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 are not in hg-git-compat mode and want to disable the tracking of
30 # git config --global remote-hg.track-branches false
32 # If you don't want to force pushes (and thus risk creating new remote heads):
33 # git config --global remote-hg.force-push false
35 # If you want the equivalent of hg's clone/pull--insecure option:
36 # git config --global remote-hg.insecure true
38 # If you want to switch to hg-git compatibility mode:
39 # git config --global remote-hg.hg-git-compat 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()
90 def get_config_bool(config, default=False):
91 value = get_config(config).rstrip('\n')
94 elif value == "false":
101 def __init__(self, path):
111 if not os.path.exists(self.path):
114 tmp = json.load(open(self.path))
116 self.tips = tmp['tips']
117 self.marks = tmp['marks']
118 self.last_mark = tmp['last-mark']
120 for rev, mark in self.marks.iteritems():
121 self.rev_marks[mark] = int(rev)
124 return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
127 json.dump(self.dict(), open(self.path, 'w'))
130 return str(self.dict())
132 def from_rev(self, rev):
133 return self.marks[str(rev)]
135 def to_rev(self, mark):
136 return self.rev_marks[mark]
140 return self.last_mark
142 def get_mark(self, rev):
144 self.marks[str(rev)] = self.last_mark
145 return self.last_mark
147 def new_mark(self, rev, mark):
148 self.marks[str(rev)] = mark
149 self.rev_marks[mark] = rev
150 self.last_mark = mark
152 def is_marked(self, rev):
153 return str(rev) in self.marks
155 def get_tip(self, branch):
156 return self.tips.get(branch, 0)
158 def set_tip(self, branch, tip):
159 self.tips[branch] = tip
163 def __init__(self, repo):
165 self.line = self.get_line()
168 return sys.stdin.readline().strip()
170 def __getitem__(self, i):
171 return self.line.split()[i]
173 def check(self, word):
174 return self.line.startswith(word)
176 def each_block(self, separator):
177 while self.line != separator:
179 self.line = self.get_line()
182 return self.each_block('')
185 self.line = self.get_line()
186 if self.line == 'done':
190 i = self.line.index(':') + 1
191 return int(self.line[i:])
194 if not self.check('data'):
196 i = self.line.index(' ') + 1
197 size = int(self.line[i:])
198 return sys.stdin.read(size)
200 def get_author(self):
204 m = RAW_AUTHOR_RE.match(self.line)
207 _, name, email, date, tz = m.groups()
208 if name and 'ext:' in name:
209 m = re.match('^(.+?) ext:\((.+)\)$', name)
212 ex = urllib.unquote(m.group(2))
214 if email != bad_mail:
216 user = '%s <%s>' % (name, email)
218 user = '<%s>' % (email)
226 tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
227 return (user, int(date), -tz)
229 def fix_file_path(path):
230 if not os.path.isabs(path):
232 return os.path.relpath(path, '/')
234 def export_files(files):
235 global marks, filenodes
239 fid = node.hex(f.filenode())
242 mark = filenodes[fid]
244 mark = marks.next_mark()
245 filenodes[fid] = mark
249 print "mark :%u" % mark
250 print "data %d" % len(d)
253 path = fix_file_path(f.path())
254 final.append((gitmode(f.flags()), mark, path))
258 def get_filechanges(repo, ctx, parent):
263 # load earliest manifest first for caching reasons
264 prev = repo[parent].manifest().copy()
269 if (cur.flags(fn) != prev.flags(fn) or cur[fn] != prev[fn]):
274 removed |= set(prev.keys())
276 return added | modified, removed
278 def fixup_user_git(user):
280 user = user.replace('"', '')
281 m = AUTHOR_RE.match(user)
284 mail = m.group(2).strip()
286 m = EMAIL_RE.match(user)
291 m = NAME_RE.match(user)
293 name = m.group(1).strip()
296 def fixup_user_hg(user):
298 # stole this from hg-git
