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 $PATH.
9 # Then you can clone with:
10 # git clone hg::/path/to/mercurial/repo/
12 from mercurial import hg, ui, bookmarks, context, util, encoding
23 # If you want to switch to hg-git compatibility mode:
24 # git config --global remote-hg.hg-git-compat true
26 # If you are not in hg-git-compat mode and want to disable the tracking of
28 # git config --global remote-hg.track-branches false
31 # Sensible defaults for git.
32 # hg bookmarks are exported as git branches, hg branches are prefixed
33 # with 'branches/', HEAD is a special case.
37 # Only hg bookmarks are exported as git branches.
38 # Commits are modified to preserve hg information and allow bidirectionality.
41 NAME_RE = re.compile('^([^<>]+)')
42 AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
43 AUTHOR_HG_RE = re.compile('^(.*?) ?<(.*?)(?:>(.+)?)?$')
44 RAW_AUTHOR_RE = re.compile('^(\w+) (?:(.+)? )?<(.*)> (\d+) ([+-]\d+)')
47 sys.stderr.write('ERROR: %s\n' % (msg % args))
51 sys.stderr.write('WARNING: %s\n' % (msg % args))
54 return 'l' in flags and '120000' or 'x' in flags and '100755' or '100644'
57 return '%+03d%02d' % (-tz / 3600, -tz % 3600 / 60)
60 m = { '100755': 'x', '120000': 'l' }
61 return m.get(mode, '')
63 def get_config(config):
64 cmd = ['git', 'config', '--get', config]
65 process = subprocess.Popen(cmd, stdout=subprocess.PIPE)
66 output, _ = process.communicate()
71 def __init__(self, path):
81 if not os.path.exists(self.path):
84 tmp = json.load(open(self.path))
86 self.tips = tmp['tips']
87 self.marks = tmp['marks']
88 self.last_mark = tmp['last-mark']
90 for rev, mark in self.marks.iteritems():
91 self.rev_marks[mark] = int(rev)
94 return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
97 json.dump(self.dict(), open(self.path, 'w'))
100 return str(self.dict())
102 def from_rev(self, rev):
103 return self.marks[str(rev)]
105 def to_rev(self, mark):
106 return self.rev_marks[mark]
108 def get_mark(self, rev):
110 self.marks[str(rev)] = self.last_mark
111 return self.last_mark
113 def new_mark(self, rev, mark):
114 self.marks[str(rev)] = mark
115 self.rev_marks[mark] = rev
116 self.last_mark = mark
118 def is_marked(self, rev):
119 return self.marks.has_key(str(rev))
121 def get_tip(self, branch):
122 return self.tips.get(branch, 0)
124 def set_tip(self, branch, tip):
125 self.tips[branch] = tip
129 def __init__(self, repo):
131 self.line = self.get_line()
134 return sys.stdin.readline().strip()
136 def __getitem__(self, i):
137 return self.line.split()[i]
139 def check(self, word):
140 return self.line.startswith(word)
142 def each_block(self, separator):
143 while self.line != separator:
145 self.line = self.get_line()
148 return self.each_block('')
151 self.line = self.get_line()
152 if self.line == 'done':
156 i = self.line.index(':') + 1
157 return int(self.line[i:])
160 if not self.check('data'):
162 i = self.line.index(' ') + 1
163 size = int(self.line[i:])
164 return sys.stdin.read(size)
166 def get_author(self):
170 m = RAW_AUTHOR_RE.match(self.line)
173 _, name, email, date, tz = m.groups()
174 if name and 'ext:' in name:
175 m = re.match('^(.+?) ext:\((.+)\)$', name)
178 ex = urllib.unquote(m.group(2))
180 if email != bad_mail:
182 user = '%s <%s>' % (name, email)
184 user = '<%s>' % (email)
192 tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
193 return (user, int(date), -tz)
197 print "M %s inline %s" % (gitmode(fc.flags()), fc.path())
198 print "data %d" % len(d)
201 def get_filechanges(repo, ctx, parent):
207 prev = repo[parent].manifest().copy()
211 if (cur.flags(fn) != prev.flags(fn) or cur[fn] != prev[fn]):
