git-svn.perlon commit git-svn: multi-init saves and reuses --tags and --branches arguments (c35b96e)
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $SVN_URL $SVN_INFO $SVN_WC $SVN_UUID
   8                $GIT_SVN_INDEX $GIT_SVN
   9                $GIT_DIR $GIT_SVN_DIR $REVDB/;
  10$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  11$VERSION = '@@GIT_VERSION@@';
  12
  13use Cwd qw/abs_path/;
  14$GIT_DIR = abs_path($ENV{GIT_DIR} || '.git');
  15$ENV{GIT_DIR} = $GIT_DIR;
  16
  17my $LC_ALL = $ENV{LC_ALL};
  18my $TZ = $ENV{TZ};
  19# make sure the svn binary gives consistent output between locales and TZs:
  20$ENV{TZ} = 'UTC';
  21$ENV{LC_ALL} = 'C';
  22$| = 1; # unbuffer STDOUT
  23
  24# If SVN:: library support is added, please make the dependencies
  25# optional and preserve the capability to use the command-line client.
  26# use eval { require SVN::... } to make it lazy load
  27# We don't use any modules not in the standard Perl distribution:
  28use Carp qw/croak/;
  29use IO::File qw//;
  30use File::Basename qw/dirname basename/;
  31use File::Path qw/mkpath/;
  32use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  33use File::Spec qw//;
  34use File::Copy qw/copy/;
  35use POSIX qw/strftime/;
  36use IPC::Open3;
  37use Memoize;
  38memoize('revisions_eq');
  39memoize('cmt_metadata');
  40memoize('get_commit_time');
  41
  42my ($SVN_PATH, $SVN, $SVN_LOG, $_use_lib);
  43$_use_lib = 1 unless $ENV{GIT_SVN_NO_LIB};
  44libsvn_load();
  45my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  46my $sha1 = qr/[a-f\d]{40}/;
  47my $sha1_short = qr/[a-f\d]{4,40}/;
  48my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
  49        $_find_copies_harder, $_l, $_cp_similarity, $_cp_remote,
  50        $_repack, $_repack_nr, $_repack_flags, $_q,
  51        $_message, $_file, $_follow_parent, $_no_metadata,
  52        $_template, $_shared, $_no_default_regex, $_no_graft_copy,
  53        $_limit, $_verbose, $_incremental, $_oneline, $_l_fmt, $_show_commit,
  54        $_version, $_upgrade, $_authors, $_branch_all_refs, @_opt_m,
  55        $_merge, $_strategy, $_dry_run, $_ignore_nodate);
  56my (@_branch_from, %tree_map, %users, %rusers, %equiv);
  57my ($_svn_co_url_revs, $_svn_pg_peg_revs);
  58my @repo_path_split_cache;
  59
  60my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
  61                'branch|b=s' => \@_branch_from,
  62                'follow-parent|follow' => \$_follow_parent,
  63                'branch-all-refs|B' => \$_branch_all_refs,
  64                'authors-file|A=s' => \$_authors,
  65                'repack:i' => \$_repack,
  66                'no-metadata' => \$_no_metadata,
  67                'quiet|q' => \$_q,
  68                'ignore-nodate' => \$_ignore_nodate,
  69                'repack-flags|repack-args|repack-opts=s' => \$_repack_flags);
  70
  71my ($_trunk, $_tags, $_branches);
  72my %multi_opts = ( 'trunk|T=s' => \$_trunk,
  73                'tags|t=s' => \$_tags,
  74                'branches|b=s' => \$_branches );
  75my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
  76my %cmt_opts = ( 'edit|e' => \$_edit,
  77                'rmdir' => \$_rmdir,
  78                'find-copies-harder' => \$_find_copies_harder,
  79                'l=i' => \$_l,
  80                'copy-similarity|C=i'=> \$_cp_similarity
  81);
  82
  83my %cmd = (
  84        fetch => [ \&fetch, "Download new revisions from SVN",
  85                        { 'revision|r=s' => \$_revision, %fc_opts } ],
  86        init => [ \&init, "Initialize a repo for tracking" .
  87                          " (requires URL argument)",
  88                          \%init_opts ],
  89        commit => [ \&commit, "Commit git revisions to SVN",
  90                        {       'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
  91        'show-ignore' => [ \&show_ignore, "Show svn:ignore listings",
  92                        { 'revision|r=i' => \$_revision } ],
  93        rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
  94                        { 'no-ignore-externals' => \$_no_ignore_ext,
  95                          'copy-remote|remote=s' => \$_cp_remote,
  96                          'upgrade' => \$_upgrade } ],
  97        'graft-branches' => [ \&graft_branches,
  98                        'Detect merges/branches from already imported history',
  99                        { 'merge-rx|m' => \@_opt_m,
 100                          'branch|b=s' => \@_branch_from,
 101                          'branch-all-refs|B' => \$_branch_all_refs,
 102                          'no-default-regex' => \$_no_default_regex,
 103                          'no-graft-copy' => \$_no_graft_copy } ],
 104        'multi-init' => [ \&multi_init,
 105                        'Initialize multiple trees (like git-svnimport)',
 106                        { %multi_opts, %fc_opts } ],
 107        'multi-fetch' => [ \&multi_fetch,
 108                        'Fetch multiple trees (like git-svnimport)',
 109                        \%fc_opts ],
 110        'log' => [ \&show_log, 'Show commit logs',
 111                        { 'limit=i' => \$_limit,
 112                          'revision|r=s' => \$_revision,
 113                          'verbose|v' => \$_verbose,
 114                          'incremental' => \$_incremental,
 115                          'oneline' => \$_oneline,
 116                          'show-commit' => \$_show_commit,
 117                          'authors-file|A=s' => \$_authors,
 118                        } ],
 119        'commit-diff' => [ \&commit_diff, 'Commit a diff between two trees',
 120                        { 'message|m=s' => \$_message,
 121                          'file|F=s' => \$_file,
 122                        %cmt_opts } ],
 123        dcommit => [ \&dcommit, 'Commit several diffs to merge with upstream',
 124                        { 'merge|m|M' => \$_merge,
 125                          'strategy|s=s' => \$_strategy,
 126                          'dry-run|n' => \$_dry_run,
 127                        %cmt_opts } ],
 128);
 129
 130my $cmd;
 131for (my $i = 0; $i < @ARGV; $i++) {
 132        if (defined $cmd{$ARGV[$i]}) {
 133                $cmd = $ARGV[$i];
 134                splice @ARGV, $i, 1;
 135                last;
 136        }
 137};
 138
 139my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 140
 141read_repo_config(\%opts);
 142my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
 143                                'version|V' => \$_version,
 144                                'id|i=s' => \$GIT_SVN);
 145exit 1 if (!$rv && $cmd ne 'log');
 146
 147set_default_vals();
 148usage(0) if $_help;
 149version() if $_version;
 150usage(1) unless defined $cmd;
 151init_vars();
 152load_authors() if $_authors;
 153load_all_refs() if $_branch_all_refs;
 154svn_compat_check() unless $_use_lib;
 155migration_check() unless $cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/;
 156$cmd{$cmd}->[0]->(@ARGV);
 157exit 0;
 158
 159####################### primary functions ######################
 160sub usage {
 161        my $exit = shift || 0;
 162        my $fd = $exit ? \*STDERR : \*STDOUT;
 163        print $fd <<"";
 164git-svn - bidirectional operations between a single Subversion tree and git
 165Usage: $0 <command> [options] [arguments]\n
 166
 167        print $fd "Available commands:\n" unless $cmd;
 168
 169        foreach (sort keys %cmd) {
 170                next if $cmd && $cmd ne $_;
 171                print $fd '  ',pack('A13',$_),$cmd{$_}->[1],"\n";
 172                foreach (keys %{$cmd{$_}->[2]}) {
 173                        # prints out arguments as they should be passed:
 174                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 175                        print $fd ' ' x 17, join(', ', map { length $_ > 1 ?
 176                                                        "--$_" : "-$_" }
 177                                                split /\|/,$_)," $x\n";
 178                }
 179        }
 180        print $fd <<"";
 181\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 182arbitrary identifier if you're tracking multiple SVN branches/repositories in
 183one git repository and want to keep them separate.  See git-svn(1) for more
 184information.
 185
 186        exit $exit;
 187}
 188
 189sub version {
 190        print "git-svn version $VERSION\n";
 191        exit 0;
 192}
 193
 194sub rebuild {
 195        if (quiet_run(qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0")) {
 196                copy_remote_ref();
 197        }
 198        $SVN_URL = shift or undef;
 199        my $newest_rev = 0;
 200        if ($_upgrade) {
 201                sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD");
 202        } else {
 203                check_upgrade_needed();
 204        }
 205
 206        my $pid = open(my $rev_list,'-|');
 207        defined $pid or croak $!;
 208        if ($pid == 0) {
 209                exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!;
 210        }
 211        my $latest;
 212        while (<$rev_list>) {
 213                chomp;
 214                my $c = $_;
 215                croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 216                my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
 217                next if (!@commit); # skip merges
 218                my ($url, $rev, $uuid) = extract_metadata($commit[$#commit]);
 219                if (!$rev || !$uuid) {
 220                        croak "Unable to extract revision or UUID from ",
 221                                "$c, $commit[$#commit]\n";
 222                }
 223
 224                # if we merged or otherwise started elsewhere, this is
 225                # how we break out of it
 226                next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
 227                next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
 228
 229                unless (defined $latest) {
 230                        if (!$SVN_URL && !$url) {
 231                                croak "SVN repository location required: $url\n";
 232                        }
 233                        $SVN_URL ||= $url;
 234                        $SVN_UUID ||= $uuid;
 235                        setup_git_svn();
 236                        $latest = $rev;
 237                }
 238                revdb_set($REVDB, $rev, $c);
 239                print "r$rev = $c\n";
 240                $newest_rev = $rev if ($rev > $newest_rev);
 241        }
 242        close $rev_list or croak $?;
 243
 244        goto out if $_use_lib;
 245        if (!chdir $SVN_WC) {
 246                svn_cmd_checkout($SVN_URL, $latest, $SVN_WC);
 247                chdir $SVN_WC or croak $!;
 248        }
 249
 250        $pid = fork;
 251        defined $pid or croak $!;
 252        if ($pid == 0) {
 253                my @svn_up = qw(svn up);
 254                push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 255                sys(@svn_up,"-r$newest_rev");
 256                $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 257                index_changes();
 258                exec('git-write-tree') or croak $!;
 259        }
 260        waitpid $pid, 0;
 261        croak $? if $?;
 262out:
 263        if ($_upgrade) {
 264                print STDERR <<"";
 265Keeping deprecated refs/head/$GIT_SVN-HEAD for now.  Please remove it
 266when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN
 267
 268        }
 269}
 270
 271sub init {
 272        my $url = shift or die "SVN repository location required " .
 273                                "as a command-line argument\n";
 274        $url =~ s!/+$!!; # strip trailing slash
 275
 276        if (my $repo_path = shift) {
 277                unless (-d $repo_path) {
 278                        mkpath([$repo_path]);
 279                }
 280                $GIT_DIR = $ENV{GIT_DIR} = $repo_path . "/.git";
 281                init_vars();
 282        }
 283
 284        $SVN_URL = $url;
 285        unless (-d $GIT_DIR) {
 286                my @init_db = ('git-init-db');
 287                push @init_db, "--template=$_template" if defined $_template;
 288                push @init_db, "--shared" if defined $_shared;
 289                sys(@init_db);
 290        }
 291        setup_git_svn();
 292}
 293
 294sub fetch {
 295        check_upgrade_needed();
 296        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 297        my $ret = $_use_lib ? fetch_lib(@_) : fetch_cmd(@_);
 298        if ($ret->{commit} && quiet_run(qw(git-rev-parse --verify
 299                                                refs/heads/master^0))) {
 300                sys(qw(git-update-ref refs/heads/master),$ret->{commit});
 301        }
 302        return $ret;
 303}
 304
 305sub fetch_cmd {
 306        my (@parents) = @_;
 307        my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 308        unless ($_revision) {
 309                $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 310        }
 311        push @log_args, "-r$_revision";
 312        push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 313
 314        my $svn_log = svn_log_raw(@log_args);
 315
 316        my $base = next_log_entry($svn_log) or croak "No base revision!\n";
 317        # don't need last_revision from grab_base_rev() because
 318        # user could've specified a different revision to skip (they
 319        # didn't want to import certain revisions into git for whatever
 320        # reason, so trust $base->{revision} instead.
 321        my (undef, $last_commit) = svn_grab_base_rev();
 322        unless (-d $SVN_WC) {
 323                svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC);
 324                chdir $SVN_WC or croak $!;
 325                read_uuid();
 326                $last_commit = git_commit($base, @parents);
 327                assert_tree($last_commit);
 328        } else {
 329                chdir $SVN_WC or croak $!;
 330                read_uuid();
 331                # looks like a user manually cp'd and svn switch'ed
 332                unless ($last_commit) {
 333                        sys(qw/svn revert -R ./);
 334                        assert_svn_wc_clean($base->{revision});
 335                        $last_commit = git_commit($base, @parents);
 336                        assert_tree($last_commit);
 337                }
 338        }
 339        my @svn_up = qw(svn up);
 340        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 341        my $last = $base;
 342        while (my $log_msg = next_log_entry($svn_log)) {
 343                if ($last->{revision} >= $log_msg->{revision}) {
 344                        croak "Out of order: last >= current: ",
 345                                "$last->{revision} >= $log_msg->{revision}\n";
 346                }
 347                # Revert is needed for cases like:
 348                # https://svn.musicpd.org/Jamming/trunk (r166:167), but
 349                # I can't seem to reproduce something like that on a test...
 350                sys(qw/svn revert -R ./);
 351                assert_svn_wc_clean($last->{revision});
 352                sys(@svn_up,"-r$log_msg->{revision}");
 353                $last_commit = git_commit($log_msg, $last_commit, @parents);
 354                $last = $log_msg;
 355        }
 356        close $svn_log->{fh};
 357        $last->{commit} = $last_commit;
 358        return $last;
 359}
 360
 361sub fetch_lib {
 362        my (@parents) = @_;
 363        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 364        my $repo;
 365        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 366        $SVN_LOG ||= libsvn_connect($repo);
 367        $SVN ||= libsvn_connect($repo);
 368        my ($last_rev, $last_commit) = svn_grab_base_rev();
 369        my ($base, $head) = libsvn_parse_revision($last_rev);
 370        if ($base > $head) {
 371                return { revision => $last_rev, commit => $last_commit }
 372        }
 373        my $index = set_index($GIT_SVN_INDEX);
 374
 375        # limit ourselves and also fork() since get_log won't release memory
 376        # after processing a revision and SVN stuff seems to leak
 377        my $inc = 1000;
 378        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
 379        read_uuid();
 380        if (defined $last_commit) {
 381                unless (-e $GIT_SVN_INDEX) {
 382                        sys(qw/git-read-tree/, $last_commit);
 383                }
 384                chomp (my $x = `git-write-tree`);
 385                my ($y) = (`git-cat-file commit $last_commit`
 386                                                        =~ /^tree ($sha1)/m);
 387                if ($y ne $x) {
 388                        unlink $GIT_SVN_INDEX or croak $!;
 389                        sys(qw/git-read-tree/, $last_commit);
 390                }
 391                chomp ($x = `git-write-tree`);
 392                if ($y ne $x) {
 393                        print STDERR "trees ($last_commit) $y != $x\n",
 394                                 "Something is seriously wrong...\n";
 395                }
 396        }
 397        while (1) {
 398                # fork, because using SVN::Pool with get_log() still doesn't
 399                # seem to help enough to keep memory usage down.
