f453c9ab98ab6d020f9cb95dedd3dfc2a086e665
   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# properties that we do not log:
  25my %SKIP = ( 'svn:wc:ra_dav:version-url' => 1,
  26             'svn:special' => 1,
  27             'svn:executable' => 1,
  28             'svn:entry:committed-rev' => 1,
  29             'svn:entry:last-author' => 1,
  30             'svn:entry:uuid' => 1,
  31             'svn:entry:committed-date' => 1,
  32);
  33
  34sub fatal (@) { print STDERR @_; exit 1 }
  35# If SVN:: library support is added, please make the dependencies
  36# optional and preserve the capability to use the command-line client.
  37# use eval { require SVN::... } to make it lazy load
  38# We don't use any modules not in the standard Perl distribution:
  39use Carp qw/croak/;
  40use IO::File qw//;
  41use File::Basename qw/dirname basename/;
  42use File::Path qw/mkpath/;
  43use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  44use File::Spec qw//;
  45use File::Copy qw/copy/;
  46use POSIX qw/strftime/;
  47use IPC::Open3;
  48use Memoize;
  49use Git qw/command command_oneline command_noisy
  50           command_output_pipe command_input_pipe command_close_pipe/;
  51memoize('revisions_eq');
  52memoize('cmt_metadata');
  53memoize('get_commit_time');
  54
  55my ($SVN, $_use_lib);
  56
  57sub nag_lib {
  58        print STDERR <<EOF;
  59! Please consider installing the SVN Perl libraries (version 1.1.0 or
  60! newer).  You will generally get better performance and fewer bugs,
  61! especially if you:
  62! 1) have a case-insensitive filesystem
  63! 2) replace symlinks with files (and vice-versa) in commits
  64
  65EOF
  66}
  67
  68$_use_lib = 1 unless $ENV{GIT_SVN_NO_LIB};
  69libsvn_load();
  70nag_lib() unless $_use_lib;
  71
  72my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  73my $sha1 = qr/[a-f\d]{40}/;
  74my $sha1_short = qr/[a-f\d]{4,40}/;
  75my $_esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
  76my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
  77        $_find_copies_harder, $_l, $_cp_similarity, $_cp_remote,
  78        $_repack, $_repack_nr, $_repack_flags, $_q,
  79        $_message, $_file, $_follow_parent, $_no_metadata,
  80        $_template, $_shared, $_no_default_regex, $_no_graft_copy,
  81        $_limit, $_verbose, $_incremental, $_oneline, $_l_fmt, $_show_commit,
  82        $_version, $_upgrade, $_authors, $_branch_all_refs, @_opt_m,
  83        $_merge, $_strategy, $_dry_run, $_ignore_nodate, $_non_recursive,
  84        $_username, $_config_dir, $_no_auth_cache, $_xfer_delta,
  85        $_pager, $_color);
  86my (@_branch_from, %tree_map, %users, %rusers, %equiv);
  87my ($_svn_co_url_revs, $_svn_pg_peg_revs, $_svn_can_do_switch);
  88my @repo_path_split_cache;
  89
  90my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
  91                'branch|b=s' => \@_branch_from,
  92                'follow-parent|follow' => \$_follow_parent,
  93                'branch-all-refs|B' => \$_branch_all_refs,
  94                'authors-file|A=s' => \$_authors,
  95                'repack:i' => \$_repack,
  96                'no-metadata' => \$_no_metadata,
  97                'quiet|q' => \$_q,
  98                'username=s' => \$_username,
  99                'config-dir=s' => \$_config_dir,
 100                'no-auth-cache' => \$_no_auth_cache,
 101                'ignore-nodate' => \$_ignore_nodate,
 102                'repack-flags|repack-args|repack-opts=s' => \$_repack_flags);
 103
 104my ($_trunk, $_tags, $_branches);
 105my %multi_opts = ( 'trunk|T=s' => \$_trunk,
 106                'tags|t=s' => \$_tags,
 107                'branches|b=s' => \$_branches );
 108my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
 109my %cmt_opts = ( 'edit|e' => \$_edit,
 110                'rmdir' => \$_rmdir,
 111                'find-copies-harder' => \$_find_copies_harder,
 112                'l=i' => \$_l,
 113                'copy-similarity|C=i'=> \$_cp_similarity
 114);
 115
 116my %cmd = (
 117        fetch => [ \&fetch, "Download new revisions from SVN",
 118                        { 'revision|r=s' => \$_revision, %fc_opts } ],
 119        init => [ \&init, "Initialize a repo for tracking" .
 120                          " (requires URL argument)",
 121                          \%init_opts ],
 122        commit => [ \&commit, "Commit git revisions to SVN",
 123                        {       'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 124        'show-ignore' => [ \&show_ignore, "Show svn:ignore listings",
 125                        { 'revision|r=i' => \$_revision } ],
 126        rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
 127                        { 'no-ignore-externals' => \$_no_ignore_ext,
 128                          'copy-remote|remote=s' => \$_cp_remote,
 129                          'upgrade' => \$_upgrade } ],
 130        'graft-branches' => [ \&graft_branches,
 131                        'Detect merges/branches from already imported history',
 132                        { 'merge-rx|m' => \@_opt_m,
 133                          'branch|b=s' => \@_branch_from,
 134                          'branch-all-refs|B' => \$_branch_all_refs,
 135                          'no-default-regex' => \$_no_default_regex,
 136                          'no-graft-copy' => \$_no_graft_copy } ],
 137        'multi-init' => [ \&multi_init,
 138                        'Initialize multiple trees (like git-svnimport)',
 139                        { %multi_opts, %init_opts,
 140                         'revision|r=i' => \$_revision,
 141                         'username=s' => \$_username,
 142                         'config-dir=s' => \$_config_dir,
 143                         'no-auth-cache' => \$_no_auth_cache,
 144                        } ],
 145        'multi-fetch' => [ \&multi_fetch,
 146                        'Fetch multiple trees (like git-svnimport)',
 147                        \%fc_opts ],
 148        'log' => [ \&show_log, 'Show commit logs',
 149                        { 'limit=i' => \$_limit,
 150                          'revision|r=s' => \$_revision,
 151                          'verbose|v' => \$_verbose,
 152                          'incremental' => \$_incremental,
 153                          'oneline' => \$_oneline,
 154                          'show-commit' => \$_show_commit,
 155                          'non-recursive' => \$_non_recursive,
 156                          'authors-file|A=s' => \$_authors,
 157                          'color' => \$_color,
 158                          'pager=s' => \$_pager,
 159                        } ],
 160        'commit-diff' => [ \&commit_diff, 'Commit a diff between two trees',
 161                        { 'message|m=s' => \$_message,
 162                          'file|F=s' => \$_file,
 163                          'revision|r=s' => \$_revision,
 164                        %cmt_opts } ],
 165        dcommit => [ \&dcommit, 'Commit several diffs to merge with upstream',
 166                        { 'merge|m|M' => \$_merge,
 167                          'strategy|s=s' => \$_strategy,
 168                          'dry-run|n' => \$_dry_run,
 169                        %cmt_opts } ],
 170);
 171
 172my $cmd;
 173for (my $i = 0; $i < @ARGV; $i++) {
 174        if (defined $cmd{$ARGV[$i]}) {
 175                $cmd = $ARGV[$i];
 176                splice @ARGV, $i, 1;
 177                last;
 178        }
 179};
 180
 181my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 182
 183read_repo_config(\%opts);
 184my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
 185                                'version|V' => \$_version,
 186                                'id|i=s' => \$GIT_SVN);
 187exit 1 if (!$rv && $cmd ne 'log');
 188
 189set_default_vals();
 190usage(0) if $_help;
 191version() if $_version;
 192usage(1) unless defined $cmd;
 193init_vars();
 194load_authors() if $_authors;
 195load_all_refs() if $_branch_all_refs;
 196svn_compat_check() unless $_use_lib;
 197migration_check() unless $cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/;
 198$cmd{$cmd}->[0]->(@ARGV);
 199exit 0;
 200
 201####################### primary functions ######################
 202sub usage {
 203        my $exit = shift || 0;
 204        my $fd = $exit ? \*STDERR : \*STDOUT;
 205        print $fd <<"";
 206git-svn - bidirectional operations between a single Subversion tree and git
 207Usage: $0 <command> [options] [arguments]\n
 208
 209        print $fd "Available commands:\n" unless $cmd;
 210
 211        foreach (sort keys %cmd) {
 212                next if $cmd && $cmd ne $_;
 213                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 214                foreach (keys %{$cmd{$_}->[2]}) {
 215                        # prints out arguments as they should be passed:
 216                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 217                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 218                                                        "--$_" : "-$_" }
 219                                                split /\|/,$_)," $x\n";
 220                }
 221        }
 222        print $fd <<"";
 223\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 224arbitrary identifier if you're tracking multiple SVN branches/repositories in
 225one git repository and want to keep them separate.  See git-svn(1) for more
 226information.
 227
 228        exit $exit;
 229}
 230
 231sub version {
 232        print "git-svn version $VERSION\n";
 233        exit 0;
 234}
 235
 236sub rebuild {
 237        if (!verify_ref("refs/remotes/$GIT_SVN^0")) {
 238                copy_remote_ref();
 239        }
 240        $SVN_URL = shift or undef;
 241        my $newest_rev = 0;
 242        if ($_upgrade) {
 243                command_noisy('update-ref',"refs/remotes/$GIT_SVN","
 244                              $GIT_SVN-HEAD");
 245        } else {
 246                check_upgrade_needed();
 247        }
 248
 249        my ($rev_list, $ctx) = command_output_pipe("rev-list",
 250                                                   "refs/remotes/$GIT_SVN");
 251        my $latest;
 252        while (<$rev_list>) {
 253                chomp;
 254                my $c = $_;
 255                croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 256                my @commit = grep(/^git-svn-id: /,
 257                                  command(qw/cat-file commit/, $c));
 258                next if (!@commit); # skip merges
 259                my ($url, $rev, $uuid) = extract_metadata($commit[$#commit]);
 260                if (!defined $rev || !$uuid) {
 261                        croak "Unable to extract revision or UUID from ",
 262                                "$c, $commit[$#commit]\n";
 263                }
 264
 265                # if we merged or otherwise started elsewhere, this is
 266                # how we break out of it
 267                next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
 268                next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
 269
 270                unless (defined $latest) {
 271                        if (!$SVN_URL && !$url) {
 272                                croak "SVN repository location required: $url\n";
 273                        }
 274                        $SVN_URL ||= $url;
 275                        $SVN_UUID ||= $uuid;
 276                        setup_git_svn();
 277                        $latest = $rev;
 278                }
 279                revdb_set($REVDB, $rev, $c);
 280                print "r$rev = $c\n";
 281                $newest_rev = $rev if ($rev > $newest_rev);
 282        }
 283        command_close_pipe($rev_list, $ctx);
 284
 285        goto out if $_use_lib;
 286        if (!chdir $SVN_WC) {
 287                svn_cmd_checkout($SVN_URL, $latest, $SVN_WC);
 288                chdir $SVN_WC or croak $!;
 289        }
 290
 291        my $pid = fork;
 292        defined $pid or croak $!;
 293        if ($pid == 0) {
 294                my @svn_up = qw(svn up);
 295                push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 296                sys(@svn_up,"-r$newest_rev");
 297                $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 298                index_changes();
 299                exec('git-write-tree') or croak $!;
 300        }
 301        waitpid $pid, 0;
 302        croak $? if $?;
 303out:
 304        if ($_upgrade) {
 305                print STDERR <<"";
 306Keeping deprecated refs/head/$GIT_SVN-HEAD for now.  Please remove it
 307when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN
 308
 309        }
 310}
 311
 312sub init {
 313        my $url = shift or die "SVN repository location required " .
 314                                "as a command-line argument\n";
 315        $url =~ s!/+$!!; # strip trailing slash
 316
 317        if (my $repo_path = shift) {
 318                unless (-d $repo_path) {
 319                        mkpath([$repo_path]);
 320                }
 321                $GIT_DIR = $ENV{GIT_DIR} = $repo_path . "/.git";
 322                init_vars();
 323        }
 324
 325        $SVN_URL = $url;
 326        unless (-d $GIT_DIR) {
 327                my @init_db = ('init-db');
 328                push @init_db, "--template=$_template" if defined $_template;
 329                push @init_db, "--shared" if defined $_shared;
 330                command_noisy(@init_db);
 331        }
 332        setup_git_svn();
 333}
 334
 335sub fetch {
 336        check_upgrade_needed();
 337        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 338        my $ret = $_use_lib ? fetch_lib(@_) : fetch_cmd(@_);
 339        if ($ret->{commit} && !verify_ref('refs/heads/master^0')) {
 340                command_noisy(qw(update-ref refs/heads/master),$ret->{commit});
 341        }
 342        return $ret;
 343}
 344
 345sub fetch_cmd {
 346        my (@parents) = @_;
 347        my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 348        unless ($_revision) {
 349                $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 350        }
 351        push @log_args, "-r$_revision";
 352        push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 353
 354        my $svn_log = svn_log_raw(@log_args);
 355
 356        my $base = next_log_entry($svn_log) or croak "No base revision!\n";
 357        # don't need last_revision from grab_base_rev() because
 358        # user could've specified a different revision to skip (they
 359        # didn't want to import certain revisions into git for whatever
 360        # reason, so trust $base->{revision} instead.
 361        my (undef, $last_commit) = svn_grab_base_rev();
 362        unless (-d $SVN_WC) {
 363                svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC);
 364                chdir $SVN_WC or croak $!;
 365                read_uuid();
 366                $last_commit = git_commit($base, @parents);
 367                assert_tree($last_commit);
 368        } else {
 369                chdir $SVN_WC or croak $!;
 370                read_uuid();
 371                # looks like a user manually cp'd and svn switch'ed
 372                unless ($last_commit) {
 373                        sys(qw/svn revert -R ./);
 374                        assert_svn_wc_clean($base->{revision});
 375                        $last_commit = git_commit($base, @parents);
 376                        assert_tree($last_commit);
 377                }
 378        }
 379        my @svn_up = qw(svn up);
 380        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 381        my $last = $base;
 382        while (my $log_msg = next_log_entry($svn_log)) {
 383                if ($last->{revision} >= $log_msg->{revision}) {
 384                        croak "Out of order: last >= current: ",
 385                                "$last->{revision} >= $log_msg->{revision}\n";
 386                }
 387                # Revert is needed for cases like:
 388                # https://svn.musicpd.org/Jamming/trunk (r166:167), but
 389                # I can't seem to reproduce something like that on a test...
 390                sys(qw/svn revert -R ./);
 391                assert_svn_wc_clean($last->{revision});
 392                sys(@svn_up,"-r$log_msg->{revision}");
 393                $last_commit = git_commit($log_msg, $last_commit, @parents);
 394                $last = $log_msg;
 395        }
 396        close $svn_log->{fh};
 397        $last->{commit} = $last_commit;
 398        return $last;
 399}
 400
 401sub fetch_lib {
 402        my (@parents) = @_;
 403        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 404        $SVN ||= libsvn_connect($SVN_URL);
 405        my ($last_rev, $last_commit) = svn_grab_base_rev();
 406        my ($base, $head) = libsvn_parse_revision($last_rev);
 407        if ($base > $head) {
 408                return { revision => $last_rev, commit => $last_commit }
 409        }
 410        my $index = set_index($GIT_SVN_INDEX);
 411
 412        # limit ourselves and also fork() since get_log won't release memory
 413        # after processing a revision and SVN stuff seems to leak
 414        my $inc = 1000;
 415        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
 416        read_uuid();
 417        if (defined $last_commit) {
 418                unless (-e $GIT_SVN_INDEX) {
 419                        command_noisy('read-tree', $last_commit);
 420                }
 421                my $x = command_oneline('write-tree');
 422                my ($y) = (command(qw/cat-file commit/, $last_commit)
 423                                                        =~ /^tree ($sha1)/m);
 424                if ($y ne $x) {
 425                        unlink $GIT_SVN_INDEX or croak $!;
 426                        command_noisy('read-tree', $last_commit);
 427                }
 428                $x = command_oneline('write-tree');
 429                if ($y ne $x) {
 430                        print STDERR "trees ($last_commit) $y != $x\n",
 431                                 "Something is seriously wrong...\n";
 432                }
 433        }
 434        while (1) {
 435                # fork, because using SVN::Pool with get_log() still doesn't
 436                # seem to help enough to keep memory usage down.
 437                defined(my $pid = fork) or croak $!;
 438                if (!$pid) {
 439                        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
 440
 441                        # Yes I'm perfectly aware that the fourth argument
 442                        # below is the limit revisions number.  Unfortunately
 443                        # performance sucks with it enabled, so it's much
 444                        # faster to fetch revision ranges instead of relying
 445                        # on the limiter.
