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+)')
61 sys
.stderr
.write('ERROR: %s\n' % (msg
% args
))
65 sys
.stderr
.write('WARNING: %s\n' % (msg
% args
))
68 return 'l' in flags
and '120000' or 'x' in flags
and '100755' or '100644'
71 return '%+03d%02d' % (-tz
/ 3600, -tz
% 3600 / 60)
74 m
= { '100755': 'x', '120000': 'l' }
75 return m
.get(mode
, '')
84 return ref
.replace('___', ' ')
87 return ref
.replace(' ', '___')
89 def get_config(config
):
90 cmd
= ['git', 'config', '--get', config
]
91 process
= subprocess
.Popen(cmd
, stdout
=subprocess
.PIPE
)
92 output
, _
= process
.communicate()
95 def get_config_bool(config
, default
=False):
96 value
= get_config(config
).rstrip('\n')
99 elif value
== "false":
106 def __init__(self
, path
, repo
):
112 if self
.version
< VERSION
:
113 if self
.version
== 1:
117 if self
.version
< VERSION
:
119 self
.version
= VERSION
129 if not os
.path
.exists(self
.path
):
132 tmp
= json
.load(open(self
.path
))
134 self
.tips
= tmp
['tips']
135 self
.marks
= tmp
['marks']
136 self
.last_mark
= tmp
['last-mark']
137 self
.version
= tmp
.get('version', 1)
139 for rev
, mark
in self
.marks
.iteritems():
140 self
.rev_marks
[mark
] = rev
142 def upgrade_one(self
):
144 return hghex(self
.repo
.changelog
.node(int(rev
)))
145 self
.tips
= dict((name
, get_id(rev
)) for name
, rev
in self
.tips
.iteritems())
146 self
.marks
= dict((get_id(rev
), mark
) for rev
, mark
in self
.marks
.iteritems())
147 self
.rev_marks
= dict((mark
, get_id(rev
)) for mark
, rev
in self
.rev_marks
.iteritems())
151 return { 'tips': self
.tips
, 'marks': self
.marks
, 'last-mark' : self
.last_mark
, 'version' : self
.version
}
154 json
.dump(self
.dict(), open(self
.path
, 'w'))
157 return str(self
.dict())
159 def from_rev(self
, rev
):
160 return self
.marks
[rev
]
162 def to_rev(self
, mark
):
163 return self
.rev_marks
[mark
]
167 return self
.last_mark
169 def get_mark(self
, rev
):
171 self
.marks
[rev
] = self
.last_mark
172 return self
.last_mark
174 def new_mark(self
, rev
, mark
):
175 self
.marks
[rev
] = mark
176 self
.rev_marks
[mark
] = rev
177 self
.last_mark
= mark
179 def is_marked(self
, rev
):
180 return rev
in self
.marks
182 def get_tip(self
, branch
):
183 return self
.tips
.get(branch
, None)
185 def set_tip(self
, branch
, tip
):
186 self
.tips
[branch
] = tip
190 def __init__(self
, repo
):
192 self
.line
= self
.get_line()
195 return sys
.stdin
.readline().strip()
197 def __getitem__(self
, i
):
198 return self
.line
.split()[i
]
200 def check(self
, word
):
201 return self
.line
.startswith(word
)
203 def each_block(self
, separator
):
204 while self
.line
!= separator
:
206 self
.line
= self
.get_line()
209 return self
.each_block('')
212 self
.line
= self
.get_line()
213 if self
.line
== 'done':
217 i
= self
.line
.index(':') + 1
218 return int(self
.line
[i
:])
221 if not self
.check('data'):
223 i
= self
.line
.index(' ') + 1
224 size
= int(self
.line
[i
:])
225 return sys
.stdin
.read(size
)
227 def get_author(self
):
231 m
= RAW_AUTHOR_RE
.match(self
.line
)
234 _
, name
, email
, date
, tz
= m
.groups()
235 if name
and 'ext:' in name
:
236 m
= re
.match('^(.+?) ext:\((.+)\)$', name
)
239 ex
= urllib
.unquote(m
.group(2))
241 if email
!= bad_mail
:
243 user
= '%s <%s>' % (name
, email
)
245 user
= '<%s>' % (email
)
253 tz
= ((tz
/ 100) * 3600) + ((tz
% 100) * 60)
254 return (user
, int(date
), -tz
)
