remote-bzr: always try to update the worktree
[git.git] / contrib / remote-helpers / git-remote-bzr
blobeb91d2814be8b9cbb7214e7cbc847012d4d2394a
1 #!/usr/bin/env python
3 # Copyright (c) 2012 Felipe Contreras
7 # Just copy to your ~/bin, or anywhere in your $PATH.
8 # Then you can clone with:
9 # % git clone bzr::/path/to/bzr/repo/or/url
11 # For example:
12 # % git clone bzr::$HOME/myrepo
13 # or
14 # % git clone bzr::lp:myrepo
17 import sys
19 import bzrlib
20 if hasattr(bzrlib, "initialize"):
21 bzrlib.initialize()
23 import bzrlib.plugin
24 bzrlib.plugin.load_plugins()
26 import bzrlib.generate_ids
27 import bzrlib.transport
28 import bzrlib.errors
29 import bzrlib.ui
31 import sys
32 import os
33 import json
34 import re
35 import StringIO
36 import atexit, shutil, hashlib, urlparse, subprocess
38 NAME_RE = re.compile('^([^<>]+)')
39 AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
40 RAW_AUTHOR_RE = re.compile('^(\w+) (.+)? <(.*)> (\d+) ([+-]\d+)')
42 def die(msg, *args):
43 sys.stderr.write('ERROR: %s\n' % (msg % args))
44 sys.exit(1)
46 def warn(msg, *args):
47 sys.stderr.write('WARNING: %s\n' % (msg % args))
49 def gittz(tz):
50 return '%+03d%02d' % (tz / 3600, tz % 3600 / 60)
52 class Marks:
54 def __init__(self, path):
55 self.path = path
56 self.tips = {}
57 self.marks = {}
58 self.rev_marks = {}
59 self.last_mark = 0
60 self.load()
62 def load(self):
63 if not os.path.exists(self.path):
64 return
66 tmp = json.load(open(self.path))
67 self.tips = tmp['tips']
68 self.marks = tmp['marks']
69 self.last_mark = tmp['last-mark']
71 for rev, mark in self.marks.iteritems():
72 self.rev_marks[mark] = rev
74 def dict(self):
75 return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
77 def store(self):
78 json.dump(self.dict(), open(self.path, 'w'))
80 def __str__(self):
81 return str(self.dict())
83 def from_rev(self, rev):
84 return self.marks[rev]
86 def to_rev(self, mark):
87 return self.rev_marks[mark]
89 def next_mark(self):
90 self.last_mark += 1
91 return self.last_mark
93 def get_mark(self, rev):
94 self.last_mark += 1
95 self.marks[rev] = self.last_mark
96 return self.last_mark
98 def is_marked(self, rev):
99 return rev in self.marks
101 def new_mark(self, rev, mark):
102 self.marks[rev] = mark
103 self.rev_marks[mark] = rev
104 self.last_mark = mark
106 def get_tip(self, branch):
107 return self.tips.get(branch, None)
109 def set_tip(self, branch, tip):
110 self.tips[branch] = tip
112 class Parser:
114 def __init__(self, repo):
115 self.repo = repo
116 self.line = self.get_line()
118 def get_line(self):
119 return sys.stdin.readline().strip()
121 def __getitem__(self, i):
122 return self.line.split()[i]
124 def check(self, word):
125 return self.line.startswith(word)
127 def each_block(self, separator):
128 while self.line != separator:
129 yield self.line
130 self.line = self.get_line()
132 def __iter__(self):
133 return self.each_block('')
135 def next(self):
136 self.line = self.get_line()
137 if self.line == 'done':
138 self.line = None
140 def get_mark(self):
141 i = self.line.index(':') + 1
142 return int(self.line[i:])
144 def get_data(self):
145 if not self.check('data'):
146 return None
147 i = self.line.index(' ') + 1
148 size = int(self.line[i:])
149 return sys.stdin.read(size)
151 def get_author(self):
152 m = RAW_AUTHOR_RE.match(self.line)
153 if not m:
154 return None
155 _, name, email, date, tz = m.groups()
156 committer = '%s <%s>' % (name, email)
157 tz = int(tz)
158 tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