 446                        libsvn_get_log(libsvn_dup_ra($SVN), [''],
 447                                        $min, $max, 0, 1, 1,
 448                                sub {
 449                                        my $log_msg;
 450                                        if ($last_commit) {
 451                                                $log_msg = libsvn_fetch(
 452                                                        $last_commit, @_);
 453                                                $last_commit = git_commit(
 454                                                        $log_msg,
 455                                                        $last_commit,
 456                                                        @parents);
 457                                        } else {
 458                                                $log_msg = libsvn_new_tree(@_);
 459                                                $last_commit = git_commit(
 460                                                        $log_msg, @parents);
 461                                        }
 462                                });
 463                        exit 0;
 464                }
 465                waitpid $pid, 0;
 466                croak $? if $?;
 467                ($last_rev, $last_commit) = svn_grab_base_rev();
 468                last if ($max >= $head);
 469                $min = $max + 1;
 470                $max += $inc;
 471                $max = $head if ($max > $head);
 472                $SVN = libsvn_connect($SVN_URL);
 473        }
 474        restore_index($index);
 475        return { revision => $last_rev, commit => $last_commit };
 476}
 477
 478sub commit {
 479        my (@commits) = @_;
 480        check_upgrade_needed();
 481        if ($_stdin || !@commits) {
 482                print "Reading from stdin...\n";
 483                @commits = ();
 484                while (<STDIN>) {
 485                        if (/\b($sha1_short)\b/o) {
 486                                unshift @commits, $1;
 487                        }
 488                }
 489        }
 490        my @revs;
 491        foreach my $c (@commits) {
 492                my @tmp = command('rev-parse',$c);
 493                if (scalar @tmp == 1) {
 494                        push @revs, $tmp[0];
 495                } elsif (scalar @tmp > 1) {
 496                        push @revs, reverse(command('rev-list',@tmp));
 497                } else {
 498                        die "Failed to rev-parse $c\n";
 499                }
 500        }
 501        $_use_lib ? commit_lib(@revs) : commit_cmd(@revs);
 502        print "Done committing ",scalar @revs," revisions to SVN\n";
 503}
 504
 505sub commit_cmd {
 506        my (@revs) = @_;
 507
 508        chdir $SVN_WC or croak "Unable to chdir $SVN_WC: $!\n";
 509        my $info = svn_info('.');
 510        my $fetched = fetch();
 511        if ($info->{Revision} != $fetched->{revision}) {
 512                print STDERR "There are new revisions that were fetched ",
 513                                "and need to be merged (or acknowledged) ",
 514                                "before committing.\n";
 515                exit 1;
 516        }
 517        $info = svn_info('.');
 518        read_uuid($info);
 519        my $last = $fetched;
 520        foreach my $c (@revs) {
 521                my $mods = svn_checkout_tree($last, $c);
 522                if (scalar @$mods == 0) {
 523                        print "Skipping, no changes detected\n";
 524                        next;
 525                }
 526                $last = svn_commit_tree($last, $c);
 527        }
 528}
 529
 530sub commit_lib {
 531        my (@revs) = @_;
 532        my ($r_last, $cmt_last) = svn_grab_base_rev();
 533        defined $r_last or die "Must have an existing revision to commit\n";
 534        my $fetched = fetch();
 535        if ($r_last != $fetched->{revision}) {
 536                print STDERR "There are new revisions that were fetched ",
 537                                "and need to be merged (or acknowledged) ",
 538                                "before committing.\n",
 539                                "last rev: $r_last\n",
 540                                " current: $fetched->{revision}\n";
 541                exit 1;
 542        }
 543        read_uuid();
 544        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 545        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
 546
 547        my $repo;
 548        set_svn_commit_env();
 549        foreach my $c (@revs) {
 550                my $log_msg = get_commit_message($c, $commit_msg);
 551
 552                # fork for each commit because there's a memory leak I
 553                # can't track down... (it's probably in the SVN code)
 554                defined(my $pid = open my $fh, '-|') or croak $!;
 555                if (!$pid) {
 556                        my $ed = SVN::Git::Editor->new(
 557                                        {       r => $r_last,
 558                                                ra => libsvn_dup_ra($SVN),
 559                                                c => $c,
 560                                                svn_path => $SVN->{svn_path},
 561                                        },
 562                                        $SVN->get_commit_editor(
 563                                                $log_msg->{msg},
 564                                                sub {
 565                                                        libsvn_commit_cb(
 566                                                                @_, $c,
 567                                                                $log_msg->{msg},
 568                                                                $r_last,
 569                                                                $cmt_last)
 570                                                },
 571                                                @lock)
 572                                        );
 573                        my $mods = libsvn_checkout_tree($cmt_last, $c, $ed);
 574                        if (@$mods == 0) {
 575                                print "No changes\nr$r_last = $cmt_last\n";
 576                                $ed->abort_edit;
 577                        } else {
 578                                $ed->close_edit;
 579                        }
 580                        exit 0;
 581                }
 582                my ($r_new, $cmt_new, $no);
 583                while (<$fh>) {
 584                        print $_;
 585                        chomp;
 586                        if (/^r(\d+) = ($sha1)$/o) {
 587                                ($r_new, $cmt_new) = ($1, $2);
 588                        } elsif ($_ eq 'No changes') {
 589                                $no = 1;
 590                        }
 591                }
 592                close $fh or exit 1;
 593                if (! defined $r_new && ! defined $cmt_new) {
 594                        unless ($no) {
 595                                die "Failed to parse revision information\n";
 596                        }
 597                } else {
 598                        ($r_last, $cmt_last) = ($r_new, $cmt_new);
 599                }
 600        }
 601        $ENV{LC_ALL} = 'C';
 602        unlink $commit_msg;
 603}
 604
 605sub dcommit {
 606        my $head = shift || 'HEAD';
 607        my $gs = "refs/remotes/$GIT_SVN";
 608        my @refs = command(qw/rev-list --no-merges/, "$gs..$head");
 609        my $last_rev;
 610        foreach my $d (reverse @refs) {
 611                if (!verify_ref("$d~1")) {
 612                        die "Commit $d\n",
 613                            "has no parent commit, and therefore ",
 614                            "nothing to diff against.\n",
 615                            "You should be working from a repository ",
 616                            "originally created by git-svn\n";
 617                }
 618                unless (defined $last_rev) {
 619                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 620                        unless (defined $last_rev) {
 621                                die "Unable to extract revision information ",
 622                                    "from commit $d~1\n";
 623                        }
 624                }
 625                if ($_dry_run) {
 626                        print "diff-tree $d~1 $d\n";
 627                } else {
 628                        if (my $r = commit_diff("$d~1", $d, undef, $last_rev)) {
 629                                $last_rev = $r;
 630                        } # else: no changes, same $last_rev
 631                }
 632        }
 633        return if $_dry_run;
 634        fetch();
 635        my @diff = command('diff-tree', $head, $gs, '--');
 636        my @finish;
 637        if (@diff) {
 638                @finish = qw/rebase/;
 639                push @finish, qw/--merge/ if $_merge;
 640                push @finish, "--strategy=$_strategy" if $_strategy;
 641                print STDERR "W: $head and $gs differ, using @finish:\n", @diff;
 642        } else {
 643                print "No changes between current $head and $gs\n",
 644                      "Resetting to the latest $gs\n";
 645                @finish = qw/reset --mixed/;
 646        }
 647        command_noisy(@finish, $gs);
 648}
 649
 650sub show_ignore {
 651        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 652        $_use_lib ? show_ignore_lib() : show_ignore_cmd();
 653}
 654
 655sub show_ignore_cmd {
 656        require File::Find or die $!;
 657        if (defined $_revision) {
 658                die "-r/--revision option doesn't work unless the Perl SVN ",
 659                        "libraries are used\n";
 660        }
 661        chdir $SVN_WC or croak $!;
 662        my %ign;
 663        File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 664                s#^\./##;
 665                @{$ign{$_}} = svn_propget_base('svn:ignore', $_);
 666                }}, no_chdir=>1},'.');
 667
 668        print "\n# /\n";
 669        foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 670        delete $ign{'.'};
 671        foreach my $i (sort keys %ign) {
 672                print "\n# ",$i,"\n";
 673                foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 674        }
 675}
 676
 677sub show_ignore_lib {
 678        my $repo;
 679        $SVN ||= libsvn_connect($SVN_URL);
 680        my $r = defined $_revision ? $_revision : $SVN->get_latest_revnum;
 681        libsvn_traverse_ignore(\*STDOUT, $SVN->{svn_path}, $r);
 682}
 683
 684sub graft_branches {
 685        my $gr_file = "$GIT_DIR/info/grafts";
 686        my ($grafts, $comments) = read_grafts($gr_file);
 687        my $gr_sha1;
 688
 689        if (%$grafts) {
 690                # temporarily disable our grafts file to make this idempotent
 691                chomp($gr_sha1 = command(qw/hash-object -w/,$gr_file));
 692                rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
 693        }
 694
 695        my $l_map = read_url_paths();
 696        my @re = map { qr/$_/is } @_opt_m if @_opt_m;
 697        unless ($_no_default_regex) {
 698                push @re, (qr/\b(?:merge|merging|merged)\s+with\s+([\w\.\-]+)/i,
 699                        qr/\b(?:merge|merging|merged)\s+([\w\.\-]+)/i,
 700                        qr/\b(?:from|of)\s+([\w\.\-]+)/i );
 701        }
 702        foreach my $u (keys %$l_map) {
 703                if (@re) {
 704                        foreach my $p (keys %{$l_map->{$u}}) {
 705                                graft_merge_msg($grafts,$l_map,$u,$p,@re);
 706                        }
 707                }
 708                unless ($_no_graft_copy) {
 709                        if ($_use_lib) {
 710                                graft_file_copy_lib($grafts,$l_map,$u);
 711                        } else {
 712                                graft_file_copy_cmd($grafts,$l_map,$u);
 713                        }
 714                }
 715        }
 716        graft_tree_joins($grafts);
 717
 718        write_grafts($grafts, $comments, $gr_file);
 719        unlink "$gr_file~$gr_sha1" if $gr_sha1;
 720}
 721
 722sub multi_init {
 723        my $url = shift;
 724        $_trunk ||= 'trunk';
 725        $_trunk =~ s#/+$##;
 726        $url =~ s#/+$## if $url;
 727        if ($_trunk !~ m#^[a-z\+]+://#) {
 728                $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#);
 729                unless ($url) {
 730                        print STDERR "E: '$_trunk' is not a complete URL ",
 731                                "and a separate URL is not specified\n";
 732                        exit 1;
 733                }
 734                $_trunk = $url . $_trunk;
 735        }
 736        my $ch_id;
 737        if ($GIT_SVN eq 'git-svn') {
 738                $ch_id = 1;
 739                $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
 740        }
 741        init_vars();
 742        unless (-d $GIT_SVN_DIR) {
 743                print "GIT_SVN_ID set to 'trunk' for $_trunk\n" if $ch_id;
 744                init($_trunk);
 745                command_noisy('repo-config', 'svn.trunk', $_trunk);
 746        }
 747        complete_url_ls_init($url, $_branches, '--branches/-b', '');
 748        complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/');
 749}
 750
 751sub multi_fetch {
 752        # try to do trunk first, since branches/tags
 753        # may be descended from it.
 754        if (-e "$GIT_DIR/svn/trunk/info/url") {
 755                fetch_child_id('trunk', @_);
 756        }
 757        rec_fetch('', "$GIT_DIR/svn", @_);
 758}
 759
 760sub show_log {
 761        my (@args) = @_;
 762        my ($r_min, $r_max);
 763        my $r_last = -1; # prevent dupes
 764        rload_authors() if $_authors;
 765        if (defined $TZ) {
 766                $ENV{TZ} = $TZ;
 767        } else {
 768                delete $ENV{TZ};
 769        }
 770        if (defined $_revision) {
 771                if ($_revision =~ /^(\d+):(\d+)$/) {
 772                        ($r_min, $r_max) = ($1, $2);
 773                } elsif ($_revision =~ /^\d+$/) {
 774                        $r_min = $r_max = $_revision;
 775                } else {
 776                        print STDERR "-r$_revision is not supported, use ",
 777                                "standard \'git log\' arguments instead\n";
 778                        exit 1;
 779                }
 780        }
 781
 782        config_pager();
 783        @args = (git_svn_log_cmd($r_min, $r_max), @args);
 784        my $log = command_output_pipe(@args);
 785        run_pager();
 786        my (@k, $c, $d);
 787
 788        while (<$log>) {
 789                if (/^${_esc_color}commit ($sha1_short)/o) {
 790                        my $cmt = $1;
 791                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
 792                                $r_last = $c->{r};
 793                                process_commit($c, $r_min, $r_max, \@k) or
 794                                                                goto out;
 795                        }
 796                        $d = undef;
 797                        $c = { c => $cmt };
 798                } elsif (/^${_esc_color}author (.+) (\d+) ([\-\+]?\d+)$/) {
 799                        get_author_info($c, $1, $2, $3);
 800                } elsif (/^${_esc_color}(?:tree|parent|committer) /) {
 801                        # ignore
 802                } elsif (/^${_esc_color}:\d{6} \d{6} $sha1_short/o) {
 803                        push @{$c->{raw}}, $_;
 804                } elsif (/^${_esc_color}[ACRMDT]\t/) {
 805                        # we could add $SVN->{svn_path} here, but that requires
 806                        # remote access at the moment (repo_path_split)...
 807                        s#^(${_esc_color})([ACRMDT])\t#$1   $2 #;
 808                        push @{$c->{changed}}, $_;
 809                } elsif (/^${_esc_color}diff /) {
 810                        $d = 1;
 811                        push @{$c->{diff}}, $_;
 812                } elsif ($d) {
 813                        push @{$c->{diff}}, $_;
 814                } elsif (/^${_esc_color}    (git-svn-id:.+)$/) {
 815                        ($c->{url}, $c->{r}, undef) = extract_metadata($1);
 816                } elsif (s/^${_esc_color}    //) {
 817                        push @{$c->{l}}, $_;
 818                }
 819        }
 820        if ($c && defined $c->{r} && $c->{r} != $r_last) {
 821                $r_last = $c->{r};
 822                process_commit($c, $r_min, $r_max, \@k);
 823        }
 824        if (@k) {
 825                my $swap = $r_max;
 826                $r_max = $r_min;
 827                $r_min = $swap;
 828                process_commit($_, $r_min, $r_max) foreach reverse @k;
 829        }
 830out:
 831        eval { command_close_pipe($log) };
 832        print '-' x72,"\n" unless $_incremental || $_oneline;
 833}
 834
 835sub commit_diff_usage {
 836        print STDERR "Usage: $0 commit-diff <tree-ish> <tree-ish> [<URL>]\n";
 837        exit 1
 838}
 839
 840sub commit_diff {
 841        if (!$_use_lib) {
 842                print STDERR "commit-diff must be used with SVN libraries\n";
 843                exit 1;
 844        }
 845        my $ta = shift or commit_diff_usage();
 846        my $tb = shift or commit_diff_usage();
 847        if (!eval { $SVN_URL = shift || file_to_s("$GIT_SVN_DIR/info/url") }) {
 848                print STDERR "Needed URL or usable git-svn id command-line\n";
 849                commit_diff_usage();
 850        }
 851        my $r = shift;
 852        unless (defined $r) {
 853                if (defined $_revision) {
 854                        $r = $_revision
 855                } else {
 856                        die "-r|--revision is a required argument\n";
 857                }
 858        }
 859        if (defined $_message && defined $_file) {
 860                print STDERR "Both --message/-m and --file/-F specified ",
 861                                "for the commit message.\n",
 862                                "I have no idea what you mean\n";
 863                exit 1;
 864        }
 865        if (defined $_file) {
 866                $_message = file_to_s($_file);
 867        } else {
 868                $_message ||= get_commit_message($tb,
 869                                        "$GIT_DIR/.svn-commit.tmp.$$")->{msg};
 870        }
 871        $SVN ||= libsvn_connect($SVN_URL);
 872        if ($r eq 'HEAD') {
 873                $r = $SVN->get_latest_revnum;
 874        } elsif ($r !~ /^\d+$/) {
 875                die "revision argument: $r not understood by git-svn\n";
 876        }
 877        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 878        my $rev_committed;
 879        my $ed = SVN::Git::Editor->new({        r => $r,
 880                                                ra => libsvn_dup_ra($SVN),
 881                                                c => $tb,
 882                                                svn_path => $SVN->{svn_path}
 883                                        },
 884                                $SVN->get_commit_editor($_message,
 885                                        sub {
 886                                                $rev_committed = $_[0];
 887                                                print "Committed $_[0]\n";
 888                                        }, @lock)
 889                                );
 890        eval {
 891                my $mods = libsvn_checkout_tree($ta, $tb, $ed);
 892                if (@$mods == 0) {
 893                        print "No changes\n$ta == $tb\n";
 894                        $ed->abort_edit;
 895                } else {
 896                        $ed->close_edit;
 897                }
 898        };
 899        fatal "$@\n" if $@;
 900        $_message = $_file = undef;
 901        return $rev_committed;
 902}
 903
 904########################### utility functions #########################
 905
 906sub cmt_showable {
 907        my ($c) = @_;
 908        return 1 if defined $c->{r};
 909        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
 910                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
 911                my @msg = command(qw/cat-file commit/, $c->{c});
 912                shift @msg while ($msg[0] ne "\n");
 913                shift @msg;
 914                @{$c->{l}} = grep !/^git-svn-id: /, @msg;
 915
 916                (undef, $c->{r}, undef) = extract_metadata(
 917                                (grep(/^git-svn-id: /, @msg))[-1]);
 918        }
 919        return defined $c->{r};
 920}
 921
 922sub log_use_color {
 923        return 1 if $_color;
 924        my ($dc, $dcvar);
 925        $dcvar = 'color.diff';
 926        $dc = `git-repo-config --get $dcvar`;
 927        if ($dc eq '') {
 928                # nothing at all; fallback to "diff.color"
 929                $dcvar = 'diff.color';
 930                $dc = `git-repo-config --get $dcvar`;
 931        }
 932        chomp($dc);
 933        if ($dc eq 'auto') {
 934                my $pc;
 935                $pc = `git-repo-config --get color.pager`;
 936                if ($pc eq '') {
 937                        # does not have it -- fallback to pager.color
 938                        $pc = `git-repo-config --bool --get pager.color`;
 939                }
 940                else {
 941                        $pc = `git-repo-config --bool --get color.pager`;
 942                        if ($?) {
 943                                $pc = 'false';
 944                        }
 945                }
 946                chomp($pc);
 947                if (-t *STDOUT || (defined $_pager && $pc eq 'true')) {
 948                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
 949                }
 950                return 0;
 951        }
 952        return 0 if $dc eq 'never';
 953        return 1 if $dc eq 'always';
 954        chomp($dc = `git-repo-config --bool --get $dcvar`);
 955        return ($dc eq 'true');
 956}
 957
 958sub git_svn_log_cmd {
 959        my ($r_min, $r_max) = @_;
 960        my @cmd = (qw/log --abbrev-commit --pretty=raw
 961                        --default/, "refs/remotes/$GIT_SVN");
 962        push @cmd, '-r' unless $_non_recursive;
 963        push @cmd, qw/--raw --name-status/ if $_verbose;
 964        push @cmd, '--color' if log_use_color();
 965        return @cmd unless defined $r_max;
 966        if ($r_max == $r_min) {
 967                push @cmd, '--max-count=1';
 968                if (my $c = revdb_get($REVDB, $r_max)) {
 969                        push @cmd, $c;
 970                }
 971        } else {
 972                my ($c_min, $c_max);
 973                $c_max = revdb_get($REVDB, $r_max);
 974                $c_min = revdb_get($REVDB, $r_min);
 975                if (defined $c_min && defined $c_max) {
 976                        if ($r_max > $r_max) {
 977                                push @cmd, "$c_min..$c_max";
 978                        } else {
 979                                push @cmd, "$c_max..$c_min";
 980                        }
 981                } elsif ($r_max > $r_min) {
 982                        push @cmd, $c_max;
 983                } else {
 984                        push @cmd, $c_min;
 985                }
