1#!/usr/bin/env python
2#
3# Copyright (c) 2012 Felipe Contreras
4#
5
6# Inspired by Rocco Rutte's hg-fast-export
7
8# Just copy to your ~/bin, or anywhere in your $PATH.
9# Then you can clone with:
10# git clone hg::/path/to/mercurial/repo/
11#
12# For remote repositories a local clone is stored in
13# "$GIT_DIR/hg/origin/clone/.hg/".
14
15from mercurial import hg, ui, bookmarks, context, util, encoding, node, error
16
17import re
18import sys
19import os
20import json
21import shutil
22import subprocess
23import urllib
24import atexit
25
26#
27# If you want to switch to hg-git compatibility mode:
28# git config --global remote-hg.hg-git-compat true
29#
30# If you are not in hg-git-compat mode and want to disable the tracking of
31# named branches:
32# git config --global remote-hg.track-branches false
33#
34# If you don't want to force pushes (and thus risk creating new remote heads):
35# git config --global remote-hg.force-push false
36#
37# If you want the equivalent of hg's clone/pull--insecure option:
38# git config remote-hg.insecure true
39#
40# git:
41# Sensible defaults for git.
42# hg bookmarks are exported as git branches, hg branches are prefixed
43# with 'branches/', HEAD is a special case.
44#
45# hg:
46# Emulate hg-git.
47# Only hg bookmarks are exported as git branches.
48# Commits are modified to preserve hg information and allow bidirectionality.
49#
50
51NAME_RE = re.compile('^([^<>]+)')
52AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
53AUTHOR_HG_RE = re.compile('^(.*?) ?<(.*?)(?:>(.+)?)?$')
54RAW_AUTHOR_RE = re.compile('^(\w+) (?:(.+)? )?<(.*)> (\d+) ([+-]\d+)')
55
56def die(msg, *args):
57 sys.stderr.write('ERROR: %s\n' % (msg % args))
58 sys.exit(1)
59
60def warn(msg, *args):
61 sys.stderr.write('WARNING: %s\n' % (msg % args))
62
63def gitmode(flags):
64 return 'l' in flags and '120000' or 'x' in flags and '100755' or '100644'
65
66def gittz(tz):
67 return '%+03d%02d' % (-tz / 3600, -tz % 3600 / 60)
68
69def hgmode(mode):
70 m = { '100755': 'x', '120000': 'l' }
71 return m.get(mode, '')
72
73def hghex(node):
74 return hg.node.hex(node)
75
76def get_config(config):
77 cmd = ['git', 'config', '--get', config]
78 process = subprocess.Popen(cmd, stdout=subprocess.PIPE)
79 output, _ = process.communicate()
80 return output
81
82class Marks:
83
84 def __init__(self, path):
85 self.path = path
86 self.tips = {}
87 self.marks = {}
88 self.rev_marks = {}
89 self.last_mark = 0
90
91 self.load()
92
93 def load(self):
94 if not os.path.exists(self.path):
95 return
96
97 tmp = json.load(open(self.path))
98
99 self.tips = tmp['tips']
100 self.marks = tmp['marks']
101 self.last_mark = tmp['last-mark']
102
103 for rev, mark in self.marks.iteritems():
104 self.rev_marks[mark] = int(rev)
105
106 def dict(self):
107 return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
108
109 def store(self):
110 json.dump(self.dict(), open(self.path, 'w'))
111
112 def __str__(self):
113 return str(self.dict())
114
115 def from_rev(self, rev):
116 return self.marks[str(rev)]
117
118 def to_rev(self, mark):
119 return self.rev_marks[mark]
120
121 def get_mark(self, rev):
122 self.last_mark += 1
123 self.marks[str(rev)] = self.last_mark
124 return self.last_mark
125
126 def new_mark(self, rev, mark):
127 self.marks[str(rev)] = mark
128 self.rev_marks[mark] = rev
129 self.last_mark = mark
130
131 def is_marked(self, rev):
132 return str(rev) in self.marks
133
134 def get_tip(self, branch):
135 return self.tips.get(branch, 0)
136
137 def set_tip(self, branch, tip):
138 self.tips[branch] = tip
139
140class Parser:
141
142 def __init__(self, repo):
143 self.repo = repo
144 self.line = self.get_line()
145
146 def get_line(self):
147 return sys.stdin.readline().strip()
148
149 def __getitem__(self, i):
150 return self.line.split()[i]
151
152 def check(self, word):
153 return self.line.startswith(word)
154
155 def each_block(self, separator):
156 while self.line != separator:
157 yield self.line
