feat: implement knowledge update functionality for bots

- Added a new hook `useKnowledgeUpdate` to manage knowledge updates.
- Integrated knowledge update feature in `BotCard` component with loading state.
- Created `updateKnowledge` API in `BotsController` to handle knowledge updates.
- Updated FastAPI router to include endpoint for knowledge updates, validating bot existence.
- Refactored retrieval logic to check and update knowledge based on document changes.

Author: sedationh

client/app/factory/list/components/BotCard.tsx

@@ -25,6 +25,7 @@ import ErrorBadgeIcon from '@/public/icons/ErrorBadgeIcon';
 import CheckBadgeIcon from '@/public/icons/CheckBadgeIcon';
 import LoadingIcon from '@/public/icons/LoadingIcon';
 import { RagTask } from '@/app/services/BotsController';
+import { useKnowledgeUpdate } from '@/app/hooks/useKnowledgeUpdate';
 
 declare type Bot = Tables<'bots'>;
 
@@ -34,6 +35,8 @@ const BotCard = (props: { bot: Bot }) => {
   const router = useRouter();
   const { deleteBot, isLoading, isSuccess } = useBotDelete();
   const { data: taskInfo } = useGetBotRagTask(bot.id, true, false);
+  const { mutate: updateKnowledge, isPending: isUpdating } =
+    useKnowledgeUpdate();
 
   useEffect(() => {
     if (isSuccess) {
@@ -69,6 +72,19 @@ const BotCard = (props: { bot: Bot }) => {
     );
   };
 
+  const handleUpdateKnowledge = () => {
+    updateKnowledge(
+      {
+        bot_id: bot.id,
+      },
+      {
+        onSuccess: () => {
+          // TODO
+        },
+      },
+    );
+  };
+
   return (
     <>
       <Card
@@ -124,21 +140,30 @@ const BotCard = (props: { bot: Bot }) => {
                 placement="top"
                 content={I18N.components.BotCard.gengXinZhiShiKu}
                 classNames={{
-                  base: [
-                    // arrow color
-                    'before:bg-[#3F3F46] dark:before:bg-white',
-                  ],
+                  base: ['before:bg-[#3F3F46] dark:before:bg-white'],
                   content: [
                     'py-2 px-4 rounded-lg shadow-xl text-white',
                     'bg-[#3F3F46]',
                   ],
                 }}
               >
-                <Image
-                  src="../images/refresh.svg"
-                  alt={I18N.components.BotCard.gengXinZhiShi}
-                  className="z-10 cursor-pointer"
-                />
+                {isUpdating ? (
+                  <div className="w-6 h-6 flex items-center justify-center">
+                    <span className="animate-spinner-ease-spin">
+                      <LoadingIcon />
+                    </span>
+                  </div>
+                ) : (
+                  <Image
+                    src="../images/refresh.svg"
+                    onClick={(e) => {
+                      e.stopPropagation();
+                      handleUpdateKnowledge();
+                    }}
+                    alt={I18N.components.BotCard.gengXinZhiShi}
+                    className="z-10 cursor-pointer"
+                  />
+                )}
               </Tooltip>
             </div>
           </div>

client/app/hooks/useKnowledgeUpdate.ts

@@ -0,0 +1,9 @@
+import { useMutation } from '@tanstack/react-query';
+import { updateKnowledge } from '../services/BotsController';
+
+export function useKnowledgeUpdate() {
+  return useMutation({
+    mutationKey: ['updateKnowledge'],
+    mutationFn: updateKnowledge
+  });
+}

client/app/services/BotsController.ts

@@ -135,3 +135,10 @@ export async function bindBotToRepo(repsConfigs: BindBotToRepoConfig[]) {
   });
   return response.data;
 }
+
+// Add knowledge update API
+export async function updateKnowledge(config: {
+  bot_id: string;
+}) {
+  return axios.post(`${apiDomain}/api/rag/update_knowledge`, config);
+}

petercat_utils/rag_helper/git_doc_task.py

@@ -142,7 +142,7 @@ def handle_tree_node(self):
         )
 
     def handle_blob_node(self):
-        retrieval.add_knowledge_by_doc(
+        retrieval.check_and_update_knowledge(
             RAGGitDocConfig(
                 repo_name=self.repo_name,
                 file_path=self.path,

petercat_utils/rag_helper/retrieval.py

@@ -197,3 +197,40 @@ def get_chunk_list(repo_name: str, page_size: int, page_number: int):
     )
     total_count = len(count_response.data)
     return {"rows": query.data, "total": total_count}
+
+
+def check_and_update_knowledge(config: RAGGitDocConfig):
+    # 初始化 GitHub loader 获取最新的文件信息
+    loader = init_github_file_loader(config)
+    latest_sha = loader.file_sha
+    
+    # 获取当前存储的文档
+    client = get_client()
+    existing_docs = (
+        client.table(TABLE_NAME)
+        .select("id, file_sha")
+        .eq("repo_name", config.repo_name)
+        .eq("file_path", config.file_path)
+        .execute()
+    )
+
+    if not existing_docs.data:
+        # 如果不存在文档，直接添加
+        return add_knowledge_by_doc(config)
+    
+    # 检查 SHA 是否变化
+    current_sha = existing_docs.data[0]["file_sha"]
+    
+    if current_sha == latest_sha:
+        return False
+    
+    # SHA 不同，需要更新
+    # 1. 删除旧文档
+    client.table(TABLE_NAME)\
+        .delete()\
+        .eq("repo_name", config.repo_name)\
+        .eq("file_path", config.file_path)\
+        .execute()
+        
+    # 2. 添加新文档
+    return add_knowledge_by_doc(config)

server/rag/router.py

@@ -1,6 +1,8 @@
 import json
-from typing import Optional
+from typing import Annotated, Optional
+from pydantic import BaseModel
 
+from auth.get_user_info import get_user_id
 from fastapi import APIRouter, Depends
 from petercat_utils.db.client.supabase import get_client
 
@@ -116,3 +118,42 @@ def get_rag_task(repo_name: str):
         return response
     except Exception as e:
         return json.dumps({"success": False, "message": str(e)})
+
+class UpdateKnowledgeRequest(BaseModel):
+    bot_id: str
+
+@router.post("/rag/update_knowledge", dependencies=[Depends(verify_rate_limit)])
+def update_knowledge(request: UpdateKnowledgeRequest, user_id: Annotated[str | None, Depends(get_user_id)] = None):
+    try:
+        # Get config from database using bot_id
+        supabase = get_client()
+        response = (
+            supabase.table("bots")
+            .select("*")
+            .eq("id", request.bot_id)
+            .eq("uid", user_id)
+            .single()
+            .execute()
+        )
+        
+        if not response.data:
+            return json.dumps({
+                "success": False,
+                "message": f"Bot with id {request.bot_id} not found"
+            })
+            
+        bot_config = RAGGitDocConfig(**response.data)
+        result = retrieval.check_and_update_knowledge(bot_config)
+        
+        if result:
+            return json.dumps({
+                "success": True,
+                "message": "Knowledge updated successfully!"
+            })
+        else:
+            return json.dumps({
+                "success": False,
+                "message": "Knowledge not updated!"
+            })
+    except Exception as e:
+        return json.dumps({"success": False, "message": str(e)})