 400                defined(my $pid = fork) or croak $!;
 401                if (!$pid) {
 402                        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
 403
 404                        # Yes I'm perfectly aware that the fourth argument
 405                        # below is the limit revisions number.  Unfortunately
 406                        # performance sucks with it enabled, so it's much
 407                        # faster to fetch revision ranges instead of relying
 408                        # on the limiter.
 409                        libsvn_get_log($SVN_LOG, '/'.$SVN_PATH,
 410                                        $min, $max, 0, 1, 1,
 411                                sub {
 412                                        my $log_msg;
 413                                        if ($last_commit) {
 414                                                $log_msg = libsvn_fetch(
 415                                                        $last_commit, @_);
 416                                                $last_commit = git_commit(
 417                                                        $log_msg,
 418                                                        $last_commit,
 419                                                        @parents);
 420                                        } else {
 421                                                $log_msg = libsvn_new_tree(@_);
 422                                                $last_commit = git_commit(
 423                                                        $log_msg, @parents);
 424                                        }
 425                                });
 426                        exit 0;
 427                }
 428                waitpid $pid, 0;
 429                croak $? if $?;
 430                ($last_rev, $last_commit) = svn_grab_base_rev();
 431                last if ($max >= $head);
 432                $min = $max + 1;
 433                $max += $inc;
 434                $max = $head if ($max > $head);
 435        }
 436        restore_index($index);
 437        return { revision => $last_rev, commit => $last_commit };
 438}
 439
 440sub commit {
 441        my (@commits) = @_;
 442        check_upgrade_needed();
 443        if ($_stdin || !@commits) {
 444                print "Reading from stdin...\n";
 445                @commits = ();
 446                while (<STDIN>) {
 447                        if (/\b($sha1_short)\b/o) {
 448                                unshift @commits, $1;
 449                        }
 450                }
 451        }
 452        my @revs;
 453        foreach my $c (@commits) {
 454                chomp(my @tmp = safe_qx('git-rev-parse',$c));
 455                if (scalar @tmp == 1) {
 456                        push @revs, $tmp[0];
 457                } elsif (scalar @tmp > 1) {
 458                        push @revs, reverse (safe_qx('git-rev-list',@tmp));
 459                } else {
 460                        die "Failed to rev-parse $c\n";
 461                }
 462        }
 463        chomp @revs;
 464        $_use_lib ? commit_lib(@revs) : commit_cmd(@revs);
 465        print "Done committing ",scalar @revs," revisions to SVN\n";
 466}
 467
 468sub commit_cmd {
 469        my (@revs) = @_;
 470
 471        chdir $SVN_WC or croak "Unable to chdir $SVN_WC: $!\n";
 472        my $info = svn_info('.');
 473        my $fetched = fetch();
 474        if ($info->{Revision} != $fetched->{revision}) {
 475                print STDERR "There are new revisions that were fetched ",
 476                                "and need to be merged (or acknowledged) ",
 477                                "before committing.\n";
 478                exit 1;
 479        }
 480        $info = svn_info('.');
 481        read_uuid($info);
 482        my $last = $fetched;
 483        foreach my $c (@revs) {
 484                my $mods = svn_checkout_tree($last, $c);
 485                if (scalar @$mods == 0) {
 486                        print "Skipping, no changes detected\n";
 487                        next;
 488                }
 489                $last = svn_commit_tree($last, $c);
 490        }
 491}
 492
 493sub commit_lib {
 494        my (@revs) = @_;
 495        my ($r_last, $cmt_last) = svn_grab_base_rev();
 496        defined $r_last or die "Must have an existing revision to commit\n";
 497        my $fetched = fetch();
 498        if ($r_last != $fetched->{revision}) {
 499                print STDERR "There are new revisions that were fetched ",
 500                                "and need to be merged (or acknowledged) ",
 501                                "before committing.\n",
 502                                "last rev: $r_last\n",
 503                                " current: $fetched->{revision}\n";
 504                exit 1;
 505        }
 506        read_uuid();
 507        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 508        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
 509
 510        my $repo;
 511        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 512        set_svn_commit_env();
 513        foreach my $c (@revs) {
 514                my $log_msg = get_commit_message($c, $commit_msg);
 515
 516                # fork for each commit because there's a memory leak I
 517                # can't track down... (it's probably in the SVN code)
 518                defined(my $pid = open my $fh, '-|') or croak $!;
 519                if (!$pid) {
 520                        $SVN_LOG = libsvn_connect($repo);
 521                        $SVN = libsvn_connect($repo);
 522                        my $ed = SVN::Git::Editor->new(
 523                                        {       r => $r_last,
 524                                                ra => $SVN,
 525                                                c => $c,
 526                                                svn_path => $SVN_PATH
 527                                        },
 528                                        $SVN->get_commit_editor(
 529                                                $log_msg->{msg},
 530                                                sub {
 531                                                        libsvn_commit_cb(
 532                                                                @_, $c,
 533                                                                $log_msg->{msg},
 534                                                                $r_last,
 535                                                                $cmt_last)
 536                                                },
 537                                                @lock)
 538                                        );
 539                        my $mods = libsvn_checkout_tree($cmt_last, $c, $ed);
 540                        if (@$mods == 0) {
 541                                print "No changes\nr$r_last = $cmt_last\n";
 542                                $ed->abort_edit;
 543                        } else {
 544                                $ed->close_edit;
 545                        }
 546                        exit 0;
 547                }
 548                my ($r_new, $cmt_new, $no);
 549                while (<$fh>) {
 550                        print $_;
 551                        chomp;
 552                        if (/^r(\d+) = ($sha1)$/o) {
 553                                ($r_new, $cmt_new) = ($1, $2);
 554                        } elsif ($_ eq 'No changes') {
 555                                $no = 1;
 556                        }
 557                }
 558                close $fh or croak $?;
 559                if (! defined $r_new && ! defined $cmt_new) {
 560                        unless ($no) {
 561                                die "Failed to parse revision information\n";
 562                        }
 563                } else {
 564                        ($r_last, $cmt_last) = ($r_new, $cmt_new);
 565                }
 566        }
 567        $ENV{LC_ALL} = 'C';
 568        unlink $commit_msg;
 569}
 570
 571sub dcommit {
 572        my $gs = "refs/remotes/$GIT_SVN";
 573        chomp(my @refs = safe_qx(qw/git-rev-list --no-merges/, "$gs..HEAD"));
 574        foreach my $d (reverse @refs) {
 575                if ($_dry_run) {
 576                        print "diff-tree $d~1 $d\n";
 577                } else {
 578                        commit_diff("$d~1", $d);
 579                }
 580        }
 581        return if $_dry_run;
 582        fetch();
 583        my @diff = safe_qx(qw/git-diff-tree HEAD/, $gs);
 584        my @finish;
 585        if (@diff) {
 586                @finish = qw/rebase/;
 587                push @finish, qw/--merge/ if $_merge;
 588                push @finish, "--strategy=$_strategy" if $_strategy;
 589                print STDERR "W: HEAD and $gs differ, using @finish:\n", @diff;
 590        } else {
 591                print "No changes between current HEAD and $gs\n",
 592                      "Hard resetting to the latest $gs\n";
 593                @finish = qw/reset --hard/;
 594        }
 595        sys('git', @finish, $gs);
 596}
 597
 598sub show_ignore {
 599        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 600        $_use_lib ? show_ignore_lib() : show_ignore_cmd();
 601}
 602
 603sub show_ignore_cmd {
 604        require File::Find or die $!;
 605        if (defined $_revision) {
 606                die "-r/--revision option doesn't work unless the Perl SVN ",
 607                        "libraries are used\n";
 608        }
 609        chdir $SVN_WC or croak $!;
 610        my %ign;
 611        File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 612                s#^\./##;
 613                @{$ign{$_}} = svn_propget_base('svn:ignore', $_);
 614                }}, no_chdir=>1},'.');
 615
 616        print "\n# /\n";
 617        foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 618        delete $ign{'.'};
 619        foreach my $i (sort keys %ign) {
 620                print "\n# ",$i,"\n";
 621                foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 622        }
 623}
 624
 625sub show_ignore_lib {
 626        my $repo;
 627        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 628        $SVN ||= libsvn_connect($repo);
 629        my $r = defined $_revision ? $_revision : $SVN->get_latest_revnum;
 630        libsvn_traverse_ignore(\*STDOUT, $SVN_PATH, $r);
 631}
 632
 633sub graft_branches {
 634        my $gr_file = "$GIT_DIR/info/grafts";
 635        my ($grafts, $comments) = read_grafts($gr_file);
 636        my $gr_sha1;
 637
 638        if (%$grafts) {
 639                # temporarily disable our grafts file to make this idempotent
 640                chomp($gr_sha1 = safe_qx(qw/git-hash-object -w/,$gr_file));
 641                rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
 642        }
 643
 644        my $l_map = read_url_paths();
 645        my @re = map { qr/$_/is } @_opt_m if @_opt_m;
 646        unless ($_no_default_regex) {
 647                push @re, (qr/\b(?:merge|merging|merged)\s+with\s+([\w\.\-]+)/i,
 648                        qr/\b(?:merge|merging|merged)\s+([\w\.\-]+)/i,
 649                        qr/\b(?:from|of)\s+([\w\.\-]+)/i );
 650        }
 651        foreach my $u (keys %$l_map) {
 652                if (@re) {
 653                        foreach my $p (keys %{$l_map->{$u}}) {
 654                                graft_merge_msg($grafts,$l_map,$u,$p,@re);
 655                        }
 656                }
 657                unless ($_no_graft_copy) {
 658                        if ($_use_lib) {
 659                                graft_file_copy_lib($grafts,$l_map,$u);
 660                        } else {
 661                                graft_file_copy_cmd($grafts,$l_map,$u);
 662                        }
 663                }
 664        }
 665        graft_tree_joins($grafts);
 666
 667        write_grafts($grafts, $comments, $gr_file);
 668        unlink "$gr_file~$gr_sha1" if $gr_sha1;
 669}
 670
 671sub multi_init {
 672        my $url = shift;
 673        $_trunk ||= 'trunk';
 674        $_trunk =~ s#/+$##;
 675        $url =~ s#/+$## if $url;
 676        if ($_trunk !~ m#^[a-z\+]+://#) {
 677                $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#);
 678                unless ($url) {
 679                        print STDERR "E: '$_trunk' is not a complete URL ",
 680                                "and a separate URL is not specified\n";
 681                        exit 1;
 682                }
 683                $_trunk = $url . $_trunk;
 684        }
 685        my $ch_id;
 686        if ($GIT_SVN eq 'git-svn') {
 687                $ch_id = 1;
 688                $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
 689        }
 690        init_vars();
 691        unless (-d $GIT_SVN_DIR) {
 692                print "GIT_SVN_ID set to 'trunk' for $_trunk\n" if $ch_id;
 693                init($_trunk);
 694                sys('git-repo-config', 'svn.trunk', $_trunk);
 695        }
 696        complete_url_ls_init($url, $_branches, '--branches/-b', '');
 697        complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/');
 698}
 699
 700sub multi_fetch {
 701        # try to do trunk first, since branches/tags
 702        # may be descended from it.
