feat(知识库): 实现文件夹功能支持

server/routers/knowledge_router.py

@@ -394,10 +394,17 @@ async def get_document_content(db_id: str, doc_id: str, current_user: User = Dep
 
 @knowledge.delete("/databases/{db_id}/documents/{doc_id}")
 async def delete_document(db_id: str, doc_id: str, current_user: User = Depends(get_admin_user)):
-    """删除文档"""
+    """删除文档或文件夹"""
     logger.debug(f"DELETE document {doc_id} info in {db_id}")
     try:
         file_meta_info = await knowledge_base.get_file_basic_info(db_id, doc_id)
+
+        # Check if it is a folder
+        is_folder = file_meta_info.get("meta", {}).get("is_folder", False)
+        if is_folder:
+            await knowledge_base.delete_folder(db_id, doc_id)
+            return {"message": "文件夹删除成功"}
+
         file_name = file_meta_info.get("meta", {}).get("filename")
 
         # 尝试从MinIO删除文件，如果失败（例如旧知识库没有MinIO实例），则忽略
@@ -1150,6 +1157,39 @@ async def get_sample_questions(db_id: str, current_user: User = Depends(get_admi
 # =============================================================================
 
 
+@knowledge.post("/databases/{db_id}/folders")
+async def create_folder(
+    db_id: str,
+    folder_name: str = Body(..., embed=True),
+    parent_id: str | None = Body(None, embed=True),
+    current_user: User = Depends(get_admin_user),
+):
+    """创建文件夹"""
+    try:
+        return await knowledge_base.create_folder(db_id, folder_name, parent_id)
+    except Exception as e:
+        logger.error(f"创建文件夹失败 {e}, {traceback.format_exc()}")
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+@knowledge.put("/databases/{db_id}/documents/{doc_id}/move")
+async def move_document(
+    db_id: str,
+    doc_id: str,
+    new_parent_id: str | None = Body(..., embed=True),
+    current_user: User = Depends(get_admin_user),
+):
+    """移动文件或文件夹"""
+    logger.debug(f"Move document {doc_id} to {new_parent_id} in {db_id}")
+    try:
+        return await knowledge_base.move_file(db_id, doc_id, new_parent_id)
+    except ValueError as e:
+        raise HTTPException(status_code=400, detail=str(e))
+    except Exception as e:
+        logger.error(f"移动文件失败 {e}, {traceback.format_exc()}")
+        raise HTTPException(status_code=500, detail=str(e))
+
+
 @knowledge.post("/files/upload")
 async def upload_file(
     file: UploadFile = File(...),

src/knowledge/base.py

@@ -210,6 +210,26 @@ def delete_database(self, db_id: str) -> dict:
 
         return {"message": "删除成功"}
 
+    def create_folder(self, db_id: str, folder_name: str, parent_id: str | None = None) -> dict:
+        """Create a folder in the database."""
+        import uuid
+
+        folder_id = f"folder-{uuid.uuid4()}"
+
+        self.files_meta[folder_id] = {
+            "file_id": folder_id,
+            "filename": folder_name,
+            "is_folder": True,
+            "parent_id": parent_id,
+            "database_id": db_id,
+            "created_at": utc_isoformat(),
+            "status": "done",
+            "path": folder_name,
+            "file_type": "folder",
+        }
+        self._save_metadata()
+        return self.files_meta[folder_id]
+
     @abstractmethod
     async def add_content(self, db_id: str, items: list[str], params: dict | None = None) -> list[dict]:
         """
@@ -307,6 +327,8 @@ def get_database_info(self, db_id: str) -> dict | None:
                     "status": file_info.get("status", "done"),
                     "created_at": created_at,
                     "processing_params": file_info.get("processing_params", None),
+                    "is_folder": file_info.get("is_folder", False),
+                    "parent_id": file_info.get("parent_id", None),
                 }
 
         # 按创建时间倒序排序文件列表
@@ -350,6 +372,8 @@ def get_databases(self) -> dict:
                         "type": file_info.get("file_type", ""),
                         "status": file_info.get("status", "done"),
                         "created_at": created_at,
+                        "is_folder": file_info.get("is_folder", False),
+                        "parent_id": file_info.get("parent_id", None),
                     }
 
             # 按创建时间倒序排序文件列表
@@ -439,6 +463,72 @@ def _check_and_fix_processing_status(self, db_id: str) -> None:
         except Exception as e:
             logger.error(f"Error checking processing status for database {db_id}: {e}")
 
