3 # Copyright (c) 2012 Felipe Contreras
7 # Just copy to your ~/bin, or anywhere in your $PATH.
8 # Then you can clone with:
9 # % git clone bzr::/path/to/bzr/repo/or/url
12 # % git clone bzr::$HOME/myrepo
14 # % git clone bzr::lp:myrepo
16 # If you want to specify which branches you want track (per repo):
17 # git config remote-bzr.branches 'trunk, devel, test'
23 if hasattr(bzrlib
, "initialize"):
27 bzrlib
.plugin
.load_plugins()
29 import bzrlib
.generate_ids
30 import bzrlib
.transport
33 import bzrlib
.urlutils
40 import atexit
, shutil
, hashlib
, urlparse
, subprocess
42 NAME_RE
= re
.compile('^([^<>]+)')
43 AUTHOR_RE
= re
.compile('^([^<>]+?)? ?<([^<>]*)>$')
44 EMAIL_RE
= re
.compile('^([^<>]+[^ \\\t<>])?\\b(?:[ \\t<>]*?)\\b([^ \\t<>]+@[^ \\t<>]+)')
45 RAW_AUTHOR_RE
= re
.compile('^(\w+) (.+)? <(.*)> (\d+) ([+-]\d+)')
48 sys
.stderr
.write('ERROR: %s\n' % (msg
% args
))
52 sys
.stderr
.write('WARNING: %s\n' % (msg
% args
))
55 return '%+03d%02d' % (tz
/ 3600, tz
% 3600 / 60)
57 def get_config(config
):
58 cmd
= ['git', 'config', '--get', config
]
59 process
= subprocess
.Popen(cmd
, stdout
=subprocess
.PIPE
)
60 output
, _
= process
.communicate()
65 def __init__(self
, path
):
74 if not os
.path
.exists(self
.path
):
77 tmp
= json
.load(open(self
.path
))
78 self
.tips
= tmp
['tips']
79 self
.marks
= tmp
['marks']
80 self
.last_mark
= tmp
['last-mark']
82 for rev
, mark
in self
.marks
.iteritems():
83 self
.rev_marks
[mark
] = rev
86 return { 'tips': self
.tips
, 'marks': self
.marks
, 'last-mark' : self
.last_mark
}
89 json
.dump(self
.dict(), open(self
.path
, 'w'))
92 return str(self
.dict())
94 def from_rev(self
, rev
):
95 return self
.marks
[rev
]
97 def to_rev(self
, mark
):
98 return self
.rev_marks
[mark
]
102 return self
.last_mark
104 def get_mark(self
, rev
):
106 self
.marks
[rev
] = self
.last_mark
107 return self
.last_mark
109 def is_marked(self
, rev
):
110 return rev
in self
.marks
112 def new_mark(self
, rev
, mark
):
113 self
.marks
[rev
] = mark
114 self
.rev_marks
[mark
] = rev
115 self
.last_mark
= mark
117 def get_tip(self
, branch
):
118 return self
.tips
.get(branch
, None)
120 def set_tip(self
, branch
, tip
):
121 self
.tips
[branch
] = tip
125 def __init__(self
, repo
):
127 self
.line
= self
.get_line()
130 return sys
.stdin
.readline().strip()
132 def __getitem__(self
, i
):
133 return self
.line
.split()[i
]
135 def check(self
, word
):
136 return self
.line
.startswith(word
)
138 def each_block(self
, separator
):
139 while self
.line
!= separator
:
141 self
.line
= self
.get_line()
144 return self
.each_block('')
147 self
.line
= self
.get_line()
148 if self
.line
== 'done':
152 i
= self
.line
.index(':') + 1
153 return int(self
.line
[i
:])
156 if not self
.check('data'):
158 i
= self
.line
.index(' ') + 1
159 size
= int(self
.line
[i
:])
160 return sys
.stdin
.read(size
)
162 def get_author(self
):
163 m
= RAW_AUTHOR_RE
.match(self
.line
)
166 _
, name
, email
, date
, tz
= m
.groups()
167 committer
= '%s <%s>' % (name
, email
)
169 tz
= ((tz
/ 100) * 3600) + ((tz
% 100) * 60)
170 return (committer
, int(date
), tz
)
172 def rev_to_mark(rev
):
174 return marks
.from_rev(rev
)
176 def mark_to_rev(mark
):
178 return marks
.to_rev(mark
)
180 def fixup_user(user
):
182 user
= user
.replace('"', '')
183 m
= AUTHOR_RE
.match(user
)
186 mail
= m
.group(2).strip()
188 m