 703        if (-e "$GIT_DIR/svn/trunk/info/url") {
 704                fetch_child_id('trunk', @_);
 705        }
 706        rec_fetch('', "$GIT_DIR/svn", @_);
 707}
 708
 709sub show_log {
 710        my (@args) = @_;
 711        my ($r_min, $r_max);
 712        my $r_last = -1; # prevent dupes
 713        rload_authors() if $_authors;
 714        if (defined $TZ) {
 715                $ENV{TZ} = $TZ;
 716        } else {
 717                delete $ENV{TZ};
 718        }
 719        if (defined $_revision) {
 720                if ($_revision =~ /^(\d+):(\d+)$/) {
 721                        ($r_min, $r_max) = ($1, $2);
 722                } elsif ($_revision =~ /^\d+$/) {
 723                        $r_min = $r_max = $_revision;
 724                } else {
 725                        print STDERR "-r$_revision is not supported, use ",
 726                                "standard \'git log\' arguments instead\n";
 727                        exit 1;
 728                }
 729        }
 730
 731        my $pid = open(my $log,'-|');
 732        defined $pid or croak $!;
 733        if (!$pid) {
 734                exec(git_svn_log_cmd($r_min,$r_max), @args) or croak $!;
 735        }
 736        setup_pager();
 737        my (@k, $c, $d);
 738
 739        while (<$log>) {
 740                if (/^commit ($sha1_short)/o) {
 741                        my $cmt = $1;
 742                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
 743                                $r_last = $c->{r};
 744                                process_commit($c, $r_min, $r_max, \@k) or
 745                                                                goto out;
 746                        }
 747                        $d = undef;
 748                        $c = { c => $cmt };
 749                } elsif (/^author (.+) (\d+) ([\-\+]?\d+)$/) {
 750                        get_author_info($c, $1, $2, $3);
 751                } elsif (/^(?:tree|parent|committer) /) {
 752                        # ignore
 753                } elsif (/^:\d{6} \d{6} $sha1_short/o) {
 754                        push @{$c->{raw}}, $_;
 755                } elsif (/^diff /) {
 756                        $d = 1;
 757                        push @{$c->{diff}}, $_;
 758                } elsif ($d) {
 759                        push @{$c->{diff}}, $_;
 760                } elsif (/^    (git-svn-id:.+)$/) {
 761                        (undef, $c->{r}, undef) = extract_metadata($1);
 762                } elsif (s/^    //) {
 763                        push @{$c->{l}}, $_;
 764                }
 765        }
 766        if ($c && defined $c->{r} && $c->{r} != $r_last) {
 767                $r_last = $c->{r};
 768                process_commit($c, $r_min, $r_max, \@k);
 769        }
 770        if (@k) {
 771                my $swap = $r_max;
 772                $r_max = $r_min;
 773                $r_min = $swap;
 774                process_commit($_, $r_min, $r_max) foreach reverse @k;
 775        }
 776out:
 777        close $log;
 778        print '-' x72,"\n" unless $_incremental || $_oneline;
 779}
 780
 781sub commit_diff_usage {
 782        print STDERR "Usage: $0 commit-diff <tree-ish> <tree-ish> [<URL>]\n";
 783        exit 1
 784}
 785
 786sub commit_diff {
 787        if (!$_use_lib) {
 788                print STDERR "commit-diff must be used with SVN libraries\n";
 789                exit 1;
 790        }
 791        my $ta = shift or commit_diff_usage();
 792        my $tb = shift or commit_diff_usage();
 793        if (!eval { $SVN_URL = shift || file_to_s("$GIT_SVN_DIR/info/url") }) {
 794                print STDERR "Needed URL or usable git-svn id command-line\n";
 795                commit_diff_usage();
 796        }
 797        if (defined $_message && defined $_file) {
 798                print STDERR "Both --message/-m and --file/-F specified ",
 799                                "for the commit message.\n",
 800                                "I have no idea what you mean\n";
 801                exit 1;
 802        }
 803        if (defined $_file) {
 804                $_message = file_to_s($_file);
 805        } else {
 806                $_message ||= get_commit_message($tb,
 807                                        "$GIT_DIR/.svn-commit.tmp.$$")->{msg};
 808        }
 809        my $repo;
 810        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 811        $SVN_LOG ||= libsvn_connect($repo);
 812        $SVN ||= libsvn_connect($repo);
 813        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 814        my $ed = SVN::Git::Editor->new({        r => $SVN->get_latest_revnum,
 815                                                ra => $SVN, c => $tb,
 816                                                svn_path => $SVN_PATH
 817                                        },
 818                                $SVN->get_commit_editor($_message,
 819                                        sub {print "Committed $_[0]\n"},@lock)
 820                                );
 821        my $mods = libsvn_checkout_tree($ta, $tb, $ed);
 822        if (@$mods == 0) {
 823                print "No changes\n$ta == $tb\n";
 824                $ed->abort_edit;
 825        } else {
 826                $ed->close_edit;
 827        }
 828        $_message = $_file = undef;
 829}
 830
 831########################### utility functions #########################
 832
 833sub cmt_showable {
 834        my ($c) = @_;
 835        return 1 if defined $c->{r};
 836        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
 837                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
 838                my @msg = safe_qx(qw/git-cat-file commit/, $c->{c});
 839                shift @msg while ($msg[0] ne "\n");
 840                shift @msg;
 841                @{$c->{l}} = grep !/^git-svn-id: /, @msg;
 842
 843                (undef, $c->{r}, undef) = extract_metadata(
 844                                (grep(/^git-svn-id: /, @msg))[-1]);
 845        }
 846        return defined $c->{r};
 847}
 848
 849sub git_svn_log_cmd {
 850        my ($r_min, $r_max) = @_;
 851        my @cmd = (qw/git-log --abbrev-commit --pretty=raw
 852                        --default/, "refs/remotes/$GIT_SVN");
 853        push @cmd, '--summary' if $_verbose;
 854        return @cmd unless defined $r_max;
 855        if ($r_max == $r_min) {
 856                push @cmd, '--max-count=1';
 857                if (my $c = revdb_get($REVDB, $r_max)) {
 858                        push @cmd, $c;
 859                }
 860        } else {
 861                my ($c_min, $c_max);
 862                $c_max = revdb_get($REVDB, $r_max);
 863                $c_min = revdb_get($REVDB, $r_min);
 864                if ($c_min && $c_max) {
 865                        if ($r_max > $r_max) {
 866                                push @cmd, "$c_min..$c_max";
 867                        } else {
 868                                push @cmd, "$c_max..$c_min";
 869                        }
 870                } elsif ($r_max > $r_min) {
 871                        push @cmd, $c_max;
 872                } else {
 873                        push @cmd, $c_min;
 874                }
 875        }
 876        return @cmd;
 877}
 878
 879sub fetch_child_id {
 880        my $id = shift;
 881        print "Fetching $id\n";
 882        my $ref = "$GIT_DIR/refs/remotes/$id";
 883        defined(my $pid = open my $fh, '-|') or croak $!;
 884        if (!$pid) {
 885                $_repack = undef;
 886                $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 887                init_vars();
 888                fetch(@_);
 889                exit 0;
 890        }
 891        while (<$fh>) {
 892                print $_;
 893                check_repack() if (/^r\d+ = $sha1/);
 894        }
 895        close $fh or croak $?;
 896}
 897
 898sub rec_fetch {
 899        my ($pfx, $p, @args) = @_;
 900        my @dir;
 901        foreach (sort <$p/*>) {
 902                if (-r "$_/info/url") {
 903                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
 904                        my $id = $pfx . basename $_;
 905                        next if $id eq 'trunk';
 906                        fetch_child_id($id, @args);
 907                } elsif (-d $_) {
 908                        push @dir, $_;
 909                }
 910        }
 911        foreach (@dir) {
 912                my $x = $_;
 913                $x =~ s!^\Q$GIT_DIR\E/svn/!!;
 914                rec_fetch($x, $_);
 915        }
 916}
 917
 918sub complete_url_ls_init {
 919        my ($url, $var, $switch, $pfx) = @_;
 920        unless ($var) {
 921                print STDERR "W: $switch not specified\n";
 922                return;
 923        }
 924        $var =~ s#/+$##;
 925        if ($var !~ m#^[a-z\+]+://#) {
 926                $var = '/' . $var if ($var !~ m#^/#);
 927                unless ($url) {
 928                        print STDERR "E: '$var' is not a complete URL ",
 929                                "and a separate URL is not specified\n";
 930                        exit 1;
 931                }
 932                $var = $url . $var;
 933        }
 934        chomp(my @ls = $_use_lib ? libsvn_ls_fullurl($var)
 935                                : safe_qx(qw/svn ls --non-interactive/, $var));
 936        my $old = $GIT_SVN;
 937        defined(my $pid = fork) or croak $!;
 938        if (!$pid) {
 939                foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
 940                        $u =~ s#/+$##;
 941                        if ($u !~ m!\Q$var\E/(.+)$!) {
 942                                print STDERR "W: Unrecognized URL: $u\n";
 943                                die "This should never happen\n";
 944                        }
 945                        # don't try to init already existing refs
 946                        my $id = $pfx.$1;
 947                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 948                        init_vars();
 949                        unless (-d $GIT_SVN_DIR) {
 950                                print "init $u => $id\n";
 951                                init($u);
 952                        }
 953                }
 954                exit 0;
 955        }
 956        waitpid $pid, 0;
 957        croak $? if $?;
 958        my ($n) = ($switch =~ /^--(\w+)/);
 959        sys('git-repo-config', "svn.$n", $var);
 960}
 961
 962sub common_prefix {
 963        my $paths = shift;
 964        my %common;
 965        foreach (@$paths) {
 966                my @tmp = split m#/#, $_;
 967                my $p = '';
 968                while (my $x = shift @tmp) {
 969                        $p .= "/$x";
 970                        $common{$p} ||= 0;
 971                        $common{$p}++;
 972                }
 973        }
 974        foreach (sort {length $b <=> length $a} keys %common) {
 975                if ($common{$_} == @$paths) {
 976                        return $_;
 977                }
 978        }
 979        return '';
 980}
 981
 982# grafts set here are 'stronger' in that they're based on actual tree
 983# matches, and won't be deleted from merge-base checking in write_grafts()
 984sub graft_tree_joins {
 985        my $grafts = shift;
 986        map_tree_joins() if (@_branch_from && !%tree_map);
 987        return unless %tree_map;
 988
 989        git_svn_each(sub {
 990                my $i = shift;
 991                defined(my $pid = open my $fh, '-|') or croak $!;
 992                if (!$pid) {
 993                        exec qw/git-rev-list --pretty=raw/,
 994                                        "refs/remotes/$i" or croak $!;
 995                }
 996                while (<$fh>) {
 997                        next unless /^commit ($sha1)$/o;
 998                        my $c = $1;
 999                        my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
1000                        next unless $tree_map{$t};
1001
1002                        my $l;
1003                        do {
1004                                $l = readline $fh;
1005                        } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
1006
1007                        my ($s, $tz) = ($1, $2);
1008                        if ($tz =~ s/^\+//) {
1009                                $s += tz_to_s_offset($tz);
1010                        } elsif ($tz =~ s/^\-//) {
1011                                $s -= tz_to_s_offset($tz);
1012                        }
1013
1014                        my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
1015
1016                        foreach my $p (@{$tree_map{$t}}) {
1017                                next if $p eq $c;
1018                                my $mb = eval {
1019                                        safe_qx('git-merge-base', $c, $p)
1020                                };
1021                                next unless ($@ || $?);
1022                                if (defined $r_a) {
1023                                        # see if SVN says it's a relative
1024                                        my ($url_b, $r_b, $uuid_b) =
1025                                                        cmt_metadata($p);
1026                                        next if (defined $url_b &&
1027                                                        defined $url_a &&
1028                                                        ($url_a eq $url_b) &&
1029                                                        ($uuid_a eq $uuid_b));
1030                                        if ($uuid_a eq $uuid_b) {
1031                                                if ($r_b < $r_a) {
1032                                                        $grafts->{$c}->{$p} = 2;
1033                                                        next;
1034                                                } elsif ($r_b > $r_a) {
1035                                                        $grafts->{$p}->{$c} = 2;
1036                                                        next;
1037                                                }
1038                                        }
1039                                }
1040                                my $ct = get_commit_time($p);
1041                                if ($ct < $s) {
1042                                        $grafts->{$c}->{$p} = 2;
1043                                } elsif ($ct > $s) {
1044                                        $grafts->{$p}->{$c} = 2;
1045                                }
1046                                # what should we do when $ct == $s ?
1047                        }
1048                }
1049                close $fh or croak $?;
1050        });
1051}
1052
1053# this isn't funky-filename safe, but good enough for now...
1054sub graft_file_copy_cmd {
1055        my ($grafts, $l_map, $u) = @_;
1056        my $paths = $l_map->{$u};
1057        my $pfx = common_prefix([keys %$paths]);
1058        $SVN_URL ||= $u.$pfx;
1059        my $pid = open my $fh, '-|';
1060        defined $pid or croak $!;
1061        unless ($pid) {
1062                my @exec = qw/svn log -v/;
1063                push @exec, "-r$_revision" if defined $_revision;
1064                exec @exec, $u.$pfx or croak $!;
1065        }
1066        my ($r, $mp) = (undef, undef);
1067        while (<$fh>) {
1068                chomp;
1069                if (/^\-{72}$/) {
1070                        $mp = $r = undef;
1071                } elsif (/^r(\d+) \| /) {
1072                        $r = $1 unless defined $r;
1073                } elsif (/^Changed paths:/) {
1074                        $mp = 1;
1075                } elsif ($mp && m#^   [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) {
1076                        my ($p1, $p0, $r0) = ($1, $2, $3);
1077                        my $c = find_graft_path_commit($paths, $p1, $r);
1078                        next unless $c;
1079                        find_graft_path_parents($grafts, $paths, $c, $p0, $r0);
1080                }
1081        }
1082}
1083
1084sub graft_file_copy_lib {
1085        my ($grafts, $l_map, $u) = @_;
1086        my $tree_paths = $l_map->{$u};
1087        my $pfx = common_prefix([keys %$tree_paths]);
1088        my ($repo, $path) = repo_path_split($u.$pfx);
1089        $SVN_LOG ||= libsvn_connect($repo);
1090        $SVN ||= libsvn_connect($repo);
1091
1092        my ($base, $head) = libsvn_parse_revision();
1093        my $inc = 1000;
1094        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
1095        my $eh = $SVN::Error::handler;
1096        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
1097        while (1) {
1098                my $pool = SVN::Pool->new;
1099                libsvn_get_log($SVN_LOG, "/$path", $min, $max, 0, 1, 1,
1100                        sub {
1101                                libsvn_graft_file_copies($grafts, $tree_paths,
1102                                                        $path, @_);
1103                        }, $pool);
1104                $pool->clear;
1105                last if ($max >= $head);
1106                $min = $max + 1;
1107                $max += $inc;
1108                $max = $head if ($max > $head);
1109        }
1110        $SVN::Error::handler = $eh;
1111}
1112
1113sub process_merge_msg_matches {
1114        my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
1115        my (@strong, @weak);
1116        foreach (@matches) {
1117                # merging with ourselves is not interesting
1118                next if $_ eq $p;
1119                if ($l_map->{$u}->{$_}) {
1120                        push @strong, $_;
1121                } else {
1122                        push @weak, $_;
1123                }
1124        }
1125        foreach my $w (@weak) {
1126                last if @strong;
1127                # no exact match, use branch name as regexp.