158 self.line = self.get_line()
159
160 def __iter__(self):
161 return self.each_block('')
162
163 def next(self):
164 self.line = self.get_line()
165 if self.line == 'done':
166 self.line = None
167
168 def get_mark(self):
169 i = self.line.index(':') + 1
170 return int(self.line[i:])
171
172 def get_data(self):
173 if not self.check('data'):
174 return None
175 i = self.line.index(' ') + 1
176 size = int(self.line[i:])
177 return sys.stdin.read(size)
178
179 def get_author(self):
180 global bad_mail
181
182 ex = None
183 m = RAW_AUTHOR_RE.match(self.line)
184 if not m:
185 return None
186 _, name, email, date, tz = m.groups()
187 if name and 'ext:' in name:
188 m = re.match('^(.+?) ext:\((.+)\)$', name)
189 if m:
190 name = m.group(1)
191 ex = urllib.unquote(m.group(2))
192
193 if email != bad_mail:
194 if name:
195 user = '%s <%s>' % (name, email)
196 else:
197 user = '<%s>' % (email)
198 else:
199 user = name
200
201 if ex:
202 user += ex
203
204 tz = int(tz)
205 tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
206 return (user, int(date), -tz)
207
208def fix_file_path(path):
209 if not os.path.isabs(path):
210 return path
211 return os.path.relpath(path, '/')
212
213def export_file(fc):
214 d = fc.data()
215 path = fix_file_path(fc.path())
216 print "M %s inline %s" % (gitmode(fc.flags()), path)
217 print "data %d" % len(d)
218 print d
219
220def get_filechanges(repo, ctx, parent):
221 modified = set()
222 added = set()
223 removed = set()
224
225 cur = ctx.manifest()
226 prev = repo[parent].manifest().copy()
227
228 for fn in cur:
229 if fn in prev:
230 if (cur.flags(fn) != prev.flags(fn) or cur[fn] != prev[fn]):
231 modified.add(fn)
232 del prev[fn]
233 else:
234 added.add(fn)
235 removed |= set(prev.keys())
236
237 return added | modified, removed
238
239def fixup_user_git(user):
240 name = mail = None
241 user = user.replace('"', '')
242 m = AUTHOR_RE.match(user)
243 if m:
244 name = m.group(1)
245 mail = m.group(2).strip()
246 else:
247 m = NAME_RE.match(user)
248 if m:
249 name = m.group(1).strip()
250 return (name, mail)
251
252def fixup_user_hg(user):
253 def sanitize(name):
254 # stole this from hg-git
255 return re.sub('[<>\n]', '?', name.lstrip('< ').rstrip('> '))
256
257 m = AUTHOR_HG_RE.match(user)
258 if m:
259 name = sanitize(m.group(1))
260 mail = sanitize(m.group(2))
261 ex = m.group(3)
262 if ex:
263 name += ' ext:(' + urllib.quote(ex) + ')'
264 else:
265 name = sanitize(user)
266 if '@' in user:
267 mail = name
268 else:
269 mail = None
270
271 return (name, mail)
272
273def fixup_user(user):
274 global mode, bad_mail
275
276 if mode == 'git':
277 name, mail = fixup_user_git(user)
278 else:
279 name, mail = fixup_user_hg(user)
280
281 if not name:
282 name = bad_name
283 if not mail:
284 mail = bad_mail
285
286 return '%s <%s>' % (name, mail)
287
288def get_repo(url, alias):
289 global dirname, peer
290
291 myui = ui.ui()
292 myui.setconfig('ui', 'interactive', 'off')
293 myui.fout = sys.stderr
294
295 try:
296 if get_config('remote-hg.insecure') == 'true\n':
297 myui.setconfig('web', 'cacerts', '')
298 except subprocess.CalledProcessError:
299 pass
300
301 if hg.islocal(url):
302 repo = hg.repository(myui, url)
303 else:
304 local_path = os.path.join(dirname, 'clone')
305 if not os.path.exists(local_path):
306 try:
307 peer, dstpeer = hg.clone(myui, {}, url, local_path, update=True, pull=True)
308 except:
309 die('Repository error')
310 repo = dstpeer.local()
311 else:
312 repo = hg.repository(myui, local_path)
313 try:
314 peer = hg.peer(myui, {}, url)
315 except:
316 die('Repository error')
317 repo.pull(peer, heads=None, force=True)
318
319 return repo
320
321def rev_to_mark(rev):
322 global marks
323 return marks.from_rev(rev)
324
325def mark_to_rev(mark):
326 global marks
327 return marks.to_rev(mark)
328
329def export_ref(repo, name, kind, head):
330 global prefix, marks, mode
331
332 ename = '%s/%s' % (kind, name)
333 tip = marks.get_tip(ename)
334
