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
)
78 def get_config(config
):
79 cmd
= ['git', 'config', '--get', config
]
80 process
= subprocess
.Popen(cmd
, stdout
=subprocess
.PIPE
)
81 output
, _
= process
.communicate()
86 def __init__(self
, path
):
96 if not os
.path
.exists(self
.path
):
99 tmp
= json
.load(open(self
.path
))
101 self
.tips
= tmp
['tips']
102 self
.marks
= tmp
['marks']
103 self
.last_mark
= tmp
['last-mark']
105 for rev
, mark
in self
.marks
.iteritems():
106 self
.rev_marks
[mark
] = int(rev
)
109 return { 'tips': self
.tips
, 'marks': self
.marks
, 'last-mark' : self
.last_mark
}
112 json
.dump(self
.dict(), open(self
.path
, 'w'))
115 return str(self
.dict())
117 def from_rev(self
, rev
):
118 return self
.marks
[str(rev
)]
120 def to_rev(self
, mark
):
121 return self
.rev_marks
[mark
]
123 def get_mark(self
, rev
):
125 self
.marks
[str(rev
)] = self
.last_mark
126 return self
.last_mark
128 def new_mark(self
, rev
, mark
):
129 self
.marks
[str(rev
)] = mark
130 self
.rev_marks
[mark
] = rev
131 self
.last_mark
= mark
133 def is_marked(self
, rev
):
134 return str(rev
) in self
.marks
136 def get_tip(self
, branch
):
137 return self
.tips
.get(branch
, 0)
139 def set_tip(self
, branch
, tip
):
140 self
.tips
[branch
] = tip
144 def __init__(self
, repo
):
146 self
.line
= self
.get_line()
149 return sys
.stdin
.readline().strip()
151 def __getitem__(self
, i
):
152 return self
.line
.split()[i
]
154 def check(self
, word
):
155 return self
.line
.startswith(word
)
157 def each_block(self
, separator
):
158 while self
.line
!= separator
:
160 self
.line
= self
.get_line()
163 return self
.each_block('')
166 self
.line
= self
.get_line()
167 if self
.line
== 'done':
171 i
= self
.line
.index(':') + 1
172 return int(self
.line
[i
:])
175 if not self
.check('data'):
177 i
= self
.line
.index(' ') + 1
178 size
= int(self
.line
[i
:])
179 return sys
.stdin
.read(size
)
181 def get_author(self
):
185 m
= RAW_AUTHOR_RE
.match(self
.line
)
188 _
, name
, email
, date
, tz
= m
.groups()
189 if name
and 'ext:' in name
:
190 m
= re
.match('^(.+?) ext:\((.+)\)$', name
)
193 ex
= urllib
.unquote(m
.group(2))
195 if email
!= bad_mail
:
197 user
= '%s <%s>' % (name
, email
)
199 user
= '<%s>' % (email
)
207 tz
= ((tz
/ 100) * 3600) + ((tz
% 100) * 60)
208 return (user
, int(date
), -tz
)
210 def fix_file_path(path
):
211 if not os
.path
.isabs(path
):
213 return os
.path
.relpath(path
, '/')
217 path
= fix_file_path(fc
.path())
218 print "M %s inline %s" % (gitmode(fc
.flags()), path
)
219 print "data %d" % len(d
)
222 def get_filechanges(repo
, ctx
, parent
):
228 prev
= repo
[parent
].manifest().copy()
232 if (cur
.flags(fn
) != prev
.flags(fn
) or cur
[fn
] != prev
[fn
]):
237 removed |
= set(prev
.keys())
239 return added | modified
, removed
241 def fixup_user_git(user
):
243 user
= user
.replace('"', '')
244 m
= AUTHOR_RE
.match(user
)
247 mail
= m
.group(2).strip()
249 m
= EMAIL_RE
.match(user
)
254 m
= NAME_RE
.match(user
)
256 name
= m
.group(1).strip()
259 def fixup_user_hg(user
):
261 # stole this from hg-git
262 return re
.sub('[<>\n]', '?', name