159 return (committer, int(date), tz)
161 def rev_to_mark(rev):
162 global marks
163 return marks.from_rev(rev)
165 def mark_to_rev(mark):
166 global marks
167 return marks.to_rev(mark)
169 def fixup_user(user):
170 name = mail = None
171 user = user.replace('"', '')
172 m = AUTHOR_RE.match(user)
173 if m:
174 name = m.group(1)
175 mail = m.group(2).strip()
176 else:
177 m = NAME_RE.match(user)
178 if m:
179 name = m.group(1).strip()
181 return '%s <%s>' % (name, mail)
183 def get_filechanges(cur, prev):
184 modified = {}
185 removed = {}
187 changes = cur.changes_from(prev)
189 def u(s):
190 return s.encode('utf-8')
192 for path, fid, kind in changes.added:
193 modified[u(path)] = fid
194 for path, fid, kind in changes.removed:
195 removed[u(path)] = None
196 for path, fid, kind, mod, _ in changes.modified:
197 modified[u(path)] = fid
198 for oldpath, newpath, fid, kind, mod, _ in changes.renamed:
199 removed[u(oldpath)] = None
200 if kind == 'directory':
201 lst = cur.list_files(from_dir=newpath, recursive=True)
202 for path, file_class, kind, fid, entry in lst:
203 if kind != 'directory':
204 modified[u(newpath + '/' + path)] = fid
205 else:
206 modified[u(newpath)] = fid
208 return modified, removed
210 def export_files(tree, files):
211 global marks, filenodes
213 final = []
214 for path, fid in files.iteritems():
215 kind = tree.kind(fid)
217 h = tree.get_file_sha1(fid)
219 if kind == 'symlink':
220 d = tree.get_symlink_target(fid)
221 mode = '120000'
222 elif kind == 'file':
224 if tree.is_executable(fid):
225 mode = '100755'
226 else:
227 mode = '100644'
229 # is the blob already exported?
230 if h in filenodes:
231 mark = filenodes[h]
232 final.append((mode, mark, path))
233 continue
235 d = tree.get_file_text(fid)
236 elif kind == 'directory':
237 continue
238 else:
239 die("Unhandled kind '%s' for path '%s'" % (kind, path))
241 mark = marks.next_mark()
242 filenodes[h] = mark
244 print "blob"
245 print "mark :%u" % mark
246 print "data %d" % len(d)
247 print d
249 final.append((mode, mark, path))
251 return final
253 def export_branch(branch, name):
254 global prefix
256 ref = '%s/heads/%s' % (prefix, name)
257 tip = marks.get_tip(name)
259 repo = branch.repository
260 repo.lock_read()
261 revs = branch.iter_merge_sorted_revisions(None, tip, 'exclude', 'forward')
262 count = 0
264 revs = [revid for revid, _, _, _ in revs if not marks.is_marked(revid)]
266 for revid in revs:
268 rev = repo.get_revision(revid)
270 parents = rev.parent_ids
271 time = rev.timestamp
272 tz = rev.timezone
273 committer = rev.committer.encode('utf-8')
274 committer = "%s %u %s" % (fixup_user(committer), time, gittz(tz))
275 authors = rev.get_apparent_authors()
276 if authors:
277 author = authors[0].encode('utf-8')
278 author = "%s %u %s" % (fixup_user(author), time, gittz(tz))
279 else:
280 author = committer
281 msg = rev.message.encode('utf-8')
283 msg += '\n'
285 if len(parents) == 0:
286 parent = bzrlib.revision.NULL_REVISION
287 else:
288 parent = parents[0]
290 cur_tree = repo.revision_tree(revid)
291 prev = repo.revision_tree(parent)
292 modified, removed = get_filechanges(cur_tree, prev)
294 modified_final = export_files(cur_tree, modified)
296 if len(parents) == 0:
297 print 'reset %s' % ref
299 print "commit %s" % ref
300 print "mark :%d" % (marks.get_mark(revid))
301 print "author %s" % (author)
302 print "committer %s" % (committer)
303 print "data %d" % (len(msg))
304 print msg
306 for i, p in enumerate(parents):
307 try:
308 m = rev_to_mark(p)
309 except KeyError:
310 # ghost?
