a865a11a3eba6798980d55b061cee593a2f405b6
   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$VERSION = '1.1.1-broken';
  12
  13use Cwd qw/abs_path/;
  14$GIT_DIR = abs_path($ENV{GIT_DIR} || '.git');
  15$ENV{GIT_DIR} = $GIT_DIR;
  16
  17my $LC_ALL = $ENV{LC_ALL};
  18my $TZ = $ENV{TZ};
  19# make sure the svn binary gives consistent output between locales and TZs:
  20$ENV{TZ} = 'UTC';
  21$ENV{LC_ALL} = 'C';
  22$| = 1; # unbuffer STDOUT
  23
  24# If SVN:: library support is added, please make the dependencies
  25# optional and preserve the capability to use the command-line client.
  26# use eval { require SVN::... } to make it lazy load
  27# We don't use any modules not in the standard Perl distribution:
  28use Carp qw/croak/;
  29use IO::File qw//;
  30use File::Basename qw/dirname basename/;
  31use File::Path qw/mkpath/;
  32use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  33use File::Spec qw//;
  34use POSIX qw/strftime/;
  35use IPC::Open3;
  36use Memoize;
  37memoize('revisions_eq');
  38memoize('cmt_metadata');
  39memoize('get_commit_time');
  40
  41my ($SVN_PATH, $SVN, $SVN_LOG, $_use_lib);
  42$_use_lib = 1 unless $ENV{GIT_SVN_NO_LIB};
  43libsvn_load();
  44my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  45my $sha1 = qr/[a-f\d]{40}/;
  46my $sha1_short = qr/[a-f\d]{4,40}/;
  47my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
  48        $_find_copies_harder, $_l, $_cp_similarity, $_cp_remote,
  49        $_repack, $_repack_nr, $_repack_flags,
  50        $_message, $_file, $_follow_parent, $_no_metadata,
  51        $_template, $_shared, $_no_default_regex, $_no_graft_copy,
  52        $_limit, $_verbose, $_incremental, $_oneline, $_l_fmt, $_show_commit,
  53        $_version, $_upgrade, $_authors, $_branch_all_refs, @_opt_m);
  54my (@_branch_from, %tree_map, %users, %rusers, %equiv);
  55my ($_svn_co_url_revs, $_svn_pg_peg_revs);
  56my @repo_path_split_cache;
  57
  58my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
  59                'branch|b=s' => \@_branch_from,
  60                'follow-parent|follow' => \$_follow_parent,
  61                'branch-all-refs|B' => \$_branch_all_refs,
  62                'authors-file|A=s' => \$_authors,
  63                'repack:i' => \$_repack,
  64                'no-metadata' => \$_no_metadata,
  65                'repack-flags|repack-args|repack-opts=s' => \$_repack_flags);
  66
  67my ($_trunk, $_tags, $_branches);
  68my %multi_opts = ( 'trunk|T=s' => \$_trunk,
  69                'tags|t=s' => \$_tags,
  70                'branches|b=s' => \$_branches );
  71my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
  72my %cmt_opts = ( 'edit|e' => \$_edit,
  73                'rmdir' => \$_rmdir,
  74                'find-copies-harder' => \$_find_copies_harder,
  75                'l=i' => \$_l,
  76                'copy-similarity|C=i'=> \$_cp_similarity
  77);
  78
  79# yes, 'native' sets "\n".  Patches to fix this for non-*nix systems welcome:
  80my %EOL = ( CR => "\015", LF => "\012", CRLF => "\015\012", native => "\012" );
  81
  82my %cmd = (
  83        fetch => [ \&fetch, "Download new revisions from SVN",
  84                        { 'revision|r=s' => \$_revision, %fc_opts } ],
  85        init => [ \&init, "Initialize a repo for tracking" .
  86                          " (requires URL argument)",
  87                          \%init_opts ],
  88        commit => [ \&commit, "Commit git revisions to SVN",
  89                        {       'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
  90        'show-ignore' => [ \&show_ignore, "Show svn:ignore listings",
  91                        { 'revision|r=i' => \$_revision } ],
  92        rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
  93                        { 'no-ignore-externals' => \$_no_ignore_ext,
  94                          'copy-remote|remote=s' => \$_cp_remote,
  95                          'upgrade' => \$_upgrade } ],
  96        'graft-branches' => [ \&graft_branches,
  97                        'Detect merges/branches from already imported history',
  98                        { 'merge-rx|m' => \@_opt_m,
  99                          'branch|b=s' => \@_branch_from,
 100                          'branch-all-refs|B' => \$_branch_all_refs,
 101                          'no-default-regex' => \$_no_default_regex,
 102                          'no-graft-copy' => \$_no_graft_copy } ],
 103        'multi-init' => [ \&multi_init,
 104                        'Initialize multiple trees (like git-svnimport)',
 105                        { %multi_opts, %fc_opts } ],
 106        'multi-fetch' => [ \&multi_fetch,
 107                        'Fetch multiple trees (like git-svnimport)',
 108                        \%fc_opts ],
 109        'log' => [ \&show_log, 'Show commit logs',
 110                        { 'limit=i' => \$_limit,
 111                          'revision|r=s' => \$_revision,
 112                          'verbose|v' => \$_verbose,
 113                          'incremental' => \$_incremental,
 114                          'oneline' => \$_oneline,
 115                          'show-commit' => \$_show_commit,
 116                          'authors-file|A=s' => \$_authors,
 117                        } ],
 118        'commit-diff' => [ \&commit_diff, 'Commit a diff between two trees',
 119                        { 'message|m=s' => \$_message,
 120                          'file|F=s' => \$_file,
 121                        %cmt_opts } ],
 122);
 123
 124my $cmd;
 125for (my $i = 0; $i < @ARGV; $i++) {
 126        if (defined $cmd{$ARGV[$i]}) {
 127                $cmd = $ARGV[$i];
 128                splice @ARGV, $i, 1;
 129                last;
 130        }
 131};
 132
 133my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 134
 135read_repo_config(\%opts);
 136my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
 137                                'version|V' => \$_version,
 138                                'id|i=s' => \$GIT_SVN);
 139exit 1 if (!$rv && $cmd ne 'log');
 140
 141set_default_vals();
 142usage(0) if $_help;
 143version() if $_version;
 144usage(1) unless defined $cmd;
 145init_vars();
 146load_authors() if $_authors;
 147load_all_refs() if $_branch_all_refs;
 148svn_compat_check() unless $_use_lib;
 149migration_check() unless $cmd =~ /^(?:init|rebuild|multi-init)$/;
 150$cmd{$cmd}->[0]->(@ARGV);
 151exit 0;
 152
 153####################### primary functions ######################
 154sub usage {
 155        my $exit = shift || 0;
 156        my $fd = $exit ? \*STDERR : \*STDOUT;
 157        print $fd <<"";
 158git-svn - bidirectional operations between a single Subversion tree and git
 159Usage: $0 <command> [options] [arguments]\n
 160
 161        print $fd "Available commands:\n" unless $cmd;
 162
 163        foreach (sort keys %cmd) {
 164                next if $cmd && $cmd ne $_;
 165                print $fd '  ',pack('A13',$_),$cmd{$_}->[1],"\n";
 166                foreach (keys %{$cmd{$_}->[2]}) {
 167                        # prints out arguments as they should be passed:
 168                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 169                        print $fd ' ' x 17, join(', ', map { length $_ > 1 ?
 170                                                        "--$_" : "-$_" }
 171                                                split /\|/,$_)," $x\n";
 172                }
 173        }
 174        print $fd <<"";
 175\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 176arbitrary identifier if you're tracking multiple SVN branches/repositories in
 177one git repository and want to keep them separate.  See git-svn(1) for more
 178information.
 179
 180        exit $exit;
 181}
 182
 183sub version {
 184        print "git-svn version $VERSION\n";
 185        exit 0;
 186}
 187
 188sub rebuild {
 189        if (quiet_run(qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0")) {
 190                copy_remote_ref();
 191        }
 192        $SVN_URL = shift or undef;
 193        my $newest_rev = 0;
 194        if ($_upgrade) {
 195                sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD");
 196        } else {
 197                check_upgrade_needed();
 198        }
 199
 200        my $pid = open(my $rev_list,'-|');
 201        defined $pid or croak $!;
 202        if ($pid == 0) {
 203                exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!;
 204        }
 205        my $latest;
 206        while (<$rev_list>) {
 207                chomp;
 208                my $c = $_;
 209                croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 210                my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
 211                next if (!@commit); # skip merges
 212                my ($url, $rev, $uuid) = extract_metadata($commit[$#commit]);
 213                if (!$rev || !$uuid) {
 214                        croak "Unable to extract revision or UUID from ",
 215                                "$c, $commit[$#commit]\n";
 216                }
 217
 218                # if we merged or otherwise started elsewhere, this is
 219                # how we break out of it
 220                next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
 221                next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
 222
 223                unless (defined $latest) {
 224                        if (!$SVN_URL && !$url) {
 225                                croak "SVN repository location required: $url\n";
 226                        }
 227                        $SVN_URL ||= $url;
 228                        $SVN_UUID ||= $uuid;
 229                        setup_git_svn();
 230                        $latest = $rev;
 231                }
 232                revdb_set($REVDB, $rev, $c);
 233                print "r$rev = $c\n";
 234                $newest_rev = $rev if ($rev > $newest_rev);
 235        }
 236        close $rev_list or croak $?;
 237
 238        goto out if $_use_lib;
 239        if (!chdir $SVN_WC) {
 240                svn_cmd_checkout($SVN_URL, $latest, $SVN_WC);
 241                chdir $SVN_WC or croak $!;
 242        }
 243
 244        $pid = fork;
 245        defined $pid or croak $!;
 246        if ($pid == 0) {
 247                my @svn_up = qw(svn up);
 248                push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 249                sys(@svn_up,"-r$newest_rev");
 250                $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 251                index_changes();
 252                exec('git-write-tree') or croak $!;
 253        }
 254        waitpid $pid, 0;
 255        croak $? if $?;
 256out:
 257        if ($_upgrade) {
 258                print STDERR <<"";
 259Keeping deprecated refs/head/$GIT_SVN-HEAD for now.  Please remove it
 260when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN
 261
 262        }
 263}
 264
 265sub init {
 266        $SVN_URL = shift or die "SVN repository location required " .
 267                                "as a command-line argument\n";
 268        $SVN_URL =~ s!/+$!!; # strip trailing slash
 269        unless (-d $GIT_DIR) {
 270                my @init_db = ('git-init-db');
 271                push @init_db, "--template=$_template" if defined $_template;
 272                push @init_db, "--shared" if defined $_shared;
 273                sys(@init_db);
 274        }
 275        setup_git_svn();
 276}
 277
 278sub fetch {
 279        check_upgrade_needed();
 280        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 281        my $ret = $_use_lib ? fetch_lib(@_) : fetch_cmd(@_);
 282        if ($ret->{commit} && quiet_run(qw(git-rev-parse --verify
 283                                                refs/heads/master^0))) {
 284                sys(qw(git-update-ref refs/heads/master),$ret->{commit});
 285        }
 286        return $ret;
 287}
 288
 289sub fetch_cmd {
 290        my (@parents) = @_;
 291        my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 292        unless ($_revision) {
 293                $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 294        }
 295        push @log_args, "-r$_revision";
 296        push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 297
 298        my $svn_log = svn_log_raw(@log_args);
 299
 300        my $base = next_log_entry($svn_log) or croak "No base revision!\n";
 301        # don't need last_revision from grab_base_rev() because
 302        # user could've specified a different revision to skip (they
 303        # didn't want to import certain revisions into git for whatever
 304        # reason, so trust $base->{revision} instead.
 305        my (undef, $last_commit) = svn_grab_base_rev();
 306        unless (-d $SVN_WC) {
 307                svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC);
 308                chdir $SVN_WC or croak $!;
 309                read_uuid();
 310                $last_commit = git_commit($base, @parents);
 311                assert_tree($last_commit);
 312        } else {
 313                chdir $SVN_WC or croak $!;
 314                read_uuid();
 315                # looks like a user manually cp'd and svn switch'ed
 316                unless ($last_commit) {
 317                        sys(qw/svn revert -R ./);
 318                        assert_svn_wc_clean($base->{revision});
 319                        $last_commit = git_commit($base, @parents);
 320                        assert_tree($last_commit);
 321                }
 322        }
 323        my @svn_up = qw(svn up);
 324        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 325        my $last = $base;
 326        while (my $log_msg = next_log_entry($svn_log)) {
 327                if ($last->{revision} >= $log_msg->{revision}) {
 328                        croak "Out of order: last >= current: ",
 329                                "$last->{revision} >= $log_msg->{revision}\n";
 330                }
 331                # Revert is needed for cases like:
 332                # https://svn.musicpd.org/Jamming/trunk (r166:167), but
 333                # I can't seem to reproduce something like that on a test...
 334                sys(qw/svn revert -R ./);
 335                assert_svn_wc_clean($last->{revision});
 336                sys(@svn_up,"-r$log_msg->{revision}");
 337                $last_commit = git_commit($log_msg, $last_commit, @parents);
 338                $last = $log_msg;
 339        }
 340        close $svn_log->{fh};
 341        $last->{commit} = $last_commit;
 342        return $last;
 343}
 344
 345sub fetch_lib {
 346        my (@parents) = @_;
 347        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 348        my $repo;
 349        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 350        $SVN_LOG ||= libsvn_connect($repo);
 351        $SVN ||= libsvn_connect($repo);
 352        my ($last_rev, $last_commit) = svn_grab_base_rev();
 353        my ($base, $head) = libsvn_parse_revision($last_rev);
 354        if ($base > $head) {
 355                return { revision => $last_rev, commit => $last_commit }
 356        }
 357        my $index = set_index($GIT_SVN_INDEX);
 358
 359        # limit ourselves and also fork() since get_log won't release memory
 360        # after processing a revision and SVN stuff seems to leak
 361        my $inc = 1000;
 362        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
 363        read_uuid();
 364        if (defined $last_commit) {
 365                unless (-e $GIT_SVN_INDEX) {
 366                        sys(qw/git-read-tree/, $last_commit);
 367                }
 368                chomp (my $x = `git-write-tree`);
 369                my ($y) = (`git-cat-file commit $last_commit`
 370                                                        =~ /^tree ($sha1)/m);
 371                if ($y ne $x) {
 372                        unlink $GIT_SVN_INDEX or croak $!;
 373                        sys(qw/git-read-tree/, $last_commit);
 374                }
 375                chomp ($x = `git-write-tree`);
 376                if ($y ne $x) {
 377                        print STDERR "trees ($last_commit) $y != $x\n",
 378                                 "Something is seriously wrong...\n";
 379                }
 380        }
 381        while (1) {
 382                # fork, because using SVN::Pool with get_log() still doesn't
 383                # seem to help enough to keep memory usage down.
 384                defined(my $pid = fork) or croak $!;
 385                if (!$pid) {
 386                        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
 387
 388                        # Yes I'm perfectly aware that the fourth argument
 389                        # below is the limit revisions number.  Unfortunately
 390                        # performance sucks with it enabled, so it's much
 391                        # faster to fetch revision ranges instead of relying
 392                        # on the limiter.
 393                        libsvn_get_log($SVN_LOG, '/'.$SVN_PATH,
 394                                        $min, $max, 0, 1, 1,
 395                                sub {
 396                                        my $log_msg;
 397                                        if ($last_commit) {
 398                                                $log_msg = libsvn_fetch(
 399                                                        $last_commit, @_);
 400                                                $last_commit = git_commit(
 401                                                        $log_msg,
 402                                                        $last_commit,
 403                                                        @parents);
 404                                        } else {
 405                                                $log_msg = libsvn_new_tree(@_);
 406                                                $last_commit = git_commit(
 407                                                        $log_msg, @parents);
 408                                        }
 409                                });
 410                        exit 0;
 411                }
 412                waitpid $pid, 0;
 413                croak $? if $?;
 414                ($last_rev, $last_commit) = svn_grab_base_rev();
 415                last if ($max >= $head);
 416                $min = $max + 1;
 417                $max += $inc;
 418                $max = $head if ($max > $head);
 419        }
 420        restore_index($index);
 421        return { revision => $last_rev, commit => $last_commit };
 422}
 423
 424sub commit {
 425        my (@commits) = @_;
 426        check_upgrade_needed();
 427        if ($_stdin || !@commits) {
 428                print "Reading from stdin...\n";
 429                @commits = ();
 430                while (<STDIN>) {
 431                        if (/\b($sha1_short)\b/o) {
 432                                unshift @commits, $1;
 433                        }
 434                }
 435        }
 436        my @revs;
 437        foreach my $c (@commits) {
 438                chomp(my @tmp = safe_qx('git-rev-parse',$c));
 439                if (scalar @tmp == 1) {
 440                        push @revs, $tmp[0];
 441                } elsif (scalar @tmp > 1) {
 442                        push @revs, reverse (safe_qx('git-rev-list',@tmp));
 443                } else {
 444                        die "Failed to rev-parse $c\n";
 445                }
 446        }
 447        chomp @revs;
 448        $_use_lib ? commit_lib(@revs) : commit_cmd(@revs);
 449        print "Done committing ",scalar @revs," revisions to SVN\n";
 450}
 451
 452sub commit_cmd {
 453        my (@revs) = @_;
 454
 455        chdir $SVN_WC or croak "Unable to chdir $SVN_WC: $!\n";
 456        my $info = svn_info('.');
 457        my $fetched = fetch();
 458        if ($info->{Revision} != $fetched->{revision}) {
 459                print STDERR "There are new revisions that were fetched ",
 460                                "and need to be merged (or acknowledged) ",
 461                                "before committing.\n";
 462                exit 1;
 463        }
 464        $info = svn_info('.');
 465        read_uuid($info);
 466        my $last = $fetched;
 467        foreach my $c (@revs) {
 468                my $mods = svn_checkout_tree($last, $c);
 469                if (scalar @$mods == 0) {
 470                        print "Skipping, no changes detected\n";
 471                        next;
 472                }
 473                $last = svn_commit_tree($last, $c);
 474        }
 475}
 476
 477sub commit_lib {
 478        my (@revs) = @_;
 479        my ($r_last, $cmt_last) = svn_grab_base_rev();
 480        defined $r_last or die "Must have an existing revision to commit\n";
 481        my $fetched = fetch();
 482        if ($r_last != $fetched->{revision}) {
 483                print STDERR "There are new revisions that were fetched ",
 484                                "and need to be merged (or acknowledged) ",
 485                                "before committing.\n",
 486                                "last rev: $r_last\n",
 487                                " current: $fetched->{revision}\n";
 488                exit 1;
 489        }
 490        read_uuid();
 491        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 492        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
 493
 494        set_svn_commit_env();
 495        foreach my $c (@revs) {
 496                my $log_msg = get_commit_message($c, $commit_msg);
 497
 498                # fork for each commit because there's a memory leak I
 499                # can't track down... (it's probably in the SVN code)
 500                defined(my $pid = open my $fh, '-|') or croak $!;
 501                if (!$pid) {
 502                        my $ed = SVN::Git::Editor->new(
 503                                        {       r => $r_last,
 504                                                ra => $SVN,
 505                                                c => $c,
 506                                                svn_path => $SVN_PATH
 507                                        },
 508                                        $SVN->get_commit_editor(
 509                                                $log_msg->{msg},
 510                                                sub {
 511                                                        libsvn_commit_cb(
 512                                                                @_, $c,
 513                                                                $log_msg->{msg},
 514                                                                $r_last,
 515                                                                $cmt_last)
 516                                                },
 517                                                @lock)
 518                                        );
 519                        my $mods = libsvn_checkout_tree($cmt_last, $c, $ed);
 520                        if (@$mods == 0) {
 521                                print "No changes\nr$r_last = $cmt_last\n";
 522                                $ed->abort_edit;
 523                        } else {
 524                                $ed->close_edit;
 525                        }
 526                        exit 0;
 527                }
 528                my ($r_new, $cmt_new, $no);
 529                while (<$fh>) {
 530                        print $_;
 531                        chomp;
 532                        if (/^r(\d+) = ($sha1)$/o) {
 533                                ($r_new, $cmt_new) = ($1, $2);
 534                        } elsif ($_ eq 'No changes') {
 535                                $no = 1;
 536                        }
 537                }
 538                close $fh or croak $?;
 539                if (! defined $r_new && ! defined $cmt_new) {
 540                        unless ($no) {
 541                                die "Failed to parse revision information\n";
 542                        }
 543                } else {
 544                        ($r_last, $cmt_last) = ($r_new, $cmt_new);
 545                }
 546        }
 547        $ENV{LC_ALL} = 'C';
 548        unlink $commit_msg;
 549}
 550
 551sub show_ignore {
 552        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 553        $_use_lib ? show_ignore_lib() : show_ignore_cmd();
 554}
 555
 556sub show_ignore_cmd {
 557        require File::Find or die $!;
 558        if (defined $_revision) {
 559                die "-r/--revision option doesn't work unless the Perl SVN ",
 560                        "libraries are used\n";
 561        }
 562        chdir $SVN_WC or croak $!;
 563        my %ign;
 564        File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 565                s#^\./##;
 566                @{$ign{$_}} = svn_propget_base('svn:ignore', $_);
 567                }}, no_chdir=>1},'.');
 568
 569        print "\n# /\n";
 570        foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 571        delete $ign{'.'};
 572        foreach my $i (sort keys %ign) {
 573                print "\n# ",$i,"\n";
 574                foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 575        }
 576}
 577
 578sub show_ignore_lib {
 579        my $repo;
 580        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 581        $SVN ||= libsvn_connect($repo);
 582        my $r = defined $_revision ? $_revision : $SVN->get_latest_revnum;
 583        libsvn_traverse_ignore(\*STDOUT, $SVN_PATH, $r);
 584}
 585
 586sub graft_branches {
 587        my $gr_file = "$GIT_DIR/info/grafts";
 588        my ($grafts, $comments) = read_grafts($gr_file);
 589        my $gr_sha1;
 590
 591        if (%$grafts) {
 592                # temporarily disable our grafts file to make this idempotent
 593                chomp($gr_sha1 = safe_qx(qw/git-hash-object -w/,$gr_file));
 594                rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
 595        }
 596
 597        my $l_map = read_url_paths();
 598        my @re = map { qr/$_/is } @_opt_m if @_opt_m;
 599        unless ($_no_default_regex) {
 600                push @re, (qr/\b(?:merge|merging|merged)\s+with\s+([\w\.\-]+)/i,
 601                        qr/\b(?:merge|merging|merged)\s+([\w\.\-]+)/i,
 602                        qr/\b(?:from|of)\s+([\w\.\-]+)/i );
 603        }
 604        foreach my $u (keys %$l_map) {
 605                if (@re) {
 606                        foreach my $p (keys %{$l_map->{$u}}) {
 607                                graft_merge_msg($grafts,$l_map,$u,$p,@re);
 608                        }
 609                }
 610                unless ($_no_graft_copy) {
 611                        if ($_use_lib) {
 612                                graft_file_copy_lib($grafts,$l_map,$u);
 613                        } else {
 614                                graft_file_copy_cmd($grafts,$l_map,$u);
 615                        }
 616                }
 617        }
 618        graft_tree_joins($grafts);
 619
 620        write_grafts($grafts, $comments, $gr_file);
 621        unlink "$gr_file~$gr_sha1" if $gr_sha1;
 622}
 623
 624sub multi_init {
 625        my $url = shift;
 626        $_trunk ||= 'trunk';
 627        $_trunk =~ s#/+$##;
 628        $url =~ s#/+$## if $url;
 629        if ($_trunk !~ m#^[a-z\+]+://#) {
 630                $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#);
 631                unless ($url) {
 632                        print STDERR "E: '$_trunk' is not a complete URL ",
 633                                "and a separate URL is not specified\n";
 634                        exit 1;
 635                }
 636                $_trunk = $url . $_trunk;
 637        }
 638        if ($GIT_SVN eq 'git-svn') {
 639                print "GIT_SVN_ID set to 'trunk' for $_trunk\n";
 640                $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
 641        }
 642        init_vars();
 643        init($_trunk);
 644        complete_url_ls_init($url, $_branches, '--branches/-b', '');
 645        complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/');
 646}
 647
 648sub multi_fetch {
 649        # try to do trunk first, since branches/tags
 650        # may be descended from it.
 651        if (-e "$GIT_DIR/svn/trunk/info/url") {
 652                fetch_child_id('trunk', @_);
 653        }
 654        rec_fetch('', "$GIT_DIR/svn", @_);
 655}
 656
 657sub show_log {
 658        my (@args) = @_;
 659        my ($r_min, $r_max);
 660        my $r_last = -1; # prevent dupes
 661        rload_authors() if $_authors;
 662        if (defined $TZ) {
 663                $ENV{TZ} = $TZ;
 664        } else {
 665                delete $ENV{TZ};
 666        }
 667        if (defined $_revision) {
 668                if ($_revision =~ /^(\d+):(\d+)$/) {
 669                        ($r_min, $r_max) = ($1, $2);
 670                } elsif ($_revision =~ /^\d+$/) {
 671                        $r_min = $r_max = $_revision;
 672                } else {
 673                        print STDERR "-r$_revision is not supported, use ",
 674                                "standard \'git log\' arguments instead\n";
 675                        exit 1;
 676                }
 677        }
 678
 679        my $pid = open(my $log,'-|');
 680        defined $pid or croak $!;
 681        if (!$pid) {
 682                exec(git_svn_log_cmd($r_min,$r_max), @args) or croak $!;
 683        }
 684        setup_pager();
 685        my (@k, $c, $d);
 686
 687        while (<$log>) {
 688                if (/^commit ($sha1_short)/o) {
 689                        my $cmt = $1;
 690                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
 691                                $r_last = $c->{r};
 692                                process_commit($c, $r_min, $r_max, \@k) or
 693                                                                goto out;
 694                        }
 695                        $d = undef;
 696                        $c = { c => $cmt };
 697                } elsif (/^author (.+) (\d+) ([\-\+]?\d+)$/) {
 698                        get_author_info($c, $1, $2, $3);
 699                } elsif (/^(?:tree|parent|committer) /) {
 700                        # ignore
 701                } elsif (/^:\d{6} \d{6} $sha1_short/o) {
 702                        push @{$c->{raw}}, $_;
 703                } elsif (/^diff /) {
 704                        $d = 1;
 705                        push @{$c->{diff}}, $_;
 706                } elsif ($d) {
 707                        push @{$c->{diff}}, $_;
 708                } elsif (/^    (git-svn-id:.+)$/) {
 709                        (undef, $c->{r}, undef) = extract_metadata($1);
 710                } elsif (s/^    //) {
 711                        push @{$c->{l}}, $_;
 712                }
 713        }
 714        if ($c && defined $c->{r} && $c->{r} != $r_last) {
 715                $r_last = $c->{r};
 716                process_commit($c, $r_min, $r_max, \@k);
 717        }
 718        if (@k) {
 719                my $swap = $r_max;
 720                $r_max = $r_min;
 721                $r_min = $swap;
 722                process_commit($_, $r_min, $r_max) foreach reverse @k;
 723        }
 724out:
 725        close $log;
 726        print '-' x72,"\n" unless $_incremental || $_oneline;
 727}
 728
 729sub commit_diff_usage {
 730        print STDERR "Usage: $0 commit-diff <tree-ish> <tree-ish> [<URL>]\n";
 731        exit 1
 732}
 733
 734sub commit_diff {
 735        if (!$_use_lib) {
 736                print STDERR "commit-diff must be used with SVN libraries\n";
 737                exit 1;
 738        }
 739        my $ta = shift or commit_diff_usage();
 740        my $tb = shift or commit_diff_usage();
 741        if (!eval { $SVN_URL = shift || file_to_s("$GIT_SVN_DIR/info/url") }) {
 742                print STDERR "Needed URL or usable git-svn id command-line\n";
 743                commit_diff_usage();
 744        }
 745        if (defined $_message && defined $_file) {
 746                print STDERR "Both --message/-m and --file/-F specified ",
 747                                "for the commit message.\n",
 748                                "I have no idea what you mean\n";
 749                exit 1;
 750        }
 751        if (defined $_file) {
 752                $_message = file_to_s($_message);
 753        } else {
 754                $_message ||= get_commit_message($tb,
 755                                        "$GIT_DIR/.svn-commit.tmp.$$")->{msg};
 756        }
 757        my $repo;
 758        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 759        $SVN_LOG ||= libsvn_connect($repo);
 760        $SVN ||= libsvn_connect($repo);
 761        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 762        my $ed = SVN::Git::Editor->new({        r => $SVN->get_latest_revnum,
 763                                                ra => $SVN, c => $tb,
 764                                                svn_path => $SVN_PATH
 765                                        },
 766                                $SVN->get_commit_editor($_message,
 767                                        sub {print "Committed $_[0]\n"},@lock)
 768                                );
 769        my $mods = libsvn_checkout_tree($ta, $tb, $ed);
 770        if (@$mods == 0) {
 771                print "No changes\n$ta == $tb\n";
 772                $ed->abort_edit;
 773        } else {
 774                $ed->close_edit;
 775        }
 776}
 777
 778########################### utility functions #########################
 779
 780sub cmt_showable {
 781        my ($c) = @_;
 782        return 1 if defined $c->{r};
 783        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
 784                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
 785                my @msg = safe_qx(qw/git-cat-file commit/, $c->{c});
 786                shift @msg while ($msg[0] ne "\n");
 787                shift @msg;
 788                @{$c->{l}} = grep !/^git-svn-id: /, @msg;
 789
 790                (undef, $c->{r}, undef) = extract_metadata(
 791                                (grep(/^git-svn-id: /, @msg))[-1]);
 792        }
 793        return defined $c->{r};
 794}
 795
 796sub git_svn_log_cmd {
 797        my ($r_min, $r_max) = @_;
 798        my @cmd = (qw/git-log --abbrev-commit --pretty=raw
 799                        --default/, "refs/remotes/$GIT_SVN");
 800        push @cmd, '--summary' if $_verbose;
 801        return @cmd unless defined $r_max;
 802        if ($r_max == $r_min) {
 803                push @cmd, '--max-count=1';
 804                if (my $c = revdb_get($REVDB, $r_max)) {
 805                        push @cmd, $c;
 806                }
 807        } else {
 808                my ($c_min, $c_max);
 809                $c_max = revdb_get($REVDB, $r_max);
 810                $c_min = revdb_get($REVDB, $r_min);
 811                if ($c_min && $c_max) {
 812                        if ($r_max > $r_max) {
 813                                push @cmd, "$c_min..$c_max";
 814                        } else {
 815                                push @cmd, "$c_max..$c_min";
 816                        }
 817                } elsif ($r_max > $r_min) {
 818                        push @cmd, $c_max;
 819                } else {
 820                        push @cmd, $c_min;
 821                }
 822        }
 823        return @cmd;
 824}
 825
 826sub fetch_child_id {
 827        my $id = shift;
 828        print "Fetching $id\n";
 829        my $ref = "$GIT_DIR/refs/remotes/$id";
 830        defined(my $pid = open my $fh, '-|') or croak $!;
 831        if (!$pid) {
 832                $_repack = undef;
 833                $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 834                init_vars();
 835                fetch(@_);
 836                exit 0;
 837        }
 838        while (<$fh>) {
 839                print $_;
 840                check_repack() if (/^r\d+ = $sha1/);
 841        }
 842        close $fh or croak $?;
 843}
 844
 845sub rec_fetch {
 846        my ($pfx, $p, @args) = @_;
 847        my @dir;
 848        foreach (sort <$p/*>) {
 849                if (-r "$_/info/url") {
 850                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
 851                        my $id = $pfx . basename $_;
 852                        next if $id eq 'trunk';
 853                        fetch_child_id($id, @args);
 854                } elsif (-d $_) {
 855                        push @dir, $_;
 856                }
 857        }
 858        foreach (@dir) {
 859                my $x = $_;
 860                $x =~ s!^\Q$GIT_DIR\E/svn/!!;
 861                rec_fetch($x, $_);
 862        }
 863}
 864
 865sub complete_url_ls_init {
 866        my ($url, $var, $switch, $pfx) = @_;
 867        unless ($var) {
 868                print STDERR "W: $switch not specified\n";
 869                return;
 870        }
 871        $var =~ s#/+$##;
 872        if ($var !~ m#^[a-z\+]+://#) {
 873                $var = '/' . $var if ($var !~ m#^/#);
 874                unless ($url) {
 875                        print STDERR "E: '$var' is not a complete URL ",
 876                                "and a separate URL is not specified\n";
 877                        exit 1;
 878                }
 879                $var = $url . $var;
 880        }
 881        chomp(my @ls = $_use_lib ? libsvn_ls_fullurl($var)
 882                                : safe_qx(qw/svn ls --non-interactive/, $var));
 883        my $old = $GIT_SVN;
 884        defined(my $pid = fork) or croak $!;
 885        if (!$pid) {
 886                foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
 887                        $u =~ s#/+$##;
 888                        if ($u !~ m!\Q$var\E/(.+)$!) {
 889                                print STDERR "W: Unrecognized URL: $u\n";
 890                                die "This should never happen\n";
 891                        }
 892                        my $id = $pfx.$1;
 893                        print "init $u => $id\n";
 894                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 895                        init_vars();
 896                        init($u);
 897                }
 898                exit 0;
 899        }
 900        waitpid $pid, 0;
 901        croak $? if $?;
 902}
 903
 904sub common_prefix {
 905        my $paths = shift;
 906        my %common;
 907        foreach (@$paths) {
 908                my @tmp = split m#/#, $_;
 909                my $p = '';
 910                while (my $x = shift @tmp) {
 911                        $p .= "/$x";
 912                        $common{$p} ||= 0;
 913                        $common{$p}++;
 914                }
 915        }
 916        foreach (sort {length $b <=> length $a} keys %common) {
 917                if ($common{$_} == @$paths) {
 918                        return $_;
 919                }
 920        }
 921        return '';
 922}
 923
 924# grafts set here are 'stronger' in that they're based on actual tree
 925# matches, and won't be deleted from merge-base checking in write_grafts()
 926sub graft_tree_joins {
 927        my $grafts = shift;
 928        map_tree_joins() if (@_branch_from && !%tree_map);
 929        return unless %tree_map;
 930
 931        git_svn_each(sub {
 932                my $i = shift;
 933                defined(my $pid = open my $fh, '-|') or croak $!;
 934                if (!$pid) {
 935                        exec qw/git-rev-list --pretty=raw/,
 936                                        "refs/remotes/$i" or croak $!;
 937                }
 938                while (<$fh>) {
 939                        next unless /^commit ($sha1)$/o;
 940                        my $c = $1;
 941                        my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
 942                        next unless $tree_map{$t};
 943
 944                        my $l;
 945                        do {
 946                                $l = readline $fh;
 947                        } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
 948
 949                        my ($s, $tz) = ($1, $2);
 950                        if ($tz =~ s/^\+//) {
 951                                $s += tz_to_s_offset($tz);
 952                        } elsif ($tz =~ s/^\-//) {
 953                                $s -= tz_to_s_offset($tz);
 954                        }
 955
 956                        my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
 957
 958                        foreach my $p (@{$tree_map{$t}}) {
 959                                next if $p eq $c;
 960                                my $mb = eval {
 961                                        safe_qx('git-merge-base', $c, $p)
 962                                };
 963                                next unless ($@ || $?);
 964                                if (defined $r_a) {
 965                                        # see if SVN says it's a relative
 966                                        my ($url_b, $r_b, $uuid_b) =
 967                                                        cmt_metadata($p);
 968                                        next if (defined $url_b &&
 969                                                        defined $url_a &&
 970                                                        ($url_a eq $url_b) &&
 971                                                        ($uuid_a eq $uuid_b));
 972                                        if ($uuid_a eq $uuid_b) {
 973                                                if ($r_b < $r_a) {
 974                                                        $grafts->{$c}->{$p} = 2;
 975                                                        next;
 976                                                } elsif ($r_b > $r_a) {
 977                                                        $grafts->{$p}->{$c} = 2;
 978                                                        next;
 979                                                }
 980                                        }
 981                                }
 982                                my $ct = get_commit_time($p);
 983                                if ($ct < $s) {
 984                                        $grafts->{$c}->{$p} = 2;
 985                                } elsif ($ct > $s) {
 986                                        $grafts->{$p}->{$c} = 2;
 987                                }
 988                                # what should we do when $ct == $s ?
 989                        }
 990                }
 991                close $fh or croak $?;
 992        });
 993}
 994
 995# this isn't funky-filename safe, but good enough for now...
 996sub graft_file_copy_cmd {
 997        my ($grafts, $l_map, $u) = @_;
 998        my $paths = $l_map->{$u};
 999        my $pfx = common_prefix([keys %$paths]);
1000        $SVN_URL ||= $u.$pfx;
1001        my $pid = open my $fh, '-|';
1002        defined $pid or croak $!;
1003        unless ($pid) {
1004                my @exec = qw/svn log -v/;
1005                push @exec, "-r$_revision" if defined $_revision;
1006                exec @exec, $u.$pfx or croak $!;
1007        }
1008        my ($r, $mp) = (undef, undef);
1009        while (<$fh>) {
1010                chomp;
1011                if (/^\-{72}$/) {
1012                        $mp = $r = undef;
1013                } elsif (/^r(\d+) \| /) {
1014                        $r = $1 unless defined $r;
1015                } elsif (/^Changed paths:/) {
1016                        $mp = 1;
1017                } elsif ($mp && m#^   [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) {
1018                        my ($p1, $p0, $r0) = ($1, $2, $3);
1019                        my $c = find_graft_path_commit($paths, $p1, $r);
1020                        next unless $c;
1021                        find_graft_path_parents($grafts, $paths, $c, $p0, $r0);
1022                }
1023        }
1024}
1025
1026sub graft_file_copy_lib {
1027        my ($grafts, $l_map, $u) = @_;
1028        my $tree_paths = $l_map->{$u};
1029        my $pfx = common_prefix([keys %$tree_paths]);
1030        my ($repo, $path) = repo_path_split($u.$pfx);
1031        $SVN_LOG ||= libsvn_connect($repo);
1032        $SVN ||= libsvn_connect($repo);
1033
1034        my ($base, $head) = libsvn_parse_revision();
1035        my $inc = 1000;
1036        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
1037        my $eh = $SVN::Error::handler;
1038        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
1039        while (1) {
1040                my $pool = SVN::Pool->new;
1041                libsvn_get_log($SVN_LOG, "/$path", $min, $max, 0, 1, 1,
1042                        sub {
1043                                libsvn_graft_file_copies($grafts, $tree_paths,
1044                                                        $path, @_);
1045                        }, $pool);
1046                $pool->clear;
1047                last if ($max >= $head);
1048                $min = $max + 1;
1049                $max += $inc;
1050                $max = $head if ($max > $head);
1051        }
1052        $SVN::Error::handler = $eh;
1053}
1054
1055sub process_merge_msg_matches {
1056        my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
1057        my (@strong, @weak);
1058        foreach (@matches) {
1059                # merging with ourselves is not interesting
1060                next if $_ eq $p;
1061                if ($l_map->{$u}->{$_}) {
1062                        push @strong, $_;
1063                } else {
1064                        push @weak, $_;
1065                }
1066        }
1067        foreach my $w (@weak) {
1068                last if @strong;
1069                # no exact match, use branch name as regexp.
1070                my $re = qr/\Q$w\E/i;
1071                foreach (keys %{$l_map->{$u}}) {
1072                        if (/$re/) {
1073                                push @strong, $l_map->{$u}->{$_};
1074                                last;
1075                        }
1076                }
1077                last if @strong;
1078                $w = basename($w);
1079                $re = qr/\Q$w\E/i;
1080                foreach (keys %{$l_map->{$u}}) {
1081                        if (/$re/) {
1082                                push @strong, $l_map->{$u}->{$_};
1083                                last;
1084                        }
1085                }
1086        }
1087        my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
1088                                        \s(?:[a-f\d\-]+)$/xsm);
1089        unless (defined $rev) {
1090                ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
1091                                        \@(?:[a-f\d\-]+)/xsm);
1092                return unless defined $rev;
1093        }
1094        foreach my $m (@strong) {
1095                my ($r0, $s0) = find_rev_before($rev, $m, 1);
1096                $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
1097        }
1098}
1099
1100sub graft_merge_msg {
1101        my ($grafts, $l_map, $u, $p, @re) = @_;
1102
1103        my $x = $l_map->{$u}->{$p};
1104        my $rl = rev_list_raw($x);
1105        while (my $c = next_rev_list_entry($rl)) {
1106                foreach my $re (@re) {
1107                        my (@br) = ($c->{m} =~ /$re/g);
1108                        next unless @br;
1109                        process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
1110                }
1111        }
1112}
1113
1114sub read_uuid {
1115        return if $SVN_UUID;
1116        if ($_use_lib) {
1117                my $pool = SVN::Pool->new;
1118                $SVN_UUID = $SVN->get_uuid($pool);
1119                $pool->clear;
1120        } else {
1121                my $info = shift || svn_info('.');
1122                $SVN_UUID = $info->{'Repository UUID'} or
1123                                        croak "Repository UUID unreadable\n";
1124        }
1125}
1126
1127sub quiet_run {
1128        my $pid = fork;
1129        defined $pid or croak $!;
1130        if (!$pid) {
1131                open my $null, '>', '/dev/null' or croak $!;
1132                open STDERR, '>&', $null or croak $!;
1133                open STDOUT, '>&', $null or croak $!;
1134                exec @_ or croak $!;
1135        }
1136        waitpid $pid, 0;
1137        return $?;
1138}
1139
1140sub repo_path_split {
1141        my $full_url = shift;
1142        $full_url =~ s#/+$##;
1143
1144        foreach (@repo_path_split_cache) {
1145                if ($full_url =~ s#$_##) {
1146                        my $u = $1;
1147                        $full_url =~ s#^/+##;
1148                        return ($u, $full_url);
1149                }
1150        }
1151
1152        my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i);
1153        $path =~ s#^/+##;
1154        my @paths = split(m#/+#, $path);
1155
1156        if ($_use_lib) {
1157                while (1) {
1158                        $SVN = libsvn_connect($url);
1159                        last if (defined $SVN &&
1160                                defined eval { $SVN->get_latest_revnum });
1161                        my $n = shift @paths || last;
1162                        $url .= "/$n";
1163                }
1164        } else {
1165                while (quiet_run(qw/svn ls --non-interactive/, $url)) {
1166                        my $n = shift @paths || last;
1167                        $url .= "/$n";
1168                }
1169        }
1170        push @repo_path_split_cache, qr/^(\Q$url\E)/;
1171        $path = join('/',@paths);
1172        return ($url, $path);
1173}
1174
1175sub setup_git_svn {
1176        defined $SVN_URL or croak "SVN repository location required\n";
1177        unless (-d $GIT_DIR) {
1178                croak "GIT_DIR=$GIT_DIR does not exist!\n";
1179        }
1180        mkpath([$GIT_SVN_DIR]);
1181        mkpath(["$GIT_SVN_DIR/info"]);
1182        open my $fh, '>>',$REVDB or croak $!;
1183        close $fh;
1184        s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1185
1186}
1187
1188sub assert_svn_wc_clean {
1189        return if $_use_lib;
1190        my ($svn_rev) = @_;
1191        croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
1192        my $lcr = svn_info('.')->{'Last Changed Rev'};
1193        if ($svn_rev != $lcr) {
1194                print STDERR "Checking for copy-tree ... ";
1195                my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
1196                                                "-r$lcr:$svn_rev")));
1197                if (@diff) {
1198                        croak "Nope!  Expected r$svn_rev, got r$lcr\n";
1199                } else {
1200                        print STDERR "OK!\n";
1201                }
1202        }
1203        my @status = grep(!/^Performing status on external/,(`svn status`));
1204        @status = grep(!/^\s*$/,@status);
1205        if (scalar @status) {
1206                print STDERR "Tree ($SVN_WC) is not clean:\n";
1207                print STDERR $_ foreach @status;
1208                croak;
1209        }
1210}
1211
1212sub get_tree_from_treeish {
1213        my ($treeish) = @_;
1214        croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1215        chomp(my $type = `git-cat-file -t $treeish`);
1216        my $expected;
1217        while ($type eq 'tag') {
1218                chomp(($treeish, $type) = `git-cat-file tag $treeish`);
1219        }
1220        if ($type eq 'commit') {
1221                $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
1222                ($expected) = ($expected =~ /^tree ($sha1)$/);
1223                die "Unable to get tree from $treeish\n" unless $expected;
1224        } elsif ($type eq 'tree') {
1225                $expected = $treeish;
1226        } else {
1227                die "$treeish is a $type, expected tree, tag or commit\n";
1228        }
1229        return $expected;
1230}
1231
1232sub assert_tree {
1233        return if $_use_lib;
1234        my ($treeish) = @_;
1235        my $expected = get_tree_from_treeish($treeish);
1236
1237        my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
1238        if (-e $tmpindex) {
1239                unlink $tmpindex or croak $!;
1240        }
1241        my $old_index = set_index($tmpindex);
1242        index_changes(1);
1243        chomp(my $tree = `git-write-tree`);
1244        restore_index($old_index);
1245        if ($tree ne $expected) {
1246                croak "Tree mismatch, Got: $tree, Expected: $expected\n";
1247        }
1248        unlink $tmpindex;
1249}
1250
1251sub parse_diff_tree {
1252        my $diff_fh = shift;
1253        local $/ = "\0";
1254        my $state = 'meta';
1255        my @mods;
1256        while (<$diff_fh>) {
1257                chomp $_; # this gets rid of the trailing "\0"
1258                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1259                                        $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1260                        push @mods, {   mode_a => $1, mode_b => $2,
1261                                        sha1_b => $3, chg => $4 };
1262                        if ($4 =~ /^(?:C|R)$/) {
1263                                $state = 'file_a';
1264                        } else {
1265                                $state = 'file_b';
1266                        }
1267                } elsif ($state eq 'file_a') {
1268                        my $x = $mods[$#mods] or croak "Empty array\n";
1269                        if ($x->{chg} !~ /^(?:C|R)$/) {
1270                                croak "Error parsing $_, $x->{chg}\n";
1271                        }
1272                        $x->{file_a} = $_;
1273                        $state = 'file_b';
1274                } elsif ($state eq 'file_b') {
1275                        my $x = $mods[$#mods] or croak "Empty array\n";
1276                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1277                                croak "Error parsing $_, $x->{chg}\n";
1278                        }
1279                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1280                                croak "Error parsing $_, $x->{chg}\n";
1281                        }
1282                        $x->{file_b} = $_;
1283                        $state = 'meta';
1284                } else {
1285                        croak "Error parsing $_\n";
1286                }
1287        }
1288        close $diff_fh or croak $?;
1289
1290        return \@mods;
1291}
1292
1293sub svn_check_prop_executable {
1294        my $m = shift;
1295        return if -l $m->{file_b};
1296        if ($m->{mode_b} =~ /755$/) {
1297                chmod((0755 &~ umask),$m->{file_b}) or croak $!;
1298                if ($m->{mode_a} !~ /755$/) {
1299                        sys(qw(svn propset svn:executable 1), $m->{file_b});
1300                }
1301                -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
1302        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
1303                sys(qw(svn propdel svn:executable), $m->{file_b});
1304                chmod((0644 &~ umask),$m->{file_b}) or croak $!;
1305                -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
1306        }
1307}
1308
1309sub svn_ensure_parent_path {
1310        my $dir_b = dirname(shift);
1311        svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
1312        mkpath([$dir_b]) unless (-d $dir_b);
1313        sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
1314}
1315
1316sub precommit_check {
1317        my $mods = shift;
1318        my (%rm_file, %rmdir_check, %added_check);
1319
1320        my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
1321        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1322                if ($m->{chg} eq 'R') {
1323                        if (-d $m->{file_b}) {
1324                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1325                        }
1326                        # dir/$file => dir/file/$file
1327                        my $dirname = dirname($m->{file_b});
1328                        while ($dirname ne File::Spec->curdir) {
1329                                if ($dirname ne $m->{file_a}) {
1330                                        $dirname = dirname($dirname);
1331                                        next;
1332                                }
1333                                err_file_to_dir("$m->{file_a} => $m->{file_b}");
1334                        }
1335                        # baz/zzz => baz (baz is a file)
1336                        $dirname = dirname($m->{file_a});
1337                        while ($dirname ne File::Spec->curdir) {
1338                                if ($dirname ne $m->{file_b}) {
1339                                        $dirname = dirname($dirname);
1340                                        next;
1341                                }
1342                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1343                        }
1344                }
1345                if ($m->{chg} =~ /^(D|R)$/) {
1346                        my $t = $1 eq 'D' ? 'file_b' : 'file_a';
1347                        $rm_file{ $m->{$t} } = 1;
1348                        my $dirname = dirname( $m->{$t} );
1349                        my $basename = basename( $m->{$t} );
1350                        $rmdir_check{$dirname}->{$basename} = 1;
1351                } elsif ($m->{chg} =~ /^(?:A|C)$/) {
1352                        if (-d $m->{file_b}) {
1353                                err_dir_to_file($m->{file_b});
1354                        }
1355                        my $dirname = dirname( $m->{file_b} );
1356                        my $basename = basename( $m->{file_b} );
1357                        $added_check{$dirname}->{$basename} = 1;
1358                        while ($dirname ne File::Spec->curdir) {
1359                                if ($rm_file{$dirname}) {
1360                                        err_file_to_dir($m->{file_b});
1361                                }
1362                                $dirname = dirname $dirname;
1363                        }
1364                }
1365        }
1366        return (\%rmdir_check, \%added_check);
1367
1368        sub err_dir_to_file {
1369                my $file = shift;
1370                print STDERR "Node change from directory to file ",
1371                                "is not supported by Subversion: ",$file,"\n";
1372                exit 1;
1373        }
1374        sub err_file_to_dir {
1375                my $file = shift;
1376                print STDERR "Node change from file to directory ",
1377                                "is not supported by Subversion: ",$file,"\n";
1378                exit 1;
1379        }
1380}
1381
1382
1383sub get_diff {
1384        my ($from, $treeish) = @_;
1385        assert_tree($from);
1386        print "diff-tree $from $treeish\n";
1387        my $pid = open my $diff_fh, '-|';
1388        defined $pid or croak $!;
1389        if ($pid == 0) {
1390                my @diff_tree = qw(git-diff-tree -z -r);
1391                if ($_cp_similarity) {
1392                        push @diff_tree, "-C$_cp_similarity";
1393                } else {
1394                        push @diff_tree, '-C';
1395                }
1396                push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1397                push @diff_tree, "-l$_l" if defined $_l;
1398                exec(@diff_tree, $from, $treeish) or croak $!;
1399        }
1400        return parse_diff_tree($diff_fh);
1401}
1402
1403sub svn_checkout_tree {
1404        my ($from, $treeish) = @_;
1405        my $mods = get_diff($from->{commit}, $treeish);
1406        return $mods unless (scalar @$mods);
1407        my ($rm, $add) = precommit_check($mods);
1408
1409        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1410        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1411                if ($m->{chg} eq 'C') {
1412                        svn_ensure_parent_path( $m->{file_b} );
1413                        sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
1414                        apply_mod_line_blob($m);
1415                        svn_check_prop_executable($m);
1416                } elsif ($m->{chg} eq 'D') {
1417                        sys(qw(svn rm --force), $m->{file_b});
1418                } elsif ($m->{chg} eq 'R') {
1419                        svn_ensure_parent_path( $m->{file_b} );
1420                        sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
1421                        apply_mod_line_blob($m);
1422                        svn_check_prop_executable($m);
1423                } elsif ($m->{chg} eq 'M') {
1424                        apply_mod_line_blob($m);
1425                        svn_check_prop_executable($m);
1426                } elsif ($m->{chg} eq 'T') {
1427                        sys(qw(svn rm --force),$m->{file_b});
1428                        apply_mod_line_blob($m);
1429                        sys(qw(svn add), $m->{file_b});
1430                        svn_check_prop_executable($m);
1431                } elsif ($m->{chg} eq 'A') {
1432                        svn_ensure_parent_path( $m->{file_b} );
1433                        apply_mod_line_blob($m);
1434                        sys(qw(svn add), $m->{file_b});
1435                        svn_check_prop_executable($m);
1436                } else {
1437                        croak "Invalid chg: $m->{chg}\n";
1438                }
1439        }
1440
1441        assert_tree($treeish);
1442        if ($_rmdir) { # remove empty directories
1443                handle_rmdir($rm, $add);
1444        }
1445        assert_tree($treeish);
1446        return $mods;
1447}
1448
1449sub libsvn_checkout_tree {
1450        my ($from, $treeish, $ed) = @_;
1451        my $mods = get_diff($from, $treeish);
1452        return $mods unless (scalar @$mods);
1453        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1454        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1455                my $f = $m->{chg};
1456                if (defined $o{$f}) {
1457                        $ed->$f($m);
1458                } else {
1459                        croak "Invalid change type: $f\n";
1460                }
1461        }
1462        $ed->rmdirs if $_rmdir;
1463        return $mods;
1464}
1465
1466# svn ls doesn't work with respect to the current working tree, but what's
1467# in the repository.  There's not even an option for it... *sigh*
1468# (added files don't show up and removed files remain in the ls listing)
1469sub svn_ls_current {
1470        my ($dir, $rm, $add) = @_;
1471        chomp(my @ls = safe_qx('svn','ls',$dir));
1472        my @ret = ();
1473        foreach (@ls) {
1474                s#/$##; # trailing slashes are evil
1475                push @ret, $_ unless $rm->{$dir}->{$_};
1476        }
1477        if (exists $add->{$dir}) {
1478                push @ret, keys %{$add->{$dir}};
1479        }
1480        return \@ret;
1481}
1482
1483sub handle_rmdir {
1484        my ($rm, $add) = @_;
1485
1486        foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
1487                my $ls = svn_ls_current($dir, $rm, $add);
1488                next if (scalar @$ls);
1489                sys(qw(svn rm --force),$dir);
1490
1491                my $dn = dirname $dir;
1492                $rm->{ $dn }->{ basename $dir } = 1;
1493                $ls = svn_ls_current($dn, $rm, $add);
1494                while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
1495                        sys(qw(svn rm --force),$dn);
1496                        $dir = basename $dn;
1497                        $dn = dirname $dn;
1498                        $rm->{ $dn }->{ $dir } = 1;
1499                        $ls = svn_ls_current($dn, $rm, $add);
1500                }
1501        }
1502}
1503
1504sub get_commit_message {
1505        my ($commit, $commit_msg) = (@_);
1506        my %log_msg = ( msg => '' );
1507        open my $msg, '>', $commit_msg or croak $!;
1508
1509        chomp(my $type = `git-cat-file -t $commit`);
1510        if ($type eq 'commit') {
1511                my $pid = open my $msg_fh, '-|';
1512                defined $pid or croak $!;
1513
1514                if ($pid == 0) {
1515                        exec(qw(git-cat-file commit), $commit) or croak $!;
1516                }
1517                my $in_msg = 0;
1518                while (<$msg_fh>) {
1519                        if (!$in_msg) {
1520                                $in_msg = 1 if (/^\s*$/);
1521                        } elsif (/^git-svn-id: /) {
1522                                # skip this, we regenerate the correct one
1523                                # on re-fetch anyways
1524                        } else {
1525                                print $msg $_ or croak $!;
1526                        }
1527                }
1528                close $msg_fh or croak $?;
1529        }
1530        close $msg or croak $!;
1531
1532        if ($_edit || ($type eq 'tree')) {
1533                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1534                system($editor, $commit_msg);
1535        }
1536
1537        # file_to_s removes all trailing newlines, so just use chomp() here:
1538        open $msg, '<', $commit_msg or croak $!;
1539        { local $/; chomp($log_msg{msg} = <$msg>); }
1540        close $msg or croak $!;
1541
1542        return \%log_msg;
1543}
1544
1545sub set_svn_commit_env {
1546        if (defined $LC_ALL) {
1547                $ENV{LC_ALL} = $LC_ALL;
1548        } else {
1549                delete $ENV{LC_ALL};
1550        }
1551}
1552
1553sub svn_commit_tree {
1554        my ($last, $commit) = @_;
1555        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
1556        my $log_msg = get_commit_message($commit, $commit_msg);
1557        my ($oneline) = ($log_msg->{msg} =~ /([^\n\r]+)/);
1558        print "Committing $commit: $oneline\n";
1559
1560        set_svn_commit_env();
1561        my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
1562        $ENV{LC_ALL} = 'C';
1563        unlink $commit_msg;
1564        my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/);
1565        if (!defined $committed) {
1566                my $out = join("\n",@ci_output);
1567                print STDERR "W: Trouble parsing \`svn commit' output:\n\n",
1568                                $out, "\n\nAssuming English locale...";
1569                ($committed) = ($out =~ /^Committed revision \d+\./sm);
1570                defined $committed or die " FAILED!\n",
1571                        "Commit output failed to parse committed revision!\n",
1572                print STDERR " OK\n";
1573        }
1574
1575        my @svn_up = qw(svn up);
1576        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
1577        if ($_optimize_commits && ($committed == ($last->{revision} + 1))) {
1578                push @svn_up, "-r$committed";
1579                sys(@svn_up);
1580                my $info = svn_info('.');
1581                my $date = $info->{'Last Changed Date'} or die "Missing date\n";
1582                if ($info->{'Last Changed Rev'} != $committed) {
1583                        croak "$info->{'Last Changed Rev'} != $committed\n"
1584                }
1585                my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1586                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1587                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1588                                         or croak "Failed to parse date: $date\n";
1589                $log_msg->{date} = "$tz $Y-$m-$d $H:$M:$S";
1590                $log_msg->{author} = $info->{'Last Changed Author'};
1591                $log_msg->{revision} = $committed;
1592                $log_msg->{msg} .= "\n";
1593                $log_msg->{parents} = [ $last->{commit} ];
1594                $log_msg->{commit} = git_commit($log_msg, $commit);
1595                return $log_msg;
1596        }
1597        # resync immediately
1598        push @svn_up, "-r$last->{revision}";
1599        sys(@svn_up);
1600        return fetch("$committed=$commit");
1601}
1602
1603sub rev_list_raw {
1604        my (@args) = @_;
1605        my $pid = open my $fh, '-|';
1606        defined $pid or croak $!;
1607        if (!$pid) {
1608                exec(qw/git-rev-list --pretty=raw/, @args) or croak $!;
1609        }
1610        return { fh => $fh, t => { } };
1611}
1612
1613sub next_rev_list_entry {
1614        my $rl = shift;
1615        my $fh = $rl->{fh};
1616        my $x = $rl->{t};
1617        while (<$fh>) {
1618                if (/^commit ($sha1)$/o) {
1619                        if ($x->{c}) {
1620                                $rl->{t} = { c => $1 };
1621                                return $x;
1622                        } else {
1623                                $x->{c} = $1;
1624                        }
1625                } elsif (/^parent ($sha1)$/o) {
1626                        $x->{p}->{$1} = 1;
1627                } elsif (s/^    //) {
1628                        $x->{m} ||= '';
1629                        $x->{m} .= $_;
1630                }
1631        }
1632        return ($x != $rl->{t}) ? $x : undef;
1633}
1634
1635# read the entire log into a temporary file (which is removed ASAP)
1636# and store the file handle + parser state
1637sub svn_log_raw {
1638        my (@log_args) = @_;
1639        my $log_fh = IO::File->new_tmpfile or croak $!;
1640        my $pid = fork;
1641        defined $pid or croak $!;
1642        if (!$pid) {
1643                open STDOUT, '>&', $log_fh or croak $!;
1644                exec (qw(svn log), @log_args) or croak $!
1645        }
1646        waitpid $pid, 0;
1647        croak $? if $?;
1648        seek $log_fh, 0, 0 or croak $!;
1649        return { state => 'sep', fh => $log_fh };
1650}
1651
1652sub next_log_entry {
1653        my $log = shift; # retval of svn_log_raw()
1654        my $ret = undef;
1655        my $fh = $log->{fh};
1656
1657        while (<$fh>) {
1658                chomp;
1659                if (/^\-{72}$/) {
1660                        if ($log->{state} eq 'msg') {
1661                                if ($ret->{lines}) {
1662                                        $ret->{msg} .= $_."\n";
1663                                        unless(--$ret->{lines}) {
1664                                                $log->{state} = 'sep';
1665                                        }
1666                                } else {
1667                                        croak "Log parse error at: $_\n",
1668                                                $ret->{revision},
1669                                                "\n";
1670                                }
1671                                next;
1672                        }
1673                        if ($log->{state} ne 'sep') {
1674                                croak "Log parse error at: $_\n",
1675                                        "state: $log->{state}\n",
1676                                        $ret->{revision},
1677                                        "\n";
1678                        }
1679                        $log->{state} = 'rev';
1680
1681                        # if we have an empty log message, put something there:
1682                        if ($ret) {
1683                                $ret->{msg} ||= "\n";
1684                                delete $ret->{lines};
1685                                return $ret;
1686                        }
1687                        next;
1688                }
1689                if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
1690                        my $rev = $1;
1691                        my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
1692                        ($lines) = ($lines =~ /(\d+)/);
1693                        my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1694                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1695                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1696                                         or croak "Failed to parse date: $date\n";
1697                        $ret = {        revision => $rev,
1698                                        date => "$tz $Y-$m-$d $H:$M:$S",
1699                                        author => $author,
1700                                        lines => $lines,
1701                                        msg => '' };
1702                        if (defined $_authors && ! defined $users{$author}) {
1703                                die "Author: $author not defined in ",
1704                                                "$_authors file\n";
1705                        }
1706                        $log->{state} = 'msg_start';
1707                        next;
1708                }
1709                # skip the first blank line of the message:
1710                if ($log->{state} eq 'msg_start' && /^$/) {
1711                        $log->{state} = 'msg';
1712                } elsif ($log->{state} eq 'msg') {
1713                        if ($ret->{lines}) {
1714                                $ret->{msg} .= $_."\n";
1715                                unless (--$ret->{lines}) {
1716                                        $log->{state} = 'sep';
1717                                }
1718                        } else {
1719                                croak "Log parse error at: $_\n",
1720                                        $ret->{revision},"\n";
1721                        }
1722                }
1723        }
1724        return $ret;
1725}
1726
1727sub svn_info {
1728        my $url = shift || $SVN_URL;
1729
1730        my $pid = open my $info_fh, '-|';
1731        defined $pid or croak $!;
1732
1733        if ($pid == 0) {
1734                exec(qw(svn info),$url) or croak $!;
1735        }
1736
1737        my $ret = {};
1738        # only single-lines seem to exist in svn info output
1739        while (<$info_fh>) {
1740                chomp $_;
1741                if (m#^([^:]+)\s*:\s*(\S.*)$#) {
1742                        $ret->{$1} = $2;
1743                        push @{$ret->{-order}}, $1;
1744                }
1745        }
1746        close $info_fh or croak $?;
1747        return $ret;
1748}
1749
1750sub sys { system(@_) == 0 or croak $? }
1751
1752sub eol_cp {
1753        my ($from, $to) = @_;
1754        my $es = svn_propget_base('svn:eol-style', $to);
1755        open my $rfd, '<', $from or croak $!;
1756        binmode $rfd or croak $!;
1757        open my $wfd, '>', $to or croak $!;
1758        binmode $wfd or croak $!;
1759        eol_cp_fd($rfd, $wfd, $es);
1760        close $rfd or croak $!;
1761        close $wfd or croak $!;
1762}
1763
1764sub eol_cp_fd {
1765        my ($rfd, $wfd, $es) = @_;
1766        my $eol = defined $es ? $EOL{$es} : undef;
1767        my $buf;
1768        use bytes;
1769        while (1) {
1770                my ($r, $w, $t);
1771                defined($r = sysread($rfd, $buf, 4096)) or croak $!;
1772                return unless $r;
1773                if ($eol) {
1774                        if ($buf =~ /\015$/) {
1775                                my $c;
1776                                defined($r = sysread($rfd,$c,1)) or croak $!;
1777                                $buf .= $c if $r > 0;
1778                        }
1779                        $buf =~ s/(?:\015\012|\015|\012)/$eol/gs;
1780                        $r = length($buf);
1781                }
1782                for ($w = 0; $w < $r; $w += $t) {
1783                        $t = syswrite($wfd, $buf, $r - $w, $w) or croak $!;
1784                }
1785        }
1786        no bytes;
1787}
1788
1789sub do_update_index {
1790        my ($z_cmd, $cmd, $no_text_base) = @_;
1791
1792        my $z = open my $p, '-|';
1793        defined $z or croak $!;
1794        unless ($z) { exec @$z_cmd or croak $! }
1795
1796        my $pid = open my $ui, '|-';
1797        defined $pid or croak $!;
1798        unless ($pid) {
1799                exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!;
1800        }
1801        local $/ = "\0";
1802        while (my $x = <$p>) {
1803                chomp $x;
1804                if (!$no_text_base && lstat $x && ! -l _ &&
1805                                svn_propget_base('svn:keywords', $x)) {
1806                        my $mode = -x _ ? 0755 : 0644;
1807                        my ($v,$d,$f) = File::Spec->splitpath($x);
1808                        my $tb = File::Spec->catfile($d, '.svn', 'tmp',
1809                                                'text-base',"$f.svn-base");
1810                        $tb =~ s#^/##;
1811                        unless (-f $tb) {
1812                                $tb = File::Spec->catfile($d, '.svn',
1813                                                'text-base',"$f.svn-base");
1814                                $tb =~ s#^/##;
1815                        }
1816                        unlink $x or croak $!;
1817                        eol_cp($tb, $x);
1818                        chmod(($mode &~ umask), $x) or croak $!;
1819                }
1820                print $ui $x,"\0";
1821        }
1822        close $ui or croak $?;
1823}
1824
1825sub index_changes {
1826        return if $_use_lib;
1827
1828        if (!-f "$GIT_SVN_DIR/info/exclude") {
1829                open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!;
1830                print $fd '.svn',"\n";
1831                close $fd or croak $!;
1832        }
1833        my $no_text_base = shift;
1834        do_update_index([qw/git-diff-files --name-only -z/],
1835                        'remove',
1836                        $no_text_base);
1837        do_update_index([qw/git-ls-files -z --others/,
1838                                "--exclude-from=$GIT_SVN_DIR/info/exclude"],
1839                        'add',
1840                        $no_text_base);
1841}
1842
1843sub s_to_file {
1844        my ($str, $file, $mode) = @_;
1845        open my $fd,'>',$file or croak $!;
1846        print $fd $str,"\n" or croak $!;
1847        close $fd or croak $!;
1848        chmod ($mode &~ umask, $file) if (defined $mode);
1849}
1850
1851sub file_to_s {
1852        my $file = shift;
1853        open my $fd,'<',$file or croak "$!: file: $file\n";
1854        local $/;
1855        my $ret = <$fd>;
1856        close $fd or croak $!;
1857        $ret =~ s/\s*$//s;
1858        return $ret;
1859}
1860
1861sub assert_revision_unknown {
1862        my $r = shift;
1863        if (my $c = revdb_get($REVDB, $r)) {
1864                croak "$r = $c already exists! Why are we refetching it?";
1865        }
1866}
1867
1868sub trees_eq {
1869        my ($x, $y) = @_;
1870        my @x = safe_qx('git-cat-file','commit',$x);
1871        my @y = safe_qx('git-cat-file','commit',$y);
1872        if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
1873                                || $y[0] !~ /^tree $sha1\n$/) {
1874                print STDERR "Trees not equal: $y[0] != $x[0]\n";
1875                return 0
1876        }
1877        return 1;
1878}
1879
1880sub git_commit {
1881        my ($log_msg, @parents) = @_;
1882        assert_revision_unknown($log_msg->{revision});
1883        map_tree_joins() if (@_branch_from && !%tree_map);
1884
1885        my (@tmp_parents, @exec_parents, %seen_parent);
1886        if (my $lparents = $log_msg->{parents}) {
1887                @tmp_parents = @$lparents
1888        }
1889        # commit parents can be conditionally bound to a particular
1890        # svn revision via: "svn_revno=commit_sha1", filter them out here:
1891        foreach my $p (@parents) {
1892                next unless defined $p;
1893                if ($p =~ /^(\d+)=($sha1_short)$/o) {
1894                        if ($1 == $log_msg->{revision}) {
1895                                push @tmp_parents, $2;
1896                        }
1897                } else {
1898                        push @tmp_parents, $p if $p =~ /$sha1_short/o;
1899                }
1900        }
1901        my $tree = $log_msg->{tree};
1902        if (!defined $tree) {
1903                my $index = set_index($GIT_SVN_INDEX);
1904                index_changes();
1905                chomp($tree = `git-write-tree`);
1906                croak $? if $?;
1907                restore_index($index);
1908        }
1909
1910        # just in case we clobber the existing ref, we still want that ref
1911        # as our parent:
1912        if (my $cur = eval { file_to_s("$GIT_DIR/refs/remotes/$GIT_SVN") }) {
1913                push @tmp_parents, $cur;
1914        }
1915
1916        if (exists $tree_map{$tree}) {
1917                foreach my $p (@{$tree_map{$tree}}) {
1918                        my $skip;
1919                        foreach (@tmp_parents) {
1920                                # see if a common parent is found
1921                                my $mb = eval {
1922                                        safe_qx('git-merge-base', $_, $p)
1923                                };
1924                                next if ($@ || $?);
1925                                $skip = 1;
1926                                last;
1927                        }
1928                        next if $skip;
1929                        my ($url_p, $r_p, $uuid_p) = cmt_metadata($p);
1930                        next if (($SVN_UUID eq $uuid_p) &&
1931                                                ($log_msg->{revision} > $r_p));
1932                        next if (defined $url_p && defined $SVN_URL &&
1933                                                ($SVN_UUID eq $uuid_p) &&
1934                                                ($url_p eq $SVN_URL));
1935                        push @tmp_parents, $p;
1936                }
1937        }
1938        foreach (@tmp_parents) {
1939                next if $seen_parent{$_};
1940                $seen_parent{$_} = 1;
1941                push @exec_parents, $_;
1942                # MAXPARENT is defined to 16 in commit-tree.c:
1943                last if @exec_parents > 16;
1944        }
1945
1946        set_commit_env($log_msg);
1947        my @exec = ('git-commit-tree', $tree);
1948        push @exec, '-p', $_  foreach @exec_parents;
1949        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1950                                                                or croak $!;
1951        print $msg_fh $log_msg->{msg} or croak $!;
1952        unless ($_no_metadata) {
1953                print $msg_fh "\ngit-svn-id: $SVN_URL\@$log_msg->{revision}",
1954                                        " $SVN_UUID\n" or croak $!;
1955        }
1956        $msg_fh->flush == 0 or croak $!;
1957        close $msg_fh or croak $!;
1958        chomp(my $commit = do { local $/; <$out_fh> });
1959        close $out_fh or croak $!;
1960        waitpid $pid, 0;
1961        croak $? if $?;
1962        if ($commit !~ /^$sha1$/o) {
1963                die "Failed to commit, invalid sha1: $commit\n";
1964        }
1965        sys('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
1966        revdb_set($REVDB, $log_msg->{revision}, $commit);
1967
1968        # this output is read via pipe, do not change:
1969        print "r$log_msg->{revision} = $commit\n";
1970        check_repack();
1971        return $commit;
1972}
1973
1974sub check_repack {
1975        if ($_repack && (--$_repack_nr == 0)) {
1976                $_repack_nr = $_repack;
1977                sys("git repack $_repack_flags");
1978        }
1979}
1980
1981sub set_commit_env {
1982        my ($log_msg) = @_;
1983        my $author = $log_msg->{author};
1984        if (!defined $author || length $author == 0) {
1985                $author = '(no author)';
1986        }
1987        my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
1988                                : ($author,"$author\@$SVN_UUID");
1989        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1990        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1991        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
1992}
1993
1994sub apply_mod_line_blob {
1995        my $m = shift;
1996        if ($m->{mode_b} =~ /^120/) {
1997                blob_to_symlink($m->{sha1_b}, $m->{file_b});
1998        } else {
1999                blob_to_file($m->{sha1_b}, $m->{file_b});
2000        }
2001}
2002
2003sub blob_to_symlink {
2004        my ($blob, $link) = @_;
2005        defined $link or croak "\$link not defined!\n";
2006        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2007        if (-l $link || -f _) {
2008                unlink $link or croak $!;
2009        }
2010
2011        my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
2012        symlink $dest, $link or croak $!;
2013}
2014
2015sub blob_to_file {
2016        my ($blob, $file) = @_;
2017        defined $file or croak "\$file not defined!\n";
2018        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2019        if (-l $file || -f _) {
2020                unlink $file or croak $!;
2021        }
2022
2023        open my $blob_fh, '>', $file or croak "$!: $file\n";
2024        my $pid = fork;
2025        defined $pid or croak $!;
2026
2027        if ($pid == 0) {
2028                open STDOUT, '>&', $blob_fh or croak $!;
2029                exec('git-cat-file','blob',$blob) or croak $!;
2030        }
2031        waitpid $pid, 0;
2032        croak $? if $?;
2033
2034        close $blob_fh or croak $!;
2035}
2036
2037sub safe_qx {
2038        my $pid = open my $child, '-|';
2039        defined $pid or croak $!;
2040        if ($pid == 0) {
2041                exec(@_) or croak $!;
2042        }
2043        my @ret = (<$child>);
2044        close $child or croak $?;
2045        die $? if $?; # just in case close didn't error out
2046        return wantarray ? @ret : join('',@ret);
2047}
2048
2049sub svn_compat_check {
2050        if ($_follow_parent) {
2051                print STDERR 'E: --follow-parent functionality is only ',
2052                                "available when SVN libraries are used\n";
2053                exit 1;
2054        }
2055        my @co_help = safe_qx(qw(svn co -h));
2056        unless (grep /ignore-externals/,@co_help) {
2057                print STDERR "W: Installed svn version does not support ",
2058                                "--ignore-externals\n";
2059                $_no_ignore_ext = 1;
2060        }
2061        if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
2062                $_svn_co_url_revs = 1;
2063        }
2064        if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) {
2065                $_svn_pg_peg_revs = 1;
2066        }
2067
2068        # I really, really hope nobody hits this...
2069        unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
2070                print STDERR <<'';
2071W: The installed svn version does not support the --stop-on-copy flag in
2072   the log command.
2073   Lets hope the directory you're tracking is not a branch or tag
2074   and was never moved within the repository...
2075
2076                $_no_stop_copy = 1;
2077        }
2078}
2079
2080# *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
2081# (and they won't honor URL@<rev> without -r<rev>, too!)
2082sub svn_cmd_checkout {
2083        my ($url, $rev, $dir) = @_;
2084        my @cmd = ('svn','co', "-r$rev");
2085        push @cmd, '--ignore-externals' unless $_no_ignore_ext;
2086        $url .= "\@$rev" if $_svn_co_url_revs;
2087        sys(@cmd, $url, $dir);
2088}
2089
2090sub check_upgrade_needed {
2091        if (!-r $REVDB) {
2092                -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2093                open my $fh, '>>',$REVDB or croak $!;
2094                close $fh;
2095        }
2096        my $old = eval {
2097                my $pid = open my $child, '-|';
2098                defined $pid or croak $!;
2099                if ($pid == 0) {
2100                        close STDERR;
2101                        exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $!;
2102                }
2103                my @ret = (<$child>);
2104                close $child or croak $?;
2105                die $? if $?; # just in case close didn't error out
2106                return wantarray ? @ret : join('',@ret);
2107        };
2108        return unless $old;
2109        my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
2110        if ($@ || !$head) {
2111                print STDERR "Please run: $0 rebuild --upgrade\n";
2112                exit 1;
2113        }
2114}
2115
2116# fills %tree_map with a reverse mapping of trees to commits.  Useful
2117# for finding parents to commit on.
2118sub map_tree_joins {
2119        my %seen;
2120        foreach my $br (@_branch_from) {
2121                my $pid = open my $pipe, '-|';
2122                defined $pid or croak $!;
2123                if ($pid == 0) {
2124                        exec(qw(git-rev-list --topo-order --pretty=raw), $br)
2125                                                                or croak $!;
2126                }
2127                while (<$pipe>) {
2128                        if (/^commit ($sha1)$/o) {
2129                                my $commit = $1;
2130
2131                                # if we've seen a commit,
2132                                # we've seen its parents
2133                                last if $seen{$commit};
2134                                my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
2135                                unless (defined $tree) {
2136                                        die "Failed to parse commit $commit\n";
2137                                }
2138                                push @{$tree_map{$tree}}, $commit;
2139                                $seen{$commit} = 1;
2140                        }
2141                }
2142                close $pipe; # we could be breaking the pipe early
2143        }
2144}
2145
2146sub load_all_refs {
2147        if (@_branch_from) {
2148                print STDERR '--branch|-b parameters are ignored when ',
2149                        "--branch-all-refs|-B is passed\n";
2150        }
2151
2152        # don't worry about rev-list on non-commit objects/tags,
2153        # it shouldn't blow up if a ref is a blob or tree...
2154        chomp(@_branch_from = `git-rev-parse --symbolic --all`);
2155}
2156
2157# '<svn username> = real-name <email address>' mapping based on git-svnimport:
2158sub load_authors {
2159        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2160        while (<$authors>) {
2161                chomp;
2162                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2163                my ($user, $name, $email) = ($1, $2, $3);
2164                $users{$user} = [$name, $email];
2165        }
2166        close $authors or croak $!;
2167}
2168
2169sub rload_authors {
2170        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2171        while (<$authors>) {
2172                chomp;
2173                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2174                my ($user, $name, $email) = ($1, $2, $3);
2175                $rusers{"$name <$email>"} = $user;
2176        }
2177        close $authors or croak $!;
2178}
2179
2180sub svn_propget_base {
2181        my ($p, $f) = @_;
2182        $f .= '@BASE' if $_svn_pg_peg_revs;
2183        return safe_qx(qw/svn propget/, $p, $f);
2184}
2185
2186sub git_svn_each {
2187        my $sub = shift;
2188        foreach (`git-rev-parse --symbolic --all`) {
2189                next unless s#^refs/remotes/##;
2190                chomp $_;
2191                next unless -f "$GIT_DIR/svn/$_/info/url";
2192                &$sub($_);
2193        }
2194}
2195
2196sub migrate_revdb {
2197        git_svn_each(sub {
2198                my $id = shift;
2199                defined(my $pid = fork) or croak $!;
2200                if (!$pid) {
2201                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2202                        init_vars();
2203                        exit 0 if -r $REVDB;
2204                        print "Upgrading svn => git mapping...\n";
2205                        -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2206                        open my $fh, '>>',$REVDB or croak $!;
2207                        close $fh;
2208                        rebuild();
2209                        print "Done upgrading. You may now delete the ",
2210                                "deprecated $GIT_SVN_DIR/revs directory\n";
2211                        exit 0;
2212                }
2213                waitpid $pid, 0;
2214                croak $? if $?;
2215        });
2216}
2217
2218sub migration_check {
2219        migrate_revdb() unless (-e $REVDB);
2220        return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
2221        print "Upgrading repository...\n";
2222        unless (-d "$GIT_DIR/svn") {
2223                mkdir "$GIT_DIR/svn" or croak $!;
2224        }
2225        print "Data from a previous version of git-svn exists, but\n\t",
2226                                "$GIT_SVN_DIR\n\t(required for this version ",
2227                                "($VERSION) of git-svn) does not.\n";
2228
2229        foreach my $x (`git-rev-parse --symbolic --all`) {
2230                next unless $x =~ s#^refs/remotes/##;
2231                chomp $x;
2232                next unless -f "$GIT_DIR/$x/info/url";
2233                my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
2234                next unless $u;
2235                my $dn = dirname("$GIT_DIR/svn/$x");
2236                mkpath([$dn]) unless -d $dn;
2237                rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
2238        }
2239        migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
2240        print "Done upgrading.\n";
2241}
2242
2243sub find_rev_before {
2244        my ($r, $id, $eq_ok) = @_;
2245        my $f = "$GIT_DIR/svn/$id/.rev_db";
2246        return (undef,undef) unless -r $f;
2247        --$r unless $eq_ok;
2248        while ($r > 0) {
2249                if (my $c = revdb_get($f, $r)) {
2250                        return ($r, $c);
2251                }
2252                --$r;
2253        }
2254        return (undef, undef);
2255}
2256
2257sub init_vars {
2258        $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
2259        $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
2260        $REVDB = "$GIT_SVN_DIR/.rev_db";
2261        $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
2262        $SVN_URL = undef;
2263        $SVN_WC = "$GIT_SVN_DIR/tree";
2264        %tree_map = ();
2265}
2266
2267# convert GetOpt::Long specs for use by git-repo-config
2268sub read_repo_config {
2269        return unless -d $GIT_DIR;
2270        my $opts = shift;
2271        foreach my $o (keys %$opts) {
2272                my $v = $opts->{$o};
2273                my ($key) = ($o =~ /^([a-z\-]+)/);
2274                $key =~ s/-//g;
2275                my $arg = 'git-repo-config';
2276                $arg .= ' --int' if ($o =~ /[:=]i$/);
2277                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
2278                if (ref $v eq 'ARRAY') {
2279                        chomp(my @tmp = `$arg --get-all svn.$key`);
2280                        @$v = @tmp if @tmp;
2281                } else {
2282                        chomp(my $tmp = `$arg --get svn.$key`);
2283                        if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
2284                                $$v = $tmp;
2285                        }
2286                }
2287        }
2288}
2289
2290sub set_default_vals {
2291        if (defined $_repack) {
2292                $_repack = 1000 if ($_repack <= 0);
2293                $_repack_nr = $_repack;
2294                $_repack_flags ||= '-d';
2295        }
2296}
2297
2298sub read_grafts {
2299        my $gr_file = shift;
2300        my ($grafts, $comments) = ({}, {});
2301        if (open my $fh, '<', $gr_file) {
2302                my @tmp;
2303                while (<$fh>) {
2304                        if (/^($sha1)\s+/) {
2305                                my $c = $1;
2306                                if (@tmp) {
2307                                        @{$comments->{$c}} = @tmp;
2308                                        @tmp = ();
2309                                }
2310                                foreach my $p (split /\s+/, $_) {
2311                                        $grafts->{$c}->{$p} = 1;
2312                                }
2313                        } else {
2314                                push @tmp, $_;
2315                        }
2316                }
2317                close $fh or croak $!;
2318                @{$comments->{'END'}} = @tmp if @tmp;
2319        }
2320        return ($grafts, $comments);
2321}
2322
2323sub write_grafts {
2324        my ($grafts, $comments, $gr_file) = @_;
2325
2326        open my $fh, '>', $gr_file or croak $!;
2327        foreach my $c (sort keys %$grafts) {
2328                if ($comments->{$c}) {
2329                        print $fh $_ foreach @{$comments->{$c}};
2330                }
2331                my $p = $grafts->{$c};
2332                my %x; # real parents
2333                delete $p->{$c}; # commits are not self-reproducing...
2334                my $pid = open my $ch, '-|';
2335                defined $pid or croak $!;
2336                if (!$pid) {
2337                        exec(qw/git-cat-file commit/, $c) or croak $!;
2338                }
2339                while (<$ch>) {
2340                        if (/^parent ($sha1)/) {
2341                                $x{$1} = $p->{$1} = 1;
2342                        } else {
2343                                last unless /^\S/;
2344                        }
2345                }
2346                close $ch; # breaking the pipe
2347
2348                # if real parents are the only ones in the grafts, drop it
2349                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2350
2351                my (@ip, @jp, $mb);
2352                my %del = %x;
2353                @ip = @jp = keys %$p;
2354                foreach my $i (@ip) {
2355                        next if $del{$i} || $p->{$i} == 2;
2356                        foreach my $j (@jp) {
2357                                next if $i eq $j || $del{$j} || $p->{$j} == 2;
2358                                $mb = eval { safe_qx('git-merge-base',$i,$j) };
2359                                next unless $mb;
2360                                chomp $mb;
2361                                next if $x{$mb};
2362                                if ($mb eq $j) {
2363                                        delete $p->{$i};
2364                                        $del{$i} = 1;
2365                                } elsif ($mb eq $i) {
2366                                        delete $p->{$j};
2367                                        $del{$j} = 1;
2368                                }
2369                        }
2370                }
2371
2372                # if real parents are the only ones in the grafts, drop it
2373                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2374
2375                print $fh $c, ' ', join(' ', sort keys %$p),"\n";
2376        }
2377        if ($comments->{'END'}) {
2378                print $fh $_ foreach @{$comments->{'END'}};
2379        }
2380        close $fh or croak $!;
2381}
2382
2383sub read_url_paths_all {
2384        my ($l_map, $pfx, $p) = @_;
2385        my @dir;
2386        foreach (<$p/*>) {
2387                if (-r "$_/info/url") {
2388                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2389                        my $id = $pfx . basename $_;
2390                        my $url = file_to_s("$_/info/url");
2391                        my ($u, $p) = repo_path_split($url);
2392                        $l_map->{$u}->{$p} = $id;
2393                } elsif (-d $_) {
2394                        push @dir, $_;
2395                }
2396        }
2397        foreach (@dir) {
2398                my $x = $_;
2399                $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
2400                read_url_paths_all($l_map, $x, $_);
2401        }
2402}
2403
2404# this one only gets ids that have been imported, not new ones
2405sub read_url_paths {
2406        my $l_map = {};
2407        git_svn_each(sub { my $x = shift;
2408                        my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
2409                        my ($u, $p) = repo_path_split($url);
2410                        $l_map->{$u}->{$p} = $x;
2411                        });
2412        return $l_map;
2413}
2414
2415sub extract_metadata {
2416        my $id = shift or return (undef, undef, undef);
2417        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
2418                                                        \s([a-f\d\-]+)$/x);
2419        if (!$rev || !$uuid || !$url) {
2420                # some of the original repositories I made had
2421                # indentifiers like this:
2422                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
2423        }
2424        return ($url, $rev, $uuid);
2425}
2426
2427sub cmt_metadata {
2428        return extract_metadata((grep(/^git-svn-id: /,
2429                safe_qx(qw/git-cat-file commit/, shift)))[-1]);
2430}
2431
2432sub get_commit_time {
2433        my $cmt = shift;
2434        defined(my $pid = open my $fh, '-|') or croak $!;
2435        if (!$pid) {
2436                exec qw/git-rev-list --pretty=raw -n1/, $cmt or croak $!;
2437        }
2438        while (<$fh>) {
2439                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
2440                my ($s, $tz) = ($1, $2);
2441                if ($tz =~ s/^\+//) {
2442                        $s += tz_to_s_offset($tz);
2443                } elsif ($tz =~ s/^\-//) {
2444                        $s -= tz_to_s_offset($tz);
2445                }
2446                close $fh;
2447                return $s;
2448        }
2449        die "Can't get commit time for commit: $cmt\n";
2450}
2451
2452sub tz_to_s_offset {
2453        my ($tz) = @_;
2454        $tz =~ s/(\d\d)$//;
2455        return ($1 * 60) + ($tz * 3600);
2456}
2457
2458sub setup_pager { # translated to Perl from pager.c
2459        return unless (-t *STDOUT);
2460        my $pager = $ENV{PAGER};
2461        if (!defined $pager) {
2462                $pager = 'less';
2463        } elsif (length $pager == 0 || $pager eq 'cat') {
2464                return;
2465        }
2466        pipe my $rfd, my $wfd or return;
2467        defined(my $pid = fork) or croak $!;
2468        if (!$pid) {
2469                open STDOUT, '>&', $wfd or croak $!;
2470                return;
2471        }
2472        open STDIN, '<&', $rfd or croak $!;
2473        $ENV{LESS} ||= '-S';
2474        exec $pager or croak "Can't run pager: $!\n";;
2475}
2476
2477sub get_author_info {
2478        my ($dest, $author, $t, $tz) = @_;
2479        $author =~ s/(?:^\s*|\s*$)//g;
2480        $dest->{a_raw} = $author;
2481        my $_a;
2482        if ($_authors) {
2483                $_a = $rusers{$author} || undef;
2484        }
2485        if (!$_a) {
2486                ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
2487        }
2488        $dest->{t} = $t;
2489        $dest->{tz} = $tz;
2490        $dest->{a} = $_a;
2491        # Date::Parse isn't in the standard Perl distro :(
2492        if ($tz =~ s/^\+//) {
2493                $t += tz_to_s_offset($tz);
2494        } elsif ($tz =~ s/^\-//) {
2495                $t -= tz_to_s_offset($tz);
2496        }
2497        $dest->{t_utc} = $t;
2498}
2499
2500sub process_commit {
2501        my ($c, $r_min, $r_max, $defer) = @_;
2502        if (defined $r_min && defined $r_max) {
2503                if ($r_min == $c->{r} && $r_min == $r_max) {
2504                        show_commit($c);
2505                        return 0;
2506                }
2507                return 1 if $r_min == $r_max;
2508                if ($r_min < $r_max) {
2509                        # we need to reverse the print order
2510                        return 0 if (defined $_limit && --$_limit < 0);
2511                        push @$defer, $c;
2512                        return 1;
2513                }
2514                if ($r_min != $r_max) {
2515                        return 1 if ($r_min < $c->{r});
2516                        return 1 if ($r_max > $c->{r});
2517                }
2518        }
2519        return 0 if (defined $_limit && --$_limit < 0);
2520        show_commit($c);
2521        return 1;
2522}
2523
2524sub show_commit {
2525        my $c = shift;
2526        if ($_oneline) {
2527                my $x = "\n";
2528                if (my $l = $c->{l}) {
2529                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2530                        $x = $l->[0];
2531                }
2532                $_l_fmt ||= 'A' . length($c->{r});
2533                print 'r',pack($_l_fmt, $c->{r}),' | ';
2534                print "$c->{c} | " if $_show_commit;
2535                print $x;
2536        } else {
2537                show_commit_normal($c);
2538        }
2539}
2540
2541sub show_commit_normal {
2542        my ($c) = @_;
2543        print '-' x72, "\nr$c->{r} | ";
2544        print "$c->{c} | " if $_show_commit;
2545        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2546                                 localtime($c->{t_utc})), ' | ';
2547        my $nr_line = 0;
2548
2549        if (my $l = $c->{l}) {
2550                while ($l->[$#$l] eq "\n" && $l->[($#$l - 1)] eq "\n") {
2551                        pop @$l;
2552                }
2553                $nr_line = scalar @$l;
2554                if (!$nr_line) {
2555                        print "1 line\n\n\n";
2556                } else {
2557                        if ($nr_line == 1) {
2558                                $nr_line = '1 line';
2559                        } else {
2560                                $nr_line .= ' lines';
2561                        }
2562                        print $nr_line, "\n\n";
2563                        print $_ foreach @$l;
2564                }
2565        } else {
2566                print "1 line\n\n";
2567
2568        }
2569        foreach my $x (qw/raw diff/) {
2570                if ($c->{$x}) {
2571                        print "\n";
2572                        print $_ foreach @{$c->{$x}}
2573                }
2574        }
2575}
2576
2577sub libsvn_load {
2578        return unless $_use_lib;
2579        $_use_lib = eval {
2580                require SVN::Core;
2581                if ($SVN::Core::VERSION lt '1.1.0') {
2582                        die "Need SVN::Core 1.1.0 or better ",
2583                                        "(got $SVN::Core::VERSION) ",
2584                                        "Falling back to command-line svn\n";
2585                }
2586                require SVN::Ra;
2587                require SVN::Delta;
2588                push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
2589                my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2590                                        $SVN::Node::dir.$SVN::Node::unknown.
2591                                        $SVN::Node::none.$SVN::Node::file.
2592                                        $SVN::Node::dir.$SVN::Node::unknown;
2593                1;
2594        };
2595}
2596
2597sub libsvn_connect {
2598        my ($url) = @_;
2599        my $auth = SVN::Core::auth_open([SVN::Client::get_simple_provider(),
2600                          SVN::Client::get_ssl_server_trust_file_provider(),
2601                          SVN::Client::get_username_provider()]);
2602        my $s = eval { SVN::Ra->new(url => $url, auth => $auth) };
2603        return $s;
2604}
2605
2606sub libsvn_get_file {
2607        my ($gui, $f, $rev) = @_;
2608        my $p = $f;
2609        return unless ($p =~ s#^\Q$SVN_PATH\E/?##);
2610
2611        my ($hash, $pid, $in, $out);
2612        my $pool = SVN::Pool->new;
2613        defined($pid = open3($in, $out, '>&STDERR',
2614                                qw/git-hash-object -w --stdin/)) or croak $!;
2615        # redirect STDOUT for SVN 1.1.x compatibility
2616        open my $stdout, '>&', \*STDOUT or croak $!;
2617        open STDOUT, '>&', $in or croak $!;
2618        my ($r, $props) = $SVN->get_file($f, $rev, \*STDOUT, $pool);
2619        $in->flush == 0 or croak $!;
2620        open STDOUT, '>&', $stdout or croak $!;
2621        close $in or croak $!;
2622        close $stdout or croak $!;
2623        $pool->clear;
2624        chomp($hash = do { local $/; <$out> });
2625        close $out or croak $!;
2626        waitpid $pid, 0;
2627        $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2628
2629        my $mode = exists $props->{'svn:executable'} ? '100755' : '100644';
2630        if (exists $props->{'svn:special'}) {
2631                $mode = '120000';
2632                my $link = `git-cat-file blob $hash`;
2633                $link =~ s/^link // or die "svn:special file with contents: <",
2634                                                $link, "> is not understood\n";
2635                defined($pid = open3($in, $out, '>&STDERR',
2636                                qw/git-hash-object -w --stdin/)) or croak $!;
2637                print $in $link;
2638                $in->flush == 0 or croak $!;
2639                close $in or croak $!;
2640                chomp($hash = do { local $/; <$out> });
2641                close $out or croak $!;
2642                waitpid $pid, 0;
2643                $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2644        }
2645        print $gui $mode,' ',$hash,"\t",$p,"\0" or croak $!;
2646}
2647
2648sub libsvn_log_entry {
2649        my ($rev, $author, $date, $msg, $parents) = @_;
2650        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2651                                         (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
2652                                or die "Unable to parse date: $date\n";
2653        if (defined $_authors && ! defined $users{$author}) {
2654                die "Author: $author not defined in $_authors file\n";
2655        }
2656        return { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
2657                author => $author, msg => $msg."\n", parents => $parents || [] }
2658}
2659
2660sub process_rm {
2661        my ($gui, $last_commit, $f) = @_;
2662        $f =~ s#^\Q$SVN_PATH\E/?## or return;
2663        # remove entire directories.
2664        if (safe_qx('git-ls-tree',$last_commit,'--',$f) =~ /^040000 tree/) {
2665                defined(my $pid = open my $ls, '-|') or croak $!;
2666                if (!$pid) {
2667                        exec(qw/git-ls-tree -r --name-only -z/,
2668                                $last_commit,'--',$f) or croak $!;
2669                }
2670                local $/ = "\0";
2671                while (<$ls>) {
2672                        print $gui '0 ',0 x 40,"\t",$_ or croak $!;
2673                }
2674                close $ls or croak $?;
2675        } else {
2676                print $gui '0 ',0 x 40,"\t",$f,"\0" or croak $!;
2677        }
2678}
2679
2680sub libsvn_fetch {
2681        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
2682        open my $gui, '| git-update-index -z --index-info' or croak $!;
2683        my @amr;
2684        foreach my $f (keys %$paths) {
2685                my $m = $paths->{$f}->action();
2686                $f =~ s#^/+##;
2687                if ($m =~ /^[DR]$/) {
2688                        process_rm($gui, $last_commit, $f);
2689                        next if $m eq 'D';
2690                        # 'R' can be file replacements, too, right?
2691                }
2692                my $pool = SVN::Pool->new;
2693                my $t = $SVN->check_path($f, $rev, $pool);
2694                if ($t == $SVN::Node::file) {
2695                        if ($m =~ /^[AMR]$/) {
2696                                push @amr, $f;
2697                        } else {
2698                                die "Unrecognized action: $m, ($f r$rev)\n";
2699                        }
2700                }
2701                $pool->clear;
2702        }
2703        libsvn_get_file($gui, $_, $rev) foreach (@amr);
2704        close $gui or croak $?;
2705        return libsvn_log_entry($rev, $author, $date, $msg, [$last_commit]);
2706}
2707
2708sub svn_grab_base_rev {
2709        defined(my $pid = open my $fh, '-|') or croak $!;
2710        if (!$pid) {
2711                open my $null, '>', '/dev/null' or croak $!;
2712                open STDERR, '>&', $null or croak $!;
2713                exec qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0"
2714                                                                or croak $!;
2715        }
2716        chomp(my $c = do { local $/; <$fh> });
2717        close $fh;
2718        if (defined $c && length $c) {
2719                my ($url, $rev, $uuid) = cmt_metadata($c);
2720                return ($rev, $c) if defined $rev;
2721        }
2722        if ($_no_metadata) {
2723                my $offset = -41; # from tail
2724                my $rl;
2725                open my $fh, '<', $REVDB or
2726                        die "--no-metadata specified and $REVDB not readable\n";
2727                seek $fh, $offset, 2;
2728                $rl = readline $fh;
2729                defined $rl or return (undef, undef);
2730                chomp $rl;
2731                while ($c ne $rl && tell $fh != 0) {
2732                        $offset -= 41;
2733                        seek $fh, $offset, 2;
2734                        $rl = readline $fh;
2735                        defined $rl or return (undef, undef);
2736                        chomp $rl;
2737                }
2738                my $rev = tell $fh;
2739                croak $! if ($rev < -1);
2740                $rev =  ($rev - 41) / 41;
2741                close $fh or croak $!;
2742                return ($rev, $c);
2743        }
2744        return (undef, undef);
2745}
2746
2747sub libsvn_parse_revision {
2748        my $base = shift;
2749        my $head = $SVN->get_latest_revnum();
2750        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
2751                return ($base + 1, $head) if (defined $base);
2752                return (0, $head);
2753        }
2754        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
2755        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
2756        if ($_revision =~ /^BASE:(\d+)$/) {
2757                return ($base + 1, $1) if (defined $base);
2758                return (0, $head);
2759        }
2760        return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
2761        die "revision argument: $_revision not understood by git-svn\n",
2762                "Try using the command-line svn client instead\n";
2763}
2764
2765sub libsvn_traverse {
2766        my ($gui, $pfx, $path, $rev) = @_;
2767        my $cwd = "$pfx/$path";
2768        my $pool = SVN::Pool->new;
2769        $cwd =~ s#^/+##g;
2770        my ($dirent, $r, $props) = $SVN->get_dir($cwd, $rev, $pool);
2771        foreach my $d (keys %$dirent) {
2772                my $t = $dirent->{$d}->kind;
2773                if ($t == $SVN::Node::dir) {
2774                        libsvn_traverse($gui, $cwd, $d, $rev);
2775                } elsif ($t == $SVN::Node::file) {
2776                        libsvn_get_file($gui, "$cwd/$d", $rev);
2777                }
2778        }
2779        $pool->clear;
2780}
2781
2782sub libsvn_traverse_ignore {
2783        my ($fh, $path, $r) = @_;
2784        $path =~ s#^/+##g;
2785        my $pool = SVN::Pool->new;
2786        my ($dirent, undef, $props) = $SVN->get_dir($path, $r, $pool);
2787        my $p = $path;
2788        $p =~ s#^\Q$SVN_PATH\E/?##;
2789        print $fh length $p ? "\n# $p\n" : "\n# /\n";
2790        if (my $s = $props->{'svn:ignore'}) {
2791                $s =~ s/[\r\n]+/\n/g;
2792                chomp $s;
2793                if (length $p == 0) {
2794                        $s =~ s#\n#\n/$p#g;
2795                        print $fh "/$s\n";
2796                } else {
2797                        $s =~ s#\n#\n/$p/#g;
2798                        print $fh "/$p/$s\n";
2799                }
2800        }
2801        foreach (sort keys %$dirent) {
2802                next if $dirent->{$_}->kind != $SVN::Node::dir;
2803                libsvn_traverse_ignore($fh, "$path/$_", $r);
2804        }
2805        $pool->clear;
2806}
2807
2808sub revisions_eq {
2809        my ($path, $r0, $r1) = @_;
2810        return 1 if $r0 == $r1;
2811        my $nr = 0;
2812        if ($_use_lib) {
2813                # should be OK to use Pool here (r1 - r0) should be small
2814                my $pool = SVN::Pool->new;
2815                libsvn_get_log($SVN, "/$path", $r0, $r1,
2816                                0, 1, 1, sub {$nr++}, $pool);
2817                $pool->clear;
2818        } else {
2819                my ($url, undef) = repo_path_split($SVN_URL);
2820                my $svn_log = svn_log_raw("$url/$path","-r$r0:$r1");
2821                while (next_log_entry($svn_log)) { $nr++ }
2822                close $svn_log->{fh};
2823        }
2824        return 0 if ($nr > 1);
2825        return 1;
2826}
2827
2828sub libsvn_find_parent_branch {
2829        my ($paths, $rev, $author, $date, $msg) = @_;
2830        my $svn_path = '/'.$SVN_PATH;
2831
2832        # look for a parent from another branch:
2833        my $i = $paths->{$svn_path} or return;
2834        my $branch_from = $i->copyfrom_path or return;
2835        my $r = $i->copyfrom_rev;
2836        print STDERR  "Found possible branch point: ",
2837                                "$branch_from => $svn_path, $r\n";
2838        $branch_from =~ s#^/##;
2839        my $l_map = {};
2840        read_url_paths_all($l_map, '', "$GIT_DIR/svn");
2841        my $url = $SVN->{url};
2842        defined $l_map->{$url} or return;
2843        my $id = $l_map->{$url}->{$branch_from};
2844        if (!defined $id && $_follow_parent) {
2845                print STDERR "Following parent: $branch_from\@$r\n";
2846                # auto create a new branch and follow it
2847                $id = basename($branch_from);
2848                $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
2849                while (-r "$GIT_DIR/svn/$id") {
2850                        # just grow a tail if we're not unique enough :x
2851                        $id .= '-';
2852                }
2853        }
2854        return unless defined $id;
2855
2856        my ($r0, $parent) = find_rev_before($r,$id,1);
2857        if ($_follow_parent && (!defined $r0 || !defined $parent)) {
2858                defined(my $pid = fork) or croak $!;
2859                if (!$pid) {
2860                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2861                        init_vars();
2862                        $SVN_URL = "$url/$branch_from";
2863                        $SVN_LOG = $SVN = undef;
2864                        setup_git_svn();
2865                        # we can't assume SVN_URL exists at r+1:
2866                        $_revision = "0:$r";
2867                        fetch_lib();
2868                        exit 0;
2869                }
2870                waitpid $pid, 0;
2871                croak $? if $?;
2872                ($r0, $parent) = find_rev_before($r,$id,1);
2873        }
2874        return unless (defined $r0 && defined $parent);
2875        if (revisions_eq($branch_from, $r0, $r)) {
2876                unlink $GIT_SVN_INDEX;
2877                print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
2878                sys(qw/git-read-tree/, $parent);
2879                return libsvn_fetch($parent, $paths, $rev,
2880                                        $author, $date, $msg);
2881        }
2882        print STDERR "Nope, branch point not imported or unknown\n";
2883        return undef;
2884}
2885
2886sub libsvn_get_log {
2887        my ($ra, @args) = @_;
2888        if ($SVN::Core::VERSION le '1.2.0') {
2889                splice(@args, 3, 1);
2890        }
2891        $ra->get_log(@args);
2892}
2893
2894sub libsvn_new_tree {
2895        if (my $log_entry = libsvn_find_parent_branch(@_)) {
2896                return $log_entry;
2897        }
2898        my ($paths, $rev, $author, $date, $msg) = @_;
2899        open my $gui, '| git-update-index -z --index-info' or croak $!;
2900        my $pool = SVN::Pool->new;
2901        libsvn_traverse($gui, '', $SVN_PATH, $rev, $pool);
2902        $pool->clear;
2903        close $gui or croak $?;
2904        return libsvn_log_entry($rev, $author, $date, $msg);
2905}
2906
2907sub find_graft_path_commit {
2908        my ($tree_paths, $p1, $r1) = @_;
2909        foreach my $x (keys %$tree_paths) {
2910                next unless ($p1 =~ /^\Q$x\E/);
2911                my $i = $tree_paths->{$x};
2912                my ($r0, $parent) = find_rev_before($r1,$i,1);
2913                return $parent if (defined $r0 && $r0 == $r1);
2914                print STDERR "r$r1 of $i not imported\n";
2915                next;
2916        }
2917        return undef;
2918}
2919
2920sub find_graft_path_parents {
2921        my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
2922        foreach my $x (keys %$tree_paths) {
2923                next unless ($p0 =~ /^\Q$x\E/);
2924                my $i = $tree_paths->{$x};
2925                my ($r, $parent) = find_rev_before($r0, $i, 1);
2926                if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
2927                        my ($url_b, undef, $uuid_b) = cmt_metadata($c);
2928                        my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
2929                        next if ($url_a && $url_b && $url_a eq $url_b &&
2930                                                        $uuid_b eq $uuid_a);
2931                        $grafts->{$c}->{$parent} = 1;
2932                }
2933        }
2934}
2935
2936sub libsvn_graft_file_copies {
2937        my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
2938        foreach (keys %$paths) {
2939                my $i = $paths->{$_};
2940                my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
2941                                        $i->copyfrom_rev);
2942                next unless (defined $p0 && defined $r0);
2943
2944                my $p1 = $_;
2945                $p1 =~ s#^/##;
2946                $p0 =~ s#^/##;
2947                my $c = find_graft_path_commit($tree_paths, $p1, $rev);
2948                next unless $c;
2949                find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
2950        }
2951}
2952
2953sub set_index {
2954        my $old = $ENV{GIT_INDEX_FILE};
2955        $ENV{GIT_INDEX_FILE} = shift;
2956        return $old;
2957}
2958
2959sub restore_index {
2960        my ($old) = @_;
2961        if (defined $old) {
2962                $ENV{GIT_INDEX_FILE} = $old;
2963        } else {
2964                delete $ENV{GIT_INDEX_FILE};
2965        }
2966}
2967
2968sub libsvn_commit_cb {
2969        my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
2970        if ($_optimize_commits && $rev == ($r_last + 1)) {
2971                my $log = libsvn_log_entry($rev,$committer,$date,$msg);
2972                $log->{tree} = get_tree_from_treeish($c);
2973                my $cmt = git_commit($log, $cmt_last, $c);
2974                my @diff = safe_qx('git-diff-tree', $cmt, $c);
2975                if (@diff) {
2976                        print STDERR "Trees differ: $cmt $c\n",
2977                                        join('',@diff),"\n";
2978                        exit 1;
2979                }
2980        } else {
2981                fetch("$rev=$c");
2982        }
2983}
2984
2985sub libsvn_ls_fullurl {
2986        my $fullurl = shift;
2987        my ($repo, $path) = repo_path_split($fullurl);
2988        $SVN ||= libsvn_connect($repo);
2989        my @ret;
2990        my $pool = SVN::Pool->new;
2991        my ($dirent, undef, undef) = $SVN->get_dir($path,
2992                                                $SVN->get_latest_revnum, $pool);
2993        foreach my $d (keys %$dirent) {
2994                if ($dirent->{$d}->kind == $SVN::Node::dir) {
2995                        push @ret, "$d/"; # add '/' for compat with cli svn
2996                }
2997        }
2998        $pool->clear;
2999        return @ret;
3000}
3001
3002
3003sub libsvn_skip_unknown_revs {
3004        my $err = shift;
3005        my $errno = $err->apr_err();
3006        # Maybe the branch we're tracking didn't
3007        # exist when the repo started, so it's
3008        # not an error if it doesn't, just continue
3009        #
3010        # Wonderfully consistent library, eh?
3011        # 160013 - svn:// and file://
3012        # 175002 - http(s)://
3013        #   More codes may be discovered later...
3014        if ($errno == 175002 || $errno == 160013) {
3015                return;
3016        }
3017        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3018};
3019
3020# Tie::File seems to be prone to offset errors if revisions get sparse,
3021# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3022# one of my favorite modules is out :<  Next up would be one of the DBM
3023# modules, but I'm not sure which is most portable...  So I'll just
3024# go with something that's plain-text, but still capable of
3025# being randomly accessed.  So here's my ultra-simple fixed-width
3026# database.  All records are 40 characters + "\n", so it's easy to seek
3027# to a revision: (41 * rev) is the byte offset.
3028# A record of 40 0s denotes an empty revision.
3029# And yes, it's still pretty fast (faster than Tie::File).
3030sub revdb_set {
3031        my ($file, $rev, $commit) = @_;
3032        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
3033        open my $fh, '+<', $file or croak $!;
3034        my $offset = $rev * 41;
3035        # assume that append is the common case:
3036        seek $fh, 0, 2 or croak $!;
3037        my $pos = tell $fh;
3038        if ($pos < $offset) {
3039                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
3040        }
3041        seek $fh, $offset, 0 or croak $!;
3042        print $fh $commit,"\n";
3043        close $fh or croak $!;
3044}
3045
3046sub revdb_get {
3047        my ($file, $rev) = @_;
3048        my $ret;
3049        my $offset = $rev * 41;
3050        open my $fh, '<', $file or croak $!;
3051        seek $fh, $offset, 0;
3052        if (tell $fh == $offset) {
3053                $ret = readline $fh;
3054                if (defined $ret) {
3055                        chomp $ret;
3056                        $ret = undef if ($ret =~ /^0{40}$/);
3057                }
3058        }
3059        close $fh or croak $!;
3060        return $ret;
3061}
3062
3063sub copy_remote_ref {
3064        my $origin = $_cp_remote ? $_cp_remote : 'origin';
3065        my $ref = "refs/remotes/$GIT_SVN";
3066        if (safe_qx('git-ls-remote', $origin, $ref)) {
3067                sys(qw/git fetch/, $origin, "$ref:$ref");
3068        } else {
3069                die "Unable to find remote reference: ",
3070                                "refs/remotes/$GIT_SVN on $origin\n";
3071        }
3072}
3073
3074package SVN::Git::Editor;
3075use vars qw/@ISA/;
3076use strict;
3077use warnings;
3078use Carp qw/croak/;
3079use IO::File;
3080
3081sub new {
3082        my $class = shift;
3083        my $git_svn = shift;
3084        my $self = SVN::Delta::Editor->new(@_);
3085        bless $self, $class;
3086        foreach (qw/svn_path c r ra /) {
3087                die "$_ required!\n" unless (defined $git_svn->{$_});
3088                $self->{$_} = $git_svn->{$_};
3089        }
3090        $self->{pool} = SVN::Pool->new;
3091        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3092        $self->{rm} = { };
3093        require Digest::MD5;
3094        return $self;
3095}
3096
3097sub split_path {
3098        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3099}
3100
3101sub repo_path {
3102        (defined $_[1] && length $_[1]) ? "$_[0]->{svn_path}/$_[1]"
3103                                        : $_[0]->{svn_path}
3104}
3105
3106sub url_path {
3107        my ($self, $path) = @_;
3108        $self->{ra}->{url} . '/' . $self->repo_path($path);
3109}
3110
3111sub rmdirs {
3112        my ($self) = @_;
3113        my $rm = $self->{rm};
3114        delete $rm->{''}; # we never delete the url we're tracking
3115        return unless %$rm;
3116
3117        foreach (keys %$rm) {
3118                my @d = split m#/#, $_;
3119                my $c = shift @d;
3120                $rm->{$c} = 1;
3121                while (@d) {
3122                        $c .= '/' . shift @d;
3123                        $rm->{$c} = 1;
3124                }
3125        }
3126        delete $rm->{$self->{svn_path}};
3127        delete $rm->{''}; # we never delete the url we're tracking
3128        return unless %$rm;
3129
3130        defined(my $pid = open my $fh,'-|') or croak $!;
3131        if (!$pid) {
3132                exec qw/git-ls-tree --name-only -r -z/, $self->{c} or croak $!;
3133        }
3134        local $/ = "\0";
3135        my @svn_path = split m#/#, $self->{svn_path};
3136        while (<$fh>) {
3137                chomp;
3138                my @dn = (@svn_path, (split m#/#, $_));
3139                while (pop @dn) {
3140                        delete $rm->{join '/', @dn};
3141                }
3142                unless (%$rm) {
3143                        close $fh;
3144                        return;
3145                }
3146        }
3147        close $fh;
3148
3149        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3150        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3151                $self->close_directory($bat->{$d}, $p);
3152                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3153                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3154                delete $bat->{$d};
3155        }
3156}
3157
3158sub open_or_add_dir {
3159        my ($self, $full_path, $baton) = @_;
3160        my $p = SVN::Pool->new;
3161        my $t = $self->{ra}->check_path($full_path, $self->{r}, $p);
3162        $p->clear;
3163        if ($t == $SVN::Node::none) {
3164                return $self->add_directory($full_path, $baton,
3165                                                undef, -1, $self->{pool});
3166        } elsif ($t == $SVN::Node::dir) {
3167                return $self->open_directory($full_path, $baton,
3168                                                $self->{r}, $self->{pool});
3169        }
3170        print STDERR "$full_path already exists in repository at ",
3171                "r$self->{r} and it is not a directory (",
3172                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3173        exit 1;
3174}
3175
3176sub ensure_path {
3177        my ($self, $path) = @_;
3178        my $bat = $self->{bat};
3179        $path = $self->repo_path($path);
3180        return $bat->{''} unless (length $path);
3181        my @p = split m#/+#, $path;
3182        my $c = shift @p;
3183        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3184        while (@p) {
3185                my $c0 = $c;
3186                $c .= '/' . shift @p;
3187                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3188        }
3189        return $bat->{$c};
3190}
3191
3192sub A {
3193        my ($self, $m) = @_;
3194        my ($dir, $file) = split_path($m->{file_b});
3195        my $pbat = $self->ensure_path($dir);
3196        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3197                                        undef, -1);
3198        $self->chg_file($fbat, $m);
3199        $self->close_file($fbat,undef,$self->{pool});
3200}
3201
3202sub C {
3203        my ($self, $m) = @_;
3204        my ($dir, $file) = split_path($m->{file_b});
3205        my $pbat = $self->ensure_path($dir);
3206        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3207                                $self->url_path($m->{file_a}), $self->{r});
3208        $self->chg_file($fbat, $m);
3209        $self->close_file($fbat,undef,$self->{pool});
3210}
3211
3212sub delete_entry {
3213        my ($self, $path, $pbat) = @_;
3214        my $rpath = $self->repo_path($path);
3215        my ($dir, $file) = split_path($rpath);
3216        $self->{rm}->{$dir} = 1;
3217        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3218}
3219
3220sub R {
3221        my ($self, $m) = @_;
3222        my ($dir, $file) = split_path($m->{file_b});
3223        my $pbat = $self->ensure_path($dir);
3224        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3225                                $self->url_path($m->{file_a}), $self->{r});
3226        $self->chg_file($fbat, $m);
3227        $self->close_file($fbat,undef,$self->{pool});
3228
3229        ($dir, $file) = split_path($m->{file_a});
3230        $pbat = $self->ensure_path($dir);
3231        $self->delete_entry($m->{file_a}, $pbat);
3232}
3233
3234sub M {
3235        my ($self, $m) = @_;
3236        my ($dir, $file) = split_path($m->{file_b});
3237        my $pbat = $self->ensure_path($dir);
3238        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3239                                $pbat,$self->{r},$self->{pool});
3240        $self->chg_file($fbat, $m);
3241        $self->close_file($fbat,undef,$self->{pool});
3242}
3243
3244sub T { shift->M(@_) }
3245
3246sub change_file_prop {
3247        my ($self, $fbat, $pname, $pval) = @_;
3248        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3249}
3250
3251sub chg_file {
3252        my ($self, $fbat, $m) = @_;
3253        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3254                $self->change_file_prop($fbat,'svn:executable','*');
3255        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3256                $self->change_file_prop($fbat,'svn:executable',undef);
3257        }
3258        my $fh = IO::File->new_tmpfile or croak $!;
3259        if ($m->{mode_b} =~ /^120/) {
3260                print $fh 'link ' or croak $!;
3261                $self->change_file_prop($fbat,'svn:special','*');
3262        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3263                $self->change_file_prop($fbat,'svn:special',undef);
3264        }
3265        defined(my $pid = fork) or croak $!;
3266        if (!$pid) {
3267                open STDOUT, '>&', $fh or croak $!;
3268                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3269        }
3270        waitpid $pid, 0;
3271        croak $? if $?;
3272        $fh->flush == 0 or croak $!;
3273        seek $fh, 0, 0 or croak $!;
3274
3275        my $md5 = Digest::MD5->new;
3276        $md5->addfile($fh) or croak $!;
3277        seek $fh, 0, 0 or croak $!;
3278
3279        my $exp = $md5->hexdigest;
3280        my $atd = $self->apply_textdelta($fbat, undef, $self->{pool});
3281        my $got = SVN::TxDelta::send_stream($fh, @$atd, $self->{pool});
3282        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3283
3284        close $fh or croak $!;
3285}
3286
3287sub D {
3288        my ($self, $m) = @_;
3289        my ($dir, $file) = split_path($m->{file_b});
3290        my $pbat = $self->ensure_path($dir);
3291        $self->delete_entry($m->{file_b}, $pbat);
3292}
3293
3294sub close_edit {
3295        my ($self) = @_;
3296        my ($p,$bat) = ($self->{pool}, $self->{bat});
3297        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3298                $self->close_directory($bat->{$_}, $p);
3299        }
3300        $self->SUPER::close_edit($p);
3301        $p->clear;
3302}
3303
3304sub abort_edit {
3305        my ($self) = @_;
3306        $self->SUPER::abort_edit($self->{pool});
3307        $self->{pool}->clear;
3308}
3309
3310__END__
3311
3312Data structures:
3313
3314$svn_log hashref (as returned by svn_log_raw)
3315{
3316        fh => file handle of the log file,
3317        state => state of the log file parser (sep/msg/rev/msg_start...)
3318}
3319
3320$log_msg hashref as returned by next_log_entry($svn_log)
3321{
3322        msg => 'whitespace-formatted log entry
3323',                                              # trailing newline is preserved
3324        revision => '8',                        # integer
3325        date => '2004-02-24T17:01:44.108345Z',  # commit date
3326        author => 'committer name'
3327};
3328
3329
3330@mods = array of diff-index line hashes, each element represents one line
3331        of diff-index output
3332
3333diff-index line ($m hash)
3334{
3335        mode_a => first column of diff-index output, no leading ':',
3336        mode_b => second column of diff-index output,
3337        sha1_b => sha1sum of the final blob,
3338        chg => change type [MCRADT],
3339        file_a => original file name of a file (iff chg is 'C' or 'R')
3340        file_b => new/current file name of a file (any chg)
3341}
3342;
3343
3344# retval of read_url_paths{,_all}();
3345$l_map = {
3346        # repository root url
3347        'https://svn.musicpd.org' => {
3348                # repository path               # GIT_SVN_ID
3349                'mpd/trunk'             =>      'trunk',
3350                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
3351        },
3352}
3353
3354Notes:
3355        I don't trust the each() function on unless I created %hash myself
3356        because the internal iterator may not have started at base.