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 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()
90 def get_config_bool(config
, default
=False):
91 value
= get_config(config
).rstrip('\n')
94 elif value
== "false":
101 def __init__(self
, path
):
111 if not os
.path
.exists(self
.path
):
114 tmp
= json
.load(open(self
.path
))
116 self
.tips
= tmp
['tips']
117 self
.marks
= tmp
['marks']
118 self
.last_mark
= tmp
['last-mark']
120 for rev
, mark
in self
.marks
.iteritems():
121 self
.rev_marks
[mark
] = int(rev
)
124 return { 'tips': self
.tips
, 'marks': self
.marks
, 'last-mark' : self
.last_mark
}
127 json
.dump(self
.dict(), open(self
.path
, 'w'))
130 return str(self
.dict())
132 def from_rev(self
, rev
):
133 return self
.marks
[str(rev
)]
135 def to_rev(self
, mark
):
136 return self
.rev_marks
[mark
]
140 return self
.last_mark
142 def get_mark(self
, rev
):
144 self
.marks
[str(rev
)] = self
.last_mark
145 return self
.last_mark
147 def new_mark(self
, rev
, mark
):
148 self
.marks
[str(rev
)] = mark
149 self
.rev_marks
[mark
] = rev
150 self
.last_mark
= mark
152 def is_marked(self
, rev
):
153 return str(rev
) in self
.marks
155 def get_tip(self
, branch
):
156 return self
.tips
.get(branch
, 0)
158 def set_tip(self
, branch
, tip
):
159 self
.tips
[branch
] = tip
163 def __init__(self
, repo
):
165 self
.line
= self
.get_line()
168 return sys
.stdin
.readline().strip()
170 def __getitem__(self
, i
):
171 return self
.line
.split()[i
]
173 def check(self
, word
):
174 return self
.line
.startswith(word
)
176 def each_block(self
, separator
):
177 while self
.line
!= separator
:
179 self
.line
= self
.get_line()
182 return self
.each_block('')
185 self
.line
= self
.get_line()
186 if self
.line
== 'done':
190 i
= self
.line
.index(':') + 1
191 return int(self
.line
[i
:])
194 if not self
.check('data'):
196 i
= self
.line
.index(' ') + 1
197 size
= int(self
.line
[i
:])
198 return sys
.stdin
.read(size
)
200 def get_author(self
):
204 m
= RAW_AUTHOR_RE
.match(self
.line
)
207 _
, name
, email
, date
, tz
= m
.groups()
208 if name
and 'ext:' in name
:
209 m
= re
.match('^(.+?) ext:\((.+)\)$', name
)
212 ex
= urllib
.unquote(m
.group(2))
214 if email
!= bad_mail
:
216 user
= '%s <%s>' % (name
, email
)
218 user
= '<%s>' % (email
)
226 tz
= ((tz
/ 100) * 3600) + ((tz
% 100) * 60)
227 return (user
, int(date
), -tz
)
229 def fix_file_path(path
):
230 if not os
.path
.isabs(path
):
232 return os
.path
.relpath(path
, '/')
234 def export_files(files
):
235 global marks
, filenodes
239 fid
= node
.hex(f
.filenode())
242 mark
= filenodes
[fid
]
244 mark
= marks
.next_mark()
245 filenodes
[fid
] = mark
249 print "mark :%u" % mark
250 print "data %d" % len(d
)
253 path
= fix_file_path(f
.path())
254 final
.append((gitmode(f
.flags()), mark
, path
))
258 def get_filechanges(repo
, ctx
, parent
):
263 # load earliest manifest first for caching reasons
264 prev
= repo
[parent
].manifest().copy()
269 if (cur
.flags(fn
) != prev
.flags(fn
) or cur
[fn
] != prev
[fn
]):
274 removed |
= set(prev
.keys())
276 return added | modified
, removed
278 def fixup_user_git(user
):
280 user
= user
.replace('"', '')
281 m
= AUTHOR_RE