216 removed |= set(prev.keys())
218 return added | modified, removed
220 def fixup_user_git(user):
222 user = user.replace('"', '')
223 m = AUTHOR_RE.match(user)
226 mail = m.group(2).strip()
228 m = NAME_RE.match(user)
230 name = m.group(1).strip()
233 def fixup_user_hg(user):
235 # stole this from hg-git
236 return re.sub('[<>\n]', '?', name.lstrip('< ').rstrip('> '))
238 m = AUTHOR_HG_RE.match(user)
240 name = sanitize(m.group(1))
241 mail = sanitize(m.group(2))
244 name += ' ext:(' + urllib.quote(ex) + ')'
246 name = sanitize(user)
254 def fixup_user(user):
255 global mode, bad_mail
258 name, mail = fixup_user_git(user)
260 name, mail = fixup_user_hg(user)
267 return '%s <%s>' % (name, mail)
269 def get_repo(url, alias):
273 myui.setconfig('ui', 'interactive', 'off')
276 repo = hg.repository(myui, url)
278 local_path = os.path.join(dirname, 'clone')
279 if not os.path.exists(local_path):
280 peer, dstpeer = hg.clone(myui, {}, url, local_path, update=False, pull=True)
281 repo = dstpeer.local()
283 repo = hg.repository(myui, local_path)
284 peer = hg.peer(myui, {}, url)
285 repo.pull(peer, heads=None, force=True)
289 def rev_to_mark(rev):
291 return marks.from_rev(rev)
293 def mark_to_rev(mark):
295 return marks.to_rev(mark)
297 def export_ref(repo, name, kind, head):
298 global prefix, marks, mode
300 ename = '%s/%s' % (kind, name)
301 tip = marks.get_tip(ename)
303 # mercurial takes too much time checking this
304 if tip and tip == head.rev():
307 revs = xrange(tip, head.rev() + 1)
310 revs = [rev for rev in revs if not marks.is_marked(rev)]
315 (manifest, user, (time, tz), files, desc, extra) = repo.changelog.read(c.node())
316 rev_branch = extra['branch']
318 author = "%s %d %s" % (fixup_user(user), time, gittz(tz))
319 if 'committer' in extra:
320 user, time, tz = extra['committer'].rsplit(' ', 2)
321 committer = "%s %s %s" % (user, time, gittz(int(tz)))
325 parents = [p for p in repo.changelog.parentrevs(rev) if p >= 0]
327 if len(parents) == 0:
328 modified = c.manifest().keys()
331 modified, removed = get_filechanges(repo, c, parents[0])
336 if rev_branch != 'default':
337 extra_msg += 'branch : %s\n' % rev_branch
341 if f not in c.manifest():
343 rename = c.filectx(f).renamed()
345 renames.append((rename[0], f))
348 extra_msg += "rename : %s => %s\n" % e
350 for key, value in extra.iteritems():
351 if key in ('author', 'committer', 'encoding', 'message', 'branch', 'hg-git'):
354 extra_msg += "extra : %s : %s\n" % (key, urllib.quote(value))
358 desc += '\n--HG--\n' + extra_msg
360 if len(parents) == 0 and rev:
361 print 'reset %s/%s' % (prefix, ename)
363 print "commit %s/%s" % (prefix, ename)
364 print "mark :%d" % (marks.get_mark(rev))
365 print "author %s" % (author)
366 print "committer %s" % (committer)
367 print "data %d" % (len(desc))
371 print "from :%s" % (rev_to_mark(parents[0]))
373 print "merge :%s" % (rev_to_mark(parents[1]))
376 export_file(c.filectx(f))
382 if (count % 100 == 0):
383 print "progress revision %d '%s' (%d/%d)" % (rev, name, count, len(revs))
384 print "#############################################################"
386 # make sure the ref is updated
387 print "reset %s/%s" % (prefix, ename)
388 print "from :%u" % rev_to_mark(rev)
391 marks.set_tip(ename, rev)
393 def export_tag(repo, tag):
394 export_ref(repo, tag, 'tags', repo[tag])
396 def export_bookmark(repo, bmark):
398 export_ref(repo, bmark, 'bookmarks', head)
400 def export_branch(repo, branch):
401 tip = get_branch_tip(repo, branch)
403 export_ref(repo, branch, 'branches', head)
405 def export_head(repo):
407 export_ref(repo, g_head[0], 'bookmarks', g_head[1])
