git-svn.perlon commit Merge branch 'np/addcommit' (78ba004)
   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 $dc;
 929        chomp($dc = `git-repo-config --get diff.color`);
 930        if ($dc eq 'auto') {
 931                if (-t *STDOUT || (defined $_pager &&
 932                    `git-repo-config --bool --get pager.color` !~ /^false/)) {
 933                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
 934                }
 935                return 0;
 936        }
 937        return 0 if $dc eq 'never';
 938        return 1 if $dc eq 'always';
 939        chomp($dc = `git-repo-config --bool --get diff.color`);
 940        $dc eq 'true';
 941}
 942
 943sub git_svn_log_cmd {
 944        my ($r_min, $r_max) = @_;
 945        my @cmd = (qw/git-log --abbrev-commit --pretty=raw
 946                        --default/, "refs/remotes/$GIT_SVN");
 947        push @cmd, '-r' unless $_non_recursive;
 948        push @cmd, qw/--raw --name-status/ if $_verbose;
 949        push @cmd, '--color' if log_use_color();
 950        return @cmd unless defined $r_max;
 951        if ($r_max == $r_min) {
 952                push @cmd, '--max-count=1';
 953                if (my $c = revdb_get($REVDB, $r_max)) {
 954                        push @cmd, $c;
 955                }
 956        } else {
 957                my ($c_min, $c_max);
 958                $c_max = revdb_get($REVDB, $r_max);
 959                $c_min = revdb_get($REVDB, $r_min);
 960                if (defined $c_min && defined $c_max) {
 961                        if ($r_max > $r_max) {
 962                                push @cmd, "$c_min..$c_max";
 963                        } else {
 964                                push @cmd, "$c_max..$c_min";
 965                        }
 966                } elsif ($r_max > $r_min) {
 967                        push @cmd, $c_max;
 968                } else {
 969                        push @cmd, $c_min;
 970                }
 971        }
 972        return @cmd;
 973}
 974
 975sub fetch_child_id {
 976        my $id = shift;
 977        print "Fetching $id\n";
 978        my $ref = "$GIT_DIR/refs/remotes/$id";
 979        defined(my $pid = open my $fh, '-|') or croak $!;
 980        if (!$pid) {
 981                $_repack = undef;
 982                $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 983                init_vars();
 984                fetch(@_);
 985                exit 0;
 986        }
 987        while (<$fh>) {
 988                print $_;
 989                check_repack() if (/^r\d+ = $sha1/);
 990        }
 991        close $fh or croak $?;
 992}
 993
 994sub rec_fetch {
 995        my ($pfx, $p, @args) = @_;
 996        my @dir;
 997        foreach (sort <$p/*>) {
 998                if (-r "$_/info/url") {
 999                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
1000                        my $id = $pfx . basename $_;
1001                        next if $id eq 'trunk';
1002                        fetch_child_id($id, @args);
1003                } elsif (-d $_) {
1004                        push @dir, $_;
1005                }
1006        }
1007        foreach (@dir) {
1008                my $x = $_;
1009                $x =~ s!^\Q$GIT_DIR\E/svn/!!;
1010                rec_fetch($x, $_);
1011        }
1012}
1013
1014sub complete_url_ls_init {
1015        my ($url, $var, $switch, $pfx) = @_;
1016        unless ($var) {
1017                print STDERR "W: $switch not specified\n";
1018                return;
1019        }
1020        $var =~ s#/+$##;
1021        if ($var !~ m#^[a-z\+]+://#) {
1022                $var = '/' . $var if ($var !~ m#^/#);
1023                unless ($url) {
1024                        print STDERR "E: '$var' is not a complete URL ",
1025                                "and a separate URL is not specified\n";
1026                        exit 1;
1027                }
1028                $var = $url . $var;
1029        }
1030        chomp(my @ls = $_use_lib ? libsvn_ls_fullurl($var)
1031                                : safe_qx(qw/svn ls --non-interactive/, $var));
1032        my $old = $GIT_SVN;
1033        defined(my $pid = fork) or croak $!;
1034        if (!$pid) {
1035                foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
1036                        $u =~ s#/+$##;
1037                        if ($u !~ m!\Q$var\E/(.+)$!) {
1038                                print STDERR "W: Unrecognized URL: $u\n";
1039                                die "This should never happen\n";
1040                        }
1041                        # don't try to init already existing refs
1042                        my $id = $pfx.$1;
1043                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
1044                        init_vars();
