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, util, encoding, node, error, extensions
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 # mercurial takes too much time checking this
377 if tip and tip == head.rev():
380 revs = xrange(tip, head.rev() + 1)
383 revs = [rev for rev in revs if not marks.is_marked(rev)]
388 (manifest, user, (time, tz), files, desc, extra) = repo.changelog.read(c.node())
389 rev_branch = extra['branch']
391 author = "%s %d %s" % (fixup_user(user), time, gittz(tz))
392 if 'committer' in extra:
393 user, time, tz = extra['committer'].rsplit(' ', 2)
394 committer = "%s %s %s" % (user, time, gittz(int(tz)))
398 parents = [p for p in repo.changelog.parentrevs(rev) if p >= 0]
400 if len(parents) == 0:
401 modified = c.manifest().keys()
404 modified, removed = get_filechanges(repo, c, parents[0])
411 if rev_branch != 'default':
412 extra_msg += 'branch : %s\n' % rev_branch
416 if f not in c.manifest():
418 rename = c.filectx(f).renamed()
420 renames.append((rename[0], f))
423 extra_msg += "rename : %s => %s\n" % e
425 for key, value in extra.iteritems():
426 if key in ('author', 'committer', 'encoding', 'message', 'branch', 'hg-git'):
429 extra_msg += "extra : %s : %s\n" % (key, urllib.quote(value))
432 desc += '\n--HG--\n' + extra_msg
434 if len(parents) == 0 and rev:
435 print 'reset %s/%s' % (prefix, ename)
437 modified_final = export_files(c.filectx(f) for f in modified)
439 print "commit %s/%s" % (prefix, ename)
440 print "mark :%d" % (marks.get_mark(rev))
441 print "author %s" % (author)
442 print "committer %s" % (committer)
443 print "data %d" % (len(desc))
447 print "from :%s" % (rev_to_mark(parents[0]))
449 print "merge :%s" % (rev_to_mark(parents[1]))
451 for f in modified_final:
452 print "M %s :%u %s" % f
454 print "D %s" % (fix_file_path(f))
458 if (count % 100 == 0):
459 print "progress revision %d '%s' (%d/%d)" % (rev, name, count, len(revs))
460 print "#############################################################"
462 # make sure the ref is updated
463 print "reset %s/%s" % (prefix, ename)
464 print "from :%u" % rev_to_mark(rev)
467 marks.set_tip(ename, rev)
469 def export_tag(repo, tag):
470 export_ref(repo, tag, 'tags', repo[hgref(tag)])
472 def export_bookmark(repo, bmark):
473 head = bmarks[hgref(bmark)]
474 export_ref(repo, bmark, 'bookmarks', head)
476 def export_branch(repo, branch):
477 tip = get_branch_tip(repo, branch)
479 export_ref(repo, branch, 'branches', head)
481 def export_head(repo):
483 export_ref(repo, g_head[0], 'bookmarks', g_head[1])
485 def do_capabilities(parser):
486 global prefix, dirname
490 print "refspec refs/heads/branches/*:%s/branches/*" % prefix
491 print "refspec refs/heads/*:%s/bookmarks/*" % prefix
492 print "refspec refs/tags/*:%s/tags/*" % prefix
494 path = os.path.join(dirname, 'marks-git')
496 if os.path.exists(path):
497 print "*import-marks %s" % path
498 print "*export-marks %s" % path
502 def branch_tip(repo, branch):
503 # older versions of mercurial don't have this
504 if hasattr(repo, 'branchtip'):
505 return repo.branchtip(branch)
507 return repo.branchtags()[branch]
509 def get_branch_tip(repo, branch):
512 heads = branches.get(hgref(branch), None)
516 # verify there's only one head
518 warn("Branch '%s' has more than one head, consider merging" % branch)
519 return branch_tip(repo, hgref(branch))
523 def list_head(repo, cur):
524 global g_head, bmarks
526 head = bookmarks.readcurrent(repo)
530 # fake bookmark from current branch
537 if head == 'default':
542 print "@refs/heads/%s HEAD" % head
543 g_head = (head, node)
546 global branches, bmarks, mode, track_branches
549 for bmark, node in bookmarks.listbookmarks(repo).iteritems():
