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         modified[newpath] = fid
 195
 196     return modified, removed
 197
 198 def export_files(tree, files):
 199     global marks, filenodes
 200
 201     final = []
 202     for path, fid in files.iteritems():
 203         kind = tree.kind(fid)
 204
 205         h = tree.get_file_sha1(fid)
 206
 207         if kind == 'symlink':
 208             d = tree.get_symlink_target(fid)
 209             mode = '120000'
 210         elif kind == 'file':
 211
 212             if tree.is_executable(fid):
 213                 mode = '100755'
 214             else:
 215                 mode = '100644'
 216
 217             # is the blog already exported?
 218             if h in filenodes:
 219                 mark = filenodes[h]
 220                 final.append((mode, mark, path))
 221                 continue
 222
 223             d = tree.get_file_text(fid)
 224         elif kind == 'directory':
 225             continue
 226         else:
 227             die("Unhandled kind '%s' for path '%s'" % (kind, path))
 228
 229         mark = marks.next_mark()
 230         filenodes[h] = mark
 231
 232         print "blob"
 233         print "mark :%u" % mark
 234         print "data %d" % len(d)
 235         print d
 236
 237         final.append((mode, mark, path))
 238
 239     return final
 240
 241 def export_branch(branch, name):
 242     global prefix, dirname
 243
 244     ref = '%s/heads/%s' % (prefix, name)
 245     tip = marks.get_tip(name)
 246
 247     repo = branch.repository
 248     repo.lock_read()
 249     revs = branch.iter_merge_sorted_revisions(None, tip, 'exclude', 'forward')
 250     count = 0
 251
 252     revs = [revid for revid, _, _, _ in revs if not marks.is_marked(revid)]
 253
 254     for revid in revs:
 255
 256         rev = repo.get_revision(revid)
 257
 258         parents = rev.parent_ids
 259         time = rev.timestamp
 260         tz = rev.timezone
 261         committer = rev.committer.encode('utf-8')
 262         committer = "%s %u %s" % (fixup_user(committer), time, gittz(tz))
 263         author = committer
 264         msg = rev.message.encode('utf-8')
 265
 266         msg += '\n'
 267
 268         if len(parents) == 0:
 269             parent = bzrlib.revision.NULL_REVISION
 270         else:
 271             parent = parents[0]
 272
 273         cur_tree = repo.revision_tree(revid)
 274         prev = repo.revision_tree(parent)
 275         modified, removed = get_filechanges(cur_tree, prev)
 276
 277         modified_final = export_files(cur_tree, modified)
 278
 279         if len(parents) == 0:
 280             print 'reset %s' % ref
 281
 282         print "commit %s" % ref
 283         print "mark :%d" % (marks.get_mark(revid))
 284         print "author %s" % (author)
 285         print "committer %s" % (committer)
 286         print "data %d" % (len(msg))
 287         print msg
 288
 289         for i, p in enumerate(parents):
 290             try:
 291                 m = rev_to_mark(p)
 292             except KeyError:
 293                 # ghost?
