python - 从 Azure Devops 将包发布到 PyPi 在 twine 上传时挂起

Question

我想从我们的 Azure Devops 发布管道将构建的 .whl 包发布到 PyPi.org，但脚本 twine upload 一直挂起，没有错误、完成或超时。所以我们的包(非常小)的实际上传不起作用。

这是如何设置的:

yaml 构建:

trigger:
- master
pr: none
pool:
  vmImage: 'ubuntu-latest'

steps:
- task: UsePythonVersion@0
  inputs:
    versionSpec: '3.7'
    addToPath: true
    architecture: 'x64'

- script: python -m pip install --upgrade pip setuptools wheel
  displayName: 'Install tools'

- script: pip install -r src/requirements.txt
  displayName: 'Install requirements'

- script: |    
    python src/setup.py bdist_wheel 
  displayName: 'Artifact creation'

- script: python -m pip install --upgrade twine
  displayName: 'Install Twine'

- task: TwineAuthenticate@1
  inputs:
    pythonUploadServiceConnection: 'AzureML PyPi feed'

- script: |
   python -m twine upload --config-file $(PYPIRC_PATH) dist/*.whl
  displayName: 'Publish to PyPi through Twine'

服务连接:

• 身份验证方法由身份验证 token 指定。
• 用于上传的 Python 存储库 URL:https://upload.pypi.org/legacy
• 端点名称:azure-pypi
• 服务连接名称:AzureML PyPi feed
• [X]授予所有管道的访问权限

发布日志(最相关的部分):

TwineAuthenticate 步骤:

Starting: TwineAuthenticate
==============================================================================
Task         : Python twine upload authenticate
Description  : Authenticate for uploading Python distributions using twine. Add '-r FeedName/EndpointName --config-file $(PYPIRC_PATH)' to your twine upload command. For feeds present in this organization, use the feed name as the repository (-r). Otherwise, use the endpoint name defined in the service connection.
Version      : 1.165.0
Author       : Microsoft Corporation
Help         : https://learn.microsoft.com/azure/devops/pipelines/tasks/package/twine-authenticate
==============================================================================
75c64e89-xxxxxredactedxxxxx exists true
Adding authentication to configuration for registry azure-pypi
Successfully added auth for 0 internal feed and 1 external endpoint.
Finishing: TwineAuthenticate

发布步骤:

Starting: Publish to PyPi through Twine
==============================================================================
Task         : Command line
Description  : Run a command line script using Bash on Linux and macOS and cmd.exe on Windows
Version      : 2.164.0
Author       : Microsoft Corporation
Help         : https://learn.microsoft.com/azure/devops/pipelines/tasks/utility/command-line
==============================================================================
Generating script.
Script contents:
python -m twine upload --config-file /home/vsts/work/_temp/twineAuthenticate/Wv8nMR/.pypirc dist/*.whl
========================== Starting Command Output ===========================
/bin/bash --noprofile --norc /home/vsts/work/_temp/0d772e31-148f-4e3c-999b-b2a43a02b287.sh
Uploading distributions to https://upload.pypi.org/legacy/

这在 30 多分钟内保持不变，所以我只是取消了发布，而没有部署包。对此有什么想法吗？

Answer 1

我现在有一个临时解决方法:

我跳过了配置文件的使用，而是执行了以下代码:

- script: |
   python -m twine upload --skip-existing --verbose -p $(pypi-api-token) -u __token__ --repository $(pypi-project-name) --repository-url https://upload.pypi.org/legacy/ dist/*.whl
  displayName: 'Publish to PyPi through Twine'

在那里，我注意到更多(更好)的异常日志记录，这向我指出了两件事:

1. 我用于配置文件的 token 被设置为项目范围，其名称与我在服务连接中指定的项目不同
2. 存储库 URL 确实必须以尾部正斜杠结尾，否则您会收到 RedirectDetected 异常。

根据上述发现，我将 yaml 片段更新为以下内容:

- script: |
   python -m twine upload --skip-existing --verbose --repository $(pypi-project-name) --config-file $(PYPIRC_PATH) dist/*.whl
  displayName: 'Publish to PyPi through Twine'

但是在执行此操作时，我现在在构建中收到以下异常消息:

Generating script.
Script contents:
python -m twine upload --skip-existing --verbose --repository arcus-azureml --config-file /home/vsts/work/_temp/twineAuthenticate/2QGKVH/.pypirc dist/*.whl
========================== Starting Command Output ===========================
/bin/bash --noprofile --norc /home/vsts/work/_temp/7605dac9-5fa9-4856-94af-e938018278a5.sh
Uploading distributions to https://upload.pypi.org/legacy/
Uploading arcus_azureml-0.0.2-py3-none-any.whl

  0%|          | 0.00/5.80k [00:00<?, ?B/s]
100%|██████████| 5.80k/5.80k [00:00<00:00, 52.4kB/s]HTTPError: 403 Client Error: Invalid or non-existent authentication information. See https://pypi.org/help/#invalid-auth for details for url: https://upload.pypi.org/legacy/

Content received from server:
<html>
 <head>
  <title>403 Invalid or non-existent authentication information. See https://pypi.org/help/#invalid-auth for details</title>
 </head>
 <body>
  <h1>403 Invalid or non-existent authentication information. See https://pypi.org/help/#invalid-auth for details</h1>
  Access was denied to this resource.<br/><br/>
Invalid or non-existent authentication information. See https://pypi.org/help/#invalid-auth for details

我还输出了pypirc文件的内容，使用CAT命令，文件内容是这样的:

[distutils]
index-servers=arcus-azureml 
[arcus-azureml]
repository=https://upload.pypi.org/legacy/
username=build
password=***

更新的解决方案

因此，安排 403 的修复方法是将我的服务连接更改为使用“用户名和密码”作为身份验证方法，而不是身份验证 token ，并进行以下设置:

• 用户名:__token__
• 密码:实际的 api token 作为密码

执行此操作后，一切都按照我想要的方式进行。