299 return re.sub('[<>\n]', '?', name.lstrip('< ').rstrip('> '))
301 m = AUTHOR_HG_RE.match(user)
303 name = sanitize(m.group(1))
304 mail = sanitize(m.group(2))
307 name += ' ext:(' + urllib.quote(ex) + ')'
309 name = sanitize(user)
317 def fixup_user(user):
318 global mode, bad_mail
321 name, mail = fixup_user_git(user)
323 name, mail = fixup_user_hg(user)
330 return '%s <%s>' % (name, mail)
332 def get_repo(url, alias):
336 myui.setconfig('ui', 'interactive', 'off')
337 myui.fout = sys.stderr
339 if get_config_bool('remote-hg.insecure'):
340 myui.setconfig('web', 'cacerts', '')
343 mod = extensions.load(myui, 'hgext.schemes', None)
349 repo = hg.repository(myui, url)
351 local_path = os.path.join(dirname, 'clone')
352 if not os.path.exists(local_path):
354 peer, dstpeer = hg.clone(myui, {}, url, local_path, update=True, pull=True)
356 die('Repository error')
357 repo = dstpeer.local()
359 repo = hg.repository(myui, local_path)
361 peer = hg.peer(myui, {}, url)
363 die('Repository error')
364 repo.pull(peer, heads=None, force=True)
368 def rev_to_mark(rev):
370 return marks.from_rev(rev)
372 def mark_to_rev(mark):
374 return marks.to_rev(mark)
376 def export_ref(repo, name, kind, head):
377 global prefix, marks, mode
379 ename = '%s/%s' % (kind, name)
380 tip = marks.get_tip(ename)
382 revs = xrange(tip, head.rev() + 1)
385 revs = [rev for rev in revs if not marks.is_marked(rev)]
390 (manifest, user, (time, tz), files, desc, extra) = repo.changelog.read(c.node())
391 rev_branch = extra['branch']
393 author = "%s %d %s" % (fixup_user(user), time, gittz(tz))
394 if 'committer' in extra:
395 user, time, tz = extra['committer'].rsplit(' ', 2)
396 committer = "%s %s %s" % (user, time, gittz(int(tz)))
400 parents = [p for p in repo.changelog.parentrevs(rev) if p >= 0]
402 if len(parents) == 0:
403 modified = c.manifest().keys()
406 modified, removed = get_filechanges(repo, c, parents[0])
413 if rev_branch != 'default':
414 extra_msg += 'branch : %s\n' % rev_branch
418 if f not in c.manifest():
420 rename = c.filectx(f).renamed()
422 renames.append((rename[0], f))
425 extra_msg += "rename : %s => %s\n" % e
427 for key, value in extra.iteritems():
428 if key in ('author', 'committer', 'encoding', 'message', 'branch', 'hg-git'):
431 extra_msg += "extra : %s : %s\n" % (key, urllib.quote(value))
434 desc += '\n--HG--\n' + extra_msg
436 if len(parents) == 0 and rev:
437 print 'reset %s/%s' % (prefix, ename)
439 modified_final = export_files(c.filectx(f) for f in modified)
441 print "commit %s/%s" % (prefix, ename)
442 print "mark :%d" % (marks.get_mark(rev))
443 print "author %s" % (author)
444 print "committer %s" % (committer)
445 print "data %d" % (len(desc))
449 print "from :%s" % (rev_to_mark(parents[0]))
451 print "merge :%s" % (rev_to_mark(parents[1]))
453 for f in modified_final:
454 print "M %s :%u %s" % f
456 print "D %s" % (fix_file_path(f))
460 if (count % 100 == 0):
461 print "progress revision %d '%s' (%d/%d)" % (rev, name, count, len(revs))
463 # make sure the ref is updated
464 print "reset %s/%s" % (prefix, ename)
465 print "from :%u" % rev_to_mark(rev)
468 marks.set_tip(ename, rev)
470 def export_tag(repo, tag):
471 export_ref(repo, tag, 'tags', repo[hgref(tag)])
473 def export_bookmark(repo, bmark):
474 head = bmarks[hgref(bmark)]
475 export_ref(repo, bmark, 'bookmarks', head)
477 def export_branch(repo, branch):
478 tip = get_branch_tip(repo, branch)
480 export_ref(repo, branch, 'branches', head)
482 def export_head(repo):
484 export_ref(repo, g_head[0], 'bookmarks', g_head[1])
486 def do_capabilities(parser):
487 global prefix, dirname
491 print "refspec refs/heads/branches/*:%s/branches/*" % prefix
