git filter-repo --path fails on gnu coreutils repo.

[gl-yziquel]

Hi.

Version: git-filter-repo v2.45.0

git clone git://git.sv.gnu.org/coreutils.git
cd coreutils
git filter-repo --path src/chroot.c

Fails as follows:

Parsed 29641 commitsTraceback (most recent call last):
  File "/home/mini-me/.pyenv/versions/3.11.8/bin/git-filter-repo", line 8, in <module>
    sys.exit(main())
             ^^^^^^
  File "/home/mini-me/.pyenv/versions/3.11.8/lib/python3.11/site-packages/git_filter_repo.py", line 4032, in main
    filter.run()
  File "/home/mini-me/.pyenv/versions/3.11.8/lib/python3.11/site-packages/git_filter_repo.py", line 3967, in run
    self._parser.run(self._input, self._output)
  File "/home/mini-me/.pyenv/versions/3.11.8/lib/python3.11/site-packages/git_filter_repo.py", line 1418, in run
    self._parse_tag()
  File "/home/mini-me/.pyenv/versions/3.11.8/lib/python3.11/site-packages/git_filter_repo.py", line 1297, in _parse_tag
    (tagger_name, tagger_email, tagger_date) = self._parse_user(b'tagger')
                                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/mini-me/.pyenv/versions/3.11.8/lib/python3.11/site-packages/git_filter_repo.py", line 1084, in _parse_user
    (name, email, when) = user_regex.match(self._currentline).groups()
                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'NoneType' object has no attribute 'groups'
fatal: stream ends early
fast-import: dumping crash report to .git/fast_import_crash_1886814

I suspect this is because the gnu coreutils repo has kind of a loooooong history going back to old version control systems. And, well, something is None at one point as a consequence of this troubled past.

Nonetheless, this points to some unwanted brittleness in git-filter-repo.

P.S.: thank you for that tool. It has proven to be invaluable to me.

[gl-yziquel]

I confirm that is an issue specific to the coreutils repository. The same operation succeeded on other repositories such as carving out cipd source code from google's luci-go repository. So, it is definitely some specific setup of coreutils that makes git filter-repo go nuts.