Andrew's git - gitweb.git/blob - 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
  17import sys
  18
  19import bzrlib
  20if hasattr(bzrlib, "initialize"):
  21    bzrlib.initialize()
  22
  23import bzrlib.plugin
  24bzrlib.plugin.load_plugins()
  25
  26import bzrlib.generate_ids
  27import bzrlib.transport
  28
  29import sys
  30import os
  31import json
  32import re
  33import StringIO
  34
  35NAME_RE = re.compile('^([^<>]+)')
  36AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
  37RAW_AUTHOR_RE = re.compile('^(\w+) (.+)? <(.*)> (\d+) ([+-]\d+)')
  38
  39def die(msg, *args):
  40    sys.stderr.write('ERROR: %s\n' % (msg % args))
  41    sys.exit(1)
  42
  43def warn(msg, *args):
  44    sys.stderr.write('WARNING: %s\n' % (msg % args))
  45
  46def gittz(tz):
  47    return '%+03d%02d' % (tz / 3600, tz % 3600 / 60)
  48
  49class 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
 109class 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
 158def rev_to_mark(rev):
 159    global marks
 160    return marks.from_rev(rev)
 161
 162def mark_to_rev(mark):
 163    global marks
 164    return marks.to_rev(mark)
 165
 166def 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
 180def 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
 204def 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
 247def 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        author = committer
 270        msg = rev.message.encode('utf-8')
 271
 272        msg += '\n'
 273
 274        if len(parents) == 0:
 275            parent = bzrlib.revision.NULL_REVISION
 276        else:
 277            parent = parents[0]
 278
 279        cur_tree = repo.revision_tree(revid)
 280        prev = repo.revision_tree(parent)
 281        modified, removed = get_filechanges(cur_tree, prev)
 282
 283        modified_final = export_files(cur_tree, modified)
 284
 285        if len(parents) == 0:
 286            print 'reset %s' % ref
 287
 288        print "commit %s" % ref
 289        print "mark :%d" % (marks.get_mark(revid))
 290        print "author %s" % (author)
 291        print "committer %s" % (committer)
 292        print "data %d" % (len(msg))
 293        print msg
 294
 295        for i, p in enumerate(parents):
 296            try:
 297                m = rev_to_mark(p)
 298            except KeyError:
 299                # ghost?
 300                continue
 301            if i == 0:
 302                print "from :%s" % m
 303            else:
 304                print "merge :%s" % m
 305
 306        for f in removed:
 307            print "D %s" % (f,)
 308        for f in modified_final:
 309            print "M %s :%u %s" % f
 310        print
 311
 312        count += 1
 313        if (count % 100 == 0):
 314            print "progress revision %s (%d/%d)" % (revid, count, len(revs))
 315            print "#############################################################"
 316
 317    repo.unlock()
 318
 319    revid = branch.last_revision()
 320
 321    # make sure the ref is updated
 322    print "reset %s" % ref
 323    print "from :%u" % rev_to_mark(revid)
 324    print
 325
 326    marks.set_tip(name, revid)
 327
 328def export_tag(repo, name):
 329    global tags
 330    try:
 331        print "reset refs/tags/%s" % name
 332        print "from :%u" % rev_to_mark(tags[name])
 333        print
 334    except KeyError:
 335        warn("TODO: fetch tag '%s'" % name)
 336
 337def do_import(parser):
 338    global dirname
 339
 340    branch = parser.repo
 341    path = os.path.join(dirname, 'marks-git')
 342
 343    print "feature done"
 344    if os.path.exists(path):
 345        print "feature import-marks=%s" % path
 346    print "feature export-marks=%s" % path
 347    sys.stdout.flush()
 348
 349    while parser.check('import'):
 350        ref = parser[1]
 351        if ref.startswith('refs/heads/'):
 352            name = ref[len('refs/heads/'):]
 353            export_branch(branch, name)
 354        if ref.startswith('refs/tags/'):
