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 to track (per repo):
17 # % git config remote.origin.bzr-branches 'trunk, devel, test'
19 # Where 'origin' is the name of the repository you want to specify the
26 if hasattr(bzrlib, "initialize"):
30 bzrlib.plugin.load_plugins()
32 import bzrlib.generate_ids
33 import bzrlib.transport
36 import bzrlib.urlutils
44 import atexit, shutil, hashlib, urlparse, subprocess
46 NAME_RE = re.compile('^([^<>]+)')
47 AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
48 EMAIL_RE = re.compile('^([^<>]+[^ \\\t<>])?\\b(?:[ \\t<>]*?)\\b([^ \\t<>]+@[^ \\t<>]+)')
49 RAW_AUTHOR_RE = re.compile('^(\w+) (.+)? <(.*)> (\d+) ([+-]\d+)')
52 sys.stderr.write('ERROR: %s\n' % (msg % args))
56 sys.stderr.write('WARNING: %s\n' % (msg % args))
59 return '%+03d%02d' % (tz / 3600, tz % 3600 / 60)
61 def get_config(config):
62 cmd = ['git', 'config', '--get', config]
63 process = subprocess.Popen(cmd, stdout=subprocess.PIPE)
64 output, _ = process.communicate()
69 def __init__(self, path):
78 if not os.path.exists(self.path):
81 tmp = json.load(open(self.path))
82 self.tips = tmp['tips']
83 self.marks = tmp['marks']
84 self.last_mark = tmp['last-mark']
86 for rev, mark in self.marks.iteritems():
87 self.rev_marks[mark] = rev
90 return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
93 json.dump(self.dict(), open(self.path, 'w'))
96 return str(self.dict())
98 def from_rev(self, rev):
99 return self.marks[rev]
101 def to_rev(self, mark):
102 return str(self.rev_marks[mark])
106 return self.last_mark
108 def get_mark(self, rev):
110 self.marks[rev] = self.last_mark
111 return self.last_mark
113 def is_marked(self, rev):
114 return rev in self.marks
116 def new_mark(self, rev, mark):
117 self.marks[rev] = mark
118 self.rev_marks[mark] = rev
119 self.last_mark = mark
121 def get_tip(self, branch):
123 return str(self.tips[branch])
127 def set_tip(self, branch, tip):
128 self.tips[branch] = tip
132 def __init__(self, repo):
134 self.line = self.get_line()
137 return sys.stdin.readline().strip()
139 def __getitem__(self, i):
140 return self.line.split()[i]
142 def check(self, word):
143 return self.line.startswith(word)
145 def each_block(self, separator):
146 while self.line != separator:
148 self.line = self.get_line()
151 return self.each_block('')
154 self.line = self.get_line()
155 if self.line == 'done':
159 i = self.line.index(':') + 1
160 return int(self.line[i:])
163 if not self.check('data'):
165 i = self.line.index(' ') + 1
166 size = int(self.line[i:])
167 return sys.stdin.read(size)
169 def get_author(self):
170 m = RAW_AUTHOR_RE.match(self.line)
173 _, name, email, date, tz = m.groups()
174 name = name.decode('utf-8')
175 committer = '%s <%s>' % (name, email)
177 tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
178 return (committer, int(date), tz)
180 def rev_to_mark(rev):
181 return marks.from_rev(rev)
183 def mark_to_rev(mark):
184 return marks.to_rev(mark)
186 def fixup_user(user):
188 user = user.replace('"', '')
189 m = AUTHOR_RE.match(user)
192 mail = m.group(2).strip()
194 m = EMAIL_RE.match(user)
199 m = NAME_RE.match(user)
201 name = m.group(1).strip()
208 return '%s <%s>' % (name, mail)
210 def get_filechanges(cur, prev):
214 changes = cur.changes_from(prev)
217 return s.encode('utf-8')
219 for path, fid, kind in changes.added:
220 modified[u(path)] = fid
221 for path, fid, kind in changes.removed:
222 removed[u(path)] = None
223 for path, fid, kind, mod, _ in changes.modified:
224 modified[u(path)] = fid
225 for oldpath, newpath, fid, kind, mod, _ in changes.renamed:
226 removed[u(oldpath)] = None
227 if kind == 'directory':
228 lst = cur.list_files(from_dir=newpath, recursive=True)
229 for path, file_class, kind, fid, entry in lst:
230 if kind != 'directory':
231 modified[u(newpath + '/' + path)] = fid
233 modified[u(newpath)] = fid
235 return modified, removed
237 def export_files(tree, files):
239 for path, fid in files.iteritems():
240 kind = tree.kind(fid)
242 h = tree.get_file_sha1(fid)
244 if kind == 'symlink':
245 d = tree.get_symlink_target(fid)
249 if tree.is_executable(fid):
254 # is the blob already exported?
