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+)')
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', '')
342 extensions
.loadall(myui
)
345 repo
= hg
.repository(myui
, url
)
346 if not os
.path
.exists(dirname
):
349 shared_path
= os
.path
.join(gitdir
, 'hg')
350 if not os
.path
.exists(shared_path
):
352 hg
.clone(myui
, {}, url
, shared_path
, update
=False, pull
=True)
354 die('Repository error')
356 if not os
.path
.exists(dirname
):
359 local_path
= os
.path
.join(dirname
, 'clone')
360 if not os
.path
.exists(local_path
):
361 hg
.share(myui
, shared_path
, local_path
, update
=False)
363 repo
= hg
.repository(myui
, local_path
)
365 peer
= hg
.peer(myui
, {}, url
)
367 die('Repository error')
368 repo
.pull(peer
, heads
=None, force
=True)
372 def rev_to_mark(rev
):
374 return marks
.from_rev(rev
)
376 def mark_to_rev(mark
):
378 return marks
.to_rev(mark
)
380 def export_ref(repo
, name
, kind
, head
):
381 global prefix
, marks
, mode
383 ename
= '%s/%s' % (kind
, name
)
384 tip
= marks
.get_tip(ename
)
386 revs
= xrange(tip
, head
.rev() + 1)
389 revs
= [rev
for rev
in revs
if not marks
.is_marked(rev
)]
394 (manifest
, user
, (time
, tz
), files
, desc
, extra
) = repo
.changelog
.read(c
.node())
395 rev_branch
= extra
['branch']
397 author
= "%s %d %s" % (fixup_user(user
), time
, gittz(tz
))
398 if 'committer' in extra
:
399 user
, time
, tz
= extra
['committer'].rsplit(' ', 2)
400 committer
= "%s %s %s" % (user
, time
, gittz(int(tz
)))
404 parents
= [p
for p
in repo
.changelog
.parentrevs(rev
) if p
>= 0]
406 if len(parents
) == 0:
407 modified
= c
.manifest().keys()
410 modified
, removed
= get_filechanges(repo
, c
, parents
[0])
417 if rev_branch
!= 'default':
418 extra_msg
+= 'branch : %s\n' % rev_branch
422 if f
not in c
.manifest():
424 rename
= c
.filectx(f
).renamed()
426 renames
.append((rename
[0], f
))
429 extra_msg
+= "rename : %s => %s\n" % e
431 for key
, value
in extra
.iteritems():
432 if key
in ('author', 'committer', 'encoding', 'message', 'branch', 'hg-git'):
435 extra_msg
+= "extra : %s : %s\n" % (key
, urllib
.quote(value
))
438 desc
+= '\n--HG--\n' + extra_msg
440 if len(parents
) == 0 and rev
:
441 print 'reset %s/%s' % (prefix
, ename
)
443 modified_final
= export_files(c
.filectx(f
) for f
in modified
)
445 print "commit %s/%s" % (prefix
, ename
)
446 print "mark :%d" % (marks
.get_mark(rev
))
447 print "author %s" % (author
)
448 print "committer %s" % (committer
)
449 print "data %d" % (len(desc
))
453 print "from :%s" % (rev_to_mark(parents
[0]))
455 print "merge :%s" % (rev_to_mark(parents
[1]))
457 for f
in modified_final
:
458 print "M %s :%u %s" % f
460 print "D %s" % (fix_file_path(f
))
464 if (count
% 100 == 0):
465 print "progress revision %d '%s' (%d/%d)" % (rev
, name
, count
, len(revs
))
467 # make sure the ref is updated
468 print "reset %s/%s" % (prefix
, ename
)
469 print "from :%u" % rev_to_mark(rev
)
472 marks
.set_tip(ename
, rev
)
474 def export_tag(repo
, tag
):
475 export_ref(repo
, tag
, 'tags', repo
[hgref(tag
)])