409 def do_capabilities(parser):
410 global prefix, dirname
414 print "refspec refs/heads/branches/*:%s/branches/*" % prefix
415 print "refspec refs/heads/*:%s/bookmarks/*" % prefix
416 print "refspec refs/tags/*:%s/tags/*" % prefix
418 path = os.path.join(dirname, 'marks-git')
420 if os.path.exists(path):
421 print "*import-marks %s" % path
422 print "*export-marks %s" % path
426 def get_branch_tip(repo, branch):
429 heads = branches.get(branch, None)
433 # verify there's only one head
435 warn("Branch '%s' has more than one head, consider merging" % branch)
436 # older versions of mercurial don't have this
437 if hasattr(repo, "branchtip"):
438 return repo.branchtip(branch)
442 def list_head(repo, cur):
443 global g_head, bmarks
445 head = bookmarks.readcurrent(repo)
449 # fake bookmark from current branch
456 if head == 'default':
460 print "@refs/heads/%s HEAD" % head
461 g_head = (head, node)
464 global branches, bmarks, mode, track_branches
467 for bmark, node in bookmarks.listbookmarks(repo).iteritems():
468 bmarks[bmark] = repo[node]
470 cur = repo.dirstate.branch()
475 for branch in repo.branchmap():
476 heads = repo.branchheads(branch)
478 branches[branch] = heads
480 for branch in branches:
481 print "? refs/heads/branches/%s" % branch
484 print "? refs/heads/%s" % bmark
486 for tag, node in repo.tagslist():
489 print "? refs/tags/%s" % tag
493 def do_import(parser):
496 path = os.path.join(dirname, 'marks-git')
499 if os.path.exists(path):
500 print "feature import-marks=%s" % path
501 print "feature export-marks=%s" % path
504 tmp = encoding.encoding
505 encoding.encoding = 'utf-8'
507 # lets get all the import lines
508 while parser.check('import'):
513 elif ref.startswith('refs/heads/branches/'):
514 branch = ref[len('refs/heads/branches/'):]
515 export_branch(repo, branch)
516 elif ref.startswith('refs/heads/'):
517 bmark = ref[len('refs/heads/'):]
518 export_bookmark(repo, bmark)
519 elif ref.startswith('refs/tags/'):
520 tag = ref[len('refs/tags/'):]
521 export_tag(repo, tag)
525 encoding.encoding = tmp
529 def parse_blob(parser):
533 mark = parser.get_mark()
535 data = parser.get_data()
536 blob_marks[mark] = data
539 def get_merge_files(repo, p1, p2, files):
540 for e in repo[p1].files():
542 if e not in repo[p1].manifest():
544 f = { 'ctx' : repo[p1][e] }
547 def parse_commit(parser):
548 global marks, blob_marks, parsed_refs
551 from_mark = merge_mark = None
556 commit_mark = parser.get_mark()
558 author = parser.get_author()
560 committer = parser.get_author()
562 data = parser.get_data()
564 if parser.check('from'):
565 from_mark = parser.get_mark()
567 if parser.check('merge'):
568 merge_mark = parser.get_mark()
570 if parser.check('merge'):
571 die('octopus merges are not supported yet')
576 if parser.check('M'):
577 t, m, mark_ref, path = line.split(' ', 3)
578 mark = int(mark_ref[1:])
579 f = { 'mode' : hgmode(m), 'data' : blob_marks[mark] }
580 elif parser.check('D'):
581 t, path = line.split(' ')
582 f = { 'deleted' : True }
584 die('Unknown file command: %s' % line)
587 def getfilectx(repo, memctx, f):
593 is_exec = of['mode'] == 'x'
594 is_link = of['mode'] == 'l'
595 rename = of.get('rename', None)
596 return context.memfilectx(f, of['data'],
597 is_link, is_exec, rename)
601 user, date, tz = author
604 if committer != author:
605 extra['committer'] = "%s %u %u" % committer
608 p1 = repo.changelog.node(mark_to_rev(from_mark))
613 p2 = repo.changelog.node(mark_to_rev(merge_mark))
618 # If files changed from any of the parents, hg wants to know, but in git if
619 # nothing changed from the first parent, nothing changed.