1045                        unless (-d $GIT_SVN_DIR) {
1046                                print "init $u => $id\n";
1047                                init($u);
1048                        }
1049                }
1050                exit 0;
1051        }
1052        waitpid $pid, 0;
1053        croak $? if $?;
1054        my ($n) = ($switch =~ /^--(\w+)/);
1055        sys('git-repo-config', "svn.$n", $var);
1056}
1057
1058sub common_prefix {
1059        my $paths = shift;
1060        my %common;
1061        foreach (@$paths) {
1062                my @tmp = split m#/#, $_;
1063                my $p = '';
1064                while (my $x = shift @tmp) {
1065                        $p .= "/$x";
1066                        $common{$p} ||= 0;
1067                        $common{$p}++;
1068                }
1069        }
1070        foreach (sort {length $b <=> length $a} keys %common) {
1071                if ($common{$_} == @$paths) {
1072                        return $_;
1073                }
1074        }
1075        return '';
1076}
1077
1078# grafts set here are 'stronger' in that they're based on actual tree
1079# matches, and won't be deleted from merge-base checking in write_grafts()
1080sub graft_tree_joins {
1081        my $grafts = shift;
1082        map_tree_joins() if (@_branch_from && !%tree_map);
1083        return unless %tree_map;
1084
1085        git_svn_each(sub {
1086                my $i = shift;
1087                defined(my $pid = open my $fh, '-|') or croak $!;
1088                if (!$pid) {
1089                        exec qw/git-rev-list --pretty=raw/,
1090                                        "refs/remotes/$i" or croak $!;
1091                }
1092                while (<$fh>) {
1093                        next unless /^commit ($sha1)$/o;
1094                        my $c = $1;
1095                        my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
1096                        next unless $tree_map{$t};
1097
1098                        my $l;
1099                        do {
1100                                $l = readline $fh;
1101                        } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
1102
1103                        my ($s, $tz) = ($1, $2);
1104                        if ($tz =~ s/^\+//) {
1105                                $s += tz_to_s_offset($tz);
1106                        } elsif ($tz =~ s/^\-//) {
1107                                $s -= tz_to_s_offset($tz);
1108                        }
1109
1110                        my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
1111
1112                        foreach my $p (@{$tree_map{$t}}) {
1113                                next if $p eq $c;
1114                                my $mb = eval {
1115                                        safe_qx('git-merge-base', $c, $p)
1116                                };
1117                                next unless ($@ || $?);
1118                                if (defined $r_a) {
1119                                        # see if SVN says it's a relative
1120                                        my ($url_b, $r_b, $uuid_b) =
1121                                                        cmt_metadata($p);
1122                                        next if (defined $url_b &&
1123                                                        defined $url_a &&
1124                                                        ($url_a eq $url_b) &&
1125                                                        ($uuid_a eq $uuid_b));
1126                                        if ($uuid_a eq $uuid_b) {
1127                                                if ($r_b < $r_a) {
1128                                                        $grafts->{$c}->{$p} = 2;
1129                                                        next;
1130                                                } elsif ($r_b > $r_a) {
1131                                                        $grafts->{$p}->{$c} = 2;
1132                                                        next;
1133                                                }
1134                                        }
1135                                }
1136                                my $ct = get_commit_time($p);
1137                                if ($ct < $s) {
1138                                        $grafts->{$c}->{$p} = 2;
1139                                } elsif ($ct > $s) {
1140                                        $grafts->{$p}->{$c} = 2;
1141                                }
1142                                # what should we do when $ct == $s ?
1143                        }
1144                }
1145                close $fh or croak $?;
1146        });
1147}
1148
1149# this isn't funky-filename safe, but good enough for now...