 355            name = ref[len('refs/tags/'):]
 356            export_tag(branch, name)
 357        parser.next()
 358
 359    print 'done'
 360
 361    sys.stdout.flush()
 362
 363def parse_blob(parser):
 364    global blob_marks
 365
 366    parser.next()
 367    mark = parser.get_mark()
 368    parser.next()
 369    data = parser.get_data()
 370    blob_marks[mark] = data
 371    parser.next()
 372
 373class CustomTree():
 374
 375    def __init__(self, repo, revid, parents, files):
 376        global files_cache
 377
 378        self.repo = repo
 379        self.revid = revid
 380        self.parents = parents
 381        self.updates = {}
 382
 383        def copy_tree(revid):
 384            files = files_cache[revid] = {}
 385            tree = repo.repository.revision_tree(revid)
 386            repo.lock_read()
 387            try:
 388                for path, entry in tree.iter_entries_by_dir():
 389                    files[path] = entry.file_id
 390            finally:
 391                repo.unlock()
 392            return files
 393
 394        if len(parents) == 0:
 395            self.base_id = bzrlib.revision.NULL_REVISION
 396            self.base_files = {}
 397        else:
 398            self.base_id = parents[0]
 399            self.base_files = files_cache.get(self.base_id, None)
 400            if not self.base_files:
 401                self.base_files = copy_tree(self.base_id)
 402
 403        self.files = files_cache[revid] = self.base_files.copy()
 404
 405        for path, f in files.iteritems():
 406            fid = self.files.get(path, None)
 407            if not fid:
 408                fid = bzrlib.generate_ids.gen_file_id(path)
 409            f['path'] = path
 410            self.updates[fid] = f
 411
 412    def last_revision(self):
 413        return self.base_id
 414
 415    def iter_changes(self):
 416        changes = []
 417
 418        def get_parent(dirname, basename):
 419            parent_fid = self.base_files.get(dirname, None)
 420            if parent_fid:
 421                return parent_fid
 422            parent_fid = self.files.get(dirname, None)
 423            if parent_fid:
 424                return parent_fid
 425            if basename == '':
 426                return None
 427            fid = bzrlib.generate_ids.gen_file_id(path)
 428            d = add_entry(fid, dirname, 'directory')
 429            return fid
 430
 431        def add_entry(fid, path, kind, mode = None):
 432            dirname, basename = os.path.split(path)
 433            parent_fid = get_parent(dirname, basename)
 434
 435            executable = False
 436            if mode == '100755':
 437                executable = True
 438            elif mode == '120000':
 439                kind = 'symlink'
 440
 441            change = (fid,
 442                    (None, path),
 443                    True,
 444                    (False, True),
 445                    (None, parent_fid),
 446                    (None, basename),
 447                    (None, kind),
 448                    (None, executable))
 449            self.files[path] = change[0]
 450            changes.append(change)
 451            return change
 452
 453        def update_entry(fid, path, kind, mode = None):
 454            dirname, basename = os.path.split(path)
 455            parent_fid = get_parent(dirname, basename)
 456
 457            executable = False
 458            if mode == '100755':
 459                executable = True
 460            elif mode == '120000':
 461                kind = 'symlink'
 462
 463            change = (fid,
 464                    (path, path),
 465                    True,
 466                    (True, True),
 467                    (None, parent_fid),
 468                    (None, basename),
 469                    (None, kind),
 470                    (None, executable))
 471            self.files[path] = change[0]
 472            changes.append(change)
 473            return change
 474
 475        def remove_entry(fid, path, kind):
