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 $PATH.
   9# Then you can clone with:
  10# git clone hg::/path/to/mercurial/repo/
  11
  12from mercurial import hg, ui, bookmarks, context, util, encoding
  13
  14import re
  15import sys
  16import os
  17import json
  18import shutil
  19import subprocess
  20import urllib
  21
  22#
  23# If you want to switch to hg-git compatibility mode:
  24# git config --global remote-hg.hg-git-compat true
  25#
  26# git:
  27# Sensible defaults for git.
  28# hg bookmarks are exported as git branches, hg branches are prefixed
  29# with 'branches/', HEAD is a special case.
  30#
  31# hg:
  32# Emulate hg-git.
  33# Only hg bookmarks are exported as git branches.
  34# Commits are modified to preserve hg information and allow biridectionality.
  35#
  36
  37NAME_RE = re.compile('^([^<>]+)')
  38AUTHOR_RE = re.compile('^([^<>]+?)? ?<([^<>]*)>$')
  39AUTHOR_HG_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 gitmode(flags):
  50    return 'l' in flags and '120000' or 'x' in flags and '100755' or '100644'
  51
  52def gittz(tz):
  53    return '%+03d%02d' % (-tz / 3600, -tz % 3600 / 60)
  54
  55def hgmode(mode):
  56    m = { '0100755': 'x', '0120000': 'l' }
  57    return m.get(mode, '')
  58
  59def get_config(config):
  60    cmd = ['git', 'config', '--get', config]
  61    process = subprocess.Popen(cmd, stdout=subprocess.PIPE)
  62    output, _ = process.communicate()
  63    return output
  64
  65class Marks:
  66
  67    def __init__(self, path):
  68        self.path = path
  69        self.tips = {}
  70        self.marks = {}
  71        self.rev_marks = {}
  72        self.last_mark = 0
  73
  74        self.load()
  75
  76    def load(self):
  77        if not os.path.exists(self.path):
  78            return
  79
  80        tmp = json.load(open(self.path))
  81
  82        self.tips = tmp['tips']
  83        self.marks = tmp['marks']
  84        self.last_mark = tmp['last-mark']
  85
  86        for rev, mark in self.marks.iteritems():
  87            self.rev_marks[mark] = int(rev)
  88
  89    def dict(self):
  90        return { 'tips': self.tips, 'marks': self.marks, 'last-mark' : self.last_mark }
  91
  92    def store(self):
  93        json.dump(self.dict(), open(self.path, 'w'))
  94
  95    def __str__(self):
  96        return str(self.dict())
  97
  98    def from_rev(self, rev):
  99        return self.marks[str(rev)]
 100
 101    def to_rev(self, mark):
 102        return self.rev_marks[mark]
 103
 104    def get_mark(self, rev):
 105        self.last_mark += 1
 106        self.marks[str(rev)] = self.last_mark
 107        return self.last_mark
 108
 109    def new_mark(self, rev, mark):
 110        self.marks[str(rev)] = mark
 111        self.rev_marks[mark] = rev
 112        self.last_mark = mark
 113
 114    def is_marked(self, rev):
 115        return self.marks.has_key(str(rev))
 116
 117    def get_tip(self, branch):
 118        return self.tips.get(branch, 0)
 119
 120    def set_tip(self, branch, tip):
 121        self.tips[branch] = tip
 122
 123class Parser:
 124
 125    def __init__(self, repo):
 126        self.repo = repo
 127        self.line = self.get_line()
 128
 129    def get_line(self):
 130        return sys.stdin.readline().strip()
 131
 132    def __getitem__(self, i):
 133        return self.line.split()[i]
 134
 135    def check(self, word):
 136        return self.line.startswith(word)
 137
 138    def each_block(self, separator):
 139        while self.line != separator:
 140            yield self.line
 141            self.line = self.get_line()
 142
 143    def __iter__(self):
 144        return self.each_block('')
 145
 146    def next(self):
 147        self.line = self.get_line()
 148        if self.line == 'done':
 149            self.line = None
 150
 151    def get_mark(self):