1150sub graft_file_copy_cmd {
1151        my ($grafts, $l_map, $u) = @_;
1152        my $paths = $l_map->{$u};
1153        my $pfx = common_prefix([keys %$paths]);
1154        $SVN_URL ||= $u.$pfx;
1155        my $pid = open my $fh, '-|';
1156        defined $pid or croak $!;
1157        unless ($pid) {
1158                my @exec = qw/svn log -v/;
1159                push @exec, "-r$_revision" if defined $_revision;
1160                exec @exec, $u.$pfx or croak $!;
1161        }
1162        my ($r, $mp) = (undef, undef);
1163        while (<$fh>) {
1164                chomp;
1165                if (/^\-{72}$/) {
1166                        $mp = $r = undef;
1167                } elsif (/^r(\d+) \| /) {
1168                        $r = $1 unless defined $r;
1169                } elsif (/^Changed paths:/) {
1170                        $mp = 1;
1171                } elsif ($mp && m#^   [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) {
1172                        my ($p1, $p0, $r0) = ($1, $2, $3);
1173                        my $c = find_graft_path_commit($paths, $p1, $r);
1174                        next unless $c;
1175                        find_graft_path_parents($grafts, $paths, $c, $p0, $r0);
1176                }
1177        }
1178}
1179
1180sub graft_file_copy_lib {
1181        my ($grafts, $l_map, $u) = @_;
1182        my $tree_paths = $l_map->{$u};
1183        my $pfx = common_prefix([keys %$tree_paths]);
1184        my ($repo, $path) = repo_path_split($u.$pfx);
1185        $SVN = libsvn_connect($repo);
1186
1187        my ($base, $head) = libsvn_parse_revision();
1188        my $inc = 1000;
1189        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
1190        my $eh = $SVN::Error::handler;
1191        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
1192        while (1) {
1193                my $pool = SVN::Pool->new;
1194                libsvn_get_log(libsvn_dup_ra($SVN), [$path],
1195                               $min, $max, 0, 2, 1,
1196                        sub {
1197                                libsvn_graft_file_copies($grafts, $tree_paths,
1198                                                        $path, @_);
1199                        }, $pool);
1200                $pool->clear;
1201                last if ($max >= $head);
1202                $min = $max + 1;
1203                $max += $inc;
1204                $max = $head if ($max > $head);
1205        }
1206        $SVN::Error::handler = $eh;
1207}
1208
1209sub process_merge_msg_matches {
1210        my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
1211        my (@strong, @weak);
1212        foreach (@matches) {
1213                # merging with ourselves is not interesting
1214                next if $_ eq $p;
1215                if ($l_map->{$u}->{$_}) {
1216                        push @strong, $_;
1217                } else {
1218                        push @weak, $_;
1219                }
1220        }
1221        foreach my $w (@weak) {
1222                last if @strong;
1223                # no exact match, use branch name as regexp.
