Andrew's git - gitweb.git/blob

   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, $_use_lib);
  43
  44sub nag_lib {
  45        print STDERR <<EOF;
  46! Please consider installing the SVN Perl libraries (version 1.1.0 or
  47! newer).  You will generally get better performance and fewer bugs,
  48! especially if you:
  49! 1) have a case-insensitive filesystem
  50! 2) replace symlinks with files (and vice-versa) in commits
  51
  52EOF
  53}
  54
  55$_use_lib = 1 unless $ENV{GIT_SVN_NO_LIB};
  56libsvn_load();
  57nag_lib() unless $_use_lib;
  58
  59my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  60my $sha1 = qr/[a-f\d]{40}/;
  61my $sha1_short = qr/[a-f\d]{4,40}/;
  62my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
  63        $_find_copies_harder, $_l, $_cp_similarity, $_cp_remote,
  64        $_repack, $_repack_nr, $_repack_flags, $_q,
  65        $_message, $_file, $_follow_parent, $_no_metadata,
  66        $_template, $_shared, $_no_default_regex, $_no_graft_copy,
  67        $_limit, $_verbose, $_incremental, $_oneline, $_l_fmt, $_show_commit,
  68        $_version, $_upgrade, $_authors, $_branch_all_refs, @_opt_m,
  69        $_merge, $_strategy, $_dry_run, $_ignore_nodate, $_non_recursive);
  70my (@_branch_from, %tree_map, %users, %rusers, %equiv);
  71my ($_svn_co_url_revs, $_svn_pg_peg_revs);
  72my @repo_path_split_cache;
  73
  74my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
  75                'branch|b=s' => \@_branch_from,
  76                'follow-parent|follow' => \$_follow_parent,
  77                'branch-all-refs|B' => \$_branch_all_refs,
  78                'authors-file|A=s' => \$_authors,
  79                'repack:i' => \$_repack,
  80                'no-metadata' => \$_no_metadata,
  81                'quiet|q' => \$_q,
  82                'ignore-nodate' => \$_ignore_nodate,
  83                'repack-flags|repack-args|repack-opts=s' => \$_repack_flags);
  84
  85my ($_trunk, $_tags, $_branches);
  86my %multi_opts = ( 'trunk|T=s' => \$_trunk,
  87                'tags|t=s' => \$_tags,
  88                'branches|b=s' => \$_branches );
  89my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
  90my %cmt_opts = ( 'edit|e' => \$_edit,
  91                'rmdir' => \$_rmdir,
  92                'find-copies-harder' => \$_find_copies_harder,
  93                'l=i' => \$_l,
  94                'copy-similarity|C=i'=> \$_cp_similarity
  95);
  96
  97my %cmd = (
  98        fetch => [ \&fetch, "Download new revisions from SVN",
  99                        { 'revision|r=s' => \$_revision, %fc_opts } ],
 100        init => [ \&init, "Initialize a repo for tracking" .
 101                          " (requires URL argument)",
 102                          \%init_opts ],
 103        commit => [ \&commit, "Commit git revisions to SVN",
 104                        {       'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 105        'show-ignore' => [ \&show_ignore, "Show svn:ignore listings",
 106                        { 'revision|r=i' => \$_revision } ],
 107        rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
 108                        { 'no-ignore-externals' => \$_no_ignore_ext,
 109                          'copy-remote|remote=s' => \$_cp_remote,
 110                          'upgrade' => \$_upgrade } ],
 111        'graft-branches' => [ \&graft_branches,
 112                        'Detect merges/branches from already imported history',
 113                        { 'merge-rx|m' => \@_opt_m,
 114                          'branch|b=s' => \@_branch_from,
 115                          'branch-all-refs|B' => \$_branch_all_refs,
 116                          'no-default-regex' => \$_no_default_regex,
 117                          'no-graft-copy' => \$_no_graft_copy } ],
 118        'multi-init' => [ \&multi_init,
 119                        'Initialize multiple trees (like git-svnimport)',
 120                        { %multi_opts, %fc_opts } ],
 121        'multi-fetch' => [ \&multi_fetch,
 122                        'Fetch multiple trees (like git-svnimport)',
 123                        \%fc_opts ],
 124        'log' => [ \&show_log, 'Show commit logs',
 125                        { 'limit=i' => \$_limit,
 126                          'revision|r=s' => \$_revision,
 127                          'verbose|v' => \$_verbose,
 128                          'incremental' => \$_incremental,
 129                          'oneline' => \$_oneline,
 130                          'show-commit' => \$_show_commit,
 131                          'non-recursive' => \$_non_recursive,
 132                          'authors-file|A=s' => \$_authors,
 133                        } ],
 134        'commit-diff' => [ \&commit_diff, 'Commit a diff between two trees',
 135                        { 'message|m=s' => \$_message,
 136                          'file|F=s' => \$_file,
 137                          'revision|r=s' => \$_revision,
 138                        %cmt_opts } ],
 139        dcommit => [ \&dcommit, 'Commit several diffs to merge with upstream',
 140                        { 'merge|m|M' => \$_merge,
 141                          'strategy|s=s' => \$_strategy,
 142                          'dry-run|n' => \$_dry_run,
 143                        %cmt_opts } ],
 144);
 145
 146my $cmd;
 147for (my $i = 0; $i < @ARGV; $i++) {
 148        if (defined $cmd{$ARGV[$i]}) {
 149                $cmd = $ARGV[$i];
 150                splice @ARGV, $i, 1;
 151                last;
 152        }
 153};
 154
 155my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 156
 157read_repo_config(\%opts);
 158my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
 159                                'version|V' => \$_version,
 160                                'id|i=s' => \$GIT_SVN);
 161exit 1 if (!$rv && $cmd ne 'log');
 162
 163set_default_vals();
 164usage(0) if $_help;
 165version() if $_version;
 166usage(1) unless defined $cmd;
 167init_vars();
 168load_authors() if $_authors;
 169load_all_refs() if $_branch_all_refs;
 170svn_compat_check() unless $_use_lib;
 171migration_check() unless $cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/;
 172$cmd{$cmd}->[0]->(@ARGV);
 173exit 0;
 174
 175####################### primary functions ######################
 176sub usage {
 177        my $exit = shift || 0;
 178        my $fd = $exit ? \*STDERR : \*STDOUT;
 179        print $fd <<"";
 180git-svn - bidirectional operations between a single Subversion tree and git
 181Usage: $0 <command> [options] [arguments]\n
 182
 183        print $fd "Available commands:\n" unless $cmd;
 184
 185        foreach (sort keys %cmd) {
 186                next if $cmd && $cmd ne $_;
 187                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 188                foreach (keys %{$cmd{$_}->[2]}) {
 189                        # prints out arguments as they should be passed:
 190                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 191                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 192                                                        "--$_" : "-$_" }
 193                                                split /\|/,$_)," $x\n";
 194                }
 195        }
 196        print $fd <<"";
 197\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 198arbitrary identifier if you're tracking multiple SVN branches/repositories in
 199one git repository and want to keep them separate.  See git-svn(1) for more
 200information.
 201
 202        exit $exit;
 203}
 204
 205sub version {
 206        print "git-svn version $VERSION\n";
 207        exit 0;
 208}
 209
 210sub rebuild {
 211        if (quiet_run(qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0")) {
 212                copy_remote_ref();
 213        }
 214        $SVN_URL = shift or undef;
 215        my $newest_rev = 0;
 216        if ($_upgrade) {
 217                sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD");
 218        } else {
 219                check_upgrade_needed();
 220        }
 221
 222        my $pid = open(my $rev_list,'-|');
 223        defined $pid or croak $!;
 224        if ($pid == 0) {
 225                exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!;
 226        }
 227        my $latest;
 228        while (<$rev_list>) {
 229                chomp;
 230                my $c = $_;
 231                croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 232                my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
 233                next if (!@commit); # skip merges
 234                my ($url, $rev, $uuid) = extract_metadata($commit[$#commit]);
 235                if (!defined $rev || !$uuid) {
 236                        croak "Unable to extract revision or UUID from ",
 237                                "$c, $commit[$#commit]\n";
 238                }
 239
 240                # if we merged or otherwise started elsewhere, this is
 241                # how we break out of it
 242                next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
 243                next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
 244
 245                unless (defined $latest) {
 246                        if (!$SVN_URL && !$url) {
 247                                croak "SVN repository location required: $url\n";
 248                        }
 249                        $SVN_URL ||= $url;
 250                        $SVN_UUID ||= $uuid;
 251                        setup_git_svn();
 252                        $latest = $rev;
 253                }
 254                revdb_set($REVDB, $rev, $c);
 255                print "r$rev = $c\n";
 256                $newest_rev = $rev if ($rev > $newest_rev);
 257        }
 258        close $rev_list or croak $?;
 259
 260        goto out if $_use_lib;
 261        if (!chdir $SVN_WC) {
 262                svn_cmd_checkout($SVN_URL, $latest, $SVN_WC);
 263                chdir $SVN_WC or croak $!;
 264        }
 265
 266        $pid = fork;
 267        defined $pid or croak $!;
 268        if ($pid == 0) {
 269                my @svn_up = qw(svn up);
 270                push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 271                sys(@svn_up,"-r$newest_rev");
 272                $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 273                index_changes();
 274                exec('git-write-tree') or croak $!;
 275        }
 276        waitpid $pid, 0;
 277        croak $? if $?;
 278out:
 279        if ($_upgrade) {
 280                print STDERR <<"";
 281Keeping deprecated refs/head/$GIT_SVN-HEAD for now.  Please remove it
 282when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN
 283
 284        }
 285}
 286
 287sub init {
 288        my $url = shift or die "SVN repository location required " .
 289                                "as a command-line argument\n";
 290        $url =~ s!/+$!!; # strip trailing slash
 291
 292        if (my $repo_path = shift) {
 293                unless (-d $repo_path) {
 294                        mkpath([$repo_path]);
 295                }
 296                $GIT_DIR = $ENV{GIT_DIR} = $repo_path . "/.git";
 297                init_vars();
 298        }
 299
 300        $SVN_URL = $url;
 301        unless (-d $GIT_DIR) {
 302                my @init_db = ('git-init-db');
 303                push @init_db, "--template=$_template" if defined $_template;
 304                push @init_db, "--shared" if defined $_shared;
 305                sys(@init_db);
 306        }
 307        setup_git_svn();
 308}
 309
 310sub fetch {
 311        check_upgrade_needed();
 312        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 313        my $ret = $_use_lib ? fetch_lib(@_) : fetch_cmd(@_);
 314        if ($ret->{commit} && quiet_run(qw(git-rev-parse --verify
 315                                                refs/heads/master^0))) {
 316                sys(qw(git-update-ref refs/heads/master),$ret->{commit});
 317        }
 318        return $ret;
 319}
 320
 321sub fetch_cmd {
 322        my (@parents) = @_;
 323        my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 324        unless ($_revision) {
 325                $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 326        }
 327        push @log_args, "-r$_revision";
 328        push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 329
 330        my $svn_log = svn_log_raw(@log_args);
 331
 332        my $base = next_log_entry($svn_log) or croak "No base revision!\n";
 333        # don't need last_revision from grab_base_rev() because
 334        # user could've specified a different revision to skip (they
 335        # didn't want to import certain revisions into git for whatever
 336        # reason, so trust $base->{revision} instead.
 337        my (undef, $last_commit) = svn_grab_base_rev();
 338        unless (-d $SVN_WC) {
 339                svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC);
 340                chdir $SVN_WC or croak $!;
 341                read_uuid();
 342                $last_commit = git_commit($base, @parents);
 343                assert_tree($last_commit);
 344        } else {
 345                chdir $SVN_WC or croak $!;
 346                read_uuid();
 347                # looks like a user manually cp'd and svn switch'ed
 348                unless ($last_commit) {
 349                        sys(qw/svn revert -R ./);
 350                        assert_svn_wc_clean($base->{revision});
 351                        $last_commit = git_commit($base, @parents);
 352                        assert_tree($last_commit);
 353                }
 354        }
 355        my @svn_up = qw(svn up);
 356        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 357        my $last = $base;
 358        while (my $log_msg = next_log_entry($svn_log)) {
 359                if ($last->{revision} >= $log_msg->{revision}) {
 360                        croak "Out of order: last >= current: ",
 361                                "$last->{revision} >= $log_msg->{revision}\n";
 362                }
 363                # Revert is needed for cases like:
 364                # https://svn.musicpd.org/Jamming/trunk (r166:167), but
 365                # I can't seem to reproduce something like that on a test...
 366                sys(qw/svn revert -R ./);
 367                assert_svn_wc_clean($last->{revision});
 368                sys(@svn_up,"-r$log_msg->{revision}");
 369                $last_commit = git_commit($log_msg, $last_commit, @parents);
 370                $last = $log_msg;
 371        }
 372        close $svn_log->{fh};
 373        $last->{commit} = $last_commit;
 374        return $last;
 375}
 376
 377sub fetch_lib {
 378        my (@parents) = @_;
 379        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 380        my $repo;
 381        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 382        $SVN_LOG ||= libsvn_connect($repo);
 383        $SVN ||= libsvn_connect($repo);
 384        my ($last_rev, $last_commit) = svn_grab_base_rev();
 385        my ($base, $head) = libsvn_parse_revision($last_rev);
 386        if ($base > $head) {
 387                return { revision => $last_rev, commit => $last_commit }
 388        }
 389        my $index = set_index($GIT_SVN_INDEX);
 390
 391        # limit ourselves and also fork() since get_log won't release memory
 392        # after processing a revision and SVN stuff seems to leak
 393        my $inc = 1000;
 394        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
 395        read_uuid();
 396        if (defined $last_commit) {
 397                unless (-e $GIT_SVN_INDEX) {
 398                        sys(qw/git-read-tree/, $last_commit);
 399                }
 400                chomp (my $x = `git-write-tree`);
 401                my ($y) = (`git-cat-file commit $last_commit`
 402                                                        =~ /^tree ($sha1)/m);
 403                if ($y ne $x) {
 404                        unlink $GIT_SVN_INDEX or croak $!;
 405                        sys(qw/git-read-tree/, $last_commit);
 406                }
 407                chomp ($x = `git-write-tree`);
 408                if ($y ne $x) {
 409                        print STDERR "trees ($last_commit) $y != $x\n",
 410                                 "Something is seriously wrong...\n";
 411                }
 412        }
 413        while (1) {
 414                # fork, because using SVN::Pool with get_log() still doesn't
 415                # seem to help enough to keep memory usage down.
 416                defined(my $pid = fork) or croak $!;
 417                if (!$pid) {
 418                        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
 419
 420                        # Yes I'm perfectly aware that the fourth argument
 421                        # below is the limit revisions number.  Unfortunately
 422                        # performance sucks with it enabled, so it's much
 423                        # faster to fetch revision ranges instead of relying
 424                        # on the limiter.
