Andrew's git - gitweb.git/blob - contrib/remote-helpers/git-remote-hg

   1#!/usr/bin/env python
   2#
   3# Copyright (c) 2012 Felipe Contreras
   4#
   5
   6# Inspired by Rocco Rutte's hg-fast-export
   7
   8# Just copy to your ~/bin, or anywhere in your $PATH.
   9# Then you can clone with:
  10# git clone hg::/path/to/mercurial/repo/
  11
  12from mercurial import hg, ui, bookmarks, context, util
  13
  14import re
  15import sys
  16import os
  17import json
  18import shutil
  19
  20NAME_RE = re.compile('^([^<>]+)')
  21AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]+)>$')
  22RAW_AUTHOR_RE = re.compile('^(\w+) (?:(.+)? )?<(.+)> (\d+) ([+-]\d+)')
  23
  24def die(msg, *args):
  25    sys.stderr.write('ERROR: %s\n' % (msg % args))
  26    sys.exit(1)
  27
  28def warn(msg, *args):
  29    sys.stderr.write('WARNING: %s\n' % (msg % args))
  30
  31def gitmode(flags):
  32    return 'l' in flags and '120000' or 'x' in flags and '100755' or '100644'
  33
  34def gittz(tz):
  35    return '%+03d%02d' % (-tz / 3600, -tz % 3600 / 60)
  36
  37def hgmode(mode):
  38    m = { '0100755': 'x', '0120000': 'l' }
  39    return m.get(mode, '')
  40
  41class Marks:
  42
  43    def __init__(self, path):
  44        self.path = path
  45        self.tips = {}
  46        self.marks = {}
  47        self.rev_marks = {}
  48        self.last_mark = 0
  49
  50        self.load()
  51
  52    def load(self):
  53        if not os.path.exists(self.path):
  54            return
  55
  56        tmp = json.load(open(self.path))
  57
  58        self.tips = tmp['tips']
  59        self.marks = tmp['marks']
  60        self.last_mark = tmp['last-mark']
  61
  62        for rev, mark in self.marks.iteritems():
  63            self.rev_marks[mark] = int(rev)
  64
  65    def dict(self):
  66        return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
  67
  68    def store(self):
  69        json.dump(self.dict(), open(self.path, 'w'))
  70
  71    def __str__(self):
  72        return str(self.dict())
  73
  74    def from_rev(self, rev):
  75        return self.marks[str(rev)]
  76
  77    def to_rev(self, mark):
  78        return self.rev_marks[mark]
  79
  80    def get_mark(self, rev):
  81        self.last_mark += 1
  82        self.marks[str(rev)] = self.last_mark
  83        return self.last_mark
  84
  85    def new_mark(self, rev, mark):
  86        self.marks[str(rev)] = mark
  87        self.rev_marks[mark] = rev
  88        self.last_mark = mark
  89
  90    def is_marked(self, rev):
  91        return self.marks.has_key(str(rev))
  92
  93    def get_tip(self, branch):
  94        return self.tips.get(branch, 0)
  95
  96    def set_tip(self, branch, tip):
  97        self.tips[branch] = tip
  98
  99class Parser:
 100
 101    def __init__(self, repo):
 102        self.repo = repo
 103        self.line = self.get_line()
 104
 105    def get_line(self):
 106        return sys.stdin.readline().strip()
 107
 108    def __getitem__(self, i):
 109        return self.line.split()[i]
 110
 111    def check(self, word):
 112        return self.line.startswith(word)
 113
 114    def each_block(self, separator):
 115        while self.line != separator:
 116            yield self.line
 117            self.line = self.get_line()
 118
 119    def __iter__(self):
 120        return self.each_block('')
 121
 122    def next(self):
 123        self.line = self.get_line()
 124        if self.line == 'done':
 125            self.line = None
 126
 127    def get_mark(self):
 128        i = self.line.index(':') + 1
 129        return int(self.line[i:])
 130
 131    def get_data(self):
 132        if not self.check('data'):
 133            return None
 134        i = self.line.index(' ') + 1
 135        size = int(self.line[i:])
 136        return sys.stdin.read(size)
 137
 138    def get_author(self):
 139        m = RAW_AUTHOR_RE.match(self.line)
 140        if not m:
 141            return None
 142        _, name, email, date, tz = m.groups()
 143
 144        if email != 'unknown':
 145            if name:
 146                user = '%s <%s>' % (name, email)
 147            else:
 148                user = '<%s>' % (email)
 149        else:
 150            user = name
 151
 152        tz = int(tz)
 153        tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
 154        return (user, int(date), -tz)
 155
 156def export_file(fc):
 157    d = fc.data()
 158    print "M %s inline %s" % (gitmode(fc.flags()), fc.path())
 159    print "data %d" % len(d)
