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
41 import atexit, shutil, hashlib, urlparse, subprocess
43 NAME_RE = re.compile('^([^<>]+)')
44 AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
45 EMAIL_RE = re.compile('^([^<>]+[^ \\\t<>])?\\b(?:[ \\t<>]*?)\\b([^ \\t<>]+@[^ \\t<>]+)')
46 RAW_AUTHOR_RE = re.compile('^(\w+) (.+)? <(.*)> (\d+) ([+-]\d+)')
49 sys.stderr.write('ERROR: %s\n' % (msg % args))
53 sys.stderr.write('WARNING: %s\n' % (msg % args))
56 return '%+03d%02d' % (tz / 3600, tz % 3600 / 60)
58 def get_config(config):
59 cmd = ['git', 'config', '--get', config]
60 process = subprocess.Popen(cmd, stdout=subprocess.PIPE)
61 output, _ = process.communicate()
66 def __init__(self, path):
75 if not os.path.exists(self.path):
78 tmp = json.load(open(self.path))
79 self.tips = tmp['tips']
80 self.marks = tmp['marks']
81 self.last_mark = tmp['last-mark']
83 for rev, mark in self.marks.iteritems():
84 self.rev_marks[mark] = rev
87 return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
90 json.dump(self.dict(), open(self.path, 'w'))
93 return str(self.dict())
95 def from_rev(self, rev):
96 return self.marks[rev]
98 def to_rev(self, mark):
99 return str(self.rev_marks[mark])
103 return self.last_mark
105 def get_mark(self, rev):
107 self.marks[rev] = self.last_mark
108 return self.last_mark
110 def is_marked(self, rev):
111 return rev in self.marks
113 def new_mark(self, rev, mark):
114 self.marks[rev] = mark
115 self.rev_marks[mark] = rev
116 self.last_mark = mark
118 def get_tip(self, branch):
120 return str(self.tips[branch])
124 def set_tip(self, branch, tip):
125 self.tips[branch] = tip
129 def __init__(self, repo):
131 self.line = self.get_line()
134 return sys.stdin.readline().strip()
136 def __getitem__(self, i):
137 return self.line.split()[i]
139 def check(self, word):
140 return self.line.startswith(word)
142 def each_block(self, separator):
143 while self.line != separator:
145 self.line = self.get_line()
148 return self.each_block('')
151 self.line = self.get_line()
152 if self.line == 'done':
156 i = self.line.index(':') + 1
157 return int(self.line[i:])
160 if not self.check('data'):
162 i = self.line.index(' ') + 1
163 size = int(self.line[i:])
164 return sys.stdin.read(size)
166 def get_author(self):
167 m = RAW_AUTHOR_RE.match(self.line)
170 _, name, email, date, tz = m.groups()
171 committer = '%s <%s>' % (name, email)
173 tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
174 return (committer, int(date), tz)
176 def rev_to_mark(rev):
178 return marks.from_rev(rev)
180 def mark_to_rev(mark):
182 return marks.to_rev(mark)
184 def fixup_user(user):
186 user = user.replace('"', '')
187 m = AUTHOR_RE.match(user)
190 mail = m.group(2).strip()
192 m = EMAIL_RE.match(user)
197 m = NAME_RE.match(user)
199 name = m.group(1).strip()
206 return '%s <%s>' % (name, mail)
208 def get_filechanges(cur, prev):
212 changes = cur.changes_from(prev)
215 return s.encode('utf-8')
217 for path, fid, kind in changes.added:
218 modified[u(path)] = fid
219 for path, fid, kind in changes.removed:
220 removed[u(path)] = None
221 for path, fid, kind, mod, _ in changes.modified:
222 modified[u(path)] = fid
223 for oldpath, newpath, fid, kind, mod, _ in changes.renamed:
224 removed[u(oldpath)] = None
225 if kind == 'directory':
226 lst = cur.list_files(from_dir=newpath, recursive=True)
227 for path, file_class, kind, fid, entry in lst:
228 if kind != 'directory':
229 modified[u(newpath + '/' + path)] = fid
231 modified[u(newpath)] = fid
233 return modified, removed
235 def export_files(tree, files):
236 global marks, filenodes
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):
281 ref = '%s/heads/%s' % (prefix, name)
282 tip = marks.get_tip(name)
284 branch = get_remote_branch(name)
285 repo = branch.repository
288 revs = branch.iter_merge_sorted_revisions(None, tip, 'exclude', 'forward')
290 tip_revno = branch.revision_id_to_revno(tip)
291 last_revno, _ = branch.last_revision_info()
292 total = last_revno - tip_revno
293 except bzrlib.errors.NoSuchRevision:
297 for revid, _, seq, _ in revs:
299 if marks.is_marked(revid):
302 rev = repo.get_revision(revid)
305 parents = rev.parent_ids
308 committer = rev.committer.encode('utf-8')
309 committer = "%s %u %s" % (fixup_user(committer), time, gittz(tz))
310 authors = rev.get_apparent_authors()
312 author = authors[0].encode('utf-8')
313 author = "%s %u %s" % (fixup_user(author), time, gittz(tz))
316 msg = rev.message.encode('utf-8')
320 if len(parents) == 0:
321 parent = bzrlib.revision.NULL_REVISION
325 cur_tree = repo.revision_tree(revid)
326 prev = repo.revision_tree(parent)
327 modified, removed = get_filechanges(cur_tree, prev)
329 modified_final = export_files(cur_tree, modified)
331 if len(parents) == 0:
332 print 'reset %s' % ref
334 print "commit %s" % ref
335 print "mark :%d" % (marks.get_mark(revid))
336 print "author %s" % (author)
337 print "committer %s" % (committer)
338 print "data %d" % (len(msg))
341 for i, p in enumerate(parents):
350 print "merge :%s" % m
354 for f in modified_final:
355 print "M %s :%u %s" % f
359 # let's skip branch revisions from the progress report
362 progress = (revno - tip_revno)
363 if (progress % 100 == 0):
365 print "progress revision %d '%s' (%d/%d)" % (revno, name, progress, total)
367 print "progress revision %d '%s' (%d)" % (revno, name, progress)
371 revid = branch.last_revision()
373 # make sure the ref is updated
374 print "reset %s" % ref
375 print "from :%u" % rev_to_mark(revid)
378 marks.set_tip(name, revid)
380 def export_tag(repo, name):
383 ref = '%s/tags/%s' % (prefix, name)
384 print "reset %s" % ref
385 print "from :%u" % rev_to_mark(tags[name])
388 def do_import(parser):
392 path = os.path.join(dirname, 'marks-git')
395 if os.path.exists(path):
396 print "feature import-marks=%s" % path
397 print "feature export-marks=%s" % path
398 print "feature force"
401 while parser.check('import'):
403 if ref.startswith('refs/heads/'):
404 name = ref[len('refs/heads/'):]
405 export_branch(repo, name)
406 if ref.startswith('refs/tags/'):
407 name = ref[len('refs/tags/'):]
408 export_tag(repo, name)
415 def parse_blob(parser):
419 mark = parser.get_mark()
421 data = parser.get_data()
422 blob_marks[mark] = data
427 def __init__(self, branch, revid, parents, files):
433 def copy_tree(revid):
434 files = files_cache[revid] = {}
436 tree = branch.repository.revision_tree(revid)
438 for path, entry in tree.iter_entries_by_dir():
439 files[path] = [entry.file_id, None]
444 if len(parents) == 0:
445 self.base_id = bzrlib.revision.NULL_REVISION
448 self.base_id = parents[0]
449 self.base_files = files_cache.get(self.base_id, None)
450 if not self.base_files:
451 self.base_files = copy_tree(self.base_id)
453 self.files = files_cache[revid] = self.base_files.copy()
456 for path, data in self.files.iteritems():
458 self.rev_files[fid] = [path, mark]
460 for path, f in files.iteritems():
461 fid, mark = self.files.get(path, [None, None])
463 fid = bzrlib.generate_ids.gen_file_id(path)
465 self.rev_files[fid] = [path, mark]
466 self.updates[fid] = f
468 def last_revision(self):
471 def iter_changes(self):
474 def get_parent(dirname, basename):
475 parent_fid, mark = self.base_files.get(dirname, [None, None])
478 parent_fid, mark = self.files.get(dirname, [None, None])
483 fid = bzrlib.generate_ids.gen_file_id(path)
484 add_entry(fid, dirname, 'directory')
487 def add_entry(fid, path, kind, mode = None):
488 dirname, basename = os.path.split(path)
489 parent_fid = get_parent(dirname, basename)
494 elif mode == '120000':
505 self.files[path] = [change[0], None]
506 changes.append(change)
508 def update_entry(fid, path, kind, mode = None):
509 dirname, basename = os.path.split(path)
510 parent_fid = get_parent(dirname, basename)
515 elif mode == '120000':
526 self.files[path] = [change[0], None]
527 changes.append(change)
529 def remove_entry(fid, path, kind):
530 dirname, basename = os.path.split(path)
531 parent_fid = get_parent(dirname, basename)
541 changes.append(change)
543 for fid, f in self.updates.iteritems():
547 remove_entry(fid, path, 'file')
550 if path in self.base_files:
551 update_entry(fid, path, 'file', f['mode'])