 986        }
 987        return @cmd;
 988}
 989
 990sub fetch_child_id {
 991        my $id = shift;
 992        print "Fetching $id\n";
 993        my $ref = "$GIT_DIR/refs/remotes/$id";
 994        defined(my $pid = open my $fh, '-|') or croak $!;
 995        if (!$pid) {
 996                $_repack = undef;
 997                $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 998                init_vars();
 999                fetch(@_);
1000                exit 0;
1001        }
1002        while (<$fh>) {
1003                print $_;
1004                check_repack() if (/^r\d+ = $sha1/);
1005        }
1006        close $fh or croak $?;
1007}
1008
1009sub rec_fetch {
1010        my ($pfx, $p, @args) = @_;
1011        my @dir;
1012        foreach (sort <$p/*>) {
1013                if (-r "$_/info/url") {
1014                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
1015                        my $id = $pfx . basename $_;
1016                        next if $id eq 'trunk';
1017                        fetch_child_id($id, @args);
1018                } elsif (-d $_) {
1019                        push @dir, $_;
1020                }
1021        }
1022        foreach (@dir) {
1023                my $x = $_;
1024                $x =~ s!^\Q$GIT_DIR\E/svn/!!;
1025                rec_fetch($x, $_);
1026        }
1027}
1028
1029sub complete_url_ls_init {
1030        my ($url, $var, $switch, $pfx) = @_;
1031        unless ($var) {
1032                print STDERR "W: $switch not specified\n";
1033                return;
1034        }
1035        $var =~ s#/+$##;
1036        if ($var !~ m#^[a-z\+]+://#) {
1037                $var = '/' . $var if ($var !~ m#^/#);
1038                unless ($url) {
1039                        print STDERR "E: '$var' is not a complete URL ",
1040                                "and a separate URL is not specified\n";
1041                        exit 1;
1042                }
1043                $var = $url . $var;
1044        }
1045        chomp(my @ls = $_use_lib ? libsvn_ls_fullurl($var)
1046                                : safe_qx(qw/svn ls --non-interactive/, $var));
1047        my $old = $GIT_SVN;
1048        defined(my $pid = fork) or croak $!;
1049        if (!$pid) {
1050                foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
1051                        $u =~ s#/+$##;
1052                        if ($u !~ m!\Q$var\E/(.+)$!) {
1053                                print STDERR "W: Unrecognized URL: $u\n";
1054                                die "This should never happen\n";
1055                        }
1056                        # don't try to init already existing refs
1057                        my $id = $pfx.$1;
1058                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
1059                        init_vars();
1060                        unless (-d $GIT_SVN_DIR) {
1061                                print "init $u => $id\n";
1062                                init($u);
1063                        }
1064                }
1065                exit 0;
1066        }
1067        waitpid $pid, 0;
1068        croak $? if $?;
1069        my ($n) = ($switch =~ /^--(\w+)/);
1070        command_noisy('repo-config', "svn.$n", $var);
1071}
1072
1073sub common_prefix {
1074        my $paths = shift;
1075        my %common;
1076        foreach (@$paths) {
1077                my @tmp = split m#/#, $_;
1078                my $p = '';
1079                while (my $x = shift @tmp) {
1080                        $p .= "/$x";
1081                        $common{$p} ||= 0;
1082                        $common{$p}++;
1083                }
1084        }
1085        foreach (sort {length $b <=> length $a} keys %common) {
1086                if ($common{$_} == @$paths) {
1087                        return $_;
1088                }
1089        }
1090        return '';
1091}
1092
1093# grafts set here are 'stronger' in that they're based on actual tree
1094# matches, and won't be deleted from merge-base checking in write_grafts()
1095sub graft_tree_joins {
1096        my $grafts = shift;
1097        map_tree_joins() if (@_branch_from && !%tree_map);
1098        return unless %tree_map;
1099
1100        git_svn_each(sub {
1101                my $i = shift;
1102                my @args = (qw/rev-list --pretty=raw/, "refs/remotes/$i");
1103                my ($fh, $ctx) = command_output_pipe(@args);
1104                while (<$fh>) {
1105                        next unless /^commit ($sha1)$/o;
1106                        my $c = $1;
1107                        my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
1108                        next unless $tree_map{$t};
1109
1110                        my $l;
1111                        do {
1112                                $l = readline $fh;
1113                        } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
1114
1115                        my ($s, $tz) = ($1, $2);
1116                        if ($tz =~ s/^\+//) {
1117                                $s += tz_to_s_offset($tz);
1118                        } elsif ($tz =~ s/^\-//) {
1119                                $s -= tz_to_s_offset($tz);
1120                        }
1121
1122                        my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
1123
1124                        foreach my $p (@{$tree_map{$t}}) {
1125                                next if $p eq $c;
1126                                my $mb = eval { command('merge-base', $c, $p) };
1127                                next unless ($@ || $?);
1128                                if (defined $r_a) {
1129                                        # see if SVN says it's a relative
1130                                        my ($url_b, $r_b, $uuid_b) =
1131                                                        cmt_metadata($p);
1132                                        next if (defined $url_b &&
1133                                                        defined $url_a &&
1134                                                        ($url_a eq $url_b) &&
1135                                                        ($uuid_a eq $uuid_b));
1136                                        if ($uuid_a eq $uuid_b) {
1137                                                if ($r_b < $r_a) {
1138                                                        $grafts->{$c}->{$p} = 2;
1139                                                        next;
1140                                                } elsif ($r_b > $r_a) {
1141                                                        $grafts->{$p}->{$c} = 2;
1142                                                        next;
1143                                                }
1144                                        }
1145                                }
1146                                my $ct = get_commit_time($p);
1147                                if ($ct < $s) {
1148                                        $grafts->{$c}->{$p} = 2;
1149                                } elsif ($ct > $s) {
1150                                        $grafts->{$p}->{$c} = 2;
1151                                }
1152                                # what should we do when $ct == $s ?
1153                        }
1154                }
1155                command_close_pipe($fh, $ctx);
1156        });
1157}
1158
1159# this isn't funky-filename safe, but good enough for now...
1160sub graft_file_copy_cmd {
1161        my ($grafts, $l_map, $u) = @_;
1162        my $paths = $l_map->{$u};
1163        my $pfx = common_prefix([keys %$paths]);
1164        $SVN_URL ||= $u.$pfx;
1165        my $pid = open my $fh, '-|';
1166        defined $pid or croak $!;
1167        unless ($pid) {
1168                my @exec = qw/svn log -v/;
1169                push @exec, "-r$_revision" if defined $_revision;
1170                exec @exec, $u.$pfx or croak $!;
1171        }
1172        my ($r, $mp) = (undef, undef);
1173        while (<$fh>) {
1174                chomp;
1175                if (/^\-{72}$/) {
1176                        $mp = $r = undef;
1177                } elsif (/^r(\d+) \| /) {
1178                        $r = $1 unless defined $r;
1179                } elsif (/^Changed paths:/) {
1180                        $mp = 1;
1181                } elsif ($mp && m#^   [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) {
1182                        my ($p1, $p0, $r0) = ($1, $2, $3);
1183                        my $c = find_graft_path_commit($paths, $p1, $r);
1184                        next unless $c;
1185                        find_graft_path_parents($grafts, $paths, $c, $p0, $r0);
1186                }
1187        }
1188}
1189
1190sub graft_file_copy_lib {
1191        my ($grafts, $l_map, $u) = @_;
1192        my $tree_paths = $l_map->{$u};
1193        my $pfx = common_prefix([keys %$tree_paths]);
1194        my ($repo, $path) = repo_path_split($u.$pfx);
1195        $SVN = libsvn_connect($repo);
1196
1197        my ($base, $head) = libsvn_parse_revision();
1198        my $inc = 1000;
1199        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
1200        my $eh = $SVN::Error::handler;
1201        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
1202        while (1) {
1203                my $pool = SVN::Pool->new;
1204                libsvn_get_log(libsvn_dup_ra($SVN), [$path],
1205                               $min, $max, 0, 2, 1,
1206                        sub {
1207                                libsvn_graft_file_copies($grafts, $tree_paths,
1208                                                        $path, @_);
1209                        }, $pool);
1210                $pool->clear;
1211                last if ($max >= $head);
1212                $min = $max + 1;
1213                $max += $inc;
1214                $max = $head if ($max > $head);
1215        }
1216        $SVN::Error::handler = $eh;
1217}
1218
1219sub process_merge_msg_matches {
1220        my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
1221        my (@strong, @weak);
1222        foreach (@matches) {
1223                # merging with ourselves is not interesting
1224                next if $_ eq $p;
1225                if ($l_map->{$u}->{$_}) {
1226                        push @strong, $_;
1227                } else {
1228                        push @weak, $_;
1229                }
1230        }
1231        foreach my $w (@weak) {
1232                last if @strong;
1233                # no exact match, use branch name as regexp.
1234                my $re = qr/\Q$w\E/i;
1235                foreach (keys %{$l_map->{$u}}) {
1236                        if (/$re/) {
1237                                push @strong, $l_map->{$u}->{$_};
1238                                last;
1239                        }
1240                }
1241                last if @strong;
1242                $w = basename($w);
1243                $re = qr/\Q$w\E/i;
1244                foreach (keys %{$l_map->{$u}}) {
1245                        if (/$re/) {
1246                                push @strong, $l_map->{$u}->{$_};
1247                                last;
1248                        }
1249                }
1250        }
1251        my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
1252                                        \s(?:[a-f\d\-]+)$/xsm);
1253        unless (defined $rev) {
1254                ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
1255                                        \@(?:[a-f\d\-]+)/xsm);
1256                return unless defined $rev;
1257        }
1258        foreach my $m (@strong) {
1259                my ($r0, $s0) = find_rev_before($rev, $m, 1);
1260                $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
1261        }
1262}
1263
1264sub graft_merge_msg {
1265        my ($grafts, $l_map, $u, $p, @re) = @_;
1266
1267        my $x = $l_map->{$u}->{$p};
1268        my $rl = rev_list_raw($x);
1269        while (my $c = next_rev_list_entry($rl)) {
1270                foreach my $re (@re) {
1271                        my (@br) = ($c->{m} =~ /$re/g);
1272                        next unless @br;
1273                        process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
1274                }
1275        }
1276}
1277
1278sub read_uuid {
1279        return if $SVN_UUID;
1280        if ($_use_lib) {
1281                my $pool = SVN::Pool->new;
1282                $SVN_UUID = $SVN->get_uuid($pool);
1283                $pool->clear;
1284        } else {
1285                my $info = shift || svn_info('.');
1286                $SVN_UUID = $info->{'Repository UUID'} or
1287                                        croak "Repository UUID unreadable\n";
1288        }
1289}
1290
1291sub verify_ref {
1292        my ($ref) = @_;
1293        eval { command_oneline([ 'rev-parse', $ref ], { STDERR => 0 }) };
1294}
1295
1296sub quiet_run {
1297        my $pid = fork;
1298        defined $pid or croak $!;
1299        if (!$pid) {
1300                open my $null, '>', '/dev/null' or croak $!;
1301                open STDERR, '>&', $null or croak $!;
1302                open STDOUT, '>&', $null or croak $!;
1303                exec @_ or croak $!;
1304        }
1305        waitpid $pid, 0;
1306        return $?;
1307}
1308
1309sub repo_path_split {
1310        my $full_url = shift;
1311        $full_url =~ s#/+$##;
1312
1313        foreach (@repo_path_split_cache) {
1314                if ($full_url =~ s#$_##) {
1315                        my $u = $1;
1316                        $full_url =~ s#^/+##;
1317                        return ($u, $full_url);
1318                }
1319        }
1320        if ($_use_lib) {
1321                my $tmp = libsvn_connect($full_url);
1322                return ($tmp->{repos_root}, $tmp->{svn_path});
1323        } else {
1324                my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i);
1325                $path =~ s#^/+##;
1326                my @paths = split(m#/+#, $path);
1327                while (quiet_run(qw/svn ls --non-interactive/, $url)) {
1328                        my $n = shift @paths || last;
1329                        $url .= "/$n";
1330                }
1331                push @repo_path_split_cache, qr/^(\Q$url\E)/;
1332                $path = join('/',@paths);
1333                return ($url, $path);
1334        }
1335}
1336
1337sub setup_git_svn {
1338        defined $SVN_URL or croak "SVN repository location required\n";
1339        unless (-d $GIT_DIR) {
1340                croak "GIT_DIR=$GIT_DIR does not exist!\n";
1341        }
1342        mkpath([$GIT_SVN_DIR]);
1343        mkpath(["$GIT_SVN_DIR/info"]);
1344        open my $fh, '>>',$REVDB or croak $!;
1345        close $fh;
1346        s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1347
1348}
1349
1350sub assert_svn_wc_clean {
1351        return if $_use_lib;
1352        my ($svn_rev) = @_;
1353        croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
1354        my $lcr = svn_info('.')->{'Last Changed Rev'};
1355        if ($svn_rev != $lcr) {
1356                print STDERR "Checking for copy-tree ... ";
1357                my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
1358                                                "-r$lcr:$svn_rev")));
1359                if (@diff) {
1360                        croak "Nope!  Expected r$svn_rev, got r$lcr\n";
1361                } else {
1362                        print STDERR "OK!\n";
1363                }
1364        }
1365        my @status = grep(!/^Performing status on external/,(`svn status`));
1366        @status = grep(!/^\s*$/,@status);
1367        @status = grep(!/^X/,@status) if $_no_ignore_ext;
1368        if (scalar @status) {
1369                print STDERR "Tree ($SVN_WC) is not clean:\n";
1370                print STDERR $_ foreach @status;
1371                croak;
1372        }
1373}
1374
1375sub get_tree_from_treeish {
1376        my ($treeish) = @_;
1377        croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1378        my $type = command_oneline(qw/cat-file -t/, $treeish);
1379        my $expected;
1380        while ($type eq 'tag') {
1381                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1382        }
1383        if ($type eq 'commit') {
1384                $expected = (grep /^tree /, command(qw/cat-file commit/,
1385                                                    $treeish))[0];
1386                ($expected) = ($expected =~ /^tree ($sha1)$/);
1387                die "Unable to get tree from $treeish\n" unless $expected;
1388        } elsif ($type eq 'tree') {
1389                $expected = $treeish;
1390        } else {
1391                die "$treeish is a $type, expected tree, tag or commit\n";
1392        }
1393        return $expected;
1394}
1395
1396sub assert_tree {
1397        return if $_use_lib;
1398        my ($treeish) = @_;
1399        my $expected = get_tree_from_treeish($treeish);
1400
1401        my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
1402        if (-e $tmpindex) {
1403                unlink $tmpindex or croak $!;
1404        }
1405        my $old_index = set_index($tmpindex);
1406        index_changes(1);
1407        my $tree = command_oneline('write-tree');
1408        restore_index($old_index);
1409        if ($tree ne $expected) {
1410                croak "Tree mismatch, Got: $tree, Expected: $expected\n";
1411        }
1412        unlink $tmpindex;
1413}
1414
1415sub get_diff {
1416        my ($from, $treeish) = @_;
1417        assert_tree($from);
1418        print "diff-tree $from $treeish\n";
1419        my @diff_tree = qw(diff-tree -z -r);
1420        if ($_cp_similarity) {
1421                push @diff_tree, "-C$_cp_similarity";
1422        } else {
1423                push @diff_tree, '-C';
1424        }
1425        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1426        push @diff_tree, "-l$_l" if defined $_l;
1427        push @diff_tree, $from, $treeish;
1428        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
1429        local $/ = "\0";
1430        my $state = 'meta';
1431        my @mods;
1432        while (<$diff_fh>) {
1433                chomp $_; # this gets rid of the trailing "\0"
1434                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1435                                        $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1436                        push @mods, {   mode_a => $1, mode_b => $2,
1437                                        sha1_b => $3, chg => $4 };
1438                        if ($4 =~ /^(?:C|R)$/) {
1439                                $state = 'file_a';
1440                        } else {
1441                                $state = 'file_b';
1442                        }
1443                } elsif ($state eq 'file_a') {
1444                        my $x = $mods[$#mods] or croak "Empty array\n";
1445                        if ($x->{chg} !~ /^(?:C|R)$/) {
1446                                croak "Error parsing $_, $x->{chg}\n";
1447                        }
1448                        $x->{file_a} = $_;
1449                        $state = 'file_b';
1450                } elsif ($state eq 'file_b') {
1451                        my $x = $mods[$#mods] or croak "Empty array\n";
1452                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1453                                croak "Error parsing $_, $x->{chg}\n";
1454                        }
1455                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1456                                croak "Error parsing $_, $x->{chg}\n";
1457                        }
1458                        $x->{file_b} = $_;
1459                        $state = 'meta';
1460                } else {
1461                        croak "Error parsing $_\n";
1462                }
1463        }
1464        command_close_pipe($diff_fh, $ctx);
1465        return \@mods;
1466}
1467
1468sub svn_check_prop_executable {
1469        my $m = shift;
1470        return if -l $m->{file_b};
1471        if ($m->{mode_b} =~ /755$/) {
1472                chmod((0755 &~ umask),$m->{file_b}) or croak $!;
1473                if ($m->{mode_a} !~ /755$/) {
1474                        sys(qw(svn propset svn:executable 1), $m->{file_b});
1475                }
1476                -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
1477        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
1478                sys(qw(svn propdel svn:executable), $m->{file_b});
1479                chmod((0644 &~ umask),$m->{file_b}) or croak $!;
1480                -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
1481        }
1482}
1483
1484sub svn_ensure_parent_path {
1485        my $dir_b = dirname(shift);
1486        svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
1487        mkpath([$dir_b]) unless (-d $dir_b);
1488        sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
1489}
1490
1491sub precommit_check {
1492        my $mods = shift;
1493        my (%rm_file, %rmdir_check, %added_check);
1494
1495        my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
1496        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1497                if ($m->{chg} eq 'R') {
1498                        if (-d $m->{file_b}) {
1499                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1500                        }
1501                        # dir/$file => dir/file/$file
1502                        my $dirname = dirname($m->{file_b});
1503                        while ($dirname ne File::Spec->curdir) {
1504                                if ($dirname ne $m->{file_a}) {
1505                                        $dirname = dirname($dirname);
1506                                        next;
1507                                }
1508                                err_file_to_dir("$m->{file_a} => $m->{file_b}");
1509                        }
1510                        # baz/zzz => baz (baz is a file)
1511                        $dirname = dirname($m->{file_a});
1512                        while ($dirname ne File::Spec->curdir) {
1513                                if ($dirname ne $m->{file_b}) {
1514                                        $dirname = dirname($dirname);
1515                                        next;
1516                                }
1517                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1518                        }
1519                }
1520                if ($m->{chg} =~ /^(D|R)$/) {
1521                        my $t = $1 eq 'D' ? 'file_b' : 'file_a';
1522                        $rm_file{ $m->{$t} } = 1;
1523                        my $dirname = dirname( $m->{$t} );
1524                        my $basename = basename( $m->{$t} );
1525                        $rmdir_check{$dirname}->{$basename} = 1;
1526                } elsif ($m->{chg} =~ /^(?:A|C)$/) {
1527                        if (-d $m->{file_b}) {
1528                                err_dir_to_file($m->{file_b});
1529                        }
1530                        my $dirname = dirname( $m->{file_b} );
1531                        my $basename = basename( $m->{file_b} );
1532                        $added_check{$dirname}->{$basename} = 1;
1533                        while ($dirname ne File::Spec->curdir) {
1534                                if ($rm_file{$dirname}) {
1535                                        err_file_to_dir($m->{file_b});
1536                                }
1537                                $dirname = dirname $dirname;
