Merge branch 'main' into feat/1.3.0

# Conflicts:
#	src/backend/bisheng/api/services/knowledge_imp.py
#	src/frontend/platform/src/pages/KnowledgePage/components/FileUploadStep2.tsx

Author: zgqgit

docker/docker-compose-uns.yml

@@ -1,7 +1,7 @@
 services:
   bisheng-unstructured:
     container_name: bisheng-unstructured
-    image: dataelement/bisheng-unstructured:v0.0.3.11
+    image: dataelement/bisheng-unstructured:v0.0.3.14
     ports:
       - "10001:10001"
     environment:

docker/docker-compose.yml

@@ -53,7 +53,7 @@ services:
 
   backend:
     container_name: bisheng-backend
-    image: dataelement/bisheng-backend:v1.1.1
+    image: dataelement/bisheng-backend:v1.2.0
     ports:
       - "7860:7860"
     environment:
@@ -93,7 +93,7 @@ services:
 
   frontend:
     container_name: bisheng-frontend
-    image: dataelement/bisheng-frontend:v1.1.1
+    image: dataelement/bisheng-frontend:v1.2.0
     ports:
       - "3001:3001"
     environment:

src/backend/bisheng/__init__.py

@@ -6,7 +6,7 @@
 
 try:
     # 通过ci去自动修改
-    __version__ = '1.1.1'
+    __version__ = '1.2.0'
 except metadata.PackageNotFoundError:
     # Case where package metadata is not available.
     __version__ = ''

src/backend/bisheng/api/services/knowledge_imp.py

@@ -448,9 +448,7 @@ def add_file_embedding(
                 metadatas.append(val["metadata"])
     for index, one in enumerate(texts):
         if len(one) > 10000:
-            raise ValueError(
-                "分段结果超长，请尝试在自定义策略中使用更多切分符（例如 \n）进行切分"
-            )
+            raise ValueError('分段结果超长，请尝试在自定义策略中使用更多切分符（例如 \\n、。、\\.）进行切分')
         # 入库时 拼接文件名和文档摘要
         texts[index] = KnowledgeUtils.aggregate_chunk_metadata(one, metadatas[index])
 
@@ -929,7 +927,7 @@ def qa_status_change(qa_id: int, target_status: int):
         qa_db.status = target_status
         QAKnoweldgeDao.update(qa_db)
     else:
-        qa_db.status = target_status
+        qa_db.status = QAStatus.PROCESSING.value
         QAKnoweldgeDao.update(qa_db)
         QA_save_knowledge(db_knowledge, qa_db)
     return qa_db

src/backend/bisheng/api/v1/schemas.py

@@ -147,7 +147,7 @@ class ChatMessage(BaseModel):
     files: Optional[list] = []
     user_id: Optional[int] = None
     message_id: Optional[int | str] = None
-    source: int = 0
+    source: Optional[int] = 0
     sender: Optional[str] = None
     receiver: Optional[dict] = None
     liked: int = 0

src/backend/bisheng/chat/client.py

@@ -245,7 +245,7 @@ async def stop_handle_message(self, message: Dict[any, any]):
         if reasoning_answer.split():
             res = await self.add_message('bot', answer, 'reasoning_answer', 'break_answer')
             await self.send_response('reasoning_answer', 'end', '', message_id=res.id if res else None)
-        elif answer.strip():
+        if answer.strip():
             res = await self.add_message('bot', answer, 'answer', 'break_answer')
             await self.send_response('answer', 'end', '', message_id=res.id if res else None)
         await self.send_response('processing', 'close', '')