 425                        libsvn_get_log($SVN_LOG, '/'.$SVN_PATH,
 426                                        $min, $max, 0, 1, 1,
 427                                sub {
 428                                        my $log_msg;
 429                                        if ($last_commit) {
 430                                                $log_msg = libsvn_fetch(
 431                                                        $last_commit, @_);
 432                                                $last_commit = git_commit(
 433                                                        $log_msg,
 434                                                        $last_commit,
 435                                                        @parents);
 436                                        } else {
 437                                                $log_msg = libsvn_new_tree(@_);
 438                                                $last_commit = git_commit(
 439                                                        $log_msg, @parents);
 440                                        }
 441                                });
 442                        exit 0;
 443                }
 444                waitpid $pid, 0;
 445                croak $? if $?;
 446                ($last_rev, $last_commit) = svn_grab_base_rev();
 447                last if ($max >= $head);
 448                $min = $max + 1;
 449                $max += $inc;
 450                $max = $head if ($max > $head);
 451        }
 452        restore_index($index);
 453        return { revision => $last_rev, commit => $last_commit };
 454}
 455
 456sub commit {
 457        my (@commits) = @_;
 458        check_upgrade_needed();
 459        if ($_stdin || !@commits) {
 460                print "Reading from stdin...\n";
 461                @commits = ();
 462                while (<STDIN>) {
 463                        if (/\b($sha1_short)\b/o) {
 464                                unshift @commits, $1;
 465                        }
 466                }
 467        }
 468        my @revs;
 469        foreach my $c (@commits) {
 470                chomp(my @tmp = safe_qx('git-rev-parse',$c));
 471                if (scalar @tmp == 1) {
 472                        push @revs, $tmp[0];
 473                } elsif (scalar @tmp > 1) {
 474                        push @revs, reverse (safe_qx('git-rev-list',@tmp));
 475                } else {
 476                        die "Failed to rev-parse $c\n";
 477                }
 478        }
 479        chomp @revs;
 480        $_use_lib ? commit_lib(@revs) : commit_cmd(@revs);
 481        print "Done committing ",scalar @revs," revisions to SVN\n";
 482}
 483
 484sub commit_cmd {
 485        my (@revs) = @_;
 486
 487        chdir $SVN_WC or croak "Unable to chdir $SVN_WC: $!\n";
 488        my $info = svn_info('.');
 489        my $fetched = fetch();
 490        if ($info->{Revision} != $fetched->{revision}) {
 491                print STDERR "There are new revisions that were fetched ",
 492                                "and need to be merged (or acknowledged) ",
 493                                "before committing.\n";
 494                exit 1;
 495        }
 496        $info = svn_info('.');
 497        read_uuid($info);
 498        my $last = $fetched;
 499        foreach my $c (@revs) {
 500                my $mods = svn_checkout_tree($last, $c);
 501                if (scalar @$mods == 0) {
 502                        print "Skipping, no changes detected\n";
 503                        next;
 504                }
 505                $last = svn_commit_tree($last, $c);
 506        }
 507}
 508
 509sub commit_lib {
 510        my (@revs) = @_;
 511        my ($r_last, $cmt_last) = svn_grab_base_rev();
 512        defined $r_last or die "Must have an existing revision to commit\n";
 513        my $fetched = fetch();
 514        if ($r_last != $fetched->{revision}) {
 515                print STDERR "There are new revisions that were fetched ",
 516                                "and need to be merged (or acknowledged) ",
 517                                "before committing.\n",
 518                                "last rev: $r_last\n",
 519                                " current: $fetched->{revision}\n";
 520                exit 1;
 521        }
 522        read_uuid();
 523        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 524        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
 525
 526        my $repo;
 527        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 528        set_svn_commit_env();
 529        foreach my $c (@revs) {
 530                my $log_msg = get_commit_message($c, $commit_msg);
 531
 532                # fork for each commit because there's a memory leak I
 533                # can't track down... (it's probably in the SVN code)
 534                defined(my $pid = open my $fh, '-|') or croak $!;
 535                if (!$pid) {
 536                        $SVN_LOG = libsvn_connect($repo);
 537                        $SVN = libsvn_connect($repo);
 538                        my $ed = SVN::Git::Editor->new(
 539                                        {       r => $r_last,
 540                                                ra => $SVN_LOG,
 541                                                c => $c,
 542                                                svn_path => $SVN_PATH
 543                                        },
 544                                        $SVN->get_commit_editor(
 545                                                $log_msg->{msg},
 546                                                sub {
 547                                                        libsvn_commit_cb(
 548                                                                @_, $c,
 549                                                                $log_msg->{msg},
 550                                                                $r_last,
 551                                                                $cmt_last)
 552                                                },
 553                                                @lock)
 554                                        );
 555                        my $mods = libsvn_checkout_tree($cmt_last, $c, $ed);
 556                        if (@$mods == 0) {
 557                                print "No changes\nr$r_last = $cmt_last\n";
 558                                $ed->abort_edit;
 559                        } else {
 560                                $ed->close_edit;
 561                        }
 562                        exit 0;
 563                }
 564                my ($r_new, $cmt_new, $no);
 565                while (<$fh>) {
 566                        print $_;
 567                        chomp;
 568                        if (/^r(\d+) = ($sha1)$/o) {
 569                                ($r_new, $cmt_new) = ($1, $2);
 570                        } elsif ($_ eq 'No changes') {
 571                                $no = 1;
 572                        }
 573                }
 574                close $fh or croak $?;
 575                if (! defined $r_new && ! defined $cmt_new) {
 576                        unless ($no) {
 577                                die "Failed to parse revision information\n";
 578                        }
 579                } else {
 580                        ($r_last, $cmt_last) = ($r_new, $cmt_new);
 581                }
 582        }
 583        $ENV{LC_ALL} = 'C';
 584        unlink $commit_msg;
 585}
 586
 587sub dcommit {
 588        my $gs = "refs/remotes/$GIT_SVN";
 589        chomp(my @refs = safe_qx(qw/git-rev-list --no-merges/, "$gs..HEAD"));
 590        my $last_rev;
 591        foreach my $d (reverse @refs) {
 592                if (quiet_run('git-rev-parse','--verify',"$d~1") != 0) {
 593                        die "Commit $d\n",
 594                            "has no parent commit, and therefore ",
 595                            "nothing to diff against.\n",
 596                            "You should be working from a repository ",
 597                            "originally created by git-svn\n";
 598                }
 599                unless (defined $last_rev) {
 600                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 601                        unless (defined $last_rev) {
 602                                die "Unable to extract revision information ",
 603                                    "from commit $d~1\n";
 604                        }
 605                }
 606                if ($_dry_run) {
 607                        print "diff-tree $d~1 $d\n";
 608                } else {
 609                        if (my $r = commit_diff("$d~1", $d, undef, $last_rev)) {
 610                                $last_rev = $r;
 611                        } # else: no changes, same $last_rev
 612                }
 613        }
 614        return if $_dry_run;
 615        fetch();
 616        my @diff = safe_qx(qw/git-diff-tree HEAD/, $gs);
 617        my @finish;
 618        if (@diff) {
 619                @finish = qw/rebase/;
 620                push @finish, qw/--merge/ if $_merge;
 621                push @finish, "--strategy=$_strategy" if $_strategy;
 622                print STDERR "W: HEAD and $gs differ, using @finish:\n", @diff;
 623        } else {
 624                print "No changes between current HEAD and $gs\n",
 625                      "Hard resetting to the latest $gs\n";
 626                @finish = qw/reset --mixed/;
 627        }
 628        sys('git', @finish, $gs);
 629}
 630
 631sub show_ignore {
 632        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 633        $_use_lib ? show_ignore_lib() : show_ignore_cmd();
 634}
 635
 636sub show_ignore_cmd {
 637        require File::Find or die $!;
 638        if (defined $_revision) {
 639                die "-r/--revision option doesn't work unless the Perl SVN ",
 640                        "libraries are used\n";
 641        }
 642        chdir $SVN_WC or croak $!;
 643        my %ign;
 644        File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 645                s#^\./##;
 646                @{$ign{$_}} = svn_propget_base('svn:ignore', $_);
 647                }}, no_chdir=>1},'.');
 648
 649        print "\n# /\n";
 650        foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 651        delete $ign{'.'};
 652        foreach my $i (sort keys %ign) {
 653                print "\n# ",$i,"\n";
 654                foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 655        }
 656}
 657
 658sub show_ignore_lib {
 659        my $repo;
 660        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 661        $SVN ||= libsvn_connect($repo);
 662        my $r = defined $_revision ? $_revision : $SVN->get_latest_revnum;
 663        libsvn_traverse_ignore(\*STDOUT, $SVN_PATH, $r);
 664}
 665
 666sub graft_branches {
 667        my $gr_file = "$GIT_DIR/info/grafts";
 668        my ($grafts, $comments) = read_grafts($gr_file);
 669        my $gr_sha1;
 670
 671        if (%$grafts) {
 672                # temporarily disable our grafts file to make this idempotent
 673                chomp($gr_sha1 = safe_qx(qw/git-hash-object -w/,$gr_file));
 674                rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
 675        }
 676
 677        my $l_map = read_url_paths();
 678        my @re = map { qr/$_/is } @_opt_m if @_opt_m;
 679        unless ($_no_default_regex) {
 680                push @re, (qr/\b(?:merge|merging|merged)\s+with\s+([\w\.\-]+)/i,
 681                        qr/\b(?:merge|merging|merged)\s+([\w\.\-]+)/i,
 682                        qr/\b(?:from|of)\s+([\w\.\-]+)/i );
 683        }
 684        foreach my $u (keys %$l_map) {
 685                if (@re) {
 686                        foreach my $p (keys %{$l_map->{$u}}) {
 687                                graft_merge_msg($grafts,$l_map,$u,$p,@re);
 688                        }
 689                }
 690                unless ($_no_graft_copy) {
 691                        if ($_use_lib) {
 692                                graft_file_copy_lib($grafts,$l_map,$u);
 693                        } else {
 694                                graft_file_copy_cmd($grafts,$l_map,$u);
 695                        }
 696                }
 697        }
 698        graft_tree_joins($grafts);
 699
 700        write_grafts($grafts, $comments, $gr_file);
 701        unlink "$gr_file~$gr_sha1" if $gr_sha1;
 702}
 703
 704sub multi_init {
 705        my $url = shift;
 706        $_trunk ||= 'trunk';
 707        $_trunk =~ s#/+$##;
 708        $url =~ s#/+$## if $url;
 709        if ($_trunk !~ m#^[a-z\+]+://#) {
 710                $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#);
 711                unless ($url) {
 712                        print STDERR "E: '$_trunk' is not a complete URL ",
 713                                "and a separate URL is not specified\n";
 714                        exit 1;
 715                }
 716                $_trunk = $url . $_trunk;
 717        }
 718        my $ch_id;
 719        if ($GIT_SVN eq 'git-svn') {
 720                $ch_id = 1;
 721                $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
 722        }
 723        init_vars();
 724        unless (-d $GIT_SVN_DIR) {
 725                print "GIT_SVN_ID set to 'trunk' for $_trunk\n" if $ch_id;
 726                init($_trunk);
 727                sys('git-repo-config', 'svn.trunk', $_trunk);
 728        }
 729        complete_url_ls_init($url, $_branches, '--branches/-b', '');
 730        complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/');
 731}
 732
 733sub multi_fetch {
 734        # try to do trunk first, since branches/tags
 735        # may be descended from it.
 736        if (-e "$GIT_DIR/svn/trunk/info/url") {
 737                fetch_child_id('trunk', @_);
 738        }
 739        rec_fetch('', "$GIT_DIR/svn", @_);
 740}
 741
 742sub show_log {
 743        my (@args) = @_;
 744        my ($r_min, $r_max);
 745        my $r_last = -1; # prevent dupes
 746        rload_authors() if $_authors;
 747        if (defined $TZ) {
 748                $ENV{TZ} = $TZ;
 749        } else {
 750                delete $ENV{TZ};
 751        }
 752        if (defined $_revision) {
 753                if ($_revision =~ /^(\d+):(\d+)$/) {
 754                        ($r_min, $r_max) = ($1, $2);
 755                } elsif ($_revision =~ /^\d+$/) {
 756                        $r_min = $r_max = $_revision;
 757                } else {
 758                        print STDERR "-r$_revision is not supported, use ",
 759                                "standard \'git log\' arguments instead\n";
 760                        exit 1;
 761                }
 762        }
 763
 764        my $pid = open(my $log,'-|');
 765        defined $pid or croak $!;
 766        if (!$pid) {
 767                exec(git_svn_log_cmd($r_min,$r_max), @args) or croak $!;
 768        }
 769        setup_pager();
 770        my (@k, $c, $d);
 771
 772        while (<$log>) {
 773                if (/^commit ($sha1_short)/o) {
 774                        my $cmt = $1;
 775                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
 776                                $r_last = $c->{r};
 777                                process_commit($c, $r_min, $r_max, \@k) or
 778                                                                goto out;
 779                        }
 780                        $d = undef;
 781                        $c = { c => $cmt };
 782                } elsif (/^author (.+) (\d+) ([\-\+]?\d+)$/) {
 783                        get_author_info($c, $1, $2, $3);
 784                } elsif (/^(?:tree|parent|committer) /) {
 785                        # ignore
 786                } elsif (/^:\d{6} \d{6} $sha1_short/o) {
 787                        push @{$c->{raw}}, $_;
 788                } elsif (/^[ACRMDT]\t/) {
 789                        # we could add $SVN_PATH here, but that requires
 790                        # remote access at the moment (repo_path_split)...