 294                 continue
 295             if i == 0:
 296                 print "from :%s" % m
 297             else:
 298                 print "merge :%s" % m
 299
 300         for f in modified_final:
 301             print "M %s :%u %s" % f
 302         for f in removed:
 303             print "D %s" % (f)
 304         print
 305
 306         count += 1
 307         if (count % 100 == 0):
 308             print "progress revision %s (%d/%d)" % (revid, count, len(revs))
 309             print "#############################################################"
 310
 311     repo.unlock()
 312
 313     revid = branch.last_revision()
 314
 315     # make sure the ref is updated
 316     print "reset %s" % ref
 317     print "from :%u" % rev_to_mark(revid)
 318     print
 319
 320     marks.set_tip(name, revid)
 321
 322 def export_tag(repo, name):
 323     global tags
 324     try:
 325         print "reset refs/tags/%s" % name
 326         print "from :%u" % rev_to_mark(tags[name])
 327         print
 328     except KeyError:
 329         warn("TODO: fetch tag '%s'" % name)
 330
 331 def do_import(parser):
 332     global dirname
 333
 334     branch = parser.repo
 335     path = os.path.join(dirname, 'marks-git')
 336
 337     print "feature done"
 338     if os.path.exists(path):
 339         print "feature import-marks=%s" % path
 340     print "feature export-marks=%s" % path
 341     sys.stdout.flush()
 342
 343     while parser.check('import'):
 344         ref = parser[1]
 345         if ref.startswith('refs/heads/'):
 346             name = ref[len('refs/heads/'):]
 347             export_branch(branch, name)
 348         if ref.startswith('refs/tags/'):
 349             name = ref[len('refs/tags/'):]
 350             export_tag(branch, name)
 351         parser.next()
 352
 353     print 'done'
 354
 355     sys.stdout.flush()
 356
 357 def parse_blob(parser):
 358     global blob_marks
 359
 360     parser.next()
 361     mark = parser.get_mark()
 362     parser.next()
 363     data = parser.get_data()
 364     blob_marks[mark] = data
 365     parser.next()
 366
 367 class CustomTree():
 368
 369     def __init__(self, repo, revid, parents, files):
 370         global files_cache
 371
 372         self.repo = repo
 373         self.revid = revid
 374         self.parents = parents
 375         self.updates = {}
 376
 377         def copy_tree(revid):
 378             files = files_cache[revid] = {}
 379             tree = repo.repository.revision_tree(revid)
 380             repo.lock_read()
 381             try:
 382                 for path, entry in tree.iter_entries_by_dir():
 383                     files[path] = entry.file_id
 384             finally:
 385                 repo.unlock()
 386             return files
 387
 388         if len(parents) == 0:
 389             self.base_id = bzrlib.revision.NULL_REVISION
 390             self.base_files = {}
 391         else:
 392             self.base_id = parents[0]
 393             self.base_files = files_cache.get(self.base_id, None)
 394             if not self.base_files:
 395                 self.base_files = copy_tree(self.base_id)
 396
 397         self.files = files_cache[revid] = self.base_files.copy()
 398
 399         for path, f in files.iteritems():
 400             fid = self.files.get(path, None)
 401             if not fid:
 402                 fid = bzrlib.generate_ids.gen_file_id(path)
 403             f['path'] = path
 404             self.updates[fid] = f
 405
 406     def last_revision(self):
 407         return self.base_id
 408
 409     def iter_changes(self):
 410         changes = []
 411
 412         def get_parent(dirname, basename):
 413             parent_fid = self.base_files.get(dirname, None)
 414             if parent_fid:
 415                 return parent_fid
 416             parent_fid = self.files.get(dirname, None)
 417             if parent_fid:
 418                 return parent_fid
 419             if basename == '':
 420                 return None
 421             fid = bzrlib.generate_ids.gen_file_id(path)
 422             d = add_entry(fid, dirname, 'directory')
 423             return fid
 424
 425         def add_entry(fid, path, kind, mode = None):
 426             dirname, basename = os.path.split(path)
 427             parent_fid = get_parent(dirname, basename)
 428
 429             executable = False
 430             if mode == '100755':
 431                 executable = True
 432             elif mode == '120000':
 433                 kind = 'symlink'
 434
 435             change = (fid,
 436                     (None, path),
 437                     True,
 438                     (False, True),
