asterisk-gitlog-find

#!/usr/bin/env python
# asterisk-gitlog-find (part of ossobv/vcutil) // wdoekes/2016
#   // Public Domain
#
# Throwaway utility to find commits in the Asterisk project in which
# you had a hand.
#
# Usage:
#
#     cd asterisk.git
#     git log > git-log.txt
#     python asterisk-gitlog-find git-log.txt wdoekes 'Walter Doekes'
#
from datetime import datetime
import re


def commitlog_to_commits(reader):
    buffer_ = []
    for line in reader:
        if line.startswith('commit '):
            if buffer_:
                yield ''.join(buffer_)
            buffer_ = []

        buffer_.append(line)

    if buffer_:
        yield ''.join(buffer_)


def logfile_to_commits(filename, matcher):
    with open(filename) as commitlog:
        for commit in commitlog_to_commits(commitlog):
            if matcher.search(commit):
                yield commit


class Commit(object):
    def __init__(self, msg, handle, modes):
        lines = msg.split('\n')
        self.commit = lines[0][7:]
        self.author = lines[1][8:]
        self.orig_date = lines[2][8:]
        self.date = datetime.strptime(
            self.orig_date[:-6], '%a %b %d %H:%M:%S %Y')  # ignored time zone!

        self.handle = handle
        self.modes = tuple(modes)
        assert not lines[3]

        msg = [i.strip() for i in lines[4:]]
        if msg[0].startswith(('Blocked revisions', 'Merged revisions')):
            while msg and (
                    msg[0].startswith(('Merged', 'Blocked')) or
                    msg[0].startswith(('https://origsvn', 'file://')) or
                    msg[0].startswith('........') or
                    msg[0] == '' or
                    (msg[0][0:1] == 'r' and msg[0][1:2].isdigit())):
                msg.pop(0)
        for i, line in enumerate(msg):
            if line.startswith('........'):
                msg = msg[0:i]
                break

        header = []
        while msg:
            line = msg.pop(0)
            if line:
                header.append(line)
            else:
                break

        self.header = ' '.join(header)
        self.msg = '\n'.join(msg)

    def __str__(self):
        return '{commit} {date} {handle}: {header} [{modes}]'.format(
            commit=self.commit[0:7], date=self.date.strftime('%Y-%m-%d'),
            handle=self.handle, header=self.header[0:72],
            modes=', '.join(self.modes))


def main(git_log, handle, fullname):
    kwargs = {'handle': re.escape(handle),
              'fullname': re.escape(fullname)}

    plain_re = re.compile(r'\b{handle}\b|\b{fullname}\b'.format(**kwargs))
    author_re = re.compile(
        r'(Author: ({handle}|{fullname})'
        r'|uploaded by[^\n]*\b({handle}|{fullname})\b'
        r'|\b({handle}|{fullname}) \(License '
        r'|Patch[ -]by:[^\n]*\b({handle}|{fullname})\b)'.format(**kwargs))
    reporter_re = re.compile(
        r'Reported[ -]by:[^\n]*\b({handle}|{fullname})\b'.format(**kwargs))
    tester_re = re.compile(
        r'Tested[ -]by:[^\n]*\b({handle}|{fullname})\b'.format(**kwargs))

    for commit in logfile_to_commits(git_log, plain_re):
        modes = []
        if author_re.search(commit):
            modes.append('author')
        if reporter_re.search(commit):
            modes.append('reporter')
        if tester_re.search(commit):
            modes.append('tester')
        if not modes:
            modes.append('other')

        commit = Commit(commit, handle, modes)
        print commit


if __name__ == '__main__':
    import sys
    try:
        git_log, handle, fullname = sys.argv[1:]
    except ValueError:
        sys.stderr.write(
            'Usage: asterisk-gitlog-find GIT_LOG_FILE HANDLE FULL_NAME\n')
        sys.exit(1)
    else:
        main(git_log, handle, fullname)