Azure Function App部署后遇AttributeError：os无add_dll_directory属性

使用Python 3.10.9构建Azure Function App，部署后出现错误：AttributeError: module 'os' has no attribute 'add_dll_directory'。本地Visual Studio调试正常，推测是因为函数App运行在Linux系统上，而pandas依赖的numpy调用了仅Windows支持的os.add_dll_directory方法导致的。

代码 #

import azure.functions as func

app = func.FunctionApp()

@app.function_name(name="myTimer")
@app.schedule(schedule="*/15 * * * *", arg_name="myTimer", run_on_startup=True, use_monitor=False)

def main(myTimer: func.TimerRequest) -> None:
    from batch_functions import price_data_download, position_data_download, trade_history_download, risk_checks
    import logging

    price_data_download()
    position_data_download()
    risk_checks()
    trade_history_download()
    if myTimer.past_due:
        logging.info('The timer is past due!')

    logging.info('Python timer trigger function executed.')

错误栈 #

Result: Failure Exception: AttributeError: module 'os' has no attribute 'add_dll_directory' Stack: File "/azure-functions-host/workers/python/3.10/LINUX/X64/azure_functions_worker/dispatcher.py", line 479, 
in _handle__invocation_request call_result = await self._loop.run_in_executor( File "/usr/local/lib/python3.10/concurrent/futures/thread.py", 
line 58, 
in run result = self.fn(*self.args, **self.kwargs) File "/azure-functions-host/workers/python/3.10/LINUX/X64/azure_functions_worker/dispatcher.py", line 752, 
in _run_sync_func return ExtensionManager.get_sync_invocation_wrapper(context, File "/azure-functions-host/workers/python/3.10/LINUX/X64/azure_functions_worker/extension.py", line 215, 
in _raw_invocation_wrapper result = function(**args) File "/home/site/wwwroot/function_app.py", line 9, 
in main from batch_functions import price_data_download, position_data_download, trade_history_download, risk_checks File "/home/site/wwwroot/batch_functions.py", line 1, 
in <module> import pandas as pd File "/home/site/wwwroot/pandas/__init__.py", line 24, 
in <module> __import__(_dependency) File "/home/site/wwwroot/numpy/__init__.py", line 112, 
in <module> _delvewheel_patch_1_5_1() File "/home/site/wwwroot/numpy/__init__.py", line 109, in _delvewheel_patch_1_5_1 os.add_dll_directory(libs_dir)

• 重新生成Linux兼容的依赖清单：本地Windows环境安装的numpy是适配Windows的版本（带delvewheel的DLL补丁），部署到Linux后不兼容。可以在WSL或Linux虚拟机里创建虚拟环境，执行pip install pandas azure-functions，再用pip freeze > requirements.txt导出依赖，用这个清单重新部署。
• 锁定兼容的numpy版本：部分高版本numpy引入的delvewheel补丁仅针对Windows，选择无需该补丁的版本，比如在requirements.txt中指定numpy<=1.23.5，重新部署安装。
• 启用Azure远程构建：部署时开启远程构建功能，让Azure在Linux环境下自动安装适配的依赖，避免本地Windows包上传导致的兼容问题。比如用Azure CLI部署时添加--build remote参数。
• 临时修改numpy初始化代码：如果无法立即更换版本，可手动修改numpy的__init__.py文件，在调用os.add_dll_directory前增加系统判断：

  def _delvewheel_patch_1_5_1():
      import os
      libs_dir = os.path.abspath(os.path.join(os.path.dirname(__file__), '.libs'))
      if os.name == 'nt':  # 仅Windows系统执行
          os.add_dll_directory(libs_dir)
  

  注意：该方法仅作为临时应急方案，后续更新numpy会覆盖修改，不推荐长期使用。

内容的提问来源于stack exchange，提问作者Haakon