++++- api/app/i18n/exceptions.py | 27 +++++++++++++++++++-------- 2 files changed, 40 insertions(+), 9 deletions(-) diff --git a/api/app/core/error_codes.py b/api/app/core/error_codes.py index a4a20cbb..77bce6b4 100644 --- a/api/app/core/error_codes.py +++ b/api/app/core/error_codes.py @@ -32,6 +32,8 @@ class BizCode(IntEnum): API_KEY_DAILY_LIMIT_EXCEEDED = 3015 API_KEY_QUOTA_EXCEEDED = 3016 API_KEY_RATE_LIMIT_EXCEEDED = 3017 + QUOTA_EXCEEDED = 3018 + RATE_LIMIT_EXCEEDED = 3019 # 资源（4xxx） NOT_FOUND = 4000 USER_NOT_FOUND = 4001 @@ -156,7 +158,8 @@ HTTP_MAPPING = { BizCode.API_KEY_QPS_LIMIT_EXCEEDED: 429, BizCode.API_KEY_DAILY_LIMIT_EXCEEDED: 429, BizCode.API_KEY_QUOTA_EXCEEDED: 429, - + BizCode.QUOTA_EXCEEDED: 402, + BizCode.MODEL_CONFIG_INVALID: 400, BizCode.API_KEY_MISSING: 400, BizCode.PROVIDER_NOT_SUPPORTED: 400, @@ -185,4 +188,21 @@ HTTP_MAPPING = { BizCode.DB_ERROR: 500, BizCode.SERVICE_UNAVAILABLE: 503, BizCode.RATE_LIMITED: 429, + BizCode.RATE_LIMIT_EXCEEDED: 429, +} + +ERROR_CODE_TO_BIZ_CODE = { + "QUOTA_EXCEEDED": BizCode.QUOTA_EXCEEDED, + "RATE_LIMIT_EXCEEDED": BizCode.RATE_LIMIT_EXCEEDED, + "API_KEY_NOT_FOUND": BizCode.API_KEY_NOT_FOUND, + "API_KEY_INVALID": BizCode.API_KEY_INVALID, + "API_KEY_EXPIRED": BizCode.API_KEY_EXPIRED, + "WORKSPACE_NOT_FOUND": BizCode.WORKSPACE_NOT_FOUND, + "WORKSPACE_NO_ACCESS": BizCode.WORKSPACE_NO_ACCESS, + "PERMISSION_DENIED": BizCode.PERMISSION_DENIED, + "TOKEN_EXPIRED": BizCode.TOKEN_EXPIRED, + "TOKEN_INVALID": BizCode.TOKEN_INVALID, + "VALIDATION_FAILED": BizCode.VALIDATION_FAILED, + "INVALID_PARAMETER": BizCode.INVALID_PARAMETER, + "MISSING_PARAMETER": BizCode.MISSING_PARAMETER, } diff --git a/api/app/i18n/exceptions.py b/api/app/i18n/exceptions.py index 9a517925..93794c39 100644 --- a/api/app/i18n/exceptions.py +++ b/api/app/i18n/exceptions.py @@ -6,12 +6,14 @@ error messages based on the current request's language. """ import logging +import time from contextvars import ContextVar from typing import Any, Dict, Optional from fastapi import HTTPException, Request from app.i18n.service import get_translation_service +from app.core.error_codes import ERROR_CODE_TO_BIZ_CODE, BizCode logger = logging.getLogger(__name__) @@ -118,15 +120,24 @@ class I18nException(HTTPException): **params ) - # Build error detail - detail = { - "error_code": self.error_code, - "message": message, - } + # Convert error_code string to BizCode value + biz_code = ERROR_CODE_TO_BIZ_CODE.get( + self.error_code, + BizCode.BAD_REQUEST + ) - # Add parameters to detail if provided - if params: - detail["params"] = params + # Build error detail in standard format for compatibility + # main.py handler expects "message" and "error_code" fields for filtering + # but we also include standard format fields + detail = { + "code": biz_code.value, + "msg": message, + "message": message, + "error_code": self.error_code, + "data": params if params else {}, + "error": message, + "time": int(time.time() * 1000), + } # Initialize HTTPException super().__init__( From ad4121b0d8efc3c9e8b5a88798f35e4aca418208 Mon Sep 17 00:00:00 2001 From: wwq Date: Tue, 21 Apr 2026 20:48:06 +0800 Subject: [PATCH 37/55] fix(api): fix API Key rate limiting and terminal user quota checks - Revert API Key rate limit handling to throw an error instead of auto-capping when exceeding the plan limit. - Optimize terminal user quota check logic to validate only during new user creation, avoiding redundant checks. - Add method to query terminal users by `workspace_id` and `other_id`. --- api/app/controllers/api_key_controller.py | 2 ++ .../controllers/public_share_controller.py | 26 +++++++++++++++++-- api/app/repositories/end_user_repository.py | 11 ++++++++ api/app/services/api_key_service.py | 14 +++++++--- 4 files changed, 47 insertions(+), 6 deletions(-) diff --git a/api/app/controllers/api_key_controller.py b/api/app/controllers/api_key_controller.py index dce8450d..6e414276 100644 --- a/api/app/controllers/api_key_controller.py +++ b/api/app/controllers/api_key_controller.py @@ -167,6 +167,8 @@ def update_api_key( return success(data=api_key_schema.ApiKey.model_validate(api_key), msg="API Key 更新成功") + except BusinessException: + raise except Exception as e: logger.error(f"未知错误: {str(e)}", extra={ "api_key_id": str(api_key_id), diff --git a/api/app/controllers/public_share_controller.py b/api/app/controllers/public_share_controller.py index 049535b5..486854ba 100644 --- a/api/app/controllers/public_share_controller.py +++ b/api/app/controllers/public_share_controller.py @@ -219,9 +219,20 @@ def list_conversations( end_user_repo = EndUserRepository(db) app_service = AppService(db) app = app_service._get_app_or_404(share.app_id) + workspace_id = app.workspace_id + + # 仅在新建终端用户时检查配额 + existing_end_user = end_user_repo.get_end_user_by_other_id(workspace_id=workspace_id, other_id=other_id) + if existing_end_user is None: + from app.core.quota_manager import _check_quota + 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") + new_end_user = end_user_repo.get_or_create_end_user( app_id=share.app_id, - workspace_id=app.workspace_id, + workspace_id=workspace_id, other_id=other_id ) logger.debug(new_end_user.id) @@ -309,7 +320,6 @@ def get_conversation( "/chat", summary="发送消息（支持流式和非流式）" ) -@check_end_user_quota async def chat( payload: conversation_schema.ChatRequest, share_data: ShareTokenData = Depends(get_share_user_id), @@ -350,6 +360,18 @@ async def chat( app_service = AppService(db) app = app_service._get_app_or_404(share.app_id) workspace_id = app.workspace_id + + # 仅在新建终端用户时检查配额，已有用户复用不受限制 + existing_end_user = end_user_repo.get_end_user_by_other_id(workspace_id=workspace_id, other_id=other_id) + logger.info(f"终端用户配额检查: workspace_id={workspace_id}, other_id={other_id}, existing={existing_end_user is not None}") + if existing_end_user is None: + from app.core.quota_manager import _check_quota + from app.models.workspace_model import Workspace + 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") + new_end_user = end_user_repo.get_or_create_end_user( app_id=share.app_id, workspace_id=workspace_id, diff --git a/api/app/repositories/end_user_repository.py b/api/app/repositories/end_user_repository.py index aad80707..aba4034f 100644 --- a/api/app/repositories/end_user_repository.py +++ b/api/app/repositories/end_user_repository.py @@ -66,6 +66,17 @@ class EndUserRepository: db_logger.error(f"查询宿主 {end_user_id} 时出错: {str(e)}") raise + def get_end_user_by_other_id(self, workspace_id: uuid.UUID, other_id: str) -> Optional["EndUser"]: + """按 workspace_id + other_id 查找终端用户，不存在返回 None""" + return ( + self.db.query(EndUser) + .filter( + EndUser.workspace_id == workspace_id, + EndUser.other_id == other_id + ) + .first() + ) + def get_or_create_end_user( self, app_id: uuid.UUID, diff --git a/api/app/services/api_key_service.py b/api/app/services/api_key_service.py index 5143ac3e..4856365a 100644 --- a/api/app/services/api_key_service.py +++ b/api/app/services/api_key_service.py @@ -51,7 +51,7 @@ class ApiKeyService: if existing: raise BusinessException(f"API Key 名称 {data.name} 已存在", BizCode.API_KEY_DUPLICATE_NAME) - # 若 rate_limit 超过租户套餐的 api_ops_rate_limit，自动截断到套餐上限 + # 若 rate_limit 超过租户套餐的 api_ops_rate_limit，直接报错 from app.models.workspace_model import Workspace from app.core.quota_manager import get_api_ops_rate_limit @@ -59,7 +59,10 @@ class ApiKeyService: if workspace: tenant_api_ops_limit = get_api_ops_rate_limit(db, workspace.tenant_id) if tenant_api_ops_limit and data.rate_limit > tenant_api_ops_limit: - data.rate_limit = tenant_api_ops_limit + raise BusinessException( + f"API Key QPS 不能超过套餐上限 {tenant_api_ops_limit}", + BizCode.BAD_REQUEST + ) # 生成 API Key api_key = generate_api_key(data.type) @@ -162,7 +165,7 @@ class ApiKeyService: if existing: raise BusinessException(f"API Key 名称 {data.name} 已存在", BizCode.API_KEY_DUPLICATE_NAME) - # 若 rate_limit 超过租户套餐的 api_ops_rate_limit，自动截断到套餐上限 + # 若 rate_limit 超过租户套餐的 api_ops_rate_limit，直接报错 if data.rate_limit is not None: from app.models.workspace_model import Workspace from app.core.quota_manager import get_api_ops_rate_limit @@ -171,7 +174,10 @@ class ApiKeyService: if workspace: tenant_api_ops_limit = get_api_ops_rate_limit(db, workspace.tenant_id) if tenant_api_ops_limit and data.rate_limit > tenant_api_ops_limit: - data.rate_limit = tenant_api_ops_limit + raise BusinessException( + f"API Key QPS 不能超过套餐上限 {tenant_api_ops_limit}", + BizCode.BAD_REQUEST + ) update_data = data.model_dump(exclude_unset=True) ApiKeyRepository.update(db, api_key_id, update_data) From 5e0d30dde8044cbcf22f106a0a947c562925baeb Mon Sep 17 00:00:00 2001 From: wwq Date: Tue, 21 Apr 2026 21:16:35 +0800 Subject: [PATCH 38/55] fix(quota): restrict quota check to new terminal user creation only - Avoid redundant quota checks for existing users on every request to optimize performance. --- api/app/controllers/service/app_api_controller.py | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/api/app/controllers/service/app_api_controller.py b/api/app/controllers/service/app_api_controller.py index a78fd842..d8aefc72 100644 --- a/api/app/controllers/service/app_api_controller.py +++ b/api/app/controllers/service/app_api_controller.py @@ -106,6 +106,16 @@ async def chat( other_id = payload.user_id workspace_id = api_key_auth.workspace_id end_user_repo = EndUserRepository(db) + + # 仅在新建终端用户时检查配额，已有用户复用不受限制 + existing_end_user = end_user_repo.get_end_user_by_other_id(workspace_id=workspace_id, other_id=other_id) + if existing_end_user is None: + from app.core.quota_manager import _check_quota + 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") + new_end_user = end_user_repo.get_or_create_end_user( app_id=app.id, workspace_id=workspace_id, From ea6fa154e09afb2d640cee52d5c6bbb828dd48af Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 22 Apr 2026 10:17:21 +0800 Subject: [PATCH 39/55] fix(web): stream add default error message --- web/src/utils/stream.ts | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) diff --git a/web/src/utils/stream.ts b/web/src/utils/stream.ts index ba966159..6ad1f785 100644 --- a/web/src/utils/stream.ts +++ b/web/src/utils/stream.ts @@ -2,7 +2,7 @@ * @Author: ZhaoYing * @Date: 2026-02-02 16:35:43 * @Last Modified by: ZhaoYing - * @Last Modified time: 2026-03-18 14:32:40 + * @Last Modified time: 2026-04-22 10:16:43 */ /** * Server-Sent Events (SSE) Stream Utility Module @@ -176,12 +176,12 @@ export const handleSSE = async (url: string, data: any, onMessage?: (data: SSEMe case 500: case 502: const errorData = await response.json(); - const errorInfo = errorData.error || i18n.t('common.serviceUpgrading'); + const errorInfo = errorData.error || errorData.msg || i18n.t('common.serviceUpgrading'); message.warning(errorInfo); throw