linkcheck: support ignored-URIs for redirects #13127
Conversation
sphinx/util/requests.py
Outdated
| # do not follow redirections that match ignored URI patterns | ||
| if resp.is_redirect: | ||
| location = resp.headers['location'] | ||
| dest = urljoin(resp.url, location) |
There was a problem hiding this comment.
I'd draw attention to this line in particular as possibly-dubious.
Servers often respond with relative Location redirects; but I think we want to match against more-fully-qualified URI paths.
There was a problem hiding this comment.
This looks good to me.
[skip ci]
| 'uri': f'http://{address}/', | ||
| 'info': f'ignored redirect: http://{address}/redirected', | ||
| } | ||
|
|
There was a problem hiding this comment.
It might be worth adding a test verifying that when the response Location is an absolute URL going to a different server (so it overrides the urljoin base entirely), it is indeed excluded.
|
I’m wondering if the change should be considered breaking? An intermediary redirect might match a pattern from |
|
Thanks @francoisfreitag for the code review! I agree with adding test coverage for a cross-domain redirect case, and I'll do that soon. I'm less sure about whether we should consider it a breaking change - so far we haven't stated anything about the handling of |
Feature or Bugfix
Purpose
linkcheckclient to URLs that the operator has configured as ignored using thelinkcheck_ignoreconfig setting. This changeset handles those redirections by returning anignoreresult immediately, without following the redirect.Detail
requests.Session.get_redirect_targetmethod so that we can raise a (Sphinx-internal) exception if we encounter a redirect to a configured ignore-URL.Relates
linkcheck_ignorein link redirection #11233.