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 for revid, _, seq, _ in revs:
291 if marks.is_marked(revid):
294 rev = repo.get_revision(revid)
297 parents = rev.parent_ids
300 committer = rev.committer.encode('utf-8')
301 committer = "%s %u %s" % (fixup_user(committer), time, gittz(tz))
302 authors = rev.get_apparent_authors()
304 author = authors[0].encode('utf-8')
305 author = "%s %u %s" % (fixup_user(author), time, gittz(tz))
308 msg = rev.message.encode('utf-8')
312 if len(parents) == 0:
313 parent = bzrlib.revision.NULL_REVISION
317 cur_tree = repo.revision_tree(revid)
318 prev = repo.revision_tree(parent)
319 modified, removed = get_filechanges(cur_tree, prev)
321 modified_final = export_files(cur_tree, modified)
323 if len(parents) == 0:
324 print 'reset %s' % ref
326 print "commit %s" % ref
327 print "mark :%d" % (marks.get_mark(revid))
328 print "author %s" % (author)
329 print "committer %s" % (committer)
330 print "data %d" % (len(msg))
333 for i, p in enumerate(parents):
342 print "merge :%s" % m
346 for f in modified_final:
347 print "M %s :%u %s" % f
351 # let's skip branch revisions from the progress report
354 progress = (revno - tip_revno)
355 if (progress % 100 == 0):
356 print "progress revision %d '%s' (%d/%d)" % (revno, name, progress, total)
360 revid = branch.last_revision()
362 # make sure the ref is updated
363 print "reset %s" % ref
364 print "from :%u" % rev_to_mark(revid)
367 marks.set_tip(name, revid)
369 def export_tag(repo, name):
372 ref = '%s/tags/%s' % (prefix, name)
373 print "reset %s" % ref
374 print "from :%u" % rev_to_mark(tags[name])
377 def do_import(parser):
381 path = os.path.join(dirname, 'marks-git')
384 if os.path.exists(path):
385 print "feature import-marks=%s" % path
386 print "feature export-marks=%s" % path
387 print "feature force"
390 while parser.check('import'):
392 if ref.startswith('refs/heads/'):
393 name = ref[len('refs/heads/'):]
394 export_branch(repo, name)
395 if ref.startswith('refs/tags/'):
396 name = ref[len('refs/tags/'):]
397 export_tag(repo, name)
404 def parse_blob(parser):
408 mark = parser.get_mark()
410 data = parser.get_data()
411 blob_marks[mark] = data
416 def __init__(self, branch, revid, parents, files):
422 def copy_tree(revid):
423 files = files_cache[revid] = {}
425 tree = branch.repository.revision_tree(revid)
427 for path, entry in tree.iter_entries_by_dir():
428 files[path] = [entry.file_id, None]
433 if len(parents) == 0:
434 self.base_id = bzrlib.revision.NULL_REVISION
437 self.base_id = parents[0]
438 self.base_files = files_cache.get(self.base_id, None)
439 if not self.base_files:
440 self.base_files = copy_tree(self.base_id)
442 self.files = files_cache[revid] = self.base_files.copy()
445 for path, data in self.files.iteritems():
447 self.rev_files[fid] = [path, mark]
449 for path, f in files.iteritems():
450 fid, mark = self.files.get(path, [None, None])
452 fid = bzrlib.generate_ids.gen_file_id(path)
454 self.rev_files[fid] = [path, mark]
455 self.updates[fid] = f
457 def last_revision(self):
460 def iter_changes(self):
463 def get_parent(dirname, basename):
464 parent_fid, mark = self.base_files.get(dirname, [None, None])
467 parent_fid, mark = self.files.get(dirname, [None, None])
472 fid = bzrlib.generate_ids.gen_file_id(path)
473 add_entry(fid, dirname, 'directory')
476 def add_entry(fid, path, kind, mode = None):
477 dirname, basename = os.path.split(path)
478 parent_fid = get_parent(dirname, basename)
483 elif mode == '120000':
494 self.files[path] = [change[0], None]
495 changes.append(change)
497 def update_entry(fid, path, kind, mode = None):
498 dirname, basename = os.path.split(path)
499 parent_fid = get_parent(dirname, basename)
504 elif mode == '120000':