335 # mercurial takes too much time checking this
336 if tip and tip == head.rev():
337 # nothing to do
338 return
339 revs = xrange(tip, head.rev() + 1)
340 count = 0
341
342 revs = [rev for rev in revs if not marks.is_marked(rev)]
343
344 for rev in revs:
345
346 c = repo[rev]
347 (manifest, user, (time, tz), files, desc, extra) = repo.changelog.read(c.node())
348 rev_branch = extra['branch']
349
350 author = "%s %d %s" % (fixup_user(user), time, gittz(tz))
351 if 'committer' in extra:
352 user, time, tz = extra['committer'].rsplit(' ', 2)
353 committer = "%s %s %s" % (user, time, gittz(int(tz)))
354 else:
355 committer = author
356
357 parents = [p for p in repo.changelog.parentrevs(rev) if p >= 0]
358
359 if len(parents) == 0:
360 modified = c.manifest().keys()
361 removed = []
362 else:
363 modified, removed = get_filechanges(repo, c, parents[0])
364
365 desc += '\n'
366
367 if mode == 'hg':
368 extra_msg = ''
369
370 if rev_branch != 'default':
371 extra_msg += 'branch : %s\n' % rev_branch
372
373 renames = []
374 for f in c.files():
375 if f not in c.manifest():
376 continue
377 rename = c.filectx(f).renamed()
378 if rename:
379 renames.append((rename[0], f))
380
381 for e in renames:
382 extra_msg += "rename : %s => %s\n" % e
383
384 for key, value in extra.iteritems():
385 if key in ('author', 'committer', 'encoding', 'message', 'branch', 'hg-git'):
386 continue
387 else:
388 extra_msg += "extra : %s : %s\n" % (key, urllib.quote(value))
389
390 if extra_msg:
391 desc += '\n--HG--\n' + extra_msg
392
393 if len(parents) == 0 and rev:
394 print 'reset %s/%s' % (prefix, ename)
395
396 print "commit %s/%s" % (prefix, ename)
397 print "mark :%d" % (marks.get_mark(rev))
398 print "author %s" % (author)
399 print "committer %s" % (committer)
400 print "data %d" % (len(desc))
401 print desc
402
403 if len(parents) > 0:
404 print "from :%s" % (rev_to_mark(parents[0]))
405 if len(parents) > 1:
406 print "merge :%s" % (rev_to_mark(parents[1]))
407
408 for f in modified:
409 export_file(c.filectx(f))
410 for f in removed:
411 print "D %s" % (fix_file_path(f))
412 print
413
414 count += 1
415 if (count % 100 == 0):
416 print "progress revision %d '%s' (%d/%d)" % (rev, name, count, len(revs))
417 print "#############################################################"
418
419 # make sure the ref is updated
420 print "reset %s/%s" % (prefix, ename)
421 print "from :%u" % rev_to_mark(rev)
422 print
423
424 marks.set_tip(ename, rev)
425
426def export_tag(repo, tag):
427 export_ref(repo, tag, 'tags', repo[tag])
428
429def export_bookmark(repo, bmark):
430 head = bmarks[bmark]
431 export_ref(repo, bmark, 'bookmarks', head)
432
433def export_branch(repo, branch):
434 tip = get_branch_tip(repo, branch)
435 head = repo[tip]
436 export_ref(repo, branch, 'branches', head)
437
438def export_head(repo):
439 global g_head
440 export_ref(repo, g_head[0], 'bookmarks', g_head[1])
441
442def do_capabilities(parser):
443 global prefix, dirname
444
445 print "import"
446 print "export"
447 print "refspec refs/heads/branches/*:%s/branches/*" % prefix
448 print "refspec refs/heads/*:%s/bookmarks/*" % prefix
449 print "refspec refs/tags/*:%s/tags/*" % prefix
450
451 path = os.path.join(dirname, 'marks-git')
452
453 if os.path.exists(path):
454 print "*import-marks %s" % path
455 print "*export-marks %s" % path
456
457 print
458
459def get_branch_tip(repo, branch):
460 global branches
461
462 heads = branches.get(branch, None)
463 if not heads:
464 return None
465
466 # verify there's only one head
467 if (len(heads) > 1):
468 warn("Branch '%s' has more than one head, consider merging" % branch)
469 # older versions of mercurial don't have this
470 if hasattr(repo, "branchtip"):
471 return repo.branchtip(branch)
472
473 return heads[0]
474
475def list_head(repo, cur):
476 global g_head, bmarks
477
478 head = bookmarks.readcurrent(repo)
479 if head:
480 node = repo[head]
481 else:
482 # fake bookmark from current branch
483 head = cur
484 node = repo['.']