256 def fix_file_path(path
):
257 if not os
.path
.isabs(path
):
259 return os
.path
.relpath(path
, '/')
261 def export_files(files
):
262 global marks
, filenodes
266 fid
= node
.hex(f
.filenode())
269 mark
= filenodes
[fid
]
271 mark
= marks
.next_mark()
272 filenodes
[fid
] = mark
276 print "mark :%u" % mark
277 print "data %d" % len(d
)
280 path
= fix_file_path(f
.path())
281 final
.append((gitmode(f
.flags()), mark
, path
))
285 def get_filechanges(repo
, ctx
, parent
):
290 # load earliest manifest first for caching reasons
291 prev
= parent
.manifest().copy()
296 if (cur
.flags(fn
) != prev
.flags(fn
) or cur
[fn
] != prev
[fn
]):
301 removed |
= set(prev
.keys())
303 return added | modified
, removed
305 def fixup_user_git(user
):
307 user
= user
.replace('"', '')
308 m
= AUTHOR_RE
.match(user
)
311 mail
= m
.group(2).strip()
313 m
= EMAIL_RE
.match(user
)
318 m
= NAME_RE
.match(user
)
320 name
= m
.group(1).strip()
323 def fixup_user_hg(user
):
325 # stole this from hg-git
326 return re
.sub('[<>\n]', '?', name
.lstrip('< ').rstrip('> '))
328 m
= AUTHOR_HG_RE
.match(user
)
330 name
= sanitize(m
.group(1))
331 mail
= sanitize(m
.group(2))
334 name
+= ' ext:(' + urllib
.quote(ex
) + ')'
336 name
= sanitize(user
)
344 def fixup_user(user
):
345 global mode
, bad_mail
348 name
, mail
= fixup_user_git(user
)
350 name
, mail
= fixup_user_hg(user
)
357 return '%s <%s>' % (name
, mail
)
359 def get_repo(url
, alias
):
363 myui
.setconfig('ui', 'interactive', 'off')
364 myui
.fout
= sys
.stderr
366 if get_config_bool('remote-hg.insecure'):
367 myui
.setconfig('web', 'cacerts', '')
369 extensions
.loadall(myui
)
372 repo
= hg
.repository(myui
, url
)
373 if not os
.path
.exists(dirname
):
376 shared_path
= os
.path
.join(gitdir
, 'hg')
377 if not os
.path
.exists(shared_path
):
379 hg
.clone(myui
, {}, url
, shared_path
, update
=False, pull
=True)
381 die('Repository error')
383 if not os
.path
.exists(dirname
):
386 local_path
= os
.path
.join(dirname
, 'clone')
387 if not os
.path
.exists(local_path
):
388 hg
.share(myui
, shared_path
, local_path
, update
=False)
390 repo
= hg
.repository(myui
, local_path
)
392 peer
= hg
.peer(myui
, {}, url
)
394 die('Repository error')
395 repo
.pull(peer
, heads
=None, force
=True)
399 def rev_to_mark(rev
):
401 return marks
.from_rev(rev
.hex())
403 def mark_to_rev(mark
):
405 return marks
.to_rev(mark
)
407 def export_ref(repo
, name
, kind
, head
):
408 global prefix
, marks
, mode
410 ename
= '%s/%s' % (kind
, name
)
411 tip
= marks
.get_tip(ename
)
412 if tip
and tip
in repo
:
413 tip
= repo
[tip
].rev()
417 revs
= xrange(tip
, head
.rev() + 1)
425 if marks
.is_marked(c
.hex()):
429 (manifest
, user
, (time
, tz
), files
, desc
, extra
) = repo
.changelog
.read(node
)
430 rev_branch
= extra
['branch']
432 author
= "%s %d %s" % (fixup_user(user
), time
, gittz(tz
))
433 if 'committer' in extra
:
434 user
, time
, tz
= extra
['committer'].rsplit(' ', 2)
435 committer
= "%s %s %s" % (user
, time
, gittz(int(tz
)))
439 parents
= [repo
[p
] for p
in repo
.changelog
.parentrevs(rev
) if p
>= 0]
441 if len(parents
) == 0:
442 modified
= c
.manifest().keys()
445 modified
, removed
= get_filechanges(repo
, c
, parents
[0])
452 if rev_branch
!= 'default':
453 extra_msg
+= 'branch : %s\n' % rev_branch
457 if f
not in c
.manifest():
459 rename
= c
.filectx(f
).renamed()
461 renames
.append((rename
[0], f
))
464 extra_msg
+= "rename : %s => %s\n" % e