257 final.append((mode, mark, path))
260 d = tree.get_file_text(fid)
261 elif kind == 'directory':
264 die("Unhandled kind '%s' for path '%s'" % (kind, path))
266 mark = marks.next_mark()
270 print "mark :%u" % mark
271 print "data %d" % len(d)
274 final.append((mode, mark, path))
278 def export_branch(repo, name):
279 ref = '%s/heads/%s' % (prefix, name)
280 tip = marks.get_tip(name)
282 branch = get_remote_branch(name)
283 repo = branch.repository
286 revs = branch.iter_merge_sorted_revisions(None, tip, 'exclude', 'forward')
288 tip_revno = branch.revision_id_to_revno(tip)
289 last_revno, _ = branch.last_revision_info()
290 total = last_revno - tip_revno
291 except bzrlib.errors.NoSuchRevision:
295 for revid, _, seq, _ in revs:
297 if marks.is_marked(revid):
300 rev = repo.get_revision(revid)
303 parents = rev.parent_ids
306 committer = rev.committer.encode('utf-8')
307 committer = "%s %u %s" % (fixup_user(committer), time, gittz(tz))
308 authors = rev.get_apparent_authors()
310 author = authors[0].encode('utf-8')
311 author = "%s %u %s" % (fixup_user(author), time, gittz(tz))
314 msg = rev.message.encode('utf-8')
318 if len(parents) == 0:
319 parent = bzrlib.revision.NULL_REVISION
323 cur_tree = repo.revision_tree(revid)
324 prev = repo.revision_tree(parent)
325 modified, removed = get_filechanges(cur_tree, prev)
327 modified_final = export_files(cur_tree, modified)
329 if len(parents) == 0:
330 print 'reset %s' % ref
332 print "commit %s" % ref
333 print "mark :%d" % (marks.get_mark(revid))
334 print "author %s" % (author)
335 print "committer %s" % (committer)
336 print "data %d" % (len(msg))
339 for i, p in enumerate(parents):
348 print "merge :%s" % m
352 for f in modified_final:
353 print "M %s :%u %s" % f
357 # let's skip branch revisions from the progress report
360 progress = (revno - tip_revno)
361 if (progress % 100 == 0):
363 print "progress revision %d '%s' (%d/%d)" % (revno, name, progress, total)
365 print "progress revision %d '%s' (%d)" % (revno, name, progress)
369 revid = branch.last_revision()
371 # make sure the ref is updated
372 print "reset %s" % ref
373 print "from :%u" % rev_to_mark(revid)
376 marks.set_tip(name, revid)
378 def export_tag(repo, name):
379 ref = '%s/tags/%s' % (prefix, name)
380 print "reset %s" % ref
381 print "from :%u" % rev_to_mark(tags[name])
384 def do_import(parser):
386 path = os.path.join(dirname, 'marks-git')
389 if os.path.exists(path):
390 print "feature import-marks=%s" % path
391 print "feature export-marks=%s" % path
392 print "feature force"
395 while parser.check('import'):
397 if ref.startswith('refs/heads/'):
398 name = ref[len('refs/heads/'):]
399 export_branch(repo, name)
400 if ref.startswith('refs/tags/'):
401 name = ref[len('refs/tags/'):]
402 export_tag(repo, name)
409 def parse_blob(parser):
411 mark = parser.get_mark()
413 data = parser.get_data()
414 blob_marks[mark] = data
419 def __init__(self, branch, revid, parents, files):
423 def copy_tree(revid):
424 files = files_cache[revid] = {}
426 tree = branch.repository.revision_tree(revid)
428 for path, entry in tree.iter_entries_by_dir():
429 files[path] = [entry.file_id, None]
434 if len(parents) == 0:
435 self.base_id = bzrlib.revision.NULL_REVISION
438 self.base_id = parents[0]
439 self.base_files = files_cache.get(self.base_id, None)
440 if not self.base_files:
441 self.base_files = copy_tree(self.base_id)
443 self.files = files_cache[revid] = self.base_files.copy()
446 for path, data in self.files.iteritems():
448 self.rev_files[fid] = [path, mark]
450 for path, f in files.iteritems():
451 fid, mark = self.files.get(path, [None, None])
453 fid = bzrlib.generate_ids.gen_file_id(path)
455 self.rev_files[fid] = [path, mark]
456 self.updates[fid] = f
458 def last_revision(self):
461 def iter_changes(self):
464 def get_parent(dirname, basename):
465 parent_fid, mark = self.base_files.get(dirname, [None, None])
468 parent_fid, mark = self.files.get(dirname, [None, None])
473 fid = bzrlib.generate_ids.gen_file_id(path)
474 add_entry(fid, dirname, 'directory')
477 def add_entry(fid, path, kind, mode=None):
478 dirname, basename = os.path.split(path)