477 def export_bookmark(repo
, bmark
):
478 head
= bmarks
[hgref(bmark
)]
479 export_ref(repo
, bmark
, 'bookmarks', head
)
481 def export_branch(repo
, branch
):
482 tip
= get_branch_tip(repo
, branch
)
484 export_ref(repo
, branch
, 'branches', head
)
486 def export_head(repo
):
488 export_ref(repo
, g_head
[0], 'bookmarks', g_head
[1])
490 def do_capabilities(parser
):
491 global prefix
, dirname
495 print "refspec refs/heads/branches/*:%s/branches/*" % prefix
496 print "refspec refs/heads/*:%s/bookmarks/*" % prefix
497 print "refspec refs/tags/*:%s/tags/*" % prefix
499 path
= os
.path
.join(dirname
, 'marks-git')
501 if os
.path
.exists(path
):
502 print "*import-marks %s" % path
503 print "*export-marks %s" % path
507 def branch_tip(repo
, branch
):
508 # older versions of mercurial don't have this
509 if hasattr(repo
, 'branchtip'):
510 return repo
.branchtip(branch
)
512 return repo
.branchtags()[branch
]
514 def get_branch_tip(repo
, branch
):
517 heads
= branches
.get(hgref(branch
), None)
521 # verify there's only one head
523 warn("Branch '%s' has more than one head, consider merging" % branch
)
524 return branch_tip(repo
, hgref(branch
))
528 def list_head(repo
, cur
):
529 global g_head
, bmarks
531 head
= bookmarks
.readcurrent(repo
)
535 # fake bookmark from current branch
542 if head
== 'default':
547 print "@refs/heads/%s HEAD" % head
548 g_head
= (head
, node
)
551 global branches
, bmarks
, track_branches
554 for bmark
, node
in bookmarks
.listbookmarks(repo
).iteritems():
555 bmarks
[bmark
] = repo
[node
]
557 cur
= repo
.dirstate
.branch()
562 for branch
in repo
.branchmap():
563 heads
= repo
.branchheads(branch
)
565 branches
[branch
] = heads
567 for branch
in branches
:
568 print "? refs/heads/branches/%s" % gitref(branch
)
571 print "? refs/heads/%s" % gitref(bmark
)
573 for tag
, node
in repo
.tagslist():
576 print "? refs/tags/%s" % gitref(tag
)
580 def do_import(parser
):
583 path
= os
.path
.join(dirname
, 'marks-git')
586 if os
.path
.exists(path
):
587 print "feature import-marks=%s" % path
588 print "feature export-marks=%s" % path
591 tmp
= encoding
.encoding
592 encoding
.encoding
= 'utf-8'
594 # lets get all the import lines
595 while parser
.check('import'):
600 elif ref
.startswith('refs/heads/branches/'):
601 branch
= ref
[len('refs/heads/branches/'):]
602 export_branch(repo
, branch
)
603 elif ref
.startswith('refs/heads/'):
604 bmark
= ref
[len('refs/heads/'):]
605 export_bookmark(repo
, bmark
)
606 elif ref
.startswith('refs/tags/'):
607 tag
= ref
[len('refs/tags/'):]
608 export_tag(repo
, tag
)
612 encoding
.encoding
= tmp
616 def parse_blob(parser
):
620 mark
= parser
.get_mark()
622 data
= parser
.get_data()
623 blob_marks
[mark
] = data
626 def get_merge_files(repo
, p1
, p2
, files
):
627 for e
in repo
[p1
].files():
629 if e
not in repo
[p1
].manifest():
631 f
= { 'ctx' : repo
[p1
][e
] }
634 def parse_commit(parser
):
635 global marks
, blob_marks
, parsed_refs
638 from_mark
= merge_mark
= None
643 commit_mark
= parser
.get_mark()
645 author
= parser