 152        i = self.line.index(':') + 1
 153        return int(self.line[i:])
 154
 155    def get_data(self):
 156        if not self.check('data'):
 157            return None
 158        i = self.line.index(' ') + 1
 159        size = int(self.line[i:])
 160        return sys.stdin.read(size)
 161
 162    def get_author(self):
 163        global bad_mail
 164
 165        ex = None
 166        m = RAW_AUTHOR_RE.match(self.line)
 167        if not m:
 168            return None
 169        _, name, email, date, tz = m.groups()
 170        if name and 'ext:' in name:
 171            m = re.match('^(.+?) ext:\((.+)\)$', name)
 172            if m:
 173                name = m.group(1)
 174                ex = urllib.unquote(m.group(2))
 175
 176        if email != bad_mail:
 177            if name:
 178                user = '%s <%s>' % (name, email)
 179            else:
 180                user = '<%s>' % (email)
 181        else:
 182            user = name
 183
 184        if ex:
 185            user += ex
 186
 187        tz = int(tz)
 188        tz = ((tz / 100) * 3600) + ((tz % 100) * 60)
 189        return (user, int(date), -tz)
 190
 191def export_file(fc):
 192    d = fc.data()
 193    print "M %s inline %s" % (gitmode(fc.flags()), fc.path())
 194    print "data %d" % len(d)
 195    print d
 196
 197def get_filechanges(repo, ctx, parent):
 198    modified = set()
 199    added = set()
 200    removed = set()
 201
 202    cur = ctx.manifest()
 203    prev = repo[parent].manifest().copy()
 204
 205    for fn in cur:
 206        if fn in prev:
 207            if (cur.flags(fn) != prev.flags(fn) or cur[fn] != prev[fn]):
 208                modified.add(fn)
 209            del prev[fn]
 210        else:
 211            added.add(fn)
 212    removed |= set(prev.keys())
 213
 214    return added | modified, removed
 215
 216def fixup_user_git(user):
 217    name = mail = None
 218    user = user.replace('"', '')
 219    m = AUTHOR_RE.match(user)
 220    if m:
 221        name = m.group(1)
 222        mail = m.group(2).strip()
 223    else:
 224        m = NAME_RE.match(user)
 225        if m:
 226            name = m.group(1).strip()
 227    return (name, mail)
 228
 229def fixup_user_hg(user):
 230    def sanitize(name):
 231        # stole this from hg-git
 232        return re.sub('[<>\n]', '?', name.lstrip('< ').rstrip('> '))
 233
 234    m = AUTHOR_HG_RE.match(user)
 235    if m:
 236        name = sanitize(m.group(1))
 237        mail = sanitize(m.group(2))
 238        ex = m.group(3)
 239        if ex:
 240            name += ' ext:(' + urllib.quote(ex) + ')'
 241    else:
 242        name = sanitize(user)
 243        if '@' in user:
 244            mail = name
 245        else:
 246            mail = None
 247
 248    return (name, mail)
 249
 250def fixup_user(user):
 251    global mode, bad_mail
 252
 253    if mode == 'git':
 254        name, mail = fixup_user_git(user)
 255    else:
 256        name, mail = fixup_user_hg(user)
 257
 258    if not name:
 259        name = bad_name
 260    if not mail:
 261        mail = bad_mail
 262
 263    return '%s <%s>' % (name, mail)
 264
 265def get_repo(url, alias):
 266    global dirname, peer
 267
 268    myui = ui.ui()
 269    myui.setconfig('ui', 'interactive', 'off')
 270
 271    if hg.islocal(url):
 272        repo = hg.repository(myui, url)
 273    else:
 274        local_path = os.path.join(dirname, 'clone')
 275        if not os.path.exists(local_path):
 276            peer, dstpeer = hg.clone(myui, {}, url, local_path, update=False, pull=True)
 277            repo = dstpeer.local()
 278        else:
 279            repo = hg.repository(myui, local_path)
 280            peer = hg.peer(myui, {}, url)
 281            repo.pull(peer, heads=None, force=True)
 282
 283    return repo
 284
 285def rev_to_mark(rev):
 286    global marks
