GitHub repo stats markdown printer, used for a project relicensing callout

gh-repo-contributions-stats.py

#!/usr/bin/python3

# by nyov, Public Domain

from __future__ import unicode_literals, print_function

import six
from six.moves import cPickle as pickle

# https://github.com/sigmavirus24/github3.py
# develop branch (1.0+)
from github3 import GitHub

TOKEN = '' # <github-api-token>
PROJECT = ('wardi','urwid') # (user, repo)
TMPFILE = 'urwid-commits.dump'

def write():
    if TOKEN:
        gh = GitHub(token=TOKEN)
    else:
        gh = GitHub() # try anon

    print('Ratelimit: %s' % gh.ratelimit_remaining)

    repo = gh.repository(*PROJECT)

    commits = list(repo.commits())

    with open(TMPFILE, 'wb') as f:
        pickle.dump(commits, f)

    print('Ratelimit: %s' % gh.ratelimit_remaining)

def read():
    github_commits = []
    unknown_commits = []
    invalid_commits = []

    with open(TMPFILE, 'rb') as f:
        c = pickle.load(f)

    commits = {}
    commits_n = {}
    for commit in c:
        if str(commit.author) in ('None',):
            unknown_commits.append(commit)
        elif str(commit.author) in ('invalid-email-address',):
            invalid_commits.append(commit)
        else:
            commits[str(commit.author)] = commit
            try:
                commits_n[str(commit.author)] += 1
            except KeyError:
                commits_n[str(commit.author)] = 1
            github_commits.append(commit)

    print(' * GitHub commit authors:  ')
    for _, commit in commits.items():
        #print('https://github.com/%s' % commit.author)
        print(
            '- [ ] @{author} ([{commits} commits](https://github.com/{user}/{repo}/commits?author={author}))  '.format(
                user = PROJECT[0],
                repo = PROJECT[1],
                author = str(commit.author),
                commits = commits_n[str(commit.author)],
            )
        )
    print()

    print(' * Git commit authors (email contact only):  ')
    unknown = []
    unknown_c = {}
    for commit in unknown_commits:
        author, email = author_for_sha(str(commit.sha))
        unknown += [(author, email)]
        try:
            unknown_c[author] += [str(commit.sha)]
        except KeyError:
            unknown_c[author] = [str(commit.sha)]
        #print(str(commit.sha), author, email)
    unknown = set(unknown)
    for author, email in unknown:
        #print(author, '<%s>' % email, ', '.join(['@%s' % c for c in unknown_c[author]]))
        print(
            '- [ ] {author} (commits: {commits} )  '.format(
                author = author,
                commits = ', '.join(['%s' % c for c in unknown_c[author]]),
            )
        )
    print()

    print(' * Other commit authors (invalid email):  ')
    invalid = []
    invalid_c = {}
    for commit in invalid_commits:
        author, email = author_for_sha(str(commit.sha))
        invalid += [(author, email)]
        try:
            invalid_c[author] += [str(commit.sha)]
        except KeyError:
            invalid_c[author] = [str(commit.sha)]
        #print(str(commit.sha), author, email)
    invalid = set(invalid)
    for author, email in invalid:
        print(
            '- [ ] {author} (commits: {commits} )  '.format(
                author = author,
                commits = ', '.join(['%s' % c for c in invalid_c[author]]),
            )
        )
    print()


## git cli, no API here
#  needs to run in the actual cloned repo directory
#  (could be done using the GH API, if you have commit access to the repo. I didn't)

import os
import subprocess

GIT = '/usr/bin/git'

def git(filestream, *params):
    cli = [GIT]
    for param in params:
        cli.extend([param])
    #cli.extend(['-', '-'])
    os.environ['PYTHONIOENCODING'] = 'utf-8'
    proc = subprocess.Popen(
        cli,
        shell=False,
        stdin=subprocess.PIPE,
        stdout=subprocess.PIPE,
        stderr=subprocess.PIPE
    )
    output = proc.communicate(filestream)[0]
    exit_code = proc.wait()
    return exit_code, output

def author_for_sha(sha):
    _, g = git(b'', 'log', '-1', str(sha))
    for line in g.decode('utf-8').split('\n'):
        if 'Author:' in line:
            author, email = line.strip('Author:').strip(' ').strip('>').split('<')
            return author.strip(), email
    return None, None



if __name__=="__main__":
    #write() # download once
    read() # test it often