 160    print d
 161
 162def get_filechanges(repo, ctx, parent):
 163    modified = set()
 164    added = set()
 165    removed = set()
 166
 167    cur = ctx.manifest()
 168    prev = repo[parent].manifest().copy()
 169
 170    for fn in cur:
 171        if fn in prev:
 172            if (cur.flags(fn) != prev.flags(fn) or cur[fn] != prev[fn]):
 173                modified.add(fn)
 174            del prev[fn]
 175        else:
 176            added.add(fn)
 177    removed |= set(prev.keys())
 178
 179    return added | modified, removed
 180
 181def fixup_user(user):
 182    user = user.replace('"', '')
 183    name = mail = None
 184    m = AUTHOR_RE.match(user)
 185    if m:
 186        name = m.group(1)
 187        mail = m.group(2).strip()
 188    else:
 189        m = NAME_RE.match(user)
 190        if m:
 191            name = m.group(1).strip()
 192
 193    if not name:
 194        name = 'Unknown'
 195    if not mail:
 196        mail = 'unknown'
 197
 198    return '%s <%s>' % (name, mail)
 199
 200def get_repo(url, alias):
 201    global dirname, peer
 202
 203    myui = ui.ui()
 204    myui.setconfig('ui', 'interactive', 'off')
 205
 206    if hg.islocal(url):
 207        repo = hg.repository(myui, url)
 208    else:
 209        local_path = os.path.join(dirname, 'clone')
 210        if not os.path.exists(local_path):
 211            peer, dstpeer = hg.clone(myui, {}, url, local_path, update=False, pull=True)
 212            repo = dstpeer.local()
 213        else:
 214            repo = hg.repository(myui, local_path)
 215            peer = hg.peer(myui, {}, url)
 216            repo.pull(peer, heads=None, force=True)
 217
 218    return repo
 219
 220def rev_to_mark(rev):
 221    global marks
 222    return marks.from_rev(rev)
 223
 224def mark_to_rev(mark):
 225    global marks
 226    return marks.to_rev(mark)
 227
 228def export_ref(repo, name, kind, head):
 229    global prefix, marks
 230
 231    ename = '%s/%s' % (kind, name)
 232    tip = marks.get_tip(ename)
 233
 234    # mercurial takes too much time checking this
 235    if tip and tip == head.rev():
 236        # nothing to do
 237        return
 238    revs = repo.revs('%u:%u' % (tip, head))
 239    count = 0
 240
 241    revs = [rev for rev in revs if not marks.is_marked(rev)]
 242
 243    for rev in revs:
 244
 245        c = repo[rev]
 246        (manifest, user, (time, tz), files, desc, extra) = repo.changelog.read(c.node())
 247        rev_branch = extra['branch']
 248
 249        author = "%s %d %s" % (fixup_user(user), time, gittz(tz))
 250        if 'committer' in extra:
 251            user, time, tz = extra['committer'].rsplit(' ', 2)
 252            committer = "%s %s %s" % (user, time, gittz(int(tz)))
 253        else:
 254            committer = author
 255
 256        parents = [p for p in repo.changelog.parentrevs(rev) if p >= 0]
 257
 258        if len(parents) == 0:
 259            modified = c.manifest().keys()
 260            removed = []
 261        else:
 262            modified, removed = get_filechanges(repo, c, parents[0])
 263
 264        if len(parents) == 0 and rev:
 265            print 'reset %s/%s' % (prefix, ename)
 266
 267        print "commit %s/%s" % (prefix, ename)
 268        print "mark :%d" % (marks.get_mark(rev))
 269        print "author %s" % (author)
 270        print "committer %s" % (committer)
 271        print "data %d" % (len(desc))
 272        print desc
 273
 274        if len(parents) > 0:
 275            print "from :%s" % (rev_to_mark(parents[0]))
 276            if len(parents) > 1:
 277                print "merge :%s" % (rev_to_mark(parents[1]))
 278
 279        for f in modified:
 280            export_file(c.filectx(f))
 281        for f in removed:
 282            print "D %s" % (f)
 283        print
 284
 285        count += 1
 286        if (count % 100 == 0):
 287            print "progress revision %d '%s' (%d/%d)" % (rev, name, count, len(revs))
 288            print "#############################################################"
 289
 290    # make sure the ref is updated
 291    print "reset %s/%s" % (prefix, ename)
 292    print "from :%u" % rev_to_mark(rev)
 293    print
 294
 295    marks.set_tip(ename, rev)
 296
 297def export_tag(repo, tag):
 298    export_ref(repo, tag, 'tags', repo[tag])
 299
 300def export_bookmark(repo, bmark):
 301    head = bmarks[bmark]
 302    export_ref(repo, bmark, 'bookmarks', head)
 303
 304def export_branch(repo, branch):
 305    tip = get_branch_tip(repo, branch)