550 bmarks[bmark] = repo[node]
552 cur = repo.dirstate.branch()
557 for branch in repo.branchmap():
558 heads = repo.branchheads(branch)
560 branches[branch] = heads
562 for branch in branches:
563 print "? refs/heads/branches/%s" % gitref(branch)
566 print "? refs/heads/%s" % gitref(bmark)
568 for tag, node in repo.tagslist():
571 print "? refs/tags/%s" % gitref(tag)
575 def do_import(parser):
578 path = os.path.join(dirname, 'marks-git')
581 if os.path.exists(path):
582 print "feature import-marks=%s" % path
583 print "feature export-marks=%s" % path
586 tmp = encoding.encoding
587 encoding.encoding = 'utf-8'
589 # lets get all the import lines
590 while parser.check('import'):
595 elif ref.startswith('refs/heads/branches/'):
596 branch = ref[len('refs/heads/branches/'):]
597 export_branch(repo, branch)
598 elif ref.startswith('refs/heads/'):
599 bmark = ref[len('refs/heads/'):]
600 export_bookmark(repo, bmark)
601 elif ref.startswith('refs/tags/'):
602 tag = ref[len('refs/tags/'):]
603 export_tag(repo, tag)
607 encoding.encoding = tmp
611 def parse_blob(parser):
615 mark = parser.get_mark()
617 data = parser.get_data()
618 blob_marks[mark] = data
621 def get_merge_files(repo, p1, p2, files):
622 for e in repo[p1].files():
624 if e not in repo[p1].manifest():
626 f = { 'ctx' : repo[p1][e] }
629 def parse_commit(parser):
630 global marks, blob_marks, parsed_refs
633 from_mark = merge_mark = None
638 commit_mark = parser.get_mark()
640 author = parser.get_author()
642 committer = parser.get_author()
644 data = parser.get_data()
646 if parser.check('from'):
647 from_mark = parser.get_mark()
649 if parser.check('merge'):
650 merge_mark = parser.get_mark()
652 if parser.check('merge'):
653 die('octopus merges are not supported yet')
655 # fast-export adds an extra newline
662 if parser.check('M'):
663 t, m, mark_ref, path = line.split(' ', 3)
664 mark = int(mark_ref[1:])
665 f = { 'mode' : hgmode(m), 'data' : blob_marks[mark] }
666 elif parser.check('D'):
667 t, path = line.split(' ', 1)
668 f = { 'deleted' : True }
670 die('Unknown file command: %s' % line)
673 def getfilectx(repo, memctx, f):
679 is_exec = of['mode'] == 'x'
680 is_link = of['mode'] == 'l'
681 rename = of.get('rename', None)
682 return context.memfilectx(f, of['data'],
683 is_link, is_exec, rename)
687 user, date, tz = author
690 if committer != author:
691 extra['committer'] = "%s %u %u" % committer
694 p1 = repo.changelog.node(mark_to_rev(from_mark))
699 p2 = repo.changelog.node(mark_to_rev(merge_mark))
704 # If files changed from any of the parents, hg wants to know, but in git if
705 # nothing changed from the first parent, nothing changed.
708 get_merge_files(repo, p1, p2, files)
710 # Check if the ref is supposed to be a named branch
711 if ref.startswith('refs/heads/branches/'):
712 branch = ref[len('refs/heads/branches/'):]
713 extra['branch'] = hgref(branch)
716 i = data.find('\n--HG--\n')
718 tmp = data[i + len('\n--HG--\n'):].strip()
719 for k, v in [e.split(' : ', 1) for e in tmp.split('\n')]:
721 old, new = v.split(' => ', 1)
722 files[new]['rename'] = old
726 ek, ev = v.split(' : ', 1)
727 extra[ek] = urllib.unquote(ev)
730 ctx = context.memctx(repo, (p1, p2), data,
731 files.keys(), getfilectx,
732 user, (date, tz), extra)
734 tmp = encoding.encoding
735 encoding.encoding = 'utf-8'
737 node = repo.commitctx(ctx)
739 encoding.encoding = tmp
741 rev = repo[node].rev()
743 parsed_refs[ref] = node
744 marks.new_mark(rev, commit_mark)
746 def parse_reset(parser):
752 if parser.check('commit'):
755 if not parser.check('from'):
757 from_mark = parser.get_mark()
760 node = parser.repo.changelog.node(mark_to_rev(from_mark))