= EMAIL_RE
.match(user
)
193 m
= NAME_RE
.match(user
)
195 name
= m
.group(1).strip()
202 return '%s <%s>' % (name
, mail
)
204 def get_filechanges(cur
, prev
):
208 changes
= cur
.changes_from(prev
)
211 return s
.encode('utf-8')
213 for path
, fid
, kind
in changes
.added
:
214 modified
[u(path
)] = fid
215 for path
, fid
, kind
in changes
.removed
:
216 removed
[u(path
)] = None
217 for path
, fid
, kind
, mod
, _
in changes
.modified
:
218 modified
[u(path
)] = fid
219 for oldpath
, newpath
, fid
, kind
, mod
, _
in changes
.renamed
:
220 removed
[u(oldpath
)] = None
221 if kind
== 'directory':
222 lst
= cur
.list_files(from_dir
=newpath
, recursive
=True)
223 for path
, file_class
, kind
, fid
, entry
in lst
:
224 if kind
!= 'directory':
225 modified
[u(newpath
+ '/' + path
)] = fid
227 modified
[u(newpath
)] = fid
229 return modified
, removed
231 def export_files(tree
, files
):
232 global marks
, filenodes
235 for path
, fid
in files
.iteritems():
236 kind
= tree
.kind(fid
)
238 h
= tree
.get_file_sha1(fid
)
240 if kind
== 'symlink':
241 d
= tree
.get_symlink_target(fid
)
245 if tree
.is_executable(fid
):
250 # is the blob already exported?
253 final
.append((mode
, mark
, path
))
256 d
= tree
.get_file_text(fid
)
257 elif kind
== 'directory':
260 die("Unhandled kind '%s' for path '%s'" % (kind
, path
))
262 mark
= marks
.next_mark()
266 print "mark :%u" % mark
267 print "data %d" % len(d
)
270 final
.append((mode
, mark
, path
))
274 def export_branch(repo
, name
):
277 ref
= '%s/heads/%s' % (prefix
, name
)
278 tip
= marks
.get_tip(name
)
280 branch
= branches
[name
]
281 repo
= branch
.repository
284 revs
= branch
.iter_merge_sorted_revisions(None, tip
, 'exclude', 'forward')
285 tip_revno
= branch
.revision_id_to_revno(tip
)
286 last_revno
, _
= branch
.last_revision_info()
287 total
= last_revno
- tip_revno
289 revs
= [revid
, seq
for revid
, _
, seq
, _
in revs
if not marks
.is_marked(revid
)]
291 for revid
, seq
in revs
:
293 rev
= repo
.get_revision(revid
)
296 parents
= rev
.parent_ids
299 committer
= rev
.committer
.encode('utf-8')
300 committer
= "%s %u %s" % (fixup_user(committer
), time
, gittz(tz
))
301 authors
= rev
.get_apparent_authors()
303 author
= authors
[0].encode('utf-8')
304 author
= "%s %u %s" % (fixup_user(author
), time
, gittz(tz
))
307 msg
= rev
.message
.encode('utf-8')
311 if len(parents
) == 0:
312 parent
= bzrlib
.revision
.NULL_REVISION
316 cur_tree
= repo
.revision_tree(revid
)
317 prev
= repo
.revision_tree(parent
)
318 modified
, removed
= get_filechanges(cur_tree
, prev
)
320 modified_final
= export_files(cur_tree
, modified
)
322 if len(parents
) == 0:
323 print 'reset %s' % ref
325 print "commit %s" % ref
326 print "mark :%d" % (marks
.get_mark(revid
))
327 print "author %s" % (author
)
328 print "committer %s" % (committer
)
329 print "data %d" % (len(msg
))
332 for i
, p
in enumerate(parents
):
341 print "merge :%s" % m
345 for f
in modified_final
:
346 print "M %s :%u %s" % f
350 # let's skip branch revisions from the progress report
353 progress
= (revno
- tip_revno
)
354 if (progress
% 100 == 0):
355 print "progress revision %d '%s' (%d/%d)" % (revno
, name
, progress
, total
)
359 revid
= branch
.last_revision()
361 # make sure the ref is updated
362 print "reset %s" % ref
363 print "from :%u" % rev_to_mark(revid
)
366 marks
.set_tip(name
, revid
)