.lstrip('< ').rstrip('> '))
264 m
= AUTHOR_HG_RE
.match(user
)
266 name
= sanitize(m
.group(1))
267 mail
= sanitize(m
.group(2))
270 name
+= ' ext:(' + urllib
.quote(ex
) + ')'
272 name
= sanitize(user
)
280 def fixup_user(user
):
281 global mode
, bad_mail
284 name
, mail
= fixup_user_git(user
)
286 name
, mail
= fixup_user_hg(user
)
293 return '%s <%s>' % (name
, mail
)
295 def get_repo(url
, alias
):
299 myui
.setconfig('ui', 'interactive', 'off')
300 myui
.fout
= sys
.stderr
303 if get_config('remote-hg.insecure') == 'true\n':
304 myui
.setconfig('web', 'cacerts', '')
305 except subprocess
.CalledProcessError
:
309 mod
= extensions
.load(myui
, 'hgext.schemes', None)
315 repo
= hg
.repository(myui
, url
)
317 local_path
= os
.path
.join(dirname
, 'clone')
318 if not os
.path
.exists(local_path
):
320 peer
, dstpeer
= hg
.clone(myui
, {}, url
, local_path
, update
=True, pull
=True)
322 die('Repository error')
323 repo
= dstpeer
.local()
325 repo
= hg
.repository(myui
, local_path
)
327 peer
= hg
.peer(myui
, {}, url
)
329 die('Repository error')
330 repo
.pull(peer
, heads
=None, force
=True)
334 def rev_to_mark(rev
):
336 return marks
.from_rev(rev
)
338 def mark_to_rev(mark
):
340 return marks
.to_rev(mark
)
342 def export_ref(repo
, name
, kind
, head
):
343 global prefix
, marks
, mode
345 ename
= '%s/%s' % (kind
, name
)
346 tip
= marks
.get_tip(ename
)
348 # mercurial takes too much time checking this
349 if tip
and tip
== head
.rev():
352 revs
= xrange(tip
, head
.rev() + 1)
355 revs
= [rev
for rev
in revs
if not marks
.is_marked(rev
)]
360 (manifest
, user
, (time
, tz
), files
, desc
, extra
) = repo
.changelog
.read(c
.node())
361 rev_branch
= extra
['branch']
363 author
= "%s %d %s" % (fixup_user(user
), time
, gittz(tz
))
364 if 'committer' in extra
:
365 user
, time
, tz
= extra
['committer'].rsplit(' ', 2)
366 committer
= "%s %s %s" % (user
, time
, gittz(int(tz
)))
370 parents
= [p
for p
in repo
.changelog
.parentrevs(rev
) if p
>= 0]
372 if len(parents
) == 0:
373 modified
= c
.manifest().keys()
376 modified
, removed
= get_filechanges(repo
, c
, parents
[0])
383 if rev_branch
!= 'default':
384 extra_msg
+= 'branch : %s\n' % rev_branch
388 if f
not in c
.manifest():
390 rename
= c
.filectx(f
).renamed()
392 renames
.append((rename
[0], f
))
395 extra_msg
+= "rename : %s => %s\n" % e
397 for key
, value
in extra
.iteritems():
398 if key
in ('author', 'committer', 'encoding', 'message', 'branch', 'hg-git'):
401 extra_msg
+= "extra : %s : %s\n" % (key
, urllib
.quote(value
))
404 desc
+= '\n--HG--\n' + extra_msg
406 if len(parents
) == 0 and rev
:
407 print 'reset %s/%s' % (prefix
, ename
)
409 print "commit %s/%s" % (prefix
, ename
)
410 print "mark :%d" % (marks
.get_mark(rev
))
411 print "author %s" % (author
)
412 print "committer %s" % (committer
)
413 print "data %d" % (len(desc
))
417 print "from :%s" % (rev_to_mark(parents
[0]))
419 print "merge :%s" % (rev_to_mark(parents
[1]))
422 export_file(c
.filectx(f
))
424 print "D %s" % (fix_file_path(f
))
428 if (count
% 100 == 0):
429 print "progress revision %d '%s' (%d/%d)" % (rev
, name