1224                my $re = qr/\Q$w\E/i;
1225                foreach (keys %{$l_map->{$u}}) {
1226                        if (/$re/) {
1227                                push @strong, $l_map->{$u}->{$_};
1228                                last;
1229                        }
1230                }
1231                last if @strong;
1232                $w = basename($w);
1233                $re = qr/\Q$w\E/i;
1234                foreach (keys %{$l_map->{$u}}) {
1235                        if (/$re/) {
1236                                push @strong, $l_map->{$u}->{$_};
1237                                last;
1238                        }
1239                }
1240        }
1241        my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
1242                                        \s(?:[a-f\d\-]+)$/xsm);
1243        unless (defined $rev) {
1244                ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
1245                                        \@(?:[a-f\d\-]+)/xsm);
1246                return unless defined $rev;
1247        }
1248        foreach my $m (@strong) {
1249                my ($r0, $s0) = find_rev_before($rev, $m, 1);
1250                $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
1251        }
1252}
1253
1254sub graft_merge_msg {
1255        my ($grafts, $l_map, $u, $p, @re) = @_;
1256
1257        my $x = $l_map->{$u}->{$p};
1258        my $rl = rev_list_raw($x);
1259        while (my $c = next_rev_list_entry($rl)) {
1260                foreach my $re (@re) {
1261                        my (@br) = ($c->{m} =~ /$re/g);
1262                        next unless @br;
1263                        process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
1264                }
1265        }
1266}
1267
1268sub read_uuid {
1269        return if $SVN_UUID;
1270        if ($_use_lib) {
1271                my $pool = SVN::Pool->new;
1272                $SVN_UUID = $SVN->get_uuid($pool);
1273                $pool->clear;
1274        } else {
1275                my $info = shift || svn_info('.');
1276                $SVN_UUID = $info->{'Repository UUID'} or
1277                                        croak "Repository UUID unreadable\n";
1278        }
1279}
1280
1281sub quiet_run {
1282        my $pid = fork;
1283        defined $pid or croak $!;
1284        if (!$pid) {
1285                open my $null, '>', '/dev/null' or croak $!;
1286                open STDERR, '>&', $null or croak $!;
1287                open STDOUT, '>&', $null or croak $!;
1288                exec @_ or croak $!;
1289        }
1290        waitpid $pid, 0;
1291        return $?;
1292}
1293
1294sub repo_path_split {
1295        my $full_url = shift;
1296        $full_url =~ s#/+$##;
1297
1298        foreach (@repo_path_split_cache) {
1299                if ($full_url =~ s#$_##) {
1300                        my $u = $1;
1301                        $full_url =~ s#^/+##;
1302                        return ($u, $full_url);
1303                }
1304        }
1305        if ($_use_lib) {
1306                my $tmp = libsvn_connect($full_url);
1307                return ($tmp->{repos_root}, $tmp->{svn_path});
1308        } else {
1309                my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i);
1310                $path =~ s#^/+##;
1311                my @paths = split(m#/+#, $path);
1312                while (quiet_run(qw/svn ls --non-interactive/, $url)) {
1313                        my $n = shift @paths || last;
1314                        $url .= "/$n";
1315                }
1316                push @repo_path_split_cache, qr/^(\Q$url\E)/;
1317                $path = join('/',@paths);
1318                return ($url, $path);
1319        }
1320}
1321
1322sub setup_git_svn {
1323        defined $SVN_URL or croak "SVN repository location required\n";
1324        unless (-d $GIT_DIR) {
1325                croak "GIT_DIR=$GIT_DIR does not exist!\n";
1326        }
1327        mkpath([$GIT_SVN_DIR]);
1328        mkpath(["$GIT_SVN_DIR/info"]);
1329        open my $fh, '>>',$REVDB or croak $!;
1330        close $fh;
1331        s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1332
1333}
1334
1335sub assert_svn_wc_clean {
1336        return if $_use_lib;
1337        my ($svn_rev) = @_;
1338        croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
1339        my $lcr = svn_info('.')->{'Last Changed Rev'};
1340        if ($svn_rev != $lcr) {
1341                print STDERR "Checking for copy-tree ... ";
1342                my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
1343                                                "-r$lcr:$svn_rev")));
1344                if (@diff) {
1345                        croak "Nope!  Expected r$svn_rev, got r$lcr\n";
1346                } else {
1347                        print STDERR "OK!\n";
1348                }
1349        }