622 get_merge_files(repo, p1, p2, files)
625 i = data.find('\n--HG--\n')
627 tmp = data[i + len('\n--HG--\n'):].strip()
628 for k, v in [e.split(' : ') for e in tmp.split('\n')]:
630 old, new = v.split(' => ', 1)
631 files[new]['rename'] = old
635 ek, ev = v.split(' : ', 1)
636 extra[ek] = urllib.unquote(ev)
639 ctx = context.memctx(repo, (p1, p2), data,
640 files.keys(), getfilectx,
641 user, (date, tz), extra)
643 tmp = encoding.encoding
644 encoding.encoding = 'utf-8'
646 node = repo.commitctx(ctx)
648 encoding.encoding = tmp
650 rev = repo[node].rev()
652 parsed_refs[ref] = node
653 marks.new_mark(rev, commit_mark)
655 def parse_reset(parser):
661 if parser.check('commit'):
664 if not parser.check('from'):
666 from_mark = parser.get_mark()
669 node = parser.repo.changelog.node(mark_to_rev(from_mark))
670 parsed_refs[ref] = node
672 def parse_tag(parser):
675 from_mark = parser.get_mark()
677 tagger = parser.get_author()
679 data = parser.get_data()
684 def do_export(parser):
685 global parsed_refs, bmarks, peer
689 for line in parser.each_block('done'):
690 if parser.check('blob'):
692 elif parser.check('commit'):
694 elif parser.check('reset'):
696 elif parser.check('tag'):
698 elif parser.check('feature'):
701 die('unhandled export command: %s' % line)
703 for ref, node in parsed_refs.iteritems():
704 if ref.startswith('refs/heads/branches'):
706 elif ref.startswith('refs/heads/'):
707 bmark = ref[len('refs/heads/'):]
709 old = bmarks[bmark].hex()
712 if not bookmarks.pushbookmark(parser.repo, bmark, old, node):
713 print "error %s" % ref
715 elif ref.startswith('refs/tags/'):
716 tag = ref[len('refs/tags/'):]
717 parser.repo.tag([tag], node, None, True, None, {})
719 # transport-helper/fast-export bugs
724 parser.repo.push(peer, force=False)
728 def fix_path(alias, repo, orig_url):
729 repo_url = util.url(repo.url())
730 url = util.url(orig_url)
731 if str(url) == str(repo_url):
733 cmd = ['git', 'config', 'remote.%s.url' % alias, "hg::%s" % repo_url]
737 global prefix, dirname, branches, bmarks
738 global marks, blob_marks, parsed_refs
739 global peer, mode, bad_mail, bad_name
740 global track_branches
746 hg_git_compat = False
747 track_branches = True
749 if get_config('remote-hg.hg-git-compat') == 'true\n':
751 track_branches = False
752 if get_config('remote-hg.track-branches') == 'false\n':
753 track_branches = False
754 except subprocess.CalledProcessError:
759 bad_mail = 'none@none'
768 alias = util.sha1(alias).hexdigest()
772 gitdir = os.environ['GIT_DIR']
773 dirname = os.path.join(gitdir, 'hg', alias)
779 repo = get_repo(url, alias)
780 prefix = 'refs/hg/%s' % alias
783 fix_path(alias, peer or repo, url)
785 if not os.path.exists(dirname):
788 marks_path = os.path.join(dirname, 'marks-hg')
789 marks = Marks(marks_path)
791 parser = Parser(repo)
793 if parser.check('capabilities'):
794 do_capabilities(parser)
795 elif parser.check('list'):
797 elif parser.check('import'):
799 elif parser.check('export'):
802 die('unhandled command: %s' % line)
808 shutil.rmtree(dirname)
810 sys.exit(main(sys.argv))