 287    return marks.from_rev(rev)
 288
 289def mark_to_rev(mark):
 290    global marks
 291    return marks.to_rev(mark)
 292
 293def export_ref(repo, name, kind, head):
 294    global prefix, marks, mode
 295
 296    ename = '%s/%s' % (kind, name)
 297    tip = marks.get_tip(ename)
 298
 299    # mercurial takes too much time checking this
 300    if tip and tip == head.rev():
 301        # nothing to do
 302        return
 303    revs = xrange(tip, head.rev() + 1)
 304    count = 0
 305
 306    revs = [rev for rev in revs if not marks.is_marked(rev)]
 307
 308    for rev in revs:
 309
 310        c = repo[rev]
 311        (manifest, user, (time, tz), files, desc, extra) = repo.changelog.read(c.node())
 312        rev_branch = extra['branch']
 313
 314        author = "%s %d %s" % (fixup_user(user), time, gittz(tz))
 315        if 'committer' in extra:
 316            user, time, tz = extra['committer'].rsplit(' ', 2)
 317            committer = "%s %s %s" % (user, time, gittz(int(tz)))
 318        else:
 319            committer = author
 320
 321        parents = [p for p in repo.changelog.parentrevs(rev) if p >= 0]
 322
 323        if len(parents) == 0:
 324            modified = c.manifest().keys()
 325            removed = []
 326        else:
 327            modified, removed = get_filechanges(repo, c, parents[0])
 328
 329        if mode == 'hg':
 330            extra_msg = ''
 331
 332            if rev_branch != 'default':
 333                extra_msg += 'branch : %s\n' % rev_branch
 334
 335            renames = []
 336            for f in c.files():
 337                if f not in c.manifest():
 338                    continue
 339                rename = c.filectx(f).renamed()
 340                if rename:
 341                    renames.append((rename[0], f))
 342
 343            for e in renames:
 344                extra_msg += "rename : %s => %s\n" % e
 345
 346            for key, value in extra.iteritems():
 347                if key in ('author', 'committer', 'encoding', 'message', 'branch', 'hg-git'):
 348                    continue
 349                else:
 350                    extra_msg += "extra : %s : %s\n" % (key, urllib.quote(value))
 351
 352            desc += '\n'
 353            if extra_msg:
 354                desc += '\n--HG--\n' + extra_msg
 355
 356        if len(parents) == 0 and rev:
 357            print 'reset %s/%s' % (prefix, ename)
 358
 359        print "commit %s/%s" % (prefix, ename)
 360        print "mark :%d" % (marks.get_mark(rev))
 361        print "author %s" % (author)
 362        print "committer %s" % (committer)
 363        print "data %d" % (len(desc))
 364        print desc
 365
 366        if len(parents) > 0:
 367            print "from :%s" % (rev_to_mark(parents[0]))
 368            if len(parents) > 1:
 369                print "merge :%s" % (rev_to_mark(parents[1]))
 370
 371        for f in modified:
 372            export_file(c.filectx(f))
 373        for f in removed:
 374            print "D %s" % (f)
 375        print
 376
 377        count += 1
 378        if (count % 100 == 0):
 379            print "progress revision %d '%s' (%d/%d)" % (rev, name, count, len(revs))
 380            print "#############################################################"
 381
 382    # make sure the ref is updated
 383    print "reset %s/%s" % (prefix, ename)
 384    print "from :%u" % rev_to_mark(rev)
 385    print
 386
 387    marks.set_tip(ename, rev)
 388
 389def export_tag(repo, tag):
 390    export_ref(repo, tag, 'tags', repo[tag])
 391
 392def export_bookmark(repo, bmark):
 393    head = bmarks[bmark]
 394    export_ref(repo, bmark, 'bookmarks', head)
 395
 396def export_branch(repo, branch):
 397    tip = get_branch_tip(repo, branch)
 398    head = repo[tip]
 399    export_ref(repo, branch, 'branches', head)
 400
 401def export_head(repo):