new Error(errorData); case 400: const error = await response.json(); - const error400 = error.error || 'Bad Request'; + const error400 = error.error || error.msg || 'Bad Request'; message.warning(error400); throw new Error(error); case 403: @@ -190,7 +190,7 @@ export const handleSSE = async (url: string, data: any, onMessage?: (data: SSEMe throw new Error(errors); case 504: const errorJson = await response.json(); - const errorMsg = errorJson.error || i18n.t('common.serverError'); + const errorMsg = errorJson.error || errorJson.msg || i18n.t('common.serverError'); message.warning(errorMsg); throw new Error(errorJson); case 401: @@ -204,6 +204,13 @@ export const handleSSE = async (url: string, data: any, onMessage?: (data: SSEMe return; } break; + default: + if (!response.ok) { + const defaultData = await response.json().catch(() => ({})); + const defaultMsg = defaultData.error || defaultData.msg; + if (defaultMsg) message.warning(defaultMsg); + throw new Error(defaultMsg || `HTTP ${response.status}`); + } } if (!response.body) throw new Error('No response body'); From 77c023102e63054edcf3e0c4c7f89e2fb84e16d0 Mon Sep 17 00:00:00 2001 From: wwq Date: Wed, 22 Apr 2026 11:10:41 +0800 Subject: [PATCH 40/55] feat(plan): bump free plan model quota from 1 to 4 - Increase the model quota for the free tier from 1 to 4. --- api/app/config/default_free_plan.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/api/app/config/default_free_plan.py b/api/app/config/default_free_plan.py index 409b4f7b..a9ac3c66 100644 --- a/api/app/config/default_free_plan.py +++ b/api/app/config/default_free_plan.py @@ -62,7 +62,7 @@ def _build_default_free_plan(): "memory_engine_quota": 1, "end_user_quota": 1, "ontology_project_quota": 3, - "model_quota": 1, + "model_quota": 4, "api_ops_rate_limit": 50, }, } From 1ea0f308ba8a2977be3bfb3d5d97fb7c17a2e84a Mon Sep 17 00:00:00 2001 From: Mark <348207283@qq.com> Date: Wed, 22 Apr 2026 11:47:32 +0800 Subject: [PATCH 41/55] [fix] celery task --- api/app/celery_worker.py | 28 ++++++++++++++++++--- api/app/core/rag/common/connection_utils.py | 8 +++--- api/app/tasks.py | 13 ++++++---- 3 files changed, 35 insertions(+), 14 deletions(-) diff --git a/api/app/celery_worker.py b/api/app/celery_worker.py index e6c041f5..9fabe15b 100644 --- a/api/app/celery_worker.py +++ b/api/app/celery_worker.py @@ -19,15 +19,35 @@ import app.tasks @worker_process_init.connect def _reinit_db_pool(**kwargs): """ - prefork 子进程启动时重建 SQLAlchemy 连接池。 + prefork 子进程启动时重建被 fork 污染的资源。 - fork() 后子进程继承了父进程的连接池和底层 TCP socket， - 多个子进程共享同一个 socket 会导致 PostgreSQL 连接损坏。 - dispose() 会关闭继承来的连接，后续操作会自动创建新连接。 + fork() 后子进程继承了父进程的： + 1. SQLAlchemy 连接池 — 多进程共享 TCP socket 导致 DB 连接损坏 + 2. ThreadPoolExecutor — fork 后线程状态不确定，第二个任务会死锁 """ + # 重建 DB 连接池 from app.db import engine engine.dispose() logger.info("DB connection pool disposed for forked worker process") + # 重建模块级 ThreadPoolExecutor（fork 后线程池不可用） + try: + from app.core.rag.deepdoc.parser import figure_parser + from concurrent.futures import ThreadPoolExecutor + figure_parser.shared_executor = ThreadPoolExecutor(max_workers=10) + logger.info("figure_parser.shared_executor recreated") + except Exception as e: + logger.warning(f"Failed to recreate figure_parser.shared_executor: {e}") + + try: + from app.core.rag.utils import libre_office + from concurrent.futures import ThreadPoolExecutor + import os + max_workers = os.cpu_count() * 2 if os.cpu_count() else 4 + libre_office.executor = ThreadPoolExecutor(max_workers=max_workers) + logger.info("libre_office.executor recreated") + except Exception