 306    head = repo[tip]
 307    export_ref(repo, branch, 'branches', head)
 308
 309def export_head(repo):
 310    global g_head
 311    export_ref(repo, g_head[0], 'bookmarks', g_head[1])
 312
 313def do_capabilities(parser):
 314    global prefix, dirname
 315
 316    print "import"
 317    print "export"
 318    print "refspec refs/heads/branches/*:%s/branches/*" % prefix
 319    print "refspec refs/heads/*:%s/bookmarks/*" % prefix
 320    print "refspec refs/tags/*:%s/tags/*" % prefix
 321
 322    path = os.path.join(dirname, 'marks-git')
 323
 324    if os.path.exists(path):
 325        print "*import-marks %s" % path
 326    print "*export-marks %s" % path
 327
 328    print
 329
 330def get_branch_tip(repo, branch):
 331    global branches
 332
 333    heads = branches.get(branch, None)
 334    if not heads:
 335        return None
 336
 337    # verify there's only one head
 338    if (len(heads) > 1):
 339        warn("Branch '%s' has more than one head, consider merging" % branch)
 340        # older versions of mercurial don't have this
 341        if hasattr(repo, "branchtip"):
 342            return repo.branchtip(branch)
 343
 344    return heads[0]
 345
 346def list_head(repo, cur):
 347    global g_head
 348
 349    head = bookmarks.readcurrent(repo)
 350    if not head:
 351        return
 352    node = repo[head]
 353    print "@refs/heads/%s HEAD" % head
 354    g_head = (head, node)
 355
 356def do_list(parser):
 357    global branches, bmarks
 358
 359    repo = parser.repo
 360    for branch in repo.branchmap():
 361        heads = repo.branchheads(branch)
 362        if len(heads):
 363            branches[branch] = heads
 364
 365    for bmark, node in bookmarks.listbookmarks(repo).iteritems():
 366        bmarks[bmark] = repo[node]
 367
 368    cur = repo.dirstate.branch()
 369
 370    list_head(repo, cur)
 371    for branch in branches:
 372        print "? refs/heads/branches/%s" % branch
 373    for bmark in bmarks:
 374        print "? refs/heads/%s" % bmark
 375
 376    for tag, node in repo.tagslist():
 377        if tag == 'tip':
 378            continue
 379        print "? refs/tags/%s" % tag
 380
 381    print
 382
 383def do_import(parser):
 384    repo = parser.repo
 385
 386    path = os.path.join(dirname, 'marks-git')
 387
 388    print "feature done"
 389    if os.path.exists(path):
 390        print "feature import-marks=%s" % path
 391    print "feature export-marks=%s" % path
 392    sys.stdout.flush()
 393
 394    # lets get all the import lines
 395    while parser.check('import'):
 396        ref = parser[1]
 397
 398        if (ref == 'HEAD'):
 399            export_head(repo)
 400        elif ref.startswith('refs/heads/branches/'):
 401            branch = ref[len('refs/heads/branches/'):]
 402            export_branch(repo, branch)
 403        elif ref.startswith('refs/heads/'):
 404            bmark = ref[len('refs/heads/'):]
 405            export_bookmark(repo, bmark)
 406        elif ref.startswith('refs/tags/'):
 407            tag = ref[len('refs/tags/'):]
 408            export_tag(repo, tag)
 409
 410        parser.next()
 411
 412    print 'done'
 413
 414def parse_blob(parser):
 415    global blob_marks
 416
 417    parser.next()
 418    mark = parser.get_mark()
 419    parser.next()
 420    data = parser.get_data()
 421    blob_marks[mark] = data
 422    parser.next()
 423    return
 424
 425def parse_commit(parser):
 426    global marks, blob_marks, bmarks, parsed_refs
 427
 428    from_mark = merge_mark = None
 429
 430    ref = parser[1]
 431    parser.next()
 432
 433    commit_mark = parser.get_mark()
 434    parser.next()
 435    author = parser.get_author()
 436    parser.next()
 437    committer = parser.get_author()
 438    parser.next()
 439    data = parser.get_data()
 440    parser.next()
 441    if parser.check('from'):
 442        from_mark = parser.get_mark()
 443        parser.next()
 444    if parser.check('merge'):
 445        merge_mark = parser.get_mark()
 446        parser.next()
 447        if parser.check('merge'):
 448            die('octopus merges are not supported yet')
 449
 450    files = {}
 451
 452    for line in parser:
 453        if parser.check('M'):
 454            t, m, mark_ref, path = line.split(' ')
 455            mark = int(mark_ref[1:])
 456            f = { 'mode' : hgmode(m), 'data' : blob_marks[mark] }
 457        elif parser.check('D'):
 458            t, path = line.split(' ')
 459            f = { 'deleted' : True }
 460        else:
 461            die('Unknown file command: %s' % line)
 462        files[path] = f
 463
 464    def getfilectx(repo, memctx, f):
 465        of = files[f]
 466        if 'deleted' in of:
 467            raise IOError
 468        is_exec = of['mode'] == 'x'
 469        is_link = of['mode'] == 'l'
 470        return context.memfilectx(f, of['data'], is_link, is_exec, None)
 471
 472    repo = parser.repo
 473
 474    user, date, tz = author
 475    extra = {}
 476
 477    if committer != author:
 478        extra['committer'] = "%s %u %u" % committer
 479
 480    if from_mark:
 481        p1 = repo.changelog.node(mark_to_rev(from_mark))
 482    else:
 483        p1 = '\0' * 20
 484
 485    if merge_mark:
 486        p2 = repo.changelog.node(mark_to_rev(merge_mark))
 487    else:
 488        p2 = '\0' * 20
 489
 490    ctx = context.memctx(repo, (p1, p2), data,
 491            files.keys(), getfilectx,
 492            user, (date, tz), extra)
 493
 494    node = repo.commitctx(ctx)
 495
 496    rev = repo[node].rev()
 497
 498    parsed_refs[ref] = node
 499
 500    marks.new_mark(rev, commit_mark)
 501
 502def parse_reset(parser):
 503    ref = parser[1]
 504    parser.next()
 505    # ugh
 506    if parser.check('commit'):
 507        parse_commit(parser)
 508        return
 509    if not parser.check('from'):
 510        return
 511    from_mark = parser.get_mark()
 512    parser.next()
 513
 514    node = parser.repo.changelog.node(mark_to_rev(from_mark))
 515    parsed_refs[ref] = node
 516
 517def parse_tag(parser):
 518    name = parser[1]
 519    parser.next()
 520    from_mark = parser.get_mark()
 521    parser.next()
 522    tagger = parser.get_author()
 523    parser.next()
 524    data = parser.get_data()
 525    parser.next()
 526
 527    # nothing to do
 528
 529def do_export(parser):
 530    global parsed_refs, bmarks, peer
 531
 532    parser.next()
 533
 534    for line in parser.each_block('done'):
 535        if parser.check('blob'):
 536            parse_blob(parser)
 537        elif parser.check('commit'):
 538            parse_commit(parser)
 539        elif parser.check('reset'):
 540            parse_reset(parser)
 541        elif parser.check('tag'):
 542            parse_tag(parser)
 543        elif parser.check('feature'):
 544            pass
 545        else:
 546            die('unhandled export command: %s' % line)
 547
 548    for ref, node in parsed_refs.iteritems():
 549        if ref.startswith('refs/heads/branches'):
 550            pass
 551        elif ref.startswith('refs/heads/'):
 552            bmark = ref[len('refs/heads/'):]
 553            if bmark in bmarks:
 554                old = bmarks[bmark].hex()
 555            else:
 556                old = ''
 557            if not bookmarks.pushbookmark(parser.repo, bmark, old, node):
 558                continue
 559        elif ref.startswith('refs/tags/'):
 560            tag = ref[len('refs/tags/'):]
 561            parser.repo.tag([tag], node, None, True, None, {})
 562        print "ok %s" % ref
 563
 564    print
 565
 566    if peer:
 567        parser.repo.push(peer, force=False)
 568
 569def main(args):
 570    global prefix, dirname, branches, bmarks
 571    global marks, blob_marks, parsed_refs
 572    global peer
 573
 574    alias = args[1]
 575    url = args[2]
 576    peer = None
 577
 578    if alias[4:] == url:
 579        is_tmp = True
 580        alias = util.sha1(alias).hexdigest()
 581    else:
 582        is_tmp = False
 583
 584    gitdir = os.environ['GIT_DIR']
 585    dirname = os.path.join(gitdir, 'hg', alias)
 586    branches = {}
 587    bmarks = {}
 588    blob_marks = {}
 589    parsed_refs = {}
 590
 591    repo = get_repo(url, alias)
 592    prefix = 'refs/hg/%s' % alias
 593
 594    if not os.path.exists(dirname):
 595        os.makedirs(dirname)
 596
 597    marks_path = os.path.join(dirname, 'marks-hg')
 598    marks = Marks(marks_path)
 599
 600    parser = Parser(repo)
 601    for line in parser:
 602        if parser.check('capabilities'):
 603            do_capabilities(parser)
 604        elif parser.check('list'):
 605            do_list(parser)
 606        elif parser.check('import'):
 607            do_import(parser)
 608        elif parser.check('export'):
 609            do_export(parser)
 610        else:
 611            die('unhandled command: %s' % line)
 612        sys.stdout.flush()
 613
 614    if not is_tmp:
 615        marks.store()
 616    else:
 617        shutil.rmtree(dirname)
 618
 619sys.exit(main(sys.argv))