+    async def delete_folder(self, db_id: str, folder_id: str) -> None:
+        """
+        Recursively delete a folder and its content.
+
+        Args:
+            db_id: Database ID
+            folder_id: Folder ID to delete
+        """
+        # Find all children
+        children = [
+            fid
+            for fid, meta in self.files_meta.items()
+            if meta.get("database_id") == db_id and meta.get("parent_id") == folder_id
+        ]
+
+        for child_id in children:
+            child_meta = self.files_meta.get(child_id)
+            if child_meta and child_meta.get("is_folder"):
+                await self.delete_folder(db_id, child_id)
+            else:
+                await self.delete_file(db_id, child_id)
+
+        # Delete the folder itself
+        # We call delete_file which should handle the actual removal.
+        # Implementations should ensure they handle folder deletion gracefully (e.g. skip vector deletion)
+        await self.delete_file(db_id, folder_id)
+
+    async def move_file(self, db_id: str, file_id: str, new_parent_id: str | None) -> dict:
+        """
+        Move a file or folder to a new parent folder.
+
+        Args:
+            db_id: Database ID
+            file_id: File/Folder ID to move
+            new_parent_id: New parent folder ID (None for root)
+
+        Returns:
+            dict: Updated metadata
+        """
+        if file_id not in self.files_meta:
+            raise ValueError(f"File {file_id} not found")
+
+        meta = self.files_meta[file_id]
+        if meta.get("database_id") != db_id:
+            raise ValueError(f"File {file_id} does not belong to database {db_id}")
+
+        # Basic cycle detection for folders
+        if meta.get("is_folder") and new_parent_id:
+            # Check if new_parent_id is a child of file_id (or is file_id itself)
+            if new_parent_id == file_id:
+                raise ValueError("Cannot move a folder into itself")
+
+            # Walk up the tree from new_parent_id
+            current = new_parent_id
+            while current:
+                parent_meta = self.files_meta.get(current)
+                if not parent_meta:
+                    break  # Should not happen if integrity is maintained
+                if current == file_id:
+                    raise ValueError("Cannot move a folder into its own subfolder")
+                current = parent_meta.get("parent_id")
+
+        meta["parent_id"] = new_parent_id
+        self._save_metadata()
+        return meta
+
     @abstractmethod
     async def delete_file(self, db_id: str, file_id: str) -> None:
         """

src/knowledge/manager.py

@@ -174,6 +174,13 @@ def _get_or_create_kb_instance(self, kb_type: str) -> KnowledgeBase:
         logger.info(f"Created {kb_type} knowledge base instance")
         return kb_instance
 
+    async def move_file(self, db_id: str, file_id: str, new_parent_id: str | None) -> dict:
+        """
+        移动文件/文件夹
+        """
+        kb_instance = self._get_kb_for_database(db_id)
+        return await kb_instance.move_file(db_id, file_id, new_parent_id)
+
     def _get_kb_for_database(self, db_id: str) -> KnowledgeBase:
         """
         根据数据库ID获取对应的知识库实例
@@ -220,8 +227,13 @@ def get_databases(self) -> dict:
 
         return {"databases": all_databases}
 
+    async def create_folder(self, db_id: str, folder_name: str, parent_id: str = None) -> dict:
+        """Create a folder in the database."""
+        kb_instance = self._get_kb_for_database(db_id)
+        return kb_instance.create_folder(db_id, folder_name, parent_id)
+
     async def create_database(
-        self, database_name: str, description: str, kb_type: str, embed_info: dict | None = None, **kwargs
+        self, database_name: str, description: str, kb_type: str = "lightrag", embed_info: dict | None = None, **kwargs
     ) -> dict:
         """
         创建数据库
@@ -315,6 +327,11 @@ def get_database_info(self, db_id: str) -> dict | None:
         except KBNotFoundError:
             return None
 
+    async def delete_folder(self, db_id: str, folder_id: str) -> None:
+        """递归删除文件夹"""
+        kb_instance = self._get_kb_for_database(db_id)
+        await kb_instance.delete_folder(db_id, folder_id)
+
     async def delete_file(self, db_id: str, file_id: str) -> None:
         """删除文件"""
         kb_instance = self._get_kb_for_database(db_id)

src/knowledge/utils/kb_utils.py

@@ -213,6 +213,7 @@ async def prepare_item_metadata(item: str, content_type: str, db_id: str, params
         "created_at": utc_isoformat(),
         "file_id": file_id,
         "content_hash": content_hash,
+        "parent_id": params.get("parent_id") if params else None,
     }
 
     # 保存处理参数到元数据

web/src/apis/knowledge_api.js

@@ -74,6 +74,34 @@ export const databaseApi = {
 // =============================================================================
 
 export const documentApi = {
+  /**
+   * 创建文件夹
+   * @param {string} dbId - 知识库ID
+   * @param {string} folderName - 文件夹名称
+   * @param {string} parentId - 父文件夹ID
+   * @returns {Promise} - 创建结果
+   */
+  createFolder: async (dbId, folderName, parentId = null) => {
+    return apiAdminPost(`/api/knowledge/databases/${dbId}/folders`, {
+      folder_name: folderName,
+      parent_id: parentId
+    })
+  },
+
+  /**
+   * 移动文档/文件夹
+   * @param {string} dbId - 知识库ID
+   * @param {string} docId - 文档/文件夹ID
+   * @param {string} newParentId - 新的父文件夹ID
+   * @returns {Promise} - 移动结果
+   */
+  moveDocument: async (dbId, docId, newParentId) => {
+    return apiAdminPut(`/api/knowledge/databases/${dbId}/documents/${docId}/move`, {
+      new_parent_id: newParentId
+    })
+  },
+
+
   /**
    * 添加文档到知识库
    * @param {string} dbId - 知识库ID