485 if not node:
486 node = repo['tip']
487 if not node:
488 return
489 if head == 'default':
490 head = 'master'
491 bmarks[head] = node
492
493 print "@refs/heads/%s HEAD" % head
494 g_head = (head, node)
495
496def do_list(parser):
497 global branches, bmarks, mode, track_branches
498
499 repo = parser.repo
500 for bmark, node in bookmarks.listbookmarks(repo).iteritems():
501 bmarks[bmark] = repo[node]
502
503 cur = repo.dirstate.branch()
504
505 list_head(repo, cur)
506
507 if track_branches:
508 for branch in repo.branchmap():
509 heads = repo.branchheads(branch)
510 if len(heads):
511 branches[branch] = heads
512
513 for branch in branches:
514 print "? refs/heads/branches/%s" % branch
515
516 for bmark in bmarks:
517 print "? refs/heads/%s" % bmark
518
519 for tag, node in repo.tagslist():
520 if tag == 'tip':
521 continue
522 print "? refs/tags/%s" % tag
523
524 print
525
526def do_import(parser):
527 repo = parser.repo
528
529 path = os.path.join(dirname, 'marks-git')
530
531 print "feature done"
532 if os.path.exists(path):
533 print "feature import-marks=%s" % path
534 print "feature export-marks=%s" % path
535 sys.stdout.flush()
536
537 tmp = encoding.encoding
538 encoding.encoding = 'utf-8'
539
540 # lets get all the import lines
541 while parser.check('import'):
542 ref = parser[1]
543
544 if (ref == 'HEAD'):
545 export_head(repo)
546 elif ref.startswith('refs/heads/branches/'):
547 branch = ref[len('refs/heads/branches/'):]
548 export_branch(repo, branch)
549 elif ref.startswith('refs/heads/'):
550 bmark = ref[len('refs/heads/'):]
551 export_bookmark(repo, bmark)
552 elif ref.startswith('refs/tags/'):
553 tag = ref[len('refs/tags/'):]
554 export_tag(repo, tag)
555
556 parser.next()
557
558 encoding.encoding = tmp
559
560 print 'done'
561
562def parse_blob(parser):
563 global blob_marks
564
565 parser.next()
566 mark = parser.get_mark()
567 parser.next()
568 data = parser.get_data()
569 blob_marks[mark] = data
570 parser.next()
571
572def get_merge_files(repo, p1, p2, files):
573 for e in repo[p1].files():
574 if e not in files:
575 if e not in repo[p1].manifest():
576 continue
577 f = { 'ctx' : repo[p1][e] }
578 files[e] = f
579
580def parse_commit(parser):
581 global marks, blob_marks, parsed_refs
582 global mode
583
584 from_mark = merge_mark = None
585
586 ref = parser[1]
587 parser.next()
588
589 commit_mark = parser.get_mark()
590 parser.next()
591 author = parser.get_author()
592 parser.next()
593 committer = parser.get_author()
594 parser.next()
595 data = parser.get_data()
596 parser.next()
597 if parser.check('from'):
598 from_mark = parser.get_mark()
599 parser.next()
600 if parser.check('merge'):
601 merge_mark = parser.get_mark()
602 parser.next()
603 if parser.check('merge'):
604 die('octopus merges are not supported yet')
605
606 files = {}
607
608 for line in parser:
609 if parser.check('M'):
610 t, m, mark_ref, path = line.split(' ', 3)
611 mark = int(mark_ref[1:])
612 f = { 'mode' : hgmode(m), 'data' : blob_marks[mark] }
613 elif parser.check('D'):
614 t, path = line.split(' ', 1)
615 f = { 'deleted' : True }
616 else:
617 die('Unknown file command: %s' % line)
618 files[path] = f
619