553 add_entry(fid, path, 'file', f['mode'])
555 self.files[path][1] = f['mark']
556 self.rev_files[fid][1] = f['mark']
560 def get_content(self, file_id):
561 path, mark = self.rev_files[file_id]
563 return blob_marks[mark]
566 tree = self.branch.repository.revision_tree(self.base_id)
567 return tree.get_file_text(file_id)
569 def get_file_with_stat(self, file_id, path=None):
570 content = self.get_content(file_id)
571 return (StringIO.StringIO(content), None)
573 def get_symlink_target(self, file_id):
574 return self.get_content(file_id)
576 def id2path(self, file_id):
577 path, mark = self.rev_files[file_id]
580 def c_style_unescape(string):
581 if string[0] == string[-1] == '"':
582 return string.decode('string-escape')[1:-1]
585 def parse_commit(parser):
586 global marks, blob_marks, parsed_refs
594 if ref.startswith('refs/heads/'):
595 name = ref[len('refs/heads/'):]
596 branch = get_remote_branch(name)
600 commit_mark = parser.get_mark()
602 author = parser.get_author()
604 committer = parser.get_author()
606 data = parser.get_data()
608 if parser.check('from'):
609 parents.append(parser.get_mark())
611 while parser.check('merge'):
612 parents.append(parser.get_mark())
615 # fast-export adds an extra newline
622 if parser.check('M'):
623 t, m, mark_ref, path = line.split(' ', 3)
624 mark = int(mark_ref[1:])
625 f = { 'mode' : m, 'mark' : mark }
626 elif parser.check('D'):
627 t, path = line.split(' ', 1)
628 f = { 'deleted' : True }
630 die('Unknown file command: %s' % line)
631 path = c_style_unescape(path).decode('utf-8')
634 committer, date, tz = committer
635 parents = [mark_to_rev(p) for p in parents]
636 revid = bzrlib.generate_ids.gen_revision_id(committer, date)
638 props['branch-nick'] = branch.nick
640 mtree = CustomTree(branch, revid, parents, files)
641 changes = mtree.iter_changes()
645 builder = branch.get_commit_builder(parents, None, date, tz, committer, props, revid)
647 list(builder.record_iter_changes(mtree, mtree.last_revision(), changes))
648 builder.finish_inventory()
649 builder.commit(data.decode('utf-8', 'replace'))
656 parsed_refs[ref] = revid
657 marks.new_mark(revid, commit_mark)
659 def parse_reset(parser):
666 if parser.check('commit'):
669 if not parser.check('from'):
671 from_mark = parser.get_mark()
674 parsed_refs[ref] = mark_to_rev(from_mark)
676 def do_export(parser):
677 global parsed_refs, dirname, transports
681 for line in parser.each_block('done'):
682 if parser.check('blob'):
684 elif parser.check('commit'):
686 elif parser.check('reset'):
688 elif parser.check('tag'):
690 elif parser.check('feature'):
693 die('unhandled export command: %s' % line)
695 for ref, revid in parsed_refs.iteritems():
696 if ref.startswith('refs/heads/'):
697 name = ref[len('refs/heads/'):]
698 branch = get_remote_branch(name)
699 branch.generate_revision_history(revid, marks.get_tip(name))
702 peer = bzrlib.branch.Branch.open(peers[name],
703 possible_transports=transports)
705 peer.bzrdir.push_branch(branch, revision_id=revid)
706 except bzrlib.errors.DivergedBranches:
707 print "error %s non-fast forward" % ref
711 wt = branch.bzrdir.open_workingtree()
713 except bzrlib.errors.NoWorkingTree:
715 elif ref.startswith('refs/tags/'):
716 # TODO: implement tag push
717 print "error %s pushing tags not supported" % ref
720 # transport-helper/fast-export bugs
727 def do_capabilities(parser):
732 print "refspec refs/heads/*:%s/heads/*" % prefix
733 print "refspec refs/tags/*:%s/tags/*" % prefix
735 path = os.path.join(dirname, 'marks-git')
737 if os.path.exists(path):
738 print "*import-marks %s" % path
739 print "*export-marks %s" % path
743 def ref_is_valid(name):
744 return not True in [c in name for c in '~^: \\']
751 for name in branches:
752 if not master_branch:
754 print "? refs/heads/%s" % name
756 branch = get_remote_branch(master_branch)
758 for tag, revid in branch.tags.get_tag_dict().items():
760 branch.revision_id_to_dotted_revno(revid)
761 except bzrlib.errors.NoSuchRevision:
763 if not ref_is_valid(tag):
765 print "? refs/tags/%s" % tag
769 print "@refs/heads/%s HEAD" % master_branch
