contrib/remote-helpers/git-remote-bzr

   1 #!/usr/bin/env python
   2 #
   3 # Copyright (c) 2012 Felipe Contreras
   4 #
   5
   6 #
   7 # Just copy to your ~/bin, or anywhere in your $PATH.
   8 # Then you can clone with:
   9 # % git clone bzr::/path/to/bzr/repo/or/url
  10 #
  11 # For example:
  12 # % git clone bzr::$HOME/myrepo
  13 # or
  14 # % git clone bzr::lp:myrepo
  15 #
  16
  17 import sys
  18
  19 import bzrlib
  20 if hasattr(bzrlib, "initialize"):
  21     bzrlib.initialize()
  22
  23 import bzrlib.plugin
  24 bzrlib.plugin.load_plugins()
  25
  26 import bzrlib.generate_ids
  27 import bzrlib.transport
  28
  29 import sys
  30 import os
  31 import json
  32 import re
  33 import StringIO
  34
  35 NAME_RE = re.compile('^([^<>]+)')
  36 AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
  37 RAW_AUTHOR_RE = re.compile('^(\w+) (.+)? <(.*)> (\d+) ([+-]\d+)')
  38
  39 def die(msg, *args):
  40     sys.stderr.write('ERROR: %s\n' % (msg % args))
  41     sys.exit(1)
  42
  43 def warn(msg, *args):
  44     sys.stderr.write('WARNING: %s\n' % (msg % args))
  45
  46 def gittz(tz):
  47     return '%+03d%02d' % (tz / 3600, tz % 3600 / 60)
  48
  49 class Marks:
  50
  51     def __init__(self, path):
  52         self.path = path
  53         self.tips = {}
  54         self.marks = {}
  55         self.rev_marks = {}
  56         self.last_mark = 0
  57         self.load()
  58
  59     def load(self):
  60         if not os.path.exists(self.path):
  61             return
  62
  63         tmp = json.load(open(self.path))
  64         self.tips = tmp['tips']
  65         self.marks = tmp['marks']
  66         self.last_mark = tmp['last-mark']
  67
  68         for rev, mark in self.marks.iteritems():
  69             self.rev_marks[mark] = rev
  70
  71     def dict(self):
  72         return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
  73
  74     def store(self):
  75         json.dump(self.dict(), open(self.path, 'w'))
  76
  77     def __str__(self):
  78         return str(self.dict())
  79
  80     def from_rev(self, rev):
  81         return self.marks[rev]
  82
  83     def to_rev(self, mark):
  84         return self.rev_marks[mark]
  85
  86     def next_mark(self):
  87         self.last_mark += 1
  88         return self.last_mark
  89
  90     def get_mark(self, rev):
  91         self.last_mark += 1
  92         self.marks[rev] = self.last_mark
  93         return self.last_mark
  94
  95     def is_marked(self, rev):
  96         return self.marks.has_key(rev)
  97
  98     def new_mark(self, rev, mark):
  99         self.marks[rev] = mark
 100         self.rev_marks[mark] = rev
 101         self.last_mark = mark
 102
 103     def get_tip(self, branch):
 104         return self.tips.get(branch, None)
 105
 106     def set_tip(self, branch, tip):
 107         self.tips[branch] = tip
 108
 109 class Parser:
 110
 111     def __init__(self, repo):
 112         self.repo = repo
 113         self.line = self.get_line()
 114
 115     def get_line(self):
 116         return sys.stdin.readline().strip()
 117
 118     def __getitem__(self, i):
 119         return self.line.split()[i]
 120
 121     def check(self, word):
 122         return self.line.startswith(word)
 123
 124     def each_block(self, separator):
 125         while self.line != separator:
 126             yield self.line
 127             self.line = self.get_line()
 128
 129     def __iter__(self):
 130         return self.each_block('')
 131
 132     def next(self):
 133         self.line = self.get_line()
 134         if self.line == 'done':
 135             self.line = None
 136
 137     def get_mark(self):
 138         i = self.line.index(':') + 1
 139         return int(self.line[i:])
 140
 141     def get_data(self):
 142         if not self.check('data'):
 143             return None
 144         i = self.line.index(' ') + 1
 145         size = int(self.line[i:])
 146         return sys.stdin.read(size)
 147
 148     def get_author(self):
 149         m = RAW_AUTHOR_RE.match(self.line)
 150         if not m:
 151             return None
 152         _, name, email, date, tz = m.groups()
 153         committer = '%s <%s>' % (name, email)
 154         tz = int(tz)
 155         tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
 156         return (committer, int(date), tz)
 157
 158 def rev_to_mark(rev):
 159     global marks
 160     return marks.from_rev(rev)
 161
 162 def mark_to_rev(mark):
 163     global marks