620 def getfilectx(repo, memctx, f):
621 of = files[f]
622 if 'deleted' in of:
623 raise IOError
624 if 'ctx' in of:
625 return of['ctx']
626 is_exec = of['mode'] == 'x'
627 is_link = of['mode'] == 'l'
628 rename = of.get('rename', None)
629 return context.memfilectx(f, of['data'],
630 is_link, is_exec, rename)
631
632 repo = parser.repo
633
634 user, date, tz = author
635 extra = {}
636
637 if committer != author:
638 extra['committer'] = "%s %u %u" % committer
639
640 if from_mark:
641 p1 = repo.changelog.node(mark_to_rev(from_mark))
642 else:
643 p1 = '\0' * 20
644
645 if merge_mark:
646 p2 = repo.changelog.node(mark_to_rev(merge_mark))
647 else:
648 p2 = '\0' * 20
649
650 #
651 # If files changed from any of the parents, hg wants to know, but in git if
652 # nothing changed from the first parent, nothing changed.
653 #
654 if merge_mark:
655 get_merge_files(repo, p1, p2, files)
656
657 # Check if the ref is supposed to be a named branch
658 if ref.startswith('refs/heads/branches/'):
659 extra['branch'] = ref[len('refs/heads/branches/'):]
660
661 if mode == 'hg':
662 i = data.find('\n--HG--\n')
663 if i >= 0:
664 tmp = data[i + len('\n--HG--\n'):].strip()
665 for k, v in [e.split(' : ', 1) for e in tmp.split('\n')]:
666 if k == 'rename':
667 old, new = v.split(' => ', 1)
668 files[new]['rename'] = old
669 elif k == 'branch':
670 extra[k] = v
671 elif k == 'extra':
672 ek, ev = v.split(' : ', 1)
673 extra[ek] = urllib.unquote(ev)
674 data = data[:i]
675
676 ctx = context.memctx(repo, (p1, p2), data,
677 files.keys(), getfilectx,
678 user, (date, tz), extra)
679
680 tmp = encoding.encoding
681 encoding.encoding = 'utf-8'
682
683 node = repo.commitctx(ctx)
684
685 encoding.encoding = tmp
686
687 rev = repo[node].rev()
688
689 parsed_refs[ref] = node
690 marks.new_mark(rev, commit_mark)
691
692def parse_reset(parser):
693 global parsed_refs
694
695 ref = parser[1]
696 parser.next()
697 # ugh
698 if parser.check('commit'):
699 parse_commit(parser)
700 return
701 if not parser.check('from'):
702 return
703 from_mark = parser.get_mark()
704 parser.next()
705
706 node = parser.repo.changelog.node(mark_to_rev(from_mark))
707 parsed_refs[ref] = node
708
709def parse_tag(parser):
710 name = parser[1]
711 parser.next()
712 from_mark = parser.get_mark()
713 parser.next()
714 tagger = parser.get_author()
715 parser.next()
716 data = parser.get_data()
717 parser.next()
718
719 # nothing to do
720
721def do_export(parser):
722 global parsed_refs, bmarks, peer
723
724 p_bmarks = []
725
726 parser.next()
727
728 for line in parser.each_block('done'):
729 if parser.check('blob'):
730 parse_blob(parser)
731 elif parser.check('commit'):
732 parse_commit(parser)
733 elif parser.check('reset'):
734 parse_reset(parser)
735 elif parser.check('tag'):
736 parse_tag(parser)
737 elif parser.check('feature'):
738 pass
739 else:
740 die('unhandled export command: %s' % line)
741
742 for ref, node in parsed_refs.iteritems():
743 if ref.startswith('refs/heads/branches'):
744 print "ok %s" % ref
745 elif ref.startswith('refs/heads/'):
746 bmark = ref[len('refs/heads/'):]
747 p_bmarks.append((bmark, node))
748 continue
749 elif ref.startswith('refs/tags/'):
