xparse-client 0.2.17__tar.gz → 0.2.19__tar.gz

{xparse_client-0.2.17 → xparse_client-0.2.19}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: xparse-client
-Version: 0.2.17
+Version: 0.2.19
 Summary: 面向Agent和RAG的新一代文档处理 AI Infra
 License-Expression: MIT
 Project-URL: Homepage, https://gitlab.intsig.net/xparse1/xparse-pipeline

{xparse_client-0.2.17 → xparse_client-0.2.19}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "xparse-client"
-version = "0.2.17"
+version = "0.2.19"
 description = "面向Agent和RAG的新一代文档处理 AI Infra"
 readme = "README.md"
 license = "MIT"

{xparse_client-0.2.17 → xparse_client-0.2.19}/xparse_client/pipeline/pipeline.py

@@ -3,6 +3,7 @@
 
 import json
 import logging
+import re
 import time
 from datetime import datetime, timezone
 from pathlib import Path
@@ -220,6 +221,67 @@ class Pipeline:
         
         return config
 
+    def _extract_error_message(self, response: requests.Response) -> Tuple[str, str]:
+        """
+        从响应中提取规范化的错误信息
+        
+        Returns:
+            Tuple[str, str]: (error_msg, x_request_id)
+        """
+        # 首先尝试从响应头中提取 x-request-id（requests的headers大小写不敏感）
+        x_request_id = response.headers.get('x-request-id', '')
+        error_msg = ''
+        
+        # 获取Content-Type
+        content_type = response.headers.get('Content-Type', '').lower()
+        
+        # 尝试解析JSON响应
+        if 'application/json' in content_type:
+            try:
+                result = response.json()
+                # 如果响应头中没有x-request-id，尝试从响应体中获取
+                if not x_request_id:
+                    x_request_id = result.get('x_request_id', '')
+                error_msg = result.get('message', result.get('msg', f'HTTP {response.status_code}'))
+                return error_msg, x_request_id
+            except:
+                pass
+        
+        # 处理HTML响应
+        if 'text/html' in content_type or response.text.strip().startswith('<'):
+            try:
+                # 从HTML中提取标题（通常包含状态码和状态文本）
+                title_match = re.search(r'<title>(.*?)</title>', response.text, re.IGNORECASE)
+                if title_match:
+                    error_msg = title_match.group(1).strip()
+                else:
+                    # 如果没有title，尝试提取h1标签
+                    h1_match = re.search(r'<h1>(.*?)</h1>', response.text, re.IGNORECASE)
+                    if h1_match:
+                        error_msg = h1_match.group(1).strip()
+                    else:
+                        error_msg = f'HTTP {response.status_code}'
+            except:
+                error_msg = f'HTTP {response.status_code}'
+        
+        # 处理纯文本响应
+        elif 'text/plain' in content_type:
+            error_msg = response.text[:200].strip() if response.text else f'HTTP {response.status_code}'
+        
+        # 其他情况
+        else:
+            if response.text:
+                # 尝试截取前200字符，但去除换行和多余空格
+                text = response.text[:200].strip()
+                # 如果包含多行，只取第一行
+                if '\n' in text:
+                    text = text.split('\n')[0].strip()
+                error_msg = text if text else f'HTTP {response.status_code}'
+            else:
+                error_msg = f'HTTP {response.status_code}'
+        
+        return error_msg, x_request_id
+
     def _call_pipeline_api(self, file_bytes: bytes, filename: str, data_source: Dict[str, Any]) -> Optional[Dict[str, Any]]:
         url = f"{self.api_base_url}/pipeline"
         max_retries = 3
@@ -263,15 +325,8 @@ class Pipeline:
                     logger.error(f"Pipeline 接口返回错误: code={result.get('code')}, message={error_msg}, x_request_id={x_request_id}")
                     return None
                 else:
-                    # 尝试解析响应获取 x_request_id 和错误信息
-                    x_request_id = ''
-                    error_msg = ''
-                    try:
-                        result = response.json()
-                        x_request_id = result.get('x_request_id', '')
-                        error_msg = result.get('message', result.get('msg', response.text[:200]))
-                    except:
-                        error_msg = response.text[:200] if response.text else f'HTTP {response.status_code}'
+                    # 使用规范化函数提取错误信息
+                    error_msg, x_request_id = self._extract_error_message(response)
                     
                     print(f"  ✗ API 错误 {response.status_code}: {error_msg}, x_request_id={x_request_id}, 重试 {try_count + 1}/{max_retries}")
                     logger.warning(f"API 错误 {response.status_code}: {error_msg}, x_request_id={x_request_id}, 重试 {try_count + 1}/{max_retries}")
@@ -374,6 +429,14 @@ class Pipeline:
             print(f"  → 读取文件...")
             file_bytes, data_source = self.source.read_file(file_path)
             data_source = data_source or {}
+            
+            # 检查文件大小，超过 100MB 则报错
+            MAX_FILE_SIZE = 100 * 1024 * 1024  # 100MB
+            file_size = len(file_bytes)
+            if file_size > MAX_FILE_SIZE:
+                file_size_mb = file_size / (1024 * 1024)
+                raise ValueError(f"文件大小过大: {file_size_mb:.2f}MB，超过100MB限制")
+            
             # 转换为毫秒时间戳字符串
             timestamp_ms = int(datetime.now(timezone.utc).timestamp() * 1000)
             data_source['date_processed'] = str(timestamp_ms)