regex failing on check of hardcoded links

[nicoa]

Describe the bug

#9800 (contained in release 4.4.0) introduced an issue on our side for extlinks that contain special characters.

the problematic part of code is re.compile(s.replace("%s", "(?P<value>.+)")). a possible fix is re.compile(re.escape(s).replace("%s", "(?P<value>.+)")).

Would you be happy to receive a PR on this?

How to Reproduce

use:

extlinks = {
    "package": (
        r"https://gitlab.com/path_to_my_project/-/packages?search[]=%s",
        "package %s",
    ),
}

this will lead to re.error: unterminated character set at position 88 when any link is found on the parsed page.

Expected behavior

No response

Your project

sorry, nothing I can share - let me know if a reproducing example is required

Screenshots

No response

OS

Unix, Mac

Python version

3.9

Sphinx version

4.4.0

Sphinx extensions

extlinks

Extra tools

No response

Additional context

No response