311 continue
312 if i == 0:
313 print "from :%s" % m
314 else:
315 print "merge :%s" % m
317 for f in removed:
318 print "D %s" % (f,)
319 for f in modified_final:
320 print "M %s :%u %s" % f
321 print
323 count += 1
324 if (count % 100 == 0):
325 print "progress revision %s (%d/%d)" % (revid, count, len(revs))
326 print "#############################################################"
328 repo.unlock()
330 revid = branch.last_revision()
332 # make sure the ref is updated
333 print "reset %s" % ref
334 print "from :%u" % rev_to_mark(revid)
335 print
337 marks.set_tip(name, revid)
339 def export_tag(repo, name):
340 global tags, prefix
342 ref = '%s/tags/%s' % (prefix, name)
343 print "reset %s" % ref
344 print "from :%u" % rev_to_mark(tags[name])
345 print
347 def do_import(parser):
348 global dirname
350 branch = parser.repo
351 path = os.path.join(dirname, 'marks-git')
353 print "feature done"
354 if os.path.exists(path):
355 print "feature import-marks=%s" % path
356 print "feature export-marks=%s" % path
357 sys.stdout.flush()
359 while parser.check('import'):
360 ref = parser[1]
361 if ref.startswith('refs/heads/'):
362 name = ref[len('refs/heads/'):]
363 export_branch(branch, name)
364 if ref.startswith('refs/tags/'):
365 name = ref[len('refs/tags/'):]
366 export_tag(branch, name)
367 parser.next()
369 print 'done'
371 sys.stdout.flush()
373 def parse_blob(parser):
374 global blob_marks
376 parser.next()
377 mark = parser.get_mark()
378 parser.next()
379 data = parser.get_data()
380 blob_marks[mark] = data
381 parser.next()
383 class CustomTree():
385 def __init__(self, repo, revid, parents, files):
386 global files_cache
388 self.updates = {}
390 def copy_tree(revid):
391 files = files_cache[revid] = {}
392 repo.lock_read()
393 tree = repo.repository.revision_tree(revid)
394 try:
395 for path, entry in tree.iter_entries_by_dir():
396 files[path] = entry.file_id
397 finally:
398 repo.unlock()
399 return files
401 if len(parents) == 0:
402 self.base_id = bzrlib.revision.NULL_REVISION
403 self.base_files = {}
404 else:
405 self.base_id = parents[0]
406 self.base_files = files_cache.get(self.base_id, None)
407 if not self.base_files:
408 self.base_files = copy_tree(self.base_id)
410 self.files = files_cache[revid] = self.base_files.copy()
412 for path, f in files.iteritems():
413 fid = self.files.get(path, None)
414 if not fid:
415 fid = bzrlib.generate_ids.gen_file_id(path)
416 f['path'] = path
417 self.updates[fid] = f
419 def last_revision(self):
420 return self.base_id
422 def iter_changes(self):
423 changes = []
425 def get_parent(dirname, basename):
426 parent_fid = self.base_files.get(dirname, None)
427 if parent_fid:
428 return parent_fid
429 parent_fid = self.files.get(dirname, None)
430 if parent_fid:
431 return parent_fid
432 if basename == '':
433 return None
434 fid = bzrlib.generate_ids.gen_file_id(path)
435 add_entry(fid, dirname, 'directory')
436 return fid
438 def add_entry(fid, path, kind, mode = None):
439 dirname, basename = os.path.split(path)
440 parent_fid = get_parent(dirname, basename)
442 executable = False
443 if mode == '100755':
444 executable = True
445 elif mode == '120000':
446 kind = 'symlink'
448 change = (fid,
449 (None, path),
450 True,
451 (False, True),
452 (None, parent_fid),
453 (None, basename),
454 (None, kind),
455 (None, executable))
456 self.files[path] = change[0]
457 changes.append(change)
459 def update_entry(fid, path, kind, mode = None):
460 dirname, basename = os.path.split(path)
461 parent_fid = get_parent(dirname, basename)
463 executable = False
464 if mode == '100755':
465 executable = True
466 elif mode == '120000':
467 kind = 'symlink'
469 change = (fid,
470 (path, path),
471 True,
472 (True, True),
473 (None, parent_fid),
474 (None, basename),