368 def export_tag(repo
, name
):
371 ref
= '%s/tags/%s' % (prefix
, name
)
372 print "reset %s" % ref
373 print "from :%u" % rev_to_mark(tags
[name
])
376 def do_import(parser
):
380 path
= os
.path
.join(dirname
, 'marks-git')
383 if os
.path
.exists(path
):
384 print "feature import-marks=%s" % path
385 print "feature export-marks=%s" % path
386 print "feature force"
389 while parser
.check('import'):
391 if ref
.startswith('refs/heads/'):
392 name
= ref
[len('refs/heads/'):]
393 export_branch(repo
, name
)
394 if ref
.startswith('refs/tags/'):
395 name
= ref
[len('refs/tags/'):]
396 export_tag(repo
, name
)
403 def parse_blob(parser
):
407 mark
= parser
.get_mark()
409 data
= parser
.get_data()
410 blob_marks
[mark
] = data
415 def __init__(self
, branch
, revid
, parents
, files
):
421 def copy_tree(revid
):
422 files
= files_cache
[revid
] = {}
424 tree
= branch
.repository
.revision_tree(revid
)
426 for path
, entry
in tree
.iter_entries_by_dir():
427 files
[path
] = [entry
.file_id
, None]
432 if len(parents
) == 0:
433 self
.base_id
= bzrlib
.revision
.NULL_REVISION
436 self
.base_id
= parents
[0]
437 self
.base_files
= files_cache
.get(self
.base_id
, None)
438 if not self
.base_files
:
439 self
.base_files
= copy_tree(self
.base_id
)
441 self
.files
= files_cache
[revid
] = self
.base_files
.copy()
444 for path
, data
in self
.files
.iteritems():
446 self
.rev_files
[fid
] = [path
, mark
]
448 for path
, f
in files
.iteritems():
449 fid
, mark
= self
.files
.get(path
, [None, None])
451 fid
= bzrlib
.generate_ids
.gen_file_id(path
)
453 self
.rev_files
[fid
] = [path
, mark
]
454 self
.updates
[fid
] = f
456 def last_revision(self
):
459 def iter_changes(self
):
462 def get_parent(dirname
, basename
):
463 parent_fid
, mark
= self
.base_files
.get(dirname
, [None, None])
466 parent_fid
, mark
= self
.files
.get(dirname
, [None, None])
471 fid
= bzrlib
.generate_ids
.gen_file_id(path
)
472 add_entry(fid
, dirname
, 'directory')
475 def add_entry(fid
, path
, kind
, mode
= None):
476 dirname
, basename
= os
.path
.split(path
)
477 parent_fid
= get_parent(dirname
, basename
)
482 elif mode
== '120000':
493 self
.files
[path
] = [change
[0], None]
494 changes
.append(change
)
496 def update_entry(fid
, path
, kind
, mode
= None):
497 dirname
, basename
= os
.path
.split(path
)
498 parent_fid
= get_parent(dirname
, basename
)
503 elif mode
== '120000':
514 self
.files
[path
] = [change
[0], None]
515 changes
.append(change
)
517 def remove_entry(fid
, path
, kind
):
518 dirname
, basename
= os
.path
.split(path
)
519 parent_fid
= get_parent(dirname
, basename
)
529 changes
.append(change
)
531 for fid
, f
in self
.updates
.iteritems():
535 remove_entry(fid
, path
, 'file')
538 if path
in self
.base_files
:
539 update_entry(fid
, path
, 'file', f
['mode'])
541 add_entry(fid
, path
, 'file', f
['mode'])
543 self
.files
[path
][1] = f
['mark']
544 self
.rev_files
[fid
][1] = f
['mark']
548 def get_content(self
, file_id
):
549 path
, mark
= self
.rev_files
[file_id
]
551 return blob_marks
[mark
]
554 tree
= self
.branch
.repository
.revision_tree(self
.base_id
)
555 return tree
.get_file_text(file_id
)
557 def get_file_with_stat(self
, file_id
, path
=None):
558 content
= self
.get_content(file_id
)
559 return (StringIO
.StringIO(content
), None)
561 def get_symlink_target(self
, file_id
):
562 return self