492 print "refspec refs/heads/*:%s/bookmarks/*" % prefix
493 print "refspec refs/tags/*:%s/tags/*" % prefix
495 path = os.path.join(dirname, 'marks-git')
497 if os.path.exists(path):
498 print "*import-marks %s" % path
499 print "*export-marks %s" % path
503 def branch_tip(repo, branch):
504 # older versions of mercurial don't have this
505 if hasattr(repo, 'branchtip'):
506 return repo.branchtip(branch)
508 return repo.branchtags()[branch]
510 def get_branch_tip(repo, branch):
513 heads = branches.get(hgref(branch), None)
517 # verify there's only one head
519 warn("Branch '%s' has more than one head, consider merging" % branch)
520 return branch_tip(repo, hgref(branch))
524 def list_head(repo, cur):
525 global g_head, bmarks
527 head = bookmarks.readcurrent(repo)
531 # fake bookmark from current branch
538 if head == 'default':
543 print "@refs/heads/%s HEAD" % head
544 g_head = (head, node)
547 global branches, bmarks, track_branches
550 for bmark, node in bookmarks.listbookmarks(repo).iteritems():
551 bmarks[bmark] = repo[node]
553 cur = repo.dirstate.branch()
558 for branch in repo.branchmap():
559 heads = repo.branchheads(branch)
561 branches[branch] = heads
563 for branch in branches:
564 print "? refs/heads/branches/%s" % gitref(branch)
567 print "? refs/heads/%s" % gitref(bmark)
569 for tag, node in repo.tagslist():
572 print "? refs/tags/%s" % gitref(tag)
576 def do_import(parser):
579 path = os.path.join(dirname, 'marks-git')
582 if os.path.exists(path):
583 print "feature import-marks=%s" % path
584 print "feature export-marks=%s" % path
587 tmp = encoding.encoding
588 encoding.encoding = 'utf-8'
590 # lets get all the import lines
591 while parser.check('import'):
596 elif ref.startswith('refs/heads/branches/'):
597 branch = ref[len('refs/heads/branches/'):]
598 export_branch(repo, branch)
599 elif ref.startswith('refs/heads/'):
600 bmark = ref[len('refs/heads/'):]
601 export_bookmark(repo, bmark)
602 elif ref.startswith('refs/tags/'):
603 tag = ref[len('refs/tags/'):]
604 export_tag(repo, tag)
608 encoding.encoding = tmp
612 def parse_blob(parser):
616 mark = parser.get_mark()
618 data = parser.get_data()
619 blob_marks[mark] = data
622 def get_merge_files(repo, p1, p2, files):
623 for e in repo[p1].files():
625 if e not in repo[p1].manifest():
627 f = { 'ctx' : repo[p1][e] }
630 def parse_commit(parser):
631 global marks, blob_marks, parsed_refs
634 from_mark = merge_mark = None
639 commit_mark = parser.get_mark()
641 author = parser.get_author()
643 committer = parser.get_author()
645 data = parser.get_data()
647 if parser.check('from'):
648 from_mark = parser.get_mark()
650 if parser.check('merge'):
651 merge_mark = parser.get_mark()
653 if parser.check('merge'):
654 die('octopus merges are not supported yet')
656 # fast-export adds an extra newline
663 if parser.check('M'):
664 t, m, mark_ref, path = line.split(' ', 3)
665 mark = int(mark_ref[1:])
666 f = { 'mode' : hgmode(m), 'data' : blob_marks[mark] }
667 elif parser.check('D'):
668 t, path = line.split(' ', 1)
669 f = { 'deleted' : True }
671 die('Unknown file command: %s' % line)
674 def getfilectx(repo, memctx, f):
680 is_exec = of['mode'] == 'x'
681 is_link = of['mode'] == 'l'
682 rename = of.get('rename', None)
683 return context.memfilectx(f, of['data'],
684 is_link, is_exec, rename)
688 user, date, tz = author
691 if committer != author:
692 extra['committer'] = "%s %u %u" % committer
695 p1 = repo.changelog.node(mark_to_rev(from_mark))
700 p2 = repo.changelog.node(mark_to_rev(merge_mark))
705 # If files changed from any of the parents, hg wants to know, but in git if
706 # nothing changed from the first parent, nothing changed.