1350        my @status = grep(!/^Performing status on external/,(`svn status`));
1351        @status = grep(!/^\s*$/,@status);
1352        @status = grep(!/^X/,@status) if $_no_ignore_ext;
1353        if (scalar @status) {
1354                print STDERR "Tree ($SVN_WC) is not clean:\n";
1355                print STDERR $_ foreach @status;
1356                croak;
1357        }
1358}
1359
1360sub get_tree_from_treeish {
1361        my ($treeish) = @_;
1362        croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1363        chomp(my $type = `git-cat-file -t $treeish`);
1364        my $expected;
1365        while ($type eq 'tag') {
1366                chomp(($treeish, $type) = `git-cat-file tag $treeish`);
1367        }
1368        if ($type eq 'commit') {
1369                $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
1370                ($expected) = ($expected =~ /^tree ($sha1)$/);
1371                die "Unable to get tree from $treeish\n" unless $expected;
1372        } elsif ($type eq 'tree') {
1373                $expected = $treeish;
1374        } else {
1375                die "$treeish is a $type, expected tree, tag or commit\n";
1376        }
1377        return $expected;
1378}
1379
1380sub assert_tree {
1381        return if $_use_lib;
1382        my ($treeish) = @_;
1383        my $expected = get_tree_from_treeish($treeish);
1384
1385        my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
1386        if (-e $tmpindex) {
1387                unlink $tmpindex or croak $!;
1388        }
1389        my $old_index = set_index($tmpindex);
1390        index_changes(1);
1391        chomp(my $tree = `git-write-tree`);
1392        restore_index($old_index);
1393        if ($tree ne $expected) {
1394                croak "Tree mismatch, Got: $tree, Expected: $expected\n";
1395        }
1396        unlink $tmpindex;
1397}
1398
1399sub parse_diff_tree {
1400        my $diff_fh = shift;
1401        local $/ = "\0";
1402        my $state = 'meta';
1403        my @mods;
1404        while (<$diff_fh>) {
1405                chomp $_; # this gets rid of the trailing "\0"
1406                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1407                                        $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1408                        push @mods, {   mode_a => $1, mode_b => $2,
1409                                        sha1_b => $3, chg => $4 };
1410                        if ($4 =~ /^(?:C|R)$/) {
1411                                $state = 'file_a';
1412                        } else {
1413                                $state = 'file_b';
1414                        }
1415                } elsif ($state eq 'file_a') {
1416                        my $x = $mods[$#mods] or croak "Empty array\n";
1417                        if ($x->{chg} !~ /^(?:C|R)$/) {
1418                                croak "Error parsing $_, $x->{chg}\n";
1419                        }
1420                        $x->{file_a} = $_;
1421                        $state = 'file_b';
1422                } elsif ($state eq 'file_b') {
1423                        my $x = $mods[$#mods] or croak "Empty array\n";
1424                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1425                                croak "Error parsing $_, $x->{chg}\n";
1426                        }
1427                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1428                                croak "Error parsing $_, $x->{chg}\n";
1429                        }
1430                        $x->{file_b} = $_;
1431                        $state = 'meta';
1432                } else {
1433                        croak "Error parsing $_\n";
1434                }
1435        }
1436        close $diff_fh or croak $?;
1437
1438        return \@mods;
1439}
1440
1441sub svn_check_prop_executable {
1442        my $m = shift;
1443        return if -l $m->{file_b};
1444        if ($m->{mode_b} =~ /755$/) {
1445                chmod((0755 &~ umask),$m->{file_b}) or croak $!;
1446                if ($m->{mode_a} !~ /755$/) {
1447                        sys(qw(svn propset svn:executable 1), $m->{file_b});
1448                }
1449                -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
1450        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
1451                sys(qw(svn propdel svn:executable), $m->{file_b});
1452                chmod((0644 &~ umask),$m->{file_b}) or croak $!;
1453                -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
1454        }
1455}
1456
1457sub svn_ensure_parent_path {
1458        my $dir_b = dirname(shift);
1459        svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
1460        mkpath([$dir_b]) unless (-d $dir_b);
1461        sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
1462}
1463
1464sub precommit_check {
1465        my $mods = shift;