 439                     (None, parent_fid),
 440                     (None, basename),
 441                     (None, kind),
 442                     (None, executable))
 443             self.files[path] = change[0]
 444             changes.append(change)
 445             return change
 446
 447         def update_entry(fid, path, kind, mode = None):
 448             dirname, basename = os.path.split(path)
 449             parent_fid = get_parent(dirname, basename)
 450
 451             executable = False
 452             if mode == '100755':
 453                 executable = True
 454             elif mode == '120000':
 455                 kind = 'symlink'
 456
 457             change = (fid,
 458                     (path, path),
 459                     True,
 460                     (True, True),
 461                     (None, parent_fid),
 462                     (None, basename),
 463                     (None, kind),
 464                     (None, executable))
 465             self.files[path] = change[0]
 466             changes.append(change)
 467             return change
 468
 469         def remove_entry(fid, path, kind):
 470             dirname, basename = os.path.split(path)
 471             parent_fid = get_parent(dirname, basename)
 472             change = (fid,
 473                     (path, None),
 474                     True,
 475                     (True, False),
 476                     (parent_fid, None),
 477                     (None, None),
 478                     (None, None),
 479                     (None, None))
 480             del self.files[path]
 481             changes.append(change)
 482             return change
 483
 484         for fid, f in self.updates.iteritems():
 485             path = f['path']
 486
 487             if 'deleted' in f:
 488                 remove_entry(fid, path, 'file')
 489                 continue
 490
 491             if path in self.base_files:
 492                 update_entry(fid, path, 'file', f['mode'])
 493             else:
 494                 add_entry(fid, path, 'file', f['mode'])
 495
 496         return changes
 497
 498     def get_file_with_stat(self, file_id, path=None):
 499         return (StringIO.StringIO(self.updates[file_id]['data']), None)
 500
 501     def get_symlink_target(self, file_id):
 502         return self.updates[file_id]['data']
 503
 504 def parse_commit(parser):
 505     global marks, blob_marks, bmarks, parsed_refs
 506     global mode
 507
 508     parents = []
 509
 510     ref = parser[1]
 511     parser.next()
 512
 513     if ref != 'refs/heads/master':
 514         die("bzr doesn't support multiple branches; use 'master'")
 515
 516     commit_mark = parser.get_mark()
 517     parser.next()
 518     author = parser.get_author()
 519     parser.next()
 520     committer = parser.get_author()
 521     parser.next()
 522     data = parser.get_data()
 523     parser.next()
 524     if parser.check('from'):
 525         parents.append(parser.get_mark())
 526         parser.next()
 527     while parser.check('merge'):
 528         parents.append(parser.get_mark())
 529         parser.next()
 530
 531     files = {}
 532
 533     for line in parser:
 534         if parser.check('M'):
 535             t, m, mark_ref, path = line.split(' ', 3)
 536             mark = int(mark_ref[1:])
 537             f = { 'mode' : m, 'data' : blob_marks[mark] }
 538         elif parser.check('D'):
 539             t, path = line.split(' ')
 540             f = { 'deleted' : True }
 541         else:
 542             die('Unknown file command: %s' % line)
 543         files[path] = f
 544
 545     repo = parser.repo
 546
 547     committer, date, tz = committer
 548     parents = [str(mark_to_rev(p)) for p in parents]
 549     revid = bzrlib.generate_ids.gen_revision_id(committer, date)
 550     props = {}
 551     props['branch-nick'] = repo.nick
 552
 553     mtree = CustomTree(repo, revid, parents, files)
 554     changes = mtree.iter_changes()
 555
 556     repo.lock_write()
 557     try:
 558         builder = repo.get_commit_builder(parents, None, date, tz, committer, props, revid)
 559         try:
 560             list(builder.record_iter_changes(mtree, mtree.last_revision(), changes))
 561             builder.finish_inventory()
 562             builder.commit(data.decode('utf-8', 'replace'))
 563         except Exception, e:
 564             builder.abort()
 565             raise
 566     finally:
 567         repo.unlock()
 568
 569     parsed_refs[ref] = revid
 570     marks.new_mark(revid, commit_mark)
 571
 572 def parse_reset(parser):
 573     global parsed_refs
 574
 575     ref = parser[1]
 576     parser.next()
 577