479 parent_fid = get_parent(dirname, basename)
484 elif mode == '120000':
495 self.files[path] = [change[0], None]
496 changes.append(change)
498 def update_entry(fid, path, kind, mode=None):
499 dirname, basename = os.path.split(path)
500 parent_fid = get_parent(dirname, basename)
505 elif mode == '120000':
516 self.files[path] = [change[0], None]
517 changes.append(change)
519 def remove_entry(fid, path, kind):
520 dirname, basename = os.path.split(path)
521 parent_fid = get_parent(dirname, basename)
531 changes.append(change)
533 for fid, f in self.updates.iteritems():
537 remove_entry(fid, path, 'file')
540 if path in self.base_files:
541 update_entry(fid, path, 'file', f['mode'])
543 add_entry(fid, path, 'file', f['mode'])
545 self.files[path][1] = f['mark']
546 self.rev_files[fid][1] = f['mark']
550 def get_content(self, file_id):
551 path, mark = self.rev_files[file_id]
553 return blob_marks[mark]
556 tree = self.branch.repository.revision_tree(self.base_id)
557 return tree.get_file_text(file_id)
559 def get_file_with_stat(self, file_id, path=None):
560 content = self.get_content(file_id)
561 return (StringIO.StringIO(content), None)
563 def get_symlink_target(self, file_id):
564 return self.get_content(file_id)
566 def id2path(self, file_id):
567 path, mark = self.rev_files[file_id]
570 def c_style_unescape(string):
571 if string[0] == string[-1] == '"':
572 return string.decode('string-escape')[1:-1]
575 def parse_commit(parser):
581 if ref.startswith('refs/heads/'):
582 name = ref[len('refs/heads/'):]
583 branch = get_remote_branch(name)
587 commit_mark = parser.get_mark()
589 author = parser.get_author()
591 committer = parser.get_author()
593 data = parser.get_data()
595 if parser.check('from'):
596 parents.append(parser.get_mark())
598 while parser.check('merge'):
599 parents.append(parser.get_mark())
602 # fast-export adds an extra newline
609 if parser.check('M'):
610 t, m, mark_ref, path = line.split(' ', 3)
611 mark = int(mark_ref[1:])
612 f = { 'mode' : m, 'mark' : mark }
613 elif parser.check('D'):
614 t, path = line.split(' ', 1)
615 f = { 'deleted' : True }
617 die('Unknown file command: %s' % line)
618 path = c_style_unescape(path).decode('utf-8')
621 committer, date, tz = committer
622 parents = [mark_to_rev(p) for p in parents]
623 revid = bzrlib.generate_ids.gen_revision_id(committer, date)
625 props['branch-nick'] = branch.nick
627 mtree = CustomTree(branch, revid, parents, files)
628 changes = mtree.iter_changes()
632 builder = branch.get_commit_builder(parents, None, date, tz, committer, props, revid)
634 list(builder.record_iter_changes(mtree, mtree.last_revision(), changes))
635 builder.finish_inventory()
636 builder.commit(data.decode('utf-8', 'replace'))
643 parsed_refs[ref] = revid
644 marks.new_mark(revid, commit_mark)
646 def parse_reset(parser):
651 if parser.check('commit'):
654 if not parser.check('from'):
656 from_mark = parser.get_mark()
659 parsed_refs[ref] = mark_to_rev(from_mark)
661 def do_export(parser):
664 for line in parser.each_block('done'):
665 if parser.check('blob'):
667 elif parser.check('commit'):
669 elif parser.check('reset'):
671 elif parser.check('tag'):
673 elif parser.check('feature'):
676 die('unhandled export command: %s' % line)
678 for ref, revid in parsed_refs.iteritems():
679 if ref.startswith('refs/heads/'):
680 name = ref[len('refs/heads/'):]
681 branch = get_remote_branch(name)
682 branch.generate_revision_history(revid, marks.get_tip(name))
685 peer = bzrlib.branch.Branch.open(peers[name],
686 possible_transports=transports)
688 peer.bzrdir.push_branch(branch, revision_id=revid)
689 except bzrlib.errors.DivergedBranches:
690 print "error %s non-fast forward" % ref
694 wt = branch.bzrdir.open_workingtree()
696 except bzrlib.errors.NoWorkingTree:
698 elif ref.startswith('refs/tags/'):
699 # TODO: implement tag push
700 print "error %s pushing tags not supported" % ref
703 # transport-helper/fast-export bugs
710 def do_capabilities(parser):
713 print "refspec refs/heads/*:%s/heads/*" % prefix
714 print "refspec refs/tags/*:%s/tags/*" % prefix
716 path = os.path.join(dirname, 'marks-git')