, count
, len(revs
))
430 print "#############################################################"
432 # make sure the ref is updated
433 print "reset %s/%s" % (prefix
, ename
)
434 print "from :%u" % rev_to_mark(rev
)
437 marks
.set_tip(ename
, rev
)
439 def export_tag(repo
, tag
):
440 export_ref(repo
, tag
, 'tags', repo
[tag
])
442 def export_bookmark(repo
, bmark
):
444 export_ref(repo
, bmark
, 'bookmarks', head
)
446 def export_branch(repo
, branch
):
447 tip
= get_branch_tip(repo
, branch
)
449 export_ref(repo
, branch
, 'branches', head
)
451 def export_head(repo
):
453 export_ref(repo
, g_head
[0], 'bookmarks', g_head
[1])
455 def do_capabilities(parser
):
456 global prefix
, dirname
460 print "refspec refs/heads/branches/*:%s/branches/*" % prefix
461 print "refspec refs/heads/*:%s/bookmarks/*" % prefix
462 print "refspec refs/tags/*:%s/tags/*" % prefix
464 path
= os
.path
.join(dirname
, 'marks-git')
466 if os
.path
.exists(path
):
467 print "*import-marks %s" % path
468 print "*export-marks %s" % path
472 def branch_tip(repo
, branch
):
473 # older versions of mercurial don't have this
474 if hasattr(repo
, 'branchtip'):
475 return repo
.branchtip(branch
)
477 return repo
.branchtags()[branch
]
479 def get_branch_tip(repo
, branch
):
482 heads
= branches
.get(branch
, None)
486 # verify there's only one head
488 warn("Branch '%s' has more than one head, consider merging" % branch
)
489 return branch_tip(repo
, branch
)
493 def list_head(repo
, cur
):
494 global g_head
, bmarks
496 head
= bookmarks
.readcurrent(repo
)
500 # fake bookmark from current branch
507 if head
== 'default':
511 print "@refs/heads/%s HEAD" % head
512 g_head
= (head
, node
)
515 global branches
, bmarks
, mode
, track_branches
518 for bmark
, node
in bookmarks
.listbookmarks(repo
).iteritems():
519 bmarks
[bmark
] = repo
[node
]
521 cur
= repo
.dirstate
.branch()
526 for branch
in repo
.branchmap():
527 heads
= repo
.branchheads(branch
)
529 branches
[branch
] = heads
531 for branch
in branches
:
532 print "? refs/heads/branches/%s" % branch
535 print "? refs/heads/%s" % bmark
537 for tag
, node
in repo
.tagslist():
540 print "? refs/tags/%s" % tag
544 def do_import(parser
):
547 path
= os
.path
.join(dirname
, 'marks-git')
550 if os
.path
.exists(path
):
551 print "feature import-marks=%s" % path
552 print "feature export-marks=%s" % path
555 tmp
= encoding
.encoding
556 encoding
.encoding
= 'utf-8'
558 # lets get all the import lines
559 while parser
.check('import'):
564 elif ref
.startswith('refs/heads/branches/'):
565 branch
= ref
[len('refs/heads/branches/'):]
566 export_branch(repo
, branch
)
567 elif ref
.startswith('refs/heads/'):
568 bmark
= ref
[len('refs/heads/'):]
569 export_bookmark(repo
, bmark
)
570 elif ref
.startswith('refs/tags/'):
571 tag
= ref
[len('refs/tags/'):]
572 export_tag(repo
, tag
)
576 encoding
.encoding
= tmp
580 def parse_blob(parser
):
584 mark
= parser
.get_mark()
586 data
= parser
.get_data()
587 blob_marks
[mark
] = data
590 def get_merge_files(repo
, p1
, p2
, files
):
591 for e
in repo
[p1
].files():
593 if e
not in repo
[p1
].manifest():
595 f
= { 'ctx' : repo