750 tag = ref[len('refs/tags/'):]
751 if mode == 'git':
752 msg = 'Added tag %s for changeset %s' % (tag, hghex(node[:6]));
753 parser.repo.tag([tag], node, msg, False, None, {})
754 else:
755 parser.repo.tag([tag], node, None, True, None, {})
756 print "ok %s" % ref
757 else:
758 # transport-helper/fast-export bugs
759 continue
760
761 if peer:
762 parser.repo.push(peer, force=force_push)
763
764 # handle bookmarks
765 for bmark, node in p_bmarks:
766 ref = 'refs/heads/' + bmark
767 new = hghex(node)
768
769 if bmark in bmarks:
770 old = bmarks[bmark].hex()
771 else:
772 old = ''
773
774 if bmark == 'master' and 'master' not in parser.repo._bookmarks:
775 # fake bookmark
776 pass
777 elif bookmarks.pushbookmark(parser.repo, bmark, old, new):
778 # updated locally
779 pass
780 else:
781 print "error %s" % ref
782 continue
783
784 if peer:
785 if not peer.pushkey('bookmarks', bmark, old, new):
786 print "error %s" % ref
787 continue
788
789 print "ok %s" % ref
790
791 print
792
793def fix_path(alias, repo, orig_url):
794 repo_url = util.url(repo.url())
795 url = util.url(orig_url)
796 if str(url) == str(repo_url):
797 return
798 cmd = ['git', 'config', 'remote.%s.url' % alias, "hg::%s" % repo_url]
799 subprocess.call(cmd)
800
801def main(args):
802 global prefix, dirname, branches, bmarks
803 global marks, blob_marks, parsed_refs
804 global peer, mode, bad_mail, bad_name
805 global track_branches, force_push, is_tmp
806
807 alias = args[1]
808 url = args[2]
809 peer = None
810
811 hg_git_compat = False
812 track_branches = True
813 force_push = True
814
815 try:
816 if get_config('remote-hg.hg-git-compat') == 'true\n':
817 hg_git_compat = True
818 track_branches = False
819 if get_config('remote-hg.track-branches') == 'false\n':
820 track_branches = False
821 if get_config('remote-hg.force-push') == 'false\n':
822 force_push = False
823 except subprocess.CalledProcessError:
824 pass
825
826 if hg_git_compat:
827 mode = 'hg'
828 bad_mail = 'none@none'
829 bad_name = ''
830 else:
831 mode = 'git'
832 bad_mail = 'unknown'
833 bad_name = 'Unknown'
834
835 if alias[4:] == url:
836 is_tmp = True
837 alias = util.sha1(alias).hexdigest()
838 else:
839 is_tmp = False
840
841 gitdir = os.environ['GIT_DIR']
842 dirname = os.path.join(gitdir, 'hg', alias)
843 branches = {}
844 bmarks = {}
845 blob_marks = {}
846 parsed_refs = {}
847 marks = None
848
849 repo = get_repo(url, alias)
850 prefix = 'refs/hg/%s' % alias
851
852 if not is_tmp:
853 fix_path(alias, peer or repo, url)
854
855 if not os.path.exists(dirname):
856 os.makedirs(dirname)
857
858 marks_path = os.path.join(dirname, 'marks-hg')
859 marks = Marks(marks_path)
860
861 parser = Parser(repo)
862 for line in parser:
863 if parser.check('capabilities'):
864 do_capabilities(parser)
865 elif parser.check('list'):
866 do_list(parser)
867 elif parser.check('import'):
868 do_import(parser)
869 elif parser.check('export'):
870 do_export(parser)
871 else:
872 die('unhandled command: %s' % line)
873 sys.stdout.flush()
874
875def bye():
876 if not marks:
877 return
878 if not is_tmp:
879 marks.store()
880 else:
881 shutil.rmtree(dirname)
882
883atexit.register(bye)
884sys.exit(main(sys.argv))