as e: + logger.warning(f"Failed to recreate libre_office.executor: {e}") + __all__ = ['celery_app'] diff --git a/api/app/core/rag/common/connection_utils.py b/api/app/core/rag/common/connection_utils.py index 349caa27..d5d0dc2a 100644 --- a/api/app/core/rag/common/connection_utils.py +++ b/api/app/core/rag/common/connection_utils.py @@ -33,18 +33,16 @@ def timeout(seconds: float | int | str = None, attempts: int = 2, *, exception: thread.daemon = True thread.start() + effective_timeout = seconds if seconds else 120 # 默认 120 秒超时 for a in range(attempts): try: - if os.environ.get("ENABLE_TIMEOUT_ASSERTION"): - result = result_queue.get(timeout=seconds) - else: - result = result_queue.get() + result = result_queue.get(timeout=effective_timeout) if isinstance(result, Exception): raise result return result except queue.Empty: pass - raise TimeoutError(f"Function '{func.__name__}' timed out after {seconds} seconds and {attempts} attempts.") + raise TimeoutError(f"Function '{func.__name__}' timed out after {effective_timeout} seconds and {attempts} attempts.") @wraps(func) async def async_wrapper(*args, **kwargs) -> Any: diff --git a/api/app/tasks.py b/api/app/tasks.py index e965a281..92843175 100644 --- a/api/app/tasks.py +++ b/api/app/tasks.py @@ -268,18 +268,21 @@ def parse_document(file_path: str, document_id: uuid.UUID): try: with open(file_path, "rb") as f: file_binary = f.read() + if not file_binary: + # NFS 上文件存在但内容为空（可能还在同步中） + raise IOError(f"File is empty (0 bytes), NFS may still be syncing: {file_path}") break - except FileNotFoundError: + except (FileNotFoundError, IOError) as e: if waited >= max_wait_seconds: - raise FileNotFoundError( - f"File not found at '{file_path}' after waiting {max_wait_seconds}s " - f"(NFS cache may be stale)" + raise type(e)( + f"File not accessible at '{file_path}' after waiting {max_wait_seconds}s: {e}" ) - logger.warning(f"File not visible yet on this node, retrying in {wait_interval}s: {file_path}") + logger.warning(f"File not ready on this node, retrying in {wait_interval}s: {file_path} ({e})") time.sleep(wait_interval) waited += wait_interval from app.core.rag.app.naive import chunk + logger.info(f"[ParseDoc] file_binary size={len(file_binary)} bytes, type={type(file_binary).__name__}, bool={bool(file_binary)}") res = chunk(filename=file_path, binary=file_binary, from_page=0, From 9159d5cbb0d9d3c35d2d80af51718ad544e51289 Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 22 Apr 2026 11:58:47 +0800 Subject: [PATCH 42/55] fix(web): i18n update --- web/src/i18n/en.ts | 3 +++ web/src/i18n/zh.ts | 3 +++ 2 files changed, 6 insertions(+) diff --git a/web/src/i18n/en.ts b/web/src/i18n/en.ts index 13634544..4d284905 100644 --- a/web/src/i18n/en.ts +++ b/web/src/i18n/en.ts @@ -451,6 +451,9 @@ export const en = { logoutApiCannotRefreshToken: 'Logout API cannot refresh token', publicApiCannotRefreshToken: 'Public API cannot refresh token', refreshTokenNotExist: 'Refresh token does not exist', + SYSTEM_DEFAULT_SCENE_CANNOT_DELETE: 'This is a system preset scene and cannot be deleted', + SYSTEM_DEFAULT_CLASS_CANNOT_DELETE: 'This scene is a system preset scene and cannot be deleted', + SYSTEM_DEFAULT_SCENE_CANNOT_UPDATE: 'This scene is a system preset scene and cannot be modified', reset: 'Reset', refresh: 'Refresh', return: 'Return', diff --git a/web/src/i18n/zh.ts b/web/src/i18n/zh.ts index 64689bc2..aea3b965 100644 --- a/web/src/i18n/zh.ts +++ b/web/src/i18n/zh.ts @@ -1130,6 +1130,9 @@ export const zh = { logoutApiCannotRefreshToken: '退出登录接口不能刷新token', publicApiCannotRefreshToken: '公共接口不能刷新token', refreshTokenNotExist: '刷新token不存在', + SYSTEM_DEFAULT_SCENE_CANNOT_DELETE: '该场景为系统预设场景，不允许删除', + SYSTEM_DEFAULT_CLASS_CANNOT_DELETE: '该场景为系统预设场景，不允许删除', + SYSTEM_DEFAULT_SCENE_CANNOT_UPDATE: '该场景为系统预设场景，不允许修改', reset: '重置', refresh: '刷新', return: '返回', From a90757745da1b67ef71c5524910e21ca2f930d7a Mon Sep 17 00:00:00 2001 From: zhaoying Date: Wed, 22 Apr 2026 13:02:42 +0800 Subject: [PATCH 43/55] fix(web): tool config --- .../Properties/ToolConfig/index.tsx | 30 +++++++++---------- 1 file changed, 15 insertions(+), 15 deletions(-) diff --git a/web/src/views/Workflow/components/Properties/ToolConfig/index.tsx b/web/src/views/Workflow/components/Properties/ToolConfig/index.tsx index d38265da..6e8bd0c0 100644 --- a/web/src/views/Workflow/components/Properties/ToolConfig/index.tsx +++ b/web/src/views/Workflow/components/Properties/ToolConfig/index.tsx @@ -1,4 +1,4 @@ -import { type FC, useEffect, useState, useMemo } from "react"; +import { type FC, useEffect, useState } from "react"; import { useTranslation } from 'react-i18next' import { Form, Select, Switch, Cascader, type CascaderProps, Tooltip } from 'antd' import type { Suggestion } from '../../Editor/plugin/AutocompletePlugin' @@ -45,15 +45,15 @@ const ToolConfig: FC<{ options: Suggestion[]; }> = ({ getToolDetail(values.tool_id) .then(res => { const detail = res as { tool_type: ToolType; } - + getTools({ tool_type: detail.tool_type }) .then(toolsRes => { const tools = toolsRes as ToolItem[] - + getToolMethods(values.tool_id) .then(methodsRes => { const response = methodsRes as Array<{ method_id: string; name: string; parameters: Parameter[] }> - + setOptionList(prevList => { return prevList.map(item => { if (item.value === detail.tool_type) { @@ -76,7 +76,7 @@ const ToolConfig: FC<{ options: Suggestion[]; }> = ({ return item }) }) - + if (response.length > 1) { const filterTarget = response.find(vo => vo.name === values.tool_parameters?.operation) if (filterTarget) { @@ -98,7 +98,7 @@ const ToolConfig: FC<{ options: Suggestion[]; }> = ({ useEffect(() => { if (values.tools && values.tools.length === 3) { const [toolType, toolId, operation] = values.tools - + // 从 optionList 中查找对应的参数 const typeOption = optionList.find(opt => opt.value === toolType) if (typeOption?.children) { @@ -155,18 +155,18 @@ const ToolConfig: FC<{ options: Suggestion[]; }> = ({ const targetOption = selectedOptions[selectedOptions.length - 1]; const curParameters = [...(targetOption.parameters ?? [])] setParameters([...curParameters]) - const inititalValue: any = { tool_id: selectedOptions[1].value, tool_parameters: {} } + const initialValue: any = { tool_id: selectedOptions[1].value, tool_parameters: { operation: undefined } } if (value[0] === 'mcp' || (value[0] === 'builtin' && selectedOptions[1]?.children && selectedOptions[1].children.length > 1)) { - inititalValue.tool_parameters.operation = value?.[2] + initialValue.tool_parameters.operation = value?.[2] } else if (value[0] === 'custom') { - inititalValue.tool_parameters.operation = selectedOptions?.[2].method_id + initialValue.tool_parameters.operation = selectedOptions?.[2].method_id } curParameters.forEach(vo => { - inititalValue.tool_parameters[vo.name] = vo.default + initialValue.tool_parameters[vo.name] = vo.default }) - form.setFieldsValue(inititalValue) + form.setFieldsValue(initialValue) } // string -> string @@ -214,9 +214,9 @@ const ToolConfig: FC<{ options: Suggestion[]; }> = ({ name="tools" label={t('workflow.config.tool.tool_id')} > - = ({ {parameter.type === 'string' && parameter.enum && parameter.enum.length > 0 ?