git-svn.perlon commit Merge branch 'maint' (eab9021)
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $SVN_URL $SVN_INFO $SVN_WC $SVN_UUID
   8                $GIT_SVN_INDEX $GIT_SVN
   9                $GIT_DIR $GIT_SVN_DIR $REVDB/;
  10$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  11$VERSION = '@@GIT_VERSION@@';
  12
  13use Cwd qw/abs_path/;
  14$GIT_DIR = abs_path($ENV{GIT_DIR} || '.git');
  15$ENV{GIT_DIR} = $GIT_DIR;
  16
  17my $LC_ALL = $ENV{LC_ALL};
  18my $TZ = $ENV{TZ};
  19# make sure the svn binary gives consistent output between locales and TZs:
  20$ENV{TZ} = 'UTC';
  21$ENV{LC_ALL} = 'C';
  22$| = 1; # unbuffer STDOUT
  23
  24# properties that we do not log:
  25my %SKIP = ( 'svn:wc:ra_dav:version-url' => 1,
  26             'svn:special' => 1,
  27             'svn:executable' => 1,
  28             'svn:entry:committed-rev' => 1,
  29             'svn:entry:last-author' => 1,
  30             'svn:entry:uuid' => 1,
  31             'svn:entry:committed-date' => 1,
  32);
  33
  34sub fatal (@) { print STDERR @_; exit 1 }
  35# If SVN:: library support is added, please make the dependencies
  36# optional and preserve the capability to use the command-line client.
  37# use eval { require SVN::... } to make it lazy load
  38# We don't use any modules not in the standard Perl distribution:
  39use Carp qw/croak/;
  40use IO::File qw//;
  41use File::Basename qw/dirname basename/;
  42use File::Path qw/mkpath/;
  43use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  44use File::Spec qw//;
  45use File::Copy qw/copy/;
  46use POSIX qw/strftime/;
  47use IPC::Open3;
  48use Memoize;
  49memoize('revisions_eq');
  50memoize('cmt_metadata');
  51memoize('get_commit_time');
  52
  53my ($SVN, $_use_lib);
  54
  55sub nag_lib {
  56        print STDERR <<EOF;
  57! Please consider installing the SVN Perl libraries (version 1.1.0 or
  58! newer).  You will generally get better performance and fewer bugs,
  59! especially if you:
  60! 1) have a case-insensitive filesystem
  61! 2) replace symlinks with files (and vice-versa) in commits
  62
  63EOF
  64}
  65
  66$_use_lib = 1 unless $ENV{GIT_SVN_NO_LIB};
  67libsvn_load();
  68nag_lib() unless $_use_lib;
  69
  70my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  71my $sha1 = qr/[a-f\d]{40}/;
  72my $sha1_short = qr/[a-f\d]{4,40}/;
  73my $_esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
  74my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
  75        $_find_copies_harder, $_l, $_cp_similarity, $_cp_remote,
  76        $_repack, $_repack_nr, $_repack_flags, $_q,
  77        $_message, $_file, $_follow_parent, $_no_metadata,
  78        $_template, $_shared, $_no_default_regex, $_no_graft_copy,
  79        $_limit, $_verbose, $_incremental, $_oneline, $_l_fmt, $_show_commit,
  80        $_version, $_upgrade, $_authors, $_branch_all_refs, @_opt_m,
  81        $_merge, $_strategy, $_dry_run, $_ignore_nodate, $_non_recursive,
  82        $_username, $_config_dir, $_no_auth_cache, $_xfer_delta,
  83        $_pager, $_color);
  84my (@_branch_from, %tree_map, %users, %rusers, %equiv);
  85my ($_svn_co_url_revs, $_svn_pg_peg_revs, $_svn_can_do_switch);
  86my @repo_path_split_cache;
  87
  88my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
  89                'branch|b=s' => \@_branch_from,
  90                'follow-parent|follow' => \$_follow_parent,
  91                'branch-all-refs|B' => \$_branch_all_refs,
  92                'authors-file|A=s' => \$_authors,
  93                'repack:i' => \$_repack,
  94                'no-metadata' => \$_no_metadata,
  95                'quiet|q' => \$_q,
  96                'username=s' => \$_username,
  97                'config-dir=s' => \$_config_dir,
  98                'no-auth-cache' => \$_no_auth_cache,
  99                'ignore-nodate' => \$_ignore_nodate,
 100                'repack-flags|repack-args|repack-opts=s' => \$_repack_flags);
 101
 102my ($_trunk, $_tags, $_branches);
 103my %multi_opts = ( 'trunk|T=s' => \$_trunk,
 104                'tags|t=s' => \$_tags,
 105                'branches|b=s' => \$_branches );
 106my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
 107my %cmt_opts = ( 'edit|e' => \$_edit,
 108                'rmdir' => \$_rmdir,
 109                'find-copies-harder' => \$_find_copies_harder,
 110                'l=i' => \$_l,
 111                'copy-similarity|C=i'=> \$_cp_similarity
 112);
 113
 114my %cmd = (
 115        fetch => [ \&fetch, "Download new revisions from SVN",
 116                        { 'revision|r=s' => \$_revision, %fc_opts } ],
 117        init => [ \&init, "Initialize a repo for tracking" .
 118                          " (requires URL argument)",
 119                          \%init_opts ],
 120        commit => [ \&commit, "Commit git revisions to SVN",
 121                        {       'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 122        'show-ignore' => [ \&show_ignore, "Show svn:ignore listings",
 123                        { 'revision|r=i' => \$_revision } ],
 124        rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
 125                        { 'no-ignore-externals' => \$_no_ignore_ext,
 126                          'copy-remote|remote=s' => \$_cp_remote,
 127                          'upgrade' => \$_upgrade } ],
 128        'graft-branches' => [ \&graft_branches,
 129                        'Detect merges/branches from already imported history',
 130                        { 'merge-rx|m' => \@_opt_m,
 131                          'branch|b=s' => \@_branch_from,
 132                          'branch-all-refs|B' => \$_branch_all_refs,
 133                          'no-default-regex' => \$_no_default_regex,
 134                          'no-graft-copy' => \$_no_graft_copy } ],
 135        'multi-init' => [ \&multi_init,
 136                        'Initialize multiple trees (like git-svnimport)',
 137                        { %multi_opts, %init_opts,
 138                         'revision|r=i' => \$_revision,
 139                         'username=s' => \$_username,
 140                         'config-dir=s' => \$_config_dir,
 141                         'no-auth-cache' => \$_no_auth_cache,
 142                        } ],
 143        'multi-fetch' => [ \&multi_fetch,
 144                        'Fetch multiple trees (like git-svnimport)',
 145                        \%fc_opts ],
 146        'log' => [ \&show_log, 'Show commit logs',
 147                        { 'limit=i' => \$_limit,
 148                          'revision|r=s' => \$_revision,
 149                          'verbose|v' => \$_verbose,
 150                          'incremental' => \$_incremental,
 151                          'oneline' => \$_oneline,
 152                          'show-commit' => \$_show_commit,
 153                          'non-recursive' => \$_non_recursive,
 154                          'authors-file|A=s' => \$_authors,
 155                          'color' => \$_color,
 156                          'pager=s' => \$_pager,
 157                        } ],
 158        'commit-diff' => [ \&commit_diff, 'Commit a diff between two trees',
 159                        { 'message|m=s' => \$_message,
 160                          'file|F=s' => \$_file,
 161                          'revision|r=s' => \$_revision,
 162                        %cmt_opts } ],
 163        dcommit => [ \&dcommit, 'Commit several diffs to merge with upstream',
 164                        { 'merge|m|M' => \$_merge,
 165                          'strategy|s=s' => \$_strategy,
 166                          'dry-run|n' => \$_dry_run,
 167                        %cmt_opts } ],
 168);
 169
 170my $cmd;
 171for (my $i = 0; $i < @ARGV; $i++) {
 172        if (defined $cmd{$ARGV[$i]}) {
 173                $cmd = $ARGV[$i];
 174                splice @ARGV, $i, 1;
 175                last;
 176        }
 177};
 178
 179my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 180
 181read_repo_config(\%opts);
 182my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
 183                                'version|V' => \$_version,
 184                                'id|i=s' => \$GIT_SVN);
 185exit 1 if (!$rv && $cmd ne 'log');
 186
 187set_default_vals();
 188usage(0) if $_help;
 189version() if $_version;
 190usage(1) unless defined $cmd;
 191init_vars();
 192load_authors() if $_authors;
 193load_all_refs() if $_branch_all_refs;
 194svn_compat_check() unless $_use_lib;
 195migration_check() unless $cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/;
 196$cmd{$cmd}->[0]->(@ARGV);
 197exit 0;
 198
 199####################### primary functions ######################
 200sub usage {
 201        my $exit = shift || 0;
 202        my $fd = $exit ? \*STDERR : \*STDOUT;
 203        print $fd <<"";
 204git-svn - bidirectional operations between a single Subversion tree and git
 205Usage: $0 <command> [options] [arguments]\n
 206
 207        print $fd "Available commands:\n" unless $cmd;
 208
 209        foreach (sort keys %cmd) {
 210                next if $cmd && $cmd ne $_;
 211                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 212                foreach (keys %{$cmd{$_}->[2]}) {
 213                        # prints out arguments as they should be passed:
 214                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 215                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 216                                                        "--$_" : "-$_" }
 217                                                split /\|/,$_)," $x\n";
 218                }
 219        }
 220        print $fd <<"";
 221\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 222arbitrary identifier if you're tracking multiple SVN branches/repositories in
 223one git repository and want to keep them separate.  See git-svn(1) for more
 224information.
 225
 226        exit $exit;
 227}
 228
 229sub version {
 230        print "git-svn version $VERSION\n";
 231        exit 0;
 232}
 233
 234sub rebuild {
 235        if (quiet_run(qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0")) {
 236                copy_remote_ref();
 237        }
 238        $SVN_URL = shift or undef;
 239        my $newest_rev = 0;
 240        if ($_upgrade) {
 241                sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD");
 242        } else {
 243                check_upgrade_needed();
 244        }
 245
 246        my $pid = open(my $rev_list,'-|');
 247        defined $pid or croak $!;
 248        if ($pid == 0) {
 249                exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!;
 250        }
 251        my $latest;
 252        while (<$rev_list>) {
 253                chomp;
 254                my $c = $_;
 255                croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 256                my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
 257                next if (!@commit); # skip merges
 258                my ($url, $rev, $uuid) = extract_metadata($commit[$#commit]);
 259                if (!defined $rev || !$uuid) {
 260                        croak "Unable to extract revision or UUID from ",
 261                                "$c, $commit[$#commit]\n";
 262                }
 263
 264                # if we merged or otherwise started elsewhere, this is
 265                # how we break out of it
 266                next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
 267                next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
 268
 269                unless (defined $latest) {
 270                        if (!$SVN_URL && !$url) {
 271                                croak "SVN repository location required: $url\n";
 272                        }
 273                        $SVN_URL ||= $url;
 274                        $SVN_UUID ||= $uuid;
 275                        setup_git_svn();
 276                        $latest = $rev;
 277                }
 278                revdb_set($REVDB, $rev, $c);
 279                print "r$rev = $c\n";
 280                $newest_rev = $rev if ($rev > $newest_rev);
 281        }
 282        close $rev_list or croak $?;
 283
 284        goto out if $_use_lib;
 285        if (!chdir $SVN_WC) {
 286                svn_cmd_checkout($SVN_URL, $latest, $SVN_WC);
 287                chdir $SVN_WC or croak $!;
 288        }
 289
 290        $pid = fork;
 291        defined $pid or croak $!;
 292        if ($pid == 0) {
 293                my @svn_up = qw(svn up);
 294                push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 295                sys(@svn_up,"-r$newest_rev");
 296                $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 297                index_changes();
 298                exec('git-write-tree') or croak $!;
 299        }
 300        waitpid $pid, 0;
 301        croak $? if $?;
 302out:
 303        if ($_upgrade) {
 304                print STDERR <<"";
 305Keeping deprecated refs/head/$GIT_SVN-HEAD for now.  Please remove it
 306when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN
 307
 308        }
 309}
 310
 311sub init {
 312        my $url = shift or die "SVN repository location required " .
 313                                "as a command-line argument\n";
 314        $url =~ s!/+$!!; # strip trailing slash
 315
 316        if (my $repo_path = shift) {
 317                unless (-d $repo_path) {
 318                        mkpath([$repo_path]);
 319                }
 320                $GIT_DIR = $ENV{GIT_DIR} = $repo_path . "/.git";
 321                init_vars();
 322        }
 323
 324        $SVN_URL = $url;
 325        unless (-d $GIT_DIR) {
 326                my @init_db = ('git-init-db');
 327                push @init_db, "--template=$_template" if defined $_template;
 328                push @init_db, "--shared" if defined $_shared;
 329                sys(@init_db);
 330        }
 331        setup_git_svn();
 332}
 333
 334sub fetch {
 335        check_upgrade_needed();
 336        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 337        my $ret = $_use_lib ? fetch_lib(@_) : fetch_cmd(@_);
 338        if ($ret->{commit} && quiet_run(qw(git-rev-parse --verify
 339                                                refs/heads/master^0))) {
 340                sys(qw(git-update-ref refs/heads/master),$ret->{commit});
 341        }
 342        return $ret;
 343}
 344
 345sub fetch_cmd {
 346        my (@parents) = @_;
 347        my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 348        unless ($_revision) {
 349                $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 350        }
 351        push @log_args, "-r$_revision";
 352        push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 353
 354        my $svn_log = svn_log_raw(@log_args);
 355
 356        my $base = next_log_entry($svn_log) or croak "No base revision!\n";
 357        # don't need last_revision from grab_base_rev() because
 358        # user could've specified a different revision to skip (they
 359        # didn't want to import certain revisions into git for whatever
 360        # reason, so trust $base->{revision} instead.
 361        my (undef, $last_commit) = svn_grab_base_rev();
 362        unless (-d $SVN_WC) {
 363                svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC);
 364                chdir $SVN_WC or croak $!;
 365                read_uuid();
 366                $last_commit = git_commit($base, @parents);
 367                assert_tree($last_commit);
 368        } else {
 369                chdir $SVN_WC or croak $!;
 370                read_uuid();
 371                # looks like a user manually cp'd and svn switch'ed
 372                unless ($last_commit) {
 373                        sys(qw/svn revert -R ./);
 374                        assert_svn_wc_clean($base->{revision});
 375                        $last_commit = git_commit($base, @parents);
 376                        assert_tree($last_commit);
 377                }
 378        }
 379        my @svn_up = qw(svn up);
 380        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 381        my $last = $base;
 382        while (my $log_msg = next_log_entry($svn_log)) {
 383                if ($last->{revision} >= $log_msg->{revision}) {
 384                        croak "Out of order: last >= current: ",
 385                                "$last->{revision} >= $log_msg->{revision}\n";
 386                }
 387                # Revert is needed for cases like:
 388                # https://svn.musicpd.org/Jamming/trunk (r166:167), but
 389                # I can't seem to reproduce something like that on a test...
 390                sys(qw/svn revert -R ./);
 391                assert_svn_wc_clean($last->{revision});
 392                sys(@svn_up,"-r$log_msg->{revision}");
 393                $last_commit = git_commit($log_msg, $last_commit, @parents);
 394                $last = $log_msg;
 395        }
 396        close $svn_log->{fh};
 397        $last->{commit} = $last_commit;
 398        return $last;
 399}
 400
 401sub fetch_lib {
 402        my (@parents) = @_;
 403        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 404        $SVN ||= libsvn_connect($SVN_URL);
 405        my ($last_rev, $last_commit) = svn_grab_base_rev();
 406        my ($base, $head) = libsvn_parse_revision($last_rev);
 407        if ($base > $head) {
 408                return { revision => $last_rev, commit => $last_commit }
 409        }
 410        my $index = set_index($GIT_SVN_INDEX);
 411
 412        # limit ourselves and also fork() since get_log won't release memory
 413        # after processing a revision and SVN stuff seems to leak
 414        my $inc = 1000;
 415        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
 416        read_uuid();
 417        if (defined $last_commit) {
 418                unless (-e $GIT_SVN_INDEX) {
 419                        sys(qw/git-read-tree/, $last_commit);
 420                }
 421                chomp (my $x = `git-write-tree`);
 422                my ($y) = (`git-cat-file commit $last_commit`
 423                                                        =~ /^tree ($sha1)/m);
 424                if ($y ne $x) {
 425                        unlink $GIT_SVN_INDEX or croak $!;
 426                        sys(qw/git-read-tree/, $last_commit);
 427                }
 428                chomp ($x = `git-write-tree`);
 429                if ($y ne $x) {
 430                        print STDERR "trees ($last_commit) $y != $x\n",
 431                                 "Something is seriously wrong...\n";
 432                }
 433        }
 434        while (1) {
 435                # fork, because using SVN::Pool with get_log() still doesn't
 436                # seem to help enough to keep memory usage down.
 437                defined(my $pid = fork) or croak $!;
 438                if (!$pid) {
 439                        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
 440
 441                        # Yes I'm perfectly aware that the fourth argument
 442                        # below is the limit revisions number.  Unfortunately
 443                        # performance sucks with it enabled, so it's much
 444                        # faster to fetch revision ranges instead of relying
 445                        # on the limiter.
 446                        libsvn_get_log(libsvn_dup_ra($SVN), [''],
 447                                        $min, $max, 0, 1, 1,
 448                                sub {
 449                                        my $log_msg;
 450                                        if ($last_commit) {
 451                                                $log_msg = libsvn_fetch(
 452                                                        $last_commit, @_);
 453                                                $last_commit = git_commit(
 454                                                        $log_msg,
 455                                                        $last_commit,
 456                                                        @parents);
 457                                        } else {
 458                                                $log_msg = libsvn_new_tree(@_);
 459                                                $last_commit = git_commit(
 460                                                        $log_msg, @parents);
 461                                        }
 462                                });
 463                        exit 0;
 464                }
 465                waitpid $pid, 0;
 466                croak $? if $?;
 467                ($last_rev, $last_commit) = svn_grab_base_rev();
 468                last if ($max >= $head);
 469                $min = $max + 1;
 470                $max += $inc;
 471                $max = $head if ($max > $head);
 472                $SVN = libsvn_connect($SVN_URL);
 473        }
 474        restore_index($index);
 475        return { revision => $last_rev, commit => $last_commit };
 476}
 477
 478sub commit {
 479        my (@commits) = @_;
 480        check_upgrade_needed();
 481        if ($_stdin || !@commits) {
 482                print "Reading from stdin...\n";
 483                @commits = ();
 484                while (<STDIN>) {
 485                        if (/\b($sha1_short)\b/o) {
 486                                unshift @commits, $1;
 487                        }
 488                }
 489        }
 490        my @revs;
 491        foreach my $c (@commits) {
 492                chomp(my @tmp = safe_qx('git-rev-parse',$c));
 493                if (scalar @tmp == 1) {
 494                        push @revs, $tmp[0];
 495                } elsif (scalar @tmp > 1) {
 496                        push @revs, reverse (safe_qx('git-rev-list',@tmp));
 497                } else {
 498                        die "Failed to rev-parse $c\n";
 499                }
 500        }
 501        chomp @revs;
 502        $_use_lib ? commit_lib(@revs) : commit_cmd(@revs);
 503        print "Done committing ",scalar @revs," revisions to SVN\n";
 504}
 505
 506sub commit_cmd {
 507        my (@revs) = @_;
 508
 509        chdir $SVN_WC or croak "Unable to chdir $SVN_WC: $!\n";
 510        my $info = svn_info('.');
 511        my $fetched = fetch();
 512        if ($info->{Revision} != $fetched->{revision}) {
 513                print STDERR "There are new revisions that were fetched ",
 514                                "and need to be merged (or acknowledged) ",
 515                                "before committing.\n";
 516                exit 1;
 517        }
 518        $info = svn_info('.');
 519        read_uuid($info);
 520        my $last = $fetched;
 521        foreach my $c (@revs) {
 522                my $mods = svn_checkout_tree($last, $c);
 523                if (scalar @$mods == 0) {
 524                        print "Skipping, no changes detected\n";
 525                        next;
 526                }
 527                $last = svn_commit_tree($last, $c);
 528        }
 529}
 530
 531sub commit_lib {
 532        my (@revs) = @_;
 533        my ($r_last, $cmt_last) = svn_grab_base_rev();
 534        defined $r_last or die "Must have an existing revision to commit\n";
 535        my $fetched = fetch();
 536        if ($r_last != $fetched->{revision}) {
 537                print STDERR "There are new revisions that were fetched ",
 538                                "and need to be merged (or acknowledged) ",
 539                                "before committing.\n",
 540                                "last rev: $r_last\n",
 541                                " current: $fetched->{revision}\n";
 542                exit 1;
 543        }
 544        read_uuid();
 545        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 546        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
 547
 548        my $repo;
 549        set_svn_commit_env();
 550        foreach my $c (@revs) {
 551                my $log_msg = get_commit_message($c, $commit_msg);
 552
 553                # fork for each commit because there's a memory leak I
 554                # can't track down... (it's probably in the SVN code)
 555                defined(my $pid = open my $fh, '-|') or croak $!;
 556                if (!$pid) {
 557                        my $ed = SVN::Git::Editor->new(
 558                                        {       r => $r_last,
 559                                                ra => libsvn_dup_ra($SVN),
 560                                                c => $c,
 561                                                svn_path => $SVN->{svn_path},
 562                                        },
 563                                        $SVN->get_commit_editor(
 564                                                $log_msg->{msg},
 565                                                sub {
 566                                                        libsvn_commit_cb(
 567                                                                @_, $c,
 568                                                                $log_msg->{msg},
 569                                                                $r_last,
 570                                                                $cmt_last)
 571                                                },
 572                                                @lock)
 573                                        );
 574                        my $mods = libsvn_checkout_tree($cmt_last, $c, $ed);
 575                        if (@$mods == 0) {
 576                                print "No changes\nr$r_last = $cmt_last\n";
 577                                $ed->abort_edit;
 578                        } else {
 579                                $ed->close_edit;
 580                        }
 581                        exit 0;
 582                }
 583                my ($r_new, $cmt_new, $no);
 584                while (<$fh>) {
 585                        print $_;
 586                        chomp;
 587                        if (/^r(\d+) = ($sha1)$/o) {
 588                                ($r_new, $cmt_new) = ($1, $2);
 589                        } elsif ($_ eq 'No changes') {
 590                                $no = 1;
 591                        }
 592                }
 593                close $fh or exit 1;
 594                if (! defined $r_new && ! defined $cmt_new) {
 595                        unless ($no) {
 596                                die "Failed to parse revision information\n";
 597                        }
 598                } else {
 599                        ($r_last, $cmt_last) = ($r_new, $cmt_new);
 600                }
 601        }
 602        $ENV{LC_ALL} = 'C';
 603        unlink $commit_msg;
 604}
 605
 606sub dcommit {
 607        my $head = shift || 'HEAD';
 608        my $gs = "refs/remotes/$GIT_SVN";
 609        chomp(my @refs = safe_qx(qw/git-rev-list --no-merges/, "$gs..$head"));
 610        my $last_rev;
 611        foreach my $d (reverse @refs) {
 612                if (quiet_run('git-rev-parse','--verify',"$d~1") != 0) {
 613                        die "Commit $d\n",
 614                            "has no parent commit, and therefore ",
 615                            "nothing to diff against.\n",
 616                            "You should be working from a repository ",
 617                            "originally created by git-svn\n";
 618                }
 619                unless (defined $last_rev) {
 620                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 621                        unless (defined $last_rev) {
 622                                die "Unable to extract revision information ",
 623                                    "from commit $d~1\n";
 624                        }
 625                }
 626                if ($_dry_run) {
 627                        print "diff-tree $d~1 $d\n";
 628                } else {
 629                        if (my $r = commit_diff("$d~1", $d, undef, $last_rev)) {
 630                                $last_rev = $r;
 631                        } # else: no changes, same $last_rev
 632                }
 633        }
 634        return if $_dry_run;
 635        fetch();
 636        my @diff = safe_qx('git-diff-tree', $head, $gs);
 637        my @finish;
 638        if (@diff) {
 639                @finish = qw/rebase/;
 640                push @finish, qw/--merge/ if $_merge;
 641                push @finish, "--strategy=$_strategy" if $_strategy;
 642                print STDERR "W: $head and $gs differ, using @finish:\n", @diff;
 643        } else {
 644                print "No changes between current $head and $gs\n",
 645                      "Resetting to the latest $gs\n";
 646                @finish = qw/reset --mixed/;
 647        }
 648        sys('git', @finish, $gs);
 649}
 650
 651sub show_ignore {
 652        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 653        $_use_lib ? show_ignore_lib() : show_ignore_cmd();
 654}
 655
 656sub show_ignore_cmd {
 657        require File::Find or die $!;
 658        if (defined $_revision) {
 659                die "-r/--revision option doesn't work unless the Perl SVN ",
 660                        "libraries are used\n";
 661        }
 662        chdir $SVN_WC or croak $!;
 663        my %ign;
 664        File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 665                s#^\./##;
 666                @{$ign{$_}} = svn_propget_base('svn:ignore', $_);
 667                }}, no_chdir=>1},'.');
 668
 669        print "\n# /\n";
 670        foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 671        delete $ign{'.'};
 672        foreach my $i (sort keys %ign) {
 673                print "\n# ",$i,"\n";
 674                foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 675        }
 676}
 677
 678sub show_ignore_lib {
 679        my $repo;
 680        $SVN ||= libsvn_connect($SVN_URL);
 681        my $r = defined $_revision ? $_revision : $SVN->get_latest_revnum;
 682        libsvn_traverse_ignore(\*STDOUT, $SVN->{svn_path}, $r);
 683}
 684
 685sub graft_branches {
 686        my $gr_file = "$GIT_DIR/info/grafts";
 687        my ($grafts, $comments) = read_grafts($gr_file);
 688        my $gr_sha1;
 689
 690        if (%$grafts) {
 691                # temporarily disable our grafts file to make this idempotent
 692                chomp($gr_sha1 = safe_qx(qw/git-hash-object -w/,$gr_file));
 693                rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
 694        }
 695
 696        my $l_map = read_url_paths();
 697        my @re = map { qr/$_/is } @_opt_m if @_opt_m;
 698        unless ($_no_default_regex) {
 699                push @re, (qr/\b(?:merge|merging|merged)\s+with\s+([\w\.\-]+)/i,
 700                        qr/\b(?:merge|merging|merged)\s+([\w\.\-]+)/i,
 701                        qr/\b(?:from|of)\s+([\w\.\-]+)/i );
 702        }
 703        foreach my $u (keys %$l_map) {
 704                if (@re) {
 705                        foreach my $p (keys %{$l_map->{$u}}) {
 706                                graft_merge_msg($grafts,$l_map,$u,$p,@re);
 707                        }
 708                }
 709                unless ($_no_graft_copy) {
 710                        if ($_use_lib) {
 711                                graft_file_copy_lib($grafts,$l_map,$u);
 712                        } else {
 713                                graft_file_copy_cmd($grafts,$l_map,$u);
 714                        }
 715                }
 716        }
 717        graft_tree_joins($grafts);
 718
 719        write_grafts($grafts, $comments, $gr_file);
 720        unlink "$gr_file~$gr_sha1" if $gr_sha1;
 721}
 722
 723sub multi_init {
 724        my $url = shift;
 725        $_trunk ||= 'trunk';
 726        $_trunk =~ s#/+$##;
 727        $url =~ s#/+$## if $url;
 728        if ($_trunk !~ m#^[a-z\+]+://#) {
 729                $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#);
 730                unless ($url) {
 731                        print STDERR "E: '$_trunk' is not a complete URL ",
 732                                "and a separate URL is not specified\n";
 733                        exit 1;
 734                }
 735                $_trunk = $url . $_trunk;
 736        }
 737        my $ch_id;
 738        if ($GIT_SVN eq 'git-svn') {
 739                $ch_id = 1;
 740                $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
 741        }
 742        init_vars();
 743        unless (-d $GIT_SVN_DIR) {
 744                print "GIT_SVN_ID set to 'trunk' for $_trunk\n" if $ch_id;
 745                init($_trunk);
 746                sys('git-repo-config', 'svn.trunk', $_trunk);
 747        }
 748        complete_url_ls_init($url, $_branches, '--branches/-b', '');
 749        complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/');
 750}
 751
 752sub multi_fetch {
 753        # try to do trunk first, since branches/tags
 754        # may be descended from it.
 755        if (-e "$GIT_DIR/svn/trunk/info/url") {
 756                fetch_child_id('trunk', @_);
 757        }
 758        rec_fetch('', "$GIT_DIR/svn", @_);
 759}
 760
 761sub show_log {
 762        my (@args) = @_;
 763        my ($r_min, $r_max);
 764        my $r_last = -1; # prevent dupes
 765        rload_authors() if $_authors;
 766        if (defined $TZ) {
 767                $ENV{TZ} = $TZ;
 768        } else {
 769                delete $ENV{TZ};
 770        }
 771        if (defined $_revision) {
 772                if ($_revision =~ /^(\d+):(\d+)$/) {
 773                        ($r_min, $r_max) = ($1, $2);
 774                } elsif ($_revision =~ /^\d+$/) {
 775                        $r_min = $r_max = $_revision;
 776                } else {
 777                        print STDERR "-r$_revision is not supported, use ",
 778                                "standard \'git log\' arguments instead\n";
 779                        exit 1;
 780                }
 781        }
 782
 783        config_pager();
 784        my $pid = open(my $log,'-|');
 785        defined $pid or croak $!;
 786        if (!$pid) {
 787                exec(git_svn_log_cmd($r_min,$r_max), @args) or croak $!;
 788        }
 789        run_pager();
 790        my (@k, $c, $d);
 791
 792        while (<$log>) {
 793                if (/^${_esc_color}commit ($sha1_short)/o) {
 794                        my $cmt = $1;
 795                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
 796                                $r_last = $c->{r};
 797                                process_commit($c, $r_min, $r_max, \@k) or
 798                                                                goto out;
 799                        }
 800                        $d = undef;
 801                        $c = { c => $cmt };
 802                } elsif (/^${_esc_color}author (.+) (\d+) ([\-\+]?\d+)$/) {
 803                        get_author_info($c, $1, $2, $3);
 804                } elsif (/^${_esc_color}(?:tree|parent|committer) /) {
 805                        # ignore
 806                } elsif (/^${_esc_color}:\d{6} \d{6} $sha1_short/o) {
 807                        push @{$c->{raw}}, $_;
 808                } elsif (/^${_esc_color}[ACRMDT]\t/) {
 809                        # we could add $SVN->{svn_path} here, but that requires
 810                        # remote access at the moment (repo_path_split)...
 811                        s#^(${_esc_color})([ACRMDT])\t#$1   $2 #;
 812                        push @{$c->{changed}}, $_;
 813                } elsif (/^${_esc_color}diff /) {
 814                        $d = 1;
 815                        push @{$c->{diff}}, $_;
 816                } elsif ($d) {
 817                        push @{$c->{diff}}, $_;
 818                } elsif (/^${_esc_color}    (git-svn-id:.+)$/) {
 819                        ($c->{url}, $c->{r}, undef) = extract_metadata($1);
 820                } elsif (s/^${_esc_color}    //) {
 821                        push @{$c->{l}}, $_;
 822                }
 823        }
 824        if ($c && defined $c->{r} && $c->{r} != $r_last) {
 825                $r_last = $c->{r};
 826                process_commit($c, $r_min, $r_max, \@k);
 827        }
 828        if (@k) {
 829                my $swap = $r_max;
 830                $r_max = $r_min;
 831                $r_min = $swap;
 832                process_commit($_, $r_min, $r_max) foreach reverse @k;
 833        }
 834out:
 835        close $log;
 836        print '-' x72,"\n" unless $_incremental || $_oneline;
 837}
 838
 839sub commit_diff_usage {
 840        print STDERR "Usage: $0 commit-diff <tree-ish> <tree-ish> [<URL>]\n";
 841        exit 1
 842}
 843
 844sub commit_diff {
 845        if (!$_use_lib) {
 846                print STDERR "commit-diff must be used with SVN libraries\n";
 847                exit 1;
 848        }
 849        my $ta = shift or commit_diff_usage();
 850        my $tb = shift or commit_diff_usage();
 851        if (!eval { $SVN_URL = shift || file_to_s("$GIT_SVN_DIR/info/url") }) {
 852                print STDERR "Needed URL or usable git-svn id command-line\n";
 853                commit_diff_usage();
 854        }
 855        my $r = shift;
 856        unless (defined $r) {
 857                if (defined $_revision) {
 858                        $r = $_revision
 859                } else {
 860                        die "-r|--revision is a required argument\n";
 861                }
 862        }
 863        if (defined $_message && defined $_file) {
 864                print STDERR "Both --message/-m and --file/-F specified ",
 865                                "for the commit message.\n",
 866                                "I have no idea what you mean\n";
 867                exit 1;
 868        }
 869        if (defined $_file) {
 870                $_message = file_to_s($_file);
 871        } else {
 872                $_message ||= get_commit_message($tb,
 873                                        "$GIT_DIR/.svn-commit.tmp.$$")->{msg};
 874        }
 875        $SVN ||= libsvn_connect($SVN_URL);
 876        if ($r eq 'HEAD') {
 877                $r = $SVN->get_latest_revnum;
 878        } elsif ($r !~ /^\d+$/) {
 879                die "revision argument: $r not understood by git-svn\n";
 880        }
 881        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 882        my $rev_committed;
 883        my $ed = SVN::Git::Editor->new({        r => $r,
 884                                                ra => libsvn_dup_ra($SVN),
 885                                                c => $tb,
 886                                                svn_path => $SVN->{svn_path}
 887                                        },
 888                                $SVN->get_commit_editor($_message,
 889                                        sub {
 890                                                $rev_committed = $_[0];
 891                                                print "Committed $_[0]\n";
 892                                        }, @lock)
 893                                );
 894        eval {
 895                my $mods = libsvn_checkout_tree($ta, $tb, $ed);
 896                if (@$mods == 0) {
 897                        print "No changes\n$ta == $tb\n";
 898                        $ed->abort_edit;
 899                } else {
 900                        $ed->close_edit;
 901                }
 902        };
 903        fatal "$@\n" if $@;
 904        $_message = $_file = undef;
 905        return $rev_committed;
 906}
 907
 908########################### utility functions #########################
 909
 910sub cmt_showable {
 911        my ($c) = @_;
 912        return 1 if defined $c->{r};
 913        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
 914                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
 915                my @msg = safe_qx(qw/git-cat-file commit/, $c->{c});
 916                shift @msg while ($msg[0] ne "\n");
 917                shift @msg;
 918                @{$c->{l}} = grep !/^git-svn-id: /, @msg;
 919
 920                (undef, $c->{r}, undef) = extract_metadata(
 921                                (grep(/^git-svn-id: /, @msg))[-1]);
 922        }
 923        return defined $c->{r};
 924}
 925
 926sub log_use_color {
 927        return 1 if $_color;
 928        my ($dc, $dcvar);
 929        $dcvar = 'color.diff';
 930        $dc = `git-repo-config --get $dcvar`;
 931        if ($dc eq '') {
 932                # nothing at all; fallback to "diff.color"
 933                $dcvar = 'diff.color';
 934                $dc = `git-repo-config --get $dcvar`;
 935        }
 936        chomp($dc);
 937        if ($dc eq 'auto') {
 938                my $pc;
 939                $pc = `git-repo-config --get color.pager`;
 940                if ($pc eq '') {
 941                        # does not have it -- fallback to pager.color
 942                        $pc = `git-repo-config --bool --get pager.color`;
 943                }
 944                else {
 945                        $pc = `git-repo-config --bool --get color.pager`;
 946                        if ($?) {
 947                                $pc = 'false';
 948                        }
 949                }
 950                chomp($pc);
 951                if (-t *STDOUT || (defined $_pager && $pc eq 'true')) {
 952                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
 953                }
 954                return 0;
 955        }
 956        return 0 if $dc eq 'never';
 957        return 1 if $dc eq 'always';
 958        chomp($dc = `git-repo-config --bool --get $dcvar`);
 959        return ($dc eq 'true');
 960}
 961
 962sub git_svn_log_cmd {
 963        my ($r_min, $r_max) = @_;
 964        my @cmd = (qw/git-log --abbrev-commit --pretty=raw
 965                        --default/, "refs/remotes/$GIT_SVN");
 966        push @cmd, '-r' unless $_non_recursive;
 967        push @cmd, qw/--raw --name-status/ if $_verbose;
 968        push @cmd, '--color' if log_use_color();
 969        return @cmd unless defined $r_max;
 970        if ($r_max == $r_min) {
 971                push @cmd, '--max-count=1';
 972                if (my $c = revdb_get($REVDB, $r_max)) {
 973                        push @cmd, $c;
 974                }
 975        } else {
 976                my ($c_min, $c_max);
 977                $c_max = revdb_get($REVDB, $r_max);
 978                $c_min = revdb_get($REVDB, $r_min);
 979                if (defined $c_min && defined $c_max) {
 980                        if ($r_max > $r_max) {
 981                                push @cmd, "$c_min..$c_max";
 982                        } else {
 983                                push @cmd, "$c_max..$c_min";
 984                        }
 985                } elsif ($r_max > $r_min) {
 986                        push @cmd, $c_max;
 987                } else {
 988                        push @cmd, $c_min;
 989                }
 990        }
 991        return @cmd;
 992}
 993
 994sub fetch_child_id {
 995        my $id = shift;
 996        print "Fetching $id\n";
 997        my $ref = "$GIT_DIR/refs/remotes/$id";
 998        defined(my $pid = open my $fh, '-|') or croak $!;
 999        if (!$pid) {
1000                $_repack = undef;
1001                $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
1002                init_vars();
1003                fetch(@_);
1004                exit 0;
1005        }
1006        while (<$fh>) {
1007                print $_;
1008                check_repack() if (/^r\d+ = $sha1/);
1009        }
1010        close $fh or croak $?;
1011}
1012
1013sub rec_fetch {
1014        my ($pfx, $p, @args) = @_;
1015        my @dir;
1016        foreach (sort <$p/*>) {
1017                if (-r "$_/info/url") {
1018                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
1019                        my $id = $pfx . basename $_;
1020                        next if $id eq 'trunk';
1021                        fetch_child_id($id, @args);
1022                } elsif (-d $_) {
1023                        push @dir, $_;
1024                }
1025        }
1026        foreach (@dir) {
1027                my $x = $_;
1028                $x =~ s!^\Q$GIT_DIR\E/svn/!!;
1029                rec_fetch($x, $_);
1030        }
1031}
1032
1033sub complete_url_ls_init {
1034        my ($url, $var, $switch, $pfx) = @_;
1035        unless ($var) {
1036                print STDERR "W: $switch not specified\n";
1037                return;
1038        }
1039        $var =~ s#/+$##;
1040        if ($var !~ m#^[a-z\+]+://#) {
1041                $var = '/' . $var if ($var !~ m#^/#);
1042                unless ($url) {
1043                        print STDERR "E: '$var' is not a complete URL ",
1044                                "and a separate URL is not specified\n";
1045                        exit 1;
1046                }
1047                $var = $url . $var;
1048        }
1049        chomp(my @ls = $_use_lib ? libsvn_ls_fullurl($var)
1050                                : safe_qx(qw/svn ls --non-interactive/, $var));
1051        my $old = $GIT_SVN;
1052        defined(my $pid = fork) or croak $!;
1053        if (!$pid) {
1054                foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
1055                        $u =~ s#/+$##;
1056                        if ($u !~ m!\Q$var\E/(.+)$!) {
1057                                print STDERR "W: Unrecognized URL: $u\n";
1058                                die "This should never happen\n";
1059                        }
1060                        # don't try to init already existing refs
1061                        my $id = $pfx.$1;
1062                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
1063                        init_vars();
1064                        unless (-d $GIT_SVN_DIR) {
1065                                print "init $u => $id\n";
1066                                init($u);
1067                        }
1068                }
1069                exit 0;
1070        }
1071        waitpid $pid, 0;
1072        croak $? if $?;
1073        my ($n) = ($switch =~ /^--(\w+)/);
1074        sys('git-repo-config', "svn.$n", $var);
1075}
1076
1077sub common_prefix {
1078        my $paths = shift;
1079        my %common;
1080        foreach (@$paths) {
1081                my @tmp = split m#/#, $_;
1082                my $p = '';
1083                while (my $x = shift @tmp) {
1084                        $p .= "/$x";
1085                        $common{$p} ||= 0;
1086                        $common{$p}++;
1087                }
1088        }
1089        foreach (sort {length $b <=> length $a} keys %common) {
1090                if ($common{$_} == @$paths) {
1091                        return $_;
1092                }
1093        }
1094        return '';
1095}
1096
1097# grafts set here are 'stronger' in that they're based on actual tree
1098# matches, and won't be deleted from merge-base checking in write_grafts()
1099sub graft_tree_joins {
1100        my $grafts = shift;
1101        map_tree_joins() if (@_branch_from && !%tree_map);
1102        return unless %tree_map;
1103
1104        git_svn_each(sub {
1105                my $i = shift;
1106                defined(my $pid = open my $fh, '-|') or croak $!;
1107                if (!$pid) {
1108                        exec qw/git-rev-list --pretty=raw/,
1109                                        "refs/remotes/$i" or croak $!;
1110                }
1111                while (<$fh>) {
1112                        next unless /^commit ($sha1)$/o;
1113                        my $c = $1;
1114                        my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
1115                        next unless $tree_map{$t};
1116
1117                        my $l;
1118                        do {
1119                                $l = readline $fh;
1120                        } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
1121
1122                        my ($s, $tz) = ($1, $2);
1123                        if ($tz =~ s/^\+//) {
1124                                $s += tz_to_s_offset($tz);
1125                        } elsif ($tz =~ s/^\-//) {
1126                                $s -= tz_to_s_offset($tz);
1127                        }
1128
1129                        my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
1130
1131                        foreach my $p (@{$tree_map{$t}}) {
1132                                next if $p eq $c;
1133                                my $mb = eval {
1134                                        safe_qx('git-merge-base', $c, $p)
1135                                };
1136                                next unless ($@ || $?);
1137                                if (defined $r_a) {
1138                                        # see if SVN says it's a relative
1139                                        my ($url_b, $r_b, $uuid_b) =
1140                                                        cmt_metadata($p);
1141                                        next if (defined $url_b &&
1142                                                        defined $url_a &&
1143                                                        ($url_a eq $url_b) &&
1144                                                        ($uuid_a eq $uuid_b));
1145                                        if ($uuid_a eq $uuid_b) {
1146                                                if ($r_b < $r_a) {
1147                                                        $grafts->{$c}->{$p} = 2;
1148                                                        next;
1149                                                } elsif ($r_b > $r_a) {
1150                                                        $grafts->{$p}->{$c} = 2;
1151                                                        next;
1152                                                }
1153                                        }
1154                                }
1155                                my $ct = get_commit_time($p);
1156                                if ($ct < $s) {
1157                                        $grafts->{$c}->{$p} = 2;
1158                                } elsif ($ct > $s) {
1159                                        $grafts->{$p}->{$c} = 2;
1160                                }
1161                                # what should we do when $ct == $s ?
1162                        }
1163                }
1164                close $fh or croak $?;
1165        });
1166}
1167
1168# this isn't funky-filename safe, but good enough for now...
1169sub graft_file_copy_cmd {
1170        my ($grafts, $l_map, $u) = @_;
1171        my $paths = $l_map->{$u};
1172        my $pfx = common_prefix([keys %$paths]);
1173        $SVN_URL ||= $u.$pfx;
1174        my $pid = open my $fh, '-|';
1175        defined $pid or croak $!;
1176        unless ($pid) {
1177                my @exec = qw/svn log -v/;
1178                push @exec, "-r$_revision" if defined $_revision;
1179                exec @exec, $u.$pfx or croak $!;
1180        }
1181        my ($r, $mp) = (undef, undef);
1182        while (<$fh>) {
1183                chomp;
1184                if (/^\-{72}$/) {
1185                        $mp = $r = undef;
1186                } elsif (/^r(\d+) \| /) {
1187                        $r = $1 unless defined $r;
1188                } elsif (/^Changed paths:/) {
1189                        $mp = 1;
1190                } elsif ($mp && m#^   [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) {
1191                        my ($p1, $p0, $r0) = ($1, $2, $3);
1192                        my $c = find_graft_path_commit($paths, $p1, $r);
1193                        next unless $c;
1194                        find_graft_path_parents($grafts, $paths, $c, $p0, $r0);
1195                }
1196        }
1197}
1198
1199sub graft_file_copy_lib {
1200        my ($grafts, $l_map, $u) = @_;
1201        my $tree_paths = $l_map->{$u};
1202        my $pfx = common_prefix([keys %$tree_paths]);
1203        my ($repo, $path) = repo_path_split($u.$pfx);
1204        $SVN = libsvn_connect($repo);
1205
1206        my ($base, $head) = libsvn_parse_revision();
1207        my $inc = 1000;
1208        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
1209        my $eh = $SVN::Error::handler;
1210        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
1211        while (1) {
1212                my $pool = SVN::Pool->new;
1213                libsvn_get_log(libsvn_dup_ra($SVN), [$path],
1214                               $min, $max, 0, 2, 1,
1215                        sub {
1216                                libsvn_graft_file_copies($grafts, $tree_paths,
1217                                                        $path, @_);
1218                        }, $pool);
1219                $pool->clear;
1220                last if ($max >= $head);
1221                $min = $max + 1;
1222                $max += $inc;
1223                $max = $head if ($max > $head);
1224        }
1225        $SVN::Error::handler = $eh;
1226}
1227
1228sub process_merge_msg_matches {
1229        my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
1230        my (@strong, @weak);
1231        foreach (@matches) {
1232                # merging with ourselves is not interesting
1233                next if $_ eq $p;
1234                if ($l_map->{$u}->{$_}) {
1235                        push @strong, $_;
1236                } else {
1237                        push @weak, $_;
1238                }
1239        }
1240        foreach my $w (@weak) {
1241                last if @strong;
1242                # no exact match, use branch name as regexp.
1243                my $re = qr/\Q$w\E/i;
1244                foreach (keys %{$l_map->{$u}}) {
1245                        if (/$re/) {
1246                                push @strong, $l_map->{$u}->{$_};
1247                                last;
1248                        }
1249                }
1250                last if @strong;
1251                $w = basename($w);
1252                $re = qr/\Q$w\E/i;
1253                foreach (keys %{$l_map->{$u}}) {
1254                        if (/$re/) {
1255                                push @strong, $l_map->{$u}->{$_};
1256                                last;
1257                        }
1258                }
1259        }
1260        my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
1261                                        \s(?:[a-f\d\-]+)$/xsm);
1262        unless (defined $rev) {
1263                ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
1264                                        \@(?:[a-f\d\-]+)/xsm);
1265                return unless defined $rev;
1266        }
1267        foreach my $m (@strong) {
1268                my ($r0, $s0) = find_rev_before($rev, $m, 1);
1269                $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
1270        }
1271}
1272
1273sub graft_merge_msg {
1274        my ($grafts, $l_map, $u, $p, @re) = @_;
1275
1276        my $x = $l_map->{$u}->{$p};
1277        my $rl = rev_list_raw($x);
1278        while (my $c = next_rev_list_entry($rl)) {
1279                foreach my $re (@re) {
1280                        my (@br) = ($c->{m} =~ /$re/g);
1281                        next unless @br;
1282                        process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
1283                }
1284        }
1285}
1286
1287sub read_uuid {
1288        return if $SVN_UUID;
1289        if ($_use_lib) {
1290                my $pool = SVN::Pool->new;
1291                $SVN_UUID = $SVN->get_uuid($pool);
1292                $pool->clear;
1293        } else {
1294                my $info = shift || svn_info('.');
1295                $SVN_UUID = $info->{'Repository UUID'} or
1296                                        croak "Repository UUID unreadable\n";
1297        }
1298}
1299
1300sub quiet_run {
1301        my $pid = fork;
1302        defined $pid or croak $!;
1303        if (!$pid) {
1304                open my $null, '>', '/dev/null' or croak $!;
1305                open STDERR, '>&', $null or croak $!;
1306                open STDOUT, '>&', $null or croak $!;
1307                exec @_ or croak $!;
1308        }
1309        waitpid $pid, 0;
1310        return $?;
1311}
1312
1313sub repo_path_split {
1314        my $full_url = shift;
1315        $full_url =~ s#/+$##;
1316
1317        foreach (@repo_path_split_cache) {
1318                if ($full_url =~ s#$_##) {
1319                        my $u = $1;
1320                        $full_url =~ s#^/+##;
1321                        return ($u, $full_url);
1322                }
1323        }
1324        if ($_use_lib) {
1325                my $tmp = libsvn_connect($full_url);
1326                return ($tmp->{repos_root}, $tmp->{svn_path});
1327        } else {
1328                my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i);
1329                $path =~ s#^/+##;
1330                my @paths = split(m#/+#, $path);
1331                while (quiet_run(qw/svn ls --non-interactive/, $url)) {
1332                        my $n = shift @paths || last;
1333                        $url .= "/$n";
1334                }
1335                push @repo_path_split_cache, qr/^(\Q$url\E)/;
1336                $path = join('/',@paths);
1337                return ($url, $path);
1338        }
1339}
1340
1341sub setup_git_svn {
1342        defined $SVN_URL or croak "SVN repository location required\n";
1343        unless (-d $GIT_DIR) {
1344                croak "GIT_DIR=$GIT_DIR does not exist!\n";
1345        }
1346        mkpath([$GIT_SVN_DIR]);
1347        mkpath(["$GIT_SVN_DIR/info"]);
1348        open my $fh, '>>',$REVDB or croak $!;
1349        close $fh;
1350        s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1351
1352}
1353
1354sub assert_svn_wc_clean {
1355        return if $_use_lib;
1356        my ($svn_rev) = @_;
1357        croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
1358        my $lcr = svn_info('.')->{'Last Changed Rev'};
1359        if ($svn_rev != $lcr) {
1360                print STDERR "Checking for copy-tree ... ";
1361                my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
1362                                                "-r$lcr:$svn_rev")));
1363                if (@diff) {
1364                        croak "Nope!  Expected r$svn_rev, got r$lcr\n";
1365                } else {
1366                        print STDERR "OK!\n";
1367                }
1368        }
1369        my @status = grep(!/^Performing status on external/,(`svn status`));
1370        @status = grep(!/^\s*$/,@status);
1371        @status = grep(!/^X/,@status) if $_no_ignore_ext;
1372        if (scalar @status) {
1373                print STDERR "Tree ($SVN_WC) is not clean:\n";
1374                print STDERR $_ foreach @status;
1375                croak;
1376        }
1377}
1378
1379sub get_tree_from_treeish {
1380        my ($treeish) = @_;
1381        croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1382        chomp(my $type = `git-cat-file -t $treeish`);
1383        my $expected;
1384        while ($type eq 'tag') {
1385                chomp(($treeish, $type) = `git-cat-file tag $treeish`);
1386        }
1387        if ($type eq 'commit') {
1388                $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
1389                ($expected) = ($expected =~ /^tree ($sha1)$/);
1390                die "Unable to get tree from $treeish\n" unless $expected;
1391        } elsif ($type eq 'tree') {
1392                $expected = $treeish;
1393        } else {
1394                die "$treeish is a $type, expected tree, tag or commit\n";
1395        }
1396        return $expected;
1397}
1398
1399sub assert_tree {
1400        return if $_use_lib;
1401        my ($treeish) = @_;
1402        my $expected = get_tree_from_treeish($treeish);
1403
1404        my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
1405        if (-e $tmpindex) {
1406                unlink $tmpindex or croak $!;
1407        }
1408        my $old_index = set_index($tmpindex);
1409        index_changes(1);
1410        chomp(my $tree = `git-write-tree`);
1411        restore_index($old_index);
1412        if ($tree ne $expected) {
1413                croak "Tree mismatch, Got: $tree, Expected: $expected\n";
1414        }
1415        unlink $tmpindex;
1416}
1417
1418sub parse_diff_tree {
1419        my $diff_fh = shift;
1420        local $/ = "\0";
1421        my $state = 'meta';
1422        my @mods;
1423        while (<$diff_fh>) {
1424                chomp $_; # this gets rid of the trailing "\0"
1425                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1426                                        $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1427                        push @mods, {   mode_a => $1, mode_b => $2,
1428                                        sha1_b => $3, chg => $4 };
1429                        if ($4 =~ /^(?:C|R)$/) {
1430                                $state = 'file_a';
1431                        } else {
1432                                $state = 'file_b';
1433                        }
1434                } elsif ($state eq 'file_a') {
1435                        my $x = $mods[$#mods] or croak "Empty array\n";
1436                        if ($x->{chg} !~ /^(?:C|R)$/) {
1437                                croak "Error parsing $_, $x->{chg}\n";
1438                        }
1439                        $x->{file_a} = $_;
1440                        $state = 'file_b';
1441                } elsif ($state eq 'file_b') {
1442                        my $x = $mods[$#mods] or croak "Empty array\n";
1443                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1444                                croak "Error parsing $_, $x->{chg}\n";
1445                        }
1446                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1447                                croak "Error parsing $_, $x->{chg}\n";
1448                        }
1449                        $x->{file_b} = $_;
1450                        $state = 'meta';
1451                } else {
1452                        croak "Error parsing $_\n";
1453                }
1454        }
1455        close $diff_fh or croak $?;
1456
1457        return \@mods;
1458}
1459
1460sub svn_check_prop_executable {
1461        my $m = shift;
1462        return if -l $m->{file_b};
1463        if ($m->{mode_b} =~ /755$/) {
1464                chmod((0755 &~ umask),$m->{file_b}) or croak $!;
1465                if ($m->{mode_a} !~ /755$/) {
1466                        sys(qw(svn propset svn:executable 1), $m->{file_b});
1467                }
1468                -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
1469        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
1470                sys(qw(svn propdel svn:executable), $m->{file_b});
1471                chmod((0644 &~ umask),$m->{file_b}) or croak $!;
1472                -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
1473        }
1474}
1475
1476sub svn_ensure_parent_path {
1477        my $dir_b = dirname(shift);
1478        svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
1479        mkpath([$dir_b]) unless (-d $dir_b);
1480        sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
1481}
1482
1483sub precommit_check {
1484        my $mods = shift;
1485        my (%rm_file, %rmdir_check, %added_check);
1486
1487        my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
1488        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1489                if ($m->{chg} eq 'R') {
1490                        if (-d $m->{file_b}) {
1491                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1492                        }
1493                        # dir/$file => dir/file/$file
1494                        my $dirname = dirname($m->{file_b});
1495                        while ($dirname ne File::Spec->curdir) {
1496                                if ($dirname ne $m->{file_a}) {
1497                                        $dirname = dirname($dirname);
1498                                        next;
1499                                }
1500                                err_file_to_dir("$m->{file_a} => $m->{file_b}");
1501                        }
1502                        # baz/zzz => baz (baz is a file)
1503                        $dirname = dirname($m->{file_a});
1504                        while ($dirname ne File::Spec->curdir) {
1505                                if ($dirname ne $m->{file_b}) {
1506                                        $dirname = dirname($dirname);
1507                                        next;
1508                                }
1509                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1510                        }
1511                }
1512                if ($m->{chg} =~ /^(D|R)$/) {
1513                        my $t = $1 eq 'D' ? 'file_b' : 'file_a';
1514                        $rm_file{ $m->{$t} } = 1;
1515                        my $dirname = dirname( $m->{$t} );
1516                        my $basename = basename( $m->{$t} );
1517                        $rmdir_check{$dirname}->{$basename} = 1;
1518                } elsif ($m->{chg} =~ /^(?:A|C)$/) {
1519                        if (-d $m->{file_b}) {
1520                                err_dir_to_file($m->{file_b});
1521                        }
1522                        my $dirname = dirname( $m->{file_b} );
1523                        my $basename = basename( $m->{file_b} );
1524                        $added_check{$dirname}->{$basename} = 1;
1525                        while ($dirname ne File::Spec->curdir) {
1526                                if ($rm_file{$dirname}) {
1527                                        err_file_to_dir($m->{file_b});
1528                                }
1529                                $dirname = dirname $dirname;
1530                        }
1531                }
1532        }
1533        return (\%rmdir_check, \%added_check);
1534
1535        sub err_dir_to_file {
1536                my $file = shift;
1537                print STDERR "Node change from directory to file ",
1538                                "is not supported by Subversion: ",$file,"\n";
1539                exit 1;
1540        }
1541        sub err_file_to_dir {
1542                my $file = shift;
1543                print STDERR "Node change from file to directory ",
1544                                "is not supported by Subversion: ",$file,"\n";
1545                exit 1;
1546        }
1547}
1548
1549
1550sub get_diff {
1551        my ($from, $treeish) = @_;
1552        assert_tree($from);
1553        print "diff-tree $from $treeish\n";
1554        my $pid = open my $diff_fh, '-|';
1555        defined $pid or croak $!;
1556        if ($pid == 0) {
1557                my @diff_tree = qw(git-diff-tree -z -r);
1558                if ($_cp_similarity) {
1559                        push @diff_tree, "-C$_cp_similarity";
1560                } else {
1561                        push @diff_tree, '-C';
1562                }
1563                push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1564                push @diff_tree, "-l$_l" if defined $_l;
1565                exec(@diff_tree, $from, $treeish) or croak $!;
1566        }
1567        return parse_diff_tree($diff_fh);
1568}
1569
1570sub svn_checkout_tree {
1571        my ($from, $treeish) = @_;
1572        my $mods = get_diff($from->{commit}, $treeish);
1573        return $mods unless (scalar @$mods);
1574        my ($rm, $add) = precommit_check($mods);
1575
1576        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1577        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1578                if ($m->{chg} eq 'C') {
1579                        svn_ensure_parent_path( $m->{file_b} );
1580                        sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
1581                        apply_mod_line_blob($m);
1582                        svn_check_prop_executable($m);
1583                } elsif ($m->{chg} eq 'D') {
1584                        sys(qw(svn rm --force), $m->{file_b});
1585                } elsif ($m->{chg} eq 'R') {
1586                        svn_ensure_parent_path( $m->{file_b} );
1587                        sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
1588                        apply_mod_line_blob($m);
1589                        svn_check_prop_executable($m);
1590                } elsif ($m->{chg} eq 'M') {
1591                        apply_mod_line_blob($m);
1592                        svn_check_prop_executable($m);
1593                } elsif ($m->{chg} eq 'T') {
1594                        svn_check_prop_executable($m);
1595                        apply_mod_line_blob($m);
1596                        if ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
1597                                sys(qw(svn propdel svn:special), $m->{file_b});
1598                        } else {
1599                                sys(qw(svn propset svn:special *),$m->{file_b});
1600                        }
1601                } elsif ($m->{chg} eq 'A') {
1602                        svn_ensure_parent_path( $m->{file_b} );
1603                        apply_mod_line_blob($m);
1604                        sys(qw(svn add), $m->{file_b});
1605                        svn_check_prop_executable($m);
1606                } else {
1607                        croak "Invalid chg: $m->{chg}\n";
1608                }
1609        }
1610
1611        assert_tree($treeish);
1612        if ($_rmdir) { # remove empty directories
1613                handle_rmdir($rm, $add);
1614        }
1615        assert_tree($treeish);
1616        return $mods;
1617}
1618
1619sub libsvn_checkout_tree {
1620        my ($from, $treeish, $ed) = @_;
1621        my $mods = get_diff($from, $treeish);
1622        return $mods unless (scalar @$mods);
1623        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1624        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1625                my $f = $m->{chg};
1626                if (defined $o{$f}) {
1627                        $ed->$f($m, $_q);
1628                } else {
1629                        croak "Invalid change type: $f\n";
1630                }
1631        }
1632        $ed->rmdirs($_q) if $_rmdir;
1633        return $mods;
1634}
1635
1636# svn ls doesn't work with respect to the current working tree, but what's
1637# in the repository.  There's not even an option for it... *sigh*
1638# (added files don't show up and removed files remain in the ls listing)
1639sub svn_ls_current {
1640        my ($dir, $rm, $add) = @_;
1641        chomp(my @ls = safe_qx('svn','ls',$dir));
1642        my @ret = ();
1643        foreach (@ls) {
1644                s#/$##; # trailing slashes are evil
1645                push @ret, $_ unless $rm->{$dir}->{$_};
1646        }
1647        if (exists $add->{$dir}) {
1648                push @ret, keys %{$add->{$dir}};
1649        }
1650        return \@ret;
1651}
1652
1653sub handle_rmdir {
1654        my ($rm, $add) = @_;
1655
1656        foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
1657                my $ls = svn_ls_current($dir, $rm, $add);
1658                next if (scalar @$ls);
1659                sys(qw(svn rm --force),$dir);
1660
1661                my $dn = dirname $dir;
1662                $rm->{ $dn }->{ basename $dir } = 1;
1663                $ls = svn_ls_current($dn, $rm, $add);
1664                while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
1665                        sys(qw(svn rm --force),$dn);
1666                        $dir = basename $dn;
1667                        $dn = dirname $dn;
1668                        $rm->{ $dn }->{ $dir } = 1;
1669                        $ls = svn_ls_current($dn, $rm, $add);
1670                }
1671        }
1672}
1673
1674sub get_commit_message {
1675        my ($commit, $commit_msg) = (@_);
1676        my %log_msg = ( msg => '' );
1677        open my $msg, '>', $commit_msg or croak $!;
1678
1679        chomp(my $type = `git-cat-file -t $commit`);
1680        if ($type eq 'commit' || $type eq 'tag') {
1681                my $pid = open my $msg_fh, '-|';
1682                defined $pid or croak $!;
1683
1684                if ($pid == 0) {
1685                        exec('git-cat-file', $type, $commit) or croak $!;
1686                }
1687                my $in_msg = 0;
1688                while (<$msg_fh>) {
1689                        if (!$in_msg) {
1690                                $in_msg = 1 if (/^\s*$/);
1691                        } elsif (/^git-svn-id: /) {
1692                                # skip this, we regenerate the correct one
1693                                # on re-fetch anyways
1694                        } else {
1695                                print $msg $_ or croak $!;
1696                        }
1697                }
1698                close $msg_fh or croak $?;
1699        }
1700        close $msg or croak $!;
1701
1702        if ($_edit || ($type eq 'tree')) {
1703                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1704                system($editor, $commit_msg);
1705        }
1706
1707        # file_to_s removes all trailing newlines, so just use chomp() here:
1708        open $msg, '<', $commit_msg or croak $!;
1709        { local $/; chomp($log_msg{msg} = <$msg>); }
1710        close $msg or croak $!;
1711
1712        return \%log_msg;
1713}
1714
1715sub set_svn_commit_env {
1716        if (defined $LC_ALL) {
1717                $ENV{LC_ALL} = $LC_ALL;
1718        } else {
1719                delete $ENV{LC_ALL};
1720        }
1721}
1722
1723sub svn_commit_tree {
1724        my ($last, $commit) = @_;
1725        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
1726        my $log_msg = get_commit_message($commit, $commit_msg);
1727        my ($oneline) = ($log_msg->{msg} =~ /([^\n\r]+)/);
1728        print "Committing $commit: $oneline\n";
1729
1730        set_svn_commit_env();
1731        my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
1732        $ENV{LC_ALL} = 'C';
1733        unlink $commit_msg;
1734        my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/);
1735        if (!defined $committed) {
1736                my $out = join("\n",@ci_output);
1737                print STDERR "W: Trouble parsing \`svn commit' output:\n\n",
1738                                $out, "\n\nAssuming English locale...";
1739                ($committed) = ($out =~ /^Committed revision \d+\./sm);
1740                defined $committed or die " FAILED!\n",
1741                        "Commit output failed to parse committed revision!\n",
1742                print STDERR " OK\n";
1743        }
1744
1745        my @svn_up = qw(svn up);
1746        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
1747        if ($_optimize_commits && ($committed == ($last->{revision} + 1))) {
1748                push @svn_up, "-r$committed";
1749                sys(@svn_up);
1750                my $info = svn_info('.');
1751                my $date = $info->{'Last Changed Date'} or die "Missing date\n";
1752                if ($info->{'Last Changed Rev'} != $committed) {
1753                        croak "$info->{'Last Changed Rev'} != $committed\n"
1754                }
1755                my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1756                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1757                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1758                                         or croak "Failed to parse date: $date\n";
1759                $log_msg->{date} = "$tz $Y-$m-$d $H:$M:$S";
1760                $log_msg->{author} = $info->{'Last Changed Author'};
1761                $log_msg->{revision} = $committed;
1762                $log_msg->{msg} .= "\n";
1763                $log_msg->{parents} = [ $last->{commit} ];
1764                $log_msg->{commit} = git_commit($log_msg, $commit);
1765                return $log_msg;
1766        }
1767        # resync immediately
1768        push @svn_up, "-r$last->{revision}";
1769        sys(@svn_up);
1770        return fetch("$committed=$commit");
1771}
1772
1773sub rev_list_raw {
1774        my (@args) = @_;
1775        my $pid = open my $fh, '-|';
1776        defined $pid or croak $!;
1777        if (!$pid) {
1778                exec(qw/git-rev-list --pretty=raw/, @args) or croak $!;
1779        }
1780        return { fh => $fh, t => { } };
1781}
1782
1783sub next_rev_list_entry {
1784        my $rl = shift;
1785        my $fh = $rl->{fh};
1786        my $x = $rl->{t};
1787        while (<$fh>) {
1788                if (/^commit ($sha1)$/o) {
1789                        if ($x->{c}) {
1790                                $rl->{t} = { c => $1 };
1791                                return $x;
1792                        } else {
1793                                $x->{c} = $1;
1794                        }
1795                } elsif (/^parent ($sha1)$/o) {
1796                        $x->{p}->{$1} = 1;
1797                } elsif (s/^    //) {
1798                        $x->{m} ||= '';
1799                        $x->{m} .= $_;
1800                }
1801        }
1802        return ($x != $rl->{t}) ? $x : undef;
1803}
1804
1805# read the entire log into a temporary file (which is removed ASAP)
1806# and store the file handle + parser state
1807sub svn_log_raw {
1808        my (@log_args) = @_;
1809        my $log_fh = IO::File->new_tmpfile or croak $!;
1810        my $pid = fork;
1811        defined $pid or croak $!;
1812        if (!$pid) {
1813                open STDOUT, '>&', $log_fh or croak $!;
1814                exec (qw(svn log), @log_args) or croak $!
1815        }
1816        waitpid $pid, 0;
1817        croak $? if $?;
1818        seek $log_fh, 0, 0 or croak $!;
1819        return { state => 'sep', fh => $log_fh };
1820}
1821
1822sub next_log_entry {
1823        my $log = shift; # retval of svn_log_raw()
1824        my $ret = undef;
1825        my $fh = $log->{fh};
1826
1827        while (<$fh>) {
1828                chomp;
1829                if (/^\-{72}$/) {
1830                        if ($log->{state} eq 'msg') {
1831                                if ($ret->{lines}) {
1832                                        $ret->{msg} .= $_."\n";
1833                                        unless(--$ret->{lines}) {
1834                                                $log->{state} = 'sep';
1835                                        }
1836                                } else {
1837                                        croak "Log parse error at: $_\n",
1838                                                $ret->{revision},
1839                                                "\n";
1840                                }
1841                                next;
1842                        }
1843                        if ($log->{state} ne 'sep') {
1844                                croak "Log parse error at: $_\n",
1845                                        "state: $log->{state}\n",
1846                                        $ret->{revision},
1847                                        "\n";
1848                        }
1849                        $log->{state} = 'rev';
1850
1851                        # if we have an empty log message, put something there:
1852                        if ($ret) {
1853                                $ret->{msg} ||= "\n";
1854                                delete $ret->{lines};
1855                                return $ret;
1856                        }
1857                        next;
1858                }
1859                if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
1860                        my $rev = $1;
1861                        my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
1862                        ($lines) = ($lines =~ /(\d+)/);
1863                        $date = '1970-01-01 00:00:00 +0000'
1864                                if ($_ignore_nodate && $date eq '(no date)');
1865                        my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1866                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1867                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1868                                         or croak "Failed to parse date: $date\n";
1869                        $ret = {        revision => $rev,
1870                                        date => "$tz $Y-$m-$d $H:$M:$S",
1871                                        author => $author,
1872                                        lines => $lines,
1873                                        msg => '' };
1874                        if (defined $_authors && ! defined $users{$author}) {
1875                                die "Author: $author not defined in ",
1876                                                "$_authors file\n";
1877                        }
1878                        $log->{state} = 'msg_start';
1879                        next;
1880                }
1881                # skip the first blank line of the message:
1882                if ($log->{state} eq 'msg_start' && /^$/) {
1883                        $log->{state} = 'msg';
1884                } elsif ($log->{state} eq 'msg') {
1885                        if ($ret->{lines}) {
1886                                $ret->{msg} .= $_."\n";
1887                                unless (--$ret->{lines}) {
1888                                        $log->{state} = 'sep';
1889                                }
1890                        } else {
1891                                croak "Log parse error at: $_\n",
1892                                        $ret->{revision},"\n";
1893                        }
1894                }
1895        }
1896        return $ret;
1897}
1898
1899sub svn_info {
1900        my $url = shift || $SVN_URL;
1901
1902        my $pid = open my $info_fh, '-|';
1903        defined $pid or croak $!;
1904
1905        if ($pid == 0) {
1906                exec(qw(svn info),$url) or croak $!;
1907        }
1908
1909        my $ret = {};
1910        # only single-lines seem to exist in svn info output
1911        while (<$info_fh>) {
1912                chomp $_;
1913                if (m#^([^:]+)\s*:\s*(\S.*)$#) {
1914                        $ret->{$1} = $2;
1915                        push @{$ret->{-order}}, $1;
1916                }
1917        }
1918        close $info_fh or croak $?;
1919        return $ret;
1920}
1921
1922sub sys { system(@_) == 0 or croak $? }
1923
1924sub do_update_index {
1925        my ($z_cmd, $cmd, $no_text_base) = @_;
1926
1927        my $z = open my $p, '-|';
1928        defined $z or croak $!;
1929        unless ($z) { exec @$z_cmd or croak $! }
1930
1931        my $pid = open my $ui, '|-';
1932        defined $pid or croak $!;
1933        unless ($pid) {
1934                exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!;
1935        }
1936        local $/ = "\0";
1937        while (my $x = <$p>) {
1938                chomp $x;
1939                if (!$no_text_base && lstat $x && ! -l _ &&
1940                                svn_propget_base('svn:keywords', $x)) {
1941                        my $mode = -x _ ? 0755 : 0644;
1942                        my ($v,$d,$f) = File::Spec->splitpath($x);
1943                        my $tb = File::Spec->catfile($d, '.svn', 'tmp',
1944                                                'text-base',"$f.svn-base");
1945                        $tb =~ s#^/##;
1946                        unless (-f $tb) {
1947                                $tb = File::Spec->catfile($d, '.svn',
1948                                                'text-base',"$f.svn-base");
1949                                $tb =~ s#^/##;
1950                        }
1951                        my @s = stat($x);
1952                        unlink $x or croak $!;
1953                        copy($tb, $x);
1954                        chmod(($mode &~ umask), $x) or croak $!;
1955                        utime $s[8], $s[9], $x;
1956                }
1957                print $ui $x,"\0";
1958        }
1959        close $ui or croak $?;
1960}
1961
1962sub index_changes {
1963        return if $_use_lib;
1964
1965        if (!-f "$GIT_SVN_DIR/info/exclude") {
1966                open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!;
1967                print $fd '.svn',"\n";
1968                close $fd or croak $!;
1969        }
1970        my $no_text_base = shift;
1971        do_update_index([qw/git-diff-files --name-only -z/],
1972                        'remove',
1973                        $no_text_base);
1974        do_update_index([qw/git-ls-files -z --others/,
1975                                "--exclude-from=$GIT_SVN_DIR/info/exclude"],
1976                        'add',
1977                        $no_text_base);
1978}
1979
1980sub s_to_file {
1981        my ($str, $file, $mode) = @_;
1982        open my $fd,'>',$file or croak $!;
1983        print $fd $str,"\n" or croak $!;
1984        close $fd or croak $!;
1985        chmod ($mode &~ umask, $file) if (defined $mode);
1986}
1987
1988sub file_to_s {
1989        my $file = shift;
1990        open my $fd,'<',$file or croak "$!: file: $file\n";
1991        local $/;
1992        my $ret = <$fd>;
1993        close $fd or croak $!;
1994        $ret =~ s/\s*$//s;
1995        return $ret;
1996}
1997
1998sub assert_revision_unknown {
1999        my $r = shift;
2000        if (my $c = revdb_get($REVDB, $r)) {
2001                croak "$r = $c already exists! Why are we refetching it?";
2002        }
2003}
2004
2005sub trees_eq {
2006        my ($x, $y) = @_;
2007        my @x = safe_qx('git-cat-file','commit',$x);
2008        my @y = safe_qx('git-cat-file','commit',$y);
2009        if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
2010                                || $y[0] !~ /^tree $sha1\n$/) {
2011                print STDERR "Trees not equal: $y[0] != $x[0]\n";
2012                return 0
2013        }
2014        return 1;
2015}
2016
2017sub git_commit {
2018        my ($log_msg, @parents) = @_;
2019        assert_revision_unknown($log_msg->{revision});
2020        map_tree_joins() if (@_branch_from && !%tree_map);
2021
2022        my (@tmp_parents, @exec_parents, %seen_parent);
2023        if (my $lparents = $log_msg->{parents}) {
2024                @tmp_parents = @$lparents
2025        }
2026        # commit parents can be conditionally bound to a particular
2027        # svn revision via: "svn_revno=commit_sha1", filter them out here:
2028        foreach my $p (@parents) {
2029                next unless defined $p;
2030                if ($p =~ /^(\d+)=($sha1_short)$/o) {
2031                        if ($1 == $log_msg->{revision}) {
2032                                push @tmp_parents, $2;
2033                        }
2034                } else {
2035                        push @tmp_parents, $p if $p =~ /$sha1_short/o;
2036                }
2037        }
2038        my $tree = $log_msg->{tree};
2039        if (!defined $tree) {
2040                my $index = set_index($GIT_SVN_INDEX);
2041                index_changes();
2042                chomp($tree = `git-write-tree`);
2043                croak $? if $?;
2044                restore_index($index);
2045        }
2046
2047        # just in case we clobber the existing ref, we still want that ref
2048        # as our parent:
2049        open my $null, '>', '/dev/null' or croak $!;
2050        open my $stderr, '>&', \*STDERR or croak $!;
2051        open STDERR, '>&', $null or croak $!;
2052        if (my $cur = eval { safe_qx('git-rev-parse',
2053                                     "refs/remotes/$GIT_SVN^0") }) {
2054                chomp $cur;
2055                push @tmp_parents, $cur;
2056        }
2057        open STDERR, '>&', $stderr or croak $!;
2058        close $stderr or croak $!;
2059        close $null or croak $!;
2060
2061        if (exists $tree_map{$tree}) {
2062                foreach my $p (@{$tree_map{$tree}}) {
2063                        my $skip;
2064                        foreach (@tmp_parents) {
2065                                # see if a common parent is found
2066                                my $mb = eval {
2067                                        safe_qx('git-merge-base', $_, $p)
2068                                };
2069                                next if ($@ || $?);
2070                                $skip = 1;
2071                                last;
2072                        }
2073                        next if $skip;
2074                        my ($url_p, $r_p, $uuid_p) = cmt_metadata($p);
2075                        next if (($SVN_UUID eq $uuid_p) &&
2076                                                ($log_msg->{revision} > $r_p));
2077                        next if (defined $url_p && defined $SVN_URL &&
2078                                                ($SVN_UUID eq $uuid_p) &&
2079                                                ($url_p eq $SVN_URL));
2080                        push @tmp_parents, $p;
2081                }
2082        }
2083        foreach (@tmp_parents) {
2084                next if $seen_parent{$_};
2085                $seen_parent{$_} = 1;
2086                push @exec_parents, $_;
2087                # MAXPARENT is defined to 16 in commit-tree.c:
2088                last if @exec_parents > 16;
2089        }
2090
2091        set_commit_env($log_msg);
2092        my @exec = ('git-commit-tree', $tree);
2093        push @exec, '-p', $_  foreach @exec_parents;
2094        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2095                                                                or croak $!;
2096        print $msg_fh $log_msg->{msg} or croak $!;
2097        unless ($_no_metadata) {
2098                print $msg_fh "\ngit-svn-id: $SVN_URL\@$log_msg->{revision}",
2099                                        " $SVN_UUID\n" or croak $!;
2100        }
2101        $msg_fh->flush == 0 or croak $!;
2102        close $msg_fh or croak $!;
2103        chomp(my $commit = do { local $/; <$out_fh> });
2104        close $out_fh or croak $!;
2105        waitpid $pid, 0;
2106        croak $? if $?;
2107        if ($commit !~ /^$sha1$/o) {
2108                die "Failed to commit, invalid sha1: $commit\n";
2109        }
2110        sys('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
2111        revdb_set($REVDB, $log_msg->{revision}, $commit);
2112
2113        # this output is read via pipe, do not change:
2114        print "r$log_msg->{revision} = $commit\n";
2115        check_repack();
2116        return $commit;
2117}
2118
2119sub check_repack {
2120        if ($_repack && (--$_repack_nr == 0)) {
2121                $_repack_nr = $_repack;
2122                sys("git repack $_repack_flags");
2123        }
2124}
2125
2126sub set_commit_env {
2127        my ($log_msg) = @_;
2128        my $author = $log_msg->{author};
2129        if (!defined $author || length $author == 0) {
2130                $author = '(no author)';
2131        }
2132        my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
2133                                : ($author,"$author\@$SVN_UUID");
2134        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
2135        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
2136        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
2137}
2138
2139sub apply_mod_line_blob {
2140        my $m = shift;
2141        if ($m->{mode_b} =~ /^120/) {
2142                blob_to_symlink($m->{sha1_b}, $m->{file_b});
2143        } else {
2144                blob_to_file($m->{sha1_b}, $m->{file_b});
2145        }
2146}
2147
2148sub blob_to_symlink {
2149        my ($blob, $link) = @_;
2150        defined $link or croak "\$link not defined!\n";
2151        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2152        if (-l $link || -f _) {
2153                unlink $link or croak $!;
2154        }
2155
2156        my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
2157        symlink $dest, $link or croak $!;
2158}
2159
2160sub blob_to_file {
2161        my ($blob, $file) = @_;
2162        defined $file or croak "\$file not defined!\n";
2163        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2164        if (-l $file || -f _) {
2165                unlink $file or croak $!;
2166        }
2167
2168        open my $blob_fh, '>', $file or croak "$!: $file\n";
2169        my $pid = fork;
2170        defined $pid or croak $!;
2171
2172        if ($pid == 0) {
2173                open STDOUT, '>&', $blob_fh or croak $!;
2174                exec('git-cat-file','blob',$blob) or croak $!;
2175        }
2176        waitpid $pid, 0;
2177        croak $? if $?;
2178
2179        close $blob_fh or croak $!;
2180}
2181
2182sub safe_qx {
2183        my $pid = open my $child, '-|';
2184        defined $pid or croak $!;
2185        if ($pid == 0) {
2186                exec(@_) or croak $!;
2187        }
2188        my @ret = (<$child>);
2189        close $child or croak $?;
2190        die $? if $?; # just in case close didn't error out
2191        return wantarray ? @ret : join('',@ret);
2192}
2193
2194sub svn_compat_check {
2195        if ($_follow_parent) {
2196                print STDERR 'E: --follow-parent functionality is only ',
2197                                "available when SVN libraries are used\n";
2198                exit 1;
2199        }
2200        my @co_help = safe_qx(qw(svn co -h));
2201        unless (grep /ignore-externals/,@co_help) {
2202                print STDERR "W: Installed svn version does not support ",
2203                                "--ignore-externals\n";
2204                $_no_ignore_ext = 1;
2205        }
2206        if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
2207                $_svn_co_url_revs = 1;
2208        }
2209        if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) {
2210                $_svn_pg_peg_revs = 1;
2211        }
2212
2213        # I really, really hope nobody hits this...
2214        unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
2215                print STDERR <<'';
2216W: The installed svn version does not support the --stop-on-copy flag in
2217   the log command.
2218   Lets hope the directory you're tracking is not a branch or tag
2219   and was never moved within the repository...
2220
2221                $_no_stop_copy = 1;
2222        }
2223}
2224
2225# *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
2226# (and they won't honor URL@<rev> without -r<rev>, too!)
2227sub svn_cmd_checkout {
2228        my ($url, $rev, $dir) = @_;
2229        my @cmd = ('svn','co', "-r$rev");
2230        push @cmd, '--ignore-externals' unless $_no_ignore_ext;
2231        $url .= "\@$rev" if $_svn_co_url_revs;
2232        sys(@cmd, $url, $dir);
2233}
2234
2235sub check_upgrade_needed {
2236        if (!-r $REVDB) {
2237                -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2238                open my $fh, '>>',$REVDB or croak $!;
2239                close $fh;
2240        }
2241        my $old = eval {
2242                my $pid = open my $child, '-|';
2243                defined $pid or croak $!;
2244                if ($pid == 0) {
2245                        close STDERR;
2246                        exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $!;
2247                }
2248                my @ret = (<$child>);
2249                close $child or croak $?;
2250                die $? if $?; # just in case close didn't error out
2251                return wantarray ? @ret : join('',@ret);
2252        };
2253        return unless $old;
2254        my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
2255        if ($@ || !$head) {
2256                print STDERR "Please run: $0 rebuild --upgrade\n";
2257                exit 1;
2258        }
2259}
2260
2261# fills %tree_map with a reverse mapping of trees to commits.  Useful
2262# for finding parents to commit on.
2263sub map_tree_joins {
2264        my %seen;
2265        foreach my $br (@_branch_from) {
2266                my $pid = open my $pipe, '-|';
2267                defined $pid or croak $!;
2268                if ($pid == 0) {
2269                        exec(qw(git-rev-list --topo-order --pretty=raw), $br)
2270                                                                or croak $!;
2271                }
2272                while (<$pipe>) {
2273                        if (/^commit ($sha1)$/o) {
2274                                my $commit = $1;
2275
2276                                # if we've seen a commit,
2277                                # we've seen its parents
2278                                last if $seen{$commit};
2279                                my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
2280                                unless (defined $tree) {
2281                                        die "Failed to parse commit $commit\n";
2282                                }
2283                                push @{$tree_map{$tree}}, $commit;
2284                                $seen{$commit} = 1;
2285                        }
2286                }
2287                close $pipe; # we could be breaking the pipe early
2288        }
2289}
2290
2291sub load_all_refs {
2292        if (@_branch_from) {
2293                print STDERR '--branch|-b parameters are ignored when ',
2294                        "--branch-all-refs|-B is passed\n";
2295        }
2296
2297        # don't worry about rev-list on non-commit objects/tags,
2298        # it shouldn't blow up if a ref is a blob or tree...
2299        chomp(@_branch_from = `git-rev-parse --symbolic --all`);
2300}
2301
2302# '<svn username> = real-name <email address>' mapping based on git-svnimport:
2303sub load_authors {
2304        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2305        while (<$authors>) {
2306                chomp;
2307                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
2308                my ($user, $name, $email) = ($1, $2, $3);
2309                $users{$user} = [$name, $email];
2310        }
2311        close $authors or croak $!;
2312}
2313
2314sub rload_authors {
2315        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2316        while (<$authors>) {
2317                chomp;
2318                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2319                my ($user, $name, $email) = ($1, $2, $3);
2320                $rusers{"$name <$email>"} = $user;
2321        }
2322        close $authors or croak $!;
2323}
2324
2325sub svn_propget_base {
2326        my ($p, $f) = @_;
2327        $f .= '@BASE' if $_svn_pg_peg_revs;
2328        return safe_qx(qw/svn propget/, $p, $f);
2329}
2330
2331sub git_svn_each {
2332        my $sub = shift;
2333        foreach (`git-rev-parse --symbolic --all`) {
2334                next unless s#^refs/remotes/##;
2335                chomp $_;
2336                next unless -f "$GIT_DIR/svn/$_/info/url";
2337                &$sub($_);
2338        }
2339}
2340
2341sub migrate_revdb {
2342        git_svn_each(sub {
2343                my $id = shift;
2344                defined(my $pid = fork) or croak $!;
2345                if (!$pid) {
2346                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2347                        init_vars();
2348                        exit 0 if -r $REVDB;
2349                        print "Upgrading svn => git mapping...\n";
2350                        -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2351                        open my $fh, '>>',$REVDB or croak $!;
2352                        close $fh;
2353                        rebuild();
2354                        print "Done upgrading. You may now delete the ",
2355                                "deprecated $GIT_SVN_DIR/revs directory\n";
2356                        exit 0;
2357                }
2358                waitpid $pid, 0;
2359                croak $? if $?;
2360        });
2361}
2362
2363sub migration_check {
2364        migrate_revdb() unless (-e $REVDB);
2365        return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
2366        print "Upgrading repository...\n";
2367        unless (-d "$GIT_DIR/svn") {
2368                mkdir "$GIT_DIR/svn" or croak $!;
2369        }
2370        print "Data from a previous version of git-svn exists, but\n\t",
2371                                "$GIT_SVN_DIR\n\t(required for this version ",
2372                                "($VERSION) of git-svn) does not.\n";
2373
2374        foreach my $x (`git-rev-parse --symbolic --all`) {
2375                next unless $x =~ s#^refs/remotes/##;
2376                chomp $x;
2377                next unless -f "$GIT_DIR/$x/info/url";
2378                my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
2379                next unless $u;
2380                my $dn = dirname("$GIT_DIR/svn/$x");
2381                mkpath([$dn]) unless -d $dn;
2382                rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
2383        }
2384        migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
2385        print "Done upgrading.\n";
2386}
2387
2388sub find_rev_before {
2389        my ($r, $id, $eq_ok) = @_;
2390        my $f = "$GIT_DIR/svn/$id/.rev_db";
2391        return (undef,undef) unless -r $f;
2392        --$r unless $eq_ok;
2393        while ($r > 0) {
2394                if (my $c = revdb_get($f, $r)) {
2395                        return ($r, $c);
2396                }
2397                --$r;
2398        }
2399        return (undef, undef);
2400}
2401
2402sub init_vars {
2403        $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
2404        $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
2405        $REVDB = "$GIT_SVN_DIR/.rev_db";
2406        $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
2407        $SVN_URL = undef;
2408        $SVN_WC = "$GIT_SVN_DIR/tree";
2409        %tree_map = ();
2410}
2411
2412# convert GetOpt::Long specs for use by git-repo-config
2413sub read_repo_config {
2414        return unless -d $GIT_DIR;
2415        my $opts = shift;
2416        foreach my $o (keys %$opts) {
2417                my $v = $opts->{$o};
2418                my ($key) = ($o =~ /^([a-z\-]+)/);
2419                $key =~ s/-//g;
2420                my $arg = 'git-repo-config';
2421                $arg .= ' --int' if ($o =~ /[:=]i$/);
2422                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
2423                if (ref $v eq 'ARRAY') {
2424                        chomp(my @tmp = `$arg --get-all svn.$key`);
2425                        @$v = @tmp if @tmp;
2426                } else {
2427                        chomp(my $tmp = `$arg --get svn.$key`);
2428                        if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
2429                                $$v = $tmp;
2430                        }
2431                }
2432        }
2433}
2434
2435sub set_default_vals {
2436        if (defined $_repack) {
2437                $_repack = 1000 if ($_repack <= 0);
2438                $_repack_nr = $_repack;
2439                $_repack_flags ||= '-d';
2440        }
2441}
2442
2443sub read_grafts {
2444        my $gr_file = shift;
2445        my ($grafts, $comments) = ({}, {});
2446        if (open my $fh, '<', $gr_file) {
2447                my @tmp;
2448                while (<$fh>) {
2449                        if (/^($sha1)\s+/) {
2450                                my $c = $1;
2451                                if (@tmp) {
2452                                        @{$comments->{$c}} = @tmp;
2453                                        @tmp = ();
2454                                }
2455                                foreach my $p (split /\s+/, $_) {
2456                                        $grafts->{$c}->{$p} = 1;
2457                                }
2458                        } else {
2459                                push @tmp, $_;
2460                        }
2461                }
2462                close $fh or croak $!;
2463                @{$comments->{'END'}} = @tmp if @tmp;
2464        }
2465        return ($grafts, $comments);
2466}
2467
2468sub write_grafts {
2469        my ($grafts, $comments, $gr_file) = @_;
2470
2471        open my $fh, '>', $gr_file or croak $!;
2472        foreach my $c (sort keys %$grafts) {
2473                if ($comments->{$c}) {
2474                        print $fh $_ foreach @{$comments->{$c}};
2475                }
2476                my $p = $grafts->{$c};
2477                my %x; # real parents
2478                delete $p->{$c}; # commits are not self-reproducing...
2479                my $pid = open my $ch, '-|';
2480                defined $pid or croak $!;
2481                if (!$pid) {
2482                        exec(qw/git-cat-file commit/, $c) or croak $!;
2483                }
2484                while (<$ch>) {
2485                        if (/^parent ($sha1)/) {
2486                                $x{$1} = $p->{$1} = 1;
2487                        } else {
2488                                last unless /^\S/;
2489                        }
2490                }
2491                close $ch; # breaking the pipe
2492
2493                # if real parents are the only ones in the grafts, drop it
2494                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2495
2496                my (@ip, @jp, $mb);
2497                my %del = %x;
2498                @ip = @jp = keys %$p;
2499                foreach my $i (@ip) {
2500                        next if $del{$i} || $p->{$i} == 2;
2501                        foreach my $j (@jp) {
2502                                next if $i eq $j || $del{$j} || $p->{$j} == 2;
2503                                $mb = eval { safe_qx('git-merge-base',$i,$j) };
2504                                next unless $mb;
2505                                chomp $mb;
2506                                next if $x{$mb};
2507                                if ($mb eq $j) {
2508                                        delete $p->{$i};
2509                                        $del{$i} = 1;
2510                                } elsif ($mb eq $i) {
2511                                        delete $p->{$j};
2512                                        $del{$j} = 1;
2513                                }
2514                        }
2515                }
2516
2517                # if real parents are the only ones in the grafts, drop it
2518                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2519
2520                print $fh $c, ' ', join(' ', sort keys %$p),"\n";
2521        }
2522        if ($comments->{'END'}) {
2523                print $fh $_ foreach @{$comments->{'END'}};
2524        }
2525        close $fh or croak $!;
2526}
2527
2528sub read_url_paths_all {
2529        my ($l_map, $pfx, $p) = @_;
2530        my @dir;
2531        foreach (<$p/*>) {
2532                if (-r "$_/info/url") {
2533                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2534                        my $id = $pfx . basename $_;
2535                        my $url = file_to_s("$_/info/url");
2536                        my ($u, $p) = repo_path_split($url);
2537                        $l_map->{$u}->{$p} = $id;
2538                } elsif (-d $_) {
2539                        push @dir, $_;
2540                }
2541        }
2542        foreach (@dir) {
2543                my $x = $_;
2544                $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
2545                read_url_paths_all($l_map, $x, $_);
2546        }
2547}
2548
2549# this one only gets ids that have been imported, not new ones
2550sub read_url_paths {
2551        my $l_map = {};
2552        git_svn_each(sub { my $x = shift;
2553                        my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
2554                        my ($u, $p) = repo_path_split($url);
2555                        $l_map->{$u}->{$p} = $x;
2556                        });
2557        return $l_map;
2558}
2559
2560sub extract_metadata {
2561        my $id = shift or return (undef, undef, undef);
2562        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
2563                                                        \s([a-f\d\-]+)$/x);
2564        if (!defined $rev || !$uuid || !$url) {
2565                # some of the original repositories I made had
2566                # identifiers like this:
2567                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
2568        }
2569        return ($url, $rev, $uuid);
2570}
2571
2572sub cmt_metadata {
2573        return extract_metadata((grep(/^git-svn-id: /,
2574                safe_qx(qw/git-cat-file commit/, shift)))[-1]);
2575}
2576
2577sub get_commit_time {
2578        my $cmt = shift;
2579        defined(my $pid = open my $fh, '-|') or croak $!;
2580        if (!$pid) {
2581                exec qw/git-rev-list --pretty=raw -n1/, $cmt or croak $!;
2582        }
2583        while (<$fh>) {
2584                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
2585                my ($s, $tz) = ($1, $2);
2586                if ($tz =~ s/^\+//) {
2587                        $s += tz_to_s_offset($tz);
2588                } elsif ($tz =~ s/^\-//) {
2589                        $s -= tz_to_s_offset($tz);
2590                }
2591                close $fh;
2592                return $s;
2593        }
2594        die "Can't get commit time for commit: $cmt\n";
2595}
2596
2597sub tz_to_s_offset {
2598        my ($tz) = @_;
2599        $tz =~ s/(\d\d)$//;
2600        return ($1 * 60) + ($tz * 3600);
2601}
2602
2603# adapted from pager.c
2604sub config_pager {
2605        $_pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2606        if (!defined $_pager) {
2607                $_pager = 'less';
2608        } elsif (length $_pager == 0 || $_pager eq 'cat') {
2609                $_pager = undef;
2610        }
2611}
2612
2613sub run_pager {
2614        return unless -t *STDOUT;
2615        pipe my $rfd, my $wfd or return;
2616        defined(my $pid = fork) or croak $!;
2617        if (!$pid) {
2618                open STDOUT, '>&', $wfd or croak $!;
2619                return;
2620        }
2621        open STDIN, '<&', $rfd or croak $!;
2622        $ENV{LESS} ||= 'FRSX';
2623        exec $_pager or croak "Can't run pager: $! ($_pager)\n";
2624}
2625
2626sub get_author_info {
2627        my ($dest, $author, $t, $tz) = @_;
2628        $author =~ s/(?:^\s*|\s*$)//g;
2629        $dest->{a_raw} = $author;
2630        my $_a;
2631        if ($_authors) {
2632                $_a = $rusers{$author} || undef;
2633        }
2634        if (!$_a) {
2635                ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
2636        }
2637        $dest->{t} = $t;
2638        $dest->{tz} = $tz;
2639        $dest->{a} = $_a;
2640        # Date::Parse isn't in the standard Perl distro :(
2641        if ($tz =~ s/^\+//) {
2642                $t += tz_to_s_offset($tz);
2643        } elsif ($tz =~ s/^\-//) {
2644                $t -= tz_to_s_offset($tz);
2645        }
2646        $dest->{t_utc} = $t;
2647}
2648
2649sub process_commit {
2650        my ($c, $r_min, $r_max, $defer) = @_;
2651        if (defined $r_min && defined $r_max) {
2652                if ($r_min == $c->{r} && $r_min == $r_max) {
2653                        show_commit($c);
2654                        return 0;
2655                }
2656                return 1 if $r_min == $r_max;
2657                if ($r_min < $r_max) {
2658                        # we need to reverse the print order
2659                        return 0 if (defined $_limit && --$_limit < 0);
2660                        push @$defer, $c;
2661                        return 1;
2662                }
2663                if ($r_min != $r_max) {
2664                        return 1 if ($r_min < $c->{r});
2665                        return 1 if ($r_max > $c->{r});
2666                }
2667        }
2668        return 0 if (defined $_limit && --$_limit < 0);
2669        show_commit($c);
2670        return 1;
2671}
2672
2673sub show_commit {
2674        my $c = shift;
2675        if ($_oneline) {
2676                my $x = "\n";
2677                if (my $l = $c->{l}) {
2678                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2679                        $x = $l->[0];
2680                }
2681                $_l_fmt ||= 'A' . length($c->{r});
2682                print 'r',pack($_l_fmt, $c->{r}),' | ';
2683                print "$c->{c} | " if $_show_commit;
2684                print $x;
2685        } else {
2686                show_commit_normal($c);
2687        }
2688}
2689
2690sub show_commit_changed_paths {
2691        my ($c) = @_;
2692        return unless $c->{changed};
2693        print "Changed paths:\n", @{$c->{changed}};
2694}
2695
2696sub show_commit_normal {
2697        my ($c) = @_;
2698        print '-' x72, "\nr$c->{r} | ";
2699        print "$c->{c} | " if $_show_commit;
2700        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2701                                 localtime($c->{t_utc})), ' | ';
2702        my $nr_line = 0;
2703
2704        if (my $l = $c->{l}) {
2705                while ($l->[$#$l] eq "\n" && $#$l > 0
2706                                          && $l->[($#$l - 1)] eq "\n") {
2707                        pop @$l;
2708                }
2709                $nr_line = scalar @$l;
2710                if (!$nr_line) {
2711                        print "1 line\n\n\n";
2712                } else {
2713                        if ($nr_line == 1) {
2714                                $nr_line = '1 line';
2715                        } else {
2716                                $nr_line .= ' lines';
2717                        }
2718                        print $nr_line, "\n";
2719                        show_commit_changed_paths($c);
2720                        print "\n";
2721                        print $_ foreach @$l;
2722                }
2723        } else {
2724                print "1 line\n";
2725                show_commit_changed_paths($c);
2726                print "\n";
2727
2728        }
2729        foreach my $x (qw/raw diff/) {
2730                if ($c->{$x}) {
2731                        print "\n";
2732                        print $_ foreach @{$c->{$x}}
2733                }
2734        }
2735}
2736
2737sub libsvn_load {
2738        return unless $_use_lib;
2739        $_use_lib = eval {
2740                require SVN::Core;
2741                if ($SVN::Core::VERSION lt '1.1.0') {
2742                        die "Need SVN::Core 1.1.0 or better ",
2743                                        "(got $SVN::Core::VERSION) ",
2744                                        "Falling back to command-line svn\n";
2745                }
2746                require SVN::Ra;
2747                require SVN::Delta;
2748                push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
2749                push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
2750                *SVN::Git::Fetcher::process_rm = *process_rm;
2751                *SVN::Git::Fetcher::safe_qx = *safe_qx;
2752                my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2753                                        $SVN::Node::dir.$SVN::Node::unknown.
2754                                        $SVN::Node::none.$SVN::Node::file.
2755                                        $SVN::Node::dir.$SVN::Node::unknown.
2756                                        $SVN::Auth::SSL::CNMISMATCH.
2757                                        $SVN::Auth::SSL::NOTYETVALID.
2758                                        $SVN::Auth::SSL::EXPIRED.
2759                                        $SVN::Auth::SSL::UNKNOWNCA.
2760                                        $SVN::Auth::SSL::OTHER;
2761                1;
2762        };
2763}
2764
2765sub _simple_prompt {
2766        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2767        $may_save = undef if $_no_auth_cache;
2768        $default_username = $_username if defined $_username;
2769        if (defined $default_username && length $default_username) {
2770                if (defined $realm && length $realm) {
2771                        print "Authentication realm: $realm\n";
2772                }
2773                $cred->username($default_username);
2774        } else {
2775                _username_prompt($cred, $realm, $may_save, $pool);
2776        }
2777        $cred->password(_read_password("Password for '" .
2778                                       $cred->username . "': ", $realm));
2779        $cred->may_save($may_save);
2780        $SVN::_Core::SVN_NO_ERROR;
2781}
2782
2783sub _ssl_server_trust_prompt {
2784        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2785        $may_save = undef if $_no_auth_cache;
2786        print "Error validating server certificate for '$realm':\n";
2787        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2788                print " - The certificate is not issued by a trusted ",
2789                      "authority. Use the\n",
2790                      "   fingerprint to validate the certificate manually!\n";
2791        }
2792        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2793                print " - The certificate hostname does not match.\n";
2794        }
2795        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2796                print " - The certificate is not yet valid.\n";
2797        }
2798        if ($failures & $SVN::Auth::SSL::EXPIRED) {
2799                print " - The certificate has expired.\n";
2800        }
2801        if ($failures & $SVN::Auth::SSL::OTHER) {
2802                print " - The certificate has an unknown error.\n";
2803        }
2804        printf( "Certificate information:\n".
2805                " - Hostname: %s\n".
2806                " - Valid: from %s until %s\n".
2807                " - Issuer: %s\n".
2808                " - Fingerprint: %s\n",
2809                map $cert_info->$_, qw(hostname valid_from valid_until
2810                                       issuer_dname fingerprint) );
2811        my $choice;
2812prompt:
2813        print $may_save ?
2814              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2815              "(R)eject or accept (t)emporarily? ";
2816        $choice = lc(substr(<STDIN> || 'R', 0, 1));
2817        if ($choice =~ /^t$/i) {
2818                $cred->may_save(undef);
2819        } elsif ($choice =~ /^r$/i) {
2820                return -1;
2821        } elsif ($may_save && $choice =~ /^p$/i) {
2822                $cred->may_save($may_save);
2823        } else {
2824                goto prompt;
2825        }
2826        $cred->accepted_failures($failures);
2827        $SVN::_Core::SVN_NO_ERROR;
2828}
2829
2830sub _ssl_client_cert_prompt {
2831        my ($cred, $realm, $may_save, $pool) = @_;
2832        $may_save = undef if $_no_auth_cache;
2833        print "Client certificate filename: ";
2834        chomp(my $filename = <STDIN>);
2835        $cred->cert_file($filename);
2836        $cred->may_save($may_save);
2837        $SVN::_Core::SVN_NO_ERROR;
2838}
2839
2840sub _ssl_client_cert_pw_prompt {
2841        my ($cred, $realm, $may_save, $pool) = @_;
2842        $may_save = undef if $_no_auth_cache;
2843        $cred->password(_read_password("Password: ", $realm));
2844        $cred->may_save($may_save);
2845        $SVN::_Core::SVN_NO_ERROR;
2846}
2847
2848sub _username_prompt {
2849        my ($cred, $realm, $may_save, $pool) = @_;
2850        $may_save = undef if $_no_auth_cache;
2851        if (defined $realm && length $realm) {
2852                print "Authentication realm: $realm\n";
2853        }
2854        my $username;
2855        if (defined $_username) {
2856                $username = $_username;
2857        } else {
2858                print "Username: ";
2859                chomp($username = <STDIN>);
2860        }
2861        $cred->username($username);
2862        $cred->may_save($may_save);
2863        $SVN::_Core::SVN_NO_ERROR;
2864}
2865
2866sub _read_password {
2867        my ($prompt, $realm) = @_;
2868        print $prompt;
2869        require Term::ReadKey;
2870        Term::ReadKey::ReadMode('noecho');
2871        my $password = '';
2872        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2873                last if $key =~ /[\012\015]/; # \n\r
2874                $password .= $key;
2875        }
2876        Term::ReadKey::ReadMode('restore');
2877        print "\n";
2878        $password;
2879}
2880
2881sub libsvn_connect {
2882        my ($url) = @_;
2883        SVN::_Core::svn_config_ensure($_config_dir, undef);
2884        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2885            SVN::Client::get_simple_provider(),
2886            SVN::Client::get_ssl_server_trust_file_provider(),
2887            SVN::Client::get_simple_prompt_provider(
2888              \&_simple_prompt, 2),
2889            SVN::Client::get_ssl_client_cert_prompt_provider(
2890              \&_ssl_client_cert_prompt, 2),
2891            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2892              \&_ssl_client_cert_pw_prompt, 2),
2893            SVN::Client::get_username_provider(),
2894            SVN::Client::get_ssl_server_trust_prompt_provider(
2895              \&_ssl_server_trust_prompt),
2896            SVN::Client::get_username_prompt_provider(
2897              \&_username_prompt, 2),
2898          ]);
2899        my $config = SVN::Core::config_get_config($_config_dir);
2900        my $ra = SVN::Ra->new(url => $url, auth => $baton,
2901                              config => $config,
2902                              pool => SVN::Pool->new,
2903                              auth_provider_callbacks => $callbacks);
2904
2905        my $df = $ENV{GIT_SVN_DELTA_FETCH};
2906        if (defined $df) {
2907                $_xfer_delta = $df;
2908        } else {
2909                $_xfer_delta = ($url =~ m#^file://#) ? undef : 1;
2910        }
2911        $ra->{svn_path} = $url;
2912        $ra->{repos_root} = $ra->get_repos_root;
2913        $ra->{svn_path} =~ s#^\Q$ra->{repos_root}\E/*##;
2914        push @repo_path_split_cache, qr/^(\Q$ra->{repos_root}\E)/;
2915        return $ra;
2916}
2917
2918sub libsvn_can_do_switch {
2919        unless (defined $_svn_can_do_switch) {
2920                my $pool = SVN::Pool->new;
2921                my $rep = eval {
2922                        $SVN->do_switch(1, '', 0, $SVN->{url},
2923                                        SVN::Delta::Editor->new, $pool);
2924                };
2925                if ($@) {
2926                        $_svn_can_do_switch = 0;
2927                } else {
2928                        $rep->abort_report($pool);
2929                        $_svn_can_do_switch = 1;
2930                }
2931                $pool->clear;
2932        }
2933        $_svn_can_do_switch;
2934}
2935
2936sub libsvn_dup_ra {
2937        my ($ra) = @_;
2938        SVN::Ra->new(map { $_ => $ra->{$_} } qw/config url
2939                     auth auth_provider_callbacks repos_root svn_path/);
2940}
2941
2942sub libsvn_get_file {
2943        my ($gui, $f, $rev, $chg, $untracked) = @_;
2944        $f =~ s#^/##;
2945        print "\t$chg\t$f\n" unless $_q;
2946
2947        my ($hash, $pid, $in, $out);
2948        my $pool = SVN::Pool->new;
2949        defined($pid = open3($in, $out, '>&STDERR',
2950                                qw/git-hash-object -w --stdin/)) or croak $!;
2951        # redirect STDOUT for SVN 1.1.x compatibility
2952        open my $stdout, '>&', \*STDOUT or croak $!;
2953        open STDOUT, '>&', $in or croak $!;
2954        my ($r, $props) = $SVN->get_file($f, $rev, \*STDOUT, $pool);
2955        $in->flush == 0 or croak $!;
2956        open STDOUT, '>&', $stdout or croak $!;
2957        close $in or croak $!;
2958        close $stdout or croak $!;
2959        $pool->clear;
2960        chomp($hash = do { local $/; <$out> });
2961        close $out or croak $!;
2962        waitpid $pid, 0;
2963        $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2964
2965        my $mode = exists $props->{'svn:executable'} ? '100755' : '100644';
2966        if (exists $props->{'svn:special'}) {
2967                $mode = '120000';
2968                my $link = `git-cat-file blob $hash`;
2969                $link =~ s/^link // or die "svn:special file with contents: <",
2970                                                $link, "> is not understood\n";
2971                defined($pid = open3($in, $out, '>&STDERR',
2972                                qw/git-hash-object -w --stdin/)) or croak $!;
2973                print $in $link;
2974                $in->flush == 0 or croak $!;
2975                close $in or croak $!;
2976                chomp($hash = do { local $/; <$out> });
2977                close $out or croak $!;
2978                waitpid $pid, 0;
2979                $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2980        }
2981        %{$untracked->{file_prop}->{$f}} = %$props;
2982        print $gui $mode,' ',$hash,"\t",$f,"\0" or croak $!;
2983}
2984
2985sub uri_encode {
2986        my ($f) = @_;
2987        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2988        $f
2989}
2990
2991sub uri_decode {
2992        my ($f) = @_;
2993        $f =~ tr/+/ /;
2994        $f =~ s/%([A-F0-9]{2})/chr hex($1)/ge;
2995        $f
2996}
2997
2998sub libsvn_log_entry {
2999        my ($rev, $author, $date, $msg, $parents, $untracked) = @_;
3000        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
3001                                         (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
3002                                or die "Unable to parse date: $date\n";
3003        if (defined $author && length $author > 0 &&
3004            defined $_authors && ! defined $users{$author}) {
3005                die "Author: $author not defined in $_authors file\n";
3006        }
3007        $msg = '' if ($rev == 0 && !defined $msg);
3008
3009        open my $un, '>>', "$GIT_SVN_DIR/unhandled.log" or croak $!;
3010        my $h;
3011        print $un "r$rev\n" or croak $!;
3012        $h = $untracked->{empty};
3013        foreach (sort keys %$h) {
3014                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
3015                print $un "  $act: ", uri_encode($_), "\n" or croak $!;
3016                warn "W: $act: $_\n";
3017        }
3018        foreach my $t (qw/dir_prop file_prop/) {
3019                $h = $untracked->{$t} or next;
3020                foreach my $path (sort keys %$h) {
3021                        my $ppath = $path eq '' ? '.' : $path;
3022                        foreach my $prop (sort keys %{$h->{$path}}) {
3023                                next if $SKIP{$prop};
3024                                my $v = $h->{$path}->{$prop};
3025                                if (defined $v) {
3026                                        print $un "  +$t: ",
3027                                                  uri_encode($ppath), ' ',
3028                                                  uri_encode($prop), ' ',
3029                                                  uri_encode($v), "\n"
3030                                                  or croak $!;
3031                                } else {
3032                                        print $un "  -$t: ",
3033                                                  uri_encode($ppath), ' ',
3034                                                  uri_encode($prop), "\n"
3035                                                  or croak $!;
3036                                }
3037                        }
3038                }
3039        }
3040        foreach my $t (qw/absent_file absent_directory/) {
3041                $h = $untracked->{$t} or next;
3042                foreach my $parent (sort keys %$h) {
3043                        foreach my $path (sort @{$h->{$parent}}) {
3044                                print $un "  $t: ",
3045                                      uri_encode("$parent/$path"), "\n"
3046                                      or croak $!;
3047                                warn "W: $t: $parent/$path ",
3048                                     "Insufficient permissions?\n";
3049                        }
3050                }
3051        }
3052
3053        # revprops (make this optional? it's an extra network trip...)
3054        my $pool = SVN::Pool->new;
3055        my $rp = $SVN->rev_proplist($rev, $pool);
3056        foreach (sort keys %$rp) {
3057                next if /^svn:(?:author|date|log)$/;
3058                print $un "  rev_prop: ", uri_encode($_), ' ',
3059                          uri_encode($rp->{$_}), "\n";
3060        }
3061        $pool->clear;
3062        close $un or croak $!;
3063
3064        { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
3065          author => $author, msg => $msg."\n", parents => $parents || [],
3066          revprops => $rp }
3067}
3068
3069sub process_rm {
3070        my ($gui, $last_commit, $f, $q) = @_;
3071        # remove entire directories.
3072        if (safe_qx('git-ls-tree',$last_commit,'--',$f) =~ /^040000 tree/) {
3073                defined(my $pid = open my $ls, '-|') or croak $!;
3074                if (!$pid) {
3075                        exec(qw/git-ls-tree -r --name-only -z/,
3076                                $last_commit,'--',$f) or croak $!;
3077                }
3078                local $/ = "\0";
3079                while (<$ls>) {
3080                        print $gui '0 ',0 x 40,"\t",$_ or croak $!;
3081                        print "\tD\t$_\n" unless $q;
3082                }
3083                print "\tD\t$f/\n" unless $q;
3084                close $ls or croak $?;
3085                return $SVN::Node::dir;
3086        } else {
3087                print $gui '0 ',0 x 40,"\t",$f,"\0" or croak $!;
3088                print "\tD\t$f\n" unless $q;
3089                return $SVN::Node::file;
3090        }
3091}
3092
3093sub libsvn_fetch {
3094        $_xfer_delta ? libsvn_fetch_delta(@_) : libsvn_fetch_full(@_);
3095}
3096
3097sub libsvn_fetch_delta {
3098        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
3099        my $pool = SVN::Pool->new;
3100        my $ed = SVN::Git::Fetcher->new({ c => $last_commit, q => $_q });
3101        my $reporter = $SVN->do_update($rev, '', 1, $ed, $pool);
3102        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3103        my (undef, $last_rev, undef) = cmt_metadata($last_commit);
3104        $reporter->set_path('', $last_rev, 0, @lock, $pool);
3105        $reporter->finish_report($pool);
3106        $pool->clear;
3107        unless ($ed->{git_commit_ok}) {
3108                die "SVN connection failed somewhere...\n";
3109        }
3110        libsvn_log_entry($rev, $author, $date, $msg, [$last_commit], $ed);
3111}
3112
3113sub libsvn_fetch_full {
3114        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
3115        open my $gui, '| git-update-index -z --index-info' or croak $!;
3116        my %amr;
3117        my $ut = { empty => {}, dir_prop => {}, file_prop => {} };
3118        my $p = $SVN->{svn_path};
3119        foreach my $f (keys %$paths) {
3120                my $m = $paths->{$f}->action();
3121                if (length $p) {
3122                        $f =~ s#^/\Q$p\E/##;
3123                        next if $f =~ m#^/#;
3124                } else {
3125                        $f =~ s#^/##;
3126                }
3127                if ($m =~ /^[DR]$/) {
3128                        my $t = process_rm($gui, $last_commit, $f, $_q);
3129                        if ($m eq 'D') {
3130                                $ut->{empty}->{$f} = 0 if $t == $SVN::Node::dir;
3131                                next;
3132                        }
3133                        # 'R' can be file replacements, too, right?
3134                }
3135                my $pool = SVN::Pool->new;
3136                my $t = $SVN->check_path($f, $rev, $pool);
3137                if ($t == $SVN::Node::file) {
3138                        if ($m =~ /^[AMR]$/) {
3139                                $amr{$f} = $m;
3140                        } else {
3141                                die "Unrecognized action: $m, ($f r$rev)\n";
3142                        }
3143                } elsif ($t == $SVN::Node::dir && $m =~ /^[AR]$/) {
3144                        my @traversed = ();
3145                        libsvn_traverse($gui, '', $f, $rev, \@traversed, $ut);
3146                        if (@traversed) {
3147                                foreach (@traversed) {
3148                                        $amr{$_} = $m;
3149                                }
3150                        } else {
3151                                my ($dir, $file) = ($f =~ m#^(.*?)/?([^/]+)$#);
3152                                delete $ut->{empty}->{$dir};
3153                                $ut->{empty}->{$f} = 1;
3154                        }
3155                }
3156                $pool->clear;
3157        }
3158        foreach (keys %amr) {
3159                libsvn_get_file($gui, $_, $rev, $amr{$_}, $ut);
3160                my ($d) = ($_ =~ m#^(.*?)/?(?:[^/]+)$#);
3161                delete $ut->{empty}->{$d};
3162        }
3163        unless (exists $ut->{dir_prop}->{''}) {
3164                my $pool = SVN::Pool->new;
3165                my (undef, undef, $props) = $SVN->get_dir('', $rev, $pool);
3166                %{$ut->{dir_prop}->{''}} = %$props;
3167                $pool->clear;
3168        }
3169        close $gui or croak $?;
3170        libsvn_log_entry($rev, $author, $date, $msg, [$last_commit], $ut);
3171}
3172
3173sub svn_grab_base_rev {
3174        defined(my $pid = open my $fh, '-|') or croak $!;
3175        if (!$pid) {
3176                open my $null, '>', '/dev/null' or croak $!;
3177                open STDERR, '>&', $null or croak $!;
3178                exec qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0"
3179                                                                or croak $!;
3180        }
3181        chomp(my $c = do { local $/; <$fh> });
3182        close $fh;
3183        if (defined $c && length $c) {
3184                my ($url, $rev, $uuid) = cmt_metadata($c);
3185                return ($rev, $c) if defined $rev;
3186        }
3187        if ($_no_metadata) {
3188                my $offset = -41; # from tail
3189                my $rl;
3190                open my $fh, '<', $REVDB or
3191                        die "--no-metadata specified and $REVDB not readable\n";
3192                seek $fh, $offset, 2;
3193                $rl = readline $fh;
3194                defined $rl or return (undef, undef);
3195                chomp $rl;
3196                while ($c ne $rl && tell $fh != 0) {
3197                        $offset -= 41;
3198                        seek $fh, $offset, 2;
3199                        $rl = readline $fh;
3200                        defined $rl or return (undef, undef);
3201                        chomp $rl;
3202                }
3203                my $rev = tell $fh;
3204                croak $! if ($rev < -1);
3205                $rev =  ($rev - 41) / 41;
3206                close $fh or croak $!;
3207                return ($rev, $c);
3208        }
3209        return (undef, undef);
3210}
3211
3212sub libsvn_parse_revision {
3213        my $base = shift;
3214        my $head = $SVN->get_latest_revnum();
3215        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
3216                return ($base + 1, $head) if (defined $base);
3217                return (0, $head);
3218        }
3219        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
3220        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
3221        if ($_revision =~ /^BASE:(\d+)$/) {
3222                return ($base + 1, $1) if (defined $base);
3223                return (0, $head);
3224        }
3225        return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
3226        die "revision argument: $_revision not understood by git-svn\n",
3227                "Try using the command-line svn client instead\n";
3228}
3229
3230sub libsvn_traverse {
3231        my ($gui, $pfx, $path, $rev, $files, $untracked) = @_;
3232        my $cwd = length $pfx ? "$pfx/$path" : $path;
3233        my $pool = SVN::Pool->new;
3234        $cwd =~ s#^\Q$SVN->{svn_path}\E##;
3235        my $nr = 0;
3236        my ($dirent, $r, $props) = $SVN->get_dir($cwd, $rev, $pool);
3237        %{$untracked->{dir_prop}->{$cwd}} = %$props;
3238        foreach my $d (keys %$dirent) {
3239                my $t = $dirent->{$d}->kind;
3240                if ($t == $SVN::Node::dir) {
3241                        my $i = libsvn_traverse($gui, $cwd, $d, $rev,
3242                                                $files, $untracked);
3243                        if ($i) {
3244                                $nr += $i;
3245                        } else {
3246                                $untracked->{empty}->{"$cwd/$d"} = 1;
3247                        }
3248                } elsif ($t == $SVN::Node::file) {
3249                        $nr++;
3250                        my $file = "$cwd/$d";
3251                        if (defined $files) {
3252                                push @$files, $file;
3253                        } else {
3254                                libsvn_get_file($gui, $file, $rev, 'A',
3255                                                $untracked);
3256                                my ($dir) = ($file =~ m#^(.*?)/?(?:[^/]+)$#);
3257                                delete $untracked->{empty}->{$dir};
3258                        }
3259                }
3260        }
3261        $pool->clear;
3262        $nr;
3263}
3264
3265sub libsvn_traverse_ignore {
3266        my ($fh, $path, $r) = @_;
3267        $path =~ s#^/+##g;
3268        my $pool = SVN::Pool->new;
3269        my ($dirent, undef, $props) = $SVN->get_dir($path, $r, $pool);
3270        my $p = $path;
3271        $p =~ s#^\Q$SVN->{svn_path}\E/##;
3272        print $fh length $p ? "\n# $p\n" : "\n# /\n";
3273        if (my $s = $props->{'svn:ignore'}) {
3274                $s =~ s/[\r\n]+/\n/g;
3275                chomp $s;
3276                if (length $p == 0) {
3277                        $s =~ s#\n#\n/$p#g;
3278                        print $fh "/$s\n";
3279                } else {
3280                        $s =~ s#\n#\n/$p/#g;
3281                        print $fh "/$p/$s\n";
3282                }
3283        }
3284        foreach (sort keys %$dirent) {
3285                next if $dirent->{$_}->kind != $SVN::Node::dir;
3286                libsvn_traverse_ignore($fh, "$path/$_", $r);
3287        }
3288        $pool->clear;
3289}
3290
3291sub revisions_eq {
3292        my ($path, $r0, $r1) = @_;
3293        return 1 if $r0 == $r1;
3294        my $nr = 0;
3295        if ($_use_lib) {
3296                # should be OK to use Pool here (r1 - r0) should be small
3297                my $pool = SVN::Pool->new;
3298                libsvn_get_log($SVN, [$path], $r0, $r1,
3299                                0, 0, 1, sub {$nr++}, $pool);
3300                $pool->clear;
3301        } else {
3302                my ($url, undef) = repo_path_split($SVN_URL);
3303                my $svn_log = svn_log_raw("$url/$path","-r$r0:$r1");
3304                while (next_log_entry($svn_log)) { $nr++ }
3305                close $svn_log->{fh};
3306        }
3307        return 0 if ($nr > 1);
3308        return 1;
3309}
3310
3311sub libsvn_find_parent_branch {
3312        my ($paths, $rev, $author, $date, $msg) = @_;
3313        my $svn_path = '/'.$SVN->{svn_path};
3314
3315        # look for a parent from another branch:
3316        my $i = $paths->{$svn_path} or return;
3317        my $branch_from = $i->copyfrom_path or return;
3318        my $r = $i->copyfrom_rev;
3319        print STDERR  "Found possible branch point: ",
3320                                "$branch_from => $svn_path, $r\n";
3321        $branch_from =~ s#^/##;
3322        my $l_map = {};
3323        read_url_paths_all($l_map, '', "$GIT_DIR/svn");
3324        my $url = $SVN->{repos_root};
3325        defined $l_map->{$url} or return;
3326        my $id = $l_map->{$url}->{$branch_from};
3327        if (!defined $id && $_follow_parent) {
3328                print STDERR "Following parent: $branch_from\@$r\n";
3329                # auto create a new branch and follow it
3330                $id = basename($branch_from);
3331                $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
3332                while (-r "$GIT_DIR/svn/$id") {
3333                        # just grow a tail if we're not unique enough :x
3334                        $id .= '-';
3335                }
3336        }
3337        return unless defined $id;
3338
3339        my ($r0, $parent) = find_rev_before($r,$id,1);
3340        if ($_follow_parent && (!defined $r0 || !defined $parent)) {
3341                defined(my $pid = fork) or croak $!;
3342                if (!$pid) {
3343                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
3344                        init_vars();
3345                        $SVN_URL = "$url/$branch_from";
3346                        $SVN = undef;
3347                        setup_git_svn();
3348                        # we can't assume SVN_URL exists at r+1:
3349                        $_revision = "0:$r";
3350                        fetch_lib();
3351                        exit 0;
3352                }
3353                waitpid $pid, 0;
3354                croak $? if $?;
3355                ($r0, $parent) = find_rev_before($r,$id,1);
3356        }
3357        return unless (defined $r0 && defined $parent);
3358        if (revisions_eq($branch_from, $r0, $r)) {
3359                unlink $GIT_SVN_INDEX;
3360                print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
3361                sys(qw/git-read-tree/, $parent);
3362                unless (libsvn_can_do_switch()) {
3363                        return libsvn_fetch_full($parent, $paths, $rev,
3364                                                $author, $date, $msg);
3365                }
3366                # do_switch works with svn/trunk >= r22312, but that is not
3367                # included with SVN 1.4.2 (the latest version at the moment),
3368                # so we can't rely on it.
3369                my $ra = libsvn_connect("$url/$branch_from");
3370                my $ed = SVN::Git::Fetcher->new({c => $parent, q => $_q});
3371                my $pool = SVN::Pool->new;
3372                my $reporter = $ra->do_switch($rev, '', 1, $SVN->{url},
3373                                              $ed, $pool);
3374                my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3375                $reporter->set_path('', $r0, 0, @lock, $pool);
3376                $reporter->finish_report($pool);
3377                $pool->clear;
3378                unless ($ed->{git_commit_ok}) {
3379                        die "SVN connection failed somewhere...\n";
3380                }
3381                return libsvn_log_entry($rev, $author, $date, $msg, [$parent]);
3382        }
3383        print STDERR "Nope, branch point not imported or unknown\n";
3384        return undef;
3385}
3386
3387sub libsvn_get_log {
3388        my ($ra, @args) = @_;
3389        $args[4]-- if $args[4] && $_xfer_delta && ! $_follow_parent;
3390        if ($SVN::Core::VERSION le '1.2.0') {
3391                splice(@args, 3, 1);
3392        }
3393        $ra->get_log(@args);
3394}
3395
3396sub libsvn_new_tree {
3397        if (my $log_entry = libsvn_find_parent_branch(@_)) {
3398                return $log_entry;
3399        }
3400        my ($paths, $rev, $author, $date, $msg) = @_;
3401        my $ut;
3402        if ($_xfer_delta) {
3403                my $pool = SVN::Pool->new;
3404                my $ed = SVN::Git::Fetcher->new({q => $_q});
3405                my $reporter = $SVN->do_update($rev, '', 1, $ed, $pool);
3406                my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3407                $reporter->set_path('', $rev, 1, @lock, $pool);
3408                $reporter->finish_report($pool);
3409                $pool->clear;
3410                unless ($ed->{git_commit_ok}) {
3411                        die "SVN connection failed somewhere...\n";
3412                }
3413                $ut = $ed;
3414        } else {
3415                $ut = { empty => {}, dir_prop => {}, file_prop => {} };
3416                open my $gui, '| git-update-index -z --index-info' or croak $!;
3417                libsvn_traverse($gui, '', $SVN->{svn_path}, $rev, undef, $ut);
3418                close $gui or croak $?;
3419        }
3420        libsvn_log_entry($rev, $author, $date, $msg, [], $ut);
3421}
3422
3423sub find_graft_path_commit {
3424        my ($tree_paths, $p1, $r1) = @_;
3425        foreach my $x (keys %$tree_paths) {
3426                next unless ($p1 =~ /^\Q$x\E/);
3427                my $i = $tree_paths->{$x};
3428                my ($r0, $parent) = find_rev_before($r1,$i,1);
3429                return $parent if (defined $r0 && $r0 == $r1);
3430                print STDERR "r$r1 of $i not imported\n";
3431                next;
3432        }
3433        return undef;
3434}
3435
3436sub find_graft_path_parents {
3437        my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
3438        foreach my $x (keys %$tree_paths) {
3439                next unless ($p0 =~ /^\Q$x\E/);
3440                my $i = $tree_paths->{$x};
3441                my ($r, $parent) = find_rev_before($r0, $i, 1);
3442                if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
3443                        my ($url_b, undef, $uuid_b) = cmt_metadata($c);
3444                        my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
3445                        next if ($url_a && $url_b && $url_a eq $url_b &&
3446                                                        $uuid_b eq $uuid_a);
3447                        $grafts->{$c}->{$parent} = 1;
3448                }
3449        }
3450}
3451
3452sub libsvn_graft_file_copies {
3453        my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
3454        foreach (keys %$paths) {
3455                my $i = $paths->{$_};
3456                my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
3457                                        $i->copyfrom_rev);
3458                next unless (defined $p0 && defined $r0);
3459
3460                my $p1 = $_;
3461                $p1 =~ s#^/##;
3462                $p0 =~ s#^/##;
3463                my $c = find_graft_path_commit($tree_paths, $p1, $rev);
3464                next unless $c;
3465                find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
3466        }
3467}
3468
3469sub set_index {
3470        my $old = $ENV{GIT_INDEX_FILE};
3471        $ENV{GIT_INDEX_FILE} = shift;
3472        return $old;
3473}
3474
3475sub restore_index {
3476        my ($old) = @_;
3477        if (defined $old) {
3478                $ENV{GIT_INDEX_FILE} = $old;
3479        } else {
3480                delete $ENV{GIT_INDEX_FILE};
3481        }
3482}
3483
3484sub libsvn_commit_cb {
3485        my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
3486        if ($_optimize_commits && $rev == ($r_last + 1)) {
3487                my $log = libsvn_log_entry($rev,$committer,$date,$msg);
3488                $log->{tree} = get_tree_from_treeish($c);
3489                my $cmt = git_commit($log, $cmt_last, $c);
3490                my @diff = safe_qx('git-diff-tree', $cmt, $c);
3491                if (@diff) {
3492                        print STDERR "Trees differ: $cmt $c\n",
3493                                        join('',@diff),"\n";
3494                        exit 1;
3495                }
3496        } else {
3497                fetch("$rev=$c");
3498        }
3499}
3500
3501sub libsvn_ls_fullurl {
3502        my $fullurl = shift;
3503        my $ra = libsvn_connect($fullurl);
3504        my @ret;
3505        my $pool = SVN::Pool->new;
3506        my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
3507        my ($dirent, undef, undef) = $ra->get_dir('', $r, $pool);
3508        foreach my $d (keys %$dirent) {
3509                if ($dirent->{$d}->kind == $SVN::Node::dir) {
3510                        push @ret, "$d/"; # add '/' for compat with cli svn
3511                }
3512        }
3513        $pool->clear;
3514        return @ret;
3515}
3516
3517
3518sub libsvn_skip_unknown_revs {
3519        my $err = shift;
3520        my $errno = $err->apr_err();
3521        # Maybe the branch we're tracking didn't
3522        # exist when the repo started, so it's
3523        # not an error if it doesn't, just continue
3524        #
3525        # Wonderfully consistent library, eh?