 476            dirname, basename = os.path.split(path)
 477            parent_fid = get_parent(dirname, basename)
 478            change = (fid,
 479                    (path, None),
 480                    True,
 481                    (True, False),
 482                    (parent_fid, None),
 483                    (None, None),
 484                    (None, None),
 485                    (None, None))
 486            del self.files[path]
 487            changes.append(change)
 488            return change
 489
 490        for fid, f in self.updates.iteritems():
 491            path = f['path']
 492
 493            if 'deleted' in f:
 494                remove_entry(fid, path, 'file')
 495                continue
 496
 497            if path in self.base_files:
 498                update_entry(fid, path, 'file', f['mode'])
 499            else:
 500                add_entry(fid, path, 'file', f['mode'])
 501
 502        return changes
 503
 504    def get_file_with_stat(self, file_id, path=None):
 505        return (StringIO.StringIO(self.updates[file_id]['data']), None)
 506
 507    def get_symlink_target(self, file_id):
 508        return self.updates[file_id]['data']
 509
 510def parse_commit(parser):
 511    global marks, blob_marks, bmarks, parsed_refs
 512    global mode
 513
 514    parents = []
 515
 516    ref = parser[1]
 517    parser.next()
 518
 519    if ref != 'refs/heads/master':
 520        die("bzr doesn't support multiple branches; use 'master'")
 521
 522    commit_mark = parser.get_mark()
 523    parser.next()
 524    author = parser.get_author()
 525    parser.next()
 526    committer = parser.get_author()
 527    parser.next()
 528    data = parser.get_data()
 529    parser.next()
 530    if parser.check('from'):
 531        parents.append(parser.get_mark())
 532        parser.next()
 533    while parser.check('merge'):
 534        parents.append(parser.get_mark())
 535        parser.next()
 536
 537    files = {}
 538
 539    for line in parser:
 540        if parser.check('M'):
 541            t, m, mark_ref, path = line.split(' ', 3)
 542            mark = int(mark_ref[1:])
 543            f = { 'mode' : m, 'data' : blob_marks[mark] }
 544        elif parser.check('D'):
 545            t, path = line.split(' ')
 546            f = { 'deleted' : True }
 547        else:
 548            die('Unknown file command: %s' % line)
 549        files[path] = f
 550
 551    repo = parser.repo
 552
 553    committer, date, tz = committer
 554    parents = [str(mark_to_rev(p)) for p in parents]
 555    revid = bzrlib.generate_ids.gen_revision_id(committer, date)
 556    props = {}
 557    props['branch-nick'] = repo.nick
 558
 559    mtree = CustomTree(repo, revid, parents, files)
 560    changes = mtree.iter_changes()
 561
 562    repo.lock_write()
 563    try:
 564        builder = repo.get_commit_builder(parents, None, date, tz, committer, props, revid)
 565        try:
 566            list(builder.record_iter_changes(mtree, mtree.last_revision(), changes))
 567            builder.finish_inventory()
 568            builder.commit(data.decode('utf-8', 'replace'))
 569        except Exception, e:
 570            builder.abort()
 571            raise
 572    finally:
 573        repo.unlock()
 574
 575    parsed_refs[ref] = revid
 576    marks.new_mark(revid, commit_mark)
 577
 578def parse_reset(parser):
 579    global parsed_refs
 580
 581    ref = parser[1]
 582    parser.next()
 583
 584    if ref != 'refs/heads/master':
 585        die("bzr doesn't support multiple branches; use 'master'")
 586
 587    # ugh
 588    if parser.check('commit'):
 589        parse_commit(parser)
 590        return
 591    if not parser.check('from'):
 592        return
 593    from_mark = parser.get_mark()
 594    parser.next()
 595
 596    parsed_refs[ref] = mark_to_rev(from_mark)
 597
 598def do_export(parser):
 599    global parsed_refs, dirname, peer
 600
 601    parser.next()
 602
 603    for line in parser.each_block('done'):