475 (None, kind),
476 (None, executable))
477 self.files[path] = change[0]
478 changes.append(change)
480 def remove_entry(fid, path, kind):
481 dirname, basename = os.path.split(path)
482 parent_fid = get_parent(dirname, basename)
483 change = (fid,
484 (path, None),
485 True,
486 (True, False),
487 (parent_fid, None),
488 (None, None),
489 (None, None),
490 (None, None))
491 del self.files[path]
492 changes.append(change)
494 for fid, f in self.updates.iteritems():
495 path = f['path']
497 if 'deleted' in f:
498 remove_entry(fid, path, 'file')
499 continue
501 if path in self.base_files:
502 update_entry(fid, path, 'file', f['mode'])
503 else:
504 add_entry(fid, path, 'file', f['mode'])
506 return changes
508 def get_file_with_stat(self, file_id, path=None):
509 mark = self.updates[file_id]['mark']
510 return (StringIO.StringIO(blob_marks[mark]), None)
512 def get_symlink_target(self, file_id):
513 mark = self.updates[file_id]['mark']
514 return blob_marks[mark]
516 def c_style_unescape(string):
517 if string[0] == string[-1] == '"':
518 return string.decode('string-escape')[1:-1]
519 return string
521 def parse_commit(parser):
522 global marks, blob_marks, parsed_refs
523 global mode
525 parents = []
527 ref = parser[1]
528 parser.next()
530 if ref != 'refs/heads/master':
531 die("bzr doesn't support multiple branches; use 'master'")
533 commit_mark = parser.get_mark()
534 parser.next()
535 author = parser.get_author()
536 parser.next()
537 committer = parser.get_author()
538 parser.next()
539 data = parser.get_data()
540 parser.next()
541 if parser.check('from'):
542 parents.append(parser.get_mark())
543 parser.next()
544 while parser.check('merge'):
545 parents.append(parser.get_mark())
546 parser.next()
548 # fast-export adds an extra newline
549 if data[-1] == '\n':
550 data = data[:-1]
552 files = {}
554 for line in parser:
555 if parser.check('M'):
556 t, m, mark_ref, path = line.split(' ', 3)
557 mark = int(mark_ref[1:])
558 f = { 'mode' : m, 'mark' : mark }
559 elif parser.check('D'):
560 t, path = line.split(' ')
561 f = { 'deleted' : True }
562 else:
563 die('Unknown file command: %s' % line)
564 path = c_style_unescape(path).decode('utf-8')
565 files[path] = f
567 repo = parser.repo
569 committer, date, tz = committer
570 parents = [str(mark_to_rev(p)) for p in parents]
571 revid = bzrlib.generate_ids.gen_revision_id(committer, date)
572 props = {}
573 props['branch-nick'] = repo.nick
575 mtree = CustomTree(repo, revid, parents, files)
576 changes = mtree.iter_changes()
578 repo.lock_write()
579 try:
580 builder = repo.get_commit_builder(parents, None, date, tz, committer, props, revid)
581 try:
582 list(builder.record_iter_changes(mtree, mtree.last_revision(), changes))
583 builder.finish_inventory()
584 builder.commit(data.decode('utf-8', 'replace'))
585 except Exception, e:
586 builder.abort()
587 raise
588 finally:
589 repo.unlock()
591 parsed_refs[ref] = revid
592 marks.new_mark(revid, commit_mark)
594 def parse_reset(parser):
595 global parsed_refs
597 ref = parser[1]
598 parser.next()
600 if ref != 'refs/heads/master':
601 die("bzr doesn't support multiple branches; use 'master'")
603 # ugh
604 if parser.check('commit'):
605 parse_commit(parser)
606 return
607 if not parser.check('from'):
608 return
609 from_mark = parser.get_mark()
610 parser.next()
612 parsed_refs[ref] = mark_to_rev(from_mark)
614 def do_export(parser):
615 global parsed_refs, dirname, peer
617 parser.next()
619 for line in parser.each_block('done'):
620 if parser.check('blob'):
621 parse_blob(parser)
622 elif parser.check('commit'):
623 parse_commit(parser)
624 elif parser.check('reset'):
625 parse_reset(parser)
626 elif parser.check('tag'):
627 pass
628 elif parser.check('feature'):
629 pass
630 else:
631 die('unhandled export command: %s' % line)
