feat(kb): 接入 Dify 知识库，移除 FastGPT；AI 聊天产出+用户文档自动入库量化

- 新增 app/dify_kb.py Dify v1 客户端(连通测试/检索/文本入库)
- kb_router 仅保留 Dify 外部检索分支
- knowledge_admin 去掉 FastGPT/RAGFlow，新增 dify 批量入库端点
- chat.py 注入 Dify RAG 上下文 + 助手回答 fire-and-forget 入库
- 前端 Knowledge.tsx 改为 Dify 单服务 UI(可通讯/可用验证+入库按钮)
- DB: external_kb_services 删除 fastgpt，qihuai 设为 https+connected

backend/python/app/kb_router.py

@@ -1,7 +1,7 @@
 """
 知识库路由层
 ============
-将检索请求智能路由到本地BM25、向量检索、或外部KB服务(FastGPT/Dify/RAGFlow)。
+将检索请求智能路由到本地BM25、向量检索、或外部KB服务(Dify)。
 根据知识库配置自动选择最优检索策略。
 
 使用方式:
@@ -9,17 +9,14 @@
     results = await kb_router_search(query, kb_id, top_k=5)
 """
 
-import os
-import json
 import logging
-import httpx
 from typing import List, Dict, Optional
 
 logger = logging.getLogger("aiflow.kb_router")
 
 
 async def _get_external_services(db_session, enabled_only: bool = True) -> List[Dict]:
-    """获取已配置的外部KB服务"""
+    """获取已配置的外部KB服务（仅 dify）"""
     from sqlalchemy import text
     try:
         query = "SELECT * FROM external_kb_services"
@@ -34,123 +31,34 @@ async def _get_external_services(db_session, enabled_only: bool = True) -> List[
         return []
 
 
-async def _search_fastgpt(api_url: str, api_key: str, query: str, top_k: int = 5) -> List[Dict]:
-    """搜索 FastGPT 知识库"""
+async def _search_dify(service: Dict, query: str, top_k: int = 5) -> List[Dict]:
+    """检索 Dify 知识库（走 app.dify_kb 客户端）"""
     try:
-        async with httpx.AsyncClient(timeout=30) as client:
-            resp = await client.post(
-                f"{api_url.rstrip('/')}/api/v1/dataset/search",
-                headers={
-                    "Authorization": f"Bearer {api_key}",
-                    "Content-Type": "application/json",
-                },
-                json={
-                    "query": query,
-                    "similarity": 0.5,
-                    "limit": top_k,
-                },
-            )
-            if resp.status_code == 200:
-                data = resp.json()
-                results = []
-                for item in data.get("data", {}).get("list", []):
-                    results.append({
-                        "content": item.get("q", item.get("content", "")),
-                        "score": float(item.get("score", 0.5)),
-                        "source": f"FastGPT: {item.get('collectionId', '')}",
-                        "metadata": item,
-                    })
-                return results
-            else:
-                logger.warning(f"FastGPT search returned {resp.status_code}")
-                return []
-    except Exception as e:
-        logger.error(f"FastGPT search error: {e}")
-        return []
-
-
-async def _search_dify(api_url: str, api_key: str, query: str, top_k: int = 5) -> List[Dict]:
-    """搜索 Dify 知识库"""
-    try:
-        async with httpx.AsyncClient(timeout=30) as client:
-            resp = await client.post(
-                f"{api_url.rstrip('/')}/v1/datasets/retrieval",
-                headers={
-                    "Authorization": f"Bearer {api_key}",
-                    "Content-Type": "application/json",
-                },
-                json={
-                    "query": query,
-                    "retrieval_mode": "semantic",
-                    "top_k": top_k,
-                },
-            )
-            if resp.status_code == 200:
-                data = resp.json()
-                results = []
-                for item in data.get("records", []):
-                    segment = item.get("segment", {})
-                    results.append({
-                        "content": segment.get("content", ""),
-                        "score": float(item.get("score", 0.5)),
-                        "source": f"Dify: {segment.get('id', '')}",
-                        "metadata": item,
-                    })
-                return results
-            else:
-                logger.warning(f"Dify search returned {resp.status_code}")
-                return []
+        from app.dify_kb import dify_retrieve
+        api_url = service.get("api_url", "")
+        api_key = service.get("api_key", "")
+        dataset_name = None
+        cfg = service.get("config")
+        if isinstance(cfg, str):
+            try:
+                import json
+                cfg = json.loads(cfg)
+            except Exception:
+                cfg = None
+        if isinstance(cfg, dict):
+            dataset_name = cfg.get("dataset_name")
+        hits = dify_retrieve(api_url, api_key, query, top_k=top_k, dataset_name=dataset_name)
+        return hits
     except Exception as e:
         logger.error(f"Dify search error: {e}")
         return []
 
