Python script that queries git to list contributors to your files. Can ignore people and detect multiple username for one contributor. Outputs CSV

list-file-contributors.py

#!/usr/bin/env python
import subprocess

# Inspiration: https://stackoverflow.com/a/11740682

r = subprocess.check_output(['git','ls-tree','-r','--name-only','origin/master','./'])

files = r.split();

ignoredPeeps = set(['Laima Tazmin', 'Cara Warner', 'Patty Delgado', 'Matt Anderson', 'r29bot'])

nameMap = {
    '[email protected]': 'Carlo Francisco',
    'jcfrancisco': 'Carlo Francisco',

    'Jen': 'Jennifer Calloway',

    'josipherceg': 'Josip Herceg',
    '[email protected]': 'Josip Herceg', 

    'Liyan.Tang': 'Liyan Tang',
    'liyantang': 'Liyan Tang',

    '[email protected]': 'Patty Delgado',

    'johnnyshankman': 'Johnny Shankman',

    'Maggie Love & Noah Lemen': 'Noah Lemen',
}

for f in files:
    peeps = subprocess.check_output([
        'git', 'log', '--follow', '--pretty=format:%an', '--', f
        ])

    # One peep per line
    peeps = peeps.split('\n')
    # Some people use multiple accounts. Consolidate to one name
    peeps = map(lambda x: nameMap.get(x, x), peeps)
    # Take out duplicates and ignoredPeeps
    peeps = set(peeps) - ignoredPeeps
    peeps = list(peeps)
    peeps.sort()

    print "{0},{1},{2}".format(f, len(peeps), ','.join(peeps))