709 get_merge_files(repo, p1, p2, files)
711 # Check if the ref is supposed to be a named branch
712 if ref.startswith('refs/heads/branches/'):
713 branch = ref[len('refs/heads/branches/'):]
714 extra['branch'] = hgref(branch)
717 i = data.find('\n--HG--\n')
719 tmp = data[i + len('\n--HG--\n'):].strip()
720 for k, v in [e.split(' : ', 1) for e in tmp.split('\n')]:
722 old, new = v.split(' => ', 1)
723 files[new]['rename'] = old
727 ek, ev = v.split(' : ', 1)
728 extra[ek] = urllib.unquote(ev)
731 ctx = context.memctx(repo, (p1, p2), data,
732 files.keys(), getfilectx,
733 user, (date, tz), extra)
735 tmp = encoding.encoding
736 encoding.encoding = 'utf-8'
738 node = repo.commitctx(ctx)
740 encoding.encoding = tmp
742 rev = repo[node].rev()
744 parsed_refs[ref] = node
745 marks.new_mark(rev, commit_mark)
747 def parse_reset(parser):
753 if parser.check('commit'):
756 if not parser.check('from'):
758 from_mark = parser.get_mark()
761 node = parser.repo.changelog.node(mark_to_rev(from_mark))
762 parsed_refs[ref] = node
764 def parse_tag(parser):
767 from_mark = parser.get_mark()
769 tagger = parser.get_author()
771 data = parser.get_data()
774 parsed_tags[name] = (tagger, data)
776 def write_tag(repo, tag, node, msg, author):
777 branch = repo[node].branch()
778 tip = branch_tip(repo, branch)
781 def getfilectx(repo, memctx, f):
783 fctx = tip.filectx(f)
785 except error.ManifestLookupError:
787 content = data + "%s %s\n" % (hghex(node), tag)
788 return context.memfilectx(f, content, False, False, None)
793 author = (None, 0, 0)
794 user, date, tz = author
796 ctx = context.memctx(repo, (p1, p2), msg,
797 ['.hgtags'], getfilectx,
798 user, (date, tz), {'branch' : branch})
800 tmp = encoding.encoding
801 encoding.encoding = 'utf-8'
803 tagnode = repo.commitctx(ctx)
805 encoding.encoding = tmp
809 def do_export(parser):
810 global parsed_refs, bmarks, peer
816 for line in parser.each_block('done'):
817 if parser.check('blob'):
819 elif parser.check('commit'):
821 elif parser.check('reset'):
823 elif parser.check('tag'):
825 elif parser.check('feature'):
828 die('unhandled export command: %s' % line)
830 for ref, node in parsed_refs.iteritems():
831 if ref.startswith('refs/heads/branches'):
832 branch = ref[len('refs/heads/branches/'):]
833 if branch in branches and node in branches[branch]:
837 elif ref.startswith('refs/heads/'):
838 bmark = ref[len('refs/heads/'):]
839 p_bmarks.append((bmark, node))
841 elif ref.startswith('refs/tags/'):
842 tag = ref[len('refs/tags/'):]
844 author, msg = parsed_tags.get(tag, (None, None))
847 msg = 'Added tag %s for changeset %s' % (tag, hghex(node[:6]));
848 write_tag(parser.repo, tag, node, msg, author)
850 fp = parser.repo.opener('localtags', 'a')
851 fp.write('%s %s\n' % (hghex(node), tag))
855 # transport-helper/fast-export bugs
859 parser.repo.push(peer, force=force_push, newbranch=True)
862 for bmark, node in p_bmarks:
863 ref = 'refs/heads/' + bmark
867 old = bmarks[bmark].hex()
874 if bmark == 'master' and 'master' not in parser.repo._bookmarks:
878 elif bookmarks.pushbookmark(parser.repo, bmark, old, new):
882 print "error %s" % ref
886 rb = peer.listkeys('bookmarks')
887 old = rb.get(bmark, '')
888 if not peer.pushkey('bookmarks', bmark, old, new):
889 print "error %s" % ref
896 def fix_path(alias, repo, orig_url):
897 url = urlparse.urlparse(orig_url, 'file')
898 if url.scheme != 'file' or os.path.isabs(url.path):
900 abs_url = urlparse.urljoin("%s/" % os.getcwd(), orig_url)
901 cmd = ['git', 'config', 'remote.%s.url' % alias, "hg::%s" % abs_url]
905 global prefix, dirname, branches, bmarks
906 global marks, blob_marks, parsed_refs
907 global peer, mode, bad_mail, bad_name
908 global track_branches, force_push, is_tmp
916 hg_git_compat = get_config_bool('remote-hg.hg-git-compat')
917 track_branches = get_config_bool('remote-hg.track-branches', True)
918 force_push = get_config_bool('remote-hg.force-push')
922 bad_mail = 'none@none'
931 alias = hashlib.sha1(alias).hexdigest()
935 gitdir = os.environ['GIT_DIR']
936 dirname = os.path.join(gitdir, 'hg', alias)
945 repo = get_repo(url, alias)
946 prefix = 'refs/hg/%s' % alias
949 fix_path(alias, peer or repo, url)
951 if not os.path.exists(dirname):
954 marks_path = os.path.join(dirname, 'marks-hg')
955 marks = Marks(marks_path)
957 if sys.platform == 'win32':
959 msvcrt.setmode(sys.stdout.fileno(), os.O_BINARY)
961 parser = Parser(repo)
963 if parser.check('capabilities'):
964 do_capabilities(parser)
965 elif parser.check('list'):
967 elif parser.check('import'):
969 elif parser.check('export'):
972 die('unhandled command: %s' % line)
981 shutil.rmtree(dirname)
984 sys.exit(main(sys.argv))