contrib / remote-helpers / git-remote-bzron commit remote-bzr: cleanup CustomTree (c80f4c7)
   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
  28import bzrlib.errors
  29import bzrlib.ui
  30
  31import sys
  32import os
  33import json
  34import re
  35import StringIO
  36import atexit, shutil, hashlib, urlparse, subprocess
  37
  38NAME_RE = re.compile('^([^<>]+)')
  39AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
  40RAW_AUTHOR_RE = re.compile('^(\w+) (.+)? <(.*)> (\d+) ([+-]\d+)')
  41
  42def die(msg, *args):
  43    sys.stderr.write('ERROR: %s\n' % (msg % args))
  44    sys.exit(1)
  45
  46def warn(msg, *args):
  47    sys.stderr.write('WARNING: %s\n' % (msg % args))
  48
  49def gittz(tz):
  50    return '%+03d%02d' % (tz / 3600, tz % 3600 / 60)
  51
  52class Marks:
  53
  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()
  61
  62    def load(self):
  63        if not os.path.exists(self.path):
  64            return
  65
  66        tmp = json.load(open(self.path))
  67        self.tips = tmp['tips']
  68        self.marks = tmp['marks']
  69        self.last_mark = tmp['last-mark']
  70
  71        for rev, mark in self.marks.iteritems():
  72            self.rev_marks[mark] = rev
  73
  74    def dict(self):
  75        return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
  76
  77    def store(self):
  78        json.dump(self.dict(), open(self.path, 'w'))
  79
  80    def __str__(self):
  81        return str(self.dict())
  82
  83    def from_rev(self, rev):
  84        return self.marks[rev]
  85
  86    def to_rev(self, mark):
  87        return self.rev_marks[mark]
  88
  89    def next_mark(self):
  90        self.last_mark += 1
  91        return self.last_mark
  92
  93    def get_mark(self, rev):
  94        self.last_mark += 1
  95        self.marks[rev] = self.last_mark
  96        return self.last_mark
  97
  98    def is_marked(self, rev):
  99        return rev in self.marks
 100
 101    def new_mark(self, rev, mark):
 102        self.marks[rev] = mark
 103        self.rev_marks[mark] = rev
 104        self.last_mark = mark
 105
 106    def get_tip(self, branch):
 107        return self.tips.get(branch, None)
 108
 109    def set_tip(self, branch, tip):
 110        self.tips[branch] = tip
 111
 112class Parser:
 113
 114    def __init__(self, repo):
 115        self.repo = repo
 116        self.line = self.get_line()
 117
 118    def get_line(self):
 119        return sys.stdin.readline().strip()
 120
 121    def __getitem__(self, i):
 122        return self.line.split()[i]
 123
 124    def check(self, word):
 125        return self.line.startswith(word)
 126
 127    def each_block(self, separator):
 128        while self.line != separator:
 129            yield self.line
 130            self.line = self.get_line()
 131
 132    def __iter__(self):
 133        return self.each_block('')
 134
 135    def next(self):
 136        self.line = self.get_line()
 137        if self.line == 'done':
 138            self.line = None
 139
 140    def get_mark(self):
 141        i = self.line.index(':') + 1
 142        return int(self.line[i:])
 143
 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)
 150
 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)
 160
 161def rev_to_mark(rev):
 162    global marks
 163    return marks.from_rev(rev)
 164
 165def mark_to_rev(mark):
 166    global marks
 167    return marks.to_rev(mark)
 168
 169def 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()
 180
 181    return '%s <%s>' % (name, mail)
 182
 183def get_filechanges(cur, prev):
 184    modified = {}
 185    removed = {}
 186
 187    changes = cur.changes_from(prev)
 188
 189    def u(s):
 190        return s.encode('utf-8')
 191
 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
 207
 208    return modified, removed
 209
 210def export_files(tree, files):
 211    global marks, filenodes
 212
 213    final = []
 214    for path, fid in files.iteritems():
 215        kind = tree.kind(fid)
 216
 217        h = tree.get_file_sha1(fid)
 218
 219        if kind == 'symlink':
 220            d = tree.get_symlink_target(fid)
 221            mode = '120000'
 222        elif kind == 'file':
 223
 224            if tree.is_executable(fid):
 225                mode = '100755'
 226            else:
 227                mode = '100644'
 228
 229            # is the blob already exported?