 402    global g_head
 403    export_ref(repo, g_head[0], 'bookmarks', g_head[1])
 404
 405def do_capabilities(parser):
 406    global prefix, dirname
 407
 408    print "import"
 409    print "export"
 410    print "refspec refs/heads/branches/*:%s/branches/*" % prefix
 411    print "refspec refs/heads/*:%s/bookmarks/*" % prefix
 412    print "refspec refs/tags/*:%s/tags/*" % prefix
 413
 414    path = os.path.join(dirname, 'marks-git')
 415
 416    if os.path.exists(path):
 417        print "*import-marks %s" % path
 418    print "*export-marks %s" % path
 419
 420    print
 421
 422def get_branch_tip(repo, branch):
 423    global branches
 424
 425    heads = branches.get(branch, None)
 426    if not heads:
 427        return None
 428
 429    # verify there's only one head
 430    if (len(heads) > 1):
 431        warn("Branch '%s' has more than one head, consider merging" % branch)
 432        # older versions of mercurial don't have this
 433        if hasattr(repo, "branchtip"):
 434            return repo.branchtip(branch)
 435
 436    return heads[0]
 437
 438def list_head(repo, cur):
 439    global g_head, bmarks
 440
 441    head = bookmarks.readcurrent(repo)
 442    if head:
 443        node = repo[head]
 444    else:
 445        # fake bookmark from current branch
 446        head = cur
 447        node = repo['.']
 448        if not node:
 449            node = repo['tip']
 450        if not node:
 451            return
 452        if head == 'default':
 453            head = 'master'
 454        bmarks[head] = node
 455
 456    print "@refs/heads/%s HEAD" % head
 457    g_head = (head, node)
 458
 459def do_list(parser):
 460    global branches, bmarks, mode, track_branches
 461
 462    repo = parser.repo
 463    for bmark, node in bookmarks.listbookmarks(repo).iteritems():
 464        bmarks[bmark] = repo[node]
 465
 466    cur = repo.dirstate.branch()
 467
 468    list_head(repo, cur)
 469
 470    if track_branches:
 471        for branch in repo.branchmap():
 472            heads = repo.branchheads(branch)
 473            if len(heads):
 474                branches[branch] = heads
 475
 476        for branch in branches:
 477            print "? refs/heads/branches/%s" % branch
 478
 479    for bmark in bmarks:
 480        print "? refs/heads/%s" % bmark
 481
 482    for tag, node in repo.tagslist():
 483        if tag == 'tip':
 484            continue
 485        print "? refs/tags/%s" % tag
 486
 487    print
 488
 489def do_import(parser):
 490    repo = parser.repo
 491
 492    path = os.path.join(dirname, 'marks-git')
 493
 494    print "feature done"
 495    if os.path.exists(path):
 496        print "feature import-marks=%s" % path
 497    print "feature export-marks=%s" % path
 498    sys.stdout.flush()
 499
 500    tmp = encoding.encoding
 501    encoding.encoding = 'utf-8'
 502
 503    # lets get all the import lines
 504    while parser.check('import'):
 505        ref = parser[1]
 506
 507        if (ref == 'HEAD'):
 508            export_head(repo)
 509        elif ref.startswith('refs/heads/branches/'):
 510            branch = ref[len('refs/heads/branches/'):]
 511            export_branch(repo, branch)
 512        elif ref.startswith('refs/heads/'):
 513            bmark = ref[len('refs/heads/'):]
 514            export_bookmark(repo, bmark)
 515        elif ref.startswith('refs/tags/'):
 516            tag = ref[len('refs/tags/'):]
 517            export_tag(repo, tag)
 518
 519        parser.next()
 520
 521    encoding.encoding = tmp
 522
 523    print 'done'
 524
 525def parse_blob(parser):
 526    global blob_marks
 527
 528    parser.next()
 529    mark = parser.get_mark()
 530    parser.next()
 531    data = parser.get_data()
 532    blob_marks[mark] = data
 533    parser.next()
 534    return
 535
 536def get_merge_files(repo, p1, p2, files):
 537    for e in repo[p1].files():
