adai/MemoryBear
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request #349 from SuanmoSuanyangTechnology/fix/multimodal

Browse Source 
fix(multimodal): temporarily limit API to image-only modality
This commit is contained in:
Mark
2026-02-06 17:28:53 +08:00
committed by GitHub
parent 12062a5440 e253fba2e9
commit 320f684354
5 changed files with 93 additions and 79 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
api/app/controllers/public_share_controller.py
View File

@@ -580,6 +580,7 @@ async def chat(
conversation_id=conversation.id, # 使用已创建的会话 ID
user_id=end_user_id, # 转换为字符串
variables=payload.variables,
files=payload.files,
config=config,
web_search=payload.web_search,
memory=payload.memory,

95
api/app/controllers/service/app_api_controller.py
View File

@@ -12,7 +12,6 @@ from app.core.exceptions import BusinessException
from app.core.logging_config import get_business_logger
from app.core.response_utils import success
from app.db import get_db
from app.dependencies import get_app_or_workspace
from app.models.app_model import App
from app.models.app_model import AppType
from app.repositories import knowledge_repository
@@ -21,9 +20,10 @@ from app.schemas import AppChatRequest, conversation_schema
from app.schemas.api_key_schema import ApiKeyAuth
from app.services import workspace_service
from app.services.app_chat_service import AppChatService, get_app_chat_service
from app.services.conversation_service import ConversationService, get_conversation_service
from app.utils.app_config_utils import dict_to_multi_agent_config, workflow_config_4_app_release, agent_config_4_app_release, multi_agent_config_4_app_release
from app.services.app_service import get_app_service, AppService
from app.services.conversation_service import ConversationService, get_conversation_service
from app.utils.app_config_utils import workflow_config_4_app_release, \
agent_config_4_app_release, multi_agent_config_4_app_release
router = APIRouter(prefix="/app", tags=["V1 - App API"])
logger = get_business_logger()
@@ -34,6 +34,7 @@ async def list_apps():
"""列出可访问的应用（占位）"""
return success(data=[], msg="App API - Coming Soon")
# /v1/app/chat
# @router.post("/chat")
@@ -73,16 +74,17 @@ def _checkAppConfig(app: App):
else:
raise BusinessException("不支持的应用类型", BizCode.AGENT_CONFIG_MISSING)
@router.post("/chat")
@require_api_key(scopes=["app"])
async def chat(
request:Request,
api_key_auth: ApiKeyAuth = None,
db: Session = Depends(get_db),
conversation_service: Annotated[ConversationService, Depends(get_conversation_service)] = None,
app_chat_service: Annotated[AppChatService, Depends(get_app_chat_service)] = None,
app_service: Annotated[AppService, Depends(get_app_service)] = None,
message: str = Body(..., description="聊天消息内容"),
request: Request,
api_key_auth: ApiKeyAuth = None,
db: Session = Depends(get_db),
conversation_service: Annotated[ConversationService, Depends(get_conversation_service)] = None,
app_chat_service: Annotated[AppChatService, Depends(get_app_chat_service)] = None,
app_service: Annotated[AppService, Depends(get_app_service)] = None,
message: str = Body(..., description="聊天消息内容"),
):
body = await request.json()
payload = AppChatRequest(**body)
@@ -98,8 +100,8 @@ async def chat(
original_user_id=other_id # Save original user_id to other_id
)
end_user_id = str(new_end_user.id)
web_search=True
memory=True
web_search = True
memory = True
# 提前验证和准备（在流式响应开始前完成）
storage_type = workspace_service.get_workspace_storage_type_without_auth(
db=db,
@@ -146,17 +148,17 @@ async def chat(
if payload.stream:
async def event_generator():
async for event in app_chat_service.agnet_chat_stream(
message=payload.message,
conversation_id=conversation.id, # 使用已创建的会话 ID
user_id= end_user_id, # 转换为字符串
variables=payload.variables,
web_search=web_search,
config=agent_config,
memory=memory,
storage_type=storage_type,
user_rag_memory_id=user_rag_memory_id,
workspace_id=workspace_id,
files=payload.files # 传递多模态文件
message=payload.message,
conversation_id=conversation.id, # 使用已创建的会话 ID
user_id=end_user_id, # 转换为字符串
variables=payload.variables,
web_search=web_search,
config=agent_config,
memory=memory,
storage_type=storage_type,
user_rag_memory_id=user_rag_memory_id,
workspace_id=workspace_id,
files=payload.files # 传递多模态文件
):
yield event
@@ -176,7 +178,7 @@ async def chat(
conversation_id=conversation.id, # 使用已创建的会话 ID
user_id=end_user_id, # 转换为字符串
variables=payload.variables,
config= agent_config,
config=agent_config,
web_search=web_search,
memory=memory,
storage_type=storage_type,
@@ -192,15 +194,15 @@ async def chat(
async def event_generator():
async for event in app_chat_service.multi_agent_chat_stream(
message=payload.message,
conversation_id=conversation.id, # 使用已创建的会话 ID
user_id=end_user_id, # 转换为字符串
variables=payload.variables,
config=config,
web_search=web_search,
memory=memory,
storage_type=storage_type,
user_rag_memory_id=user_rag_memory_id
message=payload.message,
conversation_id=conversation.id, # 使用已创建的会话 ID
user_id=end_user_id, # 转换为字符串
variables=payload.variables,
config=config,
web_search=web_search,
memory=memory,
storage_type=storage_type,
user_rag_memory_id=user_rag_memory_id
):
yield event
@@ -234,19 +236,19 @@ async def chat(
if payload.stream:
async def event_generator():
async for event in app_chat_service.workflow_chat_stream(
message=payload.message,
conversation_id=conversation.id, # 使用已创建的会话 ID
user_id=end_user_id, # 转换为字符串
variables=payload.variables,
config=config,
web_search=web_search,
memory=memory,
storage_type=storage_type,
user_rag_memory_id=user_rag_memory_id,
app_id=app.id,
workspace_id=workspace_id,
release_id=app.current_release.id,
message=payload.message,
conversation_id=conversation.id, # 使用已创建的会话 ID
user_id=end_user_id, # 转换为字符串
variables=payload.variables,
files=payload.files,
config=config,
web_search=web_search,
memory=memory,
storage_type=storage_type,
user_rag_memory_id=user_rag_memory_id,
app_id=app.id,
workspace_id=workspace_id,
release_id=app.current_release.id,
):
event_type = event.get("event", "message")
event_data = event.get("data", {})
@@ -296,4 +298,3 @@ async def chat(
from app.core.exceptions import BusinessException
from app.core.error_codes import BizCode
raise BusinessException(f"不支持的应用类型: {app_type}", BizCode.APP_TYPE_NOT_SUPPORTED)