1128                my $re = qr/\Q$w\E/i;
1129                foreach (keys %{$l_map->{$u}}) {
1130                        if (/$re/) {
1131                                push @strong, $l_map->{$u}->{$_};
1132                                last;
1133                        }
1134                }
1135                last if @strong;
1136                $w = basename($w);
1137                $re = qr/\Q$w\E/i;
1138                foreach (keys %{$l_map->{$u}}) {
1139                        if (/$re/) {
1140                                push @strong, $l_map->{$u}->{$_};
1141                                last;
1142                        }
1143                }
1144        }
1145        my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
1146                                        \s(?:[a-f\d\-]+)$/xsm);
1147        unless (defined $rev) {
1148                ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
1149                                        \@(?:[a-f\d\-]+)/xsm);
1150                return unless defined $rev;
1151        }
1152        foreach my $m (@strong) {
1153                my ($r0, $s0) = find_rev_before($rev, $m, 1);
1154                $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
1155        }
1156}
1157
1158sub graft_merge_msg {
1159        my ($grafts, $l_map, $u, $p, @re) = @_;
1160
1161        my $x = $l_map->{$u}->{$p};
1162        my $rl = rev_list_raw($x);
1163        while (my $c = next_rev_list_entry($rl)) {
1164                foreach my $re (@re) {
1165                        my (@br) = ($c->{m} =~ /$re/g);
1166                        next unless @br;
1167                        process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
1168                }
1169        }
1170}
1171
1172sub read_uuid {
1173        return if $SVN_UUID;
1174        if ($_use_lib) {
1175                my $pool = SVN::Pool->new;
1176                $SVN_UUID = $SVN->get_uuid($pool);
1177                $pool->clear;
1178        } else {
1179                my $info = shift || svn_info('.');
1180                $SVN_UUID = $info->{'Repository UUID'} or
1181                                        croak "Repository UUID unreadable\n";
1182        }
1183}
1184
1185sub quiet_run {
1186        my $pid = fork;
1187        defined $pid or croak $!;
1188        if (!$pid) {
1189                open my $null, '>', '/dev/null' or croak $!;
1190                open STDERR, '>&', $null or croak $!;
1191                open STDOUT, '>&', $null or croak $!;
1192                exec @_ or croak $!;
1193        }
1194        waitpid $pid, 0;
1195        return $?;
1196}
1197
1198sub repo_path_split {
1199        my $full_url = shift;
1200        $full_url =~ s#/+$##;
1201
1202        foreach (@repo_path_split_cache) {
1203                if ($full_url =~ s#$_##) {
1204                        my $u = $1;
1205                        $full_url =~ s#^/+##;
1206                        return ($u, $full_url);
1207                }
1208        }
1209
1210        if ($_use_lib) {
1211                my $tmp = libsvn_connect($full_url);
1212                my $url = $tmp->get_repos_root;
1213                $full_url =~ s#^\Q$url\E/*##;
1214                push @repo_path_split_cache, qr/^(\Q$url\E)/;
1215                return ($url, $full_url);
1216        } else {
1217                my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i);
1218                $path =~ s#^/+##;
1219                my @paths = split(m#/+#, $path);
1220                while (quiet_run(qw/svn ls --non-interactive/, $url)) {
1221                        my $n = shift @paths || last;
1222                        $url .= "/$n";
1223                }
1224                push @repo_path_split_cache, qr/^(\Q$url\E)/;
1225                $path = join('/',@paths);
1226                return ($url, $path);
1227        }
1228}
1229
1230sub setup_git_svn {
1231        defined $SVN_URL or croak "SVN repository location required\n";
1232        unless (-d $GIT_DIR) {
1233                croak "GIT_DIR=$GIT_DIR does not exist!\n";
1234        }
1235        mkpath([$GIT_SVN_DIR]);
1236        mkpath(["$GIT_SVN_DIR/info"]);
1237        open my $fh, '>>',$REVDB or croak $!;
1238        close $fh;
1239        s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1240
1241}
1242
1243sub assert_svn_wc_clean {
1244        return if $_use_lib;
1245        my ($svn_rev) = @_;
1246        croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
1247        my $lcr = svn_info('.')->{'Last Changed Rev'};
1248        if ($svn_rev != $lcr) {
1249                print STDERR "Checking for copy-tree ... ";
1250                my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
1251                                                "-r$lcr:$svn_rev")));
1252                if (@diff) {
1253                        croak "Nope!  Expected r$svn_rev, got r$lcr\n";
1254                } else {
1255                        print STDERR "OK!\n";
1256                }
1257        }
1258        my @status = grep(!/^Performing status on external/,(`svn status`));
1259        @status = grep(!/^\s*$/,@status);
1260        @status = grep(!/^X/,@status) if $_no_ignore_ext;
1261        if (scalar @status) {
1262                print STDERR "Tree ($SVN_WC) is not clean:\n";
1263                print STDERR $_ foreach @status;
1264                croak;
1265        }
1266}
1267
1268sub get_tree_from_treeish {
1269        my ($treeish) = @_;
1270        croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1271        chomp(my $type = `git-cat-file -t $treeish`);
1272        my $expected;
1273        while ($type eq 'tag') {
1274                chomp(($treeish, $type) = `git-cat-file tag $treeish`);
1275        }
1276        if ($type eq 'commit') {
1277                $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
1278                ($expected) = ($expected =~ /^tree ($sha1)$/);
1279                die "Unable to get tree from $treeish\n" unless $expected;
1280        } elsif ($type eq 'tree') {
1281                $expected = $treeish;
1282        } else {
1283                die "$treeish is a $type, expected tree, tag or commit\n";
1284        }
1285        return $expected;
1286}
1287
1288sub assert_tree {
1289        return if $_use_lib;
1290        my ($treeish) = @_;
1291        my $expected = get_tree_from_treeish($treeish);
1292
1293        my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
1294        if (-e $tmpindex) {
1295                unlink $tmpindex or croak $!;
1296        }
1297        my $old_index = set_index($tmpindex);
1298        index_changes(1);
1299        chomp(my $tree = `git-write-tree`);
1300        restore_index($old_index);
1301        if ($tree ne $expected) {
1302                croak "Tree mismatch, Got: $tree, Expected: $expected\n";
1303        }
1304        unlink $tmpindex;
1305}
1306
1307sub parse_diff_tree {
1308        my $diff_fh = shift;
1309        local $/ = "\0";
1310        my $state = 'meta';
1311        my @mods;
1312        while (<$diff_fh>) {
1313                chomp $_; # this gets rid of the trailing "\0"
1314                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1315                                        $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1316                        push @mods, {   mode_a => $1, mode_b => $2,
1317                                        sha1_b => $3, chg => $4 };
1318                        if ($4 =~ /^(?:C|R)$/) {
1319                                $state = 'file_a';
1320                        } else {
1321                                $state = 'file_b';
1322                        }
1323                } elsif ($state eq 'file_a') {
1324                        my $x = $mods[$#mods] or croak "Empty array\n";
1325                        if ($x->{chg} !~ /^(?:C|R)$/) {
1326                                croak "Error parsing $_, $x->{chg}\n";
1327                        }
1328                        $x->{file_a} = $_;
1329                        $state = 'file_b';
1330                } elsif ($state eq 'file_b') {
1331                        my $x = $mods[$#mods] or croak "Empty array\n";
1332                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1333                                croak "Error parsing $_, $x->{chg}\n";
1334                        }
1335                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1336                                croak "Error parsing $_, $x->{chg}\n";
1337                        }
1338                        $x->{file_b} = $_;
1339                        $state = 'meta';
1340                } else {
1341                        croak "Error parsing $_\n";
1342                }
1343        }
1344        close $diff_fh or croak $?;
1345
1346        return \@mods;
1347}
1348
1349sub svn_check_prop_executable {
1350        my $m = shift;
1351        return if -l $m->{file_b};
1352        if ($m->{mode_b} =~ /755$/) {
1353                chmod((0755 &~ umask),$m->{file_b}) or croak $!;
1354                if ($m->{mode_a} !~ /755$/) {
1355                        sys(qw(svn propset svn:executable 1), $m->{file_b});
1356                }
1357                -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
1358        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
1359                sys(qw(svn propdel svn:executable), $m->{file_b});
1360                chmod((0644 &~ umask),$m->{file_b}) or croak $!;
1361                -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
1362        }
1363}
1364
1365sub svn_ensure_parent_path {
1366        my $dir_b = dirname(shift);
1367        svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
1368        mkpath([$dir_b]) unless (-d $dir_b);
1369        sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
1370}
1371
1372sub precommit_check {
1373        my $mods = shift;
1374        my (%rm_file, %rmdir_check, %added_check);
1375
1376        my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
1377        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1378                if ($m->{chg} eq 'R') {
1379                        if (-d $m->{file_b}) {
1380                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1381                        }
1382                        # dir/$file => dir/file/$file
1383                        my $dirname = dirname($m->{file_b});
1384                        while ($dirname ne File::Spec->curdir) {
1385                                if ($dirname ne $m->{file_a}) {
1386                                        $dirname = dirname($dirname);
1387                                        next;
1388                                }
1389                                err_file_to_dir("$m->{file_a} => $m->{file_b}");
1390                        }
1391                        # baz/zzz => baz (baz is a file)
1392                        $dirname = dirname($m->{file_a});
1393                        while ($dirname ne File::Spec->curdir) {
1394                                if ($dirname ne $m->{file_b}) {
1395                                        $dirname = dirname($dirname);
1396                                        next;
1397                                }
1398                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1399                        }
1400                }
1401                if ($m->{chg} =~ /^(D|R)$/) {
1402                        my $t = $1 eq 'D' ? 'file_b' : 'file_a';
1403                        $rm_file{ $m->{$t} } = 1;
1404                        my $dirname = dirname( $m->{$t} );
1405                        my $basename = basename( $m->{$t} );
1406                        $rmdir_check{$dirname}->{$basename} = 1;
1407                } elsif ($m->{chg} =~ /^(?:A|C)$/) {
1408                        if (-d $m->{file_b}) {
1409                                err_dir_to_file($m->{file_b});
1410                        }
1411                        my $dirname = dirname( $m->{file_b} );
1412                        my $basename = basename( $m->{file_b} );
1413                        $added_check{$dirname}->{$basename} = 1;
1414                        while ($dirname ne File::Spec->curdir) {
1415                                if ($rm_file{$dirname}) {
1416                                        err_file_to_dir($m->{file_b});
1417                                }
1418                                $dirname = dirname $dirname;
1419                        }
1420                }
1421        }
1422        return (\%rmdir_check, \%added_check);
1423
1424        sub err_dir_to_file {
1425                my $file = shift;
1426                print STDERR "Node change from directory to file ",
1427                                "is not supported by Subversion: ",$file,"\n";
1428                exit 1;
1429        }
1430        sub err_file_to_dir {
1431                my $file = shift;
1432                print STDERR "Node change from file to directory ",
1433                                "is not supported by Subversion: ",$file,"\n";
1434                exit 1;
1435        }
1436}
1437
1438
1439sub get_diff {
1440        my ($from, $treeish) = @_;
1441        assert_tree($from);
1442        print "diff-tree $from $treeish\n";
1443        my $pid = open my $diff_fh, '-|';
1444        defined $pid or croak $!;
1445        if ($pid == 0) {
1446                my @diff_tree = qw(git-diff-tree -z -r);
1447                if ($_cp_similarity) {
1448                        push @diff_tree, "-C$_cp_similarity";
1449                } else {
1450                        push @diff_tree, '-C';
1451                }
1452                push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1453                push @diff_tree, "-l$_l" if defined $_l;
1454                exec(@diff_tree, $from, $treeish) or croak $!;
1455        }
1456        return parse_diff_tree($diff_fh);
1457}
1458
1459sub svn_checkout_tree {
1460        my ($from, $treeish) = @_;
1461        my $mods = get_diff($from->{commit}, $treeish);
1462        return $mods unless (scalar @$mods);
1463        my ($rm, $add) = precommit_check($mods);
1464
1465        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1466        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1467                if ($m->{chg} eq 'C') {
1468                        svn_ensure_parent_path( $m->{file_b} );
1469                        sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
1470                        apply_mod_line_blob($m);
1471                        svn_check_prop_executable($m);
1472                } elsif ($m->{chg} eq 'D') {
1473                        sys(qw(svn rm --force), $m->{file_b});
1474                } elsif ($m->{chg} eq 'R') {
1475                        svn_ensure_parent_path( $m->{file_b} );
1476                        sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
1477                        apply_mod_line_blob($m);
1478                        svn_check_prop_executable($m);
1479                } elsif ($m->{chg} eq 'M') {
1480                        apply_mod_line_blob($m);
1481                        svn_check_prop_executable($m);
1482                } elsif ($m->{chg} eq 'T') {
1483                        sys(qw(svn rm --force),$m->{file_b});
1484                        apply_mod_line_blob($m);
1485                        sys(qw(svn add), $m->{file_b});
1486                        svn_check_prop_executable($m);
1487                } elsif ($m->{chg} eq 'A') {
1488                        svn_ensure_parent_path( $m->{file_b} );
1489                        apply_mod_line_blob($m);
1490                        sys(qw(svn add), $m->{file_b});
1491                        svn_check_prop_executable($m);
1492                } else {
1493                        croak "Invalid chg: $m->{chg}\n";
1494                }
1495        }
1496
1497        assert_tree($treeish);
1498        if ($_rmdir) { # remove empty directories
1499                handle_rmdir($rm, $add);
1500        }
1501        assert_tree($treeish);
1502        return $mods;
1503}
1504
1505sub libsvn_checkout_tree {
1506        my ($from, $treeish, $ed) = @_;
1507        my $mods = get_diff($from, $treeish);
1508        return $mods unless (scalar @$mods);
1509        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1510        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1511                my $f = $m->{chg};
1512                if (defined $o{$f}) {
1513                        $ed->$f($m, $_q);
1514                } else {
1515                        croak "Invalid change type: $f\n";
1516                }
1517        }
1518        $ed->rmdirs($_q) if $_rmdir;
1519        return $mods;
1520}
1521
1522# svn ls doesn't work with respect to the current working tree, but what's
1523# in the repository.  There's not even an option for it... *sigh*
1524# (added files don't show up and removed files remain in the ls listing)
1525sub svn_ls_current {
1526        my ($dir, $rm, $add) = @_;
1527        chomp(my @ls = safe_qx('svn','ls',$dir));
1528        my @ret = ();
1529        foreach (@ls) {
1530                s#/$##; # trailing slashes are evil
1531                push @ret, $_ unless $rm->{$dir}->{$_};
1532        }
1533        if (exists $add->{$dir}) {
1534                push @ret, keys %{$add->{$dir}};
1535        }
1536        return \@ret;
1537}
1538
1539sub handle_rmdir {
1540        my ($rm, $add) = @_;
1541
1542        foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
1543                my $ls = svn_ls_current($dir, $rm, $add);
1544                next if (scalar @$ls);
1545                sys(qw(svn rm --force),$dir);
1546
1547                my $dn = dirname $dir;
1548                $rm->{ $dn }->{ basename $dir } = 1;
1549                $ls = svn_ls_current($dn, $rm, $add);
1550                while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
1551                        sys(qw(svn rm --force),$dn);
1552                        $dir = basename $dn;
1553                        $dn = dirname $dn;
1554                        $rm->{ $dn }->{ $dir } = 1;
1555                        $ls = svn_ls_current($dn, $rm, $add);
1556                }
1557        }
1558}
1559
1560sub get_commit_message {
1561        my ($commit, $commit_msg) = (@_);
1562        my %log_msg = ( msg => '' );
1563        open my $msg, '>', $commit_msg or croak $!;
1564
1565        chomp(my $type = `git-cat-file -t $commit`);
1566        if ($type eq 'commit' || $type eq 'tag') {
1567                my $pid = open my $msg_fh, '-|';
1568                defined $pid or croak $!;
1569
1570                if ($pid == 0) {
1571                        exec('git-cat-file', $type, $commit) or croak $!;
1572                }
1573                my $in_msg = 0;
1574                while (<$msg_fh>) {
1575                        if (!$in_msg) {
1576                                $in_msg = 1 if (/^\s*$/);
1577                        } elsif (/^git-svn-id: /) {
1578                                # skip this, we regenerate the correct one
1579                                # on re-fetch anyways
1580                        } else {
1581                                print $msg $_ or croak $!;
1582                        }
1583                }
1584                close $msg_fh or croak $?;
1585        }
1586        close $msg or croak $!;
1587
1588        if ($_edit || ($type eq 'tree')) {
1589                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1590                system($editor, $commit_msg);
1591        }
1592
1593        # file_to_s removes all trailing newlines, so just use chomp() here:
1594        open $msg, '<', $commit_msg or croak $!;
1595        { local $/; chomp($log_msg{msg} = <$msg>); }
1596        close $msg or croak $!;
1597
1598        return \%log_msg;
1599}
1600
1601sub set_svn_commit_env {
1602        if (defined $LC_ALL) {
1603                $ENV{LC_ALL} = $LC_ALL;
1604        } else {
1605                delete $ENV{LC_ALL};
1606        }
1607}
1608
1609sub svn_commit_tree {
1610        my ($last, $commit) = @_;
1611        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
1612        my $log_msg = get_commit_message($commit, $commit_msg);
1613        my ($oneline) = ($log_msg->{msg} =~ /([^\n\r]+)/);
1614        print "Committing $commit: $oneline\n";
1615
1616        set_svn_commit_env();
1617        my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
1618        $ENV{LC_ALL} = 'C';
1619        unlink $commit_msg;
1620        my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/);
1621        if (!defined $committed) {
1622                my $out = join("\n",@ci_output);
1623                print STDERR "W: Trouble parsing \`svn commit' output:\n\n",
1624                                $out, "\n\nAssuming English locale...";
1625                ($committed) = ($out =~ /^Committed revision \d+\./sm);
1626                defined $committed or die " FAILED!\n",
1627                        "Commit output failed to parse committed revision!\n",
1628                print STDERR " OK\n";
1629        }
1630
1631        my @svn_up = qw(svn up);
1632        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
1633        if ($_optimize_commits && ($committed == ($last->{revision} + 1))) {
1634                push @svn_up, "-r$committed";
1635                sys(@svn_up);
1636                my $info = svn_info('.');
1637                my $date = $info->{'Last Changed Date'} or die "Missing date\n";
1638                if ($info->{'Last Changed Rev'} != $committed) {
1639                        croak "$info->{'Last Changed Rev'} != $committed\n"
1640                }
1641                my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1642                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1643                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1644                                         or croak "Failed to parse date: $date\n";
1645                $log_msg->{date} = "$tz $Y-$m-$d $H:$M:$S";
1646                $log_msg->{author} = $info->{'Last Changed Author'};
1647                $log_msg->{revision} = $committed;
1648                $log_msg->{msg} .= "\n";
1649                $log_msg->{parents} = [ $last->{commit} ];
1650                $log_msg->{commit} = git_commit($log_msg, $commit);
1651                return $log_msg;
1652        }
1653        # resync immediately
1654        push @svn_up, "-r$last->{revision}";
1655        sys(@svn_up);
1656        return fetch("$committed=$commit");
1657}
1658
1659sub rev_list_raw {
1660        my (@args) = @_;
1661        my $pid = open my $fh, '-|';
1662        defined $pid or croak $!;
1663        if (!$pid) {
1664                exec(qw/git-rev-list --pretty=raw/, @args) or croak $!;
1665        }
1666        return { fh => $fh, t => { } };
1667}
1668
1669sub next_rev_list_entry {
1670        my $rl = shift;
1671        my $fh = $rl->{fh};
1672        my $x = $rl->{t};
1673        while (<$fh>) {
1674                if (/^commit ($sha1)$/o) {
1675                        if ($x->{c}) {
1676                                $rl->{t} = { c => $1 };
1677                                return $x;
1678                        } else {
1679                                $x->{c} = $1;
1680                        }
1681                } elsif (/^parent ($sha1)$/o) {
1682                        $x->{p}->{$1} = 1;
1683                } elsif (s/^    //) {
1684                        $x->{m} ||= '';
1685                        $x->{m} .= $_;
1686                }
1687        }
1688        return ($x != $rl->{t}) ? $x : undef;
1689}
1690
1691# read the entire log into a temporary file (which is removed ASAP)
1692# and store the file handle + parser state
1693sub svn_log_raw {
1694        my (@log_args) = @_;
1695        my $log_fh = IO::File->new_tmpfile or croak $!;
1696        my $pid = fork;
1697        defined $pid or croak $!;
1698        if (!$pid) {
1699                open STDOUT, '>&', $log_fh or croak $!;
1700                exec (qw(svn log), @log_args) or croak $!