 791                        s#^([ACRMDT])\t#   $1 #;
 792                        push @{$c->{changed}}, $_;
 793                } elsif (/^diff /) {
 794                        $d = 1;
 795                        push @{$c->{diff}}, $_;
 796                } elsif ($d) {
 797                        push @{$c->{diff}}, $_;
 798                } elsif (/^    (git-svn-id:.+)$/) {
 799                        ($c->{url}, $c->{r}, undef) = extract_metadata($1);
 800                } elsif (s/^    //) {
 801                        push @{$c->{l}}, $_;
 802                }
 803        }
 804        if ($c && defined $c->{r} && $c->{r} != $r_last) {
 805                $r_last = $c->{r};
 806                process_commit($c, $r_min, $r_max, \@k);
 807        }
 808        if (@k) {
 809                my $swap = $r_max;
 810                $r_max = $r_min;
 811                $r_min = $swap;
 812                process_commit($_, $r_min, $r_max) foreach reverse @k;
 813        }
 814out:
 815        close $log;
 816        print '-' x72,"\n" unless $_incremental || $_oneline;
 817}
 818
 819sub commit_diff_usage {
 820        print STDERR "Usage: $0 commit-diff <tree-ish> <tree-ish> [<URL>]\n";
 821        exit 1
 822}
 823
 824sub commit_diff {
 825        if (!$_use_lib) {
 826                print STDERR "commit-diff must be used with SVN libraries\n";
 827                exit 1;
 828        }
 829        my $ta = shift or commit_diff_usage();
 830        my $tb = shift or commit_diff_usage();
 831        if (!eval { $SVN_URL = shift || file_to_s("$GIT_SVN_DIR/info/url") }) {
 832                print STDERR "Needed URL or usable git-svn id command-line\n";
 833                commit_diff_usage();
 834        }
 835        my $r = shift;
 836        unless (defined $r) {
 837                if (defined $_revision) {
 838                        $r = $_revision
 839                } else {
 840                        die "-r|--revision is a required argument\n";
 841                }
 842        }
 843        if (defined $_message && defined $_file) {
 844                print STDERR "Both --message/-m and --file/-F specified ",
 845                                "for the commit message.\n",
 846                                "I have no idea what you mean\n";
 847                exit 1;
 848        }
 849        if (defined $_file) {
 850                $_message = file_to_s($_file);
 851        } else {
 852                $_message ||= get_commit_message($tb,
 853                                        "$GIT_DIR/.svn-commit.tmp.$$")->{msg};
 854        }
 855        my $repo;
 856        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 857        $SVN_LOG ||= libsvn_connect($repo);
 858        $SVN ||= libsvn_connect($repo);
 859        if ($r eq 'HEAD') {
 860                $r = $SVN->get_latest_revnum;
 861        } elsif ($r !~ /^\d+$/) {
 862                die "revision argument: $r not understood by git-svn\n";
 863        }
 864        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 865        my $rev_committed;
 866        my $ed = SVN::Git::Editor->new({        r => $r,
 867                                                ra => $SVN_LOG, c => $tb,
 868                                                svn_path => $SVN_PATH
 869                                        },
 870                                $SVN->get_commit_editor($_message,
 871                                        sub {
 872                                                $rev_committed = $_[0];
 873                                                print "Committed $_[0]\n";
 874                                        }, @lock)
 875                                );
 876        my $mods = libsvn_checkout_tree($ta, $tb, $ed);
 877        if (@$mods == 0) {
 878                print "No changes\n$ta == $tb\n";
 879                $ed->abort_edit;
 880        } else {
 881                $ed->close_edit;
 882        }
 883        $_message = $_file = undef;
 884        return $rev_committed;
 885}
 886
 887########################### utility functions #########################
 888
 889sub cmt_showable {
 890        my ($c) = @_;
 891        return 1 if defined $c->{r};
 892        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
 893                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
 894                my @msg = safe_qx(qw/git-cat-file commit/, $c->{c});
 895                shift @msg while ($msg[0] ne "\n");
 896                shift @msg;
 897                @{$c->{l}} = grep !/^git-svn-id: /, @msg;
 898
 899                (undef, $c->{r}, undef) = extract_metadata(
 900                                (grep(/^git-svn-id: /, @msg))[-1]);
 901        }
 902        return defined $c->{r};
 903}
 904
 905sub git_svn_log_cmd {
 906        my ($r_min, $r_max) = @_;
 907        my @cmd = (qw/git-log --abbrev-commit --pretty=raw
 908                        --default/, "refs/remotes/$GIT_SVN");
 909        push @cmd, '-r' unless $_non_recursive;
 910        push @cmd, qw/--raw --name-status/ if $_verbose;
 911        return @cmd unless defined $r_max;
 912        if ($r_max == $r_min) {
 913                push @cmd, '--max-count=1';
 914                if (my $c = revdb_get($REVDB, $r_max)) {
 915                        push @cmd, $c;
 916                }
 917        } else {
 918                my ($c_min, $c_max);
 919                $c_max = revdb_get($REVDB, $r_max);
 920                $c_min = revdb_get($REVDB, $r_min);
 921                if (defined $c_min && defined $c_max) {
 922                        if ($r_max > $r_max) {
 923                                push @cmd, "$c_min..$c_max";
 924                        } else {
 925                                push @cmd, "$c_max..$c_min";
 926                        }
 927                } elsif ($r_max > $r_min) {
 928                        push @cmd, $c_max;
 929                } else {
 930                        push @cmd, $c_min;
 931                }
 932        }
 933        return @cmd;
 934}
 935
 936sub fetch_child_id {
 937        my $id = shift;
 938        print "Fetching $id\n";
 939        my $ref = "$GIT_DIR/refs/remotes/$id";
 940        defined(my $pid = open my $fh, '-|') or croak $!;
 941        if (!$pid) {
 942                $_repack = undef;
 943                $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 944                init_vars();
 945                fetch(@_);
 946                exit 0;
 947        }
 948        while (<$fh>) {
 949                print $_;
 950                check_repack() if (/^r\d+ = $sha1/);
 951        }
 952        close $fh or croak $?;
 953}
 954
 955sub rec_fetch {
 956        my ($pfx, $p, @args) = @_;
 957        my @dir;
 958        foreach (sort <$p/*>) {
 959                if (-r "$_/info/url") {
 960                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
 961                        my $id = $pfx . basename $_;
 962                        next if $id eq 'trunk';
 963                        fetch_child_id($id, @args);
 964                } elsif (-d $_) {
 965                        push @dir, $_;
 966                }
 967        }
 968        foreach (@dir) {
 969                my $x = $_;
 970                $x =~ s!^\Q$GIT_DIR\E/svn/!!;
 971                rec_fetch($x, $_);
 972        }
 973}
 974
 975sub complete_url_ls_init {
 976        my ($url, $var, $switch, $pfx) = @_;
 977        unless ($var) {
 978                print STDERR "W: $switch not specified\n";
 979                return;
 980        }
 981        $var =~ s#/+$##;
 982        if ($var !~ m#^[a-z\+]+://#) {
 983                $var = '/' . $var if ($var !~ m#^/#);
 984                unless ($url) {
 985                        print STDERR "E: '$var' is not a complete URL ",
 986                                "and a separate URL is not specified\n";
 987                        exit 1;
 988                }
 989                $var = $url . $var;
 990        }
 991        chomp(my @ls = $_use_lib ? libsvn_ls_fullurl($var)
 992                                : safe_qx(qw/svn ls --non-interactive/, $var));
 993        my $old = $GIT_SVN;
 994        defined(my $pid = fork) or croak $!;
 995        if (!$pid) {
 996                foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
 997                        $u =~ s#/+$##;
 998                        if ($u !~ m!\Q$var\E/(.+)$!) {
 999                                print STDERR "W: Unrecognized URL: $u\n";
1000                                die "This should never happen\n";
1001                        }
1002                        # don't try to init already existing refs
1003                        my $id = $pfx.$1;
1004                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
1005                        init_vars();
1006                        unless (-d $GIT_SVN_DIR) {
1007                                print "init $u => $id\n";
1008                                init($u);
1009                        }
1010                }
1011                exit 0;
1012        }
1013        waitpid $pid, 0;
1014        croak $? if $?;
1015        my ($n) = ($switch =~ /^--(\w+)/);
1016        sys('git-repo-config', "svn.$n", $var);
1017}
1018
1019sub common_prefix {
1020        my $paths = shift;
1021        my %common;
1022        foreach (@$paths) {
1023                my @tmp = split m#/#, $_;
1024                my $p = '';
1025                while (my $x = shift @tmp) {
1026                        $p .= "/$x";
1027                        $common{$p} ||= 0;
1028                        $common{$p}++;
1029                }
1030        }
1031        foreach (sort {length $b <=> length $a} keys %common) {
1032                if ($common{$_} == @$paths) {
1033                        return $_;
1034                }
1035        }
1036        return '';
1037}
1038
1039# grafts set here are 'stronger' in that they're based on actual tree
1040# matches, and won't be deleted from merge-base checking in write_grafts()
1041sub graft_tree_joins {
1042        my $grafts = shift;
1043        map_tree_joins() if (@_branch_from && !%tree_map);
1044        return unless %tree_map;
1045
1046        git_svn_each(sub {
1047                my $i = shift;
1048                defined(my $pid = open my $fh, '-|') or croak $!;
1049                if (!$pid) {
1050                        exec qw/git-rev-list --pretty=raw/,
1051                                        "refs/remotes/$i" or croak $!;
1052                }
1053                while (<$fh>) {
1054                        next unless /^commit ($sha1)$/o;
1055                        my $c = $1;
1056                        my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
1057                        next unless $tree_map{$t};
1058
1059                        my $l;
1060                        do {
1061                                $l = readline $fh;
1062                        } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
1063
1064                        my ($s, $tz) = ($1, $2);
1065                        if ($tz =~ s/^\+//) {
1066                                $s += tz_to_s_offset($tz);
1067                        } elsif ($tz =~ s/^\-//) {
1068                                $s -= tz_to_s_offset($tz);
1069                        }
1070
1071                        my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
1072
1073                        foreach my $p (@{$tree_map{$t}}) {
1074                                next if $p eq $c;
1075                                my $mb = eval {
1076                                        safe_qx('git-merge-base', $c, $p)
1077                                };
1078                                next unless ($@ || $?);
1079                                if (defined $r_a) {
1080                                        # see if SVN says it's a relative
1081                                        my ($url_b, $r_b, $uuid_b) =
1082                                                        cmt_metadata($p);
1083                                        next if (defined $url_b &&
1084                                                        defined $url_a &&
1085                                                        ($url_a eq $url_b) &&
1086                                                        ($uuid_a eq $uuid_b));
1087                                        if ($uuid_a eq $uuid_b) {
1088                                                if ($r_b < $r_a) {
1089                                                        $grafts->{$c}->{$p} = 2;
1090                                                        next;
1091                                                } elsif ($r_b > $r_a) {
1092                                                        $grafts->{$p}->{$c} = 2;
1093                                                        next;
1094                                                }
1095                                        }
1096                                }
1097                                my $ct = get_commit_time($p);
1098                                if ($ct < $s) {
1099                                        $grafts->{$c}->{$p} = 2;
1100                                } elsif ($ct > $s) {
1101                                        $grafts->{$p}->{$c} = 2;
1102                                }
1103                                # what should we do when $ct == $s ?
1104                        }
1105                }
1106                close $fh or croak $?;
1107        });
1108}
1109
1110# this isn't funky-filename safe, but good enough for now...
1111sub graft_file_copy_cmd {
1112        my ($grafts, $l_map, $u) = @_;
1113        my $paths = $l_map->{$u};
1114        my $pfx = common_prefix([keys %$paths]);
1115        $SVN_URL ||= $u.$pfx;
1116        my $pid = open my $fh, '-|';
1117        defined $pid or croak $!;
1118        unless ($pid) {
1119                my @exec = qw/svn log -v/;
1120                push @exec, "-r$_revision" if defined $_revision;
1121                exec @exec, $u.$pfx or croak $!;
1122        }
1123        my ($r, $mp) = (undef, undef);
1124        while (<$fh>) {
1125                chomp;
1126                if (/^\-{72}$/) {
1127                        $mp = $r = undef;
1128                } elsif (/^r(\d+) \| /) {
1129                        $r = $1 unless defined $r;
1130                } elsif (/^Changed paths:/) {
1131                        $mp = 1;
1132                } elsif ($mp && m#^   [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) {
1133                        my ($p1, $p0, $r0) = ($1, $2, $3);
1134                        my $c = find_graft_path_commit($paths, $p1, $r);
1135                        next unless $c;
1136                        find_graft_path_parents($grafts, $paths, $c, $p0, $r0);
1137                }
1138        }
1139}
1140
1141sub graft_file_copy_lib {
1142        my ($grafts, $l_map, $u) = @_;
1143        my $tree_paths = $l_map->{$u};
1144        my $pfx = common_prefix([keys %$tree_paths]);
1145        my ($repo, $path) = repo_path_split($u.$pfx);
1146        $SVN_LOG ||= libsvn_connect($repo);
1147        $SVN ||= libsvn_connect($repo);
1148
1149        my ($base, $head) = libsvn_parse_revision();
1150        my $inc = 1000;
1151        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
1152        my $eh = $SVN::Error::handler;
1153        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
1154        while (1) {
1155                my $pool = SVN::Pool->new;
1156                libsvn_get_log($SVN_LOG, "/$path", $min, $max, 0, 1, 1,
1157                        sub {
1158                                libsvn_graft_file_copies($grafts, $tree_paths,
1159                                                        $path, @_);
1160                        }, $pool);
1161                $pool->clear;
1162                last if ($max >= $head);
1163                $min = $max + 1;
1164                $max += $inc;
1165                $max = $head if ($max > $head);
1166        }
1167        $SVN::Error::handler = $eh;
1168}
1169
1170sub process_merge_msg_matches {
1171        my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
1172        my (@strong, @weak);
1173        foreach (@matches) {
1174                # merging with ourselves is not interesting
1175                next if $_ eq $p;
1176                if ($l_map->{$u}->{$_}) {
1177                        push @strong, $_;
1178                } else {
1179                        push @weak, $_;
1180                }
1181        }
1182        foreach my $w (@weak) {
1183                last if @strong;
1184                # no exact match, use branch name as regexp.