.get_author()
647 committer
= parser
.get_author()
649 data
= parser
.get_data()
651 if parser
.check('from'):
652 from_mark
= parser
.get_mark()
654 if parser
.check('merge'):
655 merge_mark
= parser
.get_mark()
657 if parser
.check('merge'):
658 die('octopus merges are not supported yet')
660 # fast-export adds an extra newline
667 if parser
.check('M'):
668 t
, m
, mark_ref
, path
= line
.split(' ', 3)
669 mark
= int(mark_ref
[1:])
670 f
= { 'mode' : hgmode(m
), 'data' : blob_marks
[mark
] }
671 elif parser
.check('D'):
672 t
, path
= line
.split(' ', 1)
673 f
= { 'deleted' : True }
675 die('Unknown file command: %s' % line
)
678 def getfilectx(repo
, memctx
, f
):
684 is_exec
= of
['mode'] == 'x'
685 is_link
= of
['mode'] == 'l'
686 rename
= of
.get('rename', None)
687 return context
.memfilectx(f
, of
['data'],
688 is_link
, is_exec
, rename
)
692 user
, date
, tz
= author
695 if committer
!= author
:
696 extra
['committer'] = "%s %u %u" % committer
699 p1
= repo
.changelog
.node(mark_to_rev(from_mark
))
704 p2
= repo
.changelog
.node(mark_to_rev(merge_mark
))
709 # If files changed from any of the parents, hg wants to know, but in git if
710 # nothing changed from the first parent, nothing changed.
713 get_merge_files(repo
, p1
, p2
, files
)
715 # Check if the ref is supposed to be a named branch
716 if ref
.startswith('refs/heads/branches/'):
717 branch
= ref
[len('refs/heads/branches/'):]
718 extra
['branch'] = hgref(branch
)
721 i
= data
.find('\n--HG--\n')
723 tmp
= data
[i
+ len('\n--HG--\n'):].strip()
724 for k
, v
in [e
.split(' : ', 1) for e
in tmp
.split('\n')]:
726 old
, new
= v
.split(' => ', 1)
727 files
[new
]['rename'] = old
731 ek
, ev
= v
.split(' : ', 1)
732 extra
[ek
] = urllib
.unquote(ev
)
735 ctx
= context
.memctx(repo
, (p1
, p2
), data
,
736 files
.keys(), getfilectx
,
737 user
, (date
, tz
), extra
)
739 tmp
= encoding
.encoding
740 encoding
.encoding
= 'utf-8'
742 node
= repo
.commitctx(ctx
)
744 encoding
.encoding
= tmp
746 rev
= repo
[node
].rev()
748 parsed_refs
[ref
] = node
749 marks
.new_mark(rev
, commit_mark
)
751 def parse_reset(parser
):
757 if parser
.check('commit'):
760 if not parser
.check('from'):
762 from_mark
= parser
.get_mark()
765 node
= parser
.repo
.changelog
.node(mark_to_rev(from_mark
))
766 parsed_refs
[ref
] = node
768 def parse_tag(parser
):
771 from_mark
= parser
.get_mark()
773 tagger
= parser
.get_author()
775 data
= parser
.get_data()
778 parsed_tags
[name
] = (tagger
, data
)
780 def write_tag(repo
, tag
, node
, msg
, author
):
781 branch
= repo
[node
].branch()
782 tip
= branch_tip(repo
, branch
)
785 def getfilectx(repo
, memctx
, f
):
787 fctx
= tip
.filectx(f
)
789 except error
.ManifestLookupError
:
791 content
= data
+ "%s %s\n" % (hghex(node
), tag
)
792 return context
.memfilectx(f
, content
, False, False, None)
797 author
= (None, 0, 0)
798 user
, date
, tz
= author
800 ctx
= context
.memctx(repo
, (p1
, p2
), msg
,
801 ['.hgtags'], getfilectx
,
802 user
, (date
, tz
), {'branch' : branch
})
804 tmp
= encoding
.encoding
805 encoding