1701        }
1702        waitpid $pid, 0;
1703        croak $? if $?;
1704        seek $log_fh, 0, 0 or croak $!;
1705        return { state => 'sep', fh => $log_fh };
1706}
1707
1708sub next_log_entry {
1709        my $log = shift; # retval of svn_log_raw()
1710        my $ret = undef;
1711        my $fh = $log->{fh};
1712
1713        while (<$fh>) {
1714                chomp;
1715                if (/^\-{72}$/) {
1716                        if ($log->{state} eq 'msg') {
1717                                if ($ret->{lines}) {
1718                                        $ret->{msg} .= $_."\n";
1719                                        unless(--$ret->{lines}) {
1720                                                $log->{state} = 'sep';
1721                                        }
1722                                } else {
1723                                        croak "Log parse error at: $_\n",
1724                                                $ret->{revision},
1725                                                "\n";
1726                                }
1727                                next;
1728                        }
1729                        if ($log->{state} ne 'sep') {
1730                                croak "Log parse error at: $_\n",
1731                                        "state: $log->{state}\n",
1732                                        $ret->{revision},
1733                                        "\n";
1734                        }
1735                        $log->{state} = 'rev';
1736
1737                        # if we have an empty log message, put something there:
1738                        if ($ret) {
1739                                $ret->{msg} ||= "\n";
1740                                delete $ret->{lines};
1741                                return $ret;
1742                        }
1743                        next;
1744                }
1745                if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
1746                        my $rev = $1;
1747                        my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
1748                        ($lines) = ($lines =~ /(\d+)/);
1749                        $date = '1970-01-01 00:00:00 +0000'
1750                                if ($_ignore_nodate && $date eq '(no date)');
1751                        my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1752                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1753                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1754                                         or croak "Failed to parse date: $date\n";
1755                        $ret = {        revision => $rev,
1756                                        date => "$tz $Y-$m-$d $H:$M:$S",
1757                                        author => $author,
1758                                        lines => $lines,
1759                                        msg => '' };
1760                        if (defined $_authors && ! defined $users{$author}) {
1761                                die "Author: $author not defined in ",
1762                                                "$_authors file\n";
1763                        }
1764                        $log->{state} = 'msg_start';
1765                        next;
1766                }
1767                # skip the first blank line of the message:
1768                if ($log->{state} eq 'msg_start' && /^$/) {
1769                        $log->{state} = 'msg';
1770                } elsif ($log->{state} eq 'msg') {
1771                        if ($ret->{lines}) {
1772                                $ret->{msg} .= $_."\n";
1773                                unless (--$ret->{lines}) {
1774                                        $log->{state} = 'sep';
1775                                }
1776                        } else {
1777                                croak "Log parse error at: $_\n",
1778                                        $ret->{revision},"\n";
1779                        }
1780                }
1781        }
1782        return $ret;
1783}
1784
1785sub svn_info {
1786        my $url = shift || $SVN_URL;
1787
1788        my $pid = open my $info_fh, '-|';
1789        defined $pid or croak $!;
1790
1791        if ($pid == 0) {
1792                exec(qw(svn info),$url) or croak $!;
1793        }
1794
1795        my $ret = {};
1796        # only single-lines seem to exist in svn info output
1797        while (<$info_fh>) {
1798                chomp $_;
1799                if (m#^([^:]+)\s*:\s*(\S.*)$#) {
1800                        $ret->{$1} = $2;
1801                        push @{$ret->{-order}}, $1;
1802                }
1803        }
1804        close $info_fh or croak $?;
1805        return $ret;
1806}
1807
1808sub sys { system(@_) == 0 or croak $? }
1809
1810sub do_update_index {
1811        my ($z_cmd, $cmd, $no_text_base) = @_;
1812
1813        my $z = open my $p, '-|';
1814        defined $z or croak $!;
1815        unless ($z) { exec @$z_cmd or croak $! }
1816
1817        my $pid = open my $ui, '|-';
1818        defined $pid or croak $!;
1819        unless ($pid) {
1820                exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!;
1821        }
1822        local $/ = "\0";
1823        while (my $x = <$p>) {
1824                chomp $x;
1825                if (!$no_text_base && lstat $x && ! -l _ &&
1826                                svn_propget_base('svn:keywords', $x)) {
1827                        my $mode = -x _ ? 0755 : 0644;
1828                        my ($v,$d,$f) = File::Spec->splitpath($x);
1829                        my $tb = File::Spec->catfile($d, '.svn', 'tmp',
1830                                                'text-base',"$f.svn-base");
1831                        $tb =~ s#^/##;
1832                        unless (-f $tb) {
1833                                $tb = File::Spec->catfile($d, '.svn',
1834                                                'text-base',"$f.svn-base");
1835                                $tb =~ s#^/##;
1836                        }
1837                        my @s = stat($x);
1838                        unlink $x or croak $!;
1839                        copy($tb, $x);
1840                        chmod(($mode &~ umask), $x) or croak $!;
1841                        utime $s[8], $s[9], $x;
1842                }
1843                print $ui $x,"\0";
1844        }
1845        close $ui or croak $?;
1846}
1847
1848sub index_changes {
1849        return if $_use_lib;
1850
1851        if (!-f "$GIT_SVN_DIR/info/exclude") {
1852                open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!;
1853                print $fd '.svn',"\n";
1854                close $fd or croak $!;
1855        }
1856        my $no_text_base = shift;
1857        do_update_index([qw/git-diff-files --name-only -z/],
1858                        'remove',
1859                        $no_text_base);
1860        do_update_index([qw/git-ls-files -z --others/,
1861                                "--exclude-from=$GIT_SVN_DIR/info/exclude"],
1862                        'add',
1863                        $no_text_base);
1864}
1865
1866sub s_to_file {
1867        my ($str, $file, $mode) = @_;
1868        open my $fd,'>',$file or croak $!;
1869        print $fd $str,"\n" or croak $!;
1870        close $fd or croak $!;
1871        chmod ($mode &~ umask, $file) if (defined $mode);
1872}
1873
1874sub file_to_s {
1875        my $file = shift;
1876        open my $fd,'<',$file or croak "$!: file: $file\n";
1877        local $/;
1878        my $ret = <$fd>;
1879        close $fd or croak $!;
1880        $ret =~ s/\s*$//s;
1881        return $ret;
1882}
1883
1884sub assert_revision_unknown {
1885        my $r = shift;
1886        if (my $c = revdb_get($REVDB, $r)) {
1887                croak "$r = $c already exists! Why are we refetching it?";
1888        }
1889}
1890
1891sub trees_eq {
1892        my ($x, $y) = @_;
1893        my @x = safe_qx('git-cat-file','commit',$x);
1894        my @y = safe_qx('git-cat-file','commit',$y);
1895        if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
1896                                || $y[0] !~ /^tree $sha1\n$/) {
1897                print STDERR "Trees not equal: $y[0] != $x[0]\n";
1898                return 0
1899        }
1900        return 1;
1901}
1902
1903sub git_commit {
1904        my ($log_msg, @parents) = @_;
1905        assert_revision_unknown($log_msg->{revision});
1906        map_tree_joins() if (@_branch_from && !%tree_map);
1907
1908        my (@tmp_parents, @exec_parents, %seen_parent);
1909        if (my $lparents = $log_msg->{parents}) {
1910                @tmp_parents = @$lparents
1911        }
1912        # commit parents can be conditionally bound to a particular
1913        # svn revision via: "svn_revno=commit_sha1", filter them out here:
1914        foreach my $p (@parents) {
1915                next unless defined $p;
1916                if ($p =~ /^(\d+)=($sha1_short)$/o) {
1917                        if ($1 == $log_msg->{revision}) {
1918                                push @tmp_parents, $2;
1919                        }
1920                } else {
1921                        push @tmp_parents, $p if $p =~ /$sha1_short/o;
1922                }
1923        }
1924        my $tree = $log_msg->{tree};
1925        if (!defined $tree) {
1926                my $index = set_index($GIT_SVN_INDEX);
1927                index_changes();
1928                chomp($tree = `git-write-tree`);
1929                croak $? if $?;
1930                restore_index($index);
1931        }
1932
1933        # just in case we clobber the existing ref, we still want that ref
1934        # as our parent:
1935        if (my $cur = eval { file_to_s("$GIT_DIR/refs/remotes/$GIT_SVN") }) {
1936                push @tmp_parents, $cur;
1937        }
1938
1939        if (exists $tree_map{$tree}) {
1940                foreach my $p (@{$tree_map{$tree}}) {
1941                        my $skip;
1942                        foreach (@tmp_parents) {
1943                                # see if a common parent is found
1944                                my $mb = eval {
1945                                        safe_qx('git-merge-base', $_, $p)
1946                                };
1947                                next if ($@ || $?);
1948                                $skip = 1;
1949                                last;
1950                        }
1951                        next if $skip;
1952                        my ($url_p, $r_p, $uuid_p) = cmt_metadata($p);
1953                        next if (($SVN_UUID eq $uuid_p) &&
1954                                                ($log_msg->{revision} > $r_p));
1955                        next if (defined $url_p && defined $SVN_URL &&
1956                                                ($SVN_UUID eq $uuid_p) &&
1957                                                ($url_p eq $SVN_URL));
1958                        push @tmp_parents, $p;
1959                }
1960        }
1961        foreach (@tmp_parents) {
1962                next if $seen_parent{$_};
1963                $seen_parent{$_} = 1;
1964                push @exec_parents, $_;
1965                # MAXPARENT is defined to 16 in commit-tree.c:
1966                last if @exec_parents > 16;
1967        }
1968
1969        set_commit_env($log_msg);
1970        my @exec = ('git-commit-tree', $tree);
1971        push @exec, '-p', $_  foreach @exec_parents;
1972        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1973                                                                or croak $!;
1974        print $msg_fh $log_msg->{msg} or croak $!;
1975        unless ($_no_metadata) {
1976                print $msg_fh "\ngit-svn-id: $SVN_URL\@$log_msg->{revision}",
1977                                        " $SVN_UUID\n" or croak $!;
1978        }
1979        $msg_fh->flush == 0 or croak $!;
1980        close $msg_fh or croak $!;
1981        chomp(my $commit = do { local $/; <$out_fh> });
1982        close $out_fh or croak $!;
1983        waitpid $pid, 0;
1984        croak $? if $?;
1985        if ($commit !~ /^$sha1$/o) {
1986                die "Failed to commit, invalid sha1: $commit\n";
1987        }
1988        sys('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
1989        revdb_set($REVDB, $log_msg->{revision}, $commit);
1990
1991        # this output is read via pipe, do not change:
1992        print "r$log_msg->{revision} = $commit\n";
1993        check_repack();
1994        return $commit;
1995}
1996
1997sub check_repack {
1998        if ($_repack && (--$_repack_nr == 0)) {
1999                $_repack_nr = $_repack;
2000                sys("git repack $_repack_flags");
2001        }
2002}
2003
2004sub set_commit_env {
2005        my ($log_msg) = @_;
2006        my $author = $log_msg->{author};
2007        if (!defined $author || length $author == 0) {
2008                $author = '(no author)';
2009        }
2010        my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
2011                                : ($author,"$author\@$SVN_UUID");
2012        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
2013        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
2014        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
2015}
2016
2017sub apply_mod_line_blob {
2018        my $m = shift;
2019        if ($m->{mode_b} =~ /^120/) {
2020                blob_to_symlink($m->{sha1_b}, $m->{file_b});
2021        } else {
2022                blob_to_file($m->{sha1_b}, $m->{file_b});
2023        }
2024}
2025
2026sub blob_to_symlink {
2027        my ($blob, $link) = @_;
2028        defined $link or croak "\$link not defined!\n";
2029        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2030        if (-l $link || -f _) {
2031                unlink $link or croak $!;
2032        }
2033
2034        my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
2035        symlink $dest, $link or croak $!;
2036}
2037
2038sub blob_to_file {
2039        my ($blob, $file) = @_;
2040        defined $file or croak "\$file not defined!\n";
2041        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2042        if (-l $file || -f _) {
2043                unlink $file or croak $!;
2044        }
2045
2046        open my $blob_fh, '>', $file or croak "$!: $file\n";
2047        my $pid = fork;
2048        defined $pid or croak $!;
2049
2050        if ($pid == 0) {
2051                open STDOUT, '>&', $blob_fh or croak $!;
2052                exec('git-cat-file','blob',$blob) or croak $!;
2053        }
2054        waitpid $pid, 0;
2055        croak $? if $?;
2056
2057        close $blob_fh or croak $!;
2058}
2059
2060sub safe_qx {
2061        my $pid = open my $child, '-|';
2062        defined $pid or croak $!;
2063        if ($pid == 0) {
2064                exec(@_) or croak $!;
2065        }
2066        my @ret = (<$child>);
2067        close $child or croak $?;
2068        die $? if $?; # just in case close didn't error out
2069        return wantarray ? @ret : join('',@ret);
2070}
2071
2072sub svn_compat_check {
2073        if ($_follow_parent) {
2074                print STDERR 'E: --follow-parent functionality is only ',
2075                                "available when SVN libraries are used\n";
2076                exit 1;
2077        }
2078        my @co_help = safe_qx(qw(svn co -h));
2079        unless (grep /ignore-externals/,@co_help) {
2080                print STDERR "W: Installed svn version does not support ",
2081                                "--ignore-externals\n";
2082                $_no_ignore_ext = 1;
2083        }
2084        if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
2085                $_svn_co_url_revs = 1;
2086        }
2087        if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) {
2088                $_svn_pg_peg_revs = 1;
2089        }
2090
2091        # I really, really hope nobody hits this...
2092        unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
2093                print STDERR <<'';
2094W: The installed svn version does not support the --stop-on-copy flag in
2095   the log command.
2096   Lets hope the directory you're tracking is not a branch or tag
2097   and was never moved within the repository...
2098
2099                $_no_stop_copy = 1;
2100        }
2101}
2102
2103# *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
2104# (and they won't honor URL@<rev> without -r<rev>, too!)