466 for key
, value
in extra
.iteritems():
467 if key
in ('author', 'committer', 'encoding', 'message', 'branch', 'hg-git'):
470 extra_msg
+= "extra : %s : %s\n" % (key
, urllib
.quote(value
))
473 desc
+= '\n--HG--\n' + extra_msg
475 if len(parents
) == 0 and rev
:
476 print 'reset %s/%s' % (prefix
, ename
)
478 modified_final
= export_files(c
.filectx(f
) for f
in modified
)
480 print "commit %s/%s" % (prefix
, ename
)
481 print "mark :%d" % (marks
.get_mark(c
.hex()))
482 print "author %s" % (author
)
483 print "committer %s" % (committer
)
484 print "data %d" % (len(desc
))
488 print "from :%s" % (rev_to_mark(parents
[0]))
490 print "merge :%s" % (rev_to_mark(parents
[1]))
492 for f
in modified_final
:
493 print "M %s :%u %s" % f
495 print "D %s" % (fix_file_path(f
))
499 if (count
% 100 == 0):
500 print "progress revision %d '%s' (%d/%d)" % (rev
, name
, count
, len(revs
))
502 # make sure the ref is updated
503 print "reset %s/%s" % (prefix
, ename
)
504 print "from :%u" % rev_to_mark(head
)
507 marks
.set_tip(ename
, head
.hex())
509 def export_tag(repo
, tag
):
510 export_ref(repo
, tag
, 'tags', repo
[hgref(tag
)])
512 def export_bookmark(repo
, bmark
):
513 head
= bmarks
[hgref(bmark
)]
514 export_ref(repo
, bmark
, 'bookmarks', head
)
516 def export_branch(repo
, branch
):
517 tip
= get_branch_tip(repo
, branch
)
519 export_ref(repo
, branch
, 'branches', head
)
521 def export_head(repo
):
523 export_ref(repo
, g_head
[0], 'bookmarks', g_head
[1])
525 def do_capabilities(parser
):
526 global prefix
, dirname
530 print "refspec refs/heads/branches/*:%s/branches/*" % prefix
531 print "refspec refs/heads/*:%s/bookmarks/*" % prefix
532 print "refspec refs/tags/*:%s/tags/*" % prefix
534 path
= os
.path
.join(dirname
, 'marks-git')
536 if os
.path
.exists(path
):
537 print "*import-marks %s" % path
538 print "*export-marks %s" % path
542 def branch_tip(repo
, branch
):
543 # older versions of mercurial don't have this
544 if hasattr(repo
, 'branchtip'):
545 return repo
.branchtip(branch
)
547 return repo
.branchtags()[branch
]
549 def get_branch_tip(repo
, branch
):
552 heads
= branches
.get(hgref(branch
), None)
556 # verify there's only one head
558 warn("Branch '%s' has more than one head, consider merging" % branch
)
559 return branch_tip(repo
, hgref(branch
))
563 def list_head(repo
, cur
):
564 global g_head
, bmarks
566 head
= bookmarks
.readcurrent(repo
)
570 # fake bookmark from current branch
577 if head
== 'default':
582 print "@refs/heads/%s HEAD" % head
583 g_head
= (head
, node
)
586 global branches
, bmarks
, track_branches
589 for bmark
, node
in bookmarks
.listbookmarks(repo
).iteritems():
590 bmarks
[bmark
] = repo
[node
]
592 cur
= repo
.dirstate
.branch()
597 for branch
in repo
.branchmap():
598 heads
= repo
.branchheads(branch
)
600 branches
[branch
] = heads
602 for branch
in branches
:
603 print "? refs/heads/branches/%s" % gitref(branch
)
606 print "? refs/heads/%s" % gitref(bmark
)
608 for tag
, node
in repo
.tagslist():
611 print "? refs/tags/%s" % gitref(tag
)
615 def do_import(parser
):
618 path
= os
.path
.join(dirname
, 'marks-git')
621 if os
.path
.exists(path
):
622 print "feature import-marks=%s" % path
623 print "feature export-marks=%s" % path
624 print "feature force"
627 tmp
= encoding
.encoding
628 encoding
.encoding
= 'utf-8'
630 # lets get all the import lines
631 while parser
.check('import'):
636 elif ref