 164     return marks.to_rev(mark)
 165
 166 def fixup_user(user):
 167     name = mail = None
 168     user = user.replace('"', '')
 169     m = AUTHOR_RE.match(user)
 170     if m:
 171         name = m.group(1)
 172         mail = m.group(2).strip()
 173     else:
 174         m = NAME_RE.match(user)
 175         if m:
 176             name = m.group(1).strip()
 177
 178     return '%s <%s>' % (name, mail)
 179
 180 def get_filechanges(cur, prev):
 181     modified = {}
 182     removed = {}
 183
 184     changes = cur.changes_from(prev)
 185
 186     for path, fid, kind in changes.added:
 187         modified[path] = fid
 188     for path, fid, kind in changes.removed:
 189         removed[path] = None
 190     for path, fid, kind, mod, _ in changes.modified:
 191         modified[path] = fid
 192     for oldpath, newpath, fid, kind, mod, _ in changes.renamed:
 193         removed[oldpath] = None
 194         if kind == 'directory':
 195             lst = cur.list_files(from_dir=newpath, recursive=True)
 196             for path, file_class, kind, fid, entry in lst:
 197                 if kind != 'directory':
 198                     modified[newpath + '/' + path] = fid
 199         else:
 200             modified[newpath] = fid
 201
 202     return modified, removed
 203
 204 def export_files(tree, files):
 205     global marks, filenodes
 206
 207     final = []
 208     for path, fid in files.iteritems():
 209         kind = tree.kind(fid)
 210
 211         h = tree.get_file_sha1(fid)
 212
 213         if kind == 'symlink':
 214             d = tree.get_symlink_target(fid)
 215             mode = '120000'
 216         elif kind == 'file':
 217
 218             if tree.is_executable(fid):
 219                 mode = '100755'
 220             else:
 221                 mode = '100644'
 222
 223             # is the blog already exported?
 224             if h in filenodes:
 225                 mark = filenodes[h]
 226                 final.append((mode, mark, path))
 227                 continue
 228
 229             d = tree.get_file_text(fid)
 230         elif kind == 'directory':
 231             continue
 232         else:
 233             die("Unhandled kind '%s' for path '%s'" % (kind, path))
 234
 235         mark = marks.next_mark()
 236         filenodes[h] = mark
 237
 238         print "blob"
 239         print "mark :%u" % mark
 240         print "data %d" % len(d)
 241         print d
 242
 243         final.append((mode, mark, path))
 244
 245     return final
 246
 247 def export_branch(branch, name):
 248     global prefix, dirname
 249
 250     ref = '%s/heads/%s' % (prefix, name)
 251     tip = marks.get_tip(name)
 252
 253     repo = branch.repository
 254     repo.lock_read()
 255     revs = branch.iter_merge_sorted_revisions(None, tip, 'exclude', 'forward')
 256     count = 0
 257
 258     revs = [revid for revid, _, _, _ in revs if not marks.is_marked(revid)]
 259
 260     for revid in revs:
 261
 262         rev = repo.get_revision(revid)
 263
 264         parents = rev.parent_ids
 265         time = rev.timestamp
 266         tz = rev.timezone
 267         committer = rev.committer.encode('utf-8')
 268         committer = "%s %u %s" % (fixup_user(committer), time, gittz(tz))
 269         authors = rev.get_apparent_authors()
 270         if authors:
 271             author = authors[0].encode('utf-8')
 272             author = "%s %u %s" % (fixup_user(author), time, gittz(tz))
 273         else:
 274             author = committer
 275         msg = rev.message.encode('utf-8')
 276
 277         msg += '\n'
 278
 279         if len(parents) == 0:
 280             parent = bzrlib.revision.NULL_REVISION
 281         else:
 282             parent = parents[0]
 283
 284         cur_tree = repo.revision_tree(revid)
 285         prev = repo.revision_tree(parent)
 286         modified, removed = get_filechanges(cur_tree, prev)
 287
 288         modified_final = export_files(cur_tree, modified)
 289
 290         if len(parents) == 0:
 291             print 'reset %s' % ref
 292
 293         print "commit %s" % ref
 294         print "mark :%d" % (marks.get_mark(revid))
 295         print "author %s" % (author)
 296         print "committer %s" % (committer)
 297         print "data %d" % (len(msg))
 298         print msg
 299
 300         for i, p in enumerate(parents):
 301             try:
 302                 m = rev_to_mark(p)
 303             except KeyError:
 304                 # ghost?
