Git Contribution Analysis

git-contribution-analysis.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

import argparse
import git
import os
import os.path
import collections

VERBOSE = False

def _main():
    parser = argparse.ArgumentParser(description="Git repository contribution analysis")
    parser.add_argument('repository')
    parser.add_argument('-s', '--subdirectory', default=None)
    parser.add_argument('-v', action='store_true')
    parser.add_argument('-x', '--suffix', nargs='+')
    options = parser.parse_args()

    repo = git.Repo(options.repository)
    headcommit = repo.head.commit

    global VERBOSE
    VERBOSE = options.v

    #print repo
    #print repr(headcommit)

    print '>> Commit base contribution'
    commit_contribution = collections.defaultdict(int)
    for one in repo.iter_commits():
        commit_contribution[one.committer.email] += 1
    for k, v in commit_contribution.iteritems():
        print k, v

    print '>> Line base contribution'
    for k, v in treeanalysis(repo, headcommit, headcommit.tree, options.subdirectory, options.suffix).iteritems():
        print k, v

def treeanalysis(repo, ref, tree, subdirectory=None, suffix=None, root='', result=collections.defaultdict(int)):
    for one in tree:
        if isinstance(one, git.Tree):
            treeanalysis(repo, ref, one, subdirectory, suffix, os.path.join(root, one.name), result)
        else:
            if subdirectory and not root.startswith(subdirectory): continue
            if suffix:
                if not any([one.name.endswith(x) for x in suffix]):
                    continue
            linenum = 0
            for v in repo.blame(ref, os.path.join(root, one.name)):
                result[v[0].committer.email] += len(v[1])
                linenum += len(v[1])
            if VERBOSE:
                print os.path.join(root, one.name), one.mime_type, linenum

    return result

if __name__ == '__main__':
    _main()