 604        if parser.check('blob'):
 605            parse_blob(parser)
 606        elif parser.check('commit'):
 607            parse_commit(parser)
 608        elif parser.check('reset'):
 609            parse_reset(parser)
 610        elif parser.check('tag'):
 611            pass
 612        elif parser.check('feature'):
 613            pass
 614        else:
 615            die('unhandled export command: %s' % line)
 616
 617    repo = parser.repo
 618
 619    for ref, revid in parsed_refs.iteritems():
 620        if ref == 'refs/heads/master':
 621            repo.generate_revision_history(revid, marks.get_tip('master'))
 622            revno, revid = repo.last_revision_info()
 623            if peer:
 624                if hasattr(peer, "import_last_revision_info_and_tags"):
 625                    peer.import_last_revision_info_and_tags(repo, revno, revid)
 626                else:
 627                    peer.import_last_revision_info(repo.repository, revno, revid)
 628                wt = peer.bzrdir.open_workingtree()
 629            else:
 630                wt = repo.bzrdir.open_workingtree()
 631            wt.update()
 632        print "ok %s" % ref
 633    print
 634
 635def do_capabilities(parser):
 636    global dirname
 637
 638    print "import"
 639    print "export"
 640    print "refspec refs/heads/*:%s/heads/*" % prefix
 641
 642    path = os.path.join(dirname, 'marks-git')
 643
 644    if os.path.exists(path):
 645        print "*import-marks %s" % path
 646    print "*export-marks %s" % path
 647
 648    print
 649
 650def do_list(parser):
 651    global tags
 652    print "? refs/heads/%s" % 'master'
 653    for tag, revid in parser.repo.tags.get_tag_dict().items():
 654        print "? refs/tags/%s" % tag
 655        tags[tag] = revid
 656    print "@refs/heads/%s HEAD" % 'master'
 657    print
 658
 659def get_repo(url, alias):
 660    global dirname, peer
 661
 662    origin = bzrlib.bzrdir.BzrDir.open(url)
 663    branch = origin.open_branch()
 664
 665    if not isinstance(origin.transport, bzrlib.transport.local.LocalTransport):
 666        clone_path = os.path.join(dirname, 'clone')
 667        remote_branch = branch
 668        if os.path.exists(clone_path):
 669            # pull
 670            d = bzrlib.bzrdir.BzrDir.open(clone_path)
 671            branch = d.open_branch()
 672            result = branch.pull(remote_branch, [], None, False)
 673        else:
 674            # clone
 675            d = origin.sprout(clone_path, None,
 676                    hardlink=True, create_tree_if_local=False,
 677                    source_branch=remote_branch)
 678            branch = d.open_branch()
 679            branch.bind(remote_branch)
 680
 681        peer = remote_branch
 682    else:
 683        peer = None
 684
 685    return branch
 686
 687def main(args):
 688    global marks, prefix, dirname
 689    global tags, filenodes
 690    global blob_marks
 691    global parsed_refs
 692    global files_cache
 693
 694    alias = args[1]
 695    url = args[2]
 696
 697    prefix = 'refs/bzr/%s' % alias
 698    tags = {}
 699    filenodes = {}
 700    blob_marks = {}
 701    parsed_refs = {}
 702    files_cache = {}
 703
 704    gitdir = os.environ['GIT_DIR']
 705    dirname = os.path.join(gitdir, 'bzr', alias)
 706
 707    if not os.path.exists(dirname):
 708        os.makedirs(dirname)
 709
 710    repo = get_repo(url, alias)
 711
 712    marks_path = os.path.join(dirname, 'marks-int')
 713    marks = Marks(marks_path)
 714
 715    parser = Parser(repo)
 716    for line in parser:
 717        if parser.check('capabilities'):
 718            do_capabilities(parser)
 719        elif parser.check('list'):
 720            do_list(parser)
 721        elif parser.check('import'):
 722            do_import(parser)
 723        elif parser.check('export'):
 724            do_export(parser)
 725        else:
 726            die('unhandled command: %s' % line)
 727        sys.stdout.flush()
 728
 729    marks.store()
 730
 731sys.exit(main(sys.argv))