2105sub svn_cmd_checkout {
2106        my ($url, $rev, $dir) = @_;
2107        my @cmd = ('svn','co', "-r$rev");
2108        push @cmd, '--ignore-externals' unless $_no_ignore_ext;
2109        $url .= "\@$rev" if $_svn_co_url_revs;
2110        sys(@cmd, $url, $dir);
2111}
2112
2113sub check_upgrade_needed {
2114        if (!-r $REVDB) {
2115                -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2116                open my $fh, '>>',$REVDB or croak $!;
2117                close $fh;
2118        }
2119        my $old = eval {
2120                my $pid = open my $child, '-|';
2121                defined $pid or croak $!;
2122                if ($pid == 0) {
2123                        close STDERR;
2124                        exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $!;
2125                }
2126                my @ret = (<$child>);
2127                close $child or croak $?;
2128                die $? if $?; # just in case close didn't error out
2129                return wantarray ? @ret : join('',@ret);
2130        };
2131        return unless $old;
2132        my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
2133        if ($@ || !$head) {
2134                print STDERR "Please run: $0 rebuild --upgrade\n";
2135                exit 1;
2136        }
2137}
2138
2139# fills %tree_map with a reverse mapping of trees to commits.  Useful
2140# for finding parents to commit on.
2141sub map_tree_joins {
2142        my %seen;
2143        foreach my $br (@_branch_from) {
2144                my $pid = open my $pipe, '-|';
2145                defined $pid or croak $!;
2146                if ($pid == 0) {
2147                        exec(qw(git-rev-list --topo-order --pretty=raw), $br)
2148                                                                or croak $!;
2149                }
2150                while (<$pipe>) {
2151                        if (/^commit ($sha1)$/o) {
2152                                my $commit = $1;
2153
2154                                # if we've seen a commit,
2155                                # we've seen its parents
2156                                last if $seen{$commit};
2157                                my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
2158                                unless (defined $tree) {
2159                                        die "Failed to parse commit $commit\n";
2160                                }
2161                                push @{$tree_map{$tree}}, $commit;
2162                                $seen{$commit} = 1;
2163                        }
2164                }
2165                close $pipe; # we could be breaking the pipe early
2166        }
2167}
2168
2169sub load_all_refs {
2170        if (@_branch_from) {
2171                print STDERR '--branch|-b parameters are ignored when ',
2172                        "--branch-all-refs|-B is passed\n";
2173        }
2174
2175        # don't worry about rev-list on non-commit objects/tags,
2176        # it shouldn't blow up if a ref is a blob or tree...
2177        chomp(@_branch_from = `git-rev-parse --symbolic --all`);
2178}
2179
2180# '<svn username> = real-name <email address>' mapping based on git-svnimport:
2181sub load_authors {
2182        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2183        while (<$authors>) {
2184                chomp;
2185                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
2186                my ($user, $name, $email) = ($1, $2, $3);
2187                $users{$user} = [$name, $email];
2188        }
2189        close $authors or croak $!;
2190}
2191
2192sub rload_authors {
2193        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2194        while (<$authors>) {
2195                chomp;
2196                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2197                my ($user, $name, $email) = ($1, $2, $3);
2198                $rusers{"$name <$email>"} = $user;
2199        }
2200        close $authors or croak $!;
2201}
2202
2203sub svn_propget_base {
2204        my ($p, $f) = @_;
2205        $f .= '@BASE' if $_svn_pg_peg_revs;
2206        return safe_qx(qw/svn propget/, $p, $f);
2207}
2208
2209sub git_svn_each {
2210        my $sub = shift;
2211        foreach (`git-rev-parse --symbolic --all`) {
2212                next unless s#^refs/remotes/##;
2213                chomp $_;
2214                next unless -f "$GIT_DIR/svn/$_/info/url";
2215                &$sub($_);
2216        }
2217}
2218
2219sub migrate_revdb {
2220        git_svn_each(sub {
2221                my $id = shift;
2222                defined(my $pid = fork) or croak $!;
2223                if (!$pid) {
2224                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2225                        init_vars();
2226                        exit 0 if -r $REVDB;
2227                        print "Upgrading svn => git mapping...\n";
2228                        -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2229                        open my $fh, '>>',$REVDB or croak $!;
2230                        close $fh;
2231                        rebuild();
2232                        print "Done upgrading. You may now delete the ",
2233                                "deprecated $GIT_SVN_DIR/revs directory\n";
2234                        exit 0;
2235                }
2236                waitpid $pid, 0;
2237                croak $? if $?;
2238        });
2239}
2240
2241sub migration_check {
2242        migrate_revdb() unless (-e $REVDB);
2243        return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
2244        print "Upgrading repository...\n";
2245        unless (-d "$GIT_DIR/svn") {
2246                mkdir "$GIT_DIR/svn" or croak $!;
2247        }
2248        print "Data from a previous version of git-svn exists, but\n\t",
2249                                "$GIT_SVN_DIR\n\t(required for this version ",
2250                                "($VERSION) of git-svn) does not.\n";
2251
2252        foreach my $x (`git-rev-parse --symbolic --all`) {
2253                next unless $x =~ s#^refs/remotes/##;
2254                chomp $x;
2255                next unless -f "$GIT_DIR/$x/info/url";
2256                my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
2257                next unless $u;
2258                my $dn = dirname("$GIT_DIR/svn/$x");
2259                mkpath([$dn]) unless -d $dn;
2260                rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
2261        }
2262        migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
2263        print "Done upgrading.\n";
2264}
2265
2266sub find_rev_before {
2267        my ($r, $id, $eq_ok) = @_;
2268        my $f = "$GIT_DIR/svn/$id/.rev_db";
2269        return (undef,undef) unless -r $f;
2270        --$r unless $eq_ok;
2271        while ($r > 0) {
2272                if (my $c = revdb_get($f, $r)) {
2273                        return ($r, $c);
2274                }
2275                --$r;
2276        }
2277        return (undef, undef);
2278}
2279
2280sub init_vars {
2281        $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
2282        $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
2283        $REVDB = "$GIT_SVN_DIR/.rev_db";
2284        $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
2285        $SVN_URL = undef;
2286        $SVN_WC = "$GIT_SVN_DIR/tree";
2287        %tree_map = ();
2288}
2289
2290# convert GetOpt::Long specs for use by git-repo-config
2291sub read_repo_config {
2292        return unless -d $GIT_DIR;
2293        my $opts = shift;
2294        foreach my $o (keys %$opts) {
2295                my $v = $opts->{$o};
2296                my ($key) = ($o =~ /^([a-z\-]+)/);
2297                $key =~ s/-//g;
2298                my $arg = 'git-repo-config';
2299                $arg .= ' --int' if ($o =~ /[:=]i$/);
2300                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
2301                if (ref $v eq 'ARRAY') {
2302                        chomp(my @tmp = `$arg --get-all svn.$key`);
2303                        @$v = @tmp if @tmp;
2304                } else {
2305                        chomp(my $tmp = `$arg --get svn.$key`);
2306                        if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
2307                                $$v = $tmp;
2308                        }
2309                }
2310        }
2311}
2312
2313sub set_default_vals {
2314        if (defined $_repack) {
2315                $_repack = 1000 if ($_repack <= 0);
2316                $_repack_nr = $_repack;
2317                $_repack_flags ||= '-d';
2318        }
2319}
2320
2321sub read_grafts {
2322        my $gr_file = shift;
2323        my ($grafts, $comments) = ({}, {});
2324        if (open my $fh, '<', $gr_file) {
2325                my @tmp;
2326                while (<$fh>) {
2327                        if (/^($sha1)\s+/) {
2328                                my $c = $1;
2329                                if (@tmp) {
2330                                        @{$comments->{$c}} = @tmp;
2331                                        @tmp = ();
2332                                }
2333                                foreach my $p (split /\s+/, $_) {
2334                                        $grafts->{$c}->{$p} = 1;
2335                                }
2336                        } else {
2337                                push @tmp, $_;
2338                        }
2339                }
2340                close $fh or croak $!;
2341                @{$comments->{'END'}} = @tmp if @tmp;
2342        }
2343        return ($grafts, $comments);
2344}
2345
2346sub write_grafts {
2347        my ($grafts, $comments, $gr_file) = @_;
2348
2349        open my $fh, '>', $gr_file or croak $!;
2350        foreach my $c (sort keys %$grafts) {
2351                if ($comments->{$c}) {
2352                        print $fh $_ foreach @{$comments->{$c}};
2353                }
2354                my $p = $grafts->{$c};
2355                my %x; # real parents
2356                delete $p->{$c}; # commits are not self-reproducing...
2357                my $pid = open my $ch, '-|';
2358                defined $pid or croak $!;
2359                if (!$pid) {
2360                        exec(qw/git-cat-file commit/, $c) or croak $!;
2361                }
2362                while (<$ch>) {
2363                        if (/^parent ($sha1)/) {
2364                                $x{$1} = $p->{$1} = 1;
2365                        } else {
2366                                last unless /^\S/;
2367                        }
2368                }
2369                close $ch; # breaking the pipe
2370
2371                # if real parents are the only ones in the grafts, drop it
2372                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2373
2374                my (@ip, @jp, $mb);
2375                my %del = %x;
2376                @ip = @jp = keys %$p;
2377                foreach my $i (@ip) {
2378                        next if $del{$i} || $p->{$i} == 2;
2379                        foreach my $j (@jp) {
2380                                next if $i eq $j || $del{$j} || $p->{$j} == 2;
2381                                $mb = eval { safe_qx('git-merge-base',$i,$j) };
2382                                next unless $mb;
2383                                chomp $mb;
2384                                next if $x{$mb};
2385                                if ($mb eq $j) {
2386                                        delete $p->{$i};
2387                                        $del{$i} = 1;
2388                                } elsif ($mb eq $i) {
2389                                        delete $p->{$j};
2390                                        $del{$j} = 1;
2391                                }
2392                        }
2393                }
2394
2395                # if real parents are the only ones in the grafts, drop it
2396                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2397
2398                print $fh $c, ' ', join(' ', sort keys %$p),"\n";
2399        }
2400        if ($comments->{'END'}) {
2401                print $fh $_ foreach @{$comments->{'END'}};
2402        }
2403        close $fh or croak $!;
2404}
2405
2406sub read_url_paths_all {
2407        my ($l_map, $pfx, $p) = @_;
2408        my @dir;
2409        foreach (<$p/*>) {
2410                if (-r "$_/info/url") {
2411                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2412                        my $id = $pfx . basename $_;
2413                        my $url = file_to_s("$_/info/url");
2414                        my ($u, $p) = repo_path_split($url);
2415                        $l_map->{$u}->{$p} = $id;
2416                } elsif (-d $_) {
2417                        push @dir, $_;
2418                }
2419        }
2420        foreach (@dir) {
2421                my $x = $_;
2422                $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
2423                read_url_paths_all($l_map, $x, $_);
2424        }
2425}
2426
2427# this one only gets ids that have been imported, not new ones
2428sub read_url_paths {
2429        my $l_map = {};
2430        git_svn_each(sub { my $x = shift;
2431                        my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
2432                        my ($u, $p) = repo_path_split($url);
2433                        $l_map->{$u}->{$p} = $x;
2434                        });
2435        return $l_map;
2436}
2437
2438sub extract_metadata {
2439        my $id = shift or return (undef, undef, undef);
2440        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
2441                                                        \s([a-f\d\-]+)$/x);
2442        if (!$rev || !$uuid || !$url) {
2443                # some of the original repositories I made had
2444                # identifiers like this:
2445                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
2446        }
2447        return ($url, $rev, $uuid);
2448}
2449
2450sub cmt_metadata {
2451        return extract_metadata((grep(/^git-svn-id: /,
2452                safe_qx(qw/git-cat-file commit/, shift)))[-1]);
2453}
2454
2455sub get_commit_time {
2456        my $cmt = shift;
2457        defined(my $pid = open my $fh, '-|') or croak $!;
2458        if (!$pid) {
2459                exec qw/git-rev-list --pretty=raw -n1/, $cmt or croak $!;
2460        }
2461        while (<$fh>) {
2462                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
2463                my ($s, $tz) = ($1, $2);
2464                if ($tz =~ s/^\+//) {
2465                        $s += tz_to_s_offset($tz);
2466                } elsif ($tz =~ s/^\-//) {
2467                        $s -= tz_to_s_offset($tz);
2468                }
2469                close $fh;
2470                return $s;
2471        }
2472        die "Can't get commit time for commit: $cmt\n";
2473}
2474
2475sub tz_to_s_offset {
2476        my ($tz) = @_;
2477        $tz =~ s/(\d\d)$//;
2478        return ($1 * 60) + ($tz * 3600);
2479}
2480
2481sub setup_pager { # translated to Perl from pager.c
2482        return unless (-t *STDOUT);
2483        my $pager = $ENV{PAGER};
2484        if (!defined $pager) {
2485                $pager = 'less';
2486        } elsif (length $pager == 0 || $pager eq 'cat') {
2487                return;
2488        }
2489        pipe my $rfd, my $wfd or return;
2490        defined(my $pid = fork) or croak $!;
2491        if (!$pid) {
2492                open STDOUT, '>&', $wfd or croak $!;
2493                return;
2494        }
2495        open STDIN, '<&', $rfd or croak $!;
2496        $ENV{LESS} ||= '-S';
2497        exec $pager or croak "Can't run pager: $!\n";;
2498}
2499
2500sub get_author_info {
2501        my ($dest, $author, $t, $tz) = @_;
2502        $author =~ s/(?:^\s*|\s*$)//g;
2503        $dest->{a_raw} = $author;
2504        my $_a;
2505        if ($_authors) {
2506                $_a = $rusers{$author} || undef;
2507        }
2508        if (!$_a) {
2509                ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
2510        }
2511        $dest->{t} = $t;
2512        $dest->{tz} = $tz;
2513        $dest->{a} = $_a;
2514        # Date::Parse isn't in the standard Perl distro :(
2515        if ($tz =~ s/^\+//) {
2516                $t += tz_to_s_offset($tz);
2517        } elsif ($tz =~ s/^\-//) {
2518                $t -= tz_to_s_offset($tz);
2519        }
2520        $dest->{t_utc} = $t;
2521}
2522
2523sub process_commit {
2524        my ($c, $r_min, $r_max, $defer) = @_;
2525        if (defined $r_min && defined $r_max) {
2526                if ($r_min == $c->{r} && $r_min == $r_max) {
2527                        show_commit($c);
2528                        return 0;
2529                }
2530                return 1 if $r_min == $r_max;
2531                if ($r_min < $r_max) {
2532                        # we need to reverse the print order
2533                        return 0 if (defined $_limit && --$_limit < 0);
2534                        push @$defer, $c;
2535                        return 1;
2536                }
2537                if ($r_min != $r_max) {
2538                        return 1 if ($r_min < $c->{r});
2539                        return 1 if ($r_max > $c->{r});
2540                }
2541        }
2542        return 0 if (defined $_limit && --$_limit < 0);
2543        show_commit($c);
2544        return 1;
2545}
2546
2547sub show_commit {
2548        my $c = shift;
2549        if ($_oneline) {
2550                my $x = "\n";
2551                if (my $l = $c->{l}) {
2552                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2553                        $x = $l->[0];
2554                }
2555                $_l_fmt ||= 'A' . length($c->{r});
2556                print 'r',pack($_l_fmt, $c->{r}),' | ';
2557                print "$c->{c} | " if $_show_commit;
2558                print $x;
2559        } else {
2560                show_commit_normal($c);
2561        }
2562}
2563
2564sub show_commit_normal {
2565        my ($c) = @_;
2566        print '-' x72, "\nr$c->{r} | ";
2567        print "$c->{c} | " if $_show_commit;
2568        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2569                                 localtime($c->{t_utc})), ' | ';
2570        my $nr_line = 0;
2571
2572        if (my $l = $c->{l}) {
2573                while ($l->[$#$l] eq "\n" && $l->[($#$l - 1)] eq "\n") {
2574                        pop @$l;
2575                }
2576                $nr_line = scalar @$l;
2577                if (!$nr_line) {
2578                        print "1 line\n\n\n";
2579                } else {
2580                        if ($nr_line == 1) {
2581                                $nr_line = '1 line';
2582                        } else {
2583                                $nr_line .= ' lines';
2584                        }
2585                        print $nr_line, "\n\n";
2586                        print $_ foreach @$l;
2587                }
2588        } else {
2589                print "1 line\n\n";
2590
2591        }
2592        foreach my $x (qw/raw diff/) {
2593                if ($c->{$x}) {
2594                        print "\n";
2595                        print $_ foreach @{$c->{$x}}
2596                }
2597        }
2598}
2599
2600sub libsvn_load {
2601        return unless $_use_lib;
2602        $_use_lib = eval {
2603                require SVN::Core;
2604                if ($SVN::Core::VERSION lt '1.1.0') {
2605                        die "Need SVN::Core 1.1.0 or better ",
2606                                        "(got $SVN::Core::VERSION) ",
2607                                        "Falling back to command-line svn\n";
2608                }
2609                require SVN::Ra;
2610                require SVN::Delta;
2611                push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
2612                my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2613                                        $SVN::Node::dir.$SVN::Node::unknown.
