rag

rag/uapi_service.py

@@ -1,3 +1,4 @@
+from traceback import format_exc
 from appPublic.log import debug, error
 from typing import Dict, Any, List
 import uuid
@@ -15,23 +16,12 @@ class APIService:
         log_prefix = f"request #{request_id} " if request_id else ""
         if not b:
             error(f"{log_prefix}{service_name} 返回空响应: upappid={upappid}, apiname={apiname}")
-            raise RuntimeError(f"{service_name} 返回空响应")
+            raise RuntimeError(f"{service_name} 返回空响应\n{format_exc()}")
         try:
-            response_text = b.decode('utf-8')
+            d = json.loads(b.decode('utf-8'))
-        except UnicodeDecodeError as decode_err:
+        except Exception as e:
-            error(f"{log_prefix}{service_name} 响应解码失败: {str(decode_err)}, 原始响应: {b[:100]}")
+            error(
-            raise RuntimeError(f"响应解码失败: {str(decode_err)}")
+                f"request #{request_id} JSON 解析失败: {str(e)}, upappid={upappid}, apiname={apiname}\n{format_exc()}")
-        # 清理响应中的控制字符
-        response_text = re.sub(r'[\x00-\x1F\x7F]', '', response_text)
-        debug(f"{log_prefix}{service_name} 原始响应: {response_text[:500]}")
-        try:
-            d = json.loads(response_text)
-        except json.JSONDecodeError as json_err:
-            error(f"{log_prefix}{service_name} 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 not isinstance(d, dict):
-            error(f"{log_prefix}{service_name} 响应不是有效字典: {d}")
-            raise RuntimeError(f"{service_name} 响应格式错误")
         return d
 
     # 嵌入服务 (BAAI/bge-m3)
@@ -40,8 +30,6 @@ class APIService:
         request_id = str(uuid.uuid4())
         debug(f"Request #{request_id} started for embeddings, texts={texts[:2]}")
         try:
-            # 清理输入文本
-            texts = [re.sub(r'[\x00-\x1F\x7F]', '', text) for text in texts]
             uapi = UAPI(request, DictObject(**globals()))
             params_kw = {"input": texts}
             b = await uapi.call(upappid, apiname, user, params_kw)
@@ -64,7 +52,6 @@ class APIService:
         try:
             if not query:
                 raise ValueError("查询文本不能为空")
-            query = re.sub(r'[\x00-\x1F\x7F]', '', query)  # 清理输入
             uapi = UAPI(request, DictObject(**globals()))
             params_kw = {"query": query}
             b = await uapi.call(upappid, apiname, user, params_kw)
@@ -86,7 +73,6 @@ class APIService:
         request_id = str(uuid.uuid4())
         debug(f"Request #{request_id} started for triples extraction, text={text[:100]}")
         try:
-            text = re.sub(r'[\x00-\x1F\x7F]', '', text)  # 清理输入
             uapi = UAPI(request, DictObject(**globals()))
             params_kw = {"text": text}
             b = await uapi.call(upappid, apiname, user, params_kw)
@@ -115,9 +101,8 @@ class APIService:
                 top_n = len(results)
             else:
                 top_n = min(top_n, len(results))
-            documents = [re.sub(r'[\x00-\x1F\x7F]', '', result.get("text", str(result))) for result in results]
-            query = re.sub(r'[\x00-\x1F\x7F]', '', query)
             uapi = UAPI(request, DictObject(**globals()))
+            documents = [result.get("text", str(result)) for result in results]
             params_kw = {
                 "model": "rerank-001",
                 "query": query,
@@ -232,7 +217,6 @@ class APIService:
         request_id = str(uuid.uuid4())
         debug(f"Request #{request_id} started for Neo4j triplets matching, query={query[:100]}")
         try:
-            query = re.sub(r'[\x00-\x1F\x7F]', '', query)
             uapi = UAPI(request, DictObject(**globals()))
             params_kw = {
                 "query": query,