.startswith('refs/heads/branches/'):
637 branch
= ref
[len('refs/heads/branches/'):]
638 export_branch(repo
, branch
)
639 elif ref
.startswith('refs/heads/'):
640 bmark
= ref
[len('refs/heads/'):]
641 export_bookmark(repo
, bmark
)
642 elif ref
.startswith('refs/tags/'):
643 tag
= ref
[len('refs/tags/'):]
644 export_tag(repo
, tag
)
648 encoding
.encoding
= tmp
652 def parse_blob(parser
):
656 mark
= parser
.get_mark()
658 data
= parser
.get_data()
659 blob_marks
[mark
] = data
662 def get_merge_files(repo
, p1
, p2
, files
):
663 for e
in repo
[p1
].files():
665 if e
not in repo
[p1
].manifest():
667 f
= { 'ctx' : repo
[p1
][e
] }
670 def parse_commit(parser
):
671 global marks
, blob_marks
, parsed_refs
674 from_mark
= merge_mark
= None
679 commit_mark
= parser
.get_mark()
681 author
= parser
.get_author()
683 committer
= parser
.get_author()
685 data
= parser
.get_data()
687 if parser
.check('from'):
688 from_mark
= parser
.get_mark()
690 if parser
.check('merge'):
691 merge_mark
= parser
.get_mark()
693 if parser
.check('merge'):
694 die('octopus merges are not supported yet')
696 # fast-export adds an extra newline
703 if parser
.check('M'):
704 t
, m
, mark_ref
, path
= line
.split(' ', 3)
705 mark
= int(mark_ref
[1:])
706 f
= { 'mode' : hgmode(m
), 'data' : blob_marks
[mark
] }
707 elif parser
.check('D'):
708 t
, path
= line
.split(' ', 1)
709 f
= { 'deleted' : True }
711 die('Unknown file command: %s' % line
)
714 def getfilectx(repo
, memctx
, f
):
720 is_exec
= of
['mode'] == 'x'
721 is_link
= of
['mode'] == 'l'
722 rename
= of
.get('rename', None)
723 return context
.memfilectx(f
, of
['data'],
724 is_link
, is_exec
, rename
)
728 user
, date
, tz
= author
731 if committer
!= author
:
732 extra
['committer'] = "%s %u %u" % committer
735 p1
= mark_to_rev(from_mark
)
740 p2
= mark_to_rev(merge_mark
)
745 # If files changed from any of the parents, hg wants to know, but in git if
746 # nothing changed from the first parent, nothing changed.
749 get_merge_files(repo
, p1
, p2
, files
)
751 # Check if the ref is supposed to be a named branch
752 if ref
.startswith('refs/heads/branches/'):
753 branch
= ref
[len('refs/heads/branches/'):]
754 extra
['branch'] = hgref(branch
)
757 i
= data
.find('\n--HG--\n')
759 tmp
= data
[i
+ len('\n--HG--\n'):].strip()
760 for k
, v
in [e
.split(' : ', 1) for e
in tmp
.split('\n')]:
762 old
, new
= v
.split(' => ', 1)
763 files
[new
]['rename'] = old
767 ek
, ev
= v
.split(' : ', 1)
768 extra
[ek
] = urllib
.unquote(ev
)
771 ctx
= context
.memctx(repo
, (p1
, p2
), data
,
772 files
.keys(), getfilectx
,
773 user
, (date
, tz
), extra
)
775 tmp
= encoding
.encoding
776 encoding
.encoding
= 'utf-8'
778 node
= hghex(repo
.commitctx(ctx
))
780 encoding
.encoding
= tmp
782 parsed_refs
[ref
] = node
783 marks
.new_mark(node
, commit_mark
)
785 def parse_reset(parser
):
791 if parser
.check('commit'):
794 if not parser
.check('from'):
796 from_mark
= parser
.get_mark()
799 rev
= mark_to_rev(from_mark
)
800 parsed_refs
[ref
] = rev
802 def parse_tag(parser
):
805 from_mark
= parser
.get_mark()
807 tagger
= parser
.get_author()
809 data
= parser
.get_data()
812 parsed_tags
[name
] = (tagger
, data
)
814 def write_tag(repo
, tag
, node
, msg
, author
):
815 branch
= repo
[node
].branch()
816 tip
= branch_tip(repo
, branch
)
819 def getfilectx(repo
, memctx
, f
):
821 fctx
= tip
.filectx(f
)
823 except error
.ManifestLookupError