2614                                        $SVN::Node::none.$SVN::Node::file.
2615                                        $SVN::Node::dir.$SVN::Node::unknown;
2616                1;
2617        };
2618}
2619
2620sub libsvn_connect {
2621        my ($url) = @_;
2622        my $auth = SVN::Core::auth_open([SVN::Client::get_simple_provider(),
2623                          SVN::Client::get_ssl_server_trust_file_provider(),
2624                          SVN::Client::get_username_provider()]);
2625        my $s = eval { SVN::Ra->new(url => $url, auth => $auth) };
2626        return $s;
2627}
2628
2629sub libsvn_get_file {
2630        my ($gui, $f, $rev) = @_;
2631        my $p = $f;
2632        if (length $SVN_PATH > 0) {
2633                return unless ($p =~ s#^\Q$SVN_PATH\E/##);
2634        }
2635
2636        my ($hash, $pid, $in, $out);
2637        my $pool = SVN::Pool->new;
2638        defined($pid = open3($in, $out, '>&STDERR',
2639                                qw/git-hash-object -w --stdin/)) or croak $!;
2640        # redirect STDOUT for SVN 1.1.x compatibility
2641        open my $stdout, '>&', \*STDOUT or croak $!;
2642        open STDOUT, '>&', $in or croak $!;
2643        my ($r, $props) = $SVN->get_file($f, $rev, \*STDOUT, $pool);
2644        $in->flush == 0 or croak $!;
2645        open STDOUT, '>&', $stdout or croak $!;
2646        close $in or croak $!;
2647        close $stdout or croak $!;
2648        $pool->clear;
2649        chomp($hash = do { local $/; <$out> });
2650        close $out or croak $!;
2651        waitpid $pid, 0;
2652        $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2653
2654        my $mode = exists $props->{'svn:executable'} ? '100755' : '100644';
2655        if (exists $props->{'svn:special'}) {
2656                $mode = '120000';
2657                my $link = `git-cat-file blob $hash`;
2658                $link =~ s/^link // or die "svn:special file with contents: <",
2659                                                $link, "> is not understood\n";
2660                defined($pid = open3($in, $out, '>&STDERR',
2661                                qw/git-hash-object -w --stdin/)) or croak $!;
2662                print $in $link;
2663                $in->flush == 0 or croak $!;
2664                close $in or croak $!;
2665                chomp($hash = do { local $/; <$out> });
2666                close $out or croak $!;
2667                waitpid $pid, 0;
2668                $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2669        }
2670        print $gui $mode,' ',$hash,"\t",$p,"\0" or croak $!;
2671}
2672
2673sub libsvn_log_entry {
2674        my ($rev, $author, $date, $msg, $parents) = @_;
2675        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2676                                         (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
2677                                or die "Unable to parse date: $date\n";
2678        if (defined $_authors && ! defined $users{$author}) {
2679                die "Author: $author not defined in $_authors file\n";
2680        }
2681        $msg = '' if ($rev == 0 && !defined $msg);
2682        return { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
2683                author => $author, msg => $msg."\n", parents => $parents || [] }
2684}
2685
2686sub process_rm {
2687        my ($gui, $last_commit, $f) = @_;
2688        $f =~ s#^\Q$SVN_PATH\E/?## or return;
2689        # remove entire directories.
2690        if (safe_qx('git-ls-tree',$last_commit,'--',$f) =~ /^040000 tree/) {
2691                defined(my $pid = open my $ls, '-|') or croak $!;
2692                if (!$pid) {
2693                        exec(qw/git-ls-tree -r --name-only -z/,
2694                                $last_commit,'--',$f) or croak $!;
2695                }
2696                local $/ = "\0";
2697                while (<$ls>) {
2698                        print $gui '0 ',0 x 40,"\t",$_ or croak $!;
2699                }
2700                close $ls or croak $?;
2701        } else {
2702                print $gui '0 ',0 x 40,"\t",$f,"\0" or croak $!;
2703        }
2704}
2705
2706sub libsvn_fetch {
2707        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
2708        open my $gui, '| git-update-index -z --index-info' or croak $!;
2709        my @amr;
2710        foreach my $f (keys %$paths) {
2711                my $m = $paths->{$f}->action();
2712                $f =~ s#^/+##;
2713                if ($m =~ /^[DR]$/) {
2714                        print "\t$m\t$f\n" unless $_q;
2715                        process_rm($gui, $last_commit, $f);
2716                        next if $m eq 'D';
2717                        # 'R' can be file replacements, too, right?
2718                }
2719                my $pool = SVN::Pool->new;
2720                my $t = $SVN->check_path($f, $rev, $pool);
2721                if ($t == $SVN::Node::file) {
2722                        if ($m =~ /^[AMR]$/) {
2723                                push @amr, [ $m, $f ];
2724                        } else {
2725                                die "Unrecognized action: $m, ($f r$rev)\n";
2726                        }
2727                } elsif ($t == $SVN::Node::dir && $m =~ /^[AR]$/) {
2728                        my @traversed = ();
2729                        libsvn_traverse($gui, '', $f, $rev, \@traversed);
2730                        foreach (@traversed) {
2731                                push @amr, [ $m, $_ ]
2732                        }
2733                }
2734                $pool->clear;
2735        }
2736        foreach (@amr) {
2737                print "\t$_->[0]\t$_->[1]\n" unless $_q;
2738                libsvn_get_file($gui, $_->[1], $rev)
2739        }
2740        close $gui or croak $?;
2741        return libsvn_log_entry($rev, $author, $date, $msg, [$last_commit]);
2742}
2743
2744sub svn_grab_base_rev {
2745        defined(my $pid = open my $fh, '-|') or croak $!;
2746        if (!$pid) {
2747                open my $null, '>', '/dev/null' or croak $!;
2748                open STDERR, '>&', $null or croak $!;
2749                exec qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0"
2750                                                                or croak $!;
2751        }
2752        chomp(my $c = do { local $/; <$fh> });
2753        close $fh;
2754        if (defined $c && length $c) {
2755                my ($url, $rev, $uuid) = cmt_metadata($c);
2756                return ($rev, $c) if defined $rev;
2757        }
2758        if ($_no_metadata) {
2759                my $offset = -41; # from tail
2760                my $rl;
2761                open my $fh, '<', $REVDB or
2762                        die "--no-metadata specified and $REVDB not readable\n";
2763                seek $fh, $offset, 2;
2764                $rl = readline $fh;
2765                defined $rl or return (undef, undef);
2766                chomp $rl;
2767                while ($c ne $rl && tell $fh != 0) {
2768                        $offset -= 41;
2769                        seek $fh, $offset, 2;
2770                        $rl = readline $fh;
2771                        defined $rl or return (undef, undef);
2772                        chomp $rl;
2773                }
2774                my $rev = tell $fh;
2775                croak $! if ($rev < -1);
2776                $rev =  ($rev - 41) / 41;
2777                close $fh or croak $!;
2778                return ($rev, $c);
2779        }
2780        return (undef, undef);
2781}
2782
2783sub libsvn_parse_revision {
2784        my $base = shift;
2785        my $head = $SVN->get_latest_revnum();
2786        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
2787                return ($base + 1, $head) if (defined $base);
2788                return (0, $head);
2789        }
2790        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
2791        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
2792        if ($_revision =~ /^BASE:(\d+)$/) {
2793                return ($base + 1, $1) if (defined $base);
2794                return (0, $head);
2795        }
2796        return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
2797        die "revision argument: $_revision not understood by git-svn\n",
2798                "Try using the command-line svn client instead\n";
2799}
2800
2801sub libsvn_traverse {
2802        my ($gui, $pfx, $path, $rev, $files) = @_;
2803        my $cwd = "$pfx/$path";
2804        my $pool = SVN::Pool->new;
2805        $cwd =~ s#^/+##g;
2806        my ($dirent, $r, $props) = $SVN->get_dir($cwd, $rev, $pool);
2807        foreach my $d (keys %$dirent) {
2808                my $t = $dirent->{$d}->kind;
2809                if ($t == $SVN::Node::dir) {
2810                        libsvn_traverse($gui, $cwd, $d, $rev, $files);
2811                } elsif ($t == $SVN::Node::file) {
2812                        my $file = "$cwd/$d";
2813                        if (defined $files) {
2814                                push @$files, $file;
2815                        } else {
2816                                print "\tA\t$file\n" unless $_q;
2817                                libsvn_get_file($gui, $file, $rev);
2818                        }
2819                }
2820        }
2821        $pool->clear;
2822}
2823
2824sub libsvn_traverse_ignore {
2825        my ($fh, $path, $r) = @_;
2826        $path =~ s#^/+##g;
2827        my $pool = SVN::Pool->new;
2828        my ($dirent, undef, $props) = $SVN->get_dir($path, $r, $pool);
2829        my $p = $path;
2830        $p =~ s#^\Q$SVN_PATH\E/?##;
2831        print $fh length $p ? "\n# $p\n" : "\n# /\n";
2832        if (my $s = $props->{'svn:ignore'}) {
2833                $s =~ s/[\r\n]+/\n/g;
2834                chomp $s;
2835                if (length $p == 0) {
2836                        $s =~ s#\n#\n/$p#g;
2837                        print $fh "/$s\n";
2838                } else {
2839                        $s =~ s#\n#\n/$p/#g;
2840                        print $fh "/$p/$s\n";
2841                }
2842        }
2843        foreach (sort keys %$dirent) {
2844                next if $dirent->{$_}->kind != $SVN::Node::dir;
2845                libsvn_traverse_ignore($fh, "$path/$_", $r);
2846        }
2847        $pool->clear;
2848}
2849
2850sub revisions_eq {
2851        my ($path, $r0, $r1) = @_;
2852        return 1 if $r0 == $r1;
2853        my $nr = 0;
2854        if ($_use_lib) {
2855                # should be OK to use Pool here (r1 - r0) should be small
2856                my $pool = SVN::Pool->new;
2857                libsvn_get_log($SVN, "/$path", $r0, $r1,
2858                                0, 1, 1, sub {$nr++}, $pool);
2859                $pool->clear;
2860        } else {
2861                my ($url, undef) = repo_path_split($SVN_URL);
2862                my $svn_log = svn_log_raw("$url/$path","-r$r0:$r1");
2863                while (next_log_entry($svn_log)) { $nr++ }
2864                close $svn_log->{fh};
2865        }
2866        return 0 if ($nr > 1);
2867        return 1;
2868}
2869
2870sub libsvn_find_parent_branch {
2871        my ($paths, $rev, $author, $date, $msg) = @_;
2872        my $svn_path = '/'.$SVN_PATH;
2873
2874        # look for a parent from another branch:
2875        my $i = $paths->{$svn_path} or return;
2876        my $branch_from = $i->copyfrom_path or return;
2877        my $r = $i->copyfrom_rev;
2878        print STDERR  "Found possible branch point: ",
2879                                "$branch_from => $svn_path, $r\n";
2880        $branch_from =~ s#^/##;
2881        my $l_map = {};
2882        read_url_paths_all($l_map, '', "$GIT_DIR/svn");
2883        my $url = $SVN->{url};
2884        defined $l_map->{$url} or return;
2885        my $id = $l_map->{$url}->{$branch_from};
2886        if (!defined $id && $_follow_parent) {
2887                print STDERR "Following parent: $branch_from\@$r\n";
2888                # auto create a new branch and follow it
2889                $id = basename($branch_from);
2890                $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
2891                while (-r "$GIT_DIR/svn/$id") {
2892                        # just grow a tail if we're not unique enough :x
2893                        $id .= '-';
2894                }
2895        }
2896        return unless defined $id;
2897
2898        my ($r0, $parent) = find_rev_before($r,$id,1);
2899        if ($_follow_parent && (!defined $r0 || !defined $parent)) {
2900                defined(my $pid = fork) or croak $!;
2901                if (!$pid) {
2902                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2903                        init_vars();
2904                        $SVN_URL = "$url/$branch_from";
2905                        $SVN_LOG = $SVN = undef;
2906                        setup_git_svn();
2907                        # we can't assume SVN_URL exists at r+1:
2908                        $_revision = "0:$r";
2909                        fetch_lib();
2910                        exit 0;
2911                }
2912                waitpid $pid, 0;
2913                croak $? if $?;
2914                ($r0, $parent) = find_rev_before($r,$id,1);
2915        }
2916        return unless (defined $r0 && defined $parent);
2917        if (revisions_eq($branch_from, $r0, $r)) {
2918                unlink $GIT_SVN_INDEX;
2919                print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
2920                sys(qw/git-read-tree/, $parent);
2921                return libsvn_fetch($parent, $paths, $rev,
2922                                        $author, $date, $msg);
2923        }
2924        print STDERR "Nope, branch point not imported or unknown\n";
2925        return undef;
2926}
2927
2928sub libsvn_get_log {
2929        my ($ra, @args) = @_;
2930        if ($SVN::Core::VERSION le '1.2.0') {
2931                splice(@args, 3, 1);
2932        }
2933        $ra->get_log(@args);
2934}
2935
2936sub libsvn_new_tree {
2937        if (my $log_entry = libsvn_find_parent_branch(@_)) {
2938                return $log_entry;
2939        }
2940        my ($paths, $rev, $author, $date, $msg) = @_;
2941        open my $gui, '| git-update-index -z --index-info' or croak $!;
2942        libsvn_traverse($gui, '', $SVN_PATH, $rev);
2943        close $gui or croak $?;
2944        return libsvn_log_entry($rev, $author, $date, $msg);
2945}
2946
2947sub find_graft_path_commit {
2948        my ($tree_paths, $p1, $r1) = @_;
2949        foreach my $x (keys %$tree_paths) {
2950                next unless ($p1 =~ /^\Q$x\E/);
2951                my $i = $tree_paths->{$x};
2952                my ($r0, $parent) = find_rev_before($r1,$i,1);
2953                return $parent if (defined $r0 && $r0 == $r1);
2954                print STDERR "r$r1 of $i not imported\n";
2955                next;
2956        }
2957        return undef;
2958}
2959
2960sub find_graft_path_parents {
2961        my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
2962        foreach my $x (keys %$tree_paths) {
2963                next unless ($p0 =~ /^\Q$x\E/);
2964                my $i = $tree_paths->{$x};
2965                my ($r, $parent) = find_rev_before($r0, $i, 1);
2966                if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
2967                        my ($url_b, undef, $uuid_b) = cmt_metadata($c);
2968                        my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
2969                        next if ($url_a && $url_b && $url_a eq $url_b &&
2970                                                        $uuid_b eq $uuid_a);
2971                        $grafts->{$c}->{$parent} = 1;
2972                }
2973        }
2974}
2975
2976sub libsvn_graft_file_copies {
2977        my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
2978        foreach (keys %$paths) {
2979                my $i = $paths->{$_};
2980                my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
2981                                        $i->copyfrom_rev);
2982                next unless (defined $p0 && defined $r0);
2983
2984                my $p1 = $_;
2985                $p1 =~ s#^/##;
2986                $p0 =~ s#^/##;
2987                my $c = find_graft_path_commit($tree_paths, $p1, $rev);
2988                next unless $c;
2989                find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
2990        }
2991}
2992
2993sub set_index {
2994        my $old = $ENV{GIT_INDEX_FILE};
2995        $ENV{GIT_INDEX_FILE} = shift;
2996        return $old;
2997}
2998
2999sub restore_index {
3000        my ($old) = @_;
3001        if (defined $old) {
3002                $ENV{GIT_INDEX_FILE} = $old;
3003        } else {
3004                delete $ENV{GIT_INDEX_FILE};
3005        }
3006}
3007
3008sub libsvn_commit_cb {
3009        my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
3010        if ($_optimize_commits && $rev == ($r_last + 1)) {
3011                my $log = libsvn_log_entry($rev,$committer,$date,$msg);
3012                $log->{tree} = get_tree_from_treeish($c);
3013                my $cmt = git_commit($log, $cmt_last, $c);
3014                my @diff = safe_qx('git-diff-tree', $cmt, $c);
3015                if (@diff) {
3016                        print STDERR "Trees differ: $cmt $c\n",
3017                                        join('',@diff),"\n";
3018                        exit 1;
3019                }
3020        } else {
3021                fetch("$rev=$c");
3022        }
3023}
3024
3025sub libsvn_ls_fullurl {
3026        my $fullurl = shift;
3027        my ($repo, $path) = repo_path_split($fullurl);
3028        $SVN ||= libsvn_connect($repo);
3029        my @ret;
3030        my $pool = SVN::Pool->new;
3031        my ($dirent, undef, undef) = $SVN->get_dir($path,
3032                                                $SVN->get_latest_revnum, $pool);
3033        foreach my $d (keys %$dirent) {
3034                if ($dirent->{$d}->kind == $SVN::Node::dir) {
3035                        push @ret, "$d/"; # add '/' for compat with cli svn
3036                }
3037        }
3038        $pool->clear;
3039        return @ret;
3040}
3041
3042
3043sub libsvn_skip_unknown_revs {
3044        my $err = shift;
3045        my $errno = $err->apr_err();
3046        # Maybe the branch we're tracking didn't
3047        # exist when the repo started, so it's
3048        # not an error if it doesn't, just continue
3049        #
3050        # Wonderfully consistent library, eh?