[p1
][e
] }
598 def parse_commit(parser
):
599 global marks
, blob_marks
, parsed_refs
602 from_mark
= merge_mark
= None
607 commit_mark
= parser
.get_mark()
609 author
= parser
.get_author()
611 committer
= parser
.get_author()
613 data
= parser
.get_data()
615 if parser
.check('from'):
616 from_mark
= parser
.get_mark()
618 if parser
.check('merge'):
619 merge_mark
= parser
.get_mark()
621 if parser
.check('merge'):
622 die('octopus merges are not supported yet')
627 if parser
.check('M'):
628 t
, m
, mark_ref
, path
= line
.split(' ', 3)
629 mark
= int(mark_ref
[1:])
630 f
= { 'mode' : hgmode(m
), 'data' : blob_marks
[mark
] }
631 elif parser
.check('D'):
632 t
, path
= line
.split(' ', 1)
633 f
= { 'deleted' : True }
635 die('Unknown file command: %s' % line
)
638 def getfilectx(repo
, memctx
, f
):
644 is_exec
= of
['mode'] == 'x'
645 is_link
= of
['mode'] == 'l'
646 rename
= of
.get('rename', None)
647 return context
.memfilectx(f
, of
['data'],
648 is_link
, is_exec
, rename
)
652 user
, date
, tz
= author
655 if committer
!= author
:
656 extra
['committer'] = "%s %u %u" % committer
659 p1
= repo
.changelog
.node(mark_to_rev(from_mark
))
664 p2
= repo
.changelog
.node(mark_to_rev(merge_mark
))
669 # If files changed from any of the parents, hg wants to know, but in git if
670 # nothing changed from the first parent, nothing changed.
673 get_merge_files(repo
, p1
, p2
, files
)
675 # Check if the ref is supposed to be a named branch
676 if ref
.startswith('refs/heads/branches/'):
677 extra
['branch'] = ref
[len('refs/heads/branches/'):]
680 i
= data
.find('\n--HG--\n')
682 tmp
= data
[i
+ len('\n--HG--\n'):].strip()
683 for k
, v
in [e
.split(' : ', 1) for e
in tmp
.split('\n')]:
685 old
, new
= v
.split(' => ', 1)
686 files
[new
]['rename'] = old
690 ek
, ev
= v
.split(' : ', 1)
691 extra
[ek
] = urllib
.unquote(ev
)
694 ctx
= context
.memctx(repo
, (p1
, p2
), data
,
695 files
.keys(), getfilectx
,
696 user
, (date
, tz
), extra
)
698 tmp
= encoding
.encoding
699 encoding
.encoding
= 'utf-8'
701 node
= repo
.commitctx(ctx
)
703 encoding
.encoding
= tmp
705 rev
= repo
[node
].rev()
707 parsed_refs
[ref
] = node
708 marks
.new_mark(rev
, commit_mark
)
710 def parse_reset(parser
):
716 if parser
.check('commit'):
719 if not parser
.check('from'):
721 from_mark
= parser
.get_mark()
724 node
= parser
.repo
.changelog
.node(mark_to_rev(from_mark
))
725 parsed_refs
[ref
] = node
727 def parse_tag(parser
):
730 from_mark
= parser
.get_mark()
732 tagger
= parser
.get_author()
734 data
= parser
.get_data()
737 parsed_tags
[name
] = (tagger
, data
)
739 def write_tag(repo
, tag
, node
, msg
, author
):
740 branch
= repo
[node
].branch()
741 tip
= branch_tip(repo
, branch
)
744 def getfilectx(repo
, memctx
, f
):
746 fctx
= tip
.filectx(f
)
748 except error
.ManifestLookupError
:
750 content
= data
+ "%s %s\n" % (hghex(node
), tag
)
751 return context
.memfilectx(f
, content
, False, False, None)
756 author
= (None, 0, 0)
757 user
, date
, tz
= author
759 ctx
= context
.memctx(repo
, (p1
, p2
), msg
,
760 ['.hgtags'], getfilectx
,
761 user
, (date
, tz
), {'branch' : branch