 538        if e not in files:
 539            if e not in repo[p1].manifest():
 540                continue
 541            f = { 'ctx' : repo[p1][e] }
 542            files[e] = f
 543
 544def parse_commit(parser):
 545    global marks, blob_marks, bmarks, parsed_refs
 546    global mode
 547
 548    from_mark = merge_mark = None
 549
 550    ref = parser[1]
 551    parser.next()
 552
 553    commit_mark = parser.get_mark()
 554    parser.next()
 555    author = parser.get_author()
 556    parser.next()
 557    committer = parser.get_author()
 558    parser.next()
 559    data = parser.get_data()
 560    parser.next()
 561    if parser.check('from'):
 562        from_mark = parser.get_mark()
 563        parser.next()
 564    if parser.check('merge'):
 565        merge_mark = parser.get_mark()
 566        parser.next()
 567        if parser.check('merge'):
 568            die('octopus merges are not supported yet')
 569
 570    files = {}
 571
 572    for line in parser:
 573        if parser.check('M'):
 574            t, m, mark_ref, path = line.split(' ', 3)
 575            mark = int(mark_ref[1:])
 576            f = { 'mode' : hgmode(m), 'data' : blob_marks[mark] }
 577        elif parser.check('D'):
 578            t, path = line.split(' ')
 579            f = { 'deleted' : True }
 580        else:
 581            die('Unknown file command: %s' % line)
 582        files[path] = f
 583
 584    def getfilectx(repo, memctx, f):
 585        of = files[f]
 586        if 'deleted' in of:
 587            raise IOError
 588        if 'ctx' in of:
 589            return of['ctx']
 590        is_exec = of['mode'] == 'x'
 591        is_link = of['mode'] == 'l'
 592        rename = of.get('rename', None)
 593        return context.memfilectx(f, of['data'],
 594                is_link, is_exec, rename)
 595
 596    repo = parser.repo
 597
 598    user, date, tz = author
 599    extra = {}
 600
 601    if committer != author:
 602        extra['committer'] = "%s %u %u" % committer
 603
 604    if from_mark:
 605        p1 = repo.changelog.node(mark_to_rev(from_mark))
 606    else:
 607        p1 = '\0' * 20
 608
 609    if merge_mark:
 610        p2 = repo.changelog.node(mark_to_rev(merge_mark))
 611    else:
 612        p2 = '\0' * 20
 613
 614    #
 615    # If files changed from any of the parents, hg wants to know, but in git if
 616    # nothing changed from the first parent, nothing changed.
 617    #
 618    if merge_mark:
 619        get_merge_files(repo, p1, p2, files)
 620
 621    if mode == 'hg':
 622        i = data.find('\n--HG--\n')
 623        if i >= 0:
 624            tmp = data[i + len('\n--HG--\n'):].strip()
 625            for k, v in [e.split(' : ') for e in tmp.split('\n')]:
 626                if k == 'rename':
 627                    old, new = v.split(' => ', 1)
 628                    files[new]['rename'] = old
 629                elif k == 'branch':
 630                    extra[k] = v
 631                elif k == 'extra':
 632                    ek, ev = v.split(' : ', 1)
 633                    extra[ek] = urllib.unquote(ev)
 634            data = data[:i]
 635
 636    ctx = context.memctx(repo, (p1, p2), data,
 637            files.keys(), getfilectx,
 638            user, (date, tz), extra)
 639
 640    tmp = encoding.encoding
 641    encoding.encoding = 'utf-8'
 642
 643    node = repo.commitctx(ctx)
 644
 645    encoding.encoding = tmp
 646
 647    rev = repo[node].rev()
 648
 649    parsed_refs[ref] = node
 650
 651    marks.new_mark(rev, commit_mark)
 652
 653def parse_reset(parser):
 654    ref = parser[1]
 655    parser.next()
 656    # ugh
 657    if parser.check('commit'):
 658        parse_commit(parser)
 659        return
 660    if not parser.check('from'):
 661        return
 662    from_mark = parser.get_mark()
 663    parser.next()
 664
