.gitignore

@@ -19,7 +19,7 @@ Cargo.lock
 
 # MSVC Windows builds of rustc generate these, which store debugging information
 *.pdb
-.trie
+*.trie
 
 .idea/
 .vscode/

docs/question/README.md

@@ -70,34 +70,6 @@ deploy:
 
 **回答：** Ragflow原生解析器心跳触发的问题，不影响正常使用，可忽略，官方回答可参考：https://github.com/infiniflow/ragflow/issues/6700
 
-## 问题 11：为什么添加ollama时，无法联通？
-
-**回答：** ollama需要预先设置为对所有网络接口开放
-
-修改配置文件：
-```bash
-vim /etc/systemd/system/ollama.service
-```
-
-[Service] 下添加：
-
-```bash
-Environment="OLLAMA_HOST=0.0.0.0"
-```
-
-重新载入配置文件，重启ollama。
-
-```bash
-systemctl daemon-reload
-systemctl restart ollama
-```
-
-## 问题 12：在后台知识库连接测试中显示 text-embedding-v3，无法连通？
-
-**回答：** 出现此情况原因是使用 ragflow 创建的初始用户(创建时间最早的用户)，默认自带了通义千问的模型配置，可登陆此用户，在模型管理中移除通义千问的模型配置，并添加新的嵌入模型配置，后台会自动读取最新添加的模型配置信息。
-
-嵌入模型仅支持 bge-m3 模型，联网API仅支持硅基流动平台，可免费调用该嵌入模型：https://cloud.siliconflow.cn/i/bjDoFhPf
-
 --- 
 
 

management/server/routes/knowledgebases/routes.py

@@ -159,6 +159,24 @@ 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"""
@@ -224,8 +242,8 @@ def set_system_embedding_config_route():
 
 
 @knowledgebase_bp.route("/documents/<doc_id>/parse", methods=["POST"])
-def parse_document(doc_id):
+def parse_document_async(doc_id):  # 函数名改为 async 以区分
-    """开始解析文档"""
+    """开始异步解析单个文档"""
     if request.method == "OPTIONS":
         response = success_response({})
         response.headers.add("Access-Control-Allow-Methods", "POST")
@@ -233,7 +251,7 @@ def parse_document(doc_id):
         return response
 
     try:
-        result = KnowledgebaseService.parse_document(doc_id)
+        result = KnowledgebaseService.parse_document(doc_id)  # 调用同步版本
         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

@@ -1,9 +1,7 @@
-from typing import Any, Dict, Type, TypeVar
-
 from peewee import Model
+from typing import Type, TypeVar, Dict, Any
 
-T = TypeVar("T", bound=Model)
+T = TypeVar('T', bound=Model)
-
 
 class BaseService:
     model: Type[T]
@@ -22,4 +20,6 @@ class BaseService:
     
     @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

@@ -1,15 +1,16 @@
 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]:

management/server/services/knowledgebases/__init__.py

@@ -1,4 +1,3 @@
 import logging
 
 logger = logging.getLogger(__name__)
-logger.setLevel(logging.INFO)

management/server/services/knowledgebases/document_parser.py

@@ -484,8 +484,11 @@ def perform_parse(doc_id, doc_info, file_info, embedding_config, kb_info):
 
         # 4. 更新文本块的图像信息
         if image_info_list and chunk_ids_list:
-
+            conn = None
+            cursor = None
             try:
+                conn = get_db_connection()
+                cursor = conn.cursor()
 
                 # 为每个文本块找到最近的图片
                 for i, chunk_id in enumerate(chunk_ids_list):
@@ -513,7 +516,11 @@ def perform_parse(doc_id, doc_info, file_info, embedding_config, kb_info):
             except Exception as e:
                 logger.error(f"[Parser-ERROR] 更新文本块图片关联失败: {e}")
                 raise Exception(f"[Parser-ERROR] 更新文本块图片关联失败: {e}")
-
+            finally:
+                if cursor:
+                    cursor.close()
+                if conn:
+                    conn.close()
 
         # 5. 更新最终状态
         process_duration = time.time() - start_time

management/server/services/knowledgebases/service.py

@@ -711,19 +711,12 @@ class KnowledgebaseService:
             cursor = conn.cursor(dictionary=True)
 
             # 先检查文档是否存在
-            # check_query = """
-            #     SELECT 
-            #         d.kb_id, 
-            #         kb.created_by AS tenant_id  -- 获取 tenant_id (knowledgebase的创建者)
-            #     FROM document d
-            #     JOIN knowledgebase kb ON d.kb_id = kb.id -- JOIN knowledgebase 表
-            #     WHERE d.id = %s
-            # """
             check_query = """
                 SELECT 
                     d.kb_id, 
-                    d.created_by AS tenant_id
+                    kb.created_by AS tenant_id  -- 获取 tenant_id (knowledgebase的创建者)
                 FROM document d
+                JOIN knowledgebase kb ON d.kb_id = kb.id -- JOIN knowledgebase 表
                 WHERE d.id = %s
             """
             cursor.execute(check_query, (doc_id,))
@@ -1048,7 +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
+                ORDER BY create_time DESC  # 如果一个用户可能有多个embedding配置，取最早的
                 LIMIT 1
             """
             cursor.execute(query_embedding_config, (earliest_user_id,))

management/web/src/common/apis/tables/index.ts

@@ -11,7 +11,7 @@ export function createTableDataApi(data: Tables.CreateOrUpdateTableRequestData)
 }
 
 /** 删 */
-export function deleteTableDataApi(id: string) {
+export function deleteTableDataApi(id: number) {
   return request({
     url: `api/v1/users/${id}`,
     method: "delete"
@@ -42,7 +42,7 @@ export function getTableDataApi(params: Tables.TableRequestData) {
  * @param password 新密码
  * @returns BaseResponse
  */
-export function resetPasswordApi(userId: string, password: string) {
+export function resetPasswordApi(userId: number, password: string) {
   return request({
     url: `api/v1/users/${userId}/reset-password`,
     method: "put",

management/web/src/common/apis/tables/type.ts

@@ -1,5 +1,5 @@
 export interface CreateOrUpdateTableRequestData {
-  id?: string
+  id?: number
   username: string
   email?: string
   password?: string
@@ -21,7 +21,7 @@ export interface TableRequestData {
 }
 
 export interface TableData {
-  id: string
+  id: number
   username: string
   email: string
   createTime: string

management/web/src/pages/user-management/index.vue

@@ -39,7 +39,7 @@ const formRules: FormRules<CreateOrUpdateTableRequestData> = {
 // #region 重置密码
 const resetPasswordDialogVisible = ref<boolean>(false)
 const resetPasswordFormRef = ref<FormInstance | null>(null)
-const currentUserId = ref<string | undefined>(undefined) // 用于存储当前要重置密码的用户ID
+const currentUserId = ref<number | undefined>(undefined) // 用于存储当前要重置密码的用户ID
 const resetPasswordFormData = reactive({
   password: ""
 })