Fix 法规对话

backend/app/shared/async_utils.py

@@ -0,0 +1,38 @@
+"""Async utility helpers for bridging sync generators to async FastAPI routes."""
+
+from __future__ import annotations
+
+import asyncio
+import threading
+from typing import AsyncGenerator, Generator, TypeVar
+
+T = TypeVar("T")
+
+_SENTINEL = object()
+
+
+async def iter_in_thread(sync_gen: Generator[T, None, None]) -> AsyncGenerator[T, None]:
+    """Yield items from a synchronous generator without blocking the event loop.
+
+    Runs the generator in a daemon thread, forwarding items via an asyncio.Queue.
+    Use this to wrap blocking LLM streaming generators inside async FastAPI routes.
+    """
+    loop = asyncio.get_running_loop()
+    queue: asyncio.Queue = asyncio.Queue(maxsize=32)
+
+    def _drain() -> None:
+        try:
+            for item in sync_gen:
+                future = asyncio.run_coroutine_threadsafe(queue.put(item), loop)
+                future.result()
+        finally:
+            asyncio.run_coroutine_threadsafe(queue.put(_SENTINEL), loop).result()
+
+    thread = threading.Thread(target=_drain, daemon=True)
+    thread.start()
+
+    while True:
+        item = await queue.get()
+        if item is _SENTINEL:
+            break
+        yield item  # type: ignore[misc]

backend/app/shared/bootstrap.py

@@ -19,6 +19,7 @@ from app.infrastructure.storage.json_document_repository import JsonDocumentRepo
 from app.infrastructure.storage.minio_binary_store import MinioDocumentBinaryStore
 from app.infrastructure.storage.postgres_document_repository import PostgresDocumentRepository
 from app.infrastructure.storage.postgres_parse_artifact_store import PostgresParseArtifactStore
+from app.infrastructure.vectorstore.bm25_retriever import BM25Retriever
 from app.infrastructure.vectorstore.dense_retriever import DenseRetriever
 from app.infrastructure.vectorstore.milvus_vector_index import MilvusVectorIndex
 from app.infrastructure.vectorstore.cross_encoder_reranker import OpenAICompatibleReranker
@@ -87,6 +88,13 @@ def get_reranker():
     return None
 
 
+@lru_cache
+def get_bm25_retriever() -> BM25Retriever | None:
+    """Return BM25 retriever if rank_bm25 + jieba are installed, else None."""
+    retriever = BM25Retriever(vector_index=get_vector_index())
+    return retriever if retriever.available else None
+
+
 @lru_cache
 def get_retrieval_service() -> KnowledgeRetrievalService:
     """Return retrieval service."""
@@ -96,6 +104,7 @@ def get_retrieval_service() -> KnowledgeRetrievalService:
     )
     return KnowledgeRetrievalService(
         retriever=retriever,
+        bm25_retriever=get_bm25_retriever(),
         reranker=get_reranker(),
         reranker_top_k=settings.reranker_top_k,
     )