How can I use pypi-mirror behind a proxy?

[tt2000]

If I use pip to download a package, I add --proxy http://proxy.url:port. Is there a similar option for pypi-mirror?

[montag451]

Hi @tt2000,

For the moment there is no way to specify a proxy using a command line switch but I will add one since it's something pretty common to be behind a corporate proxy. For the moment, your best shot is to try to specify the proxy using an environment variable (HTTP_PROXY and/or HTTPS_PROXY)

[montag451]

I've added a --proxy option to the download command. Could you test it? If it's OK for you I will publish a new version on PyPI

[tt2000]

How can I test it? Just cloning your master branch and running e. g. python .\pypi_mirror.py download -d test_download --proxy http://proxy.url:1234 numpy throws errors (not sure if related to proxy). Do I have to install it differently? It seems to try to install dependencies:

Collecting numpy
  Using cached numpy-1.21.2.zip (10.3 MB)
  Installing build dependencies ... error

Should it not just download the dependencies instead of installing them?

[montag451]

pypi-mirror uses pip under the hood to download a package and its dependencies. When pip downloads a source distribution (e.g numpy-1.21.2.zip) it runs the setup script of the package to find out the dependencies of the package because source distributions do not have the Requires-Dist field in their metadata file (i.e the PKG-INFO inside the tarball or the zip file, see pypa/setuptools#1805). Setup scripts can do a lot of things and the setup script of numpy is a pretty complicated one. To prevent this kind of things to happen, I strongly advise you to download only binary distribution (i.e wheel files) using the -b option of the download command:

python .\pypi_mirror.py download -d test_download --proxy http://proxy.url:1234 -b numpy

[johnziebro]

Just noting that using the -b flag also allowed mirroring of the osmnx module, which has a complicated install process with pip (considered experimental) and conda installation is generally recommended. Flagging binary also significantly speeds up the process by bypassing local building.

[tt2000]

@montag451 Testing was successful, I can confirm that the new proxy argument works. Thx for the very fast responses and good service!

[montag451]

@tt2000 Perfect, I will publish a new version of the package on PyPI. Thx for the bug report and the testing! I will close this issue as soon as the new version is published.

[montag451]

I've published a new version on PyPI so I will close this issue. Thx again for the bug report and the testing!