1466        my (%rm_file, %rmdir_check, %added_check);
1467
1468        my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
1469        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1470                if ($m->{chg} eq 'R') {
1471                        if (-d $m->{file_b}) {
1472                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1473                        }
1474                        # dir/$file => dir/file/$file
1475                        my $dirname = dirname($m->{file_b});
1476                        while ($dirname ne File::Spec->curdir) {
1477                                if ($dirname ne $m->{file_a}) {
1478                                        $dirname = dirname($dirname);
1479                                        next;
1480                                }
1481                                err_file_to_dir("$m->{file_a} => $m->{file_b}");
1482                        }
1483                        # baz/zzz => baz (baz is a file)
1484                        $dirname = dirname($m->{file_a});
1485                        while ($dirname ne File::Spec->curdir) {
1486                                if ($dirname ne $m->{file_b}) {
1487                                        $dirname = dirname($dirname);
1488                                        next;
1489                                }
1490                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1491                        }
1492                }
1493                if ($m->{chg} =~ /^(D|R)$/) {
1494                        my $t = $1 eq 'D' ? 'file_b' : 'file_a';
1495                        $rm_file{ $m->{$t} } = 1;
1496                        my $dirname = dirname( $m->{$t} );
1497                        my $basename = basename( $m->{$t} );
1498                        $rmdir_check{$dirname}->{$basename} = 1;
1499                } elsif ($m->{chg} =~ /^(?:A|C)$/) {
1500                        if (-d $m->{file_b}) {
1501                                err_dir_to_file($m->{file_b});
1502                        }
1503                        my $dirname = dirname( $m->{file_b} );
1504                        my $basename = basename( $m->{file_b} );
1505                        $added_check{$dirname}->{$basename} = 1;
1506                        while ($dirname ne File::Spec->curdir) {
1507                                if ($rm_file{$dirname}) {
1508                                        err_file_to_dir($m->{file_b});
1509                                }
1510                                $dirname = dirname $dirname;
1511                        }
1512                }
1513        }
1514        return (\%rmdir_check, \%added_check);
1515
1516        sub err_dir_to_file {
1517                my $file = shift;
1518                print STDERR "Node change from directory to file ",
1519                                "is not supported by Subversion: ",$file,"\n";
1520                exit 1;
1521        }
1522        sub err_file_to_dir {
1523                my $file = shift;
1524                print STDERR "Node change from file to directory ",
1525                                "is not supported by Subversion: ",$file,"\n";
1526                exit 1;
1527        }
1528}
1529
1530
1531sub get_diff {
1532        my ($from, $treeish) = @_;
1533        assert_tree($from);
1534        print "diff-tree $from $treeish\n";
1535        my $pid = open my $diff_fh, '-|';
1536        defined $pid or croak $!;
1537        if ($pid == 0) {
1538                my @diff_tree = qw(git-diff-tree -z -r);
1539                if ($_cp_similarity) {
1540                        push @diff_tree, "-C$_cp_similarity";
1541                } else {
1542                        push @diff_tree, '-C';
1543                }
1544                push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1545                push @diff_tree, "-l$_l" if defined $_l;
1546                exec(@diff_tree, $from, $treeish) or croak $!;
1547        }
1548        return parse_diff_tree($diff_fh);
1549}
1550
1551sub svn_checkout_tree {
1552        my ($from, $treeish) = @_;
1553        my $mods = get_diff($from->{commit}, $treeish);
1554        return $mods unless (scalar @$mods);
1555        my ($rm, $add) = precommit_check($mods);
1556
1557        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1558        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1559                if ($m->{chg} eq 'C') {
1560                        svn_ensure_parent_path( $m->{file_b} );
1561                        sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
1562                        apply_mod_line_blob($m);
1563                        svn_check_prop_executable($m);
1564                } elsif ($m->{chg} eq 'D') {
1565                        sys(qw(svn rm --force), $m->{file_b});
1566                } elsif ($m->{chg} eq 'R') {
1567                        svn_ensure_parent_path( $m->{file_b} );
