list-top-authors

#!/usr/bin/python

import git
import sys
import time

if len(sys.argv) < 4:
    print 'Usage: list-top-authors repo startrev threshold'
    print
    print 'Lists authors of commits in given repo, starting with startrev'
    print 'The threshold defines minimal amount of commits author needs to be listed'
    sys.exit()

threshold = int(sys.argv[3])

repo = git.Repo(sys.argv[1])

authors = {}

for commit in repo.iter_commits('%s..master' % sys.argv[2]):
    author = commit.author.name
    full_author = '%s <%s>' % (commit.author.name, commit.author.email)
    date = commit.authored_date
    if not author in authors:
        authors[author] = {
            'first': date,
            'last': date,
            'count': 1,
            'name': full_author,
        }
    else:
        if authors[author]['first'] > date:
            authors[author]['first'] = date
        if authors[author]['last'] < date:
            authors[author]['last'] = date
        authors[author]['count'] += 1

authors = authors.values()
authors.sort(lambda a, b: -1 * cmp(a['count'], b['count']))

for a in authors:
    if a['count'] < threshold:
        break
    first = time.strftime("%b %Y", time.gmtime(a['first']))
    last = time.strftime("%b %Y", time.gmtime(a['last']))
    # Ignore people with short interval
    if first == last:
        continue
    print '%s (%s - %s)' % (a['name'].encode('utf-8'), first, last)