.match(user
)
284 mail
= m
.group(2).strip()
286 m
= EMAIL_RE
.match(user
)
291 m
= NAME_RE
.match(user
)
293 name
= m
.group(1).strip()
296 def fixup_user_hg(user
):
298 # stole this from hg-git
299 return re
.sub('[<>\n]', '?', name
.lstrip('< ').rstrip('> '))
301 m
= AUTHOR_HG_RE
.match(user
)
303 name
= sanitize(m
.group(1))
304 mail
= sanitize(m
.group(2))
307 name
+= ' ext:(' + urllib
.quote(ex
) + ')'
309 name
= sanitize(user
)
317 def fixup_user(user
):
318 global mode
, bad_mail
321 name
, mail
= fixup_user_git(user
)
323 name
, mail
= fixup_user_hg(user
)
330 return '%s <%s>' % (name
, mail
)
332 def get_repo(url
, alias
):
336 myui
.setconfig('ui', 'interactive', 'off')
337 myui
.fout
= sys
.stderr
339 if get_config_bool('remote-hg.insecure'):
340 myui
.setconfig('web', 'cacerts', '')
343 mod
= extensions
.load(myui
, 'hgext.schemes', None)
349 repo
= hg
.repository(myui
, url
)
351 local_path
= os
.path
.join(dirname
, 'clone')
352 if not os
.path
.exists(local_path
):
354 peer
, dstpeer
= hg
.clone(myui
, {}, url
, local_path
, update
=True, pull
=True)
356 die('Repository error')
357 repo
= dstpeer
.local()
359 repo
= hg
.repository(myui
, local_path
)
361 peer
= hg
.peer(myui
, {}, url
)
363 die('Repository error')
364 repo
.pull(peer
, heads
=None, force
=True)
366 rb
= peer
.listkeys('bookmarks')
367 bookmarks
.updatefromremote(myui
, repo
, rb
, url
)
371 def rev_to_mark(rev
):
373 return marks
.from_rev(rev
)
375 def mark_to_rev(mark
):
377 return marks
.to_rev(mark
)
379 def export_ref(repo
, name
, kind
, head
):
380 global prefix
, marks
, mode
382 ename
= '%s/%s' % (kind
, name
)
383 tip
= marks
.get_tip(ename
)
385 revs
= xrange(tip
, head
.rev() + 1)
388 revs
= [rev
for rev
in revs
if not marks
.is_marked(rev
)]
393 (manifest
, user
, (time
, tz
), files
, desc
, extra
) = repo
.changelog
.read(c
.node())
394 rev_branch
= extra
['branch']
396 author
= "%s %d %s" % (fixup_user(user
), time
, gittz(tz
))
397 if 'committer' in extra
:
398 user
, time
, tz
= extra
['committer'].rsplit(' ', 2)
399 committer
= "%s %s %s" % (user
, time
, gittz(int(tz
)))
403 parents
= [p
for p
in repo
.changelog
.parentrevs(rev
) if p
>= 0]
405 if len(parents
) == 0:
406 modified
= c
.manifest().keys()
409 modified
, removed
= get_filechanges(repo
, c
, parents
[0])
416 if rev_branch
!= 'default':
417 extra_msg
+= 'branch : %s\n' % rev_branch
421 if f
not in c
.manifest():
423 rename
= c
.filectx(f
).renamed()
425 renames
.append((rename
[0], f
))
428 extra_msg
+= "rename : %s => %s\n" % e
430 for key
, value
in extra
.iteritems():
431 if key
in ('author', 'committer', 'encoding', 'message', 'branch', 'hg-git'):
434 extra_msg
+= "extra : %s : %s\n" % (key
, urllib
.quote(value
))
437 desc
+= '\n--HG--\n' + extra_msg
439 if len(parents
) == 0 and rev
:
440 print 'reset %s/%s' % (prefix
, ename
)
442 modified_final
= export_files(c
.filectx(f
) for f
in modified
)
444 print "commit %s/%s" % (prefix
, ename
)
445 print "mark :%d" % (marks
.get_mark(rev
))
446 print "author %s" % (author
)
447 print "committer %s" % (committer
)
448 print "data %d" % (len(desc
))
452 print "from :%s" % (rev_to_mark(parents
[0]))