:
825 content
= data
+ "%s %s\n" % (node
, tag
)
826 return context
.memfilectx(f
, content
, False, False, None)
831 author
= (None, 0, 0)
832 user
, date
, tz
= author
834 ctx
= context
.memctx(repo
, (p1
, p2
), msg
,
835 ['.hgtags'], getfilectx
,
836 user
, (date
, tz
), {'branch' : branch
})
838 tmp
= encoding
.encoding
839 encoding
.encoding
= 'utf-8'
841 tagnode
= repo
.commitctx(ctx
)
843 encoding
.encoding
= tmp
847 def do_export(parser
):
848 global parsed_refs
, bmarks
, peer
854 for line
in parser
.each_block('done'):
855 if parser
.check('blob'):
857 elif parser
.check('commit'):
859 elif parser
.check('reset'):
861 elif parser
.check('tag'):
863 elif parser
.check('feature'):
866 die('unhandled export command: %s' % line
)
868 for ref
, node
in parsed_refs
.iteritems():
870 if ref
.startswith('refs/heads/branches'):
871 branch
= ref
[len('refs/heads/branches/'):]
872 if branch
in branches
and bnode
in branches
[branch
]:
876 elif ref
.startswith('refs/heads/'):
877 bmark
= ref
[len('refs/heads/'):]
878 p_bmarks
.append((bmark
, node
))
880 elif ref
.startswith('refs/tags/'):
881 tag
= ref
[len('refs/tags/'):]
883 author
, msg
= parsed_tags
.get(tag
, (None, None))
886 msg
= 'Added tag %s for changeset %s' % (tag
, node
[:12]);
887 write_tag(parser
.repo
, tag
, node
, msg
, author
)
889 fp
= parser
.repo
.opener('localtags', 'a')
890 fp
.write('%s %s\n' % (node
, tag
))
894 # transport-helper/fast-export bugs
898 parser
.repo
.push(peer
, force
=force_push
, newbranch
=True)
899 remote_bmarks
= peer
.listkeys('bookmarks')
902 for bmark
, node
in p_bmarks
:
903 ref
= 'refs/heads/' + bmark
907 old
= bmarks
[bmark
].hex()
914 if bmark
== 'master' and 'master' not in parser
.repo
._bookmarks
:
918 elif bookmarks
.pushbookmark(parser
.repo
, bmark
, old
, new
):
922 print "error %s" % ref
926 old
= remote_bmarks
.get(bmark
, '')
927 if not peer
.pushkey('bookmarks', bmark
, old
, new
):
928 print "error %s" % ref
935 def fix_path(alias
, repo
, orig_url
):
936 url
= urlparse
.urlparse(orig_url
, 'file')
937 if url
.scheme
!= 'file' or os
.path
.isabs(url
.path
):
939 abs_url
= urlparse
.urljoin("%s/" % os
.getcwd(), orig_url
)
940 cmd
= ['git', 'config', 'remote.%s.url' % alias
, "hg::%s" % abs_url
]
944 global prefix
, gitdir
, dirname
, branches
, bmarks
945 global marks
, blob_marks
, parsed_refs
946 global peer
, mode
, bad_mail
, bad_name
947 global track_branches
, force_push
, is_tmp
955 hg_git_compat
= get_config_bool('remote-hg.hg-git-compat')
956 track_branches
= get_config_bool('remote-hg.track-branches', True)
957 force_push
= get_config_bool('remote-hg.force-push')
961 bad_mail
= 'none@none'
970 alias
= hashlib
.sha1(alias
).hexdigest()
974 gitdir
= os
.environ
['GIT_DIR']
975 dirname
= os
.path
.join(gitdir
, 'hg', alias
)
984 repo
= get_repo(url
, alias
)
985 prefix
= 'refs/hg/%s' % alias
988 fix_path(alias
, peer
or repo
, url
)
990 marks_path
= os
.path
.join(dirname
, 'marks-hg')
991 marks
= Marks(marks_path
, repo
)
993 if sys
.platform
== 'win32':
995 msvcrt
.setmode(sys
.stdout
.fileno(), os
.O_BINARY
)
997 parser
= Parser(repo
)
999 if parser
.check('capabilities'):
1000 do_capabilities(parser
)
1001 elif parser
.check('list'):
1003 elif parser
.check('import'):
1005 elif parser
.check('export'):
1008 die('unhandled command: %s' % line
)
1017 shutil
.rmtree(dirname
)
1019 atexit
.register(bye
)
1020 sys
.exit(main(sys
.argv
))