.get_content(file_id
)
564 def id2path(self
, file_id
):
565 path
, mark
= self
.rev_files
[file_id
]
568 def c_style_unescape(string
):
569 if string
[0] == string
[-1] == '"':
570 return string
.decode('string-escape')[1:-1]
573 def parse_commit(parser
):
574 global marks
, blob_marks
, parsed_refs
582 if ref
.startswith('refs/heads/'):
583 name
= ref
[len('refs/heads/'):]
584 branch
= branches
[name
]
588 commit_mark
= parser
.get_mark()
590 author
= parser
.get_author()
592 committer
= parser
.get_author()
594 data
= parser
.get_data()
596 if parser
.check('from'):
597 parents
.append(parser
.get_mark())
599 while parser
.check('merge'):
600 parents
.append(parser
.get_mark())
603 # fast-export adds an extra newline
610 if parser
.check('M'):
611 t
, m
, mark_ref
, path
= line
.split(' ', 3)
612 mark
= int(mark_ref
[1:])
613 f
= { 'mode' : m
, 'mark' : mark
}
614 elif parser
.check('D'):
615 t
, path
= line
.split(' ')
616 f
= { 'deleted' : True }
618 die('Unknown file command: %s' % line
)
619 path
= c_style_unescape(path
).decode('utf-8')
622 committer
, date
, tz
= committer
623 parents
= [str(mark_to_rev(p
)) for p
in parents
]
624 revid
= bzrlib
.generate_ids
.gen_revision_id(committer
, date
)
626 props
['branch-nick'] = branch
.nick
628 mtree
= CustomTree(branch
, revid
, parents
, files
)
629 changes
= mtree
.iter_changes()
633 builder
= branch
.get_commit_builder(parents
, None, date
, tz
, committer
, props
, revid
)
635 list(builder
.record_iter_changes(mtree
, mtree
.last_revision(), changes
))
636 builder
.finish_inventory()
637 builder
.commit(data
.decode('utf-8', 'replace'))
644 parsed_refs
[ref
] = revid
645 marks
.new_mark(revid
, commit_mark
)
647 def parse_reset(parser
):
654 if parser
.check('commit'):
657 if not parser
.check('from'):
659 from_mark
= parser
.get_mark()
662 parsed_refs
[ref
] = mark_to_rev(from_mark
)
664 def do_export(parser
):
665 global parsed_refs
, dirname
669 for line
in parser
.each_block('done'):
670 if parser
.check('blob'):
672 elif parser
.check('commit'):
674 elif parser
.check('reset'):
676 elif parser
.check('tag'):
678 elif parser
.check('feature'):
681 die('unhandled export command: %s' % line
)
683 for ref
, revid
in parsed_refs
.iteritems():
684 name
= ref
[len('refs/heads/'):]
685 branch
= branches
[name
]
686 branch
.generate_revision_history(revid
, marks
.get_tip(name
))
691 peer
.bzrdir
.push_branch(branch
, revision_id
=revid
)
692 except bzrlib
.errors
.DivergedBranches
:
693 print "error %s non-fast forward" % ref
697 wt
= branch
.bzrdir
.open_workingtree()
699 except bzrlib
.errors
.NoWorkingTree
:
706 def do_capabilities(parser
):
711 print "refspec refs/heads/*:%s/heads/*" % prefix
712 print "refspec refs/tags/*:%s/tags/*" % prefix
714 path
= os
.path
.join(dirname
, 'marks-git')
716 if os
.path
.exists(path
):
717 print "*import-marks %s" % path
718 print "*export-marks %s" % path
722 def ref_is_valid(name
):
723 return not True in [c
in name
for c
in '~^: \\']
730 for name
in branches
:
731 if not master_branch
:
733 print "? refs/heads/%s" % name
735 branch
= branches
[master_branch
]
737 for tag
, revid
in branch
.tags
.get_tag_dict().items():
739 branch
.revision_id_to_dotted_revno(revid
)
740 except bzrlib
.errors
.NoSuchRevision
:
742 if not ref_is_valid(tag
):
744 print "? refs/tags/%s" % tag
748 print "@refs/heads/%s HEAD" % master_branch