515 self.files[path] = [change[0], None]
516 changes.append(change)
518 def remove_entry(fid, path, kind):
519 dirname, basename = os.path.split(path)
520 parent_fid = get_parent(dirname, basename)
530 changes.append(change)
532 for fid, f in self.updates.iteritems():
536 remove_entry(fid, path, 'file')
539 if path in self.base_files:
540 update_entry(fid, path, 'file', f['mode'])
542 add_entry(fid, path, 'file', f['mode'])
544 self.files[path][1] = f['mark']
545 self.rev_files[fid][1] = f['mark']
549 def get_content(self, file_id):
550 path, mark = self.rev_files[file_id]
552 return blob_marks[mark]
555 tree = self.branch.repository.revision_tree(self.base_id)
556 return tree.get_file_text(file_id)
558 def get_file_with_stat(self, file_id, path=None):
559 content = self.get_content(file_id)
560 return (StringIO.StringIO(content), None)
562 def get_symlink_target(self, file_id):
563 return self.get_content(file_id)
565 def id2path(self, file_id):
566 path, mark = self.rev_files[file_id]
569 def c_style_unescape(string):
570 if string[0] == string[-1] == '"':
571 return string.decode('string-escape')[1:-1]
574 def parse_commit(parser):
575 global marks, blob_marks, parsed_refs
583 if ref.startswith('refs/heads/'):
584 name = ref[len('refs/heads/'):]
585 branch = branches[name]
589 commit_mark = parser.get_mark()
591 author = parser.get_author()
593 committer = parser.get_author()
595 data = parser.get_data()
597 if parser.check('from'):
598 parents.append(parser.get_mark())
600 while parser.check('merge'):
601 parents.append(parser.get_mark())
604 # fast-export adds an extra newline
611 if parser.check('M'):
612 t, m, mark_ref, path = line.split(' ', 3)
613 mark = int(mark_ref[1:])
614 f = { 'mode' : m, 'mark' : mark }
615 elif parser.check('D'):
616 t, path = line.split(' ')
617 f = { 'deleted' : True }
619 die('Unknown file command: %s' % line)
620 path = c_style_unescape(path).decode('utf-8')
623 committer, date, tz = committer
624 parents = [str(mark_to_rev(p)) for p in parents]
625 revid = bzrlib.generate_ids.gen_revision_id(committer, date)
627 props['branch-nick'] = branch.nick
629 mtree = CustomTree(branch, revid, parents, files)
630 changes = mtree.iter_changes()
634 builder = branch.get_commit_builder(parents, None, date, tz, committer, props, revid)
636 list(builder.record_iter_changes(mtree, mtree.last_revision(), changes))
637 builder.finish_inventory()
638 builder.commit(data.decode('utf-8', 'replace'))
645 parsed_refs[ref] = revid
646 marks.new_mark(revid, commit_mark)
648 def parse_reset(parser):
655 if parser.check('commit'):
658 if not parser.check('from'):
660 from_mark = parser.get_mark()
663 parsed_refs[ref] = mark_to_rev(from_mark)
665 def do_export(parser):
666 global parsed_refs, dirname
670 for line in parser.each_block('done'):
671 if parser.check('blob'):
673 elif parser.check('commit'):
675 elif parser.check('reset'):
677 elif parser.check('tag'):
679 elif parser.check('feature'):
682 die('unhandled export command: %s' % line)
684 for ref, revid in parsed_refs.iteritems():
685 name = ref[len('refs/heads/'):]
686 branch = branches[name]
687 branch.generate_revision_history(revid, marks.get_tip(name))
690 peer = bzrlib.branch.Branch.open(peers[name])
692 peer.bzrdir.push_branch(branch, revision_id=revid)
693 except bzrlib.errors.DivergedBranches:
694 print "error %s non-fast forward" % ref
698 wt = branch.bzrdir.open_workingtree()
700 except bzrlib.errors.NoWorkingTree:
707 def do_capabilities(parser):
712 print "refspec refs/heads/*:%s/heads/*" % prefix
713 print "refspec refs/tags/*:%s/tags/*" % prefix
715 path = os.path.join(dirname, 'marks-git')
717 if os.path.exists(path):
718 print "*import-marks %s" % path
719 print "*export-marks %s" % path
