gem5/util/maint/git-patch-to-hg-patch

138 lines
4.3 KiB
Text
Raw Normal View History

#!/usr/bin/env python
#
# This file originated from the moz-git-tools repo on GitHub
# (https://github.com/mozilla/moz-git-tools), which contains the
# following LICENSE notice:
#
# <quote>
# Except for git-new-workdir, which is covered under GPLv2, the code
# in this repository is placed into the public domain via CC0.
#
# http://creativecommons.org/publicdomain/zero/1.0/legalcode
# </quote>
r"""Git format-patch to hg importable patch.
(Who knew this was so complicated?)
>>> process(StringIO('From 3ce1ccc06 Mon Sep 17 00:00:00 2001\nFrom: fromuser\nSubject: subject\n\nRest of patch.\nMore patch.\n'))
'# HG changeset patch\n# User fromuser\n\nsubject\n\nRest of patch.\nMore patch.\n'
>>> 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'))
'# 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'
>>> 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.'))
'# HG changeset patch\n# User f\n\nBug 655877 - Dont treat SVG text frames as being positioned. r=?\n\nPatch.'
"""
# Original author: bholley
import sys
import re
import fileinput
import email, email.parser, email.header, email.utils
import math
from cStringIO import StringIO
from itertools import takewhile
def decode_header(hdr_string):
r"""Clean up weird encoding crap.
>>> clean_header('[PATCH] =?UTF-8?q?Bug=20655877=20r=3D=3F?=')
'[PATCH] Bug 655877 r=?'
"""
rv = []
hdr = email.header.Header(hdr_string, maxlinelen=float('inf'))
for (part, encoding) in email.header.decode_header(hdr):
if encoding is None:
rv.append(part)
else:
rv.append(part.decode(encoding).encode('utf-8'))
return ' '.join(rv)
def clean_header(hdr_string):
r"""Transform a header split over many lines into a header split only where
linebreaks are intended. This is important because hg cares about the first
line of the commit message.
Also clean up weird encoding crap.
>>> clean_header('Foo\n bar\n baz')
'Foo bar baz'
>>> clean_header('Foo\n bar\nSpam\nEggs')
'Foo bar\nSpam\nEggs'
"""
lines = []
curline = ''
for line in decode_header(hdr_string).split('\n'):
if not line.startswith(' '):
lines.append(curline)
curline = ''
curline += line
lines.append(curline)
return '\n'.join(lines[1:])
def process(git_patch_file):
parser = email.parser.Parser()
msg = parser.parse(git_patch_file)
from_hdr = clean_header(msg['From'])
commit_title = clean_header(msg['subject'])
if not len(commit_title) or not len(from_hdr):
sys.stderr.write("%s does not look like a valid git patch file, skipping\n"
% git_patch_file.name)
return
parsed_from = email.utils.parseaddr(from_hdr)
nuke_prefix = r"\[PATCH( \d+/\d+)?\] "
match = re.match(nuke_prefix, commit_title)
if match:
commit_title = commit_title[match.end():]
patch_body = msg.get_payload()
# git format-patch wraps the diff (including trailing whitespace):
# ---
# <diff>
# --
# 2.0.3
# This doesn't hurt parsing the diff at all, but the version number is
# nonsense once the git specific items have been stripped
patch_body = re.sub(r'--\s?\n[0-9\.]+\n$', '', patch_body)
return '\n'.join(['# HG changeset patch',
'# User %s <%s>' % parsed_from,
'',
commit_title,
'',
patch_body])
if __name__ == "__main__":
if len(sys.argv) > 1 and sys.argv[1] == '--test':
import doctest
doctest.testmod()
sys.exit(0)
# If there were no arguments, do stdin->stdout.
filelist = sys.argv[1:]
if not filelist:
lines = process(sys.stdin)
sys.stdout.writelines(lines)
sys.exit(0)
# Otherwise, we take a list of files.
for filename in filelist:
# Read the lines.
f = open(filename, 'r')
lines = process(f)
f.close()
# Process.
if lines:
# Write them back to the same file.
f = open(filename, 'w')
f.writelines(lines)
f.close()