更新了chromadb的打印的符号

configs/kb_config.py.example

@@ -3,7 +3,7 @@ import os
 # 默认使用的知识库
 DEFAULT_KNOWLEDGE_BASE = "samples"
 
-# 默认向量库/全文检索引擎类型。可选：faiss, milvus(离线) & zilliz(在线), pgvector, 全文检索引擎es, chromadb
+# 默认向量库/全文检索引擎类型。可选：faiss, milvus(离线) & zilliz(在线), pgvector, chromadb 全文检索引擎es
 DEFAULT_VS_TYPE = "faiss"
 
 # 缓存向量库数量（针对FAISS）

requirements.txt

@@ -44,10 +44,11 @@ llama-index==0.9.35
 # dashscope==1.13.6 # qwen
 # volcengine==1.0.119 # fangzhou
 # uncomment libs if you want to use corresponding vector store
-# pymilvus==2.3.4
+# pymilvus==2.3.6
 # psycopg2==2.9.9
-# pgvector==0.2.4
+# pgvector>=0.2.4
 # chromadb==0.4.13
+
 #flash-attn==2.4.2 # For Orion-14B-Chat and Qwen-14B-Chat
 #autoawq==0.1.8 # For Int4
 #rapidocr_paddle[gpu]==1.3.11 # gpu accelleration for ocr of pdf and image files

requirements_api.txt

@@ -50,9 +50,11 @@ pyjwt==2.8.0
 # metaphor-python~=0.1.23
 
 # volcengine>=1.0.119
-# pymilvus>=2.3.4
+# pymilvus==2.3.6
 # psycopg2==2.9.9
 # pgvector>=0.2.4
+# chromadb==0.4.13
+
 #flash-attn==2.4.2 # For Orion-14B-Chat and Qwen-14B-Chat
 #autoawq==0.1.8 # For Int4
 #rapidocr_paddle[gpu]==1.3.11 # gpu accelleration for ocr of pdf and image files

requirements_lite.txt

@@ -33,6 +33,7 @@ watchdog~=3.0.0
 # psycopg2==2.9.9
 # pgvector>=0.2.4
 # chromadb==0.4.13
+
 # jq==1.6.0
 # beautifulsoup4~=4.12.2
 # pysrt~=1.1.2

server/knowledge_base/kb_service/chromadb_kb_service.py

@@ -15,7 +15,7 @@ def _get_result_to_documents(get_result: GetResult) -> List[Document]:
     if not get_result['documents']:
         return []
 
-    _metadatas = get_result['metadatas'] if get_result['metadatas'] else [{}] * len(get_result['documents']) 
+    _metadatas = get_result['metadatas'] if get_result['metadatas'] else [{}] * len(get_result['documents'])
 
     document_list = []
     for page_content, metadata in zip(get_result['documents'], _metadatas):
@@ -23,13 +23,13 @@ def _get_result_to_documents(get_result: GetResult) -> List[Document]:
 
     return document_list
 
+
 def _results_to_docs_and_scores(results: Any) -> List[Tuple[Document, float]]:
     """
     from langchain_community.vectorstores.chroma import Chroma
     """
     return [
         # TODO: Chroma can do batch querying,
-        # we shouldn't hard code to the 1st result
         (Document(page_content=result[0], metadata=result[1] or {}), result[2])
         for result in zip(
             results["documents"][0],
@@ -40,14 +40,12 @@ def _results_to_docs_and_scores(results: Any) -> List[Tuple[Document, float]]:
 
 
 class ChromaKBService(KBService):
-
     vs_path: str
     kb_path: str
 
     client = None
     collection = None
 
-
     def vs_type(self) -> str:
         return SupportedVSType.CHROMADB
 
@@ -75,16 +73,16 @@ class ChromaKBService(KBService):
             if not str(e) == f"Collection {self.kb_name} does not exist.":
                 raise e
 
-    def do_search(self, query: str, top_k: int, score_threshold: float = SCORE_THRESHOLD) -> List[Tuple[Document, float]]:
+    def do_search(self, query: str, top_k: int, score_threshold: float = SCORE_THRESHOLD) -> List[
+        Tuple[Document, float]]:
         embed_func = EmbeddingsFunAdapter(self.embed_model)
         embeddings = embed_func.embed_query(query)
         query_result: QueryResult = self.collection.query(query_embeddings=embeddings, n_results=top_k)
-        return  _results_to_docs_and_scores(query_result)
+        return _results_to_docs_and_scores(query_result)
 
     def do_add_doc(self, docs: List[Document], **kwargs) -> List[Dict]:
         doc_infos = []
         data = self._docs_to_embeddings(docs)
-        print(data)
         ids = [str(uuid.uuid1()) for _ in range(len(data["texts"]))]
         for _id, text, embedding, metadata in zip(ids, data["texts"], data["embeddings"], data["metadatas"]):
             self.collection.add(ids=_id, embeddings=embedding, metadatas=metadata, documents=text)

server/knowledge_base/kb_service/milvus_kb_service.py

@@ -2,6 +2,7 @@ from typing import List, Dict, Optional
 
 from langchain.schema import Document
 from langchain.vectorstores.milvus import Milvus
+import os
 
 from configs import kbs_config
 
@@ -85,6 +86,7 @@ class MilvusKBService(KBService):
     def do_delete_doc(self, kb_file: KnowledgeFile, **kwargs):
         if self.milvus.col:
             filepath = kb_file.filepath.replace('\\', '\\\\')
+            filename = os.path.basename(filepath)
             delete_list = [item.get("pk") for item in
                            self.milvus.col.query(expr=f'source == "{filepath}"', output_fields=["pk"])]
             self.milvus.col.delete(expr=f'pk in {delete_list}')