761 parsed_refs[ref] = node
763 def parse_tag(parser):
766 from_mark = parser.get_mark()
768 tagger = parser.get_author()
770 data = parser.get_data()
773 parsed_tags[name] = (tagger, data)
775 def write_tag(repo, tag, node, msg, author):
776 branch = repo[node].branch()
777 tip = branch_tip(repo, branch)
780 def getfilectx(repo, memctx, f):
782 fctx = tip.filectx(f)
784 except error.ManifestLookupError:
786 content = data + "%s %s\n" % (hghex(node), tag)
787 return context.memfilectx(f, content, False, False, None)
792 author = (None, 0, 0)
793 user, date, tz = author
795 ctx = context.memctx(repo, (p1, p2), msg,
796 ['.hgtags'], getfilectx,
797 user, (date, tz), {'branch' : branch})
799 tmp = encoding.encoding
800 encoding.encoding = 'utf-8'
802 tagnode = repo.commitctx(ctx)
804 encoding.encoding = tmp
808 def do_export(parser):
809 global parsed_refs, bmarks, peer
815 for line in parser.each_block('done'):
816 if parser.check('blob'):
818 elif parser.check('commit'):
820 elif parser.check('reset'):
822 elif parser.check('tag'):
824 elif parser.check('feature'):
827 die('unhandled export command: %s' % line)
829 for ref, node in parsed_refs.iteritems():
830 if ref.startswith('refs/heads/branches'):
831 branch = ref[len('refs/heads/branches/'):]
832 if branch in branches and node in branches[branch]:
836 elif ref.startswith('refs/heads/'):
837 bmark = ref[len('refs/heads/'):]
838 p_bmarks.append((bmark, node))
840 elif ref.startswith('refs/tags/'):
841 tag = ref[len('refs/tags/'):]
843 author, msg = parsed_tags.get(tag, (None, None))
846 msg = 'Added tag %s for changeset %s' % (tag, hghex(node[:6]));
847 write_tag(parser.repo, tag, node, msg, author)
849 fp = parser.repo.opener('localtags', 'a')
850 fp.write('%s %s\n' % (hghex(node), tag))
854 # transport-helper/fast-export bugs
858 parser.repo.push(peer, force=force_push)
861 for bmark, node in p_bmarks:
862 ref = 'refs/heads/' + bmark
866 old = bmarks[bmark].hex()
873 if bmark == 'master' and 'master' not in parser.repo._bookmarks:
876 elif bookmarks.pushbookmark(parser.repo, bmark, old, new):
880 print "error %s" % ref
884 rb = peer.listkeys('bookmarks')
885 old = rb.get(bmark, '')
886 if not peer.pushkey('bookmarks', bmark, old, new):
887 print "error %s" % ref
894 def fix_path(alias, repo, orig_url):
895 url = urlparse.urlparse(orig_url, 'file')
896 if url.scheme != 'file' or os.path.isabs(url.path):
898 abs_url = urlparse.urljoin("%s/" % os.getcwd(), orig_url)
899 cmd = ['git', 'config', 'remote.%s.url' % alias, "hg::%s" % abs_url]
903 global prefix, dirname, branches, bmarks
904 global marks, blob_marks, parsed_refs
905 global peer, mode, bad_mail, bad_name
906 global track_branches, force_push, is_tmp
914 hg_git_compat = False
915 track_branches = True
919 if get_config('remote-hg.hg-git-compat') == 'true\n':
921 track_branches = False
922 if get_config('remote-hg.track-branches') == 'false\n':
923 track_branches = False
924 if get_config('remote-hg.force-push') == 'false\n':
926 except subprocess.CalledProcessError:
931 bad_mail = 'none@none'
940 alias = util.sha1(alias).hexdigest()
944 gitdir = os.environ['GIT_DIR']
945 dirname = os.path.join(gitdir, 'hg', alias)
954 repo = get_repo(url, alias)
955 prefix = 'refs/hg/%s' % alias
958 fix_path(alias, peer or repo, url)
960 if not os.path.exists(dirname):
963 marks_path = os.path.join(dirname, 'marks-hg')
964 marks = Marks(marks_path)
966 parser = Parser(repo)
968 if parser.check('capabilities'):
969 do_capabilities(parser)
970 elif parser.check('list'):
972 elif parser.check('import'):
974 elif parser.check('export'):
977 die('unhandled command: %s' % line)
986 shutil.rmtree(dirname)
989 sys.exit(main(sys.argv))