1185                my $re = qr/\Q$w\E/i;
1186                foreach (keys %{$l_map->{$u}}) {
1187                        if (/$re/) {
1188                                push @strong, $l_map->{$u}->{$_};
1189                                last;
1190                        }
1191                }
1192                last if @strong;
1193                $w = basename($w);
1194                $re = qr/\Q$w\E/i;
1195                foreach (keys %{$l_map->{$u}}) {
1196                        if (/$re/) {
1197                                push @strong, $l_map->{$u}->{$_};
1198                                last;
1199                        }
1200                }
1201        }
1202        my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
1203                                        \s(?:[a-f\d\-]+)$/xsm);
1204        unless (defined $rev) {
1205                ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
1206                                        \@(?:[a-f\d\-]+)/xsm);
1207                return unless defined $rev;
1208        }
1209        foreach my $m (@strong) {
1210                my ($r0, $s0) = find_rev_before($rev, $m, 1);
1211                $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
1212        }
1213}
1214
1215sub graft_merge_msg {
1216        my ($grafts, $l_map, $u, $p, @re) = @_;
1217
1218        my $x = $l_map->{$u}->{$p};
1219        my $rl = rev_list_raw($x);
1220        while (my $c = next_rev_list_entry($rl)) {
1221                foreach my $re (@re) {
1222                        my (@br) = ($c->{m} =~ /$re/g);
1223                        next unless @br;
1224                        process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
1225                }
1226        }
1227}
1228
1229sub read_uuid {
1230        return if $SVN_UUID;
1231        if ($_use_lib) {
1232                my $pool = SVN::Pool->new;
1233                $SVN_UUID = $SVN->get_uuid($pool);
1234                $pool->clear;
1235        } else {
1236                my $info = shift || svn_info('.');
1237                $SVN_UUID = $info->{'Repository UUID'} or
1238                                        croak "Repository UUID unreadable\n";
1239        }
1240}
1241
1242sub quiet_run {
1243        my $pid = fork;
1244        defined $pid or croak $!;
1245        if (!$pid) {
1246                open my $null, '>', '/dev/null' or croak $!;
1247                open STDERR, '>&', $null or croak $!;
1248                open STDOUT, '>&', $null or croak $!;
1249                exec @_ or croak $!;
1250        }
1251        waitpid $pid, 0;
1252        return $?;
1253}
1254
1255sub repo_path_split {
1256        my $full_url = shift;
1257        $full_url =~ s#/+$##;
1258
1259        foreach (@repo_path_split_cache) {
1260                if ($full_url =~ s#$_##) {
1261                        my $u = $1;
1262                        $full_url =~ s#^/+##;
1263                        return ($u, $full_url);
1264                }
1265        }
1266
1267        if ($_use_lib) {
1268                my $tmp = libsvn_connect($full_url);
1269                my $url = $tmp->get_repos_root;
1270                $full_url =~ s#^\Q$url\E/*##;
1271                push @repo_path_split_cache, qr/^(\Q$url\E)/;
1272                return ($url, $full_url);
1273        } else {
1274                my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i);
1275                $path =~ s#^/+##;
1276                my @paths = split(m#/+#, $path);
1277                while (quiet_run(qw/svn ls --non-interactive/, $url)) {
1278                        my $n = shift @paths || last;
1279                        $url .= "/$n";
1280                }
1281                push @repo_path_split_cache, qr/^(\Q$url\E)/;
1282                $path = join('/',@paths);
1283                return ($url, $path);
1284        }
1285}
1286
1287sub setup_git_svn {
1288        defined $SVN_URL or croak "SVN repository location required\n";
1289        unless (-d $GIT_DIR) {
1290                croak "GIT_DIR=$GIT_DIR does not exist!\n";
1291        }
1292        mkpath([$GIT_SVN_DIR]);
1293        mkpath(["$GIT_SVN_DIR/info"]);
1294        open my $fh, '>>',$REVDB or croak $!;
1295        close $fh;
1296        s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1297
1298}
1299
1300sub assert_svn_wc_clean {
1301        return if $_use_lib;
1302        my ($svn_rev) = @_;
1303        croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
1304        my $lcr = svn_info('.')->{'Last Changed Rev'};
1305        if ($svn_rev != $lcr) {
1306                print STDERR "Checking for copy-tree ... ";
1307                my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
1308                                                "-r$lcr:$svn_rev")));
1309                if (@diff) {
1310                        croak "Nope!  Expected r$svn_rev, got r$lcr\n";
1311                } else {
1312                        print STDERR "OK!\n";
1313                }
1314        }
1315        my @status = grep(!/^Performing status on external/,(`svn status`));
1316        @status = grep(!/^\s*$/,@status);
1317        @status = grep(!/^X/,@status) if $_no_ignore_ext;
1318        if (scalar @status) {
1319                print STDERR "Tree ($SVN_WC) is not clean:\n";
1320                print STDERR $_ foreach @status;
1321                croak;
1322        }
1323}
1324
1325sub get_tree_from_treeish {
1326        my ($treeish) = @_;
1327        croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1328        chomp(my $type = `git-cat-file -t $treeish`);
1329        my $expected;
1330        while ($type eq 'tag') {
1331                chomp(($treeish, $type) = `git-cat-file tag $treeish`);
1332        }
1333        if ($type eq 'commit') {
1334                $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
1335                ($expected) = ($expected =~ /^tree ($sha1)$/);
1336                die "Unable to get tree from $treeish\n" unless $expected;
1337        } elsif ($type eq 'tree') {
1338                $expected = $treeish;
1339        } else {
1340                die "$treeish is a $type, expected tree, tag or commit\n";
1341        }
1342        return $expected;
1343}
1344
1345sub assert_tree {
1346        return if $_use_lib;
1347        my ($treeish) = @_;
1348        my $expected = get_tree_from_treeish($treeish);
1349
1350        my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
1351        if (-e $tmpindex) {
1352                unlink $tmpindex or croak $!;
1353        }
1354        my $old_index = set_index($tmpindex);
1355        index_changes(1);
1356        chomp(my $tree = `git-write-tree`);
1357        restore_index($old_index);
1358        if ($tree ne $expected) {
1359                croak "Tree mismatch, Got: $tree, Expected: $expected\n";
1360        }
1361        unlink $tmpindex;
1362}
1363
1364sub parse_diff_tree {
1365        my $diff_fh = shift;
1366        local $/ = "\0";
1367        my $state = 'meta';
1368        my @mods;
1369        while (<$diff_fh>) {
1370                chomp $_; # this gets rid of the trailing "\0"
1371                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1372                                        $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1373                        push @mods, {   mode_a => $1, mode_b => $2,
1374                                        sha1_b => $3, chg => $4 };
1375                        if ($4 =~ /^(?:C|R)$/) {
1376                                $state = 'file_a';
1377                        } else {
1378                                $state = 'file_b';
1379                        }
1380                } elsif ($state eq 'file_a') {
1381                        my $x = $mods[$#mods] or croak "Empty array\n";
1382                        if ($x->{chg} !~ /^(?:C|R)$/) {
1383                                croak "Error parsing $_, $x->{chg}\n";
1384                        }
1385                        $x->{file_a} = $_;
1386                        $state = 'file_b';
1387                } elsif ($state eq 'file_b') {
1388                        my $x = $mods[$#mods] or croak "Empty array\n";
1389                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1390                                croak "Error parsing $_, $x->{chg}\n";
1391                        }
1392                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1393                                croak "Error parsing $_, $x->{chg}\n";
1394                        }
1395                        $x->{file_b} = $_;
1396                        $state = 'meta';
1397                } else {
1398                        croak "Error parsing $_\n";
1399                }
1400        }
1401        close $diff_fh or croak $?;
1402
1403        return \@mods;
1404}
1405
1406sub svn_check_prop_executable {
1407        my $m = shift;
1408        return if -l $m->{file_b};
1409        if ($m->{mode_b} =~ /755$/) {
1410                chmod((0755 &~ umask),$m->{file_b}) or croak $!;
1411                if ($m->{mode_a} !~ /755$/) {
1412                        sys(qw(svn propset svn:executable 1), $m->{file_b});
1413                }
1414                -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
1415        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
1416                sys(qw(svn propdel svn:executable), $m->{file_b});
1417                chmod((0644 &~ umask),$m->{file_b}) or croak $!;
1418                -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
1419        }
1420}
1421
1422sub svn_ensure_parent_path {
1423        my $dir_b = dirname(shift);
1424        svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
1425        mkpath([$dir_b]) unless (-d $dir_b);
1426        sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
1427}
1428
1429sub precommit_check {
1430        my $mods = shift;
1431        my (%rm_file, %rmdir_check, %added_check);
1432
1433        my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
1434        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1435                if ($m->{chg} eq 'R') {
1436                        if (-d $m->{file_b}) {
1437                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1438                        }
1439                        # dir/$file => dir/file/$file
1440                        my $dirname = dirname($m->{file_b});
1441                        while ($dirname ne File::Spec->curdir) {
1442                                if ($dirname ne $m->{file_a}) {
1443                                        $dirname = dirname($dirname);
1444                                        next;
1445                                }
1446                                err_file_to_dir("$m->{file_a} => $m->{file_b}");
1447                        }
1448                        # baz/zzz => baz (baz is a file)
1449                        $dirname = dirname($m->{file_a});
1450                        while ($dirname ne File::Spec->curdir) {
1451                                if ($dirname ne $m->{file_b}) {
1452                                        $dirname = dirname($dirname);
1453                                        next;
1454                                }
1455                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1456                        }
1457                }
1458                if ($m->{chg} =~ /^(D|R)$/) {
1459                        my $t = $1 eq 'D' ? 'file_b' : 'file_a';
1460                        $rm_file{ $m->{$t} } = 1;
1461                        my $dirname = dirname( $m->{$t} );
1462                        my $basename = basename( $m->{$t} );
1463                        $rmdir_check{$dirname}->{$basename} = 1;
1464                } elsif ($m->{chg} =~ /^(?:A|C)$/) {
1465                        if (-d $m->{file_b}) {
1466                                err_dir_to_file($m->{file_b});
1467                        }
1468                        my $dirname = dirname( $m->{file_b} );
1469                        my $basename = basename( $m->{file_b} );
1470                        $added_check{$dirname}->{$basename} = 1;
1471                        while ($dirname ne File::Spec->curdir) {
1472                                if ($rm_file{$dirname}) {
1473                                        err_file_to_dir($m->{file_b});
1474                                }
1475                                $dirname = dirname $dirname;
1476                        }
1477                }
1478        }
1479        return (\%rmdir_check, \%added_check);
1480
1481        sub err_dir_to_file {
1482                my $file = shift;
1483                print STDERR "Node change from directory to file ",
1484                                "is not supported by Subversion: ",$file,"\n";
1485                exit 1;
1486        }
1487        sub err_file_to_dir {
1488                my $file = shift;
1489                print STDERR "Node change from file to directory ",
1490                                "is not supported by Subversion: ",$file,"\n";
1491                exit 1;
1492        }
1493}
1494
1495
1496sub get_diff {
1497        my ($from, $treeish) = @_;
1498        assert_tree($from);
1499        print "diff-tree $from $treeish\n";
1500        my $pid = open my $diff_fh, '-|';
1501        defined $pid or croak $!;
1502        if ($pid == 0) {
1503                my @diff_tree = qw(git-diff-tree -z -r);
1504                if ($_cp_similarity) {
1505                        push @diff_tree, "-C$_cp_similarity";
1506                } else {
1507                        push @diff_tree, '-C';
1508                }
1509                push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1510                push @diff_tree, "-l$_l" if defined $_l;
1511                exec(@diff_tree, $from, $treeish) or croak $!;
1512        }
1513        return parse_diff_tree($diff_fh);
1514}
1515
1516sub svn_checkout_tree {
1517        my ($from, $treeish) = @_;
1518        my $mods = get_diff($from->{commit}, $treeish);
1519        return $mods unless (scalar @$mods);
1520        my ($rm, $add) = precommit_check($mods);
1521
1522        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1523        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1524                if ($m->{chg} eq 'C') {
1525                        svn_ensure_parent_path( $m->{file_b} );
1526                        sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
1527                        apply_mod_line_blob($m);
1528                        svn_check_prop_executable($m);
1529                } elsif ($m->{chg} eq 'D') {
1530                        sys(qw(svn rm --force), $m->{file_b});
1531                } elsif ($m->{chg} eq 'R') {
1532                        svn_ensure_parent_path( $m->{file_b} );
1533                        sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
1534                        apply_mod_line_blob($m);
1535                        svn_check_prop_executable($m);
1536                } elsif ($m->{chg} eq 'M') {
1537                        apply_mod_line_blob($m);
1538                        svn_check_prop_executable($m);
1539                } elsif ($m->{chg} eq 'T') {
1540                        svn_check_prop_executable($m);
1541                        apply_mod_line_blob($m);
1542                        if ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
1543                                sys(qw(svn propdel svn:special), $m->{file_b});
1544                        } else {
1545                                sys(qw(svn propset svn:special *),$m->{file_b});
1546                        }
1547                } elsif ($m->{chg} eq 'A') {
1548                        svn_ensure_parent_path( $m->{file_b} );
1549                        apply_mod_line_blob($m);
1550                        sys(qw(svn add), $m->{file_b});
1551                        svn_check_prop_executable($m);
1552                } else {
1553                        croak "Invalid chg: $m->{chg}\n";
1554                }
1555        }
1556
1557        assert_tree($treeish);
1558        if ($_rmdir) { # remove empty directories
1559                handle_rmdir($rm, $add);
1560        }
1561        assert_tree($treeish);
1562        return $mods;
1563}
1564
1565sub libsvn_checkout_tree {
1566        my ($from, $treeish, $ed) = @_;
1567        my $mods = get_diff($from, $treeish);
1568        return $mods unless (scalar @$mods);
1569        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1570        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1571                my $f = $m->{chg};
1572                if (defined $o{$f}) {
1573                        $ed->$f($m, $_q);
1574                } else {
1575                        croak "Invalid change type: $f\n";
1576                }
1577        }
1578        $ed->rmdirs($_q) if $_rmdir;
1579        return $mods;
1580}
1581
1582# svn ls doesn't work with respect to the current working tree, but what's
1583# in the repository.  There's not even an option for it... *sigh*
1584# (added files don't show up and removed files remain in the ls listing)
1585sub svn_ls_current {
1586        my ($dir, $rm, $add) = @_;
1587        chomp(my @ls = safe_qx('svn','ls',$dir));
1588        my @ret = ();
1589        foreach (@ls) {
1590                s#/$##; # trailing slashes are evil
1591                push @ret, $_ unless $rm->{$dir}->{$_};
1592        }
1593        if (exists $add->{$dir}) {
1594                push @ret, keys %{$add->{$dir}};
1595        }
1596        return \@ret;
1597}
1598
1599sub handle_rmdir {
1600        my ($rm, $add) = @_;
1601
1602        foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
1603                my $ls = svn_ls_current($dir, $rm, $add);
1604                next if (scalar @$ls);
1605                sys(qw(svn rm --force),$dir);
1606
1607                my $dn = dirname $dir;
1608                $rm->{ $dn }->{ basename $dir } = 1;
1609                $ls = svn_ls_current($dn, $rm, $add);
1610                while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
1611                        sys(qw(svn rm --force),$dn);
1612                        $dir = basename $dn;
1613                        $dn = dirname $dn;
1614                        $rm->{ $dn }->{ $dir } = 1;
1615                        $ls = svn_ls_current($dn, $rm, $add);
1616                }
1617        }
1618}
1619
1620sub get_commit_message {
1621        my ($commit, $commit_msg) = (@_);
1622        my %log_msg = ( msg => '' );
1623        open my $msg, '>', $commit_msg or croak $!;
1624
1625        chomp(my $type = `git-cat-file -t $commit`);
1626        if ($type eq 'commit' || $type eq 'tag') {
1627                my $pid = open my $msg_fh, '-|';
1628                defined $pid or croak $!;
1629
1630                if ($pid == 0) {
1631                        exec('git-cat-file', $type, $commit) or croak $!;
1632                }
1633                my $in_msg = 0;
1634                while (<$msg_fh>) {
1635                        if (!$in_msg) {
1636                                $in_msg = 1 if (/^\s*$/);
1637                        } elsif (/^git-svn-id: /) {
1638                                # skip this, we regenerate the correct one
1639                                # on re-fetch anyways
1640                        } else {
1641                                print $msg $_ or croak $!;
1642                        }
1643                }
1644                close $msg_fh or croak $?;
1645        }
1646        close $msg or croak $!;
1647
1648        if ($_edit || ($type eq 'tree')) {
1649                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1650                system($editor, $commit_msg);
1651        }
1652
1653        # file_to_s removes all trailing newlines, so just use chomp() here:
1654        open $msg, '<', $commit_msg or croak $!;
1655        { local $/; chomp($log_msg{msg} = <$msg>); }
1656        close $msg or croak $!;
1657
1658        return \%log_msg;
1659}
1660
1661sub set_svn_commit_env {
1662        if (defined $LC_ALL) {
1663                $ENV{LC_ALL} = $LC_ALL;
1664        } else {
1665                delete $ENV{LC_ALL};
1666        }
1667}
1668
1669sub svn_commit_tree {
1670        my ($last, $commit) = @_;
1671        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
1672        my $log_msg = get_commit_message($commit, $commit_msg);
1673        my ($oneline) = ($log_msg->{msg} =~ /([^\n\r]+)/);
1674        print "Committing $commit: $oneline\n";
1675
1676        set_svn_commit_env();
1677        my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
1678        $ENV{LC_ALL} = 'C';
1679        unlink $commit_msg;
1680        my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/);
1681        if (!defined $committed) {
1682                my $out = join("\n",@ci_output);
1683                print STDERR "W: Trouble parsing \`svn commit' output:\n\n",
1684                                $out, "\n\nAssuming English locale...";
1685                ($committed) = ($out =~ /^Committed revision \d+\./sm);
1686                defined $committed or die " FAILED!\n",
1687                        "Commit output failed to parse committed revision!\n",
1688                print STDERR " OK\n";
1689        }
1690
1691        my @svn_up = qw(svn up);
1692        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
1693        if ($_optimize_commits && ($committed == ($last->{revision} + 1))) {
1694                push @svn_up, "-r$committed";
1695                sys(@svn_up);
1696                my $info = svn_info('.');
1697                my $date = $info->{'Last Changed Date'} or die "Missing date\n";
1698                if ($info->{'Last Changed Rev'} != $committed) {
1699                        croak "$info->{'Last Changed Rev'} != $committed\n"
1700                }
1701                my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1702                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1703                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1704                                         or croak "Failed to parse date: $date\n";
1705                $log_msg->{date} = "$tz $Y-$m-$d $H:$M:$S";
1706                $log_msg->{author} = $info->{'Last Changed Author'};
1707                $log_msg->{revision} = $committed;
1708                $log_msg->{msg} .= "\n";
1709                $log_msg->{parents} = [ $last->{commit} ];
1710                $log_msg->{commit} = git_commit($log_msg, $commit);
1711                return $log_msg;
1712        }
1713        # resync immediately
1714        push @svn_up, "-r$last->{revision}";
1715        sys(@svn_up);
1716        return fetch("$committed=$commit");
1717}
1718
1719sub rev_list_raw {
1720        my (@args) = @_;
1721        my $pid = open my $fh, '-|';
1722        defined $pid or croak $!;
1723        if (!$pid) {
1724                exec(qw/git-rev-list --pretty=raw/, @args) or croak $!;
1725        }
1726        return { fh => $fh, t => { } };
1727}
1728
1729sub next_rev_list_entry {
1730        my $rl = shift;
1731        my $fh = $rl->{fh};
1732        my $x = $rl->{t};
1733        while (<$fh>) {
1734                if (/^commit ($sha1)$/o) {
1735                        if ($x->{c}) {
1736                                $rl->{t} = { c => $1 };
1737                                return $x;
1738                        } else {
1739                                $x->{c} = $1;
1740                        }
1741                } elsif (/^parent ($sha1)$/o) {
1742                        $x->{p}->{$1} = 1;
1743                } elsif (s/^    //) {
1744                        $x->{m} ||= '';
1745                        $x->{m} .= $_;
1746                }
1747        }
1748        return ($x != $rl->{t}) ? $x : undef;
1749}
1750
1751# read the entire log into a temporary file (which is removed ASAP)
1752# and store the file handle + parser state
1753sub svn_log_raw {
1754        my (@log_args) = @_;
1755        my $log_fh = IO::File->new_tmpfile or croak $!;
1756        my $pid = fork;
1757        defined $pid or croak $!;
1758        if (!$pid) {
1759                open STDOUT, '>&', $log_fh or croak $!;
1760                exec (qw(svn log), @log_args) or croak $!