1538                        }
1539                }
1540        }
1541        return (\%rmdir_check, \%added_check);
1542
1543        sub err_dir_to_file {
1544                my $file = shift;
1545                print STDERR "Node change from directory to file ",
1546                                "is not supported by Subversion: ",$file,"\n";
1547                exit 1;
1548        }
1549        sub err_file_to_dir {
1550                my $file = shift;
1551                print STDERR "Node change from file to directory ",
1552                                "is not supported by Subversion: ",$file,"\n";
1553                exit 1;
1554        }
1555}
1556
1557
1558sub svn_checkout_tree {
1559        my ($from, $treeish) = @_;
1560        my $mods = get_diff($from->{commit}, $treeish);
1561        return $mods unless (scalar @$mods);
1562        my ($rm, $add) = precommit_check($mods);
1563
1564        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1565        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1566                if ($m->{chg} eq 'C') {
1567                        svn_ensure_parent_path( $m->{file_b} );
1568                        sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
1569                        apply_mod_line_blob($m);
1570                        svn_check_prop_executable($m);
1571                } elsif ($m->{chg} eq 'D') {
1572                        sys(qw(svn rm --force), $m->{file_b});
1573                } elsif ($m->{chg} eq 'R') {
1574                        svn_ensure_parent_path( $m->{file_b} );
1575                        sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
1576                        apply_mod_line_blob($m);
1577                        svn_check_prop_executable($m);
1578                } elsif ($m->{chg} eq 'M') {
1579                        apply_mod_line_blob($m);
1580                        svn_check_prop_executable($m);
1581                } elsif ($m->{chg} eq 'T') {
1582                        svn_check_prop_executable($m);
1583                        apply_mod_line_blob($m);
1584                        if ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
1585                                sys(qw(svn propdel svn:special), $m->{file_b});
1586                        } else {
1587                                sys(qw(svn propset svn:special *),$m->{file_b});
1588                        }
1589                } elsif ($m->{chg} eq 'A') {
1590                        svn_ensure_parent_path( $m->{file_b} );
1591                        apply_mod_line_blob($m);
1592                        sys(qw(svn add), $m->{file_b});
1593                        svn_check_prop_executable($m);
1594                } else {
1595                        croak "Invalid chg: $m->{chg}\n";
1596                }
1597        }
1598
1599        assert_tree($treeish);
1600        if ($_rmdir) { # remove empty directories
1601                handle_rmdir($rm, $add);
1602        }
1603        assert_tree($treeish);
1604        return $mods;
1605}
1606
1607sub libsvn_checkout_tree {
1608        my ($from, $treeish, $ed) = @_;
1609        my $mods = get_diff($from, $treeish);
1610        return $mods unless (scalar @$mods);
1611        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1612        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1613                my $f = $m->{chg};
1614                if (defined $o{$f}) {
1615                        $ed->$f($m, $_q);
1616                } else {
1617                        croak "Invalid change type: $f\n";
1618                }
1619        }
1620        $ed->rmdirs($_q) if $_rmdir;
1621        return $mods;
1622}
1623
1624# svn ls doesn't work with respect to the current working tree, but what's
1625# in the repository.  There's not even an option for it... *sigh*
1626# (added files don't show up and removed files remain in the ls listing)
1627sub svn_ls_current {
1628        my ($dir, $rm, $add) = @_;
1629        chomp(my @ls = safe_qx('svn','ls',$dir));
1630        my @ret = ();
1631        foreach (@ls) {
1632                s#/$##; # trailing slashes are evil
1633                push @ret, $_ unless $rm->{$dir}->{$_};
1634        }
1635        if (exists $add->{$dir}) {
1636                push @ret, keys %{$add->{$dir}};
1637        }
1638        return \@ret;
1639}
1640
1641sub handle_rmdir {
1642        my ($rm, $add) = @_;
1643
1644        foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
1645                my $ls = svn_ls_current($dir, $rm, $add);
1646                next if (scalar @$ls);
1647                sys(qw(svn rm --force),$dir);
1648
1649                my $dn = dirname $dir;
1650                $rm->{ $dn }->{ basename $dir } = 1;
1651                $ls = svn_ls_current($dn, $rm, $add);
1652                while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
1653                        sys(qw(svn rm --force),$dn);
1654                        $dir = basename $dn;
1655                        $dn = dirname $dn;
1656                        $rm->{ $dn }->{ $dir } = 1;
1657                        $ls = svn_ls_current($dn, $rm, $add);
1658                }
1659        }
1660}
1661
1662sub get_commit_message {
1663        my ($commit, $commit_msg) = (@_);
1664        my %log_msg = ( msg => '' );
1665        open my $msg, '>', $commit_msg or croak $!;
1666
1667        my $type = command_oneline(qw/cat-file -t/, $commit);
1668        if ($type eq 'commit' || $type eq 'tag') {
1669                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1670                                                         $type, $commit);
1671                my $in_msg = 0;
1672                while (<$msg_fh>) {
1673                        if (!$in_msg) {
1674                                $in_msg = 1 if (/^\s*$/);
1675                        } elsif (/^git-svn-id: /) {
1676                                # skip this, we regenerate the correct one
1677                                # on re-fetch anyways
1678                        } else {
1679                                print $msg $_ or croak $!;
1680                        }
1681                }
1682                command_close_pipe($msg_fh, $ctx);
1683        }
1684        close $msg or croak $!;
1685
1686        if ($_edit || ($type eq 'tree')) {
1687                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1688                system($editor, $commit_msg);
1689        }
1690
1691        # file_to_s removes all trailing newlines, so just use chomp() here:
1692        open $msg, '<', $commit_msg or croak $!;
1693        { local $/; chomp($log_msg{msg} = <$msg>); }
1694        close $msg or croak $!;
1695
1696        return \%log_msg;
1697}
1698
1699sub set_svn_commit_env {
1700        if (defined $LC_ALL) {
1701                $ENV{LC_ALL} = $LC_ALL;
1702        } else {
1703                delete $ENV{LC_ALL};
1704        }
1705}
1706
1707sub svn_commit_tree {
1708        my ($last, $commit) = @_;
1709        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
1710        my $log_msg = get_commit_message($commit, $commit_msg);
1711        my ($oneline) = ($log_msg->{msg} =~ /([^\n\r]+)/);
1712        print "Committing $commit: $oneline\n";
1713
1714        set_svn_commit_env();
1715        my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
1716        $ENV{LC_ALL} = 'C';
1717        unlink $commit_msg;
1718        my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/);
1719        if (!defined $committed) {
1720                my $out = join("\n",@ci_output);
1721                print STDERR "W: Trouble parsing \`svn commit' output:\n\n",
1722                                $out, "\n\nAssuming English locale...";
1723                ($committed) = ($out =~ /^Committed revision \d+\./sm);
1724                defined $committed or die " FAILED!\n",
1725                        "Commit output failed to parse committed revision!\n",
1726                print STDERR " OK\n";
1727        }
1728
1729        my @svn_up = qw(svn up);
1730        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
1731        if ($_optimize_commits && ($committed == ($last->{revision} + 1))) {
1732                push @svn_up, "-r$committed";
1733                sys(@svn_up);
1734                my $info = svn_info('.');
1735                my $date = $info->{'Last Changed Date'} or die "Missing date\n";
1736                if ($info->{'Last Changed Rev'} != $committed) {
1737                        croak "$info->{'Last Changed Rev'} != $committed\n"
1738                }
1739                my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1740                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1741                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1742                                         or croak "Failed to parse date: $date\n";
1743                $log_msg->{date} = "$tz $Y-$m-$d $H:$M:$S";
1744                $log_msg->{author} = $info->{'Last Changed Author'};
1745                $log_msg->{revision} = $committed;
1746                $log_msg->{msg} .= "\n";
1747                $log_msg->{parents} = [ $last->{commit} ];
1748                $log_msg->{commit} = git_commit($log_msg, $commit);
1749                return $log_msg;
1750        }
1751        # resync immediately
1752        push @svn_up, "-r$last->{revision}";
1753        sys(@svn_up);
1754        return fetch("$committed=$commit");
1755}
1756
1757sub rev_list_raw {
1758        my ($fh, $c) = command_output_pipe(qw/rev-list --pretty=raw/, @_);
1759        return { fh => $fh, ctx => $c, t => { } };
1760}
1761
1762sub next_rev_list_entry {
1763        my $rl = shift;
1764        my $fh = $rl->{fh};
1765        my $x = $rl->{t};
1766        while (<$fh>) {
1767                if (/^commit ($sha1)$/o) {
1768                        if ($x->{c}) {
1769                                $rl->{t} = { c => $1 };
1770                                return $x;
1771                        } else {
1772                                $x->{c} = $1;
1773                        }
1774                } elsif (/^parent ($sha1)$/o) {
1775                        $x->{p}->{$1} = 1;
1776                } elsif (s/^    //) {
1777                        $x->{m} ||= '';
1778                        $x->{m} .= $_;
1779                }
1780        }
1781        command_close_pipe($fh, $rl->{ctx});
1782        return ($x != $rl->{t}) ? $x : undef;
1783}
1784
1785# read the entire log into a temporary file (which is removed ASAP)
1786# and store the file handle + parser state
1787sub svn_log_raw {
1788        my (@log_args) = @_;
1789        my $log_fh = IO::File->new_tmpfile or croak $!;
1790        my $pid = fork;
1791        defined $pid or croak $!;
1792        if (!$pid) {
1793                open STDOUT, '>&', $log_fh or croak $!;
1794                exec (qw(svn log), @log_args) or croak $!
1795        }
1796        waitpid $pid, 0;
1797        croak $? if $?;
1798        seek $log_fh, 0, 0 or croak $!;
1799        return { state => 'sep', fh => $log_fh };
1800}
1801
1802sub next_log_entry {
1803        my $log = shift; # retval of svn_log_raw()
1804        my $ret = undef;
1805        my $fh = $log->{fh};
1806
1807        while (<$fh>) {
1808                chomp;
1809                if (/^\-{72}$/) {
1810                        if ($log->{state} eq 'msg') {
1811                                if ($ret->{lines}) {
1812                                        $ret->{msg} .= $_."\n";
1813                                        unless(--$ret->{lines}) {
1814                                                $log->{state} = 'sep';
1815                                        }
1816                                } else {
1817                                        croak "Log parse error at: $_\n",
1818                                                $ret->{revision},
1819                                                "\n";
1820                                }
1821                                next;
1822                        }
1823                        if ($log->{state} ne 'sep') {
1824                                croak "Log parse error at: $_\n",
1825                                        "state: $log->{state}\n",
1826                                        $ret->{revision},
1827                                        "\n";
1828                        }
1829                        $log->{state} = 'rev';
1830
1831                        # if we have an empty log message, put something there:
1832                        if ($ret) {
1833                                $ret->{msg} ||= "\n";
1834                                delete $ret->{lines};
1835                                return $ret;
1836                        }
1837                        next;
1838                }
1839                if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
1840                        my $rev = $1;
1841                        my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
1842                        ($lines) = ($lines =~ /(\d+)/);
1843                        $date = '1970-01-01 00:00:00 +0000'
1844                                if ($_ignore_nodate && $date eq '(no date)');
1845                        my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1846                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1847                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1848                                         or croak "Failed to parse date: $date\n";
1849                        $ret = {        revision => $rev,
1850                                        date => "$tz $Y-$m-$d $H:$M:$S",
1851                                        author => $author,
1852                                        lines => $lines,
1853                                        msg => '' };
1854                        if (defined $_authors && ! defined $users{$author}) {
1855                                die "Author: $author not defined in ",
1856                                                "$_authors file\n";
1857                        }
1858                        $log->{state} = 'msg_start';
1859                        next;
1860                }
1861                # skip the first blank line of the message:
1862                if ($log->{state} eq 'msg_start' && /^$/) {
1863                        $log->{state} = 'msg';
1864                } elsif ($log->{state} eq 'msg') {
1865                        if ($ret->{lines}) {
1866                                $ret->{msg} .= $_."\n";
1867                                unless (--$ret->{lines}) {
1868                                        $log->{state} = 'sep';
1869                                }
1870                        } else {
1871                                croak "Log parse error at: $_\n",
1872                                        $ret->{revision},"\n";
1873                        }
1874                }
1875        }
1876        return $ret;
1877}
1878
1879sub svn_info {
1880        my $url = shift || $SVN_URL;
1881
1882        my $pid = open my $info_fh, '-|';
1883        defined $pid or croak $!;
1884
1885        if ($pid == 0) {
1886                exec(qw(svn info),$url) or croak $!;
1887        }
1888
1889        my $ret = {};
1890        # only single-lines seem to exist in svn info output
1891        while (<$info_fh>) {
1892                chomp $_;
1893                if (m#^([^:]+)\s*:\s*(\S.*)$#) {
1894                        $ret->{$1} = $2;
1895                        push @{$ret->{-order}}, $1;
1896                }
1897        }
1898        close $info_fh or croak $?;
1899        return $ret;
1900}
1901
1902sub sys { system(@_) == 0 or croak $? }
1903
1904sub do_update_index {
1905        my ($z_cmd, $cmd, $no_text_base) = @_;
1906
1907        my ($p, $pctx) = command_output_pipe(@$z_cmd);
1908
1909        my ($ui, $uctx) = command_input_pipe('update-index',
1910                                             "--$cmd",'-z','--stdin');
1911        local $/ = "\0";
1912        while (my $x = <$p>) {
1913                chomp $x;
1914                if (!$no_text_base && lstat $x && ! -l _ &&
1915                                svn_propget_base('svn:keywords', $x)) {
1916                        my $mode = -x _ ? 0755 : 0644;
1917                        my ($v,$d,$f) = File::Spec->splitpath($x);
1918                        my $tb = File::Spec->catfile($d, '.svn', 'tmp',
1919                                                'text-base',"$f.svn-base");
1920                        $tb =~ s#^/##;
1921                        unless (-f $tb) {
1922                                $tb = File::Spec->catfile($d, '.svn',
1923                                                'text-base',"$f.svn-base");
1924                                $tb =~ s#^/##;
1925                        }
1926                        my @s = stat($x);
1927                        unlink $x or croak $!;
1928                        copy($tb, $x);
1929                        chmod(($mode &~ umask), $x) or croak $!;
1930                        utime $s[8], $s[9], $x;
1931                }
1932                print $ui $x,"\0";
1933        }
1934        command_close_pipe($p, $pctx);
1935        command_close_pipe($ui, $uctx);
1936}
1937
1938sub index_changes {
1939        return if $_use_lib;
1940
1941        if (!-f "$GIT_SVN_DIR/info/exclude") {
1942                open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!;
1943                print $fd '.svn',"\n";
1944                close $fd or croak $!;
1945        }
1946        my $no_text_base = shift;
1947        do_update_index([qw/diff-files --name-only -z/],
1948                        'remove',
1949                        $no_text_base);
1950        do_update_index([qw/ls-files -z --others/,
1951                                "--exclude-from=$GIT_SVN_DIR/info/exclude"],
1952                        'add',
1953                        $no_text_base);
1954}
1955
1956sub s_to_file {
1957        my ($str, $file, $mode) = @_;
1958        open my $fd,'>',$file or croak $!;
1959        print $fd $str,"\n" or croak $!;
1960        close $fd or croak $!;
1961        chmod ($mode &~ umask, $file) if (defined $mode);
1962}
1963
1964sub file_to_s {
1965        my $file = shift;
1966        open my $fd,'<',$file or croak "$!: file: $file\n";
1967        local $/;
1968        my $ret = <$fd>;
1969        close $fd or croak $!;
1970        $ret =~ s/\s*$//s;
1971        return $ret;
1972}
1973
1974sub assert_revision_unknown {
1975        my $r = shift;
1976        if (my $c = revdb_get($REVDB, $r)) {
1977                croak "$r = $c already exists! Why are we refetching it?";
1978        }
1979}
1980
1981sub trees_eq {
1982        my ($x, $y) = @_;
1983        my @x = command(qw/cat-file commit/,$x);
1984        my @y = command(qw/cat-file commit/,$y);
1985        if (($y[0] ne $x[0]) || $x[0] ne "tree $sha1"
1986                             || $y[0] ne "tree $sha1") {
1987                print STDERR "Trees not equal: $y[0] != $x[0]\n";
1988                return 0
1989        }
1990        return 1;
1991}
1992
1993sub git_commit {
1994        my ($log_msg, @parents) = @_;
1995        assert_revision_unknown($log_msg->{revision});
1996        map_tree_joins() if (@_branch_from && !%tree_map);
1997
1998        my (@tmp_parents, @exec_parents, %seen_parent);
1999        if (my $lparents = $log_msg->{parents}) {
2000                @tmp_parents = @$lparents
2001        }
2002        # commit parents can be conditionally bound to a particular
2003        # svn revision via: "svn_revno=commit_sha1", filter them out here:
2004        foreach my $p (@parents) {
2005                next unless defined $p;
2006                if ($p =~ /^(\d+)=($sha1_short)$/o) {
2007                        if ($1 == $log_msg->{revision}) {
2008                                push @tmp_parents, $2;
2009                        }
2010                } else {
2011                        push @tmp_parents, $p if $p =~ /$sha1_short/o;
2012                }
2013        }
2014        my $tree = $log_msg->{tree};
2015        if (!defined $tree) {
2016                my $index = set_index($GIT_SVN_INDEX);
2017                index_changes();
2018                $tree = command_oneline('write-tree');
2019                croak $? if $?;
2020                restore_index($index);
2021        }
2022
2023        # just in case we clobber the existing ref, we still want that ref
2024        # as our parent:
2025        if (my $cur = verify_ref("refs/remotes/$GIT_SVN^0")) {
2026                chomp $cur;
2027                push @tmp_parents, $cur;
2028        }
2029
2030        if (exists $tree_map{$tree}) {
2031                foreach my $p (@{$tree_map{$tree}}) {
2032                        my $skip;
2033                        foreach (@tmp_parents) {
2034                                # see if a common parent is found
2035                                my $mb = eval { command('merge-base', $_, $p) };
2036                                next if ($@ || $?);
2037                                $skip = 1;
2038                                last;
2039                        }
2040                        next if $skip;
2041                        my ($url_p, $r_p, $uuid_p) = cmt_metadata($p);
2042                        next if (($SVN_UUID eq $uuid_p) &&
2043                                                ($log_msg->{revision} > $r_p));
2044                        next if (defined $url_p && defined $SVN_URL &&
2045                                                ($SVN_UUID eq $uuid_p) &&
2046                                                ($url_p eq $SVN_URL));
2047                        push @tmp_parents, $p;
2048                }
2049        }
2050        foreach (@tmp_parents) {
2051                next if $seen_parent{$_};
2052                $seen_parent{$_} = 1;
2053                push @exec_parents, $_;
2054                # MAXPARENT is defined to 16 in commit-tree.c:
2055                last if @exec_parents > 16;
2056        }
2057
2058        set_commit_env($log_msg);
2059        my @exec = ('git-commit-tree', $tree);
2060        push @exec, '-p', $_  foreach @exec_parents;
2061        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2062                                                                or croak $!;
2063        print $msg_fh $log_msg->{msg} or croak $!;
2064        unless ($_no_metadata) {
2065                print $msg_fh "\ngit-svn-id: $SVN_URL\@$log_msg->{revision}",
2066                                        " $SVN_UUID\n" or croak $!;
2067        }
2068        $msg_fh->flush == 0 or croak $!;
2069        close $msg_fh or croak $!;
2070        chomp(my $commit = do { local $/; <$out_fh> });
2071        close $out_fh or croak $!;
2072        waitpid $pid, 0;
2073        croak $? if $?;
2074        if ($commit !~ /^$sha1$/o) {
2075                die "Failed to commit, invalid sha1: $commit\n";
2076        }
2077        command_noisy('update-ref',"refs/remotes/$GIT_SVN",$commit);
2078        revdb_set($REVDB, $log_msg->{revision}, $commit);
2079
2080        # this output is read via pipe, do not change:
2081        print "r$log_msg->{revision} = $commit\n";
2082        check_repack();
2083        return $commit;
2084}
2085
2086sub check_repack {
2087        if ($_repack && (--$_repack_nr == 0)) {
2088                $_repack_nr = $_repack;
2089                # repack doesn't use any arguments with spaces in them, does it?
2090                command_noisy('repack', split(/\s+/, $_repack_flags));
2091        }
2092}
2093
2094sub set_commit_env {
2095        my ($log_msg) = @_;
2096        my $author = $log_msg->{author};
2097        if (!defined $author || length $author == 0) {
2098                $author = '(no author)';
2099        }
2100        my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
2101                                : ($author,"$author\@$SVN_UUID");
2102        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
2103        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
2104        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
2105}
2106
2107sub apply_mod_line_blob {
2108        my $m = shift;
2109        if ($m->{mode_b} =~ /^120/) {
2110                blob_to_symlink($m->{sha1_b}, $m->{file_b});
2111        } else {
2112                blob_to_file($m->{sha1_b}, $m->{file_b});
2113        }
2114}
2115
2116sub blob_to_symlink {
2117        my ($blob, $link) = @_;
2118        defined $link or croak "\$link not defined!\n";
2119        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2120        if (-l $link || -f _) {
2121                unlink $link or croak $!;
2122        }
2123
2124        my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
2125        symlink $dest, $link or croak $!;
2126}
2127
2128sub blob_to_file {
2129        my ($blob, $file) = @_;
2130        defined $file or croak "\$file not defined!\n";
2131        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2132        if (-l $file || -f _) {
2133                unlink $file or croak $!;
2134        }
2135
2136        open my $blob_fh, '>', $file or croak "$!: $file\n";
2137        my $pid = fork;
2138        defined $pid or croak $!;
2139
2140        if ($pid == 0) {
2141                open STDOUT, '>&', $blob_fh or croak $!;
2142                exec('git-cat-file','blob',$blob) or croak $!;
2143        }
2144        waitpid $pid, 0;
2145        croak $? if $?;
2146
2147        close $blob_fh or croak $!;
2148}
2149
2150sub safe_qx {
2151        my $pid = open my $child, '-|';
2152        defined $pid or croak $!;
2153        if ($pid == 0) {
2154                exec(@_) or croak $!;
2155        }
2156        my @ret = (<$child>);
2157        close $child or croak $?;
2158        die $? if $?; # just in case close didn't error out
2159        return wantarray ? @ret : join('',@ret);
2160}
2161
2162sub svn_compat_check {
2163        if ($_follow_parent) {
2164                print STDERR 'E: --follow-parent functionality is only ',
2165                                "available when SVN libraries are used\n";
2166                exit 1;
2167        }
2168        my @co_help = safe_qx(qw(svn co -h));
2169        unless (grep /ignore-externals/,@co_help) {
2170                print STDERR "W: Installed svn version does not support ",
2171                                "--ignore-externals\n";
2172                $_no_ignore_ext = 1;
2173        }
2174        if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
2175                $_svn_co_url_revs = 1;
2176        }
2177        if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) {
2178                $_svn_pg_peg_revs = 1;
2179        }
2180
2181        # I really, really hope nobody hits this...
2182        unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
2183                print STDERR <<'';
2184W: The installed svn version does not support the --stop-on-copy flag in
2185   the log command.
2186   Lets hope the directory you're tracking is not a branch or tag
2187   and was never moved within the repository...
2188
2189                $_no_stop_copy = 1;
2190        }
2191}
2192
2193# *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
2194# (and they won't honor URL@<rev> without -r<rev>, too!)
2195sub svn_cmd_checkout {
2196        my ($url, $rev, $dir) = @_;
2197        my @cmd = ('svn','co', "-r$rev");
2198        push @cmd, '--ignore-externals' unless $_no_ignore_ext;
2199        $url .= "\@$rev" if $_svn_co_url_revs;
2200        sys(@cmd, $url, $dir);
2201}
2202
2203sub check_upgrade_needed {
2204        if (!-r $REVDB) {
2205                -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2206                open my $fh, '>>',$REVDB or croak $!;
2207                close $fh;
2208        }
2209        return unless eval {
2210                command([qw/rev-parse --verify/,"$GIT_SVN-HEAD^0"],
2211                        {STDERR => 0});
2212        };
2213        my $head = eval { command('rev-parse',"refs/remotes/$GIT_SVN") };
2214        if ($@ || !$head) {
2215                print STDERR "Please run: $0 rebuild --upgrade\n";
2216                exit 1;
2217        }
2218}
2219
2220# fills %tree_map with a reverse mapping of trees to commits.  Useful
2221# for finding parents to commit on.
2222sub map_tree_joins {
2223        my %seen;
2224        foreach my $br (@_branch_from) {
2225                my $pipe = command_output_pipe(qw/rev-list
2226                                            --topo-order --pretty=raw/, $br);
2227                while (<$pipe>) {
2228                        if (/^commit ($sha1)$/o) {
2229                                my $commit = $1;
2230
2231                                # if we've seen a commit,
2232                                # we've seen its parents
2233                                last if $seen{$commit};
2234                                my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
2235                                unless (defined $tree) {
2236                                        die "Failed to parse commit $commit\n";
2237                                }
2238                                push @{$tree_map{$tree}}, $commit;
2239                                $seen{$commit} = 1;
2240                        }
2241                }
2242                eval { command_close_pipe($pipe) };
2243        }
2244}
2245
2246sub load_all_refs {
2247        if (@_branch_from) {
2248                print STDERR '--branch|-b parameters are ignored when ',
2249                        "--branch-all-refs|-B is passed\n";
2250        }
2251
2252        # don't worry about rev-list on non-commit objects/tags,
2253        # it shouldn't blow up if a ref is a blob or tree...
2254        @_branch_from = command(qw/rev-parse --symbolic --all/);
2255}
2256
2257# '<svn username> = real-name <email address>' mapping based on git-svnimport:
2258sub load_authors {
2259        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2260        while (<$authors>) {
2261                chomp;
2262                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
2263                my ($user, $name, $email) = ($1, $2, $3);
2264                $users{$user} = [$name, $email];
2265        }
2266        close $authors or croak $!;
2267}
2268
2269sub rload_authors {
2270        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2271        while (<$authors>) {
2272                chomp;
2273                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2274                my ($user, $name, $email) = ($1, $2, $3);
2275                $rusers{"$name <$email>"} = $user;
2276        }
2277        close $authors or croak $!;
2278}
2279
2280sub svn_propget_base {
2281        my ($p, $f) = @_;
2282        $f .= '@BASE' if $_svn_pg_peg_revs;
2283        return safe_qx(qw/svn propget/, $p, $f);
2284}
2285
2286sub git_svn_each {
2287        my $sub = shift;
2288        foreach (command(qw/rev-parse --symbolic --all/)) {
2289                next unless s#^refs/remotes/##;
2290                chomp $_;
2291                next unless -f "$GIT_DIR/svn/$_/info/url";
2292                &$sub($_);
2293        }
2294}
2295
2296sub migrate_revdb {
2297        git_svn_each(sub {
2298                my $id = shift;
2299                defined(my $pid = fork) or croak $!;
2300                if (!$pid) {
2301                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2302                        init_vars();
2303                        exit 0 if -r $REVDB;
2304                        print "Upgrading svn => git mapping...\n";
2305                        -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2306                        open my $fh, '>>',$REVDB or croak $!;
2307                        close $fh;
2308                        rebuild();
2309                        print "Done upgrading. You may now delete the ",
2310                                "deprecated $GIT_SVN_DIR/revs directory\n";
2311                        exit 0;
2312                }
2313                waitpid $pid, 0;
2314                croak $? if $?;
2315        });
2316}
2317
2318sub migration_check {
2319        migrate_revdb() unless (-e $REVDB);
2320        return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
2321        print "Upgrading repository...\n";
2322        unless (-d "$GIT_DIR/svn") {
2323                mkdir "$GIT_DIR/svn" or croak $!;
2324        }
2325        print "Data from a previous version of git-svn exists, but\n\t",
2326                                "$GIT_SVN_DIR\n\t(required for this version ",
2327                                "($VERSION) of git-svn) does not.\n";
2328
2329        foreach my $x (command(qw/rev-parse --symbolic --all/)) {
2330                next unless $x =~ s#^refs/remotes/##;
2331                chomp $x;
2332                next unless -f "$GIT_DIR/$x/info/url";
2333                my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
2334                next unless $u;
2335                my $dn = dirname("$GIT_DIR/svn/$x");
2336                mkpath([$dn]) unless -d $dn;
2337                rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
2338        }
2339        migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
2340        print "Done upgrading.\n";
2341}
2342
2343sub find_rev_before {
2344        my ($r, $id, $eq_ok) = @_;
2345        my $f = "$GIT_DIR/svn/$id/.rev_db";
2346        return (undef,undef) unless -r $f;
2347        --$r unless $eq_ok;
2348        while ($r > 0) {
2349                if (my $c = revdb_get($f, $r)) {
2350                        return ($r, $c);
2351                }
2352                --$r;
2353        }
2354        return (undef, undef);
2355}
2356
2357sub init_vars {
2358        $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
2359        $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
2360        $REVDB = "$GIT_SVN_DIR/.rev_db";
2361        $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
2362        $SVN_URL = undef;
2363        $SVN_WC = "$GIT_SVN_DIR/tree";
2364        %tree_map = ();
2365}
2366
2367# convert GetOpt::Long specs for use by git-repo-config
2368sub read_repo_config {
2369        return unless -d $GIT_DIR;
2370        my $opts = shift;
2371        foreach my $o (keys %$opts) {
2372                my $v = $opts->{$o};
2373                my ($key) = ($o =~ /^([a-z\-]+)/);
2374                $key =~ s/-//g;
2375                my $arg = 'git-repo-config';
2376                $arg .= ' --int' if ($o =~ /[:=]i$/);
2377                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
2378                if (ref $v eq 'ARRAY') {
2379                        chomp(my @tmp = `$arg --get-all svn.$key`);
2380                        @$v = @tmp if @tmp;
2381                } else {
2382                        chomp(my $tmp = `$arg --get svn.$key`);
2383                        if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
2384                                $$v = $tmp;
2385                        }
2386                }
2387        }
2388}
2389
2390sub set_default_vals {
2391        if (defined $_repack) {
2392                $_repack = 1000 if ($_repack <= 0);
2393                $_repack_nr = $_repack;
2394                $_repack_flags ||= '-d';
2395        }
2396}
2397
2398sub read_grafts {
2399        my $gr_file = shift;
2400        my ($grafts, $comments) = ({}, {});
2401        if (open my $fh, '<', $gr_file) {
2402                my @tmp;
2403                while (<$fh>) {
2404                        if (/^($sha1)\s+/) {
2405                                my $c = $1;
2406                                if (@tmp) {
2407                                        @{$comments->{$c}} = @tmp;
2408                                        @tmp = ();
2409                                }
2410                                foreach my $p (split /\s+/, $_) {
2411                                        $grafts->{$c}->{$p} = 1;
2412                                }
2413                        } else {
2414                                push @tmp, $_;
2415                        }
2416                }
2417                close $fh or croak $!;
2418                @{$comments->{'END'}} = @tmp if @tmp;
2419        }
2420        return ($grafts, $comments);
2421}
2422
2423sub write_grafts {
2424        my ($grafts, $comments, $gr_file) = @_;
2425
2426        open my $fh, '>', $gr_file or croak $!;
2427        foreach my $c (sort keys %$grafts) {
2428                if ($comments->{$c}) {
2429                        print $fh $_ foreach @{$comments->{$c}};
2430                }
2431                my $p = $grafts->{$c};
2432                my %x; # real parents
2433                delete $p->{$c}; # commits are not self-reproducing...
