Stream chat completions from model providers

ai_knowledge_base_v2/apps/backend/app/api/chat.py

@@ -21,6 +21,7 @@ from app.schemas.chat import (
 )
 from app.services.chat_service import ChatService
 from app.services.chat_queue_service import chat_queue_manager, load_chat_queue_config
+from app.services.chat_stream_service import ChatStreamService
 
 router = APIRouter()
 
@@ -124,8 +125,8 @@ def _chat_stream(payload: ChatCompletionRequest, db: Session, current_user: User
             activeCount=queue_request.active_count,
             waitingCount=queue_request.waiting_count,
         )
-        answer = ChatService.create_answer(db, current_user, payload.sessionId, payload.message)
-        yield from _sse_chunks(answer)
+        for chunk in ChatStreamService.stream_answer(db, current_user, payload.sessionId, payload.message):
+            yield _sse_event("content", content=chunk)
     except HTTPException as exc:
         yield _sse_event("error", message=str(exc.detail))
     except Exception:
@@ -138,14 +139,6 @@ def _chat_stream(payload: ChatCompletionRequest, db: Session, current_user: User
 
     yield _sse_done()
 
-
-def _sse_chunks(answer: str) -> Iterator[str]:
-    chunk_size = 12
-    for index in range(0, len(answer), chunk_size):
-        chunk = answer[index : index + chunk_size]
-        yield _sse_event("content", content=chunk)
-
-
 def _sse_event(event_type: str, **payload: object) -> str:
     return f"data: {json.dumps({'type': event_type, **payload}, ensure_ascii=False)}\n\n"