454 print "merge :%s" % (rev_to_mark(parents
[1]))
456 for f
in modified_final
:
457 print "M %s :%u %s" % f
459 print "D %s" % (fix_file_path(f
))
463 if (count
% 100 == 0):
464 print "progress revision %d '%s' (%d/%d)" % (rev
, name
, count
, len(revs
))
466 # make sure the ref is updated
467 print "reset %s/%s" % (prefix
, ename
)
468 print "from :%u" % rev_to_mark(rev
)
471 marks
.set_tip(ename
, rev
)
473 def export_tag(repo
, tag
):
474 export_ref(repo
, tag
, 'tags', repo
[hgref(tag
)])
476 def export_bookmark(repo
, bmark
):
477 head
= bmarks
[hgref(bmark
)]
478 export_ref(repo
, bmark
, 'bookmarks', head
)
480 def export_branch(repo
, branch
):
481 tip
= get_branch_tip(repo
, branch
)
483 export_ref(repo
, branch
, 'branches', head
)
485 def export_head(repo
):
487 export_ref(repo
, g_head
[0], 'bookmarks', g_head
[1])
489 def do_capabilities(parser
):
490 global prefix
, dirname
494 print "refspec refs/heads/branches/*:%s/branches/*" % prefix
495 print "refspec refs/heads/*:%s/bookmarks/*" % prefix
496 print "refspec refs/tags/*:%s/tags/*" % prefix
498 path
= os
.path
.join(dirname
, 'marks-git')
500 if os
.path
.exists(path
):
501 print "*import-marks %s" % path
502 print "*export-marks %s" % path
506 def branch_tip(repo
, branch
):
507 # older versions of mercurial don't have this
508 if hasattr(repo
, 'branchtip'):
509 return repo
.branchtip(branch
)
511 return repo
.branchtags()[branch
]
513 def get_branch_tip(repo
, branch
):
516 heads
= branches
.get(hgref(branch
), None)
520 # verify there's only one head
522 warn("Branch '%s' has more than one head, consider merging" % branch
)
523 return branch_tip(repo
, hgref(branch
))
527 def list_head(repo
, cur
):
528 global g_head
, bmarks
530 head
= bookmarks
.readcurrent(repo
)
534 # fake bookmark from current branch
541 if head
== 'default':
546 print "@refs/heads/%s HEAD" % head
547 g_head
= (head
, node
)
550 global branches
, bmarks
, track_branches
553 for bmark
, node
in bookmarks
.listbookmarks(repo
).iteritems():
554 bmarks
[bmark
] = repo
[node
]
556 cur
= repo
.dirstate
.branch()
561 for branch
in repo
.branchmap():
562 heads
= repo
.branchheads(branch
)
564 branches
[branch
] = heads
566 for branch
in branches
:
567 print "? refs/heads/branches/%s" % gitref(branch
)
570 print "? refs/heads/%s" % gitref(bmark
)
572 for tag
, node
in repo
.tagslist():
575 print "? refs/tags/%s" % gitref(tag
)
579 def do_import(parser
):
582 path
= os
.path
.join(dirname
, 'marks-git')
585 if os
.path
.exists(path
):
586 print "feature import-marks=%s" % path
587 print "feature export-marks=%s" % path
590 tmp
= encoding
.encoding
591 encoding
.encoding
= 'utf-8'
593 # lets get all the import lines
594 while parser
.check('import'):
599 elif ref
.startswith('refs/heads/branches/'):
600 branch
= ref
[len('refs/heads/branches/'):]
601 export_branch(repo
, branch
)
602 elif ref
.startswith('refs/heads/'):
603 bmark
= ref
[len('refs/heads/'):]
604 export_bookmark(repo
, bmark
)
605 elif ref
.startswith('refs/tags/'):
606 tag
= ref
[len('refs/tags/'):]
607 export_tag(repo
, tag
)
611 encoding
.encoding
= tmp
615 def parse_blob(parser
):
619 mark
= parser
.get_mark()
621 data
= parser
.get_data()