 305                 continue
 306             if i == 0:
 307                 print "from :%s" % m
 308             else:
 309                 print "merge :%s" % m
 310
 311         for f in removed:
 312             print "D %s" % (f,)
 313         for f in modified_final:
 314             print "M %s :%u %s" % f
 315         print
 316
 317         count += 1
 318         if (count % 100 == 0):
 319             print "progress revision %s (%d/%d)" % (revid, count, len(revs))
 320             print "#############################################################"
 321
 322     repo.unlock()
 323
 324     revid = branch.last_revision()
 325
 326     # make sure the ref is updated
 327     print "reset %s" % ref
 328     print "from :%u" % rev_to_mark(revid)
 329     print
 330
 331     marks.set_tip(name, revid)
 332
 333 def export_tag(repo, name):
 334     global tags
 335     try:
 336         print "reset refs/tags/%s" % name
 337         print "from :%u" % rev_to_mark(tags[name])
 338         print
 339     except KeyError:
 340         warn("TODO: fetch tag '%s'" % name)
 341
 342 def do_import(parser):
 343     global dirname
 344
 345     branch = parser.repo
 346     path = os.path.join(dirname, 'marks-git')
 347
 348     print "feature done"
 349     if os.path.exists(path):
 350         print "feature import-marks=%s" % path
 351     print "feature export-marks=%s" % path
 352     sys.stdout.flush()
 353
 354     while parser.check('import'):
 355         ref = parser[1]
 356         if ref.startswith('refs/heads/'):
 357             name = ref[len('refs/heads/'):]
 358             export_branch(branch, name)
 359         if ref.startswith('refs/tags/'):
 360             name = ref[len('refs/tags/'):]
 361             export_tag(branch, name)
 362         parser.next()
 363
 364     print 'done'
 365
 366     sys.stdout.flush()
 367
 368 def parse_blob(parser):
 369     global blob_marks
 370
 371     parser.next()
 372     mark = parser.get_mark()
 373     parser.next()
 374     data = parser.get_data()
 375     blob_marks[mark] = data
 376     parser.next()
 377
 378 class CustomTree():
 379
 380     def __init__(self, repo, revid, parents, files):
 381         global files_cache
 382
 383         self.repo = repo
 384         self.revid = revid
 385         self.parents = parents
 386         self.updates = {}
 387
 388         def copy_tree(revid):
 389             files = files_cache[revid] = {}
 390             tree = repo.repository.revision_tree(revid)
 391             repo.lock_read()
 392             try:
 393                 for path, entry in tree.iter_entries_by_dir():
 394                     files[path] = entry.file_id
 395             finally:
 396                 repo.unlock()
 397             return files
 398
 399         if len(parents) == 0:
 400             self.base_id = bzrlib.revision.NULL_REVISION
 401             self.base_files = {}
 402         else:
 403             self.base_id = parents[0]
 404             self.base_files = files_cache.get(self.base_id, None)
 405             if not self.base_files:
 406                 self.base_files = copy_tree(self.base_id)
 407
 408         self.files = files_cache[revid] = self.base_files.copy()
 409
 410         for path, f in files.iteritems():
 411             fid = self.files.get(path, None)
 412             if not fid:
 413                 fid = bzrlib.generate_ids.gen_file_id(path)
 414             f['path'] = path
 415             self.updates[fid] = f
 416
 417     def last_revision(self):
 418         return self.base_id
 419
 420     def iter_changes(self):
 421         changes = []
 422
 423         def get_parent(dirname, basename):
 424             parent_fid = self.base_files.get(dirname, None)
 425             if parent_fid:
 426                 return parent_fid
 427             parent_fid = self.files.get(dirname, None)
 428             if parent_fid:
 429                 return parent_fid
 430             if basename == '':
 431                 return None
 432             fid = bzrlib.generate_ids.gen_file_id(path)
 433             d = add_entry(fid, dirname, 'directory')
 434             return fid
 435
 436         def add_entry(fid, path, kind, mode = None):
 437             dirname, basename = os.path.split(path)
 438             parent_fid = get_parent(dirname, basename)
 439
 440             executable = False
 441             if mode == '100755':
 442                 executable = True
 443             elif mode == '120000':
