util/maint/git-patch-to-hg-patch

   1 #!/usr/bin/env python2
   2 #
   3 # This file originated from the moz-git-tools repo on GitHub
   4 # (https://github.com/mozilla/moz-git-tools), which contains the
   5 # following LICENSE notice:
   6 #
   7 # <quote>
   8 # Except for git-new-workdir, which is covered under GPLv2, the code
   9 # in this repository is placed into the public domain via CC0.
  10 #
  11 # http://creativecommons.org/publicdomain/zero/1.0/legalcode
  12 # </quote>
  13
  14 r"""Git format-patch to hg importable patch.
  15
  16 (Who knew this was so complicated?)
  17
  18 >>> process(StringIO('From 3ce1ccc06 Mon Sep 17 00:00:00 2001\nFrom: fromuser\nSubject: subject\n\nRest of patch.\nMore patch.\n'))
  19 '# HG changeset patch\n# User fromuser\n\nsubject\n\nRest of patch.\nMore patch.\n'
  20
  21 >>> process(StringIO('From: fromuser\nSubject: A very long subject line.  Lorem ipsum dolor sit amet, consectetur adipiscing elit. Morbi faucibus, arcu sit amet\n\nRest of patch.\nMore patch.\n'))
  22 '# HG changeset patch\n# User fromuser\n\nA very long subject line.  Lorem ipsum dolor sit amet, consectetur adipiscing elit. Morbi faucibus, arcu sit amet\n\nRest of patch.\nMore patch.\n'
  23
  24 >>> process(StringIO('From: f\nSubject: =?UTF-8?q?Bug=20655877=20-=20Dont=20treat=20SVG=20text=20frames=20?= =?UTF-8?q?as=20being=20positioned.=20r=3D=3F?=\n\nPatch.'))
  25 '# HG changeset patch\n# User f\n\nBug 655877 - Dont treat SVG text frames as being positioned. r=?\n\nPatch.'
  26 """
  27
  28 # Original author: bholley
  29
  30 import sys
  31 import re
  32 import fileinput
  33 import email, email.parser, email.header, email.utils
  34 import math
  35 from cStringIO import StringIO
  36 from itertools import takewhile
  37
  38 def decode_header(hdr_string):
  39   r"""Clean up weird encoding crap.
  40
  41   >>> clean_header('[PATCH] =?UTF-8?q?Bug=20655877=20r=3D=3F?=')
  42   '[PATCH] Bug 655877 r=?'
  43   """
  44   rv = []
  45   hdr = email.header.Header(hdr_string, maxlinelen=float('inf'))
  46   for (part, encoding) in email.header.decode_header(hdr):
  47     if encoding is None:
  48       rv.append(part)
  49     else:
  50       rv.append(part.decode(encoding).encode('utf-8'))
  51   return ' '.join(rv)
  52
  53 def clean_header(hdr_string):
  54   r"""Transform a header split over many lines into a header split only where
  55   linebreaks are intended.  This is important because hg cares about the first
  56   line of the commit message.
  57
  58   Also clean up weird encoding crap.
  59
  60   >>> clean_header('Foo\n bar\n baz')
  61   'Foo bar baz'
  62   >>> clean_header('Foo\n bar\nSpam\nEggs')
  63   'Foo bar\nSpam\nEggs'
  64   """
  65
  66   lines = []
  67   curline = ''
  68   for line in decode_header(hdr_string).split('\n'):
  69     if not line.startswith(' '):
  70       lines.append(curline)
  71       curline = ''
  72     curline += line
  73   lines.append(curline)
  74   return '\n'.join(lines[1:])
  75
  76 def process(git_patch_file):
  77   parser = email.parser.Parser()
  78   msg = parser.parse(git_patch_file)
  79   from_hdr = clean_header(msg['From'])
  80   commit_title = clean_header(msg['subject'])
  81   if not len(commit_title) or not len(from_hdr):
  82     sys.stderr.write("%s does not look like a valid git patch file, skipping\n"
  83                      % git_patch_file.name)
  84     return
  85
  86   parsed_from = email.utils.parseaddr(from_hdr)
  87   nuke_prefix = r"\[PATCH( \d+/\d+)?\] "
  88   match = re.match(nuke_prefix, commit_title)
  89   if match:
  90     commit_title = commit_title[match.end():]
  91
  92   patch_body = msg.get_payload()
  93
  94   # git format-patch wraps the diff (including trailing whitespace):
  95   #   ---
  96   #   <diff>
  97   #   --
  98   #   2.0.3
  99   # This doesn't hurt parsing the diff at all, but the version number is
 100   # nonsense once the git specific items have been stripped
 101   patch_body = re.sub(r'--\s?\n[0-9\.]+\n$', '', patch_body)
 102
 103   return '\n'.join(['# HG changeset patch',
 104                     '# User %s <%s>' % parsed_from,
 105                     '',
 106                     commit_title,
 107                     '',
 108                     patch_body])
 109
 110 if __name__ == "__main__":
 111   if len(sys.argv) > 1 and sys.argv[1] == '--test':
 112     import doctest
 113     doctest.testmod()
 114     sys.exit(0)
 115
 116   # If there were no arguments, do stdin->stdout.
 117   filelist = sys.argv[1:]
 118   if not filelist:
 119     lines = process(sys.stdin)
 120     sys.stdout.writelines(lines)
 121     sys.exit(0)
 122
 123   # Otherwise, we take a list of files.
 124   for filename in filelist:
 125
 126     # Read the lines.
 127     f = open(filename, 'r')
 128     lines = process(f)
 129     f.close()
 130
 131     # Process.
 132
 133     if lines:
 134       # Write them back to the same file.
 135       f = open(filename, 'w')
 136       f.writelines(lines)
 137       f.close()