2434                my $ch = command_output_pipe(qw/cat-file commit/, $c);
2435                while (<$ch>) {
2436                        if (/^parent ($sha1)/) {
2437                                $x{$1} = $p->{$1} = 1;
2438                        } else {
2439                                last unless /^\S/;
2440                        }
2441                }
2442                eval { command_close_pipe($ch) }; # breaking the pipe
2443
2444                # if real parents are the only ones in the grafts, drop it
2445                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2446
2447                my (@ip, @jp, $mb);
2448                my %del = %x;
2449                @ip = @jp = keys %$p;
2450                foreach my $i (@ip) {
2451                        next if $del{$i} || $p->{$i} == 2;
2452                        foreach my $j (@jp) {
2453                                next if $i eq $j || $del{$j} || $p->{$j} == 2;
2454                                $mb = eval { command('merge-base', $i, $j) };
2455                                next unless $mb;
2456                                chomp $mb;
2457                                next if $x{$mb};
2458                                if ($mb eq $j) {
2459                                        delete $p->{$i};
2460                                        $del{$i} = 1;
2461                                } elsif ($mb eq $i) {
2462                                        delete $p->{$j};
2463                                        $del{$j} = 1;
2464                                }
2465                        }
2466                }
2467
2468                # if real parents are the only ones in the grafts, drop it
2469                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2470
2471                print $fh $c, ' ', join(' ', sort keys %$p),"\n";
2472        }
2473        if ($comments->{'END'}) {
2474                print $fh $_ foreach @{$comments->{'END'}};
2475        }
2476        close $fh or croak $!;
2477}
2478
2479sub read_url_paths_all {
2480        my ($l_map, $pfx, $p) = @_;
2481        my @dir;
2482        foreach (<$p/*>) {
2483                if (-r "$_/info/url") {
2484                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2485                        my $id = $pfx . basename $_;
2486                        my $url = file_to_s("$_/info/url");
2487                        my ($u, $p) = repo_path_split($url);
2488                        $l_map->{$u}->{$p} = $id;
2489                } elsif (-d $_) {
2490                        push @dir, $_;
2491                }
2492        }
2493        foreach (@dir) {
2494                my $x = $_;
2495                $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
2496                read_url_paths_all($l_map, $x, $_);
2497        }
2498}
2499
2500# this one only gets ids that have been imported, not new ones
2501sub read_url_paths {
2502        my $l_map = {};
2503        git_svn_each(sub { my $x = shift;
2504                        my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
2505                        my ($u, $p) = repo_path_split($url);
2506                        $l_map->{$u}->{$p} = $x;
2507                        });
2508        return $l_map;
2509}
2510
2511sub extract_metadata {
2512        my $id = shift or return (undef, undef, undef);
2513        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
2514                                                        \s([a-f\d\-]+)$/x);
2515        if (!defined $rev || !$uuid || !$url) {
2516                # some of the original repositories I made had
2517                # identifiers like this:
2518                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
2519        }
2520        return ($url, $rev, $uuid);
2521}
2522
2523sub cmt_metadata {
2524        return extract_metadata((grep(/^git-svn-id: /,
2525                command(qw/cat-file commit/, shift)))[-1]);
2526}
2527
2528sub get_commit_time {
2529        my $cmt = shift;
2530        my $fh = command_output_pipe(qw/rev-list --pretty=raw -n1/, $cmt);
2531        while (<$fh>) {
2532                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
2533                my ($s, $tz) = ($1, $2);
2534                if ($tz =~ s/^\+//) {
2535                        $s += tz_to_s_offset($tz);
2536                } elsif ($tz =~ s/^\-//) {
2537                        $s -= tz_to_s_offset($tz);
2538                }
2539                eval { command_close_pipe($fh) };
2540                return $s;
2541        }
2542        die "Can't get commit time for commit: $cmt\n";
2543}
2544
2545sub tz_to_s_offset {
2546        my ($tz) = @_;
2547        $tz =~ s/(\d\d)$//;
2548        return ($1 * 60) + ($tz * 3600);
2549}
2550
2551# adapted from pager.c
2552sub config_pager {
2553        $_pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2554        if (!defined $_pager) {
2555                $_pager = 'less';
2556        } elsif (length $_pager == 0 || $_pager eq 'cat') {
2557                $_pager = undef;
2558        }
2559}
2560
2561sub run_pager {
2562        return unless -t *STDOUT;
2563        pipe my $rfd, my $wfd or return;
2564        defined(my $pid = fork) or croak $!;
2565        if (!$pid) {
2566                open STDOUT, '>&', $wfd or croak $!;
2567                return;
2568        }
2569        open STDIN, '<&', $rfd or croak $!;
2570        $ENV{LESS} ||= 'FRSX';
2571        exec $_pager or croak "Can't run pager: $! ($_pager)\n";
2572}
2573
2574sub get_author_info {
2575        my ($dest, $author, $t, $tz) = @_;
2576        $author =~ s/(?:^\s*|\s*$)//g;
2577        $dest->{a_raw} = $author;
2578        my $_a;
2579        if ($_authors) {
2580                $_a = $rusers{$author} || undef;
2581        }
2582        if (!$_a) {
2583                ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
2584        }
2585        $dest->{t} = $t;
2586        $dest->{tz} = $tz;
2587        $dest->{a} = $_a;
2588        # Date::Parse isn't in the standard Perl distro :(
2589        if ($tz =~ s/^\+//) {
2590                $t += tz_to_s_offset($tz);
2591        } elsif ($tz =~ s/^\-//) {
2592                $t -= tz_to_s_offset($tz);
2593        }
2594        $dest->{t_utc} = $t;
2595}
2596
2597sub process_commit {
2598        my ($c, $r_min, $r_max, $defer) = @_;
2599        if (defined $r_min && defined $r_max) {
2600                if ($r_min == $c->{r} && $r_min == $r_max) {
2601                        show_commit($c);
2602                        return 0;
2603                }
2604                return 1 if $r_min == $r_max;
2605                if ($r_min < $r_max) {
2606                        # we need to reverse the print order
2607                        return 0 if (defined $_limit && --$_limit < 0);
2608                        push @$defer, $c;
2609                        return 1;
2610                }
2611                if ($r_min != $r_max) {
2612                        return 1 if ($r_min < $c->{r});
2613                        return 1 if ($r_max > $c->{r});
2614                }
2615        }
2616        return 0 if (defined $_limit && --$_limit < 0);
2617        show_commit($c);
2618        return 1;
2619}
2620
2621sub show_commit {
2622        my $c = shift;
2623        if ($_oneline) {
2624                my $x = "\n";
2625                if (my $l = $c->{l}) {
2626                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2627                        $x = $l->[0];
2628                }
2629                $_l_fmt ||= 'A' . length($c->{r});
2630                print 'r',pack($_l_fmt, $c->{r}),' | ';
2631                print "$c->{c} | " if $_show_commit;
2632                print $x;
2633        } else {
2634                show_commit_normal($c);
2635        }
2636}
2637
2638sub show_commit_changed_paths {
2639        my ($c) = @_;
2640        return unless $c->{changed};
2641        print "Changed paths:\n", @{$c->{changed}};
2642}
2643
2644sub show_commit_normal {
2645        my ($c) = @_;
2646        print '-' x72, "\nr$c->{r} | ";
2647        print "$c->{c} | " if $_show_commit;
2648        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2649                                 localtime($c->{t_utc})), ' | ';
2650        my $nr_line = 0;
2651
2652        if (my $l = $c->{l}) {
2653                while ($l->[$#$l] eq "\n" && $#$l > 0
2654                                          && $l->[($#$l - 1)] eq "\n") {
2655                        pop @$l;
2656                }
2657                $nr_line = scalar @$l;
2658                if (!$nr_line) {
2659                        print "1 line\n\n\n";
2660                } else {
2661                        if ($nr_line == 1) {
2662                                $nr_line = '1 line';
2663                        } else {
2664                                $nr_line .= ' lines';
2665                        }
2666                        print $nr_line, "\n";
2667                        show_commit_changed_paths($c);
2668                        print "\n";
2669                        print $_ foreach @$l;
2670                }
2671        } else {
2672                print "1 line\n";
2673                show_commit_changed_paths($c);
2674                print "\n";
2675
2676        }
2677        foreach my $x (qw/raw diff/) {
2678                if ($c->{$x}) {
2679                        print "\n";
2680                        print $_ foreach @{$c->{$x}}
2681                }
2682        }
2683}
2684
2685sub libsvn_load {
2686        return unless $_use_lib;
2687        $_use_lib = eval {
2688                require SVN::Core;
2689                if ($SVN::Core::VERSION lt '1.1.0') {
2690                        die "Need SVN::Core 1.1.0 or better ",
2691                                        "(got $SVN::Core::VERSION) ",
2692                                        "Falling back to command-line svn\n";
2693                }
2694                require SVN::Ra;
2695                require SVN::Delta;
2696                push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
2697                push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
2698                *SVN::Git::Fetcher::process_rm = *process_rm;
2699                my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2700                                        $SVN::Node::dir.$SVN::Node::unknown.
2701                                        $SVN::Node::none.$SVN::Node::file.
2702                                        $SVN::Node::dir.$SVN::Node::unknown.
2703                                        $SVN::Auth::SSL::CNMISMATCH.
2704                                        $SVN::Auth::SSL::NOTYETVALID.
2705                                        $SVN::Auth::SSL::EXPIRED.
2706                                        $SVN::Auth::SSL::UNKNOWNCA.
2707                                        $SVN::Auth::SSL::OTHER;
2708                1;
2709        };
2710}
2711
2712sub _simple_prompt {
2713        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2714        $may_save = undef if $_no_auth_cache;
2715        $default_username = $_username if defined $_username;
2716        if (defined $default_username && length $default_username) {
2717                if (defined $realm && length $realm) {
2718                        print "Authentication realm: $realm\n";
2719                }
2720                $cred->username($default_username);
2721        } else {
2722                _username_prompt($cred, $realm, $may_save, $pool);
2723        }
2724        $cred->password(_read_password("Password for '" .
2725                                       $cred->username . "': ", $realm));
2726        $cred->may_save($may_save);
2727        $SVN::_Core::SVN_NO_ERROR;
2728}
2729
2730sub _ssl_server_trust_prompt {
2731        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2732        $may_save = undef if $_no_auth_cache;
2733        print "Error validating server certificate for '$realm':\n";
2734        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2735                print " - The certificate is not issued by a trusted ",
2736                      "authority. Use the\n",
2737                      "   fingerprint to validate the certificate manually!\n";
2738        }
2739        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2740                print " - The certificate hostname does not match.\n";
2741        }
2742        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2743                print " - The certificate is not yet valid.\n";
2744        }
2745        if ($failures & $SVN::Auth::SSL::EXPIRED) {
2746                print " - The certificate has expired.\n";
2747        }
2748        if ($failures & $SVN::Auth::SSL::OTHER) {
2749                print " - The certificate has an unknown error.\n";
2750        }
2751        printf( "Certificate information:\n".
2752                " - Hostname: %s\n".
2753                " - Valid: from %s until %s\n".
2754                " - Issuer: %s\n".
2755                " - Fingerprint: %s\n",
2756                map $cert_info->$_, qw(hostname valid_from valid_until
2757                                       issuer_dname fingerprint) );
2758        my $choice;
2759prompt:
2760        print $may_save ?
2761              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2762              "(R)eject or accept (t)emporarily? ";
2763        $choice = lc(substr(<STDIN> || 'R', 0, 1));
2764        if ($choice =~ /^t$/i) {
2765                $cred->may_save(undef);
2766        } elsif ($choice =~ /^r$/i) {
2767                return -1;
2768        } elsif ($may_save && $choice =~ /^p$/i) {
2769                $cred->may_save($may_save);
2770        } else {
2771                goto prompt;
2772        }
2773        $cred->accepted_failures($failures);
2774        $SVN::_Core::SVN_NO_ERROR;
2775}
2776
2777sub _ssl_client_cert_prompt {
2778        my ($cred, $realm, $may_save, $pool) = @_;
2779        $may_save = undef if $_no_auth_cache;
2780        print "Client certificate filename: ";
2781        chomp(my $filename = <STDIN>);
2782        $cred->cert_file($filename);
2783        $cred->may_save($may_save);
2784        $SVN::_Core::SVN_NO_ERROR;
2785}
2786
2787sub _ssl_client_cert_pw_prompt {
2788        my ($cred, $realm, $may_save, $pool) = @_;
2789        $may_save = undef if $_no_auth_cache;
2790        $cred->password(_read_password("Password: ", $realm));
2791        $cred->may_save($may_save);
2792        $SVN::_Core::SVN_NO_ERROR;
2793}
2794
2795sub _username_prompt {
2796        my ($cred, $realm, $may_save, $pool) = @_;
2797        $may_save = undef if $_no_auth_cache;
2798        if (defined $realm && length $realm) {
2799                print "Authentication realm: $realm\n";
2800        }
2801        my $username;
2802        if (defined $_username) {
2803                $username = $_username;
2804        } else {
2805                print "Username: ";
2806                chomp($username = <STDIN>);
2807        }
2808        $cred->username($username);
2809        $cred->may_save($may_save);
2810        $SVN::_Core::SVN_NO_ERROR;
2811}
2812
2813sub _read_password {
2814        my ($prompt, $realm) = @_;
2815        print $prompt;
2816        require Term::ReadKey;
2817        Term::ReadKey::ReadMode('noecho');
2818        my $password = '';
2819        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2820                last if $key =~ /[\012\015]/; # \n\r
2821                $password .= $key;
2822        }
2823        Term::ReadKey::ReadMode('restore');
2824        print "\n";
2825        $password;
2826}
2827
2828sub libsvn_connect {
2829        my ($url) = @_;
2830        SVN::_Core::svn_config_ensure($_config_dir, undef);
2831        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2832            SVN::Client::get_simple_provider(),
2833            SVN::Client::get_ssl_server_trust_file_provider(),
2834            SVN::Client::get_simple_prompt_provider(
2835              \&_simple_prompt, 2),
2836            SVN::Client::get_ssl_client_cert_prompt_provider(
2837              \&_ssl_client_cert_prompt, 2),
2838            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2839              \&_ssl_client_cert_pw_prompt, 2),
2840            SVN::Client::get_username_provider(),
2841            SVN::Client::get_ssl_server_trust_prompt_provider(
2842              \&_ssl_server_trust_prompt),
2843            SVN::Client::get_username_prompt_provider(
2844              \&_username_prompt, 2),
2845          ]);
2846        my $config = SVN::Core::config_get_config($_config_dir);
2847        my $ra = SVN::Ra->new(url => $url, auth => $baton,
2848                              config => $config,
2849                              pool => SVN::Pool->new,
2850                              auth_provider_callbacks => $callbacks);
2851
2852        my $df = $ENV{GIT_SVN_DELTA_FETCH};
2853        if (defined $df) {
2854                $_xfer_delta = $df;
2855        } else {
2856                $_xfer_delta = ($url =~ m#^file://#) ? undef : 1;
2857        }
2858        $ra->{svn_path} = $url;
2859        $ra->{repos_root} = $ra->get_repos_root;
2860        $ra->{svn_path} =~ s#^\Q$ra->{repos_root}\E/*##;
2861        push @repo_path_split_cache, qr/^(\Q$ra->{repos_root}\E)/;
2862        return $ra;
2863}
2864
2865sub libsvn_can_do_switch {
2866        unless (defined $_svn_can_do_switch) {
2867                my $pool = SVN::Pool->new;
2868                my $rep = eval {
2869                        $SVN->do_switch(1, '', 0, $SVN->{url},
2870                                        SVN::Delta::Editor->new, $pool);
2871                };
2872                if ($@) {
2873                        $_svn_can_do_switch = 0;
2874                } else {
2875                        $rep->abort_report($pool);
2876                        $_svn_can_do_switch = 1;
2877                }
2878                $pool->clear;
2879        }
2880        $_svn_can_do_switch;
2881}
2882
2883sub libsvn_dup_ra {
2884        my ($ra) = @_;
2885        SVN::Ra->new(map { $_ => $ra->{$_} } qw/config url
2886                     auth auth_provider_callbacks repos_root svn_path/);
2887}
2888
2889sub libsvn_get_file {
2890        my ($gui, $f, $rev, $chg, $untracked) = @_;
2891        $f =~ s#^/##;
2892        print "\t$chg\t$f\n" unless $_q;
2893
2894        my ($hash, $pid, $in, $out);
2895        my $pool = SVN::Pool->new;
2896        defined($pid = open3($in, $out, '>&STDERR',
2897                                qw/git-hash-object -w --stdin/)) or croak $!;
2898        # redirect STDOUT for SVN 1.1.x compatibility
2899        open my $stdout, '>&', \*STDOUT or croak $!;
2900        open STDOUT, '>&', $in or croak $!;
2901        my ($r, $props) = $SVN->get_file($f, $rev, \*STDOUT, $pool);
2902        $in->flush == 0 or croak $!;
2903        open STDOUT, '>&', $stdout or croak $!;
2904        close $in or croak $!;
2905        close $stdout or croak $!;
2906        $pool->clear;
2907        chomp($hash = do { local $/; <$out> });
2908        close $out or croak $!;
2909        waitpid $pid, 0;
2910        $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2911
2912        my $mode = exists $props->{'svn:executable'} ? '100755' : '100644';
2913        if (exists $props->{'svn:special'}) {
2914                $mode = '120000';
2915                my $link = `git-cat-file blob $hash`; # no chomping symlinks
2916                $link =~ s/^link // or die "svn:special file with contents: <",
2917                                                $link, "> is not understood\n";
2918                defined($pid = open3($in, $out, '>&STDERR',
2919                                qw/git-hash-object -w --stdin/)) or croak $!;
2920                print $in $link;
2921                $in->flush == 0 or croak $!;
2922                close $in or croak $!;
2923                chomp($hash = do { local $/; <$out> });
2924                close $out or croak $!;
2925                waitpid $pid, 0;
2926                $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2927        }
2928        %{$untracked->{file_prop}->{$f}} = %$props;
2929        print $gui $mode,' ',$hash,"\t",$f,"\0" or croak $!;
2930}
2931
2932sub uri_encode {
2933        my ($f) = @_;
2934        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2935        $f
2936}
2937
2938sub uri_decode {
2939        my ($f) = @_;
2940        $f =~ tr/+/ /;
2941        $f =~ s/%([A-F0-9]{2})/chr hex($1)/ge;
2942        $f
2943}
2944
2945sub libsvn_log_entry {
2946        my ($rev, $author, $date, $msg, $parents, $untracked) = @_;
2947        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2948                                         (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
2949                                or die "Unable to parse date: $date\n";
2950        if (defined $author && length $author > 0 &&
2951            defined $_authors && ! defined $users{$author}) {
2952                die "Author: $author not defined in $_authors file\n";
2953        }
2954        $msg = '' if ($rev == 0 && !defined $msg);
2955
2956        open my $un, '>>', "$GIT_SVN_DIR/unhandled.log" or croak $!;
2957        my $h;
2958        print $un "r$rev\n" or croak $!;
2959        $h = $untracked->{empty};
2960        foreach (sort keys %$h) {
2961                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2962                print $un "  $act: ", uri_encode($_), "\n" or croak $!;
2963                warn "W: $act: $_\n";
2964        }
2965        foreach my $t (qw/dir_prop file_prop/) {
2966                $h = $untracked->{$t} or next;
2967                foreach my $path (sort keys %$h) {
2968                        my $ppath = $path eq '' ? '.' : $path;
2969                        foreach my $prop (sort keys %{$h->{$path}}) {
2970                                next if $SKIP{$prop};
2971                                my $v = $h->{$path}->{$prop};
2972                                if (defined $v) {
2973                                        print $un "  +$t: ",
2974                                                  uri_encode($ppath), ' ',
2975                                                  uri_encode($prop), ' ',
2976                                                  uri_encode($v), "\n"
2977                                                  or croak $!;
2978                                } else {
2979                                        print $un "  -$t: ",
2980                                                  uri_encode($ppath), ' ',
2981                                                  uri_encode($prop), "\n"
2982                                                  or croak $!;
2983                                }
2984                        }
2985                }
2986        }
2987        foreach my $t (qw/absent_file absent_directory/) {
2988                $h = $untracked->{$t} or next;
2989                foreach my $parent (sort keys %$h) {
2990                        foreach my $path (sort @{$h->{$parent}}) {
2991                                print $un "  $t: ",
2992                                      uri_encode("$parent/$path"), "\n"
2993                                      or croak $!;
2994                                warn "W: $t: $parent/$path ",
2995                                     "Insufficient permissions?\n";
2996                        }
2997                }
2998        }
2999
3000        # revprops (make this optional? it's an extra network trip...)
3001        my $pool = SVN::Pool->new;
3002        my $rp = $SVN->rev_proplist($rev, $pool);
3003        foreach (sort keys %$rp) {
3004                next if /^svn:(?:author|date|log)$/;
3005                print $un "  rev_prop: ", uri_encode($_), ' ',
3006                          uri_encode($rp->{$_}), "\n";
3007        }
3008        $pool->clear;
3009        close $un or croak $!;
3010
3011        { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
3012          author => $author, msg => $msg."\n", parents => $parents || [],
3013          revprops => $rp }
3014}
3015
3016sub process_rm {
3017        my ($gui, $last_commit, $f, $q) = @_;
3018        # remove entire directories.
3019        if (command('ls-tree',$last_commit,'--',$f) =~ /^040000 tree/) {
3020                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3021                                                     -r --name-only -z/,
3022                                                     $last_commit,'--',$f);
3023                local $/ = "\0";
3024                while (<$ls>) {
3025                        print $gui '0 ',0 x 40,"\t",$_ or croak $!;
3026                        print "\tD\t$_\n" unless $q;
3027                }
3028                print "\tD\t$f/\n" unless $q;
3029                command_close_pipe($ls, $ctx);
3030                return $SVN::Node::dir;
3031        } else {
3032                print $gui '0 ',0 x 40,"\t",$f,"\0" or croak $!;
3033                print "\tD\t$f\n" unless $q;
3034                return $SVN::Node::file;
3035        }
3036}
3037
3038sub libsvn_fetch {
3039        $_xfer_delta ? libsvn_fetch_delta(@_) : libsvn_fetch_full(@_);
3040}
3041
3042sub libsvn_fetch_delta {
3043        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
3044        my $pool = SVN::Pool->new;
3045        my $ed = SVN::Git::Fetcher->new({ c => $last_commit, q => $_q });
3046        my $reporter = $SVN->do_update($rev, '', 1, $ed, $pool);
3047        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3048        my (undef, $last_rev, undef) = cmt_metadata($last_commit);
3049        $reporter->set_path('', $last_rev, 0, @lock, $pool);
3050        $reporter->finish_report($pool);
3051        $pool->clear;
3052        unless ($ed->{git_commit_ok}) {
3053                die "SVN connection failed somewhere...\n";
3054        }
3055        libsvn_log_entry($rev, $author, $date, $msg, [$last_commit], $ed);
3056}
3057
3058sub libsvn_fetch_full {
3059        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
3060        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3061        my %amr;
3062        my $ut = { empty => {}, dir_prop => {}, file_prop => {} };
3063        my $p = $SVN->{svn_path};
3064        foreach my $f (keys %$paths) {
3065                my $m = $paths->{$f}->action();
3066                if (length $p) {
3067                        $f =~ s#^/\Q$p\E/##;
3068                        next if $f =~ m#^/#;
3069                } else {
3070                        $f =~ s#^/##;
3071                }
3072                if ($m =~ /^[DR]$/) {
3073                        my $t = process_rm($gui, $last_commit, $f, $_q);
3074                        if ($m eq 'D') {
3075                                $ut->{empty}->{$f} = 0 if $t == $SVN::Node::dir;
3076                                next;
3077                        }
3078                        # 'R' can be file replacements, too, right?
3079                }
3080                my $pool = SVN::Pool->new;
3081                my $t = $SVN->check_path($f, $rev, $pool);
3082                if ($t == $SVN::Node::file) {
3083                        if ($m =~ /^[AMR]$/) {
3084                                $amr{$f} = $m;
3085                        } else {
3086                                die "Unrecognized action: $m, ($f r$rev)\n";
3087                        }
3088                } elsif ($t == $SVN::Node::dir && $m =~ /^[AR]$/) {
3089                        my @traversed = ();
3090                        libsvn_traverse($gui, '', $f, $rev, \@traversed, $ut);
3091                        if (@traversed) {
3092                                foreach (@traversed) {
3093                                        $amr{$_} = $m;
3094                                }
3095                        } else {
3096                                my ($dir, $file) = ($f =~ m#^(.*?)/?([^/]+)$#);
3097                                delete $ut->{empty}->{$dir};
3098                                $ut->{empty}->{$f} = 1;
3099                        }
3100                }
3101                $pool->clear;
3102        }
3103        foreach (keys %amr) {
3104                libsvn_get_file($gui, $_, $rev, $amr{$_}, $ut);
3105                my ($d) = ($_ =~ m#^(.*?)/?(?:[^/]+)$#);
3106                delete $ut->{empty}->{$d};
3107        }
3108        unless (exists $ut->{dir_prop}->{''}) {
3109                my $pool = SVN::Pool->new;
3110                my (undef, undef, $props) = $SVN->get_dir('', $rev, $pool);
3111                %{$ut->{dir_prop}->{''}} = %$props;
3112                $pool->clear;
3113        }
3114        command_close_pipe($gui, $ctx);
3115        libsvn_log_entry($rev, $author, $date, $msg, [$last_commit], $ut);
3116}
3117
3118sub svn_grab_base_rev {
3119        my $c = eval { command_oneline([qw/rev-parse --verify/,
3120                                        "refs/remotes/$GIT_SVN^0"],
3121                                        { STDERR => 0 }) };
3122        if (defined $c && length $c) {
3123                my ($url, $rev, $uuid) = cmt_metadata($c);
3124                return ($rev, $c) if defined $rev;
3125        }
3126        if ($_no_metadata) {
3127                my $offset = -41; # from tail
3128                my $rl;
3129                open my $fh, '<', $REVDB or
3130                        die "--no-metadata specified and $REVDB not readable\n";
3131                seek $fh, $offset, 2;
3132                $rl = readline $fh;
3133                defined $rl or return (undef, undef);
3134                chomp $rl;
3135                while ($c ne $rl && tell $fh != 0) {
3136                        $offset -= 41;
3137                        seek $fh, $offset, 2;
3138                        $rl = readline $fh;
3139                        defined $rl or return (undef, undef);
3140                        chomp $rl;
3141                }
3142                my $rev = tell $fh;
3143                croak $! if ($rev < -1);
3144                $rev =  ($rev - 41) / 41;
3145                close $fh or croak $!;
3146                return ($rev, $c);
3147        }
3148        return (undef, undef);
3149}
3150
3151sub libsvn_parse_revision {
3152        my $base = shift;
3153        my $head = $SVN->get_latest_revnum();
3154        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
3155                return ($base + 1, $head) if (defined $base);
3156                return (0, $head);
3157        }
3158        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
3159        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
3160        if ($_revision =~ /^BASE:(\d+)$/) {
3161                return ($base + 1, $1) if (defined $base);
3162                return (0, $head);
3163        }
3164        return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
3165        die "revision argument: $_revision not understood by git-svn\n",
3166                "Try using the command-line svn client instead\n";
3167}
3168
3169sub libsvn_traverse {
3170        my ($gui, $pfx, $path, $rev, $files, $untracked) = @_;
3171        my $cwd = length $pfx ? "$pfx/$path" : $path;
3172        my $pool = SVN::Pool->new;
3173        $cwd =~ s#^\Q$SVN->{svn_path}\E##;
3174        my $nr = 0;
3175        my ($dirent, $r, $props) = $SVN->get_dir($cwd, $rev, $pool);
3176        %{$untracked->{dir_prop}->{$cwd}} = %$props;
3177        foreach my $d (keys %$dirent) {
3178                my $t = $dirent->{$d}->kind;
3179                if ($t == $SVN::Node::dir) {
3180                        my $i = libsvn_traverse($gui, $cwd, $d, $rev,
3181                                                $files, $untracked);
3182                        if ($i) {
3183                                $nr += $i;
3184                        } else {
3185                                $untracked->{empty}->{"$cwd/$d"} = 1;
3186                        }
3187                } elsif ($t == $SVN::Node::file) {
3188                        $nr++;
3189                        my $file = "$cwd/$d";
3190                        if (defined $files) {
3191                                push @$files, $file;
3192                        } else {
3193                                libsvn_get_file($gui, $file, $rev, 'A',
3194                                                $untracked);
3195                                my ($dir) = ($file =~ m#^(.*?)/?(?:[^/]+)$#);
3196                                delete $untracked->{empty}->{$dir};
3197                        }
3198                }
3199        }
3200        $pool->clear;
3201        $nr;
3202}
3203
3204sub libsvn_traverse_ignore {
3205        my ($fh, $path, $r) = @_;
3206        $path =~ s#^/+##g;
3207        my $pool = SVN::Pool->new;
3208        my ($dirent, undef, $props) = $SVN->get_dir($path, $r, $pool);
3209        my $p = $path;
3210        $p =~ s#^\Q$SVN->{svn_path}\E/##;
3211        print $fh length $p ? "\n# $p\n" : "\n# /\n";
3212        if (my $s = $props->{'svn:ignore'}) {
3213                $s =~ s/[\r\n]+/\n/g;
3214                chomp $s;
3215                if (length $p == 0) {
3216                        $s =~ s#\n#\n/$p#g;
3217                        print $fh "/$s\n";
3218                } else {
3219                        $s =~ s#\n#\n/$p/#g;
3220                        print $fh "/$p/$s\n";
3221                }
3222        }
3223        foreach (sort keys %$dirent) {
3224                next if $dirent->{$_}->kind != $SVN::Node::dir;
3225                libsvn_traverse_ignore($fh, "$path/$_", $r);
3226        }
3227        $pool->clear;
3228}
3229
3230sub revisions_eq {
3231        my ($path, $r0, $r1) = @_;
3232        return 1 if $r0 == $r1;
3233        my $nr = 0;
3234        if ($_use_lib) {
3235                # should be OK to use Pool here (r1 - r0) should be small
3236                my $pool = SVN::Pool->new;
3237                libsvn_get_log($SVN, [$path], $r0, $r1,
3238                                0, 0, 1, sub {$nr++}, $pool);
3239                $pool->clear;
3240        } else {
3241                my ($url, undef) = repo_path_split($SVN_URL);
3242                my $svn_log = svn_log_raw("$url/$path","-r$r0:$r1");
3243                while (next_log_entry($svn_log)) { $nr++ }
3244                close $svn_log->{fh};
3245        }
3246        return 0 if ($nr > 1);
3247        return 1;
3248}
3249
3250sub libsvn_find_parent_branch {
3251        my ($paths, $rev, $author, $date, $msg) = @_;
3252        my $svn_path = '/'.$SVN->{svn_path};
3253
3254        # look for a parent from another branch:
3255        my $i = $paths->{$svn_path} or return;
3256        my $branch_from = $i->copyfrom_path or return;
3257        my $r = $i->copyfrom_rev;
3258        print STDERR  "Found possible branch point: ",
3259                                "$branch_from => $svn_path, $r\n";
3260        $branch_from =~ s#^/##;
3261        my $l_map = {};
3262        read_url_paths_all($l_map, '', "$GIT_DIR/svn");
3263        my $url = $SVN->{repos_root};
3264        defined $l_map->{$url} or return;
3265        my $id = $l_map->{$url}->{$branch_from};
3266        if (!defined $id && $_follow_parent) {
3267                print STDERR "Following parent: $branch_from\@$r\n";
3268                # auto create a new branch and follow it
3269                $id = basename($branch_from);
3270                $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
3271                while (-r "$GIT_DIR/svn/$id") {
3272                        # just grow a tail if we're not unique enough :x
3273                        $id .= '-';
3274                }
3275        }
3276        return unless defined $id;
3277
3278        my ($r0, $parent) = find_rev_before($r,$id,1);
3279        if ($_follow_parent && (!defined $r0 || !defined $parent)) {
3280                defined(my $pid = fork) or croak $!;
3281                if (!$pid) {
3282                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
3283                        init_vars();
3284                        $SVN_URL = "$url/$branch_from";
3285                        $SVN = undef;
3286                        setup_git_svn();
3287                        # we can't assume SVN_URL exists at r+1:
3288                        $_revision = "0:$r";
3289                        fetch_lib();
3290                        exit 0;
3291                }
3292                waitpid $pid, 0;
3293                croak $? if $?;
3294                ($r0, $parent) = find_rev_before($r,$id,1);
3295        }
3296        return unless (defined $r0 && defined $parent);
3297        if (revisions_eq($branch_from, $r0, $r)) {
3298                unlink $GIT_SVN_INDEX;
3299                print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
3300                command_noisy('read-tree', $parent);
3301                unless (libsvn_can_do_switch()) {
3302                        return libsvn_fetch_full($parent, $paths, $rev,
3303                                                $author, $date, $msg);
3304                }
3305                # do_switch works with svn/trunk >= r22312, but that is not
3306                # included with SVN 1.4.2 (the latest version at the moment),
3307                # so we can't rely on it.
3308                my $ra = libsvn_connect("$url/$branch_from");
3309                my $ed = SVN::Git::Fetcher->new({c => $parent, q => $_q });
3310                my $pool = SVN::Pool->new;
3311                my $reporter = $ra->do_switch($rev, '', 1, $SVN->{url},
3312                                              $ed, $pool);
3313                my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3314                $reporter->set_path('', $r0, 0, @lock, $pool);
3315                $reporter->finish_report($pool);
3316                $pool->clear;
3317                unless ($ed->{git_commit_ok}) {
3318                        die "SVN connection failed somewhere...\n";
3319                }
3320                return libsvn_log_entry($rev, $author, $date, $msg, [$parent]);
3321        }
3322        print STDERR "Nope, branch point not imported or unknown\n";
3323        return undef;
3324}
3325
3326sub libsvn_get_log {
3327        my ($ra, @args) = @_;
3328        $args[4]-- if $args[4] && $_xfer_delta && ! $_follow_parent;
3329        if ($SVN::Core::VERSION le '1.2.0') {
3330                splice(@args, 3, 1);
3331        }
3332        $ra->get_log(@args);
3333}
3334
3335sub libsvn_new_tree {
3336        if (my $log_entry = libsvn_find_parent_branch(@_)) {
3337                return $log_entry;
3338        }
3339        my ($paths, $rev, $author, $date, $msg) = @_;
3340        my $ut;
3341        if ($_xfer_delta) {
3342                my $pool = SVN::Pool->new;
3343                my $ed = SVN::Git::Fetcher->new({q => $_q});
3344                my $reporter = $SVN->do_update($rev, '', 1, $ed, $pool);
3345                my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3346                $reporter->set_path('', $rev, 1, @lock, $pool);
3347                $reporter->finish_report($pool);
3348                $pool->clear;
3349                unless ($ed->{git_commit_ok}) {
3350                        die "SVN connection failed somewhere...\n";
3351                }
3352                $ut = $ed;
3353        } else {
3354                $ut = { empty => {}, dir_prop => {}, file_prop => {} };
3355                my ($gui, $ctx) = command_input_pipe(qw/update-index
3356                                                     -z --index-info/);
3357                libsvn_traverse($gui, '', $SVN->{svn_path}, $rev, undef, $ut);
3358                command_close_pipe($gui, $ctx);
3359        }
3360        libsvn_log_entry($rev, $author, $date, $msg, [], $ut);
3361}
3362
3363sub find_graft_path_commit {
3364        my ($tree_paths, $p1, $r1) = @_;
3365        foreach my $x (keys %$tree_paths) {
3366                next unless ($p1 =~ /^\Q$x\E/);
3367                my $i = $tree_paths->{$x};
3368                my ($r0, $parent) = find_rev_before($r1,$i,1);
3369                return $parent if (defined $r0 && $r0 == $r1);
3370                print STDERR "r$r1 of $i not imported\n";
3371                next;
3372        }
3373        return undef;
3374}
3375
3376sub find_graft_path_parents {
3377        my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
3378        foreach my $x (keys %$tree_paths) {
3379                next unless ($p0 =~ /^\Q$x\E/);
3380                my $i = $tree_paths->{$x};
3381                my ($r, $parent) = find_rev_before($r0, $i, 1);
3382                if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
3383                        my ($url_b, undef, $uuid_b) = cmt_metadata($c);
3384                        my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
3385                        next if ($url_a && $url_b && $url_a eq $url_b &&
3386                                                        $uuid_b eq $uuid_a);
3387                        $grafts->{$c}->{$parent} = 1;
3388                }
3389        }
3390}
3391
3392sub libsvn_graft_file_copies {
3393        my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
3394        foreach (keys %$paths) {
3395                my $i = $paths->{$_};
3396                my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
3397                                        $i->copyfrom_rev);
3398                next unless (defined $p0 && defined $r0);
3399
3400                my $p1 = $_;
3401                $p1 =~ s#^/##;
3402                $p0 =~ s#^/##;
3403                my $c = find_graft_path_commit($tree_paths, $p1, $rev);
3404                next unless $c;
3405                find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
3406        }
3407}
3408
3409sub set_index {
3410        my $old = $ENV{GIT_INDEX_FILE};
3411        $ENV{GIT_INDEX_FILE} = shift;
3412        return $old;
3413}
3414
3415sub restore_index {
3416        my ($old) = @_;
3417        if (defined $old) {
3418                $ENV{GIT_INDEX_FILE} = $old;
3419        } else {
3420                delete $ENV{GIT_INDEX_FILE};
3421        }
3422}
3423
3424sub libsvn_commit_cb {
3425        my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
3426        if ($_optimize_commits && $rev == ($r_last + 1)) {
3427                my $log = libsvn_log_entry($rev,$committer,$date,$msg);
3428                $log->{tree} = get_tree_from_treeish($c);
3429                my $cmt = git_commit($log, $cmt_last, $c);
3430                my @diff = command('diff-tree', $cmt, $c);
3431                if (@diff) {
3432                        print STDERR "Trees differ: $cmt $c\n",
3433                                        join('',@diff),"\n";
3434                        exit 1;
3435                }
3436        } else {
3437                fetch("$rev=$c");
3438        }
3439}
3440
3441sub libsvn_ls_fullurl {
3442        my $fullurl = shift;
3443        my $ra = libsvn_connect($fullurl);
3444        my @ret;
3445        my $pool = SVN::Pool->new;
3446        my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
3447        my ($dirent, undef, undef) = $ra->get_dir('', $r, $pool);
3448        foreach my $d (keys %$dirent) {
3449                if ($dirent->{$d}->kind == $SVN::Node::dir) {
3450                        push @ret, "$d/"; # add '/' for compat with cli svn
3451                }
3452        }
3453        $pool->clear;
3454        return @ret;
3455}
3456
3457
3458sub libsvn_skip_unknown_revs {
3459        my $err = shift;
3460        my $errno = $err->apr_err();
3461        # Maybe the branch we're tracking didn't
3462        # exist when the repo started, so it's
3463        # not an error if it doesn't, just continue
3464        #
3465        # Wonderfully consistent library, eh?
3466        # 160013 - svn:// and file://
3467        # 175002 - http(s)://
3468        # 175007 - http(s):// (this repo required authorization, too...)
3469        #   More codes may be discovered later...
3470        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3471                return;
3472        }
3473        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3474};
3475
3476# Tie::File seems to be prone to offset errors if revisions get sparse,
3477# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3478# one of my favorite modules is out :<  Next up would be one of the DBM
3479# modules, but I'm not sure which is most portable...  So I'll just
3480# go with something that's plain-text, but still capable of
3481# being randomly accessed.  So here's my ultra-simple fixed-width
3482# database.  All records are 40 characters + "\n", so it's easy to seek
3483# to a revision: (41 * rev) is the byte offset.
3484# A record of 40 0s denotes an empty revision.
3485# And yes, it's still pretty fast (faster than Tie::File).
3486sub revdb_set {
3487        my ($file, $rev, $commit) = @_;
3488        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
3489        open my $fh, '+<', $file or croak $!;
3490        my $offset = $rev * 41;
3491        # assume that append is the common case:
3492        seek $fh, 0, 2 or croak $!;
3493        my $pos = tell $fh;
3494        if ($pos < $offset) {
3495                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
3496        }
3497        seek $fh, $offset, 0 or croak $!;
3498        print $fh $commit,"\n";
3499        close $fh or croak $!;
3500}
3501
3502sub revdb_get {
3503        my ($file, $rev) = @_;
3504        my $ret;
3505        my $offset = $rev * 41;
3506        open my $fh, '<', $file or croak $!;
3507        seek $fh, $offset, 0;
3508        if (tell $fh == $offset) {
3509                $ret = readline $fh;
3510                if (defined $ret) {
3511                        chomp $ret;
3512                        $ret = undef if ($ret =~ /^0{40}$/);
3513                }
3514        }
3515        close $fh or croak $!;
3516        return $ret;
3517}
3518
3519sub copy_remote_ref {
3520        my $origin = $_cp_remote ? $_cp_remote : 'origin';
3521        my $ref = "refs/remotes/$GIT_SVN";
3522        if (command('ls-remote', $origin, $ref)) {
3523                command_noisy('fetch', $origin, "$ref:$ref");
3524        } elsif ($_cp_remote && !$_upgrade) {
3525                die "Unable to find remote reference: ",
3526                                "refs/remotes/$GIT_SVN on $origin\n";
3527        }
3528}
3529package SVN::Git::Fetcher;
3530use vars qw/@ISA/;
3531use strict;
3532use warnings;
3533use Carp qw/croak/;
3534use IO::File qw//;
3535use Git qw/command command_oneline command_noisy
3536           command_output_pipe command_input_pipe command_close_pipe/;
3537
3538# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3539sub new {
3540        my ($class, $git_svn) = @_;
3541        my $self = SVN::Delta::Editor->new;
3542        bless $self, $class;
3543        $self->{c} = $git_svn->{c} if exists $git_svn->{c};
3544        $self->{q} = $git_svn->{q};
3545        $self->{empty} = {};
3546        $self->{dir_prop} = {};
3547        $self->{file_prop} = {};
3548        $self->{absent_dir} = {};
3549        $self->{absent_file} = {};
3550        ($self->{gui}, $self->{ctx}) = command_input_pipe(
3551                                             qw/update-index -z --index-info/);
3552        require Digest::MD5;
3553        $self;
3554}
3555
3556sub open_root {
3557        { path => '' };
3558}
3559
3560sub open_directory {
3561        my ($self, $path, $pb, $rev) = @_;
3562        { path => $path };
3563}
3564
3565sub delete_entry {
3566        my ($self, $path, $rev, $pb) = @_;
3567        my $t = process_rm($self->{gui}, $self->{c}, $path, $self->{q});
3568        $self->{empty}->{$path} = 0 if $t == $SVN::Node::dir;
3569        undef;
3570}
3571
3572sub open_file {
3573        my ($self, $path, $pb, $rev) = @_;
3574        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--',$path)
3575                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3576        unless (defined $mode && defined $blob) {
3577                die "$path was not found in commit $self->{c} (r$rev)\n";
3578        }
3579        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3580          pool => SVN::Pool->new, action => 'M' };
3581}
3582
3583sub add_file {
3584        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3585        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3586        delete $self->{empty}->{$dir};
3587        { path => $path, mode_a => 100644, mode_b => 100644,
3588          pool => SVN::Pool->new, action => 'A' };
3589}
3590
3591sub add_directory {
3592        my ($self, $path, $cp_path, $cp_rev) = @_;
3593        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3594        delete $self->{empty}->{$dir};
3595        $self->{empty}->{$path} = 1;
3596        { path => $path };
3597}
3598
3599sub change_dir_prop {
3600        my ($self, $db, $prop, $value) = @_;
3601        $self->{dir_prop}->{$db->{path}} ||= {};
3602        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3603        undef;
3604}
3605
3606sub absent_directory {
3607        my ($self, $path, $pb) = @_;
3608        $self->{absent_dir}->{$pb->{path}} ||= [];
3609        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3610        undef;
3611}
3612
3613sub absent_file {
3614        my ($self, $path, $pb) = @_;
3615        $self->{absent_file}->{$pb->{path}} ||= [];
3616        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3617        undef;
3618}
3619
3620sub change_file_prop {
3621        my ($self, $fb, $prop, $value) = @_;
3622        if ($prop eq 'svn:executable') {
3623                if ($fb->{mode_b} != 120000) {
3624                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3625                }
3626        } elsif ($prop eq 'svn:special') {
3627                $fb->{mode_b} = defined $value ? 120000 : 100644;
3628        } else {
3629                $self->{file_prop}->{$fb->{path}} ||= {};
3630                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3631        }
3632        undef;
3633}
3634
3635sub apply_textdelta {
3636        my ($self, $fb, $exp) = @_;
3637        my $fh = IO::File->new_tmpfile;
3638        $fh->autoflush(1);
3639        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3640        # (but $base does not,) so dup() it for reading in close_file
3641        open my $dup, '<&', $fh or croak $!;
3642        my $base = IO::File->new_tmpfile;
3643        $base->autoflush(1);
3644        if ($fb->{blob}) {
3645                defined (my $pid = fork) or croak $!;
3646                if (!$pid) {
3647                        open STDOUT, '>&', $base or croak $!;
3648                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
3649                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
3650                }
3651                waitpid $pid, 0;
3652                croak $? if $?;
3653
3654                if (defined $exp) {
3655                        seek $base, 0, 0 or croak $!;
3656                        my $md5 = Digest::MD5->new;
3657                        $md5->addfile($base);
3658                        my $got = $md5->hexdigest;
3659                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3660                            "expected: $exp\n",
3661                            "     got: $got\n" if ($got ne $exp);
3662                }
3663        }
3664        seek $base, 0, 0 or croak $!;
3665        $fb->{fh} = $dup;
3666        $fb->{base} = $base;
3667        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
3668}
3669
3670sub close_file {
3671        my ($self, $fb, $exp) = @_;
3672        my $hash;
3673        my $path = $fb->{path};
3674        if (my $fh = $fb->{fh}) {
3675                seek($fh, 0, 0) or croak $!;
3676                my $md5 = Digest::MD5->new;
3677                $md5->addfile($fh);
3678                my $got = $md5->hexdigest;
3679                die "Checksum mismatch: $path\n",
3680                    "expected: $exp\n    got: $got\n" if ($got ne $exp);
3681                seek($fh, 0, 0) or croak $!;
3682                if ($fb->{mode_b} == 120000) {
3683                        read($fh, my $buf, 5) == 5 or croak $!;
3684                        $buf eq 'link ' or die "$path has mode 120000",
3685                                               "but is not a link\n";
3686                }
3687                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
3688                if (!$pid) {
3689                        open STDIN, '<&', $fh or croak $!;
3690                        exec qw/git-hash-object -w --stdin/ or croak $!;
3691                }
3692                chomp($hash = do { local $/; <$out> });
3693                close $out or croak $!;
3694                close $fh or croak $!;
3695                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3696                close $fb->{base} or croak $!;
3697        } else {
3698                $hash = $fb->{blob} or die "no blob information\n";
3699        }
3700        $fb->{pool}->clear;
3701        my $gui = $self->{gui};
3702        print $gui "$fb->{mode_b} $hash\t$path\0" or croak $!;
3703        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
3704        undef;
3705}
3706
3707sub abort_edit {
3708        my $self = shift;
3709        eval { command_close_pipe($self->{gui}, $self->{ctx}) };
3710        $self->SUPER::abort_edit(@_);
3711}
3712
3713sub close_edit {
3714        my $self = shift;
3715        command_close_pipe($self->{gui}, $self->{ctx});
3716        $self->{git_commit_ok} = 1;
3717        $self->SUPER::close_edit(@_);
3718}
3719
3720package SVN::Git::Editor;
3721use vars qw/@ISA/;
3722use strict;
3723use warnings;
3724use Carp qw/croak/;
3725use IO::File;
3726use Git qw/command command_oneline command_noisy
3727           command_output_pipe command_input_pipe command_close_pipe/;
3728
3729sub new {
3730        my $class = shift;
3731        my $git_svn = shift;
3732        my $self = SVN::Delta::Editor->new(@_);
3733        bless $self, $class;
3734        foreach (qw/svn_path c r ra /) {
3735                die "$_ required!\n" unless (defined $git_svn->{$_});
3736                $self->{$_} = $git_svn->{$_};
3737        }
3738        $self->{pool} = SVN::Pool->new;
3739        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3740        $self->{rm} = { };
3741        require Digest::MD5;
3742        return $self;
3743}
3744
3745sub split_path {
3746        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3747}
3748
3749sub repo_path {
3750        (defined $_[1] && length $_[1]) ? $_[1] : ''
3751}
3752
3753sub url_path {
3754        my ($self, $path) = @_;
3755        $self->{ra}->{url} . '/' . $self->repo_path($path);
3756}
3757
3758sub rmdirs {
3759        my ($self, $q) = @_;
3760        my $rm = $self->{rm};
3761        delete $rm->{''}; # we never delete the url we're tracking
3762        return unless %$rm;
3763
3764        foreach (keys %$rm) {
3765                my @d = split m#/#, $_;
3766                my $c = shift @d;
3767                $rm->{$c} = 1;
3768                while (@d) {
3769                        $c .= '/' . shift @d;
3770                        $rm->{$c} = 1;
3771                }
3772        }
3773        delete $rm->{$self->{svn_path}};
3774        delete $rm->{''}; # we never delete the url we're tracking
3775        return unless %$rm;
3776
3777        my ($fh, $ctx) = command_output_pipe(
3778                                   qw/ls-tree --name-only -r -z/, $self->{c});
3779        local $/ = "\0";
3780        while (<$fh>) {
3781                chomp;
3782                my @dn = split m#/#, $_;
3783                while (pop @dn) {
3784                        delete $rm->{join '/', @dn};
3785                }
3786                unless (%$rm) {
3787                        eval { command_close_pipe($fh) };
3788                        return;
3789                }
3790        }
3791        command_close_pipe($fh, $ctx);
3792
3793        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3794        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3795                $self->close_directory($bat->{$d}, $p);
3796                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3797                print "\tD+\t/$d/\n" unless $q;
3798                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3799                delete $bat->{$d};
3800        }
3801}
3802
3803sub open_or_add_dir {
3804        my ($self, $full_path, $baton) = @_;
3805        my $p = SVN::Pool->new;
3806        my $t = $self->{ra}->check_path($full_path, $self->{r}, $p);
3807        $p->clear;
3808        if ($t == $SVN::Node::none) {
3809                return $self->add_directory($full_path, $baton,
3810                                                undef, -1, $self->{pool});
3811        } elsif ($t == $SVN::Node::dir) {
3812                return $self->open_directory($full_path, $baton,
3813                                                $self->{r}, $self->{pool});
3814        }
3815        print STDERR "$full_path already exists in repository at ",
3816                "r$self->{r} and it is not a directory (",
3817                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3818        exit 1;
3819}
3820
3821sub ensure_path {
3822        my ($self, $path) = @_;
3823        my $bat = $self->{bat};
3824        $path = $self->repo_path($path);
3825        return $bat->{''} unless (length $path);
3826        my @p = split m#/+#, $path;
3827        my $c = shift @p;
3828        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3829        while (@p) {
3830                my $c0 = $c;
3831                $c .= '/' . shift @p;
3832                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3833        }
3834        return $bat->{$c};
3835}
3836
3837sub A {
3838        my ($self, $m, $q) = @_;
3839        my ($dir, $file) = split_path($m->{file_b});
3840        my $pbat = $self->ensure_path($dir);
3841        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3842                                        undef, -1);
3843        print "\tA\t$m->{file_b}\n" unless $q;
3844        $self->chg_file($fbat, $m);
3845        $self->close_file($fbat,undef,$self->{pool});
3846}
3847
3848sub C {
3849        my ($self, $m, $q) = @_;
3850        my ($dir, $file) = split_path($m->{file_b});
3851        my $pbat = $self->ensure_path($dir);
3852        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3853                                $self->url_path($m->{file_a}), $self->{r});
3854        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $q;
3855        $self->chg_file($fbat, $m);
3856        $self->close_file($fbat,undef,$self->{pool});
3857}
3858
3859sub delete_entry {
3860        my ($self, $path, $pbat) = @_;
3861        my $rpath = $self->repo_path($path);
3862        my ($dir, $file) = split_path($rpath);
3863        $self->{rm}->{$dir} = 1;
3864        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3865}
3866
3867sub R {
3868        my ($self, $m, $q) = @_;
3869        my ($dir, $file) = split_path($m->{file_b});
3870        my $pbat = $self->ensure_path($dir);
3871        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3872                                $self->url_path($m->{file_a}), $self->{r});
3873        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $q;
3874        $self->chg_file($fbat, $m);
3875        $self->close_file($fbat,undef,$self->{pool});
3876
3877        ($dir, $file) = split_path($m->{file_a});
3878        $pbat = $self->ensure_path($dir);
3879        $self->delete_entry($m->{file_a}, $pbat);
3880}
3881
3882sub M {
3883        my ($self, $m, $q) = @_;
3884        my ($dir, $file) = split_path($m->{file_b});
3885        my $pbat = $self->ensure_path($dir);
3886        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3887                                $pbat,$self->{r},$self->{pool});
3888        print "\t$m->{chg}\t$m->{file_b}\n" unless $q;
3889        $self->chg_file($fbat, $m);
3890        $self->close_file($fbat,undef,$self->{pool});
3891}
3892
3893sub T { shift->M(@_) }
3894
3895sub change_file_prop {
3896        my ($self, $fbat, $pname, $pval) = @_;
3897        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3898}
3899
3900sub chg_file {
3901        my ($self, $fbat, $m) = @_;
3902        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3903                $self->change_file_prop($fbat,'svn:executable','*');
3904        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3905                $self->change_file_prop($fbat,'svn:executable',undef);
3906        }
3907        my $fh = IO::File->new_tmpfile or croak $!;
3908        if ($m->{mode_b} =~ /^120/) {
3909                print $fh 'link ' or croak $!;
3910                $self->change_file_prop($fbat,'svn:special','*');
3911        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3912                $self->change_file_prop($fbat,'svn:special',undef);
3913        }
3914        defined(my $pid = fork) or croak $!;
3915        if (!$pid) {
3916                open STDOUT, '>&', $fh or croak $!;
3917                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3918        }
3919        waitpid $pid, 0;
3920        croak $? if $?;
3921        $fh->flush == 0 or croak $!;
3922        seek $fh, 0, 0 or croak $!;
3923
3924        my $md5 = Digest::MD5->new;
3925        $md5->addfile($fh) or croak $!;
3926        seek $fh, 0, 0 or croak $!;
3927
3928        my $exp = $md5->hexdigest;
3929        my $pool = SVN::Pool->new;
3930        my $atd = $self->apply_textdelta($fbat, undef, $pool);
3931        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3932        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3933        $pool->clear;
3934
3935        close $fh or croak $!;
3936}
3937
3938sub D {
3939        my ($self, $m, $q) = @_;
3940        my ($dir, $file) = split_path($m->{file_b});
3941        my $pbat = $self->ensure_path($dir);
3942        print "\tD\t$m->{file_b}\n" unless $q;
3943        $self->delete_entry($m->{file_b}, $pbat);
3944}
3945
3946sub close_edit {
3947        my ($self) = @_;
3948        my ($p,$bat) = ($self->{pool}, $self->{bat});
3949        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3950                $self->close_directory($bat->{$_}, $p);
3951        }
3952        $self->SUPER::close_edit($p);
3953        $p->clear;
3954}
3955
3956sub abort_edit {
3957        my ($self) = @_;
3958        $self->SUPER::abort_edit($self->{pool});
3959        $self->{pool}->clear;
3960}
3961
3962__END__
3963
3964Data structures:
3965
3966$svn_log hashref (as returned by svn_log_raw)
3967{
3968        fh => file handle of the log file,
3969        state => state of the log file parser (sep/msg/rev/msg_start...)
3970}
3971
3972$log_msg hashref as returned by next_log_entry($svn_log)
3973{
3974        msg => 'whitespace-formatted log entry
3975',                                              # trailing newline is preserved
3976        revision => '8',                        # integer
3977        date => '2004-02-24T17:01:44.108345Z',  # commit date
3978        author => 'committer name'
3979};
3980
3981
3982@mods = array of diff-index line hashes, each element represents one line
3983        of diff-index output
3984
3985diff-index line ($m hash)
3986{
3987        mode_a => first column of diff-index output, no leading ':',
3988        mode_b => second column of diff-index output,
3989        sha1_b => sha1sum of the final blob,
3990        chg => change type [MCRADT],
3991        file_a => original file name of a file (iff chg is 'C' or 'R')
3992        file_b => new/current file name of a file (any chg)
3993}
3994;
3995
3996# retval of read_url_paths{,_all}();
3997$l_map = {
3998        # repository root url
3999        'https://svn.musicpd.org' => {
4000                # repository path               # GIT_SVN_ID
4001                'mpd/trunk'             =>      'trunk',
4002                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
4003        },
4004}
4005
4006Notes:
4007        I don't trust the each() function on unless I created %hash myself
4008        because the internal iterator may not have started at base.