1568                        sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
1569                        apply_mod_line_blob($m);
1570                        svn_check_prop_executable($m);
1571                } elsif ($m->{chg} eq 'M') {
1572                        apply_mod_line_blob($m);
1573                        svn_check_prop_executable($m);
1574                } elsif ($m->{chg} eq 'T') {
1575                        svn_check_prop_executable($m);
1576                        apply_mod_line_blob($m);
1577                        if ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
1578                                sys(qw(svn propdel svn:special), $m->{file_b});
1579                        } else {
1580                                sys(qw(svn propset svn:special *),$m->{file_b});
1581                        }
1582                } elsif ($m->{chg} eq 'A') {
1583                        svn_ensure_parent_path( $m->{file_b} );
1584                        apply_mod_line_blob($m);
1585                        sys(qw(svn add), $m->{file_b});
1586                        svn_check_prop_executable($m);
1587                } else {
1588                        croak "Invalid chg: $m->{chg}\n";
1589                }
1590        }
1591
1592        assert_tree($treeish);
1593        if ($_rmdir) { # remove empty directories
1594                handle_rmdir($rm, $add);
1595        }
1596        assert_tree($treeish);
1597        return $mods;
1598}
1599
1600sub libsvn_checkout_tree {
1601        my ($from, $treeish, $ed) = @_;
1602        my $mods = get_diff($from, $treeish);
1603        return $mods unless (scalar @$mods);
1604        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1605        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1606                my $f = $m->{chg};
1607                if (defined $o{$f}) {
1608                        $ed->$f($m, $_q);
1609                } else {
1610                        croak "Invalid change type: $f\n";
1611                }
1612        }
1613        $ed->rmdirs($_q) if $_rmdir;
1614        return $mods;
1615}
1616
1617# svn ls doesn't work with respect to the current working tree, but what's
1618# in the repository.  There's not even an option for it... *sigh*
1619# (added files don't show up and removed files remain in the ls listing)
1620sub svn_ls_current {
1621        my ($dir, $rm, $add) = @_;
1622        chomp(my @ls = safe_qx('svn','ls',$dir));
1623        my @ret = ();
1624        foreach (@ls) {
1625                s#/$##; # trailing slashes are evil
1626                push @ret, $_ unless $rm->{$dir}->{$_};
1627        }
1628        if (exists $add->{$dir}) {
1629                push @ret, keys %{$add->{$dir}};
1630        }
1631        return \@ret;
1632}
1633
1634sub handle_rmdir {
1635        my ($rm, $add) = @_;
1636
1637        foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
1638                my $ls = svn_ls_current($dir, $rm, $add);
1639                next if (scalar @$ls);
1640                sys(qw(svn rm --force),$dir);
1641
1642                my $dn = dirname $dir;
1643                $rm->{ $dn }->{ basename $dir } = 1;
1644                $ls = svn_ls_current($dn, $rm, $add);
1645                while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
1646                        sys(qw(svn rm --force),$dn);
1647                        $dir = basename $dn;
1648                        $dn = dirname $dn;
1649                        $rm->{ $dn }->{ $dir } = 1;
1650                        $ls = svn_ls_current($dn, $rm, $add);
1651                }
1652        }
1653}
1654
1655sub get_commit_message {
1656        my ($commit, $commit_msg) = (@_);
1657        my %log_msg = ( msg => '' );
1658        open my $msg, '>', $commit_msg or croak $!;
1659
1660        chomp(my $type = `git-cat-file -t $commit`);
1661        if ($type eq 'commit' || $type eq 'tag') {
1662                my $pid = open my $msg_fh, '-|';
1663                defined $pid or croak $!;
1664
1665                if ($pid == 0) {
1666                        exec('git-cat-file', $type, $commit) or croak $!;
1667                }
1668                my $in_msg = 0;
1669                while (<$msg_fh>) {
1670                        if (!$in_msg) {
1671                                $in_msg = 1 if (/^\s*$/);
1672                        } elsif (/^git-svn-id: /) {
1673                                # skip this, we regenerate the correct one
1674                                # on re-fetch anyways
1675                        } else {
1676                                print $msg $_ or croak $!;
1677                        }
1678                }
1679                close $msg_fh or croak $?;
1680        }
1681        close $msg or croak $!;
1682
1683        if ($_edit || ($type eq 'tree')) {