20
api/app/schemas/app_schema.py
View File

@@ -15,6 +15,13 @@ class FileType(StrEnum):
AUDIO = "audio"
VIDEO = "video"
@classmethod
def trans(cls, value: str) -> 'FileType':
if value.startswith("image"):
return cls.IMAGE
# TODO: other file type support
raise RuntimeError("Unsupport file type")
class TransferMethod(str, Enum):
"""文件传输方式枚举"""
@@ -28,7 +35,13 @@ class FileInput(BaseModel):
transfer_method: TransferMethod = Field(..., description="传输方式: local_file/remote_url")
upload_file_id: Optional[uuid.UUID] = Field(None, description="已上传文件ID（local_file时必填）")
url: Optional[str] = Field(None, description="远程URL（remote_url时必填）")
@field_validator("type", mode="before")
@classmethod
def validate_type(cls, v):
"""验证文件类型"""
return FileType.trans(v)
@field_validator("upload_file_id")
@classmethod
def validate_local_file(cls, v, info):
@@ -36,7 +49,7 @@ class FileInput(BaseModel):
if info.data.get("transfer_method") == TransferMethod.LOCAL_FILE and not v:
raise ValueError("transfer_method 为 local_file 时，upload_file_id 不能为空")
return v
@field_validator("url")
@classmethod
def validate_remote_url(cls, v, info):
@@ -82,6 +95,7 @@ class ToolConfig(BaseModel):
tool_id: Optional[str] = Field(default=None, description="工具ID")
operation: Optional[str] = Field(default=None, description="工具特定配置")
class SkillConfig(BaseModel):
"""技能配置"""
enabled: bool = Field(default=True, description="是否启用该技能")
@@ -216,7 +230,7 @@ class AgentConfigUpdate(BaseModel):
# 工具配置
tools: Optional[List[ToolConfig]] = Field(default_factory=list, description="工具列表")
# 技能配置
skills: Optional[SkillConfig] = Field(default=dict, description="关联的技能列表")