622 blob_marks
[mark
] = data
625 def get_merge_files(repo
, p1
, p2
, files
):
626 for e
in repo
[p1
].files():
628 if e
not in repo
[p1
].manifest():
630 f
= { 'ctx' : repo
[p1
][e
] }
633 def parse_commit(parser
):
634 global marks
, blob_marks
, parsed_refs
637 from_mark
= merge_mark
= None
642 commit_mark
= parser
.get_mark()
644 author
= parser
.get_author()
646 committer
= parser
.get_author()
648 data
= parser
.get_data()
650 if parser
.check('from'):
651 from_mark
= parser
.get_mark()
653 if parser
.check('merge'):
654 merge_mark
= parser
.get_mark()
656 if parser
.check('merge'):
657 die('octopus merges are not supported yet')
659 # fast-export adds an extra newline
666 if parser
.check('M'):
667 t
, m
, mark_ref
, path
= line
.split(' ', 3)
668 mark
= int(mark_ref
[1:])
669 f
= { 'mode' : hgmode(m
), 'data' : blob_marks
[mark
] }
670 elif parser
.check('D'):
671 t
, path
= line
.split(' ', 1)
672 f
= { 'deleted' : True }
674 die('Unknown file command: %s' % line
)
677 def getfilectx(repo
, memctx
, f
):
683 is_exec
= of
['mode'] == 'x'
684 is_link
= of
['mode'] == 'l'
685 rename
= of
.get('rename', None)
686 return context
.memfilectx(f
, of
['data'],
687 is_link
, is_exec
, rename
)
691 user
, date
, tz
= author
694 if committer
!= author
:
695 extra
['committer'] = "%s %u %u" % committer
698 p1
= repo
.changelog
.node(mark_to_rev(from_mark
))
703 p2
= repo
.changelog
.node(mark_to_rev(merge_mark
))
708 # If files changed from any of the parents, hg wants to know, but in git if
709 # nothing changed from the first parent, nothing changed.
712 get_merge_files(repo
, p1
, p2
, files
)
714 # Check if the ref is supposed to be a named branch
715 if ref
.startswith('refs/heads/branches/'):
716 branch
= ref
[len('refs/heads/branches/'):]
717 extra
['branch'] = hgref(branch
)
720 i
= data
.find('\n--HG--\n')
722 tmp
= data
[i
+ len('\n--HG--\n'):].strip()
723 for k
, v
in [e
.split(' : ', 1) for e
in tmp
.split('\n')]:
725 old
, new
= v
.split(' => ', 1)
726 files
[new
]['rename'] = old
730 ek
, ev
= v
.split(' : ', 1)
731 extra
[ek
] = urllib
.unquote(ev
)
734 ctx
= context
.memctx(repo
, (p1
, p2
), data
,
735 files
.keys(), getfilectx
,
736 user
, (date
, tz
), extra
)
738 tmp
= encoding
.encoding
739 encoding
.encoding
= 'utf-8'
741 node
= repo
.commitctx(ctx
)
743 encoding
.encoding
= tmp
745 rev
= repo
[node
].rev()
747 parsed_refs
[ref
] = node
748 marks
.new_mark(rev
, commit_mark
)
750 def parse_reset(parser
):
756 if parser
.check('commit'):
759 if not parser
.check('from'):
761 from_mark
= parser
.get_mark()
764 node
= parser
.repo
.changelog
.node(mark_to_rev(from_mark
))
765 parsed_refs
[ref
] = node
767 def parse_tag(parser
):
770 from_mark
= parser
.get_mark()
772 tagger
= parser
.get_author()
774 data
= parser
.get_data()
777 parsed_tags
[name
] = (tagger
, data
)
779 def write_tag(repo
, tag
, node
, msg
, author
):
780 branch
= repo
[node
].branch()
781 tip
= branch_tip(repo
, branch
)
784 def getfilectx(repo
, memctx
, f
):
786 fctx
= tip
.filectx(f
)
788 except error
.ManifestLookupError
:
790 content
= data
+ "%s %s\n" % (hghex(node
), tag
)
791 return context
.memfilectx(f
, content
, False, False, None)