})
763 tmp
= encoding
.encoding
764 encoding
.encoding
= 'utf-8'
766 tagnode
= repo
.commitctx(ctx
)
768 encoding
.encoding
= tmp
772 def do_export(parser
):
773 global parsed_refs
, bmarks
, peer
779 for line
in parser
.each_block('done'):
780 if parser
.check('blob'):
782 elif parser
.check('commit'):
784 elif parser
.check('reset'):
786 elif parser
.check('tag'):
788 elif parser
.check('feature'):
791 die('unhandled export command: %s' % line
)
793 for ref
, node
in parsed_refs
.iteritems():
794 if ref
.startswith('refs/heads/branches'):
795 branch
= ref
[len('refs/heads/branches/'):]
796 if branch
in branches
and node
in branches
[branch
]:
800 elif ref
.startswith('refs/heads/'):
801 bmark
= ref
[len('refs/heads/'):]
802 p_bmarks
.append((bmark
, node
))
804 elif ref
.startswith('refs/tags/'):
805 tag
= ref
[len('refs/tags/'):]
806 author
, msg
= parsed_tags
.get(tag
, (None, None))
809 msg
= 'Added tag %s for changeset %s' % (tag
, hghex(node
[:6]));
810 write_tag(parser
.repo
, tag
, node
, msg
, author
)
812 fp
= parser
.repo
.opener('localtags', 'a')
813 fp
.write('%s %s\n' % (hghex(node
), tag
))
817 # transport-helper/fast-export bugs
821 parser
.repo
.push(peer
, force
=force_push
)
824 for bmark
, node
in p_bmarks
:
825 ref
= 'refs/heads/' + bmark
829 old
= bmarks
[bmark
].hex()
833 if bmark
== 'master' and 'master' not in parser
.repo
._bookmarks
:
836 elif bookmarks
.pushbookmark(parser
.repo
, bmark
, old
, new
):
840 print "error %s" % ref
844 rb
= peer
.listkeys('bookmarks')
845 old
= rb
.get(bmark
, '')
846 if not peer
.pushkey('bookmarks', bmark
, old
, new
):
847 print "error %s" % ref
854 def fix_path(alias
, repo
, orig_url
):
855 url
= urlparse
.urlparse(orig_url
, 'file')
856 if url
.scheme
!= 'file' or os
.path
.isabs(url
.path
):
858 abs_url
= urlparse
.urljoin("%s/" % os
.getcwd(), orig_url
)
859 cmd
= ['git', 'config', 'remote.%s.url' % alias
, "hg::%s" % abs_url
]
863 global prefix
, dirname
, branches
, bmarks
864 global marks
, blob_marks
, parsed_refs
865 global peer
, mode
, bad_mail
, bad_name
866 global track_branches
, force_push
, is_tmp
873 hg_git_compat
= False
874 track_branches
= True
878 if get_config('remote-hg.hg-git-compat') == 'true\n':
880 track_branches
= False
881 if get_config('remote-hg.track-branches') == 'false\n':
882 track_branches
= False
883 if get_config('remote-hg.force-push') == 'false\n':
885 except subprocess
.CalledProcessError
:
890 bad_mail
= 'none@none'
899 alias
= util
.sha1(alias
).hexdigest()
903 gitdir
= os
.environ
['GIT_DIR']
904 dirname
= os
.path
.join(gitdir
, 'hg', alias
)
912 repo
= get_repo(url
, alias
)
913 prefix
= 'refs/hg/%s' % alias
916 fix_path(alias
, peer
or repo
, url
)
918 if not os
.path
.exists(dirname
):
921 marks_path
= os
.path
.join(dirname
, 'marks-hg')
922 marks
= Marks(marks_path
)
924 parser
= Parser(repo
)
926 if parser
.check('capabilities'):
927 do_capabilities(parser
)
928 elif parser
.check('list'):
930 elif parser
.check('import'):
932 elif parser
.check('export'):
935 die('unhandled command: %s' % line
)
944 shutil
.rmtree(dirname
)
947 sys
.exit(main(sys
.argv
))