4
api/app/services/app_chat_service.py
View File

@@ -655,6 +655,7 @@ class AppChatService:
workspace_id: uuid.UUID,
user_id: str = None,
variables: Optional[Dict[str, Any]] = None,
files: Optional[List[FileInput]] = None,
web_search: bool = False,
memory: bool = True,
storage_type: Optional[str] = None,
@@ -669,7 +670,8 @@ class AppChatService:
variables=variables,
conversation_id=str(conversation_id),
stream=True,
user_id=user_id
user_id=user_id,
files=files
)
async for event in workflow_service.run_stream(
app_id=app_id,

52
api/app/services/workflow_service.py
View File

@@ -446,19 +446,9 @@ class WorkflowService:
code=BizCode.CONFIG_MISSING,
message=f"工作流配置不存在: app_id={app_id}"
)
files = []
if payload.files:
for file in payload.files:
files.append(
{
"type": file.type,
"url": await self.multimodal_service.get_file_url(file),
"__file": True
}
)
input_data = {"message": payload.message, "variables": payload.variables,
"conversation_id": payload.conversation_id, "files": files}
"conversation_id": payload.conversation_id, "files": [file.model_dump() for file in payload.files]}
# 转换 conversation_id 为 UUID
conversation_id_uuid = uuid.UUID(payload.conversation_id) if payload.conversation_id else None
@@ -488,6 +478,17 @@ class WorkflowService:
from app.core.workflow.executor import execute_workflow
try:
files = []
if payload.files:
for file in payload.files:
files.append(
{
"type": file.type,
"url": await self.multimodal_service.get_file_url(file),
"__file": True
}
)
input_data["files"] = files
# 更新状态为运行中
self.update_execution_status(execution.execution_id, "running")
@@ -634,19 +635,8 @@ class WorkflowService:
message=f"工作流配置不存在: app_id={app_id}"
)
files = []
if payload.files:
for file in payload.files:
files.append(
{
"type": file.type,
"url": await self.multimodal_service.get_file_url(file),
"__file": True
}
)
input_data = {"message": payload.message, "variables": payload.variables,
"conversation_id": payload.conversation_id, "files": files}
"conversation_id": payload.conversation_id, "files": [file.model_dump() for file in payload.files]}
# 转换 conversation_id 为 UUID
conversation_id_uuid = uuid.UUID(payload.conversation_id) if payload.conversation_id else None
@@ -670,12 +660,18 @@ class WorkflowService:
"execution_config": config.execution_config
}
# 4. 获取工作空间 ID（从 app 获取）
# 5. 流式执行工作流
try:
# 更新状态为运行中
files = []
if payload.files:
for file in payload.files:
files.append(
{
"type": file.type,
"url": await self.multimodal_service.get_file_url(file),
"__file": True
}
)
input_data["files"] = files
self.update_execution_status(execution.execution_id, "running")
executions = self.execution_repo.get_by_conversation_id(conversation_id=conversation_id_uuid)