3 # Copyright (c) 2012 Felipe Contreras
6 # Inspired by Rocco Rutte's hg-fast-export
8 # Just copy to your ~/bin, or anywhere in your $PATH.
9 # Then you can clone with:
10 # git clone hg::/path/to/mercurial/repo/
12 from mercurial import hg, ui, bookmarks, context, util, encoding
23 # If you want to switch to hg-git compatibility mode:
24 # git config --global remote-hg.hg-git-compat true
27 # Sensible defaults for git.
28 # hg bookmarks are exported as git branches, hg branches are prefixed
29 # with 'branches/', HEAD is a special case.
33 # Only hg bookmarks are exported as git branches.
34 # Commits are modified to preserve hg information and allow bidirectionality.
37 NAME_RE = re.compile('^([^<>]+)')
38 AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
39 AUTHOR_HG_RE = re.compile('^(.*?) ?<(.*?)(?:>(.+)?)?$')
40 RAW_AUTHOR_RE = re.compile('^(\w+) (?:(.+)? )?<(.*)> (\d+) ([+-]\d+)')
43 sys.stderr.write('ERROR: %s\n' % (msg % args))
47 sys.stderr.write('WARNING: %s\n' % (msg % args))
50 return 'l' in flags and '120000' or 'x' in flags and '100755' or '100644'
53 return '%+03d%02d' % (-tz / 3600, -tz % 3600 / 60)
56 m = { '100755': 'x', '120000': 'l' }
57 return m.get(mode, '')
59 def get_config(config):
60 cmd = ['git', 'config', '--get', config]
61 process = subprocess.Popen(cmd, stdout=subprocess.PIPE)
62 output, _ = process.communicate()
67 def __init__(self, path):
77 if not os.path.exists(self.path):
80 tmp = json.load(open(self.path))
82 self.tips = tmp['tips']
83 self.marks = tmp['marks']
84 self.last_mark = tmp['last-mark']
86 for rev, mark in self.marks.iteritems():
87 self.rev_marks[mark] = int(rev)
90 return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
93 json.dump(self.dict(), open(self.path, 'w'))
96 return str(self.dict())
98 def from_rev(self, rev):
99 return self.marks[str(rev)]
101 def to_rev(self, mark):
102 return self.rev_marks[mark]
104 def get_mark(self, rev):
106 self.marks[str(rev)] = self.last_mark
107 return self.last_mark
109 def new_mark(self, rev, mark):
110 self.marks[str(rev)] = mark
111 self.rev_marks[mark] = rev
112 self.last_mark = mark
114 def is_marked(self, rev):
115 return self.marks.has_key(str(rev))
117 def get_tip(self, branch):
118 return self.tips.get(branch, 0)
120 def set_tip(self, branch, tip):
121 self.tips[branch] = tip
125 def __init__(self, repo):
127 self.line = self.get_line()
130 return sys.stdin.readline().strip()
132 def __getitem__(self, i):
133 return self.line.split()[i]
135 def check(self, word):
136 return self.line.startswith(word)
138 def each_block(self, separator):
139 while self.line != separator:
141 self.line = self.get_line()
144 return self.each_block('')
147 self.line = self.get_line()
148 if self.line == 'done':
152 i = self.line.index(':') + 1
153 return int(self.line[i:])
156 if not self.check('data'):
158 i = self.line.index(' ') + 1
159 size = int(self.line[i:])
160 return sys.stdin.read(size)
162 def get_author(self):
166 m = RAW_AUTHOR_RE.match(self.line)
169 _, name, email, date, tz = m.groups()
170 if name and 'ext:' in name:
171 m = re.match('^(.+?) ext:\((.+)\)$', name)
174 ex = urllib.unquote(m.group(2))
176 if email != bad_mail:
178 user = '%s <%s>' % (name, email)
180 user = '<%s>' % (email)
188 tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
189 return (user, int(date), -tz)
193 print "M %s inline %s" % (gitmode(fc.flags()), fc.path())
194 print "data %d" % len(d)
197 def get_filechanges(repo, ctx, parent):
203 prev = repo[parent].manifest().copy()
207 if (cur.flags(fn) != prev.flags(fn) or cur[fn] != prev[fn]):
212 removed |= set(prev.keys())
214 return added | modified, removed
216 def fixup_user_git(user):
218 user = user.replace('"', '')
219 m = AUTHOR_RE.match(user)
222 mail = m.group(2).strip()
224 m = NAME_RE.match(user)
226 name = m.group(1).strip()
229 def fixup_user_hg(user):
231 # stole this from hg-git
232 return re.sub('[<>\n]', '?', name.lstrip('< ').rstrip('> '))
234 m = AUTHOR_HG_RE.match(user)
236 name = sanitize(m.group(1))
237 mail = sanitize(m.group(2))
240 name += ' ext:(' + urllib.quote(ex) + ')'
242 name = sanitize(user)
250 def fixup_user(user):
251 global mode, bad_mail
254 name, mail = fixup_user_git(user)
256 name, mail = fixup_user_hg(user)
263 return '%s <%s>' % (name, mail)
265 def get_repo(url, alias):
269 myui.setconfig('ui', 'interactive', 'off')
272 repo = hg.repository(myui, url)
274 local_path = os.path.join(dirname, 'clone')
275 if not os.path.exists(local_path):
276 peer, dstpeer = hg.clone(myui, {}, url, local_path, update=False, pull=True)
277 repo = dstpeer.local()
279 repo = hg.repository(myui, local_path)
280 peer = hg.peer(myui, {}, url)
281 repo.pull(peer, heads=None, force=True)
285 def rev_to_mark(rev):
287 return marks.from_rev(rev)
289 def mark_to_rev(mark):
291 return marks.to_rev(mark)
293 def export_ref(repo, name, kind, head):
294 global prefix, marks, mode
296 ename = '%s/%s' % (kind, name)
297 tip = marks.get_tip(ename)
299 # mercurial takes too much time checking this
300 if tip and tip == head.rev():
303 revs = xrange(tip, head.rev() + 1)
306 revs = [rev for rev in revs if not marks.is_marked(rev)]
311 (manifest, user, (time, tz), files, desc, extra) = repo.changelog.read(c.node())
312 rev_branch = extra['branch']
314 author = "%s %d %s" % (fixup_user(user), time, gittz(tz))
315 if 'committer' in extra:
316 user, time, tz = extra['committer'].rsplit(' ', 2)
317 committer = "%s %s %s" % (user, time, gittz(int(tz)))
321 parents = [p for p in repo.changelog.parentrevs(rev) if p >= 0]
323 if len(parents) == 0:
324 modified = c.manifest().keys()
327 modified, removed = get_filechanges(repo, c, parents[0])
334 if rev_branch != 'default':
335 extra_msg += 'branch : %s\n' % rev_branch
339 if f not in c.manifest():
341 rename = c.filectx(f).renamed()
343 renames.append((rename[0], f))
346 extra_msg += "rename : %s => %s\n" % e
348 for key, value in extra.iteritems():
349 if key in ('author', 'committer', 'encoding', 'message', 'branch', 'hg-git'):
352 extra_msg += "extra : %s : %s\n" % (key, urllib.quote(value))
355 desc += '\n--HG--\n' + extra_msg
357 if len(parents) == 0 and rev:
358 print 'reset %s/%s' % (prefix, ename)
360 print "commit %s/%s" % (prefix, ename)
361 print "mark :%d" % (marks.get_mark(rev))
362 print "author %s" % (author)
363 print "committer %s" % (committer)
364 print "data %d" % (len(desc))
368 print "from :%s" % (rev_to_mark(parents[0]))
370 print "merge :%s" % (rev_to_mark(parents[1]))
373 export_file(c.filectx(f))
379 if (count % 100 == 0):
380 print "progress revision %d '%s' (%d/%d)" % (rev, name, count, len(revs))
381 print "#############################################################"
383 # make sure the ref is updated
384 print "reset %s/%s" % (prefix, ename)
385 print "from :%u" % rev_to_mark(rev)
388 marks.set_tip(ename, rev)
390 def export_tag(repo, tag):
391 export_ref(repo, tag, 'tags', repo[tag])
393 def export_bookmark(repo, bmark):
395 export_ref(repo, bmark, 'bookmarks', head)
397 def export_branch(repo, branch):
398 tip = get_branch_tip(repo, branch)
400 export_ref(repo, branch, 'branches', head)
402 def export_head(repo):
404 export_ref(repo, g_head[0], 'bookmarks', g_head[1])
406 def do_capabilities(parser):
407 global prefix, dirname
411 print "refspec refs/heads/branches/*:%s/branches/*" % prefix
412 print "refspec refs/heads/*:%s/bookmarks/*" % prefix
413 print "refspec refs/tags/*:%s/tags/*" % prefix
415 path = os.path.join(dirname, 'marks-git')
417 if os.path.exists(path):
418 print "*import-marks %s" % path
419 print "*export-marks %s" % path
423 def get_branch_tip(repo, branch):
426 heads = branches.get(branch, None)
430 # verify there's only one head
432 warn("Branch '%s' has more than one head, consider merging" % branch)
433 # older versions of mercurial don't have this
434 if hasattr(repo, "branchtip"):
435 return repo.branchtip(branch)
439 def list_head(repo, cur):
440 global g_head, bmarks
442 head = bookmarks.readcurrent(repo)
446 # fake bookmark from current branch
453 if head == 'default':
457 print "@refs/heads/%s HEAD" % head
458 g_head = (head, node)
461 global branches, bmarks, mode, track_branches
464 for bmark, node in bookmarks.listbookmarks(repo).iteritems():
465 bmarks[bmark] = repo[node]
467 cur = repo.dirstate.branch()
472 for branch in repo.branchmap():
473 heads = repo.branchheads(branch)
475 branches[branch] = heads
477 for branch in branches:
478 print "? refs/heads/branches/%s" % branch
481 print "? refs/heads/%s" % bmark
483 for tag, node in repo.tagslist():
486 print "? refs/tags/%s" % tag
490 def do_import(parser):
493 path = os.path.join(dirname, 'marks-git')
496 if os.path.exists(path):
497 print "feature import-marks=%s" % path
498 print "feature export-marks=%s" % path
501 tmp = encoding.encoding
502 encoding.encoding = 'utf-8'
504 # lets get all the import lines
505 while parser.check('import'):
510 elif ref.startswith('refs/heads/branches/'):
511 branch = ref[len('refs/heads/branches/'):]
512 export_branch(repo, branch)
513 elif ref.startswith('refs/heads/'):
514 bmark = ref[len('refs/heads/'):]
515 export_bookmark(repo, bmark)
516 elif ref.startswith('refs/tags/'):
517 tag = ref[len('refs/tags/'):]
518 export_tag(repo, tag)
522 encoding.encoding = tmp
526 def parse_blob(parser):
530 mark = parser.get_mark()
532 data = parser.get_data()
533 blob_marks[mark] = data
537 def get_merge_files(repo, p1, p2, files):
538 for e in repo[p1].files():
540 if e not in repo[p1].manifest():
542 f = { 'ctx' : repo[p1][e] }
545 def parse_commit(parser):
546 global marks, blob_marks, bmarks, parsed_refs
549 from_mark = merge_mark = None
554 commit_mark = parser.get_mark()
556 author = parser.get_author()
558 committer = parser.get_author()
560 data = parser.get_data()
562 if parser.check('from'):
563 from_mark = parser.get_mark()
565 if parser.check('merge'):
566 merge_mark = parser.get_mark()
568 if parser.check('merge'):
569 die('octopus merges are not supported yet')
574 if parser.check('M'):
575 t, m, mark_ref, path = line.split(' ', 3)
576 mark = int(mark_ref[1:])
577 f = { 'mode' : hgmode(m), 'data' : blob_marks[mark] }
578 elif parser.check('D'):
579 t, path = line.split(' ')
580 f = { 'deleted' : True }
582 die('Unknown file command: %s' % line)
585 def getfilectx(repo, memctx, f):
591 is_exec = of['mode'] == 'x'
592 is_link = of['mode'] == 'l'
593 rename = of.get('rename', None)
594 return context.memfilectx(f, of['data'],
595 is_link, is_exec, rename)
599 user, date, tz = author
602 if committer != author:
603 extra['committer'] = "%s %u %u" % committer
606 p1 = repo.changelog.node(mark_to_rev(from_mark))
611 p2 = repo.changelog.node(mark_to_rev(merge_mark))
616 # If files changed from any of the parents, hg wants to know, but in git if
617 # nothing changed from the first parent, nothing changed.
620 get_merge_files(repo, p1, p2, files)
623 i = data.find('\n--HG--\n')
625 tmp = data[i + len('\n--HG--\n'):].strip()
626 for k, v in [e.split(' : ') for e in tmp.split('\n')]:
628 old, new = v.split(' => ', 1)
629 files[new]['rename'] = old
633 ek, ev = v.split(' : ', 1)
634 extra[ek] = urllib.unquote(ev)
637 ctx = context.memctx(repo, (p1, p2), data,
638 files.keys(), getfilectx,
639 user, (date, tz), extra)
641 tmp = encoding.encoding
642 encoding.encoding = 'utf-8'
644 node = repo.commitctx(ctx)
646 encoding.encoding = tmp
648 rev = repo[node].rev()
650 parsed_refs[ref] = node
652 marks.new_mark(rev, commit_mark)
654 def parse_reset(parser):
658 if parser.check('commit'):
661 if not parser.check('from'):
663 from_mark = parser.get_mark()
666 node = parser.repo.changelog.node(mark_to_rev(from_mark))
667 parsed_refs[ref] = node
669 def parse_tag(parser):
672 from_mark = parser.get_mark()
674 tagger = parser.get_author()
676 data = parser.get_data()
681 def do_export(parser):
682 global parsed_refs, bmarks, peer
686 for line in parser.each_block('done'):
687 if parser.check('blob'):
689 elif parser.check('commit'):
691 elif parser.check('reset'):
693 elif parser.check('tag'):
695 elif parser.check('feature'):
698 die('unhandled export command: %s' % line)
700 for ref, node in parsed_refs.iteritems():
701 if ref.startswith('refs/heads/branches'):
703 elif ref.startswith('refs/heads/'):
704 bmark = ref[len('refs/heads/'):]
706 old = bmarks[bmark].hex()
709 if not bookmarks.pushbookmark(parser.repo, bmark, old, node):
711 elif ref.startswith('refs/tags/'):
712 tag = ref[len('refs/tags/'):]
713 parser.repo.tag([tag], node, None, True, None, {})
715 # transport-helper/fast-export bugs
722 parser.repo.push(peer, force=False)
724 def fix_path(alias, repo, orig_url):
725 repo_url = util.url(repo.url())
726 url = util.url(orig_url)
727 if str(url) == str(repo_url):
729 cmd = ['git', 'config', 'remote.%s.url' % alias, "hg::%s" % repo_url]
733 global prefix, dirname, branches, bmarks
734 global marks, blob_marks, parsed_refs
735 global peer, mode, bad_mail, bad_name
736 global track_branches
742 hg_git_compat = False
743 track_branches = True
745 if get_config('remote-hg.hg-git-compat') == 'true\n':
747 track_branches = False
748 if get_config('remote-hg.track-branches') == 'false\n':
749 track_branches = False
750 except subprocess.CalledProcessError:
755 bad_mail = 'none@none'
764 alias = util.sha1(alias).hexdigest()
768 gitdir = os.environ['GIT_DIR']
769 dirname = os.path.join(gitdir, 'hg', alias)
775 repo = get_repo(url, alias)
776 prefix = 'refs/hg/%s' % alias
779 fix_path(alias, peer or repo, url)
781 if not os.path.exists(dirname):
784 marks_path = os.path.join(dirname, 'marks-hg')
785 marks = Marks(marks_path)
787 parser = Parser(repo)
789 if parser.check('capabilities'):
790 do_capabilities(parser)
791 elif parser.check('list'):
793 elif parser.check('import'):
795 elif parser.check('export'):
798 die('unhandled command: %s' % line)
804 shutil.rmtree(dirname)
806 sys.exit(main(sys.argv))