2.21 判断是否为招标文件时禁用判断是否扫描件 增加逻辑确保返回字符串为utf-8格式

flask_app/general/json_utils.py

@@ -470,6 +470,8 @@ def transform_json_values(data):
     elif isinstance(data, (int, float)):
         return str(data)
     elif isinstance(data, str):
+        # 替换掉超出 BMP 范围的字符（即 4 字节字符），用 ? 代替
+        data = re.sub(r'[^\u0000-\uffff]', '?', data)    #确保返回utf-8数据
         return data.replace('\n', '<br>')
     else:
         return data

flask_app/routes/judge_zbfile.py

@@ -47,7 +47,7 @@ def judge_zbfile() -> Any:              #判断是否是招标文件
             start_time = time.time()
             downloaded_filename = os.path.join(output_folder, "ztbfile")
             logger.info(f"接收到的url:{file_url}")
-            downloaded_filepath, file_type = download_file(file_url, downloaded_filename,True)
+            downloaded_filepath, file_type = download_file(file_url, downloaded_filename,False)   #判断是否为招标文件时不需要处理扫描型pdf
 
             if not downloaded_filepath or file_type == 4:
                 log_error_unique_id(unique_id, 4)

flask_app/run_serve.py

@@ -1,7 +1,6 @@
 #flask_app/run_serve.py
 import threading
 import time
-
 import requests
 from waitress import serve
 from flask_app.start_up import create_app