chores(knowledgebases): 清理相关冗余代码

- 删除了多余的 parse_document 函数
- 重命名 parse_document_async 为 parse_document，作为主要的文档解析接口

management/server/routes/knowledgebases/routes.py

@@ -159,24 +159,6 @@ def delete_document(doc_id):
         return error_response(str(e))
 
 
-@knowledgebase_bp.route("/documents/<doc_id>/parse", methods=["POST"])
-def parse_document(doc_id):
-    """开始解析文档"""
-    # 处理 OPTIONS 预检请求
-    if request.method == "OPTIONS":
-        response = success_response({})
-        # 添加 CORS 相关头
-        response.headers.add("Access-Control-Allow-Methods", "POST")
-        response.headers.add("Access-Control-Allow-Headers", "Content-Type,Authorization")
-        return response
-
-    try:
-        result = KnowledgebaseService.async_parse_document(doc_id)
-        return success_response(data=result)
-    except Exception as e:
-        return error_response(str(e), code=500)
-
-
 @knowledgebase_bp.route("/documents/<doc_id>/parse/progress", methods=["GET"])
 def get_parse_progress(doc_id):
     """获取文档解析进度"""
@@ -242,8 +224,8 @@ def set_system_embedding_config_route():
 
 
 @knowledgebase_bp.route("/documents/<doc_id>/parse", methods=["POST"])
-def parse_document_async(doc_id):  # 函数名改为 async 以区分
-    """开始异步解析单个文档"""
+def parse_document(doc_id):
+    """开始解析文档"""
     if request.method == "OPTIONS":
         response = success_response({})
         response.headers.add("Access-Control-Allow-Methods", "POST")
@@ -251,7 +233,7 @@ def parse_document_async(doc_id):  # 函数名改为 async 以区分
         return response
 
     try:
-        result = KnowledgebaseService.parse_document(doc_id)  # 调用同步版本
+        result = KnowledgebaseService.parse_document(doc_id)
         if result.get("success"):
             return success_response(data={"message": f"文档 {doc_id} 同步解析完成。", "details": result})
         else:

management/server/services/files/base_service.py

@@ -1,25 +1,25 @@
-from peewee import Model
-from typing import Type, TypeVar, Dict, Any
+from typing import Any, Dict, Type, TypeVar
+
+from peewee import Model
+
+T = TypeVar("T", bound=Model)
 
-T = TypeVar('T', bound=Model)
 
 class BaseService:
     model: Type[T]
-    
+
     @classmethod
     def get_by_id(cls, id: str) -> T:
         return cls.model.get_by_id(id)
-    
-    @classmethod 
+
+    @classmethod
     def insert(cls, data: Dict[str, Any]) -> T:
         return cls.model.create(**data)
-    
+
     @classmethod
     def delete_by_id(cls, id: str) -> int:
         return cls.model.delete().where(cls.model.id == id).execute()
-    
+
     @classmethod
     def query(cls, **kwargs) -> list[T]:
-        return list(cls.model.select().where(*[
-            getattr(cls.model, k) == v for k, v in kwargs.items()
-        ]))
+        return list(cls.model.select().where(*[getattr(cls.model, k) == v for k, v in kwargs.items()]))

management/server/services/files/file2document_service.py

@@ -1,21 +1,20 @@
 from peewee import *
+
 from .base_service import BaseService
 from .models import File2Document
 
+
 class File2DocumentService(BaseService):
     model = File2Document
-    
+
     @classmethod
     def create_mapping(cls, file_id: str, document_id: str) -> File2Document:
-        return cls.insert({
-            'file_id': file_id,
-            'document_id': document_id
-        })
-    
+        return cls.insert({"file_id": file_id, "document_id": document_id})
+
     @classmethod
     def get_by_document_id(cls, document_id: str) -> list[File2Document]:
         return cls.query(document_id=document_id)
-    
+
     @classmethod
     def get_by_file_id(cls, file_id: str) -> list[File2Document]:
-        return cls.query(file_id=file_id)
+        return cls.query(file_id=file_id)

management/server/services/knowledgebases/service.py

@@ -1041,7 +1041,7 @@ class KnowledgebaseService:
                 SELECT llm_name, api_key, api_base
                 FROM tenant_llm
                 WHERE tenant_id = %s AND model_type = 'embedding'
-                ORDER BY create_time DESC  # 如果一个用户可能有多个embedding配置，取最早的
+                ORDER BY create_time DESC
                 LIMIT 1
             """
             cursor.execute(query_embedding_config, (earliest_user_id,))