723 def ref_is_valid(name):
724 return not True in [c in name for c in '~^: \\']
731 for name in branches:
732 if not master_branch:
734 print "? refs/heads/%s" % name
736 branch = branches[master_branch]
738 for tag, revid in branch.tags.get_tag_dict().items():
740 branch.revision_id_to_dotted_revno(revid)
741 except bzrlib.errors.NoSuchRevision:
743 if not ref_is_valid(tag):
745 print "? refs/tags/%s" % tag
749 print "@refs/heads/%s HEAD" % master_branch
752 def get_remote_branch(origin, remote_branch, name):
753 global dirname, peers
755 branch_path = os.path.join(dirname, 'clone', name)
756 if os.path.exists(branch_path):
758 d = bzrlib.bzrdir.BzrDir.open(branch_path)
759 branch = d.open_branch()
761 branch.pull(remote_branch, [], None, False)
762 except bzrlib.errors.DivergedBranches:
763 # use remote branch for now
767 d = origin.sprout(branch_path, None,
768 hardlink=True, create_tree_if_local=False,
769 force_new_repo=False,
770 source_branch=remote_branch)
771 branch = d.open_branch()
775 def find_branches(repo, wanted):
776 transport = repo.user_transport
778 for fn in transport.iter_files_recursive():
779 if not fn.endswith('.bzr/branch-format'):
782 name = subdir = fn[:-len('/.bzr/branch-format')]
783 name = name if name != '' else 'master'
784 name = name.replace('/', '+')
786 if wanted and not name in wanted:
790 cur = transport.clone(subdir)
791 branch = bzrlib.branch.Branch.open_from_transport(cur)
792 except bzrlib.errors.NotBranchError:
797 def get_repo(url, alias):
798 global dirname, peer, branches
800 normal_url = bzrlib.urlutils.normalize_url(url)
801 origin = bzrlib.bzrdir.BzrDir.open(url)
802 is_local = isinstance(origin.transport, bzrlib.transport.local.LocalTransport)
804 shared_path = os.path.join(gitdir, 'bzr')
806 shared_dir = bzrlib.bzrdir.BzrDir.open(shared_path)
807 except bzrlib.errors.NotBranchError:
808 shared_dir = bzrlib.bzrdir.BzrDir.create(shared_path)
810 shared_repo = shared_dir.open_repository()
811 except bzrlib.errors.NoRepositoryPresent:
812 shared_repo = shared_dir.create_repository(shared=True)
815 clone_path = os.path.join(dirname, 'clone')
816 if not os.path.exists(clone_path):
820 repo = origin.open_repository()
821 except bzrlib.errors.NoRepositoryPresent:
825 branch = origin.open_branch()
828 peers[name] = branch.base
829 branches[name] = get_remote_branch(origin, branch, name)
831 branches[name] = branch
833 return branch.repository
837 wanted = get_config('remote-bzr.branches').rstrip().split(', ')
839 wanted = [e for e in wanted if e]
841 for name, branch in find_branches(repo, wanted):
844 peers[name] = branch.base
845 branches[name] = get_remote_branch(origin, branch, name)
847 branches[name] = branch
851 def fix_path(alias, orig_url):
852 url = urlparse.urlparse(orig_url, 'file')
853 if url.scheme != 'file' or os.path.isabs(url.path):
855 abs_url = urlparse.urljoin("%s/" % os.getcwd(), orig_url)
856 cmd = ['git', 'config', 'remote.%s.url' % alias, "bzr::%s" % abs_url]
860 global marks, prefix, gitdir, dirname
861 global tags, filenodes
866 global branches, peers
882 alias = hashlib.sha1(alias).hexdigest()
886 prefix = 'refs/bzr/%s' % alias
887 gitdir = os.environ['GIT_DIR']
888 dirname = os.path.join(gitdir, 'bzr', alias)
893 if not os.path.exists(dirname):
896 bzrlib.ui.ui_factory.be_quiet(True)
898 repo = get_repo(url, alias)
900 marks_path = os.path.join(dirname, 'marks-int')
901 marks = Marks(marks_path)
903 parser = Parser(repo)
905 if parser.check('capabilities'):
906 do_capabilities(parser)
907 elif parser.check('list'):
909 elif parser.check('import'):
911 elif parser.check('export'):
914 die('unhandled command: %s' % line)
923 shutil.rmtree(dirname)
926 sys.exit(main(sys.argv))