1761        }
1762        waitpid $pid, 0;
1763        croak $? if $?;
1764        seek $log_fh, 0, 0 or croak $!;
1765        return { state => 'sep', fh => $log_fh };
1766}
1767
1768sub next_log_entry {
1769        my $log = shift; # retval of svn_log_raw()
1770        my $ret = undef;
1771        my $fh = $log->{fh};
1772
1773        while (<$fh>) {
1774                chomp;
1775                if (/^\-{72}$/) {
1776                        if ($log->{state} eq 'msg') {
1777                                if ($ret->{lines}) {
1778                                        $ret->{msg} .= $_."\n";
1779                                        unless(--$ret->{lines}) {
1780                                                $log->{state} = 'sep';
1781                                        }
1782                                } else {
1783                                        croak "Log parse error at: $_\n",
1784                                                $ret->{revision},
1785                                                "\n";
1786                                }
1787                                next;
1788                        }
1789                        if ($log->{state} ne 'sep') {
1790                                croak "Log parse error at: $_\n",
1791                                        "state: $log->{state}\n",
1792                                        $ret->{revision},
1793                                        "\n";
1794                        }
1795                        $log->{state} = 'rev';
1796
1797                        # if we have an empty log message, put something there:
1798                        if ($ret) {
1799                                $ret->{msg} ||= "\n";
1800                                delete $ret->{lines};
1801                                return $ret;
1802                        }
1803                        next;
1804                }
1805                if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
1806                        my $rev = $1;
1807                        my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
1808                        ($lines) = ($lines =~ /(\d+)/);
1809                        $date = '1970-01-01 00:00:00 +0000'
1810                                if ($_ignore_nodate && $date eq '(no date)');
1811                        my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1812                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1813                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1814                                         or croak "Failed to parse date: $date\n";
1815                        $ret = {        revision => $rev,
1816                                        date => "$tz $Y-$m-$d $H:$M:$S",
1817                                        author => $author,
1818                                        lines => $lines,
1819                                        msg => '' };
1820                        if (defined $_authors && ! defined $users{$author}) {
1821                                die "Author: $author not defined in ",
1822                                                "$_authors file\n";
1823                        }
1824                        $log->{state} = 'msg_start';
1825                        next;
1826                }
1827                # skip the first blank line of the message:
1828                if ($log->{state} eq 'msg_start' && /^$/) {
1829                        $log->{state} = 'msg';
1830                } elsif ($log->{state} eq 'msg') {
1831                        if ($ret->{lines}) {
1832                                $ret->{msg} .= $_."\n";
1833                                unless (--$ret->{lines}) {
1834                                        $log->{state} = 'sep';
1835                                }
1836                        } else {
1837                                croak "Log parse error at: $_\n",
1838                                        $ret->{revision},"\n";
1839                        }
1840                }
1841        }
1842        return $ret;
1843}
1844
1845sub svn_info {
1846        my $url = shift || $SVN_URL;
1847
1848        my $pid = open my $info_fh, '-|';
1849        defined $pid or croak $!;
1850
1851        if ($pid == 0) {
1852                exec(qw(svn info),$url) or croak $!;
1853        }
1854
1855        my $ret = {};
1856        # only single-lines seem to exist in svn info output
1857        while (<$info_fh>) {
1858                chomp $_;
1859                if (m#^([^:]+)\s*:\s*(\S.*)$#) {
1860                        $ret->{$1} = $2;
1861                        push @{$ret->{-order}}, $1;
1862                }
1863        }
1864        close $info_fh or croak $?;
1865        return $ret;
1866}
1867
1868sub sys { system(@_) == 0 or croak $? }
1869
1870sub do_update_index {
1871        my ($z_cmd, $cmd, $no_text_base) = @_;
1872
1873        my $z = open my $p, '-|';
1874        defined $z or croak $!;
1875        unless ($z) { exec @$z_cmd or croak $! }
1876
1877        my $pid = open my $ui, '|-';
1878        defined $pid or croak $!;
1879        unless ($pid) {
1880                exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!;
1881        }
1882        local $/ = "\0";
1883        while (my $x = <$p>) {
1884                chomp $x;
1885                if (!$no_text_base && lstat $x && ! -l _ &&
1886                                svn_propget_base('svn:keywords', $x)) {
1887                        my $mode = -x _ ? 0755 : 0644;
1888                        my ($v,$d,$f) = File::Spec->splitpath($x);
1889                        my $tb = File::Spec->catfile($d, '.svn', 'tmp',
1890                                                'text-base',"$f.svn-base");
1891                        $tb =~ s#^/##;
1892                        unless (-f $tb) {
1893                                $tb = File::Spec->catfile($d, '.svn',
1894                                                'text-base',"$f.svn-base");
1895                                $tb =~ s#^/##;
1896                        }
1897                        my @s = stat($x);
1898                        unlink $x or croak $!;
1899                        copy($tb, $x);
1900                        chmod(($mode &~ umask), $x) or croak $!;
1901                        utime $s[8], $s[9], $x;
1902                }
1903                print $ui $x,"\0";
1904        }
1905        close $ui or croak $?;
1906}
1907
1908sub index_changes {
1909        return if $_use_lib;
1910
1911        if (!-f "$GIT_SVN_DIR/info/exclude") {
1912                open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!;
1913                print $fd '.svn',"\n";
1914                close $fd or croak $!;
1915        }
1916        my $no_text_base = shift;
1917        do_update_index([qw/git-diff-files --name-only -z/],
1918                        'remove',
1919                        $no_text_base);
1920        do_update_index([qw/git-ls-files -z --others/,
1921                                "--exclude-from=$GIT_SVN_DIR/info/exclude"],
1922                        'add',
1923                        $no_text_base);
1924}
1925
1926sub s_to_file {
1927        my ($str, $file, $mode) = @_;
1928        open my $fd,'>',$file or croak $!;
1929        print $fd $str,"\n" or croak $!;
1930        close $fd or croak $!;
1931        chmod ($mode &~ umask, $file) if (defined $mode);
1932}
1933
1934sub file_to_s {
1935        my $file = shift;
1936        open my $fd,'<',$file or croak "$!: file: $file\n";
1937        local $/;
1938        my $ret = <$fd>;
1939        close $fd or croak $!;
1940        $ret =~ s/\s*$//s;
1941        return $ret;
1942}
1943
1944sub assert_revision_unknown {
1945        my $r = shift;
1946        if (my $c = revdb_get($REVDB, $r)) {
1947                croak "$r = $c already exists! Why are we refetching it?";
1948        }
1949}
1950
1951sub trees_eq {
1952        my ($x, $y) = @_;
1953        my @x = safe_qx('git-cat-file','commit',$x);
1954        my @y = safe_qx('git-cat-file','commit',$y);
1955        if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
1956                                || $y[0] !~ /^tree $sha1\n$/) {
1957                print STDERR "Trees not equal: $y[0] != $x[0]\n";
1958                return 0
1959        }
1960        return 1;
1961}
1962
1963sub git_commit {
1964        my ($log_msg, @parents) = @_;
1965        assert_revision_unknown($log_msg->{revision});
1966        map_tree_joins() if (@_branch_from && !%tree_map);
1967
1968        my (@tmp_parents, @exec_parents, %seen_parent);
1969        if (my $lparents = $log_msg->{parents}) {
1970                @tmp_parents = @$lparents
1971        }
1972        # commit parents can be conditionally bound to a particular
1973        # svn revision via: "svn_revno=commit_sha1", filter them out here:
1974        foreach my $p (@parents) {
1975                next unless defined $p;
1976                if ($p =~ /^(\d+)=($sha1_short)$/o) {
1977                        if ($1 == $log_msg->{revision}) {
1978                                push @tmp_parents, $2;
1979                        }
1980                } else {
1981                        push @tmp_parents, $p if $p =~ /$sha1_short/o;
1982                }
1983        }
1984        my $tree = $log_msg->{tree};
1985        if (!defined $tree) {
1986                my $index = set_index($GIT_SVN_INDEX);
1987                index_changes();
1988                chomp($tree = `git-write-tree`);
1989                croak $? if $?;
1990                restore_index($index);
1991        }
1992
1993        # just in case we clobber the existing ref, we still want that ref
1994        # as our parent:
1995        if (my $cur = eval { file_to_s("$GIT_DIR/refs/remotes/$GIT_SVN") }) {
1996                push @tmp_parents, $cur;
1997        }
1998
1999        if (exists $tree_map{$tree}) {
2000                foreach my $p (@{$tree_map{$tree}}) {
2001                        my $skip;
2002                        foreach (@tmp_parents) {
2003                                # see if a common parent is found
2004                                my $mb = eval {
2005                                        safe_qx('git-merge-base', $_, $p)
2006                                };
2007                                next if ($@ || $?);
2008                                $skip = 1;
2009                                last;
2010                        }
2011                        next if $skip;
2012                        my ($url_p, $r_p, $uuid_p) = cmt_metadata($p);
2013                        next if (($SVN_UUID eq $uuid_p) &&
2014                                                ($log_msg->{revision} > $r_p));
2015                        next if (defined $url_p && defined $SVN_URL &&
2016                                                ($SVN_UUID eq $uuid_p) &&
2017                                                ($url_p eq $SVN_URL));
2018                        push @tmp_parents, $p;
2019                }
2020        }
2021        foreach (@tmp_parents) {
2022                next if $seen_parent{$_};
2023                $seen_parent{$_} = 1;
2024                push @exec_parents, $_;
2025                # MAXPARENT is defined to 16 in commit-tree.c:
2026                last if @exec_parents > 16;
2027        }
2028
2029        set_commit_env($log_msg);
2030        my @exec = ('git-commit-tree', $tree);
2031        push @exec, '-p', $_  foreach @exec_parents;
2032        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2033                                                                or croak $!;
2034        print $msg_fh $log_msg->{msg} or croak $!;
2035        unless ($_no_metadata) {
2036                print $msg_fh "\ngit-svn-id: $SVN_URL\@$log_msg->{revision}",
2037                                        " $SVN_UUID\n" or croak $!;
2038        }
2039        $msg_fh->flush == 0 or croak $!;
2040        close $msg_fh or croak $!;
2041        chomp(my $commit = do { local $/; <$out_fh> });
2042        close $out_fh or croak $!;
2043        waitpid $pid, 0;
2044        croak $? if $?;
2045        if ($commit !~ /^$sha1$/o) {
2046                die "Failed to commit, invalid sha1: $commit\n";
2047        }
2048        sys('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
2049        revdb_set($REVDB, $log_msg->{revision}, $commit);
2050
2051        # this output is read via pipe, do not change:
2052        print "r$log_msg->{revision} = $commit\n";
2053        check_repack();
2054        return $commit;
2055}
2056
2057sub check_repack {
2058        if ($_repack && (--$_repack_nr == 0)) {
2059                $_repack_nr = $_repack;
2060                sys("git repack $_repack_flags");
2061        }
2062}
2063
2064sub set_commit_env {
2065        my ($log_msg) = @_;
2066        my $author = $log_msg->{author};
2067        if (!defined $author || length $author == 0) {
2068                $author = '(no author)';
2069        }
2070        my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
2071                                : ($author,"$author\@$SVN_UUID");
2072        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
2073        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
2074        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
2075}
2076
2077sub apply_mod_line_blob {
2078        my $m = shift;
2079        if ($m->{mode_b} =~ /^120/) {
2080                blob_to_symlink($m->{sha1_b}, $m->{file_b});
2081        } else {
2082                blob_to_file($m->{sha1_b}, $m->{file_b});
2083        }
2084}
2085
2086sub blob_to_symlink {
2087        my ($blob, $link) = @_;
2088        defined $link or croak "\$link not defined!\n";
2089        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2090        if (-l $link || -f _) {
2091                unlink $link or croak $!;
2092        }
2093
2094        my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
2095        symlink $dest, $link or croak $!;
2096}
2097
2098sub blob_to_file {
2099        my ($blob, $file) = @_;
2100        defined $file or croak "\$file not defined!\n";
2101        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2102        if (-l $file || -f _) {
2103                unlink $file or croak $!;
2104        }
2105
2106        open my $blob_fh, '>', $file or croak "$!: $file\n";
2107        my $pid = fork;
2108        defined $pid or croak $!;
2109
2110        if ($pid == 0) {
2111                open STDOUT, '>&', $blob_fh or croak $!;
2112                exec('git-cat-file','blob',$blob) or croak $!;
2113        }
2114        waitpid $pid, 0;
2115        croak $? if $?;
2116
2117        close $blob_fh or croak $!;
2118}
2119
2120sub safe_qx {
2121        my $pid = open my $child, '-|';
2122        defined $pid or croak $!;
2123        if ($pid == 0) {
2124                exec(@_) or croak $!;
2125        }
2126        my @ret = (<$child>);
2127        close $child or croak $?;
2128        die $? if $?; # just in case close didn't error out
2129        return wantarray ? @ret : join('',@ret);
2130}
2131
2132sub svn_compat_check {
2133        if ($_follow_parent) {
2134                print STDERR 'E: --follow-parent functionality is only ',
2135                                "available when SVN libraries are used\n";
2136                exit 1;
2137        }
2138        my @co_help = safe_qx(qw(svn co -h));
2139        unless (grep /ignore-externals/,@co_help) {
2140                print STDERR "W: Installed svn version does not support ",
2141                                "--ignore-externals\n";
2142                $_no_ignore_ext = 1;
2143        }
2144        if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
2145                $_svn_co_url_revs = 1;
2146        }
2147        if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) {
2148                $_svn_pg_peg_revs = 1;
2149        }
2150
2151        # I really, really hope nobody hits this...
2152        unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
2153                print STDERR <<'';
2154W: The installed svn version does not support the --stop-on-copy flag in
2155   the log command.
2156   Lets hope the directory you're tracking is not a branch or tag
2157   and was never moved within the repository...
2158
2159                $_no_stop_copy = 1;
2160        }
2161}
2162
2163# *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
2164# (and they won't honor URL@<rev> without -r<rev>, too!)