 578     if ref != 'refs/heads/master':
 579         die("bzr doesn't support multiple branches; use 'master'")
 580
 581     # ugh
 582     if parser.check('commit'):
 583         parse_commit(parser)
 584         return
 585     if not parser.check('from'):
 586         return
 587     from_mark = parser.get_mark()
 588     parser.next()
 589
 590     parsed_refs[ref] = mark_to_rev(from_mark)
 591
 592 def do_export(parser):
 593     global parsed_refs, dirname, peer
 594
 595     parser.next()
 596
 597     for line in parser.each_block('done'):
 598         if parser.check('blob'):
 599             parse_blob(parser)
 600         elif parser.check('commit'):
 601             parse_commit(parser)
 602         elif parser.check('reset'):
 603             parse_reset(parser)
 604         elif parser.check('tag'):
 605             pass
 606         elif parser.check('feature'):
 607             pass
 608         else:
 609             die('unhandled export command: %s' % line)
 610
 611     repo = parser.repo
 612
 613     for ref, revid in parsed_refs.iteritems():
 614         if ref == 'refs/heads/master':
 615             repo.generate_revision_history(revid, marks.get_tip('master'))
 616             revno, revid = repo.last_revision_info()
 617             if peer:
 618                 if hasattr(peer, "import_last_revision_info_and_tags"):
 619                     peer.import_last_revision_info_and_tags(repo, revno, revid)
 620                 else:
 621                     peer.import_last_revision_info(repo.repository, revno, revid)
 622                 wt = peer.bzrdir.open_workingtree()
 623             else:
 624                 wt = repo.bzrdir.open_workingtree()
 625             wt.update()
 626         print "ok %s" % ref
 627     print
 628
 629 def do_capabilities(parser):
 630     global dirname
 631
 632     print "import"
 633     print "export"
 634     print "refspec refs/heads/*:%s/heads/*" % prefix
 635
 636     path = os.path.join(dirname, 'marks-git')
 637
 638     if os.path.exists(path):
 639         print "*import-marks %s" % path
 640     print "*export-marks %s" % path
 641
 642     print
 643
 644 def do_list(parser):
 645     global tags
 646     print "? refs/heads/%s" % 'master'
 647     for tag, revid in parser.repo.tags.get_tag_dict().items():
 648         print "? refs/tags/%s" % tag
 649         tags[tag] = revid
 650     print "@refs/heads/%s HEAD" % 'master'
 651     print
 652
 653 def get_repo(url, alias):
 654     global dirname, peer
 655
 656     origin = bzrlib.bzrdir.BzrDir.open(url)
 657     branch = origin.open_branch()
 658
 659     if not isinstance(origin.transport, bzrlib.transport.local.LocalTransport):
 660         clone_path = os.path.join(dirname, 'clone')
 661         remote_branch = branch
 662         if os.path.exists(clone_path):
 663             # pull
 664             d = bzrlib.bzrdir.BzrDir.open(clone_path)
 665             branch = d.open_branch()
 666             result = branch.pull(remote_branch, [], None, False)
 667         else:
 668             # clone
 669             d = origin.sprout(clone_path, None,
 670                     hardlink=True, create_tree_if_local=False,
 671                     source_branch=remote_branch)
 672             branch = d.open_branch()
 673             branch.bind(remote_branch)
 674
 675         peer = remote_branch
 676     else:
 677         peer = None
 678
 679     return branch
 680
 681 def main(args):
 682     global marks, prefix, dirname
 683     global tags, filenodes
 684     global blob_marks
 685     global parsed_refs
 686     global files_cache
 687
 688     alias = args[1]
 689     url = args[2]
 690
 691     prefix = 'refs/bzr/%s' % alias
 692     tags = {}
 693     filenodes = {}
 694     blob_marks = {}
 695     parsed_refs = {}
 696     files_cache = {}
 697
 698     gitdir = os.environ['GIT_DIR']
 699     dirname = os.path.join(gitdir, 'bzr', alias)
 700
 701     if not os.path.exists(dirname):
 702         os.makedirs(dirname)
 703
 704     repo = get_repo(url, alias)
 705
 706     marks_path = os.path.join(dirname, 'marks-int')
 707     marks = Marks(marks_path)
 708
 709     parser = Parser(repo)
 710     for line in parser:
 711         if parser.check('capabilities'):
 712             do_capabilities(parser)
 713         elif parser.check('list'):
 714             do_list(parser)
 715         elif parser.check('import'):
 716             do_import(parser)
 717         elif parser.check('export'):
 718             do_export(parser)
 719         else:
 720             die('unhandled command: %s' % line)
 721         sys.stdout.flush()
 722
 723     marks.store()
 724
 725 sys.exit(main(sys.argv))