718 if os.path.exists(path):
719 print "*import-marks %s" % path
720 print "*export-marks %s" % path
724 def ref_is_valid(name):
725 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 = get_remote_branch(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 clone(path, remote_branch):
753 bdir = bzrlib.bzrdir.BzrDir.create(path, possible_transports=transports)
754 except bzrlib.errors.AlreadyControlDirError:
755 bdir = bzrlib.bzrdir.BzrDir.open(path, possible_transports=transports)
756 repo = bdir.find_repository()
757 repo.fetch(remote_branch.repository)
758 return remote_branch.sprout(bdir, repository=repo)
760 def get_remote_branch(name):
761 remote_branch = bzrlib.branch.Branch.open(branches[name],
762 possible_transports=transports)
763 if isinstance(remote_branch.user_transport, bzrlib.transport.local.LocalTransport):
766 branch_path = os.path.join(dirname, 'clone', name)
769 branch = bzrlib.branch.Branch.open(branch_path,
770 possible_transports=transports)
771 except bzrlib.errors.NotBranchError:
773 branch = clone(branch_path, remote_branch)
777 branch.pull(remote_branch, overwrite=True)
778 except bzrlib.errors.DivergedBranches:
779 # use remote branch for now
784 def find_branches(repo):
785 transport = repo.bzrdir.root_transport
787 for fn in transport.iter_files_recursive():
788 if not fn.endswith('.bzr/branch-format'):
791 name = subdir = fn[:-len('/.bzr/branch-format')]
792 name = name if name != '' else 'master'
793 name = name.replace('/', '+')
796 cur = transport.clone(subdir)
797 branch = bzrlib.branch.Branch.open_from_transport(cur)
798 except bzrlib.errors.NotBranchError:
801 yield name, branch.base
803 def get_repo(url, alias):
804 normal_url = bzrlib.urlutils.normalize_url(url)
805 origin = bzrlib.bzrdir.BzrDir.open(url, possible_transports=transports)
806 is_local = isinstance(origin.transport, bzrlib.transport.local.LocalTransport)
808 shared_path = os.path.join(gitdir, 'bzr')
810 shared_dir = bzrlib.bzrdir.BzrDir.open(shared_path,
811 possible_transports=transports)
812 except bzrlib.errors.NotBranchError:
813 shared_dir = bzrlib.bzrdir.BzrDir.create(shared_path,
814 possible_transports=transports)
816 shared_repo = shared_dir.open_repository()
817 except bzrlib.errors.NoRepositoryPresent:
818 shared_repo = shared_dir.create_repository(shared=True)
821 clone_path = os.path.join(dirname, 'clone')
822 if not os.path.exists(clone_path):
825 # check and remove old organization
827 bdir = bzrlib.bzrdir.BzrDir.open(clone_path,
828 possible_transports=transports)
829 bdir.destroy_repository()
830 except bzrlib.errors.NotBranchError:
832 except bzrlib.errors.NoRepositoryPresent:
835 wanted = get_config('remote.%s.bzr-branches' % alias).rstrip().split(', ')
837 wanted = [e for e in wanted if e]
839 wanted = get_config('remote-bzr.branches').rstrip().split(', ')
841 wanted = [e for e in wanted if e]
845 repo = origin.open_repository()
846 if not repo.user_transport.listable():
847 # this repository is not usable for us
848 raise bzrlib.errors.NoRepositoryPresent(repo.bzrdir)
849 except bzrlib.errors.NoRepositoryPresent:
853 def list_wanted(url, wanted):
855 subdir = name if name != 'master' else ''
856 yield name, bzrlib.urlutils.join(url, subdir)
858 branch_list = list_wanted(url, wanted)
860 branch_list = find_branches(repo)
862 for name, url in branch_list:
869 def fix_path(alias, orig_url):
870 url = urlparse.urlparse(orig_url, 'file')
871 if url.scheme != 'file' or os.path.isabs(url.path):
873 abs_url = urlparse.urljoin("%s/" % os.getcwd(), orig_url)
874 cmd = ['git', 'config', 'remote.%s.url' % alias, "bzr::%s" % abs_url]
878 global marks, prefix, gitdir, dirname
879 global tags, filenodes
884 global branches, peers
902 alias = hashlib.sha1(alias).hexdigest()
906 prefix = 'refs/bzr/%s' % alias
907 gitdir = os.environ['GIT_DIR']
908 dirname = os.path.join(gitdir, 'bzr', alias)
913 if not os.path.exists(dirname):
916 if hasattr(bzrlib.ui.ui_factory, 'be_quiet'):
917 bzrlib.ui.ui_factory.be_quiet(True)
919 repo = get_repo(url, alias)
921 marks_path = os.path.join(dirname, 'marks-int')
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))