751 def get_remote_branch(origin
, remote_branch
, name
):
752 global dirname
, peers
754 branch_path
= os
.path
.join(dirname
, 'clone', name
)
755 if os
.path
.exists(branch_path
):
757 d
= bzrlib
.bzrdir
.BzrDir
.open(branch_path
)
758 branch
= d
.open_branch()
760 branch
.pull(remote_branch
, [], None, False)
761 except bzrlib
.errors
.DivergedBranches
:
762 # use remote branch for now
766 d
= origin
.sprout(branch_path
, None,
767 hardlink
=True, create_tree_if_local
=False,
768 force_new_repo
=False,
769 source_branch
=remote_branch
)
770 branch
= d
.open_branch()
774 def find_branches(repo
, wanted
):
775 transport
= repo
.user_transport
777 for fn
in transport
.iter_files_recursive():
778 if not fn
.endswith('.bzr/branch-format'):
781 name
= subdir
= fn
[:-len('/.bzr/branch-format')]
782 name
= name
if name
!= '' else 'master'
783 name
= name
.replace('/', '+')
785 if wanted
and not name
in wanted
:
789 cur
= transport
.clone(subdir
)
790 branch
= bzrlib
.branch
.Branch
.open_from_transport(cur
)
791 except bzrlib
.errors
.NotBranchError
:
796 def get_repo(url
, alias
):
797 global dirname
, peer
, branches
799 normal_url
= bzrlib
.urlutils
.normalize_url(url
)
800 origin
= bzrlib
.bzrdir
.BzrDir
.open(url
)
801 is_local
= isinstance(origin
.transport
, bzrlib
.transport
.local
.LocalTransport
)
803 shared_path
= os
.path
.join(gitdir
, 'bzr')
805 shared_dir
= bzrlib
.bzrdir
.BzrDir
.open(shared_path
)
806 except bzrlib
.errors
.NotBranchError
:
807 shared_dir
= bzrlib
.bzrdir
.BzrDir
.create(shared_path
)
809 shared_repo
= shared_dir
.open_repository()
810 except bzrlib
.errors
.NoRepositoryPresent
:
811 shared_repo
= shared_dir
.create_repository(shared
=True)
814 clone_path
= os
.path
.join(dirname
, 'clone')
815 if not os
.path
.exists(clone_path
):
819 repo
= origin
.open_repository()
820 except bzrlib
.errors
.NoRepositoryPresent
:
824 branch
= origin
.open_branch()
828 branches
[name
] = get_remote_branch(origin
, branch
, name
)
830 branches
[name
] = branch
832 return branch
.repository
836 wanted
= get_config('remote-bzr.branches').rstrip().split(', ')
838 wanted
= [e
for e
in wanted
if e
]
840 for name
, branch
in find_branches(repo
, wanted
):
844 branches
[name
] = get_remote_branch(origin
, branch
, name
)
846 branches
[name
] = branch
850 def fix_path(alias
, orig_url
):
851 url
= urlparse
.urlparse(orig_url
, 'file')
852 if url
.scheme
!= 'file' or os
.path
.isabs(url
.path
):
854 abs_url
= urlparse
.urljoin("%s/" % os
.getcwd(), orig_url
)
855 cmd
= ['git', 'config', 'remote.%s.url' % alias
, "bzr::%s" % abs_url
]
859 global marks
, prefix
, gitdir
, dirname
860 global tags
, filenodes
865 global branches
, peers
881 alias
= hashlib
.sha1(alias
).hexdigest()
885 prefix
= 'refs/bzr/%s' % alias
886 gitdir
= os
.environ
['GIT_DIR']
887 dirname
= os
.path
.join(gitdir
, 'bzr', alias
)
892 if not os
.path
.exists(dirname
):
895 bzrlib
.ui
.ui_factory
.be_quiet(True)
897 repo
= get_repo(url
, alias
)
899 marks_path
= os
.path
.join(dirname
, 'marks-int')
900 marks
= Marks(marks_path
)
902 parser
= Parser(repo
)
904 if parser
.check('capabilities'):
905 do_capabilities(parser
)
906 elif parser
.check('list'):
908 elif parser
.check('import'):
910 elif parser
.check('export'):
913 die('unhandled command: %s' % line
)
922 shutil
.rmtree(dirname
)
925 sys
.exit(main(sys
.argv
))