3526        # 160013 - svn:// and file://
3527        # 175002 - http(s)://
3528        # 175007 - http(s):// (this repo required authorization, too...)
3529        #   More codes may be discovered later...
3530        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3531                return;
3532        }
3533        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3534};
3535
3536# Tie::File seems to be prone to offset errors if revisions get sparse,
3537# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3538# one of my favorite modules is out :<  Next up would be one of the DBM
3539# modules, but I'm not sure which is most portable...  So I'll just
3540# go with something that's plain-text, but still capable of
3541# being randomly accessed.  So here's my ultra-simple fixed-width
3542# database.  All records are 40 characters + "\n", so it's easy to seek
3543# to a revision: (41 * rev) is the byte offset.
3544# A record of 40 0s denotes an empty revision.
3545# And yes, it's still pretty fast (faster than Tie::File).
3546sub revdb_set {
3547        my ($file, $rev, $commit) = @_;
3548        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
3549        open my $fh, '+<', $file or croak $!;
3550        my $offset = $rev * 41;
3551        # assume that append is the common case:
3552        seek $fh, 0, 2 or croak $!;
3553        my $pos = tell $fh;
3554        if ($pos < $offset) {
3555                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
3556        }
3557        seek $fh, $offset, 0 or croak $!;
3558        print $fh $commit,"\n";
3559        close $fh or croak $!;
3560}
3561
3562sub revdb_get {
3563        my ($file, $rev) = @_;
3564        my $ret;
3565        my $offset = $rev * 41;
3566        open my $fh, '<', $file or croak $!;
3567        seek $fh, $offset, 0;
3568        if (tell $fh == $offset) {
3569                $ret = readline $fh;
3570                if (defined $ret) {
3571                        chomp $ret;
3572                        $ret = undef if ($ret =~ /^0{40}$/);
3573                }
3574        }
3575        close $fh or croak $!;
3576        return $ret;
3577}
3578
3579sub copy_remote_ref {
3580        my $origin = $_cp_remote ? $_cp_remote : 'origin';
3581        my $ref = "refs/remotes/$GIT_SVN";
3582        if (safe_qx('git-ls-remote', $origin, $ref)) {
3583                sys(qw/git fetch/, $origin, "$ref:$ref");
3584        } elsif ($_cp_remote && !$_upgrade) {
3585                die "Unable to find remote reference: ",
3586                                "refs/remotes/$GIT_SVN on $origin\n";
3587        }
3588}
3589package SVN::Git::Fetcher;
3590use vars qw/@ISA/;
3591use strict;
3592use warnings;
3593use Carp qw/croak/;
3594use IO::File qw//;
3595
3596# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3597sub new {
3598        my ($class, $git_svn) = @_;
3599        my $self = SVN::Delta::Editor->new;
3600        bless $self, $class;
3601        open my $gui, '| git-update-index -z --index-info' or croak $!;
3602        $self->{gui} = $gui;
3603        $self->{c} = $git_svn->{c} if exists $git_svn->{c};
3604        $self->{q} = $git_svn->{q};
3605        $self->{empty} = {};
3606        $self->{dir_prop} = {};
3607        $self->{file_prop} = {};
3608        $self->{absent_dir} = {};
3609        $self->{absent_file} = {};
3610        require Digest::MD5;
3611        $self;
3612}
3613
3614sub open_root {
3615        { path => '' };
3616}
3617
3618sub open_directory {
3619        my ($self, $path, $pb, $rev) = @_;
3620        { path => $path };
3621}
3622
3623sub delete_entry {
3624        my ($self, $path, $rev, $pb) = @_;
3625        my $t = process_rm($self->{gui}, $self->{c}, $path, $self->{q});
3626        $self->{empty}->{$path} = 0 if $t == $SVN::Node::dir;
3627        undef;
3628}
3629
3630sub open_file {
3631        my ($self, $path, $pb, $rev) = @_;
3632        my ($mode, $blob) = (safe_qx('git-ls-tree',$self->{c},'--',$path)
3633                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3634        unless (defined $mode && defined $blob) {
3635                die "$path was not found in commit $self->{c} (r$rev)\n";
3636        }
3637        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3638          pool => SVN::Pool->new, action => 'M' };
3639}
3640
3641sub add_file {
3642        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3643        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3644        delete $self->{empty}->{$dir};
3645        { path => $path, mode_a => 100644, mode_b => 100644,
3646          pool => SVN::Pool->new, action => 'A' };
3647}
3648
3649sub add_directory {
3650        my ($self, $path, $cp_path, $cp_rev) = @_;
3651        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3652        delete $self->{empty}->{$dir};
3653        $self->{empty}->{$path} = 1;
3654        { path => $path };
3655}
3656
3657sub change_dir_prop {
3658        my ($self, $db, $prop, $value) = @_;
3659        $self->{dir_prop}->{$db->{path}} ||= {};
3660        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3661        undef;
3662}
3663
3664sub absent_directory {
3665        my ($self, $path, $pb) = @_;
3666        $self->{absent_dir}->{$pb->{path}} ||= [];
3667        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3668        undef;
3669}
3670
3671sub absent_file {
3672        my ($self, $path, $pb) = @_;
3673        $self->{absent_file}->{$pb->{path}} ||= [];
3674        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3675        undef;
3676}
3677
3678sub change_file_prop {
3679        my ($self, $fb, $prop, $value) = @_;
3680        if ($prop eq 'svn:executable') {
3681                if ($fb->{mode_b} != 120000) {
3682                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3683                }
3684        } elsif ($prop eq 'svn:special') {
3685                $fb->{mode_b} = defined $value ? 120000 : 100644;
3686        } else {
3687                $self->{file_prop}->{$fb->{path}} ||= {};
3688                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3689        }
3690        undef;
3691}
3692
3693sub apply_textdelta {
3694        my ($self, $fb, $exp) = @_;
3695        my $fh = IO::File->new_tmpfile;
3696        $fh->autoflush(1);
3697        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3698        # (but $base does not,) so dup() it for reading in close_file
3699        open my $dup, '<&', $fh or croak $!;
3700        my $base = IO::File->new_tmpfile;
3701        $base->autoflush(1);
3702        if ($fb->{blob}) {
3703                defined (my $pid = fork) or croak $!;
3704                if (!$pid) {
3705                        open STDOUT, '>&', $base or croak $!;
3706                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
3707                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
3708                }
3709                waitpid $pid, 0;
3710                croak $? if $?;
3711
3712                if (defined $exp) {
3713                        seek $base, 0, 0 or croak $!;
3714                        my $md5 = Digest::MD5->new;
3715                        $md5->addfile($base);
3716                        my $got = $md5->hexdigest;
3717                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3718                            "expected: $exp\n",
3719                            "     got: $got\n" if ($got ne $exp);
3720                }
3721        }
3722        seek $base, 0, 0 or croak $!;
3723        $fb->{fh} = $dup;
3724        $fb->{base} = $base;
3725        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
3726}
3727
3728sub close_file {
3729        my ($self, $fb, $exp) = @_;
3730        my $hash;
3731        my $path = $fb->{path};
3732        if (my $fh = $fb->{fh}) {
3733                seek($fh, 0, 0) or croak $!;
3734                my $md5 = Digest::MD5->new;
3735                $md5->addfile($fh);
3736                my $got = $md5->hexdigest;
3737                die "Checksum mismatch: $path\n",
3738                    "expected: $exp\n    got: $got\n" if ($got ne $exp);
3739                seek($fh, 0, 0) or croak $!;
3740                if ($fb->{mode_b} == 120000) {
3741                        read($fh, my $buf, 5) == 5 or croak $!;
3742                        $buf eq 'link ' or die "$path has mode 120000",
3743                                               "but is not a link\n";
3744                }
3745                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
3746                if (!$pid) {
3747                        open STDIN, '<&', $fh or croak $!;
3748                        exec qw/git-hash-object -w --stdin/ or croak $!;
3749                }
3750                chomp($hash = do { local $/; <$out> });
3751                close $out or croak $!;
3752                close $fh or croak $!;
3753                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3754                close $fb->{base} or croak $!;
3755        } else {
3756                $hash = $fb->{blob} or die "no blob information\n";
3757        }
3758        $fb->{pool}->clear;
3759        my $gui = $self->{gui};
3760        print $gui "$fb->{mode_b} $hash\t$path\0" or croak $!;
3761        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
3762        undef;
3763}
3764
3765sub abort_edit {
3766        my $self = shift;
3767        close $self->{gui};
3768        $self->SUPER::abort_edit(@_);
3769}
3770
3771sub close_edit {
3772        my $self = shift;
3773        close $self->{gui} or croak $!;
3774        $self->{git_commit_ok} = 1;
3775        $self->SUPER::close_edit(@_);
3776}
3777
3778package SVN::Git::Editor;
3779use vars qw/@ISA/;
3780use strict;
3781use warnings;
3782use Carp qw/croak/;
3783use IO::File;
3784
3785sub new {
3786        my $class = shift;
3787        my $git_svn = shift;
3788        my $self = SVN::Delta::Editor->new(@_);
3789        bless $self, $class;
3790        foreach (qw/svn_path c r ra /) {
3791                die "$_ required!\n" unless (defined $git_svn->{$_});
3792                $self->{$_} = $git_svn->{$_};
3793        }
3794        $self->{pool} = SVN::Pool->new;
3795        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3796        $self->{rm} = { };
3797        require Digest::MD5;
3798        return $self;
3799}
3800
3801sub split_path {
3802        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3803}
3804
3805sub repo_path {
3806        (defined $_[1] && length $_[1]) ? $_[1] : ''
3807}
3808
3809sub url_path {
3810        my ($self, $path) = @_;
3811        $self->{ra}->{url} . '/' . $self->repo_path($path);
3812}
3813
3814sub rmdirs {
3815        my ($self, $q) = @_;
3816        my $rm = $self->{rm};
3817        delete $rm->{''}; # we never delete the url we're tracking
3818        return unless %$rm;
3819
3820        foreach (keys %$rm) {
3821                my @d = split m#/#, $_;
3822                my $c = shift @d;
3823                $rm->{$c} = 1;
3824                while (@d) {
3825                        $c .= '/' . shift @d;
3826                        $rm->{$c} = 1;
3827                }
3828        }
3829        delete $rm->{$self->{svn_path}};
3830        delete $rm->{''}; # we never delete the url we're tracking
3831        return unless %$rm;
3832
3833        defined(my $pid = open my $fh,'-|') or croak $!;
3834        if (!$pid) {
3835                exec qw/git-ls-tree --name-only -r -z/, $self->{c} or croak $!;
3836        }
3837        local $/ = "\0";
3838        while (<$fh>) {
3839                chomp;
3840                my @dn = split m#/#, $_;
3841                while (pop @dn) {
3842                        delete $rm->{join '/', @dn};
3843                }
3844                unless (%$rm) {
3845                        close $fh;
3846                        return;
3847                }
3848        }
3849        close $fh;
3850
3851        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3852        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3853                $self->close_directory($bat->{$d}, $p);
3854                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3855                print "\tD+\t/$d/\n" unless $q;
3856                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3857                delete $bat->{$d};
3858        }
3859}
3860
3861sub open_or_add_dir {
3862        my ($self, $full_path, $baton) = @_;
3863        my $p = SVN::Pool->new;
3864        my $t = $self->{ra}->check_path($full_path, $self->{r}, $p);
3865        $p->clear;
3866        if ($t == $SVN::Node::none) {
3867                return $self->add_directory($full_path, $baton,
3868                                                undef, -1, $self->{pool});
3869        } elsif ($t == $SVN::Node::dir) {
3870                return $self->open_directory($full_path, $baton,
3871                                                $self->{r}, $self->{pool});
3872        }
3873        print STDERR "$full_path already exists in repository at ",
3874                "r$self->{r} and it is not a directory (",
3875                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3876        exit 1;
3877}
3878
3879sub ensure_path {
3880        my ($self, $path) = @_;
3881        my $bat = $self->{bat};
3882        $path = $self->repo_path($path);
3883        return $bat->{''} unless (length $path);
3884        my @p = split m#/+#, $path;
3885        my $c = shift @p;
3886        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3887        while (@p) {
3888                my $c0 = $c;
3889                $c .= '/' . shift @p;
3890                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3891        }
3892        return $bat->{$c};
3893}
3894
3895sub A {
3896        my ($self, $m, $q) = @_;
3897        my ($dir, $file) = split_path($m->{file_b});
3898        my $pbat = $self->ensure_path($dir);
3899        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3900                                        undef, -1);
3901        print "\tA\t$m->{file_b}\n" unless $q;
3902        $self->chg_file($fbat, $m);
3903        $self->close_file($fbat,undef,$self->{pool});
3904}
3905
3906sub C {
3907        my ($self, $m, $q) = @_;
3908        my ($dir, $file) = split_path($m->{file_b});
3909        my $pbat = $self->ensure_path($dir);
3910        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3911                                $self->url_path($m->{file_a}), $self->{r});
3912        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $q;
3913        $self->chg_file($fbat, $m);
3914        $self->close_file($fbat,undef,$self->{pool});
3915}
3916
3917sub delete_entry {
3918        my ($self, $path, $pbat) = @_;
3919        my $rpath = $self->repo_path($path);
3920        my ($dir, $file) = split_path($rpath);
3921        $self->{rm}->{$dir} = 1;
3922        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3923}
3924
3925sub R {
3926        my ($self, $m, $q) = @_;
3927        my ($dir, $file) = split_path($m->{file_b});
3928        my $pbat = $self->ensure_path($dir);
3929        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3930                                $self->url_path($m->{file_a}), $self->{r});
3931        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $q;
3932        $self->chg_file($fbat, $m);
3933        $self->close_file($fbat,undef,$self->{pool});
3934
3935        ($dir, $file) = split_path($m->{file_a});
3936        $pbat = $self->ensure_path($dir);
3937        $self->delete_entry($m->{file_a}, $pbat);
3938}
3939
3940sub M {
3941        my ($self, $m, $q) = @_;
3942        my ($dir, $file) = split_path($m->{file_b});
3943        my $pbat = $self->ensure_path($dir);
3944        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3945                                $pbat,$self->{r},$self->{pool});
3946        print "\t$m->{chg}\t$m->{file_b}\n" unless $q;
3947        $self->chg_file($fbat, $m);
3948        $self->close_file($fbat,undef,$self->{pool});
3949}
3950
3951sub T { shift->M(@_) }
3952
3953sub change_file_prop {
3954        my ($self, $fbat, $pname, $pval) = @_;
3955        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3956}
3957
3958sub chg_file {
3959        my ($self, $fbat, $m) = @_;
3960        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3961                $self->change_file_prop($fbat,'svn:executable','*');
3962        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3963                $self->change_file_prop($fbat,'svn:executable',undef);
3964        }
3965        my $fh = IO::File->new_tmpfile or croak $!;
3966        if ($m->{mode_b} =~ /^120/) {
3967                print $fh 'link ' or croak $!;
3968                $self->change_file_prop($fbat,'svn:special','*');
3969        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3970                $self->change_file_prop($fbat,'svn:special',undef);
3971        }
3972        defined(my $pid = fork) or croak $!;
3973        if (!$pid) {
3974                open STDOUT, '>&', $fh or croak $!;
3975                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3976        }
3977        waitpid $pid, 0;
3978        croak $? if $?;
3979        $fh->flush == 0 or croak $!;
3980        seek $fh, 0, 0 or croak $!;
3981
3982        my $md5 = Digest::MD5->new;
3983        $md5->addfile($fh) or croak $!;
3984        seek $fh, 0, 0 or croak $!;
3985
3986        my $exp = $md5->hexdigest;
3987        my $pool = SVN::Pool->new;
3988        my $atd = $self->apply_textdelta($fbat, undef, $pool);
3989        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3990        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3991        $pool->clear;
3992
3993        close $fh or croak $!;
3994}
3995
3996sub D {
3997        my ($self, $m, $q) = @_;
3998        my ($dir, $file) = split_path($m->{file_b});
3999        my $pbat = $self->ensure_path($dir);
4000        print "\tD\t$m->{file_b}\n" unless $q;
4001        $self->delete_entry($m->{file_b}, $pbat);
4002}
4003
4004sub close_edit {
4005        my ($self) = @_;
4006        my ($p,$bat) = ($self->{pool}, $self->{bat});
4007        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4008                $self->close_directory($bat->{$_}, $p);
4009        }
4010        $self->SUPER::close_edit($p);
4011        $p->clear;
4012}
4013
4014sub abort_edit {
4015        my ($self) = @_;
4016        $self->SUPER::abort_edit($self->{pool});
4017        $self->{pool}->clear;
4018}
4019
4020__END__
4021
4022Data structures:
4023
4024$svn_log hashref (as returned by svn_log_raw)
4025{
4026        fh => file handle of the log file,
4027        state => state of the log file parser (sep/msg/rev/msg_start...)
4028}
4029
4030$log_msg hashref as returned by next_log_entry($svn_log)
4031{
4032        msg => 'whitespace-formatted log entry
4033',                                              # trailing newline is preserved
4034        revision => '8',                        # integer
4035        date => '2004-02-24T17:01:44.108345Z',  # commit date
4036        author => 'committer name'
4037};
4038
4039
4040@mods = array of diff-index line hashes, each element represents one line
4041        of diff-index output
4042
4043diff-index line ($m hash)
4044{
4045        mode_a => first column of diff-index output, no leading ':',
4046        mode_b => second column of diff-index output,
4047        sha1_b => sha1sum of the final blob,
4048        chg => change type [MCRADT],
4049        file_a => original file name of a file (iff chg is 'C' or 'R')
4050        file_b => new/current file name of a file (any chg)
4051}
4052;
4053
4054# retval of read_url_paths{,_all}();
4055$l_map = {
4056        # repository root url
4057        'https://svn.musicpd.org' => {
4058                # repository path               # GIT_SVN_ID
4059                'mpd/trunk'             =>      'trunk',
4060                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
4061        },
4062}
4063
4064Notes:
4065        I don't trust the each() function on unless I created %hash myself
4066        because the internal iterator may not have started at base.