772 def clone(path, remote_branch):
775 bdir = bzrlib.bzrdir.BzrDir.create(path, possible_transports=transports)
776 except bzrlib.errors.AlreadyControlDirError:
777 bdir = bzrlib.bzrdir.BzrDir.open(path, possible_transports=transports)
778 repo = bdir.find_repository()
779 repo.fetch(remote_branch.repository)
780 return remote_branch.sprout(bdir, repository=repo)
782 def get_remote_branch(name):
783 global dirname, branches, transports
785 remote_branch = bzrlib.branch.Branch.open(branches[name],
786 possible_transports=transports)
787 if isinstance(remote_branch.user_transport, bzrlib.transport.local.LocalTransport):
790 branch_path = os.path.join(dirname, 'clone', name)
793 branch = bzrlib.branch.Branch.open(branch_path,
794 possible_transports=transports)
795 except bzrlib.errors.NotBranchError:
797 branch = clone(branch_path, remote_branch)
801 branch.pull(remote_branch, overwrite=True)
802 except bzrlib.errors.DivergedBranches:
803 # use remote branch for now
808 def find_branches(repo):
809 transport = repo.bzrdir.root_transport
811 for fn in transport.iter_files_recursive():
812 if not fn.endswith('.bzr/branch-format'):
815 name = subdir = fn[:-len('/.bzr/branch-format')]
816 name = name if name != '' else 'master'
817 name = name.replace('/', '+')
820 cur = transport.clone(subdir)
821 branch = bzrlib.branch.Branch.open_from_transport(cur)
822 except bzrlib.errors.NotBranchError:
825 yield name, branch.base
827 def get_repo(url, alias):
828 global dirname, peer, branches, transports
830 normal_url = bzrlib.urlutils.normalize_url(url)
831 origin = bzrlib.bzrdir.BzrDir.open(url, possible_transports=transports)
832 is_local = isinstance(origin.transport, bzrlib.transport.local.LocalTransport)
834 shared_path = os.path.join(gitdir, 'bzr')
836 shared_dir = bzrlib.bzrdir.BzrDir.open(shared_path,
837 possible_transports=transports)
838 except bzrlib.errors.NotBranchError:
839 shared_dir = bzrlib.bzrdir.BzrDir.create(shared_path,
840 possible_transports=transports)
842 shared_repo = shared_dir.open_repository()
843 except bzrlib.errors.NoRepositoryPresent:
844 shared_repo = shared_dir.create_repository(shared=True)
847 clone_path = os.path.join(dirname, 'clone')
848 if not os.path.exists(clone_path):
851 # check and remove old organization
853 bdir = bzrlib.bzrdir.BzrDir.open(clone_path,
854 possible_transports=transports)
855 bdir.destroy_repository()
856 except bzrlib.errors.NotBranchError:
858 except bzrlib.errors.NoRepositoryPresent:
861 wanted = get_config('remote-bzr.branches').rstrip().split(', ')
863 wanted = [e for e in wanted if e]
867 repo = origin.open_repository()
868 if not repo.user_transport.listable():
869 # this repository is not usable for us
870 raise bzrlib.errors.NoRepositoryPresent(repo.bzrdir)
871 except bzrlib.errors.NoRepositoryPresent:
875 def list_wanted(url, wanted):
877 subdir = name if name != 'master' else ''
878 yield name, bzrlib.urlutils.join(url, subdir)
880 branch_list = list_wanted(url, wanted)
882 branch_list = find_branches(repo)
884 for name, url in branch_list:
891 def fix_path(alias, orig_url):
892 url = urlparse.urlparse(orig_url, 'file')
893 if url.scheme != 'file' or os.path.isabs(url.path):
895 abs_url = urlparse.urljoin("%s/" % os.getcwd(), orig_url)
896 cmd = ['git', 'config', 'remote.%s.url' % alias, "bzr::%s" % abs_url]
900 global marks, prefix, gitdir, dirname
901 global tags, filenodes
906 global branches, peers
924 alias = hashlib.sha1(alias).hexdigest()
928 prefix = 'refs/bzr/%s' % alias
929 gitdir = os.environ['GIT_DIR']
930 dirname = os.path.join(gitdir, 'bzr', alias)
935 if not os.path.exists(dirname):
938 if hasattr(bzrlib.ui.ui_factory, 'be_quiet'):
939 bzrlib.ui.ui_factory.be_quiet(True)
941 repo = get_repo(url, alias)
943 marks_path = os.path.join(dirname, 'marks-int')
944 marks = Marks(marks_path)
946 parser = Parser(repo)
948 if parser.check('capabilities'):
949 do_capabilities(parser)
950 elif parser.check('list'):
952 elif parser.check('import'):
954 elif parser.check('export'):
957 die('unhandled command: %s' % line)
966 shutil.rmtree(dirname)
969 sys.exit(main(sys.argv))