python -m pip install parsegitlog
I've been working on a project where I need to get information on every commit from serveral hundred repos, and using GitPython. GitPython is great, but it was a little slow for what I was trying to do.
The method of parsing the gitlog in this project is much faster for my use case.
Note that this project is much more narrow in scope than GitPython, and relies
on parsing the output of git log directly, so might be more fragile.
You might find this useful if you only need to analyze the commits for a repo, and not perform any other git operations progromatically.
You can either import this as a module or run it from the command line.
python -m parsegitlog --help
python -m parsegitlog
from parsegitlog import get_commits
repo_path = '/path/to/my/git/repository'
get_commits(repo_path)get_commits will return a list of dicts, each dict representing a single
commit from the repository.
There is also a get_commits_df function that will return the commits as a
pandas dataframe (assumes pandas is available.)
Merge commits will show up multiple times. This is intended behavior.
Although they appear to be duplicates, the files_changed, insertions, and
deletions for each will be slightly different, and there will be duplicate
entries for each merge commit corresponding to the number of parents in the
merge. This is because, depending on which parent we compare to, the difference
from the parent (in terms of files changed, insertions, and deletions) won't be
the same (every other property should be the same).