796 author
= (None, 0, 0)
797 user
, date
, tz
= author
799 ctx
= context
.memctx(repo
, (p1
, p2
), msg
,
800 ['.hgtags'], getfilectx
,
801 user
, (date
, tz
), {'branch' : branch
})
803 tmp
= encoding
.encoding
804 encoding
.encoding
= 'utf-8'
806 tagnode
= repo
.commitctx(ctx
)
808 encoding
.encoding
= tmp
812 def do_export(parser
):
813 global parsed_refs
, bmarks
, peer
819 for line
in parser
.each_block('done'):
820 if parser
.check('blob'):
822 elif parser
.check('commit'):
824 elif parser
.check('reset'):
826 elif parser
.check('tag'):
828 elif parser
.check('feature'):
831 die('unhandled export command: %s' % line
)
833 for ref
, node
in parsed_refs
.iteritems():
834 if ref
.startswith('refs/heads/branches'):
835 branch
= ref
[len('refs/heads/branches/'):]
836 if branch
in branches
and node
in branches
[branch
]:
840 elif ref
.startswith('refs/heads/'):
841 bmark
= ref
[len('refs/heads/'):]
842 p_bmarks
.append((bmark
, node
))
844 elif ref
.startswith('refs/tags/'):
845 tag
= ref
[len('refs/tags/'):]
847 author
, msg
= parsed_tags
.get(tag
, (None, None))
850 msg
= 'Added tag %s for changeset %s' % (tag
, hghex(node
[:6]));
851 write_tag(parser
.repo
, tag
, node
, msg
, author
)
853 fp
= parser
.repo
.opener('localtags', 'a')
854 fp
.write('%s %s\n' % (hghex(node
), tag
))
858 # transport-helper/fast-export bugs
862 parser
.repo
.push(peer
, force
=force_push
, newbranch
=True)
865 for bmark
, node
in p_bmarks
:
866 ref
= 'refs/heads/' + bmark
870 old
= bmarks
[bmark
].hex()
877 if bmark
== 'master' and 'master' not in parser
.repo
._bookmarks
:
881 elif bookmarks
.pushbookmark(parser
.repo
, bmark
, old
, new
):
885 print "error %s" % ref
889 rb
= peer
.listkeys('bookmarks')
890 old
= rb
.get(bmark
, '')
891 if not peer
.pushkey('bookmarks', bmark
, old
, new
):
892 print "error %s" % ref
899 def fix_path(alias
, repo
, orig_url
):
900 url
= urlparse
.urlparse(orig_url
, 'file')
901 if url
.scheme
!= 'file' or os
.path
.isabs(url
.path
):
903 abs_url
= urlparse
.urljoin("%s/" % os
.getcwd(), orig_url
)
904 cmd
= ['git', 'config', 'remote.%s.url' % alias
, "hg::%s" % abs_url
]
908 global prefix
, dirname
, branches
, bmarks
909 global marks
, blob_marks
, parsed_refs
910 global peer
, mode
, bad_mail
, bad_name
911 global track_branches
, force_push
, is_tmp
919 hg_git_compat
= get_config_bool('remote-hg.hg-git-compat')
920 track_branches
= get_config_bool('remote-hg.track-branches', True)
921 force_push
= get_config_bool('remote-hg.force-push')
925 bad_mail
= 'none@none'
934 alias
= hashlib
.sha1(alias
).hexdigest()
938 gitdir
= os
.environ
['GIT_DIR']
939 dirname
= os
.path
.join(gitdir
, 'hg', alias
)
948 repo
= get_repo(url
, alias
)
949 prefix
= 'refs/hg/%s' % alias
952 fix_path(alias
, peer
or repo
, url
)
954 if not os
.path
.exists(dirname
):
957 marks_path
= os
.path
.join(dirname
, 'marks-hg')
958 marks
= Marks(marks_path
)
960 parser
= Parser(repo
)
962 if parser
.check('capabilities'):
963 do_capabilities(parser
)
964 elif parser
.check('list'):
966 elif parser
.check('import'):
968 elif parser
.check('export'):
971 die('unhandled command: %s' % line
)
980 shutil
.rmtree(dirname
)
983 sys
.exit(main(sys
.argv
))