2165sub svn_cmd_checkout {
2166        my ($url, $rev, $dir) = @_;
2167        my @cmd = ('svn','co', "-r$rev");
2168        push @cmd, '--ignore-externals' unless $_no_ignore_ext;
2169        $url .= "\@$rev" if $_svn_co_url_revs;
2170        sys(@cmd, $url, $dir);
2171}
2172
2173sub check_upgrade_needed {
2174        if (!-r $REVDB) {
2175                -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2176                open my $fh, '>>',$REVDB or croak $!;
2177                close $fh;
2178        }
2179        my $old = eval {
2180                my $pid = open my $child, '-|';
2181                defined $pid or croak $!;
2182                if ($pid == 0) {
2183                        close STDERR;
2184                        exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $!;
2185                }
2186                my @ret = (<$child>);
2187                close $child or croak $?;
2188                die $? if $?; # just in case close didn't error out
2189                return wantarray ? @ret : join('',@ret);
2190        };
2191        return unless $old;
2192        my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
2193        if ($@ || !$head) {
2194                print STDERR "Please run: $0 rebuild --upgrade\n";
2195                exit 1;
2196        }
2197}
2198
2199# fills %tree_map with a reverse mapping of trees to commits.  Useful
2200# for finding parents to commit on.
2201sub map_tree_joins {
2202        my %seen;
2203        foreach my $br (@_branch_from) {
2204                my $pid = open my $pipe, '-|';
2205                defined $pid or croak $!;
2206                if ($pid == 0) {
2207                        exec(qw(git-rev-list --topo-order --pretty=raw), $br)
2208                                                                or croak $!;
2209                }
2210                while (<$pipe>) {
2211                        if (/^commit ($sha1)$/o) {
2212                                my $commit = $1;
2213
2214                                # if we've seen a commit,
2215                                # we've seen its parents
2216                                last if $seen{$commit};
2217                                my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
2218                                unless (defined $tree) {
2219                                        die "Failed to parse commit $commit\n";
2220                                }
2221                                push @{$tree_map{$tree}}, $commit;
2222                                $seen{$commit} = 1;
2223                        }
2224                }
2225                close $pipe; # we could be breaking the pipe early
2226        }
2227}
2228
2229sub load_all_refs {
2230        if (@_branch_from) {
2231                print STDERR '--branch|-b parameters are ignored when ',
2232                        "--branch-all-refs|-B is passed\n";
2233        }
2234
2235        # don't worry about rev-list on non-commit objects/tags,
2236        # it shouldn't blow up if a ref is a blob or tree...
2237        chomp(@_branch_from = `git-rev-parse --symbolic --all`);
2238}
2239
2240# '<svn username> = real-name <email address>' mapping based on git-svnimport:
2241sub load_authors {
2242        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2243        while (<$authors>) {
2244                chomp;
2245                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
2246                my ($user, $name, $email) = ($1, $2, $3);
2247                $users{$user} = [$name, $email];
2248        }
2249        close $authors or croak $!;
2250}
2251
2252sub rload_authors {
2253        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2254        while (<$authors>) {
2255                chomp;
2256                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2257                my ($user, $name, $email) = ($1, $2, $3);
2258                $rusers{"$name <$email>"} = $user;
2259        }
2260        close $authors or croak $!;
2261}
2262
2263sub svn_propget_base {
2264        my ($p, $f) = @_;
2265        $f .= '@BASE' if $_svn_pg_peg_revs;
2266        return safe_qx(qw/svn propget/, $p, $f);
2267}
2268
2269sub git_svn_each {
2270        my $sub = shift;
2271        foreach (`git-rev-parse --symbolic --all`) {
2272                next unless s#^refs/remotes/##;
2273                chomp $_;
2274                next unless -f "$GIT_DIR/svn/$_/info/url";
2275                &$sub($_);
2276        }
2277}
2278
2279sub migrate_revdb {
2280        git_svn_each(sub {
2281                my $id = shift;
2282                defined(my $pid = fork) or croak $!;
2283                if (!$pid) {
2284                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2285                        init_vars();
2286                        exit 0 if -r $REVDB;
2287                        print "Upgrading svn => git mapping...\n";
2288                        -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2289                        open my $fh, '>>',$REVDB or croak $!;
2290                        close $fh;
2291                        rebuild();
2292                        print "Done upgrading. You may now delete the ",
2293                                "deprecated $GIT_SVN_DIR/revs directory\n";
2294                        exit 0;
2295                }
2296                waitpid $pid, 0;
2297                croak $? if $?;
2298        });
2299}
2300
2301sub migration_check {
2302        migrate_revdb() unless (-e $REVDB);
2303        return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
2304        print "Upgrading repository...\n";
2305        unless (-d "$GIT_DIR/svn") {
2306                mkdir "$GIT_DIR/svn" or croak $!;
2307        }
2308        print "Data from a previous version of git-svn exists, but\n\t",
2309                                "$GIT_SVN_DIR\n\t(required for this version ",
2310                                "($VERSION) of git-svn) does not.\n";
2311
2312        foreach my $x (`git-rev-parse --symbolic --all`) {
2313                next unless $x =~ s#^refs/remotes/##;
2314                chomp $x;
2315                next unless -f "$GIT_DIR/$x/info/url";
2316                my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
2317                next unless $u;
2318                my $dn = dirname("$GIT_DIR/svn/$x");
2319                mkpath([$dn]) unless -d $dn;
2320                rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
2321        }
2322        migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
2323        print "Done upgrading.\n";
2324}
2325
2326sub find_rev_before {
2327        my ($r, $id, $eq_ok) = @_;
2328        my $f = "$GIT_DIR/svn/$id/.rev_db";
2329        return (undef,undef) unless -r $f;
2330        --$r unless $eq_ok;
2331        while ($r > 0) {
2332                if (my $c = revdb_get($f, $r)) {
2333                        return ($r, $c);
2334                }
2335                --$r;
2336        }
2337        return (undef, undef);
2338}
2339
2340sub init_vars {
2341        $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
2342        $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
2343        $REVDB = "$GIT_SVN_DIR/.rev_db";
2344        $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
2345        $SVN_URL = undef;
2346        $SVN_WC = "$GIT_SVN_DIR/tree";
2347        %tree_map = ();
2348}
2349
2350# convert GetOpt::Long specs for use by git-repo-config
2351sub read_repo_config {
2352        return unless -d $GIT_DIR;
2353        my $opts = shift;
2354        foreach my $o (keys %$opts) {
2355                my $v = $opts->{$o};
2356                my ($key) = ($o =~ /^([a-z\-]+)/);
2357                $key =~ s/-//g;
2358                my $arg = 'git-repo-config';
2359                $arg .= ' --int' if ($o =~ /[:=]i$/);
2360                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
2361                if (ref $v eq 'ARRAY') {
2362                        chomp(my @tmp = `$arg --get-all svn.$key`);
2363                        @$v = @tmp if @tmp;
2364                } else {
2365                        chomp(my $tmp = `$arg --get svn.$key`);
2366                        if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
2367                                $$v = $tmp;
2368                        }
2369                }
2370        }
2371}
2372
2373sub set_default_vals {
2374        if (defined $_repack) {
2375                $_repack = 1000 if ($_repack <= 0);
2376                $_repack_nr = $_repack;
2377                $_repack_flags ||= '-d';
2378        }
2379}
2380
2381sub read_grafts {
2382        my $gr_file = shift;
2383        my ($grafts, $comments) = ({}, {});
2384        if (open my $fh, '<', $gr_file) {
2385                my @tmp;
2386                while (<$fh>) {
2387                        if (/^($sha1)\s+/) {
2388                                my $c = $1;
2389                                if (@tmp) {
2390                                        @{$comments->{$c}} = @tmp;
2391                                        @tmp = ();
2392                                }
2393                                foreach my $p (split /\s+/, $_) {
2394                                        $grafts->{$c}->{$p} = 1;
2395                                }
2396                        } else {
2397                                push @tmp, $_;
2398                        }
2399                }
2400                close $fh or croak $!;
2401                @{$comments->{'END'}} = @tmp if @tmp;
2402        }
2403        return ($grafts, $comments);
2404}
2405
2406sub write_grafts {
2407        my ($grafts, $comments, $gr_file) = @_;
2408
2409        open my $fh, '>', $gr_file or croak $!;
2410        foreach my $c (sort keys %$grafts) {
2411                if ($comments->{$c}) {
2412                        print $fh $_ foreach @{$comments->{$c}};
2413                }
2414                my $p = $grafts->{$c};
2415                my %x; # real parents
2416                delete $p->{$c}; # commits are not self-reproducing...
2417                my $pid = open my $ch, '-|';
2418                defined $pid or croak $!;
2419                if (!$pid) {
2420                        exec(qw/git-cat-file commit/, $c) or croak $!;
2421                }
2422                while (<$ch>) {
2423                        if (/^parent ($sha1)/) {
2424                                $x{$1} = $p->{$1} = 1;
2425                        } else {
2426                                last unless /^\S/;
2427                        }
2428                }
2429                close $ch; # breaking the pipe
2430
2431                # if real parents are the only ones in the grafts, drop it
2432                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2433
2434                my (@ip, @jp, $mb);
2435                my %del = %x;
2436                @ip = @jp = keys %$p;
2437                foreach my $i (@ip) {
2438                        next if $del{$i} || $p->{$i} == 2;
2439                        foreach my $j (@jp) {
2440                                next if $i eq $j || $del{$j} || $p->{$j} == 2;
2441                                $mb = eval { safe_qx('git-merge-base',$i,$j) };
2442                                next unless $mb;
2443                                chomp $mb;
2444                                next if $x{$mb};
2445                                if ($mb eq $j) {
2446                                        delete $p->{$i};
2447                                        $del{$i} = 1;
2448                                } elsif ($mb eq $i) {
2449                                        delete $p->{$j};
2450                                        $del{$j} = 1;
2451                                }
2452                        }
2453                }
2454
2455                # if real parents are the only ones in the grafts, drop it
2456                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2457
2458                print $fh $c, ' ', join(' ', sort keys %$p),"\n";
2459        }
2460        if ($comments->{'END'}) {
2461                print $fh $_ foreach @{$comments->{'END'}};
2462        }
2463        close $fh or croak $!;
2464}
2465
2466sub read_url_paths_all {
2467        my ($l_map, $pfx, $p) = @_;
2468        my @dir;
2469        foreach (<$p/*>) {
2470                if (-r "$_/info/url") {
2471                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2472                        my $id = $pfx . basename $_;
2473                        my $url = file_to_s("$_/info/url");
2474                        my ($u, $p) = repo_path_split($url);
2475                        $l_map->{$u}->{$p} = $id;
2476                } elsif (-d $_) {
2477                        push @dir, $_;
2478                }
2479        }
2480        foreach (@dir) {
2481                my $x = $_;
2482                $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
2483                read_url_paths_all($l_map, $x, $_);
2484        }
2485}
2486
2487# this one only gets ids that have been imported, not new ones
2488sub read_url_paths {
2489        my $l_map = {};
2490        git_svn_each(sub { my $x = shift;
2491                        my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
2492                        my ($u, $p) = repo_path_split($url);
2493                        $l_map->{$u}->{$p} = $x;
2494                        });
2495        return $l_map;
2496}
2497
2498sub extract_metadata {
2499        my $id = shift or return (undef, undef, undef);
2500        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
2501                                                        \s([a-f\d\-]+)$/x);
2502        if (!defined $rev || !$uuid || !$url) {
2503                # some of the original repositories I made had
2504                # identifiers like this:
2505                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
2506        }
2507        return ($url, $rev, $uuid);
2508}
2509
2510sub cmt_metadata {
2511        return extract_metadata((grep(/^git-svn-id: /,
2512                safe_qx(qw/git-cat-file commit/, shift)))[-1]);
2513}
2514
2515sub get_commit_time {
2516        my $cmt = shift;
2517        defined(my $pid = open my $fh, '-|') or croak $!;
2518        if (!$pid) {
2519                exec qw/git-rev-list --pretty=raw -n1/, $cmt or croak $!;
2520        }
2521        while (<$fh>) {
2522                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
2523                my ($s, $tz) = ($1, $2);
2524                if ($tz =~ s/^\+//) {
2525                        $s += tz_to_s_offset($tz);
2526                } elsif ($tz =~ s/^\-//) {
2527                        $s -= tz_to_s_offset($tz);
2528                }
2529                close $fh;
2530                return $s;
2531        }
2532        die "Can't get commit time for commit: $cmt\n";
2533}
2534
2535sub tz_to_s_offset {
2536        my ($tz) = @_;
2537        $tz =~ s/(\d\d)$//;
2538        return ($1 * 60) + ($tz * 3600);
2539}
2540
2541sub setup_pager { # translated to Perl from pager.c
2542        return unless (-t *STDOUT);
2543        my $pager = $ENV{PAGER};
2544        if (!defined $pager) {
2545                $pager = 'less';
2546        } elsif (length $pager == 0 || $pager eq 'cat') {
2547                return;
2548        }
2549        pipe my $rfd, my $wfd or return;
2550        defined(my $pid = fork) or croak $!;
2551        if (!$pid) {
2552                open STDOUT, '>&', $wfd or croak $!;
2553                return;
2554        }
2555        open STDIN, '<&', $rfd or croak $!;
2556        $ENV{LESS} ||= '-S';
2557        exec $pager or croak "Can't run pager: $!\n";;
2558}
2559
2560sub get_author_info {
2561        my ($dest, $author, $t, $tz) = @_;
2562        $author =~ s/(?:^\s*|\s*$)//g;
2563        $dest->{a_raw} = $author;
2564        my $_a;
2565        if ($_authors) {
2566                $_a = $rusers{$author} || undef;
2567        }
2568        if (!$_a) {
2569                ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
2570        }
2571        $dest->{t} = $t;
2572        $dest->{tz} = $tz;
2573        $dest->{a} = $_a;
2574        # Date::Parse isn't in the standard Perl distro :(
2575        if ($tz =~ s/^\+//) {
2576                $t += tz_to_s_offset($tz);
2577        } elsif ($tz =~ s/^\-//) {
2578                $t -= tz_to_s_offset($tz);
2579        }
2580        $dest->{t_utc} = $t;
2581}
2582
2583sub process_commit {
2584        my ($c, $r_min, $r_max, $defer) = @_;
2585        if (defined $r_min && defined $r_max) {
2586                if ($r_min == $c->{r} && $r_min == $r_max) {
2587                        show_commit($c);
2588                        return 0;
2589                }
2590                return 1 if $r_min == $r_max;
2591                if ($r_min < $r_max) {
2592                        # we need to reverse the print order
2593                        return 0 if (defined $_limit && --$_limit < 0);
2594                        push @$defer, $c;
2595                        return 1;
2596                }
2597                if ($r_min != $r_max) {
2598                        return 1 if ($r_min < $c->{r});
2599                        return 1 if ($r_max > $c->{r});
2600                }
2601        }
2602        return 0 if (defined $_limit && --$_limit < 0);
2603        show_commit($c);
2604        return 1;
2605}
2606
2607sub show_commit {
2608        my $c = shift;
2609        if ($_oneline) {
2610                my $x = "\n";
2611                if (my $l = $c->{l}) {
2612                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2613                        $x = $l->[0];
2614                }
2615                $_l_fmt ||= 'A' . length($c->{r});
2616                print 'r',pack($_l_fmt, $c->{r}),' | ';
2617                print "$c->{c} | " if $_show_commit;
2618                print $x;
2619        } else {
2620                show_commit_normal($c);
2621        }
2622}
2623
2624sub show_commit_changed_paths {
2625        my ($c) = @_;
2626        return unless $c->{changed};
2627        print "Changed paths:\n", @{$c->{changed}};
2628}
2629
2630sub show_commit_normal {
2631        my ($c) = @_;
2632        print '-' x72, "\nr$c->{r} | ";
2633        print "$c->{c} | " if $_show_commit;
2634        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2635                                 localtime($c->{t_utc})), ' | ';
2636        my $nr_line = 0;
2637
2638        if (my $l = $c->{l}) {
2639                while ($l->[$#$l] eq "\n" && $#$l > 0
2640                                          && $l->[($#$l - 1)] eq "\n") {
2641                        pop @$l;
2642                }
2643                $nr_line = scalar @$l;
2644                if (!$nr_line) {
2645                        print "1 line\n\n\n";
2646                } else {
2647                        if ($nr_line == 1) {
2648                                $nr_line = '1 line';
2649                        } else {
2650                                $nr_line .= ' lines';
2651                        }
2652                        print $nr_line, "\n";
2653                        show_commit_changed_paths($c);
2654                        print "\n";
2655                        print $_ foreach @$l;
2656                }
2657        } else {
2658                print "1 line\n";
2659                show_commit_changed_paths($c);
2660                print "\n";
2661
2662        }
2663        foreach my $x (qw/raw diff/) {
2664                if ($c->{$x}) {
2665                        print "\n";
2666                        print $_ foreach @{$c->{$x}}
2667                }
2668        }
2669}
2670
2671sub libsvn_load {
2672        return unless $_use_lib;
2673        $_use_lib = eval {
2674                require SVN::Core;
2675                if ($SVN::Core::VERSION lt '1.1.0') {
2676                        die "Need SVN::Core 1.1.0 or better ",
2677                                        "(got $SVN::Core::VERSION) ",
2678                                        "Falling back to command-line svn\n";
2679                }
2680                require SVN::Ra;
2681                require SVN::Delta;
2682                push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
2683                my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2684                                        $SVN::Node::dir.$SVN::Node::unknown.