 230            if h in filenodes:
 231                mark = filenodes[h]
 232                final.append((mode, mark, path))
 233                continue
 234
 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))
 240
 241        mark = marks.next_mark()
 242        filenodes[h] = mark
 243
 244        print "blob"
 245        print "mark :%u" % mark
 246        print "data %d" % len(d)
 247        print d
 248
 249        final.append((mode, mark, path))
 250
 251    return final
 252
 253def export_branch(branch, name):
 254    global prefix
 255
 256    ref = '%s/heads/%s' % (prefix, name)
 257    tip = marks.get_tip(name)
 258
 259    repo = branch.repository
 260    repo.lock_read()
 261    revs = branch.iter_merge_sorted_revisions(None, tip, 'exclude', 'forward')
 262    count = 0
 263
 264    revs = [revid for revid, _, _, _ in revs if not marks.is_marked(revid)]
 265
 266    for revid in revs:
 267
 268        rev = repo.get_revision(revid)
 269
 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')
 282
 283        msg += '\n'
 284
 285        if len(parents) == 0:
 286            parent = bzrlib.revision.NULL_REVISION
 287        else:
 288            parent = parents[0]
 289
 290        cur_tree = repo.revision_tree(revid)
 291        prev = repo.revision_tree(parent)
 292        modified, removed = get_filechanges(cur_tree, prev)
 293
 294        modified_final = export_files(cur_tree, modified)
 295
 296        if len(parents) == 0:
 297            print 'reset %s' % ref
 298
 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
 305
 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
 316
 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
 322
 323        count += 1
 324        if (count % 100 == 0):
 325            print "progress revision %s (%d/%d)" % (revid, count, len(revs))
 326            print "#############################################################"
 327
 328    repo.unlock()
 329
 330    revid = branch.last_revision()
 331
 332    # make sure the ref is updated
 333    print "reset %s" % ref
 334    print "from :%u" % rev_to_mark(revid)
 335    print
 336
 337    marks.set_tip(name, revid)
 338
 339def export_tag(repo, name):
 340    global tags, prefix
 341
 342    ref = '%s/tags/%s' % (prefix, name)
 343    print "reset %s" % ref
 344    print "from :%u" % rev_to_mark(tags[name])
 345    print
 346
 347def do_import(parser):
 348    global dirname
 349
 350    branch = parser.repo
 351    path = os.path.join(dirname, 'marks-git')
 352
 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()
 358
 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()
 368
 369    print 'done'
 370
 371    sys.stdout.flush()
 372
 373def parse_blob(parser):
 374    global blob_marks
 375
 376    parser.next()
 377    mark = parser.get_mark()
 378    parser.next()
 379    data = parser.get_data()
 380    blob_marks[mark] = data
 381    parser.next()
 382
 383class CustomTree():
 384
 385    def __init__(self, repo, revid, parents, files):
 386        global files_cache
 387
 388        self.updates = {}
 389
 390        def copy_tree(revid):
 391            files = files_cache[revid] = {}
 392            tree = repo.repository.revision_tree(revid)
 393            repo.lock_read()
 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
 400
 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)
 409
 410        self.files = files_cache[revid] = self.base_files.copy()
 411
 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
 418
 419    def last_revision(self):
 420        return self.base_id
 421
 422    def iter_changes(self):
 423        changes = []
 424
 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
 437
 438        def add_entry(fid, path, kind, mode = None):
 439            dirname, basename = os.path.split(path)
 440            parent_fid = get_parent(dirname, basename)
 441
 442            executable = False
 443            if mode == '100755':
 444                executable = True
 445            elif mode == '120000':
 446                kind = 'symlink'
 447
 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)
 458
 459        def update_entry(fid, path, kind, mode = None):
 460            dirname, basename = os.path.split(path)
 461            parent_fid = get_parent(dirname, basename)
 462
 463            executable = False
 464            if mode == '100755':
 465                executable = True
 466            elif mode == '120000':