src/backend/bisheng/chat/handlers.py

@@ -10,6 +10,7 @@
 from bisheng.chat.utils import judge_source, process_graph, process_source_document
 from bisheng.database.base import session_getter
 from bisheng.database.models.report import Report
+from bisheng.database.models.message import ChatMessage as ChatMessageDB, ChatMessageDao
 from bisheng.interface.importing.utils import import_by_type
 from bisheng.interface.initialize.loading import instantiate_llm
 from bisheng.settings import settings
@@ -67,6 +68,38 @@ async def process_stop(self, session: ChatManager, client_id: str, chat_id: str,
             else:
                 logger.error(f'act=auto_gen act={action}')
         else:
+            # 将流式输出的内容写到数据库内
+            answer = ''
+            reasoning_answer = ''
+            while not self.stream_queue.empty():
+                msg = self.stream_queue.get()
+                if msg.get('type') == 'answer':
+                    answer += msg.get('content', '')
+                elif msg.get('type') == 'reasoning':
+                    reasoning_answer += msg.get('content', '')
+            if reasoning_answer.strip():
+                chat_message = ChatMessageDB(flow_id=client_id, chat_id=chat_id,
+                                             message=reasoning_answer,
+                                             category='answer',
+                                             type='end',
+                                             user_id=user_id,
+                                             remark='break_answer',
+                                             is_bot=True)
+                if chat_id:
+                    db_message = ChatMessageDao.insert_one(chat_message)
+                    await session.send_json(client_id, chat_id, ChatMessage(**db_message.model_dump(), message_id=db_message.id), add=False)
+
+            if answer.strip():
+                chat_message = ChatMessageDB(flow_id=client_id, chat_id=chat_id,
+                                             message=answer,
+                                             category='answer',
+                                             type='end',
+                                             user_id=user_id,
+                                             remark='break_answer',
+                                             is_bot=True)
+                if chat_id:
+                    db_message = ChatMessageDao.insert_one(chat_message)
+                    await session.send_json(client_id, chat_id, ChatMessage(**db_message.model_dump(), message_id=db_message.id), add=False)
             # 普通技能的stop
             res = thread_pool.cancel_task([key])  # 将进行中的任务进行cancel
             if res[0]:
@@ -75,18 +108,7 @@ async def process_stop(self, session: ChatManager, client_id: str, chat_id: str,
                 close = ChatResponse(type='close')
                 await session.send_json(client_id, chat_id, res, add=False)
                 await session.send_json(client_id, chat_id, close, add=False)
-        answer = ''
-        # 记录中止后产生的流式输出内容
-        while not self.stream_queue.empty():
-            answer += self.stream_queue.get()
-        if answer.strip():
-            chat_message = ChatMessage(message=answer,
-                                       category='answer',
-                                       type='end',
-                                       user_id=user_id,
-                                       remark='break_answer',
-                                       is_bot=True)
-            session.chat_history.add_message(client_id, chat_id, chat_message)
+
         logger.info('process_stop done')
 
     async def process_report(self,

src/backend/bisheng/chat/manager.py

@@ -51,6 +51,7 @@ def add_message(
         from bisheng.database.models.message import ChatMessage
         message.flow_id = client_id
         message.chat_id = chat_id
+        db_message = None
         if chat_id and (message.message or message.intermediate_steps
                         or message.files) and message.type != 'stream':
             msg = message.copy()
@@ -67,6 +68,7 @@ def add_message(
 
         if not isinstance(message, FileResponse):
             self.notify()
+        return db_message
 
     def empty_history(self, client_id: str, chat_id: str):
         """Empty the chat history for a client."""

src/backend/bisheng/database/models/knowledge_file.py

@@ -5,7 +5,7 @@
 
 # if TYPE_CHECKING:
 from pydantic import field_validator
-from sqlalchemy import JSON, Column, DateTime, String, or_, text
+from sqlalchemy import JSON, Column, DateTime, String, or_, text, Text
 from sqlmodel import Field, delete, func, select
 
 from bisheng.database.base import session_getter
@@ -97,7 +97,7 @@ class KnowledgeFile(KnowledgeFileBase, table=True):
 class QAKnowledge(QAKnowledgeBase, table=True):
     id: Optional[int] = Field(default=None, primary_key=True)
     questions: Optional[List[str]] = Field(default=None, sa_column=Column(JSON))
-    answers: Optional[str] = Field(default=None, sa_column=Column(String(length=2048)))
+    answers: Optional[str] = Field(default=None, sa_column=Column(Text))
 
 
 class KnowledgeFileRead(KnowledgeFileBase):

src/backend/bisheng/worker/workflow/redis_callback.py

@@ -153,7 +153,7 @@ async def get_response_until_break(self) -> AsyncIterator[ChatResponse]:
                         break
                     yield chat_response
                 break
-            elif status_info['status'] == WorkflowStatus.WAITING.value and time.time() - status_info['time'] > 10:
+            elif status_info['status'] in [WorkflowStatus.WAITING.value, WorkflowStatus.INPUT_OVER.value] and time.time() - status_info['time'] > 10:
                 # 10秒内没有收到状态更新，说明workflow没有启动，可能是celery worker线程数已满
                 self.set_workflow_status(WorkflowStatus.FAILED.value, 'workflow task execute busy')
                 yield self.build_chat_response(WorkflowEventType.Error.value, 'over',