633 repo = parser.repo
635 for ref, revid in parsed_refs.iteritems():
636 if ref == 'refs/heads/master':
637 repo.generate_revision_history(revid, marks.get_tip('master'))
638 if peer:
639 try:
640 repo.push(peer, stop_revision=revid)
641 except bzrlib.errors.DivergedBranches:
642 print "error %s non-fast forward" % ref
643 continue
645 try:
646 wt = repo.bzrdir.open_workingtree()
647 wt.update()
648 except bzrlib.errors.NoWorkingTree:
649 pass
651 print "ok %s" % ref
653 print
655 def do_capabilities(parser):
656 global dirname
658 print "import"
659 print "export"
660 print "refspec refs/heads/*:%s/heads/*" % prefix
661 print "refspec refs/tags/*:%s/tags/*" % prefix
663 path = os.path.join(dirname, 'marks-git')
665 if os.path.exists(path):
666 print "*import-marks %s" % path
667 print "*export-marks %s" % path
669 print
671 def ref_is_valid(name):
672 return not True in [c in name for c in '~^: \\']
674 def do_list(parser):
675 global tags
676 print "? refs/heads/%s" % 'master'
678 branch = parser.repo
679 branch.lock_read()
680 for tag, revid in branch.tags.get_tag_dict().items():
681 try:
682 branch.revision_id_to_dotted_revno(revid)
683 except bzrlib.errors.NoSuchRevision:
684 continue
685 if not ref_is_valid(tag):
686 continue
687 print "? refs/tags/%s" % tag
688 tags[tag] = revid
689 branch.unlock()
690 print "@refs/heads/%s HEAD" % 'master'
691 print
693 def get_repo(url, alias):
694 global dirname, peer
696 origin = bzrlib.bzrdir.BzrDir.open(url)
697 branch = origin.open_branch()
699 if not isinstance(origin.transport, bzrlib.transport.local.LocalTransport):
700 clone_path = os.path.join(dirname, 'clone')
701 remote_branch = branch
702 if os.path.exists(clone_path):
703 # pull
704 d = bzrlib.bzrdir.BzrDir.open(clone_path)
705 branch = d.open_branch()
706 result = branch.pull(remote_branch, [], None, False)
707 else:
708 # clone
709 d = origin.sprout(clone_path, None,
710 hardlink=True, create_tree_if_local=False,
711 source_branch=remote_branch)
712 branch = d.open_branch()
713 branch.bind(remote_branch)
715 peer = remote_branch
716 else:
717 peer = None
719 return branch
721 def fix_path(alias, orig_url):
722 url = urlparse.urlparse(orig_url, 'file')
723 if url.scheme != 'file' or os.path.isabs(url.path):
724 return
725 abs_url = urlparse.urljoin("%s/" % os.getcwd(), orig_url)
726 cmd = ['git', 'config', 'remote.%s.url' % alias, "bzr::%s" % abs_url]
727 subprocess.call(cmd)
729 def main(args):
730 global marks, prefix, dirname
731 global tags, filenodes
732 global blob_marks
733 global parsed_refs
734 global files_cache
735 global is_tmp
737 alias = args[1]
738 url = args[2]
740 tags = {}
741 filenodes = {}
742 blob_marks = {}
743 parsed_refs = {}
744 files_cache = {}
745 marks = None
747 if alias[5:] == url:
748 is_tmp = True
749 alias = hashlib.sha1(alias).hexdigest()
750 else:
751 is_tmp = False
753 prefix = 'refs/bzr/%s' % alias
754 gitdir = os.environ['GIT_DIR']
755 dirname = os.path.join(gitdir, 'bzr', alias)
757 if not is_tmp:
758 fix_path(alias, url)
760 if not os.path.exists(dirname):
761 os.makedirs(dirname)
763 bzrlib.ui.ui_factory.be_quiet(True)
765 repo = get_repo(url, alias)
767 marks_path = os.path.join(dirname, 'marks-int')
768 marks = Marks(marks_path)
770 parser = Parser(repo)
771 for line in parser:
772 if parser.check('capabilities'):
773 do_capabilities(parser)
774 elif parser.check('list'):
775 do_list(parser)
776 elif parser.check('import'):
777 do_import(parser)
778 elif parser.check('export'):
779 do_export(parser)
780 else:
781 die('unhandled command: %s' % line)
782 sys.stdout.flush()
784 def bye():
785 if not marks:
786 return
787 if not is_tmp:
788 marks.store()
789 else:
790 shutil.rmtree(dirname)
792 atexit.register(bye)
793 sys.exit(main(sys.argv))