darcs_bundle_to_diff

#!/usr/bin/env python2
from __future__ import print_function


import itertools as it, operator as op, functools as ft
from os.path import join, normpath
import os, sys, re


def main():

	import argparse
	parser = argparse.ArgumentParser(
		description='Convert darcs bundle to a unified diff.')
	parser.add_argument('bundle', nargs='?',
		help='Path to darcs bundle file (default: use stdin).')
	parser.add_argument('-s', '--skip-files',
		metavar='regexp', action='append', default=list(),
		help='Skip hunks from files with paths (as given by darcs),'
			' matching given regexp. Can be specified multiple times.')
	parser.add_argument('-n', '--limit', type=int, help='Max number of hunks to export.')
	parser.add_argument('--debug', action='store_true', help='Verbose operation mode.')
	optz = parser.parse_args()

	import logging
	logging.basicConfig(
		format='%(levelname)s :: %(message)s',
		level=logging.WARNING
			if not optz.debug else logging.DEBUG )
	log = logging.getLogger()


	def panic(msg='Invalid file format'):
		raise ValueError(msg)

	def skip_lines(src, empty=True, stop_on=list()):
		src, src_peek = it.tee(src)
		for line in src_peek:
			line_is_empty = not line.strip()
			if empty and line_is_empty:
				next(src)
				continue
			elif not empty:
				if stop_on:
					for regexp in stop_on:
						if re.search(regexp, line): return src
					else: regexp = False
				if not line_is_empty:
					next(src)
					continue
				else:
					skip_lines(src, empty=True)
					break
			break
		return src

	def skip_to_hunk(src, line=''):
		for line in it.chain([line], src):
			if line.strip() == '}':
				log.debug('No more hunks left - next commit')
				return None
			match = re.search(r'^hunk (?P<path>.*) (?P<line>\d+)$', line)
			if match: break
		else:
			log.debug('No more lines left')
			return None
		log.debug('Next hunk start: {}'.format(line.strip()))
		return match


	with (sys.stdin if not optz.bundle else open(optz.bundle)) as src:
		for line in src:
			# section -- New patches:
			if re.search(r'^New patches:$', line): break
		else: panic()

		while True:
			src = skip_lines(src, empty=True)

			# -- [interfaces.py: 'which -> that' grammar cleanup.
			line = next(src)

			if re.search(r'^Context:$', line):
				log.debug('Reached darcs-context, all done')
				return

			log.debug('Next commit start: {}'.format(line.strip()))
			match = re.search(r'\[(.*)$', line)
			if not match: panic('Not a commit-start line, bailing out: {}'.format(line.strip()))
			commit_summary = match.group(1)

			# -- david-sarah@jacaranda.org**20111205024024
			# --  Ignore-this: 2de8d902dd7b473fb5c15701fcd8e0f7
			src = skip_lines(src, empty=False, stop_on=r'^[')
			commit_message = list()
			for line in src:
				if line.startswith(']'): break
				commit_message.append(line.strip())

			for line in src:
				# -- hunk ./src/allmydata/interfaces.py 33
				match = re.search(r'^hunk (?P<path>.*) (?P<line>\d+)$', line)
				if not match: continue
				break


			while match:
				path, line_no = match.group('path'), int(match.group('line'))
				log.debug('Hunk: {} {}'.format(path, line_no))

				if optz.skip_files:
					for regexp in optz.skip_files:
						if re.search(regexp, path): break
					else: regexp = None
					if regexp:
						log.debug('Skipping hunk')
						match = skip_to_hunk(src)
						continue

				diff = list()
				change_src = change_dst = 0
				for line in src:
					# -- -class RIStubClient(RemoteInterface):
					if line[0] not in '-+ ':
						match = skip_to_hunk(src, line)
						break
					else:
						diff.append(line)
						if line[0] in '- ': change_src += 1
						if line[0] in '+ ': change_dst += 1

				paths, new = list(normpath(join(k, path)) for k in 'ab'), False
				line_no_src = line_no_dst = line_no
				if line_no == 1 and change_src == 0:
					log.debug('Detected file creation: {}'.format(path))
					new, line_no_src = True, 0
				print('diff --git {} {}'.format(*paths))
				print('--- {}\n+++ {}'.format(*(paths if not new else ['/dev/null', paths[1]])))
				print(
					'@@ -{line_no_src},{change_src} +{line_no_dst},{change_dst} @@'\
					.format( line_no_src=line_no_src, line_no_dst=line_no_dst,
						change_src=change_src, change_dst=change_dst ) )
				print(''.join(diff), end='')

				if optz.limit is not None:
					optz.limit -= 1
					if optz.limit <= 0:
						log.debug('Reached hunk-limit, exiting')
						return






if __name__ == '__main__': main()