 467                kind = 'symlink'
 468
 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)
 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
 494        for fid, f in self.updates.iteritems():
 495            path = f['path']
 496
 497            if 'deleted' in f:
 498                remove_entry(fid, path, 'file')
 499                continue
 500
 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'])
 505
 506        return changes
 507
 508    def get_file_with_stat(self, file_id, path=None):
 509        return (StringIO.StringIO(self.updates[file_id]['data']), None)
 510
 511    def get_symlink_target(self, file_id):
 512        return self.updates[file_id]['data']
 513
 514def c_style_unescape(string):
 515    if string[0] == string[-1] == '"':
 516        return string.decode('string-escape')[1:-1]
 517    return string
 518
 519def parse_commit(parser):
 520    global marks, blob_marks, parsed_refs
 521    global mode
 522
 523    parents = []
 524
 525    ref = parser[1]
 526    parser.next()
 527
 528    if ref != 'refs/heads/master':
 529        die("bzr doesn't support multiple branches; use 'master'")
 530
 531    commit_mark = parser.get_mark()
 532    parser.next()
 533    author = parser.get_author()
 534    parser.next()
 535    committer = parser.get_author()
 536    parser.next()
 537    data = parser.get_data()
 538    parser.next()
 539    if parser.check('from'):
 540        parents.append(parser.get_mark())
 541        parser.next()
 542    while parser.check('merge'):
 543        parents.append(parser.get_mark())
 544        parser.next()
 545
 546    # fast-export adds an extra newline
 547    if data[-1] == '\n':
 548        data = data[:-1]
 549
 550    files = {}
 551
 552    for line in parser:
 553        if parser.check('M'):
 554            t, m, mark_ref, path = line.split(' ', 3)
 555            mark = int(mark_ref[1:])
 556            f = { 'mode' : m, 'data' : blob_marks[mark] }
 557        elif parser.check('D'):
 558            t, path = line.split(' ')
 559            f = { 'deleted' : True }
 560        else:
 561            die('Unknown file command: %s' % line)
 562        path = c_style_unescape(path).decode('utf-8')
 563        files[path] = f
 564
 565    repo = parser.repo
 566
 567    committer, date, tz = committer
 568    parents = [str(mark_to_rev(p)) for p in parents]
 569    revid = bzrlib.generate_ids.gen_revision_id(committer, date)
 570    props = {}
 571    props['branch-nick'] = repo.nick
 572
 573    mtree = CustomTree(repo, revid, parents, files)
 574    changes = mtree.iter_changes()
 575
 576    repo.lock_write()
 577    try:
 578        builder = repo.get_commit_builder(parents, None, date, tz, committer, props, revid)
 579        try:
 580            list(builder.record_iter_changes(mtree, mtree.last_revision(), changes))
 581            builder.finish_inventory()
 582            builder.commit(data.decode('utf-8', 'replace'))
 583        except Exception, e:
 584            builder.abort()
 585            raise
 586    finally:
 587        repo.unlock()
 588
 589    parsed_refs[ref] = revid
 590    marks.new_mark(revid, commit_mark)
 591
 592def parse_reset(parser):
 593    global parsed_refs
 594
 595    ref = parser[1]
 596    parser.next()
 597
 598    if ref != 'refs/heads/master':
 599        die("bzr doesn't support multiple branches; use 'master'")
 600
 601    # ugh
 602    if parser.check('commit'):
 603        parse_commit(parser)
 604        return
 605    if not parser.check('from'):
 606        return
 607    from_mark = parser.get_mark()
 608    parser.next()
 609
 610    parsed_refs[ref] = mark_to_rev(from_mark)
 611
 612def do_export(parser):
 613    global parsed_refs, dirname, peer
 614
 615    parser.next()
 616
 617    for line in parser.each_block('done'):
 618        if parser.check('blob'):
 619            parse_blob(parser)
 620        elif parser.check('commit'):
 621            parse_commit(parser)
 622        elif parser.check('reset'):
 623            parse_reset(parser)
 624        elif parser.check('tag'):
 625            pass
 626        elif parser.check('feature'):
 627            pass
 628        else:
 629            die('unhandled export command: %s' % line)
 630
 631    repo = parser.repo
 632
 633    for ref, revid in parsed_refs.iteritems():
 634        if ref == 'refs/heads/master':
 635            repo.generate_revision_history(revid, marks.get_tip('master'))
 636            if peer:
 637                try:
 638                    repo.push(peer, stop_revision=revid)
 639                except bzrlib.errors.DivergedBranches:
 640                    print "error %s non-fast forward" % ref
 641                    continue
 642            else:
 643                wt = repo.bzrdir.open_workingtree()
 644                wt.update()
 645        print "ok %s" % ref
 646
 647    print
 648
 649def do_capabilities(parser):
 650    global dirname
 651
 652    print "import"
 653    print "export"
 654    print "refspec refs/heads/*:%s/heads/*" % prefix
 655    print "refspec refs/tags/*:%s/tags/*" % prefix
 656
 657    path = os.path.join(dirname, 'marks-git')
 658
 659    if os.path.exists(path):
 660        print "*import-marks %s" % path
 661    print "*export-marks %s" % path
 662
 663    print
 664
 665def ref_is_valid(name):
 666    return not True in [c in name for c in '~^: \\']
 667
 668def do_list(parser):
 669    global tags
 670    print "? refs/heads/%s" % 'master'
 671
 672    branch = parser.repo
 673    branch.lock_read()
 674    for tag, revid in branch.tags.get_tag_dict().items():
 675        try:
 676            branch.revision_id_to_dotted_revno(revid)
 677        except bzrlib.errors.NoSuchRevision:
 678            continue
 679        if not ref_is_valid(tag):
 680            continue
 681        print "? refs/tags/%s" % tag
 682        tags[tag] = revid
 683    branch.unlock()
 684    print "@refs/heads/%s HEAD" % 'master'
 685    print
 686
 687def get_repo(url, alias):
 688    global dirname, peer
 689
 690    origin = bzrlib.bzrdir.BzrDir.open(url)
 691    branch = origin.open_branch()
 692
 693    if not isinstance(origin.transport, bzrlib.transport.local.LocalTransport):
 694        clone_path = os.path.join(dirname, 'clone')
 695        remote_branch = branch
 696        if os.path.exists(clone_path):
 697            # pull
 698            d = bzrlib.bzrdir.BzrDir.open(clone_path)
 699            branch = d.open_branch()
 700            result = branch.pull(remote_branch, [], None, False)
 701        else:
 702            # clone
 703            d = origin.sprout(clone_path, None,
 704                    hardlink=True, create_tree_if_local=False,
 705                    source_branch=remote_branch)
 706            branch = d.open_branch()
 707            branch.bind(remote_branch)
 708
 709        peer = remote_branch
 710    else:
 711        peer = None
 712
 713    return branch
 714
 715def fix_path(alias, orig_url):
 716    url = urlparse.urlparse(orig_url, 'file')
 717    if url.scheme != 'file' or os.path.isabs(url.path):
 718        return
 719    abs_url = urlparse.urljoin("%s/" % os.getcwd(), orig_url)
 720    cmd = ['git', 'config', 'remote.%s.url' % alias, "bzr::%s" % abs_url]
 721    subprocess.call(cmd)
 722
 723def main(args):
 724    global marks, prefix, dirname
 725    global tags, filenodes
 726    global blob_marks
 727    global parsed_refs
 728    global files_cache
 729    global is_tmp
 730
 731    alias = args[1]
 732    url = args[2]
 733
 734    tags = {}
 735    filenodes = {}
 736    blob_marks = {}
 737    parsed_refs = {}
 738    files_cache = {}
 739    marks = None
 740
 741    if alias[5:] == url:
 742        is_tmp = True
 743        alias = hashlib.sha1(alias).hexdigest()
 744    else:
 745        is_tmp = False
 746
 747    prefix = 'refs/bzr/%s' % alias
 748    gitdir = os.environ['GIT_DIR']
 749    dirname = os.path.join(gitdir, 'bzr', alias)
 750
 751    if not is_tmp:
 752        fix_path(alias, url)
 753
 754    if not os.path.exists(dirname):
 755        os.makedirs(dirname)
 756
 757    bzrlib.ui.ui_factory.be_quiet(True)
 758
 759    repo = get_repo(url, alias)
 760
 761    marks_path = os.path.join(dirname, 'marks-int')
 762    marks = Marks(marks_path)
 763
 764    parser = Parser(repo)
 765    for line in parser:
 766        if parser.check('capabilities'):
 767            do_capabilities(parser)
 768        elif parser.check('list'):
 769            do_list(parser)
 770        elif parser.check('import'):
 771            do_import(parser)
 772        elif parser.check('export'):
 773            do_export(parser)
 774        else:
 775            die('unhandled command: %s' % line)
 776        sys.stdout.flush()
 777
 778def bye():
 779    if not marks:
 780        return
 781    if not is_tmp:
 782        marks.store()
 783    else:
 784        shutil.rmtree(dirname)
 785
 786atexit.register(bye)
 787sys.exit(main(sys.argv))