 665    node = parser.repo.changelog.node(mark_to_rev(from_mark))
 666    parsed_refs[ref] = node
 667
 668def parse_tag(parser):
 669    name = parser[1]
 670    parser.next()
 671    from_mark = parser.get_mark()
 672    parser.next()
 673    tagger = parser.get_author()
 674    parser.next()
 675    data = parser.get_data()
 676    parser.next()
 677
 678    # nothing to do
 679
 680def do_export(parser):
 681    global parsed_refs, bmarks, peer
 682
 683    parser.next()
 684
 685    for line in parser.each_block('done'):
 686        if parser.check('blob'):
 687            parse_blob(parser)
 688        elif parser.check('commit'):
 689            parse_commit(parser)
 690        elif parser.check('reset'):
 691            parse_reset(parser)
 692        elif parser.check('tag'):
 693            parse_tag(parser)
 694        elif parser.check('feature'):
 695            pass
 696        else:
 697            die('unhandled export command: %s' % line)
 698
 699    for ref, node in parsed_refs.iteritems():
 700        if ref.startswith('refs/heads/branches'):
 701            pass
 702        elif ref.startswith('refs/heads/'):
 703            bmark = ref[len('refs/heads/'):]
 704            if bmark in bmarks:
 705                old = bmarks[bmark].hex()
 706            else:
 707                old = ''
 708            if not bookmarks.pushbookmark(parser.repo, bmark, old, node):
 709                continue
 710        elif ref.startswith('refs/tags/'):
 711            tag = ref[len('refs/tags/'):]
 712            parser.repo.tag([tag], node, None, True, None, {})
 713        else:
 714            # transport-helper/fast-export bugs
 715            continue
 716        print "ok %s" % ref
 717
 718    print
 719
 720    if peer:
 721        parser.repo.push(peer, force=False)
 722
 723def main(args):
 724    global prefix, dirname, branches, bmarks
 725    global marks, blob_marks, parsed_refs
 726    global peer, mode, bad_mail, bad_name
 727    global track_branches
 728
 729    alias = args[1]
 730    url = args[2]
 731    peer = None
 732
 733    hg_git_compat = False
 734    track_branches = True
 735    try:
 736        if get_config('remote-hg.hg-git-compat') == 'true\n':
 737            hg_git_compat = True
 738            track_branches = False
 739        if get_config('remote-hg.track-branches') == 'false\n':
 740            track_branches = False
 741    except subprocess.CalledProcessError:
 742        pass
 743
 744    if hg_git_compat:
 745        mode = 'hg'
 746        bad_mail = 'none@none'
 747        bad_name = ''
 748    else:
 749        mode = 'git'
 750        bad_mail = 'unknown'
 751        bad_name = 'Unknown'
 752
 753    if alias[4:] == url:
 754        is_tmp = True
 755        alias = util.sha1(alias).hexdigest()
 756    else:
 757        is_tmp = False
 758
 759    gitdir = os.environ['GIT_DIR']
 760    dirname = os.path.join(gitdir, 'hg', alias)
 761    branches = {}
 762    bmarks = {}
 763    blob_marks = {}
 764    parsed_refs = {}
 765
 766    repo = get_repo(url, alias)
 767    prefix = 'refs/hg/%s' % alias
 768
 769    if not os.path.exists(dirname):
 770        os.makedirs(dirname)
 771
 772    marks_path = os.path.join(dirname, 'marks-hg')
 773    marks = Marks(marks_path)
 774
 775    parser = Parser(repo)
 776    for line in parser:
 777        if parser.check('capabilities'):
 778            do_capabilities(parser)
 779        elif parser.check('list'):
 780            do_list(parser)
 781        elif parser.check('import'):
 782            do_import(parser)
 783        elif parser.check('export'):
 784            do_export(parser)
 785        else:
 786            die('unhandled command: %s' % line)
 787        sys.stdout.flush()
 788
 789    if not is_tmp:
 790        marks.store()
 791    else:
 792        shutil.rmtree(dirname)
 793
 794sys.exit(main(sys.argv))