Geany ChangeLog, AUTHORS & COMMITTERS from Git log

gistfile1.py

#!/usr/bin/env python

'''
Helper script for Geany's distribution files generated by Git.

TODO: Use .mailmap or other means to use preferred name and emails.
'''

AUTHORS_HEADER = 'The following people have authored commits:\n'
AUTHORS_FOOTER = '''\nSee the COMMITTERS, ChangeLog and THANKS files for \
a full list of contributors.'''

COMMITTERS_HEADER = 'The following people have committed code:\n'
COMMITTERS_FOOTER = '''\nSee the AUTHORS, ChangeLog and THANKS files for \
a full list of contributors.'''


import codecs
import os
import subprocess
import sys

from collections import namedtuple
from optparse import OptionParser


Person = namedtuple ('Person', 'name emails count')


def find_person (people, name, emails):
  '''
  Lookup person in people by name or any of their email addresses.
  '''
  for person in people:
    if person.name == name:
      return person
    elif any (email in person.emails for email in emails):
      return person
  return None


def command_get_lines (args, workdir='.'):
  '''
  Get a list of UTF-8 encoded lines with trailing whitespace stripped
  off from the stdout of a subprocess.
  '''
  cd = os.getcwd ()
  os.chdir (workdir)
  proc = subprocess.Popen (args, stdout=subprocess.PIPE)
  fobj = codecs.EncodedFile (proc.stdout, 'UTF-8', 'UTF-8', errors='replace')
  lines = [ line.rstrip () for line in fobj ]
  os.chdir (cd)
  return lines


def utf8_ljust (s, length, fill_char=u' '):
  '''
  Pad the right side of s with fill_char until its as long as length.
  Built-in string.ljust () doesn't seem to work properly with unicode.
  '''
  if len (fill_char) != 1:
    raise ValueError ('fill_char must be exactly 1 character long')
  s = unicode (s, 'UTF-8', errors='replace')
  while len (s) < length:
    s = s + fill_char
  return s


def get_people (d, fmt):
  '''
  Gets a list of unique People from the git log in directory d.  The
  fmt is passed as the format for git log's --pretty argument.
  '''
  lines = command_get_lines ([ 'git', 'log',
                               '--encoding=UTF-8',
                               '--pretty=format:%s' % fmt ], d)
  people = []
  for line in lines:
    name = line.split ('|')[0]
    emails = [line.split ('|')[1]]
    person = find_person (people, name, emails)
    if person:
      people.remove (person)
      person = Person (person.name, person.emails, person.count + 1)
      for e in emails:
        if e not in person.emails:
          person.emails.append (e)
          person.emails.sort ()
      people.append (person)
    else:
      people.append (Person (name, emails, 1))
  return sorted (people, key=lambda p: p.count, reverse=True)



def print_people (d, fmt, header='', footer=''):
  '''
  Prints the unique People for git log in directory d passing the fmt
  as the format to git log's --pretty option.  The people are printed
  in the standard format of `Firstname Lastname <EmailAddress>` with
  all of the email addresses left-aligned.
  '''
  if header:
    print header
  people = get_people (d, fmt)
  max_len = max ([ len (person.name) for person in people ])
  for person in people:
    name = utf8_ljust (person.name, max_len).encode ('utf8', 'replace')
    emails = person.emails[0].encode ('utf8', 'replace')
    sys.stdout.write ("%s <%s>\n" % (name, emails))
  if footer:
    print footer


def print_changelog (d):
  args = [ 'git', 'log', '--shortstat', '--no-color', '--encoding=UTF-8' ]
  cd = os.getcwd ()
  os.chdir (d)
  proc = subprocess.Popen (args, stdout=subprocess.PIPE)
  fobj = codecs.EncodedFile (proc.stdout, 'UTF-8', 'UTF-8', errors='replace')
  for line in fobj:
    print line.strip ()
  os.chdir (cd)


def main ():

  p = OptionParser(usage='Usage: %prog [options] GITDIR')

  p.add_option ('-c', '--committers', default=False, action='store_true',
    dest='committers', help='print a list of committers')
  p.add_option ('-a', '--authors', default=False, action='store_true',
    dest='authors', help='print a list of authors')
  p.add_option ('-l', '--changlog', default=False, action='store_true',
    dest='changelog', help='print a change log')

  opts, args = p.parse_args ()

  if len (args) < 1:
    p.error ('please specify the root Git directory')
  elif not os.path.isdir (os.path.join (args[0], '.git')):
    p.error ('.git directory is missing in %s' % args[0])

  if int (opts.committers) + int (opts.authors) + int (opts.changelog) > 1:
    p.error ('options -c, -a and -l are mutually exclusive')

  if opts.committers:
    print_people (args[0], '%cn|%ce', COMMITTERS_HEADER, COMMITTERS_FOOTER)
  elif opts.authors:
    print_people (args[0], '%an|%ae', AUTHORS_HEADER, AUTHORS_FOOTER)
  elif opts.changelog:
    print_changelog (args[0])
  else:
    p.error ('no option specified')


if __name__ == '__main__': main ()