 444                 kind = 'symlink'
 445
 446             change = (fid,
 447                     (None, path),
 448                     True,
 449                     (False, True),
 450                     (None, parent_fid),
 451                     (None, basename),
 452                     (None, kind),
 453                     (None, executable))
 454             self.files[path] = change[0]
 455             changes.append(change)
 456             return change
 457
 458         def update_entry(fid, path, kind, mode = None):
 459             dirname, basename = os.path.split(path)
 460             parent_fid = get_parent(dirname, basename)
 461
 462             executable = False
 463             if mode == '100755':
 464                 executable = True
 465             elif mode == '120000':
 466                 kind = 'symlink'
 467
 468             change = (fid,
 469                     (path, path),
 470                     True,
 471                     (True, True),
 472                     (None, parent_fid),
 473                     (None, basename),
 474                     (None, kind),
 475                     (None, executable))
 476             self.files[path] = change[0]
 477             changes.append(change)
 478             return change
 479
 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)
 493             return change
 494
 495         for fid, f in self.updates.iteritems():
 496             path = f['path']
 497
 498             if 'deleted' in f:
 499                 remove_entry(fid, path, 'file')
 500                 continue
 501
 502             if path in self.base_files:
 503                 update_entry(fid, path, 'file', f['mode'])
 504             else:
 505                 add_entry(fid, path, 'file', f['mode'])
 506
 507         return changes
 508
 509     def get_file_with_stat(self, file_id, path=None):
 510         return (StringIO.StringIO(self.updates[file_id]['data']), None)
 511
 512     def get_symlink_target(self, file_id):
 513         return self.updates[file_id]['data']
 514
 515 def parse_commit(parser):
 516     global marks, blob_marks, bmarks, parsed_refs
 517     global mode
 518
 519     parents = []
 520
 521     ref = parser[1]
 522     parser.next()
 523
 524     if ref != 'refs/heads/master':
 525         die("bzr doesn't support multiple branches; use 'master'")
 526
 527     commit_mark = parser.get_mark()
 528     parser.next()
 529     author = parser.get_author()
 530     parser.next()
 531     committer = parser.get_author()
 532     parser.next()
 533     data = parser.get_data()
 534     parser.next()
 535     if parser.check('from'):
 536         parents.append(parser.get_mark())
 537         parser.next()
 538     while parser.check('merge'):
 539         parents.append(parser.get_mark())
 540         parser.next()
 541
 542     files = {}
 543
 544     for line in parser:
 545         if parser.check('M'):
 546             t, m, mark_ref, path = line.split(' ', 3)
 547             mark = int(mark_ref[1:])
 548             f = { 'mode' : m, 'data' : blob_marks[mark] }
 549         elif parser.check('D'):
 550             t, path = line.split(' ')
 551             f = { 'deleted' : True }
 552         else:
 553             die('Unknown file command: %s' % line)
 554         files[path] = f
 555
 556     repo = parser.repo
 557
 558     committer, date, tz = committer
 559     parents = [str(mark_to_rev(p)) for p in parents]
 560     revid = bzrlib.generate_ids.gen_revision_id(committer, date)
 561     props = {}
 562     props['branch-nick'] = repo.nick
 563
 564     mtree = CustomTree(repo, revid, parents, files)
 565     changes = mtree.iter_changes()
 566
 567     repo.lock_write()
 568     try:
 569         builder = repo.get_commit_builder(parents, None, date, tz, committer, props, revid)
 570         try:
 571             list(builder.record_iter_changes(mtree, mtree.last_revision(), changes))
 572             builder.finish_inventory()
 573             builder.commit(data.decode('utf-8', 'replace'))
 574         except Exception, e:
 575             builder.abort()
 576             raise
 577     finally:
 578         repo.unlock()
 579
 580     parsed_refs[ref] = revid
 581     marks.new_mark(revid, commit_mark)
 582
 583 def parse_reset(parser):
 584     global parsed_refs
 585
 586     ref = parser[1]
 587     parser.next()
 588
 589     if ref != 'refs/heads/master':
 590         die("bzr doesn't support multiple branches; use 'master'")
 591
 592     # ugh
