.

scripts/rag_build_query.py

@@ -1,4 +1,8 @@
 import os
+from typing import List
+import asyncio
+import faiss
+from sentence_transformers import SentenceTransformer
 
 from llama_index import (
     SimpleDirectoryReader,
@@ -7,47 +11,23 @@ from llama_index import (
     PromptTemplate,
     ServiceContext,
 )
-from llama_index.embeddings.base import BaseEmbedding
+from llama_index.embeddings.huggingface import HuggingFaceEmbedding
 from llama_index.vector_stores.faiss import FaissVectorStore
 from app.core.config import settings
 from scripts.permissions import get_user_allowed_indexes
-import faiss
-from typing import List
-import asyncio
-from sentence_transformers import SentenceTransformer
-from llama_index.embeddings.huggingface import HuggingFaceEmbedding
 
 USER_INDEX_PATH = "index_data"
 USER_DOC_PATH = "docs"
 
-
-class CustomEmbedding(BaseEmbedding):
-    model: SentenceTransformer  # ✅ 显式声明是字段
-
-    def __init__(self, model_name: str):
-        from sentence_transformers import SentenceTransformer
-        self.model = SentenceTransformer(model_name)
-
-    # 同步方法（必须实现）
-    def _get_text_embedding(self, text: str) -> List[float]:
-        return self.model.encode(text).tolist()
-
+# ✅ 替代 CustomEmbedding，用于 bge-m3 模型，自动加前缀
+class BGEEmbedding(HuggingFaceEmbedding):
     def _get_query_embedding(self, query: str) -> List[float]:
-        return self.model.encode(query).tolist()
-
-    def _get_text_embeddings(self, texts: List[str]) -> List[List[float]]:
-        return self.model.encode(texts).tolist()
+        prefix = "Represent this sentence for searching relevant passages: "
+        return super()._get_query_embedding(prefix + query)
 
     def _get_query_embeddings(self, queries: List[str]) -> List[List[float]]:
-        return self.model.encode(queries).tolist()
-
-    # 异步方法（必须实现，哪怕用同步方式包起来）
-    async def _aget_query_embedding(self, query: str) -> List[float]:
-        return self._get_query_embedding(query)
-
-    async def _aget_query_embeddings(self, queries: List[str]) -> List[List[float]]:
-        return self._get_query_embeddings(queries)
-
+        prefix = "Represent this sentence for searching relevant passages: "
+        return super()._get_query_embeddings([prefix + q for q in queries])
 
 def build_user_index(user_id: str):
     doc_dir = os.path.join(USER_DOC_PATH, user_id)
@@ -55,13 +35,17 @@ def build_user_index(user_id: str):
         raise FileNotFoundError(f"文档目录不存在: {doc_dir}")
     
     documents = SimpleDirectoryReader(doc_dir).load_data()
-    embed_model = HuggingFaceEmbedding(model_name=settings.MODEL_NAME)
+    embed_model = BGEEmbedding(model_name=settings.MODEL_NAME)
 
     service_context = ServiceContext.from_defaults(embed_model=embed_model, llm=None)
 
+    # ✅ 指定正确维度：bge-m3 是 1024
+    faiss_index = faiss.IndexFlatL2(1024)
+    vector_store = FaissVectorStore(faiss_index=faiss_index)
+
     index = VectorStoreIndex.from_documents(
         documents,
-        vector_store=FaissVectorStore(),
+        vector_store=vector_store,
         service_context=service_context
     )
     
@@ -70,7 +54,7 @@ def build_user_index(user_id: str):
     print(f"[BUILD] 为用户 {user_id} 构建并保存了索引 → {index_path}")
 
 def query_user_rag(user_id: str, question: str, top_k: int = 4) -> str:
-    embed_model = HuggingFaceEmbedding(model_name=settings.MODEL_NAME)
+    embed_model = BGEEmbedding(model_name=settings.MODEL_NAME)
     service_context = ServiceContext.from_defaults(embed_model=embed_model, llm=None)
 
     all_nodes = []