From c25a9c05ca40ce6af43ed370827360e28b701f07 Mon Sep 17 00:00:00 2001
From: wangmeihua <13383952685@163.com>
Date: Wed, 13 Aug 2025 15:50:31 +0800
Subject: [PATCH] rag

---
 rag/uapi_service.py | 30 ++++++++++++++++++++++++------
 1 file changed, 24 insertions(+), 6 deletions(-)

diff --git a/rag/uapi_service.py b/rag/uapi_service.py
index 7b1c81f..80cbaf5 100644
--- a/rag/uapi_service.py
+++ b/rag/uapi_service.py
@@ -68,21 +68,39 @@ class APIService:
     async def extract_triples(self, request, text: str, upappid: str, apiname: str, user: str) -> list:
         """调用三元组抽取服务"""
         request_id = str(uuid.uuid4())
-        debug(f"Request #{request_id} started for triples extraction")
+        debug(f"Request #{request_id} started for triples extraction, text={text[:100]}")
         try:
+            # 清理输入文本，移除控制字符
+            import re
+            text = re.sub(r'[\x00-\x1F\x7F]', '', text)  # 移除 ASCII 控制字符
             uapi = UAPI(request, DictObject(**globals()))
             params_kw = {"text": text}
             b = await uapi.call(upappid, apiname, user, params_kw)
-            d = json.loads(b.decode('utf-8'))
-            if d.get("object") != "list":
-                error(f"request #{request_id} invalid response format: {d}")
+            if not b:
+                error(f"request #{request_id} 三元组抽取服务返回空响应: upappid={upappid}, apiname={apiname}")
+                raise RuntimeError("三元组抽取服务返回空响应")
+            try:
+                response_text = b.decode('utf-8')
+            except UnicodeDecodeError as decode_err:
+                error(f"request #{request_id} 响应解码失败: {str(decode_err)}, 原始响应: {b[:100]}")
+                raise RuntimeError(f"响应解码失败: {str(decode_err)}")
+            debug(f"request #{request_id} 三元组抽取服务原始响应: {response_text[:500]}")
+            # 清理响应字符串，移除控制字符
+            response_text = re.sub(r'[\x00-\x1F\x7F]', '', response_text)
+            try:
+                d = json.loads(response_text)
+            except json.JSONDecodeError as json_err:
+                error(f"request #{request_id} JSON 解析失败: {str(json_err)}, 响应内容: {response_text[:500]}")
+                raise RuntimeError(f"JSON 解析失败: {str(json_err)}")
+            if d.get("object") != "list" or not d.get("data"):
+                error(f"request #{request_id} 三元组抽取服务响应格式错误: {d}")
                 raise RuntimeError("三元组抽取服务响应格式错误")
             triples = d["data"]
             debug(f"request #{request_id} extracted {len(triples)} triples")
             return triples
         except Exception as e:
-            error(f"request #{request_id} failed to extract triples: {str(e)}")
-            raise RuntimeError(f"三元组抽取服务调用失败: {str(e)}")
+            error(f"request #{request_id} failed to extract triples: {str(e)}, upappid={upappid}, apiname={apiname}")
+            return []  # 根据业务需求返回空列表
 
     # 重排序服务 (BAAI/bge-reranker-v2-m3)
     async def rerank_results(self, request, query: str, results: list, top_n: int, upappid: str, apiname: str, user: str) -> list: