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 $PATH.
9 # Then you can clone with:
10 # git clone hg::/path/to/mercurial/repo/
12 # For remote repositories a local clone is stored in
13 # "$GIT_DIR/hg/origin/clone/.hg/".
15 from mercurial import hg, ui, bookmarks, context, encoding, node, error, extensions
25 import urlparse, hashlib
28 # If you want to switch to hg-git compatibility mode:
29 # git config --global remote-hg.hg-git-compat true
31 # If you are not in hg-git-compat mode and want to disable the tracking of
33 # git config --global remote-hg.track-branches false
35 # If you don't want to force pushes (and thus risk creating new remote heads):
36 # git config --global remote-hg.force-push false
38 # If you want the equivalent of hg's clone/pull--insecure option:
39 # git config remote-hg.insecure true
42 # Sensible defaults for git.
43 # hg bookmarks are exported as git branches, hg branches are prefixed
44 # with 'branches/', HEAD is a special case.
48 # Only hg bookmarks are exported as git branches.
49 # Commits are modified to preserve hg information and allow bidirectionality.
52 NAME_RE = re.compile('^([^<>]+)')
53 AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
54 EMAIL_RE = re.compile('^([^<>]+[^ \\\t<>])?\\b(?:[ \\t<>]*?)\\b([^ \\t<>]+@[^ \\t<>]+)')
55 AUTHOR_HG_RE = re.compile('^(.*?) ?<(.*?)(?:>(.+)?)?$')
56 RAW_AUTHOR_RE = re.compile('^(\w+) (?:(.+)? )?<(.*)> (\d+) ([+-]\d+)')
59 sys.stderr.write('ERROR: %s\n' % (msg % args))
63 sys.stderr.write('WARNING: %s\n' % (msg % args))
66 return 'l' in flags and '120000' or 'x' in flags and '100755' or '100644'
69 return '%+03d%02d' % (-tz / 3600, -tz % 3600 / 60)
72 m = { '100755': 'x', '120000': 'l' }
73 return m.get(mode, '')
76 return hg.node.hex(node)
79 return ref.replace('___', ' ')
82 return ref.replace(' ', '___')
84 def get_config(config):
85 cmd = ['git', 'config', '--get', config]
86 process = subprocess.Popen(cmd, stdout=subprocess.PIPE)
87 output, _ = process.communicate()
92 def __init__(self, path):
102 if not os.path.exists(self.path):
105 tmp = json.load(open(self.path))
107 self.tips = tmp['tips']
108 self.marks = tmp['marks']
109 self.last_mark = tmp['last-mark']
111 for rev, mark in self.marks.iteritems():
112 self.rev_marks[mark] = int(rev)
115 return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
118 json.dump(self.dict(), open(self.path, 'w'))
121 return str(self.dict())
123 def from_rev(self, rev):
124 return self.marks[str(rev)]
126 def to_rev(self, mark):
127 return self.rev_marks[mark]
131 return self.last_mark
133 def get_mark(self, rev):
135 self.marks[str(rev)] = self.last_mark
136 return self.last_mark
138 def new_mark(self, rev, mark):
139 self.marks[str(rev)] = mark
140 self.rev_marks[mark] = rev
141 self.last_mark = mark
143 def is_marked(self, rev):
144 return str(rev) in self.marks
146 def get_tip(self, branch):
147 return self.tips.get(branch, 0)
149 def set_tip(self, branch, tip):
150 self.tips[branch] = tip
154 def __init__(self, repo):
156 self.line = self.get_line()
159 return sys.stdin.readline().strip()
161 def __getitem__(self, i):
162 return self.line.split()[i]
164 def check(self, word):
165 return self.line.startswith(word)
167 def each_block(self, separator):
168 while self.line != separator:
170 self.line = self.get_line()
173 return self.each_block('')
176 self.line = self.get_line()
177 if self.line == 'done':
181 i = self.line.index(':') + 1
182 return int(self.line[i:])
185 if not self.check('data'):
187 i = self.line.index(' ') + 1
188 size = int(self.line[i:])
189 return sys.stdin.read(size)
191 def get_author(self):
195 m = RAW_AUTHOR_RE.match(self.line)
198 _, name, email, date, tz = m.groups()
199 if name and 'ext:' in name:
200 m = re.match('^(.+?) ext:\((.+)\)$', name)
203 ex = urllib.unquote(m.group(2))
205 if email != bad_mail:
207 user = '%s <%s>' % (name, email)
209 user = '<%s>' % (email)
217 tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
218 return (user, int(date), -tz)
220 def fix_file_path(path):
221 if not os.path.isabs(path):
223 return os.path.relpath(path, '/')
225 def export_files(files):
226 global marks, filenodes
230 fid = node.hex(f.filenode())
233 mark = filenodes[fid]
235 mark = marks.next_mark()
236 filenodes[fid] = mark
240 print "mark :%u" % mark
241 print "data %d" % len(d)
244 path = fix_file_path(f.path())
245 final.append((gitmode(f.flags()), mark, path))
249 def get_filechanges(repo, ctx, parent):
254 # load earliest manifest first for caching reasons
255 prev = repo[parent].manifest().copy()
260 if (cur.flags(fn) != prev.flags(fn) or cur[fn] != prev[fn]):
265 removed |= set(prev.keys())
267 return added | modified, removed
269 def fixup_user_git(user):
271 user = user.replace('"', '')
272 m = AUTHOR_RE.match(user)
275 mail = m.group(2).strip()
277 m = EMAIL_RE.match(user)
282 m = NAME_RE.match(user)
284 name = m.group(1).strip()
287 def fixup_user_hg(user):
289 # stole this from hg-git
290 return re.sub('[<>\n]', '?', name.lstrip('< ').rstrip('> '))
292 m = AUTHOR_HG_RE.match(user)
294 name = sanitize(m.group(1))
295 mail = sanitize(m.group(2))
298 name += ' ext:(' + urllib.quote(ex) + ')'
300 name = sanitize(user)
308 def fixup_user(user):
309 global mode, bad_mail
312 name, mail = fixup_user_git(user)
314 name, mail = fixup_user_hg(user)
321 return '%s <%s>' % (name, mail)
323 def get_repo(url, alias):
327 myui.setconfig('ui', 'interactive', 'off')
328 myui.fout = sys.stderr
331 if get_config('remote-hg.insecure') == 'true\n':
332 myui.setconfig('web', 'cacerts', '')
333 except subprocess.CalledProcessError:
337 mod = extensions.load(myui, 'hgext.schemes', None)
343 repo = hg.repository(myui, url)
345 local_path = os.path.join(dirname, 'clone')
346 if not os.path.exists(local_path):
348 peer, dstpeer = hg.clone(myui, {}, url, local_path, update=True, pull=True)
350 die('Repository error')
351 repo = dstpeer.local()
353 repo = hg.repository(myui, local_path)
355 peer = hg.peer(myui, {}, url)
357 die('Repository error')
358 repo.pull(peer, heads=None, force=True)
362 def rev_to_mark(rev):
364 return marks.from_rev(rev)
366 def mark_to_rev(mark):
368 return marks.to_rev(mark)
370 def export_ref(repo, name, kind, head):
371 global prefix, marks, mode
373 ename = '%s/%s' % (kind, name)
374 tip = marks.get_tip(ename)
376 revs = xrange(tip, head.rev() + 1)
379 revs = [rev for rev in revs if not marks.is_marked(rev)]
384 (manifest, user, (time, tz), files, desc, extra) = repo.changelog.read(c.node())
385 rev_branch = extra['branch']
387 author = "%s %d %s" % (fixup_user(user), time, gittz(tz))
388 if 'committer' in extra:
389 user, time, tz = extra['committer'].rsplit(' ', 2)
390 committer = "%s %s %s" % (user, time, gittz(int(tz)))
394 parents = [p for p in repo.changelog.parentrevs(rev) if p >= 0]
396 if len(parents) == 0:
397 modified = c.manifest().keys()
400 modified, removed = get_filechanges(repo, c, parents[0])
407 if rev_branch != 'default':
408 extra_msg += 'branch : %s\n' % rev_branch
412 if f not in c.manifest():
414 rename = c.filectx(f).renamed()
416 renames.append((rename[0], f))
419 extra_msg += "rename : %s => %s\n" % e
421 for key, value in extra.iteritems():
422 if key in ('author', 'committer', 'encoding', 'message', 'branch', 'hg-git'):
425 extra_msg += "extra : %s : %s\n" % (key, urllib.quote(value))
428 desc += '\n--HG--\n' + extra_msg
430 if len(parents) == 0 and rev:
431 print 'reset %s/%s' % (prefix, ename)
433 modified_final = export_files(c.filectx(f) for f in modified)
435 print "commit %s/%s" % (prefix, ename)
436 print "mark :%d" % (marks.get_mark(rev))
437 print "author %s" % (author)
438 print "committer %s" % (committer)
439 print "data %d" % (len(desc))
443 print "from :%s" % (rev_to_mark(parents[0]))
445 print "merge :%s" % (rev_to_mark(parents[1]))
447 for f in modified_final:
448 print "M %s :%u %s" % f
450 print "D %s" % (fix_file_path(f))
454 if (count % 100 == 0):
455 print "progress revision %d '%s' (%d/%d)" % (rev, name, count, len(revs))
456 print "#############################################################"
458 # make sure the ref is updated
459 print "reset %s/%s" % (prefix, ename)
460 print "from :%u" % rev_to_mark(rev)
463 marks.set_tip(ename, rev)
465 def export_tag(repo, tag):
466 export_ref(repo, tag, 'tags', repo[hgref(tag)])
468 def export_bookmark(repo, bmark):
469 head = bmarks[hgref(bmark)]
470 export_ref(repo, bmark, 'bookmarks', head)
472 def export_branch(repo, branch):
473 tip = get_branch_tip(repo, branch)
475 export_ref(repo, branch, 'branches', head)
477 def export_head(repo):
479 export_ref(repo, g_head[0], 'bookmarks', g_head[1])
481 def do_capabilities(parser):
482 global prefix, dirname
486 print "refspec refs/heads/branches/*:%s/branches/*" % prefix
487 print "refspec refs/heads/*:%s/bookmarks/*" % prefix
488 print "refspec refs/tags/*:%s/tags/*" % prefix
490 path = os.path.join(dirname, 'marks-git')
492 if os.path.exists(path):
493 print "*import-marks %s" % path
494 print "*export-marks %s" % path
498 def branch_tip(repo, branch):
499 # older versions of mercurial don't have this
500 if hasattr(repo, 'branchtip'):
501 return repo.branchtip(branch)
503 return repo.branchtags()[branch]
505 def get_branch_tip(repo, branch):
508 heads = branches.get(hgref(branch), None)
512 # verify there's only one head
514 warn("Branch '%s' has more than one head, consider merging" % branch)
515 return branch_tip(repo, hgref(branch))
519 def list_head(repo, cur):
520 global g_head, bmarks
522 head = bookmarks.readcurrent(repo)
526 # fake bookmark from current branch
533 if head == 'default':
538 print "@refs/heads/%s HEAD" % head
539 g_head = (head, node)
542 global branches, bmarks, mode, track_branches
545 for bmark, node in bookmarks.listbookmarks(repo).iteritems():
546 bmarks[bmark] = repo[node]
548 cur = repo.dirstate.branch()
553 for branch in repo.branchmap():
554 heads = repo.branchheads(branch)
556 branches[branch] = heads
558 for branch in branches:
559 print "? refs/heads/branches/%s" % gitref(branch)
562 print "? refs/heads/%s" % gitref(bmark)
564 for tag, node in repo.tagslist():
567 print "? refs/tags/%s" % gitref(tag)
571 def do_import(parser):
574 path = os.path.join(dirname, 'marks-git')
577 if os.path.exists(path):
578 print "feature import-marks=%s" % path
579 print "feature export-marks=%s" % path
582 tmp = encoding.encoding
583 encoding.encoding = 'utf-8'
585 # lets get all the import lines
586 while parser.check('import'):
591 elif ref.startswith('refs/heads/branches/'):
592 branch = ref[len('refs/heads/branches/'):]
593 export_branch(repo, branch)
594 elif ref.startswith('refs/heads/'):
595 bmark = ref[len('refs/heads/'):]
596 export_bookmark(repo, bmark)
597 elif ref.startswith('refs/tags/'):
598 tag = ref[len('refs/tags/'):]
599 export_tag(repo, tag)
603 encoding.encoding = tmp
607 def parse_blob(parser):
611 mark = parser.get_mark()
613 data = parser.get_data()
614 blob_marks[mark] = data
617 def get_merge_files(repo, p1, p2, files):
618 for e in repo[p1].files():
620 if e not in repo[p1].manifest():
622 f = { 'ctx' : repo[p1][e] }
625 def parse_commit(parser):
626 global marks, blob_marks, parsed_refs
629 from_mark = merge_mark = None
634 commit_mark = parser.get_mark()
636 author = parser.get_author()
638 committer = parser.get_author()
640 data = parser.get_data()
642 if parser.check('from'):
643 from_mark = parser.get_mark()
645 if parser.check('merge'):
646 merge_mark = parser.get_mark()
648 if parser.check('merge'):
649 die('octopus merges are not supported yet')
651 # fast-export adds an extra newline
658 if parser.check('M'):
659 t, m, mark_ref, path = line.split(' ', 3)
660 mark = int(mark_ref[1:])
661 f = { 'mode' : hgmode(m), 'data' : blob_marks[mark] }
662 elif parser.check('D'):
663 t, path = line.split(' ', 1)
664 f = { 'deleted' : True }
666 die('Unknown file command: %s' % line)
669 def getfilectx(repo, memctx, f):
675 is_exec = of['mode'] == 'x'
676 is_link = of['mode'] == 'l'
677 rename = of.get('rename', None)
678 return context.memfilectx(f, of['data'],
679 is_link, is_exec, rename)
683 user, date, tz = author
686 if committer != author:
687 extra['committer'] = "%s %u %u" % committer
690 p1 = repo.changelog.node(mark_to_rev(from_mark))
695 p2 = repo.changelog.node(mark_to_rev(merge_mark))
700 # If files changed from any of the parents, hg wants to know, but in git if
701 # nothing changed from the first parent, nothing changed.
704 get_merge_files(repo, p1, p2, files)
706 # Check if the ref is supposed to be a named branch
707 if ref.startswith('refs/heads/branches/'):
708 branch = ref[len('refs/heads/branches/'):]
709 extra['branch'] = hgref(branch)
712 i = data.find('\n--HG--\n')
714 tmp = data[i + len('\n--HG--\n'):].strip()
715 for k, v in [e.split(' : ', 1) for e in tmp.split('\n')]:
717 old, new = v.split(' => ', 1)
718 files[new]['rename'] = old
722 ek, ev = v.split(' : ', 1)
723 extra[ek] = urllib.unquote(ev)
726 ctx = context.memctx(repo, (p1, p2), data,
727 files.keys(), getfilectx,
728 user, (date, tz), extra)
730 tmp = encoding.encoding
731 encoding.encoding = 'utf-8'
733 node = repo.commitctx(ctx)
735 encoding.encoding = tmp
737 rev = repo[node].rev()
739 parsed_refs[ref] = node
740 marks.new_mark(rev, commit_mark)
742 def parse_reset(parser):
748 if parser.check('commit'):
751 if not parser.check('from'):
753 from_mark = parser.get_mark()
756 node = parser.repo.changelog.node(mark_to_rev(from_mark))
757 parsed_refs[ref] = node
759 def parse_tag(parser):
762 from_mark = parser.get_mark()
764 tagger = parser.get_author()
766 data = parser.get_data()
769 parsed_tags[name] = (tagger, data)
771 def write_tag(repo, tag, node, msg, author):
772 branch = repo[node].branch()
773 tip = branch_tip(repo, branch)
776 def getfilectx(repo, memctx, f):
778 fctx = tip.filectx(f)
780 except error.ManifestLookupError:
782 content = data + "%s %s\n" % (hghex(node), tag)
783 return context.memfilectx(f, content, False, False, None)
788 author = (None, 0, 0)
789 user, date, tz = author
791 ctx = context.memctx(repo, (p1, p2), msg,
792 ['.hgtags'], getfilectx,
793 user, (date, tz), {'branch' : branch})
795 tmp = encoding.encoding
796 encoding.encoding = 'utf-8'
798 tagnode = repo.commitctx(ctx)
800 encoding.encoding = tmp
804 def do_export(parser):
805 global parsed_refs, bmarks, peer
811 for line in parser.each_block('done'):
812 if parser.check('blob'):
814 elif parser.check('commit'):
816 elif parser.check('reset'):
818 elif parser.check('tag'):
820 elif parser.check('feature'):
823 die('unhandled export command: %s' % line)
825 for ref, node in parsed_refs.iteritems():
826 if ref.startswith('refs/heads/branches'):
827 branch = ref[len('refs/heads/branches/'):]
828 if branch in branches and node in branches[branch]:
832 elif ref.startswith('refs/heads/'):
833 bmark = ref[len('refs/heads/'):]
834 p_bmarks.append((bmark, node))
836 elif ref.startswith('refs/tags/'):
837 tag = ref[len('refs/tags/'):]
839 author, msg = parsed_tags.get(tag, (None, None))
842 msg = 'Added tag %s for changeset %s' % (tag, hghex(node[:6]));
843 write_tag(parser.repo, tag, node, msg, author)
845 fp = parser.repo.opener('localtags', 'a')
846 fp.write('%s %s\n' % (hghex(node), tag))
850 # transport-helper/fast-export bugs
854 parser.repo.push(peer, force=force_push)
857 for bmark, node in p_bmarks:
858 ref = 'refs/heads/' + bmark
862 old = bmarks[bmark].hex()
869 if bmark == 'master' and 'master' not in parser.repo._bookmarks:
872 elif bookmarks.pushbookmark(parser.repo, bmark, old, new):
876 print "error %s" % ref
880 rb = peer.listkeys('bookmarks')
881 old = rb.get(bmark, '')
882 if not peer.pushkey('bookmarks', bmark, old, new):
883 print "error %s" % ref
890 def fix_path(alias, repo, orig_url):
891 url = urlparse.urlparse(orig_url, 'file')
892 if url.scheme != 'file' or os.path.isabs(url.path):
894 abs_url = urlparse.urljoin("%s/" % os.getcwd(), orig_url)
895 cmd = ['git', 'config', 'remote.%s.url' % alias, "hg::%s" % abs_url]
899 global prefix, dirname, branches, bmarks
900 global marks, blob_marks, parsed_refs
901 global peer, mode, bad_mail, bad_name
902 global track_branches, force_push, is_tmp
910 hg_git_compat = False
911 track_branches = True
915 if get_config('remote-hg.hg-git-compat') == 'true\n':
917 track_branches = False
918 if get_config('remote-hg.track-branches') == 'false\n':
919 track_branches = False
920 if get_config('remote-hg.force-push') == 'false\n':
922 except subprocess.CalledProcessError:
927 bad_mail = 'none@none'
936 alias = hashlib.sha1(alias).hexdigest()
940 gitdir = os.environ['GIT_DIR']
941 dirname = os.path.join(gitdir, 'hg', alias)
950 repo = get_repo(url, alias)
951 prefix = 'refs/hg/%s' % alias
954 fix_path(alias, peer or repo, url)
956 if not os.path.exists(dirname):
959 marks_path = os.path.join(dirname, 'marks-hg')
960 marks = Marks(marks_path)
962 parser = Parser(repo)
964 if parser.check('capabilities'):
965 do_capabilities(parser)
966 elif parser.check('list'):
968 elif parser.check('import'):
970 elif parser.check('export'):
973 die('unhandled command: %s' % line)
982 shutil.rmtree(dirname)
985 sys.exit(main(sys.argv))