Apply tls_verify and tls_cacerts config to ImageDownloader (#11715)

[ntouran]

Subject: Apply tls_verify and tls_cacerts config to ImageDownloader

Feature or Bugfix

• Bugfix

Purpose

• Apply existing requests configuration to a request where it is needed in some cases.
• To fully test, you need a self-signed certificate to put remote images on via URL.

Detail

• Fixes the issue defined in tls_verify and tls_cacerts config options are not applied to requests in ImageDownloader #11715

Relates

• tls_verify and tls_cacerts config options are not applied to requests in ImageDownloader #11715

[picnixz]

If possible, can you check where other tls_verify and tls_cacerets values are not applied? Like, are there other places where the same issue could happen?

[ntouran]

If possible, can you check where other tls_verify and tls_cacerets values are not applied? Like, are there other places where the same issue could happen?

I looked around to see if there were any more obvious cases at first but will double check. Good idea.

Ok:

$ git grep 'from sphinx.util import' | grep requests
sphinx/builders/linkcheck.py:from sphinx.util import encode_uri, logging, requests
sphinx/ext/intersphinx.py:from sphinx.util import logging, requests
sphinx/transforms/post_transforms/images.py:from sphinx.util import logging, requests
tests/test_build_linkcheck.py:from sphinx.util import requests

The 3 clients of the requests util here are:

• linkcheck -- yes, does use this config correctly in HEAD and GET requests
• intersphinx -- yes, does use proper config already
• ImageDownloader -- now does correct usage applied by this PR

So I think we're good!

[picnixz]

Good then! Could you add a CHANGES entry please?

@AA-Turner I think we're good unless you want to wrap small PRs for the day when 7.3.0 is planned to be released.

[ntouran]

Could you add a CHANGES entry please?

Done!

[picnixz]

Perfect! In the future, if you want to contribute again, don't force push commits since it makes incremental review harder (we squash them at the end anyway).

[ntouran]

Perfect! In the future, if you want to contribute again, don't force push commits since it makes incremental review harder (we squash them at the end anyway).

Ok sorry. The 2nd commit on the followup had the wrong email address, so I only force pushed an update to the email to that one commit, hoping the tooling would show just that, but I see what you're saying, and can understand that.