 
-async def _search_ragflow(api_url: str, api_key: str, query: str, top_k: int = 5) -> List[Dict]:
-    """搜索 RAGFlow 知识库"""
-    try:
-        async with httpx.AsyncClient(timeout=30) as client:
-            resp = await client.post(
-                f"{api_url.rstrip('/')}/api/v1/retrieval",
-                headers={
-                    "Authorization": f"Bearer {api_key}",
-                    "Content-Type": "application/json",
-                },
-                json={
-                    "question": query,
-                    "top_k": top_k,
-                },
-            )
-            if resp.status_code == 200:
-                data = resp.json()
-                results = []
-                for item in data.get("data", {}).get("chunks", []):
-                    results.append({
-                        "content": item.get("content", ""),
-                        "score": float(item.get("similarity", 0.5)),
-                        "source": f"RAGFlow: {item.get('document_id', '')}",
-                        "metadata": item,
-                    })
-                return results
-            else:
-                logger.warning(f"RAGFlow search returned {resp.status_code}")
-                return []
-    except Exception as e:
-        logger.error(f"RAGFlow search error: {e}")
-        return []
-
-
 async def _search_external(service: Dict, query: str, top_k: int = 5) -> List[Dict]:
-    """路由到对应的外部KB服务"""
+    """路由到对应的外部KB服务（当前仅支持 dify）"""
     service_type = service.get("service_type", "")
-    api_url = service.get("api_url", "")
-    api_key = service.get("api_key", "")
-
-    if service_type == "fastgpt":
-        return await _search_fastgpt(api_url, api_key, query, top_k)
-    elif service_type == "dify":
-        return await _search_dify(api_url, api_key, query, top_k)
-    elif service_type == "ragflow":
-        return await _search_ragflow(api_url, api_key, query, top_k)
+    if service_type == "dify":
+        return await _search_dify(service, query, top_k)
     else:
         logger.warning(f"Unknown external KB type: {service_type}")
         return []
@@ -171,7 +79,7 @@ async def kb_router_search(
     检索策略:
     1. 本地 BM25 检索 (MySQL, 已有)
     2. 向量语义检索 (PostgreSQL+pgvector, 新增)
-    3. 外部KB检索 (FastGPT/Dify/RAGFlow, 按配置)
+    3. 外部KB检索 (Dify, 按配置)
 
     融合方式: Reciprocal Rank Fusion (RRF)
 
@@ -218,7 +126,7 @@ async def kb_router_search(
         except Exception as e:
             logger.warning(f"Vector search failed: {e}")
 
-    # 3. 外部KB检索
+    # 3. 外部KB检索 (Dify)
     if use_external and db_session:
         services = await _get_external_services(db_session)
         for service in services:
@@ -228,7 +136,7 @@ async def kb_router_search(
                     all_results.append({
                         "content": item.get("content", ""),
                         "score": item.get("score", 0.5),
-                        "source": item.get("source", f"external_{service['service_type']}"),
+                        "source": item.get("source", "external_dify"),
                         "rank": rank + 1,
                         "metadata": item.get("metadata", {}),
                     })