 593     if parser.check('commit'):
 594         parse_commit(parser)
 595         return
 596     if not parser.check('from'):
 597         return
 598     from_mark = parser.get_mark()
 599     parser.next()
 600
 601     parsed_refs[ref] = mark_to_rev(from_mark)
 602
 603 def do_export(parser):
 604     global parsed_refs, dirname, peer
 605
 606     parser.next()
 607
 608     for line in parser.each_block('done'):
 609         if parser.check('blob'):
 610             parse_blob(parser)
 611         elif parser.check('commit'):
 612             parse_commit(parser)
 613         elif parser.check('reset'):
 614             parse_reset(parser)
 615         elif parser.check('tag'):
 616             pass
 617         elif parser.check('feature'):
 618             pass
 619         else:
 620             die('unhandled export command: %s' % line)
 621
 622     repo = parser.repo
 623
 624     for ref, revid in parsed_refs.iteritems():
 625         if ref == 'refs/heads/master':
 626             repo.generate_revision_history(revid, marks.get_tip('master'))
 627             revno, revid = repo.last_revision_info()
 628             if peer:
 629                 if hasattr(peer, "import_last_revision_info_and_tags"):
 630                     peer.import_last_revision_info_and_tags(repo, revno, revid)
 631                 else:
 632                     peer.import_last_revision_info(repo.repository, revno, revid)
 633             else:
 634                 wt = repo.bzrdir.open_workingtree()
 635                 wt.update()
 636         print "ok %s" % ref
 637     print
 638
 639 def do_capabilities(parser):
 640     global dirname
 641
 642     print "import"
 643     print "export"
 644     print "refspec refs/heads/*:%s/heads/*" % prefix
 645
 646     path = os.path.join(dirname, 'marks-git')
 647
 648     if os.path.exists(path):
 649         print "*import-marks %s" % path
 650     print "*export-marks %s" % path
 651
 652     print
 653
 654 def do_list(parser):
 655     global tags
 656     print "? refs/heads/%s" % 'master'
 657     for tag, revid in parser.repo.tags.get_tag_dict().items():
 658         print "? refs/tags/%s" % tag
 659         tags[tag] = revid
 660     print "@refs/heads/%s HEAD" % 'master'
 661     print
 662
 663 def get_repo(url, alias):
 664     global dirname, peer
 665
 666     origin = bzrlib.bzrdir.BzrDir.open(url)
 667     branch = origin.open_branch()
 668
 669     if not isinstance(origin.transport, bzrlib.transport.local.LocalTransport):
 670         clone_path = os.path.join(dirname, 'clone')
 671         remote_branch = branch
 672         if os.path.exists(clone_path):
 673             # pull
 674             d = bzrlib.bzrdir.BzrDir.open(clone_path)
 675             branch = d.open_branch()
 676             result = branch.pull(remote_branch, [], None, False)
 677         else:
 678             # clone
 679             d = origin.sprout(clone_path, None,
 680                     hardlink=True, create_tree_if_local=False,
 681                     source_branch=remote_branch)
 682             branch = d.open_branch()
 683             branch.bind(remote_branch)
 684
 685         peer = remote_branch
 686     else:
 687         peer = None
 688
 689     return branch
 690
 691 def main(args):
 692     global marks, prefix, dirname
 693     global tags, filenodes
 694     global blob_marks
 695     global parsed_refs
 696     global files_cache
 697
 698     alias = args[1]
 699     url = args[2]
 700
 701     prefix = 'refs/bzr/%s' % alias
 702     tags = {}
 703     filenodes = {}
 704     blob_marks = {}
 705     parsed_refs = {}
 706     files_cache = {}
 707
 708     gitdir = os.environ['GIT_DIR']
 709     dirname = os.path.join(gitdir, 'bzr', alias)
 710
 711     if not os.path.exists(dirname):
 712         os.makedirs(dirname)
 713
 714     repo = get_repo(url, alias)
 715
 716     marks_path = os.path.join(dirname, 'marks-int')
 717     marks = Marks(marks_path)
 718
 719     parser = Parser(repo)
 720     for line in parser:
 721         if parser.check('capabilities'):
 722             do_capabilities(parser)
 723         elif parser.check('list'):
 724             do_list(parser)
 725         elif parser.check('import'):
 726             do_import(parser)
 727         elif parser.check('export'):
 728             do_export(parser)
 729         else:
 730             die('unhandled command: %s' % line)
 731         sys.stdout.flush()
 732
 733     marks.store()
 734
 735 sys.exit(main(sys.argv))