2685                                        $SVN::Node::none.$SVN::Node::file.
2686                                        $SVN::Node::dir.$SVN::Node::unknown;
2687                1;
2688        };
2689}
2690
2691sub libsvn_connect {
2692        my ($url) = @_;
2693        my $auth = SVN::Core::auth_open([SVN::Client::get_simple_provider(),
2694                          SVN::Client::get_ssl_server_trust_file_provider(),
2695                          SVN::Client::get_username_provider()]);
2696        my $s = eval { SVN::Ra->new(url => $url, auth => $auth) };
2697        return $s;
2698}
2699
2700sub libsvn_get_file {
2701        my ($gui, $f, $rev, $chg) = @_;
2702        my $p = $f;
2703        if (length $SVN_PATH > 0) {
2704                return unless ($p =~ s#^\Q$SVN_PATH\E/##);
2705        }
2706        print "\t$chg\t$f\n" unless $_q;
2707
2708        my ($hash, $pid, $in, $out);
2709        my $pool = SVN::Pool->new;
2710        defined($pid = open3($in, $out, '>&STDERR',
2711                                qw/git-hash-object -w --stdin/)) or croak $!;
2712        # redirect STDOUT for SVN 1.1.x compatibility
2713        open my $stdout, '>&', \*STDOUT or croak $!;
2714        open STDOUT, '>&', $in or croak $!;
2715        my ($r, $props) = $SVN->get_file($f, $rev, \*STDOUT, $pool);
2716        $in->flush == 0 or croak $!;
2717        open STDOUT, '>&', $stdout or croak $!;
2718        close $in or croak $!;
2719        close $stdout or croak $!;
2720        $pool->clear;
2721        chomp($hash = do { local $/; <$out> });
2722        close $out or croak $!;
2723        waitpid $pid, 0;
2724        $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2725
2726        my $mode = exists $props->{'svn:executable'} ? '100755' : '100644';
2727        if (exists $props->{'svn:special'}) {
2728                $mode = '120000';
2729                my $link = `git-cat-file blob $hash`;
2730                $link =~ s/^link // or die "svn:special file with contents: <",
2731                                                $link, "> is not understood\n";
2732                defined($pid = open3($in, $out, '>&STDERR',
2733                                qw/git-hash-object -w --stdin/)) or croak $!;
2734                print $in $link;
2735                $in->flush == 0 or croak $!;
2736                close $in or croak $!;
2737                chomp($hash = do { local $/; <$out> });
2738                close $out or croak $!;
2739                waitpid $pid, 0;
2740                $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2741        }
2742        print $gui $mode,' ',$hash,"\t",$p,"\0" or croak $!;
2743}
2744
2745sub libsvn_log_entry {
2746        my ($rev, $author, $date, $msg, $parents) = @_;
2747        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2748                                         (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
2749                                or die "Unable to parse date: $date\n";
2750        if (defined $_authors && ! defined $users{$author}) {
2751                die "Author: $author not defined in $_authors file\n";
2752        }
2753        $msg = '' if ($rev == 0 && !defined $msg);
2754        return { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
2755                author => $author, msg => $msg."\n", parents => $parents || [] }
2756}
2757
2758sub process_rm {
2759        my ($gui, $last_commit, $f) = @_;
2760        $f =~ s#^\Q$SVN_PATH\E/?## or return;
2761        # remove entire directories.
2762        if (safe_qx('git-ls-tree',$last_commit,'--',$f) =~ /^040000 tree/) {
2763                defined(my $pid = open my $ls, '-|') or croak $!;
2764                if (!$pid) {
2765                        exec(qw/git-ls-tree -r --name-only -z/,
2766                                $last_commit,'--',$f) or croak $!;
2767                }
2768                local $/ = "\0";
2769                while (<$ls>) {
2770                        print $gui '0 ',0 x 40,"\t",$_ or croak $!;
2771                }
2772                close $ls or croak $?;
2773        } else {
2774                print $gui '0 ',0 x 40,"\t",$f,"\0" or croak $!;
2775        }
2776}
2777
2778sub libsvn_fetch {
2779        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
2780        open my $gui, '| git-update-index -z --index-info' or croak $!;
2781        my @amr;
2782        foreach my $f (keys %$paths) {
2783                my $m = $paths->{$f}->action();
2784                $f =~ s#^/+##;
2785                if ($m =~ /^[DR]$/) {
2786                        print "\t$m\t$f\n" unless $_q;
2787                        process_rm($gui, $last_commit, $f);
2788                        next if $m eq 'D';
2789                        # 'R' can be file replacements, too, right?
2790                }
2791                my $pool = SVN::Pool->new;
2792                my $t = $SVN->check_path($f, $rev, $pool);
2793                if ($t == $SVN::Node::file) {
2794                        if ($m =~ /^[AMR]$/) {
2795                                push @amr, [ $m, $f ];
2796                        } else {
2797                                die "Unrecognized action: $m, ($f r$rev)\n";
2798                        }
2799                } elsif ($t == $SVN::Node::dir && $m =~ /^[AR]$/) {
2800                        my @traversed = ();
2801                        libsvn_traverse($gui, '', $f, $rev, \@traversed);
2802                        foreach (@traversed) {
2803                                push @amr, [ $m, $_ ]
2804                        }
2805                }
2806                $pool->clear;
2807        }
2808        foreach (@amr) {
2809                libsvn_get_file($gui, $_->[1], $rev, $_->[0]);
2810        }
2811        close $gui or croak $?;
2812        return libsvn_log_entry($rev, $author, $date, $msg, [$last_commit]);
2813}
2814
2815sub svn_grab_base_rev {
2816        defined(my $pid = open my $fh, '-|') or croak $!;
2817        if (!$pid) {
2818                open my $null, '>', '/dev/null' or croak $!;
2819                open STDERR, '>&', $null or croak $!;
2820                exec qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0"
2821                                                                or croak $!;
2822        }
2823        chomp(my $c = do { local $/; <$fh> });
2824        close $fh;
2825        if (defined $c && length $c) {
2826                my ($url, $rev, $uuid) = cmt_metadata($c);
2827                return ($rev, $c) if defined $rev;
2828        }
2829        if ($_no_metadata) {
2830                my $offset = -41; # from tail
2831                my $rl;
2832                open my $fh, '<', $REVDB or
2833                        die "--no-metadata specified and $REVDB not readable\n";
2834                seek $fh, $offset, 2;
2835                $rl = readline $fh;
2836                defined $rl or return (undef, undef);
2837                chomp $rl;
2838                while ($c ne $rl && tell $fh != 0) {
2839                        $offset -= 41;
2840                        seek $fh, $offset, 2;
2841                        $rl = readline $fh;
2842                        defined $rl or return (undef, undef);
2843                        chomp $rl;
2844                }
2845                my $rev = tell $fh;
2846                croak $! if ($rev < -1);
2847                $rev =  ($rev - 41) / 41;
2848                close $fh or croak $!;
2849                return ($rev, $c);
2850        }
2851        return (undef, undef);
2852}
2853
2854sub libsvn_parse_revision {
2855        my $base = shift;
2856        my $head = $SVN->get_latest_revnum();
2857        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
2858                return ($base + 1, $head) if (defined $base);
2859                return (0, $head);
2860        }
2861        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
2862        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
2863        if ($_revision =~ /^BASE:(\d+)$/) {
2864                return ($base + 1, $1) if (defined $base);
2865                return (0, $head);
2866        }
2867        return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
2868        die "revision argument: $_revision not understood by git-svn\n",
2869                "Try using the command-line svn client instead\n";
2870}
2871
2872sub libsvn_traverse {
2873        my ($gui, $pfx, $path, $rev, $files) = @_;
2874        my $cwd = "$pfx/$path";
2875        my $pool = SVN::Pool->new;
2876        $cwd =~ s#^/+##g;
2877        my ($dirent, $r, $props) = $SVN->get_dir($cwd, $rev, $pool);
2878        foreach my $d (keys %$dirent) {
2879                my $t = $dirent->{$d}->kind;
2880                if ($t == $SVN::Node::dir) {
2881                        libsvn_traverse($gui, $cwd, $d, $rev, $files);
2882                } elsif ($t == $SVN::Node::file) {
2883                        my $file = "$cwd/$d";
2884                        if (defined $files) {
2885                                push @$files, $file;
2886                        } else {
2887                                libsvn_get_file($gui, $file, $rev, 'A');
2888                        }
2889                }
2890        }
2891        $pool->clear;
2892}
2893
2894sub libsvn_traverse_ignore {
2895        my ($fh, $path, $r) = @_;
2896        $path =~ s#^/+##g;
2897        my $pool = SVN::Pool->new;
2898        my ($dirent, undef, $props) = $SVN->get_dir($path, $r, $pool);
2899        my $p = $path;
2900        $p =~ s#^\Q$SVN_PATH\E/?##;
2901        print $fh length $p ? "\n# $p\n" : "\n# /\n";
2902        if (my $s = $props->{'svn:ignore'}) {
2903                $s =~ s/[\r\n]+/\n/g;
2904                chomp $s;
2905                if (length $p == 0) {
2906                        $s =~ s#\n#\n/$p#g;
2907                        print $fh "/$s\n";
2908                } else {
2909                        $s =~ s#\n#\n/$p/#g;
2910                        print $fh "/$p/$s\n";
2911                }
2912        }
2913        foreach (sort keys %$dirent) {
2914                next if $dirent->{$_}->kind != $SVN::Node::dir;
2915                libsvn_traverse_ignore($fh, "$path/$_", $r);
2916        }
2917        $pool->clear;
2918}
2919
2920sub revisions_eq {
2921        my ($path, $r0, $r1) = @_;
2922        return 1 if $r0 == $r1;
2923        my $nr = 0;
2924        if ($_use_lib) {
2925                # should be OK to use Pool here (r1 - r0) should be small
2926                my $pool = SVN::Pool->new;
2927                libsvn_get_log($SVN, "/$path", $r0, $r1,
2928                                0, 1, 1, sub {$nr++}, $pool);
2929                $pool->clear;
2930        } else {
2931                my ($url, undef) = repo_path_split($SVN_URL);
2932                my $svn_log = svn_log_raw("$url/$path","-r$r0:$r1");
2933                while (next_log_entry($svn_log)) { $nr++ }
2934                close $svn_log->{fh};
2935        }
2936        return 0 if ($nr > 1);
2937        return 1;
2938}
2939
2940sub libsvn_find_parent_branch {
2941        my ($paths, $rev, $author, $date, $msg) = @_;
2942        my $svn_path = '/'.$SVN_PATH;
2943
2944        # look for a parent from another branch:
2945        my $i = $paths->{$svn_path} or return;
2946        my $branch_from = $i->copyfrom_path or return;
2947        my $r = $i->copyfrom_rev;
2948        print STDERR  "Found possible branch point: ",
2949                                "$branch_from => $svn_path, $r\n";
2950        $branch_from =~ s#^/##;
2951        my $l_map = {};
2952        read_url_paths_all($l_map, '', "$GIT_DIR/svn");
2953        my $url = $SVN->{url};
2954        defined $l_map->{$url} or return;
2955        my $id = $l_map->{$url}->{$branch_from};
2956        if (!defined $id && $_follow_parent) {
2957                print STDERR "Following parent: $branch_from\@$r\n";
2958                # auto create a new branch and follow it
2959                $id = basename($branch_from);
2960                $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
2961                while (-r "$GIT_DIR/svn/$id") {
2962                        # just grow a tail if we're not unique enough :x
2963                        $id .= '-';
2964                }
2965        }
2966        return unless defined $id;
2967
2968        my ($r0, $parent) = find_rev_before($r,$id,1);
2969        if ($_follow_parent && (!defined $r0 || !defined $parent)) {
2970                defined(my $pid = fork) or croak $!;
2971                if (!$pid) {
2972                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2973                        init_vars();
2974                        $SVN_URL = "$url/$branch_from";
2975                        $SVN_LOG = $SVN = undef;
2976                        setup_git_svn();
2977                        # we can't assume SVN_URL exists at r+1:
2978                        $_revision = "0:$r";
2979                        fetch_lib();
2980                        exit 0;
2981                }
2982                waitpid $pid, 0;
2983                croak $? if $?;
2984                ($r0, $parent) = find_rev_before($r,$id,1);
2985        }
2986        return unless (defined $r0 && defined $parent);
2987        if (revisions_eq($branch_from, $r0, $r)) {
2988                unlink $GIT_SVN_INDEX;
2989                print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
2990                sys(qw/git-read-tree/, $parent);
2991                return libsvn_fetch($parent, $paths, $rev,
2992                                        $author, $date, $msg);
2993        }
2994        print STDERR "Nope, branch point not imported or unknown\n";
2995        return undef;
2996}
2997
2998sub libsvn_get_log {
2999        my ($ra, @args) = @_;
3000        if ($SVN::Core::VERSION le '1.2.0') {
3001                splice(@args, 3, 1);
3002        }
3003        $ra->get_log(@args);
3004}
3005
3006sub libsvn_new_tree {
3007        if (my $log_entry = libsvn_find_parent_branch(@_)) {
3008                return $log_entry;
3009        }
3010        my ($paths, $rev, $author, $date, $msg) = @_;
3011        open my $gui, '| git-update-index -z --index-info' or croak $!;
3012        libsvn_traverse($gui, '', $SVN_PATH, $rev);
3013        close $gui or croak $?;
3014        return libsvn_log_entry($rev, $author, $date, $msg);
3015}
3016
3017sub find_graft_path_commit {
3018        my ($tree_paths, $p1, $r1) = @_;
3019        foreach my $x (keys %$tree_paths) {
3020                next unless ($p1 =~ /^\Q$x\E/);
3021                my $i = $tree_paths->{$x};
3022                my ($r0, $parent) = find_rev_before($r1,$i,1);
3023                return $parent if (defined $r0 && $r0 == $r1);
3024                print STDERR "r$r1 of $i not imported\n";
3025                next;
3026        }
3027        return undef;
3028}
3029
3030sub find_graft_path_parents {
3031        my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
3032        foreach my $x (keys %$tree_paths) {
3033                next unless ($p0 =~ /^\Q$x\E/);
3034                my $i = $tree_paths->{$x};
3035                my ($r, $parent) = find_rev_before($r0, $i, 1);
3036                if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
3037                        my ($url_b, undef, $uuid_b) = cmt_metadata($c);
3038                        my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
3039                        next if ($url_a && $url_b && $url_a eq $url_b &&
3040                                                        $uuid_b eq $uuid_a);
3041                        $grafts->{$c}->{$parent} = 1;
3042                }
3043        }
3044}
3045
3046sub libsvn_graft_file_copies {
3047        my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
3048        foreach (keys %$paths) {
3049                my $i = $paths->{$_};
3050                my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
3051                                        $i->copyfrom_rev);
3052                next unless (defined $p0 && defined $r0);
3053
3054                my $p1 = $_;
3055                $p1 =~ s#^/##;
3056                $p0 =~ s#^/##;
3057                my $c = find_graft_path_commit($tree_paths, $p1, $rev);
3058                next unless $c;
3059                find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
3060        }
3061}
3062
3063sub set_index {
3064        my $old = $ENV{GIT_INDEX_FILE};
3065        $ENV{GIT_INDEX_FILE} = shift;
3066        return $old;
3067}
3068
3069sub restore_index {
3070        my ($old) = @_;
3071        if (defined $old) {
3072                $ENV{GIT_INDEX_FILE} = $old;
3073        } else {
3074                delete $ENV{GIT_INDEX_FILE};
3075        }
3076}
3077
3078sub libsvn_commit_cb {
3079        my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
3080        if ($_optimize_commits && $rev == ($r_last + 1)) {
3081                my $log = libsvn_log_entry($rev,$committer,$date,$msg);
3082                $log->{tree} = get_tree_from_treeish($c);
3083                my $cmt = git_commit($log, $cmt_last, $c);
3084                my @diff = safe_qx('git-diff-tree', $cmt, $c);
3085                if (@diff) {
3086                        print STDERR "Trees differ: $cmt $c\n",
3087                                        join('',@diff),"\n";
3088                        exit 1;
3089                }
3090        } else {
3091                fetch("$rev=$c");
3092        }
3093}
3094
3095sub libsvn_ls_fullurl {
3096        my $fullurl = shift;
3097        my ($repo, $path) = repo_path_split($fullurl);
3098        $SVN ||= libsvn_connect($repo);
3099        my @ret;
3100        my $pool = SVN::Pool->new;
3101        my ($dirent, undef, undef) = $SVN->get_dir($path,
3102                                                $SVN->get_latest_revnum, $pool);
3103        foreach my $d (keys %$dirent) {
3104                if ($dirent->{$d}->kind == $SVN::Node::dir) {
3105                        push @ret, "$d/"; # add '/' for compat with cli svn
3106                }
3107        }
3108        $pool->clear;
3109        return @ret;
3110}
3111
3112
3113sub libsvn_skip_unknown_revs {
3114        my $err = shift;
3115        my $errno = $err->apr_err();
3116        # Maybe the branch we're tracking didn't
3117        # exist when the repo started, so it's
3118        # not an error if it doesn't, just continue
3119        #
3120        # Wonderfully consistent library, eh?