3051        # 160013 - svn:// and file://
3052        # 175002 - http(s)://
3053        #   More codes may be discovered later...
3054        if ($errno == 175002 || $errno == 160013) {
3055                return;
3056        }
3057        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3058};
3059
3060# Tie::File seems to be prone to offset errors if revisions get sparse,
3061# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3062# one of my favorite modules is out :<  Next up would be one of the DBM
3063# modules, but I'm not sure which is most portable...  So I'll just
3064# go with something that's plain-text, but still capable of
3065# being randomly accessed.  So here's my ultra-simple fixed-width
3066# database.  All records are 40 characters + "\n", so it's easy to seek
3067# to a revision: (41 * rev) is the byte offset.
3068# A record of 40 0s denotes an empty revision.
3069# And yes, it's still pretty fast (faster than Tie::File).
3070sub revdb_set {
3071        my ($file, $rev, $commit) = @_;
3072        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
3073        open my $fh, '+<', $file or croak $!;
3074        my $offset = $rev * 41;
3075        # assume that append is the common case:
3076        seek $fh, 0, 2 or croak $!;
3077        my $pos = tell $fh;
3078        if ($pos < $offset) {
3079                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
3080        }
3081        seek $fh, $offset, 0 or croak $!;
3082        print $fh $commit,"\n";
3083        close $fh or croak $!;
3084}
3085
3086sub revdb_get {
3087        my ($file, $rev) = @_;
3088        my $ret;
3089        my $offset = $rev * 41;
3090        open my $fh, '<', $file or croak $!;
3091        seek $fh, $offset, 0;
3092        if (tell $fh == $offset) {
3093                $ret = readline $fh;
3094                if (defined $ret) {
3095                        chomp $ret;
3096                        $ret = undef if ($ret =~ /^0{40}$/);
3097                }
3098        }
3099        close $fh or croak $!;
3100        return $ret;
3101}
3102
3103sub copy_remote_ref {
3104        my $origin = $_cp_remote ? $_cp_remote : 'origin';
3105        my $ref = "refs/remotes/$GIT_SVN";
3106        if (safe_qx('git-ls-remote', $origin, $ref)) {
3107                sys(qw/git fetch/, $origin, "$ref:$ref");
3108        } else {
3109                die "Unable to find remote reference: ",
3110                                "refs/remotes/$GIT_SVN on $origin\n";
3111        }
3112}
3113
3114package SVN::Git::Editor;
3115use vars qw/@ISA/;
3116use strict;
3117use warnings;
3118use Carp qw/croak/;
3119use IO::File;
3120
3121sub new {
3122        my $class = shift;
3123        my $git_svn = shift;
3124        my $self = SVN::Delta::Editor->new(@_);
3125        bless $self, $class;
3126        foreach (qw/svn_path c r ra /) {
3127                die "$_ required!\n" unless (defined $git_svn->{$_});
3128                $self->{$_} = $git_svn->{$_};
3129        }
3130        $self->{pool} = SVN::Pool->new;
3131        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3132        $self->{rm} = { };
3133        require Digest::MD5;
3134        return $self;
3135}
3136
3137sub split_path {
3138        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3139}
3140
3141sub repo_path {
3142        (defined $_[1] && length $_[1]) ? "$_[0]->{svn_path}/$_[1]"
3143                                        : $_[0]->{svn_path}
3144}
3145
3146sub url_path {
3147        my ($self, $path) = @_;
3148        $self->{ra}->{url} . '/' . $self->repo_path($path);
3149}
3150
3151sub rmdirs {
3152        my ($self, $q) = @_;
3153        my $rm = $self->{rm};
3154        delete $rm->{''}; # we never delete the url we're tracking
3155        return unless %$rm;
3156
3157        foreach (keys %$rm) {
3158                my @d = split m#/#, $_;
3159                my $c = shift @d;
3160                $rm->{$c} = 1;
3161                while (@d) {
3162                        $c .= '/' . shift @d;
3163                        $rm->{$c} = 1;
3164                }
3165        }
3166        delete $rm->{$self->{svn_path}};
3167        delete $rm->{''}; # we never delete the url we're tracking
3168        return unless %$rm;
3169
3170        defined(my $pid = open my $fh,'-|') or croak $!;
3171        if (!$pid) {
3172                exec qw/git-ls-tree --name-only -r -z/, $self->{c} or croak $!;
3173        }
3174        local $/ = "\0";
3175        my @svn_path = split m#/#, $self->{svn_path};
3176        while (<$fh>) {
3177                chomp;
3178                my @dn = (@svn_path, (split m#/#, $_));
3179                while (pop @dn) {
3180                        delete $rm->{join '/', @dn};
3181                }
3182                unless (%$rm) {
3183                        close $fh;
3184                        return;
3185                }
3186        }
3187        close $fh;
3188
3189        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3190        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3191                $self->close_directory($bat->{$d}, $p);
3192                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3193                print "\tD+\t/$d/\n" unless $q;
3194                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3195                delete $bat->{$d};
3196        }
3197}
3198
3199sub open_or_add_dir {
3200        my ($self, $full_path, $baton) = @_;
3201        my $p = SVN::Pool->new;
3202        my $t = $self->{ra}->check_path($full_path, $self->{r}, $p);
3203        $p->clear;
3204        if ($t == $SVN::Node::none) {
3205                return $self->add_directory($full_path, $baton,
3206                                                undef, -1, $self->{pool});
3207        } elsif ($t == $SVN::Node::dir) {
3208                return $self->open_directory($full_path, $baton,
3209                                                $self->{r}, $self->{pool});
3210        }
3211        print STDERR "$full_path already exists in repository at ",
3212                "r$self->{r} and it is not a directory (",
3213                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3214        exit 1;
3215}
3216
3217sub ensure_path {
3218        my ($self, $path) = @_;
3219        my $bat = $self->{bat};
3220        $path = $self->repo_path($path);
3221        return $bat->{''} unless (length $path);
3222        my @p = split m#/+#, $path;
3223        my $c = shift @p;
3224        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3225        while (@p) {
3226                my $c0 = $c;
3227                $c .= '/' . shift @p;
3228                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3229        }
3230        return $bat->{$c};
3231}
3232
3233sub A {
3234        my ($self, $m, $q) = @_;
3235        my ($dir, $file) = split_path($m->{file_b});
3236        my $pbat = $self->ensure_path($dir);
3237        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3238                                        undef, -1);
3239        print "\tA\t$m->{file_b}\n" unless $q;
3240        $self->chg_file($fbat, $m);
3241        $self->close_file($fbat,undef,$self->{pool});
3242}
3243
3244sub C {
3245        my ($self, $m, $q) = @_;
3246        my ($dir, $file) = split_path($m->{file_b});
3247        my $pbat = $self->ensure_path($dir);
3248        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3249                                $self->url_path($m->{file_a}), $self->{r});
3250        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $q;
3251        $self->chg_file($fbat, $m);
3252        $self->close_file($fbat,undef,$self->{pool});
3253}
3254
3255sub delete_entry {
3256        my ($self, $path, $pbat) = @_;
3257        my $rpath = $self->repo_path($path);
3258        my ($dir, $file) = split_path($rpath);
3259        $self->{rm}->{$dir} = 1;
3260        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3261}
3262
3263sub R {
3264        my ($self, $m, $q) = @_;
3265        my ($dir, $file) = split_path($m->{file_b});
3266        my $pbat = $self->ensure_path($dir);
3267        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3268                                $self->url_path($m->{file_a}), $self->{r});
3269        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $q;
3270        $self->chg_file($fbat, $m);
3271        $self->close_file($fbat,undef,$self->{pool});
3272
3273        ($dir, $file) = split_path($m->{file_a});
3274        $pbat = $self->ensure_path($dir);
3275        $self->delete_entry($m->{file_a}, $pbat);
3276}
3277
3278sub M {
3279        my ($self, $m, $q) = @_;
3280        my ($dir, $file) = split_path($m->{file_b});
3281        my $pbat = $self->ensure_path($dir);
3282        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3283                                $pbat,$self->{r},$self->{pool});
3284        print "\t$m->{chg}\t$m->{file_b}\n" unless $q;
3285        $self->chg_file($fbat, $m);
3286        $self->close_file($fbat,undef,$self->{pool});
3287}
3288
3289sub T { shift->M(@_) }
3290
3291sub change_file_prop {
3292        my ($self, $fbat, $pname, $pval) = @_;
3293        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3294}
3295
3296sub chg_file {
3297        my ($self, $fbat, $m) = @_;
3298        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3299                $self->change_file_prop($fbat,'svn:executable','*');
3300        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3301                $self->change_file_prop($fbat,'svn:executable',undef);
3302        }
3303        my $fh = IO::File->new_tmpfile or croak $!;
3304        if ($m->{mode_b} =~ /^120/) {
3305                print $fh 'link ' or croak $!;
3306                $self->change_file_prop($fbat,'svn:special','*');
3307        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3308                $self->change_file_prop($fbat,'svn:special',undef);
3309        }
3310        defined(my $pid = fork) or croak $!;
3311        if (!$pid) {
3312                open STDOUT, '>&', $fh or croak $!;
3313                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3314        }
3315        waitpid $pid, 0;
3316        croak $? if $?;
3317        $fh->flush == 0 or croak $!;
3318        seek $fh, 0, 0 or croak $!;
3319
3320        my $md5 = Digest::MD5->new;
3321        $md5->addfile($fh) or croak $!;
3322        seek $fh, 0, 0 or croak $!;
3323
3324        my $exp = $md5->hexdigest;
3325        my $atd = $self->apply_textdelta($fbat, undef, $self->{pool});
3326        my $got = SVN::TxDelta::send_stream($fh, @$atd, $self->{pool});
3327        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3328
3329        close $fh or croak $!;
3330}
3331
3332sub D {
3333        my ($self, $m, $q) = @_;
3334        my ($dir, $file) = split_path($m->{file_b});
3335        my $pbat = $self->ensure_path($dir);
3336        print "\tD\t$m->{file_b}\n" unless $q;
3337        $self->delete_entry($m->{file_b}, $pbat);
3338}
3339
3340sub close_edit {
3341        my ($self) = @_;
3342        my ($p,$bat) = ($self->{pool}, $self->{bat});
3343        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3344                $self->close_directory($bat->{$_}, $p);
3345        }
3346        $self->SUPER::close_edit($p);
3347        $p->clear;
3348}
3349
3350sub abort_edit {
3351        my ($self) = @_;
3352        $self->SUPER::abort_edit($self->{pool});
3353        $self->{pool}->clear;
3354}
3355
3356__END__
3357
3358Data structures:
3359
3360$svn_log hashref (as returned by svn_log_raw)
3361{
3362        fh => file handle of the log file,
3363        state => state of the log file parser (sep/msg/rev/msg_start...)
3364}
3365
3366$log_msg hashref as returned by next_log_entry($svn_log)
3367{
3368        msg => 'whitespace-formatted log entry
3369',                                              # trailing newline is preserved
3370        revision => '8',                        # integer
3371        date => '2004-02-24T17:01:44.108345Z',  # commit date
3372        author => 'committer name'
3373};
3374
3375
3376@mods = array of diff-index line hashes, each element represents one line
3377        of diff-index output
3378
3379diff-index line ($m hash)
3380{
3381        mode_a => first column of diff-index output, no leading ':',
3382        mode_b => second column of diff-index output,
3383        sha1_b => sha1sum of the final blob,
3384        chg => change type [MCRADT],
3385        file_a => original file name of a file (iff chg is 'C' or 'R')
3386        file_b => new/current file name of a file (any chg)
3387}
3388;
3389
3390# retval of read_url_paths{,_all}();
3391$l_map = {
3392        # repository root url
3393        'https://svn.musicpd.org' => {
3394                # repository path               # GIT_SVN_ID
3395                'mpd/trunk'             =>      'trunk',
3396                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
3397        },
3398}
3399
3400Notes:
3401        I don't trust the each() function on unless I created %hash myself
3402        because the internal iterator may not have started at base.