.encoding
= 'utf-8'
807 tagnode
= repo
.commitctx(ctx
)
809 encoding
.encoding
= tmp
813 def do_export(parser
):
814 global parsed_refs
, bmarks
, peer
820 for line
in parser
.each_block('done'):
821 if parser
.check('blob'):
823 elif parser
.check('commit'):
825 elif parser
.check('reset'):
827 elif parser
.check('tag'):
829 elif parser
.check('feature'):
832 die('unhandled export command: %s' % line
)
834 for ref
, node
in parsed_refs
.iteritems():
835 if ref
.startswith('refs/heads/branches'):
836 branch
= ref
[len('refs/heads/branches/'):]
837 if branch
in branches
and node
in branches
[branch
]:
841 elif ref
.startswith('refs/heads/'):
842 bmark
= ref
[len('refs/heads/'):]
843 p_bmarks
.append((bmark
, node
))
845 elif ref
.startswith('refs/tags/'):
846 tag
= ref
[len('refs/tags/'):]
848 author
, msg
= parsed_tags
.get(tag
, (None, None))
851 msg
= 'Added tag %s for changeset %s' % (tag
, hghex(node
[:6]));
852 write_tag(parser
.repo
, tag
, node
, msg
, author
)
854 fp
= parser
.repo
.opener('localtags', 'a')
855 fp
.write('%s %s\n' % (hghex(node
), tag
))
859 # transport-helper/fast-export bugs
863 parser
.repo
.push(peer
, force
=force_push
, newbranch
=True)
864 remote_bmarks
= peer
.listkeys('bookmarks')
867 for bmark
, node
in p_bmarks
:
868 ref
= 'refs/heads/' + bmark
872 old
= bmarks
[bmark
].hex()
879 if bmark
== 'master' and 'master' not in parser
.repo
._bookmarks
:
883 elif bookmarks
.pushbookmark(parser
.repo
, bmark
, old
, new
):
887 print "error %s" % ref
891 old
= remote_bmarks
.get(bmark
, '')
892 if not peer
.pushkey('bookmarks', bmark
, old
, new
):
893 print "error %s" % ref
900 def fix_path(alias
, repo
, orig_url
):
901 url
= urlparse
.urlparse(orig_url
, 'file')
902 if url
.scheme
!= 'file' or os
.path
.isabs(url
.path
):
904 abs_url
= urlparse
.urljoin("%s/" % os
.getcwd(), orig_url
)
905 cmd
= ['git', 'config', 'remote.%s.url' % alias
, "hg::%s" % abs_url
]
909 global prefix
, gitdir
, dirname
, branches
, bmarks
910 global marks
, blob_marks
, parsed_refs
911 global peer
, mode
, bad_mail
, bad_name
912 global track_branches
, force_push
, is_tmp
920 hg_git_compat
= get_config_bool('remote-hg.hg-git-compat')
921 track_branches
= get_config_bool('remote-hg.track-branches', True)
922 force_push
= get_config_bool('remote-hg.force-push')
926 bad_mail
= 'none@none'
935 alias
= hashlib
.sha1(alias
).hexdigest()
939 gitdir
= os
.environ
['GIT_DIR']
940 dirname
= os
.path
.join(gitdir
, 'hg', alias
)
949 repo
= get_repo(url
, alias
)
950 prefix
= 'refs/hg/%s' % alias
953 fix_path(alias
, peer
or repo
, url
)
955 marks_path
= os
.path
.join(dirname
, 'marks-hg')
956 marks
= Marks(marks_path
)
958 if sys
.platform
== 'win32':
960 msvcrt
.setmode(sys
.stdout
.fileno(), os
.O_BINARY
)
962 parser
= Parser(repo
)
964 if parser
.check('capabilities'):
965 do_capabilities(parser
)
966 elif parser
.check('list'):
968 elif parser
.check('import'):
970 elif parser
.check('export'):
973 die('unhandled command: %s' % line
)
982 shutil
.rmtree(dirname
)
985 sys
.exit(main(sys
.argv
))