3121        # 160013 - svn:// and file://
3122        # 175002 - http(s)://
3123        #   More codes may be discovered later...
3124        if ($errno == 175002 || $errno == 160013) {
3125                return;
3126        }
3127        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3128};
3129
3130# Tie::File seems to be prone to offset errors if revisions get sparse,
3131# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3132# one of my favorite modules is out :<  Next up would be one of the DBM
3133# modules, but I'm not sure which is most portable...  So I'll just
3134# go with something that's plain-text, but still capable of
3135# being randomly accessed.  So here's my ultra-simple fixed-width
3136# database.  All records are 40 characters + "\n", so it's easy to seek
3137# to a revision: (41 * rev) is the byte offset.
3138# A record of 40 0s denotes an empty revision.
3139# And yes, it's still pretty fast (faster than Tie::File).
3140sub revdb_set {
3141        my ($file, $rev, $commit) = @_;
3142        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
3143        open my $fh, '+<', $file or croak $!;
3144        my $offset = $rev * 41;
3145        # assume that append is the common case:
3146        seek $fh, 0, 2 or croak $!;
3147        my $pos = tell $fh;
3148        if ($pos < $offset) {
3149                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
3150        }
3151        seek $fh, $offset, 0 or croak $!;
3152        print $fh $commit,"\n";
3153        close $fh or croak $!;
3154}
3155
3156sub revdb_get {
3157        my ($file, $rev) = @_;
3158        my $ret;
3159        my $offset = $rev * 41;
3160        open my $fh, '<', $file or croak $!;
3161        seek $fh, $offset, 0;
3162        if (tell $fh == $offset) {
3163                $ret = readline $fh;
3164                if (defined $ret) {
3165                        chomp $ret;
3166                        $ret = undef if ($ret =~ /^0{40}$/);
3167                }
3168        }
3169        close $fh or croak $!;
3170        return $ret;
3171}
3172
3173sub copy_remote_ref {
3174        my $origin = $_cp_remote ? $_cp_remote : 'origin';
3175        my $ref = "refs/remotes/$GIT_SVN";
3176        if (safe_qx('git-ls-remote', $origin, $ref)) {
3177                sys(qw/git fetch/, $origin, "$ref:$ref");
3178        } elsif ($_cp_remote && !$_upgrade) {
3179                die "Unable to find remote reference: ",
3180                                "refs/remotes/$GIT_SVN on $origin\n";
3181        }
3182}
3183
3184package SVN::Git::Editor;
3185use vars qw/@ISA/;
3186use strict;
3187use warnings;
3188use Carp qw/croak/;
3189use IO::File;
3190
3191sub new {
3192        my $class = shift;
3193        my $git_svn = shift;
3194        my $self = SVN::Delta::Editor->new(@_);
3195        bless $self, $class;
3196        foreach (qw/svn_path c r ra /) {
3197                die "$_ required!\n" unless (defined $git_svn->{$_});
3198                $self->{$_} = $git_svn->{$_};
3199        }
3200        $self->{pool} = SVN::Pool->new;
3201        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3202        $self->{rm} = { };
3203        require Digest::MD5;
3204        return $self;
3205}
3206
3207sub split_path {
3208        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3209}
3210
3211sub repo_path {
3212        (defined $_[1] && length $_[1]) ? "$_[0]->{svn_path}/$_[1]"
3213                                        : $_[0]->{svn_path}
3214}
3215
3216sub url_path {
3217        my ($self, $path) = @_;
3218        $self->{ra}->{url} . '/' . $self->repo_path($path);
3219}
3220
3221sub rmdirs {
3222        my ($self, $q) = @_;
3223        my $rm = $self->{rm};
3224        delete $rm->{''}; # we never delete the url we're tracking
3225        return unless %$rm;
3226
3227        foreach (keys %$rm) {
3228                my @d = split m#/#, $_;
3229                my $c = shift @d;
3230                $rm->{$c} = 1;
3231                while (@d) {
3232                        $c .= '/' . shift @d;
3233                        $rm->{$c} = 1;
3234                }
3235        }
3236        delete $rm->{$self->{svn_path}};
3237        delete $rm->{''}; # we never delete the url we're tracking
3238        return unless %$rm;
3239
3240        defined(my $pid = open my $fh,'-|') or croak $!;
3241        if (!$pid) {
3242                exec qw/git-ls-tree --name-only -r -z/, $self->{c} or croak $!;
3243        }
3244        local $/ = "\0";
3245        my @svn_path = split m#/#, $self->{svn_path};
3246        while (<$fh>) {
3247                chomp;
3248                my @dn = (@svn_path, (split m#/#, $_));
3249                while (pop @dn) {
3250                        delete $rm->{join '/', @dn};
3251                }
3252                unless (%$rm) {
3253                        close $fh;
3254                        return;
3255                }
3256        }
3257        close $fh;
3258
3259        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3260        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3261                $self->close_directory($bat->{$d}, $p);
3262                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3263                print "\tD+\t/$d/\n" unless $q;
3264                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3265                delete $bat->{$d};
3266        }
3267}
3268
3269sub open_or_add_dir {
3270        my ($self, $full_path, $baton) = @_;
3271        my $p = SVN::Pool->new;
3272        my $t = $self->{ra}->check_path($full_path, $self->{r}, $p);
3273        $p->clear;
3274        if ($t == $SVN::Node::none) {
3275                return $self->add_directory($full_path, $baton,
3276                                                undef, -1, $self->{pool});
3277        } elsif ($t == $SVN::Node::dir) {
3278                return $self->open_directory($full_path, $baton,
3279                                                $self->{r}, $self->{pool});
3280        }
3281        print STDERR "$full_path already exists in repository at ",
3282                "r$self->{r} and it is not a directory (",
3283                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3284        exit 1;
3285}
3286
3287sub ensure_path {
3288        my ($self, $path) = @_;
3289        my $bat = $self->{bat};
3290        $path = $self->repo_path($path);
3291        return $bat->{''} unless (length $path);
3292        my @p = split m#/+#, $path;
3293        my $c = shift @p;
3294        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3295        while (@p) {
3296                my $c0 = $c;
3297                $c .= '/' . shift @p;
3298                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3299        }
3300        return $bat->{$c};
3301}
3302
3303sub A {
3304        my ($self, $m, $q) = @_;
3305        my ($dir, $file) = split_path($m->{file_b});
3306        my $pbat = $self->ensure_path($dir);
3307        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3308                                        undef, -1);
3309        print "\tA\t$m->{file_b}\n" unless $q;
3310        $self->chg_file($fbat, $m);
3311        $self->close_file($fbat,undef,$self->{pool});
3312}
3313
3314sub C {
3315        my ($self, $m, $q) = @_;
3316        my ($dir, $file) = split_path($m->{file_b});
3317        my $pbat = $self->ensure_path($dir);
3318        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3319                                $self->url_path($m->{file_a}), $self->{r});
3320        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $q;
3321        $self->chg_file($fbat, $m);
3322        $self->close_file($fbat,undef,$self->{pool});
3323}
3324
3325sub delete_entry {
3326        my ($self, $path, $pbat) = @_;
3327        my $rpath = $self->repo_path($path);
3328        my ($dir, $file) = split_path($rpath);
3329        $self->{rm}->{$dir} = 1;
3330        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3331}
3332
3333sub R {
3334        my ($self, $m, $q) = @_;
3335        my ($dir, $file) = split_path($m->{file_b});
3336        my $pbat = $self->ensure_path($dir);
3337        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3338                                $self->url_path($m->{file_a}), $self->{r});
3339        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $q;
3340        $self->chg_file($fbat, $m);
3341        $self->close_file($fbat,undef,$self->{pool});
3342
3343        ($dir, $file) = split_path($m->{file_a});
3344        $pbat = $self->ensure_path($dir);
3345        $self->delete_entry($m->{file_a}, $pbat);
3346}
3347
3348sub M {
3349        my ($self, $m, $q) = @_;
3350        my ($dir, $file) = split_path($m->{file_b});
3351        my $pbat = $self->ensure_path($dir);
3352        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3353                                $pbat,$self->{r},$self->{pool});
3354        print "\t$m->{chg}\t$m->{file_b}\n" unless $q;
3355        $self->chg_file($fbat, $m);
3356        $self->close_file($fbat,undef,$self->{pool});
3357}
3358
3359sub T { shift->M(@_) }
3360
3361sub change_file_prop {
3362        my ($self, $fbat, $pname, $pval) = @_;
3363        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3364}
3365
3366sub chg_file {
3367        my ($self, $fbat, $m) = @_;
3368        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3369                $self->change_file_prop($fbat,'svn:executable','*');
3370        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3371                $self->change_file_prop($fbat,'svn:executable',undef);
3372        }
3373        my $fh = IO::File->new_tmpfile or croak $!;
3374        if ($m->{mode_b} =~ /^120/) {
3375                print $fh 'link ' or croak $!;
3376                $self->change_file_prop($fbat,'svn:special','*');
3377        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3378                $self->change_file_prop($fbat,'svn:special',undef);
3379        }
3380        defined(my $pid = fork) or croak $!;
3381        if (!$pid) {
3382                open STDOUT, '>&', $fh or croak $!;
3383                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3384        }
3385        waitpid $pid, 0;
3386        croak $? if $?;
3387        $fh->flush == 0 or croak $!;
3388        seek $fh, 0, 0 or croak $!;
3389
3390        my $md5 = Digest::MD5->new;
3391        $md5->addfile($fh) or croak $!;
3392        seek $fh, 0, 0 or croak $!;
3393
3394        my $exp = $md5->hexdigest;
3395        my $pool = SVN::Pool->new;
3396        my $atd = $self->apply_textdelta($fbat, undef, $pool);
3397        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3398        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3399        $pool->clear;
3400
3401        close $fh or croak $!;
3402}
3403
3404sub D {
3405        my ($self, $m, $q) = @_;
3406        my ($dir, $file) = split_path($m->{file_b});
3407        my $pbat = $self->ensure_path($dir);
3408        print "\tD\t$m->{file_b}\n" unless $q;
3409        $self->delete_entry($m->{file_b}, $pbat);
3410}
3411
3412sub close_edit {
3413        my ($self) = @_;
3414        my ($p,$bat) = ($self->{pool}, $self->{bat});
3415        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3416                $self->close_directory($bat->{$_}, $p);
3417        }
3418        $self->SUPER::close_edit($p);
3419        $p->clear;
3420}
3421
3422sub abort_edit {
3423        my ($self) = @_;
3424        $self->SUPER::abort_edit($self->{pool});
3425        $self->{pool}->clear;
3426}
3427
3428__END__
3429
3430Data structures:
3431
3432$svn_log hashref (as returned by svn_log_raw)
3433{
3434        fh => file handle of the log file,
3435        state => state of the log file parser (sep/msg/rev/msg_start...)
3436}
3437
3438$log_msg hashref as returned by next_log_entry($svn_log)
3439{
3440        msg => 'whitespace-formatted log entry
3441',                                              # trailing newline is preserved
3442        revision => '8',                        # integer
3443        date => '2004-02-24T17:01:44.108345Z',  # commit date
3444        author => 'committer name'
3445};
3446
3447
3448@mods = array of diff-index line hashes, each element represents one line
3449        of diff-index output
3450
3451diff-index line ($m hash)
3452{
3453        mode_a => first column of diff-index output, no leading ':',
3454        mode_b => second column of diff-index output,
3455        sha1_b => sha1sum of the final blob,
3456        chg => change type [MCRADT],
3457        file_a => original file name of a file (iff chg is 'C' or 'R')
3458        file_b => new/current file name of a file (any chg)
3459}
3460;
3461
3462# retval of read_url_paths{,_all}();
3463$l_map = {
3464        # repository root url
3465        'https://svn.musicpd.org' => {
3466                # repository path               # GIT_SVN_ID
3467                'mpd/trunk'             =>      'trunk',
3468                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
3469        },
3470}
3471
3472Notes:
3473        I don't trust the each() function on unless I created %hash myself
3474        because the internal iterator may not have started at base.