1684                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1685                system($editor, $commit_msg);
1686        }
1687
1688        # file_to_s removes all trailing newlines, so just use chomp() here:
1689        open $msg, '<', $commit_msg or croak $!;
1690        { local $/; chomp($log_msg{msg} = <$msg>); }
1691        close $msg or croak $!;
1692
1693        return \%log_msg;
1694}
1695
1696sub set_svn_commit_env {
1697        if (defined $LC_ALL) {
1698                $ENV{LC_ALL} = $LC_ALL;
1699        } else {
1700                delete $ENV{LC_ALL};
1701        }
1702}
1703
1704sub svn_commit_tree {
1705        my ($last, $commit) = @_;
1706        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
1707        my $log_msg = get_commit_message($commit, $commit_msg);
1708        my ($oneline) = ($log_msg->{msg} =~ /([^\n\r]+)/);
1709        print "Committing $commit: $oneline\n";
1710
1711        set_svn_commit_env();
1712        my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
1713        $ENV{LC_ALL} = 'C';
1714        unlink $commit_msg;
1715        my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/);
1716        if (!defined $committed) {
1717                my $out = join("\n",@ci_output);
1718                print STDERR "W: Trouble parsing \`svn commit' output:\n\n",
1719                                $out, "\n\nAssuming English locale...";
1720                ($committed) = ($out =~ /^Committed revision \d+\./sm);
1721                defined $committed or die " FAILED!\n",
1722                        "Commit output failed to parse committed revision!\n",
1723                print STDERR " OK\n";
1724        }
1725
1726        my @svn_up = qw(svn up);
1727        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
1728        if ($_optimize_commits && ($committed == ($last->{revision} + 1))) {
1729                push @svn_up, "-r$committed";
1730                sys(@svn_up);
1731                my $info = svn_info('.');
1732                my $date = $info->{'Last Changed Date'} or die "Missing date\n";
1733                if ($info->{'Last Changed Rev'} != $committed) {
1734                        croak "$info->{'Last Changed Rev'} != $committed\n"
1735                }
1736                my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1737                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1738                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1739                                         or croak "Failed to parse date: $date\n";
1740                $log_msg->{date} = "$tz $Y-$m-$d $H:$M:$S";
1741                $log_msg->{author} = $info->{'Last Changed Author'};
1742                $log_msg->{revision} = $committed;
1743                $log_msg->{msg} .= "\n";
1744                $log_msg->{parents} = [ $last->{commit} ];
1745                $log_msg->{commit} = git_commit($log_msg, $commit);
1746                return $log_msg;
1747        }
1748        # resync immediately
1749        push @svn_up, "-r$last->{revision}";
1750        sys(@svn_up);
1751        return fetch("$committed=$commit");
1752}
1753
1754sub rev_list_raw {
1755        my (@args) = @_;
1756        my $pid = open my $fh, '-|';
1757        defined $pid or croak $!;
1758        if (!$pid) {
1759                exec(qw/git-rev-list --pretty=raw/, @args) or croak $!;
1760        }
1761        return { fh => $fh, t => { } };
1762}
1763
1764sub next_rev_list_entry {
1765        my $rl = shift;
1766        my $fh = $rl->{fh};
1767        my $x = $rl->{t};
1768        while (<$fh>) {
1769                if (/^commit ($sha1)$/o) {
1770                        if ($x->{c}) {
1771                                $rl->{t} = { c => $1 };
1772                                return $x;
1773                        } else {
1774                                $x->{c} = $1;
1775                        }
1776                } elsif (/^parent ($sha1)$/o) {
1777                        $x->{p}->{$1} = 1;
1778                } elsif (s/^    //) {
1779                        $x->{m} ||= '';
1780                        $x->{m} .= $_;
1781                }
1782        }
1783        return ($x != $rl->{t}) ? $x : undef;
1784}
1785
1786# read the entire log into a temporary file (which is removed ASAP)
1787# and store the file handle + parser state
1788sub svn_log_raw {
1789        my (@log_args) = @_;
1790        my $log_fh = IO::File->new_tmpfile or croak $!;
1791        my $pid = fork;
1792        defined $pid or croak $!;
1793        if (!$pid) {
1794                open STDOUT, '>&', $log_fh or croak $!;
1795                exec (qw(svn log), @log_args) or croak $!
1796        }
1797        waitpid $pid, 0;
1798        croak $? if $?;
1799        seek $log_fh, 0, 0 or croak $!;
1800        return { state => 'sep', fh => $log_fh };
1801}
1802
1803sub next_log_entry {
1804        my $log = shift; # retval of svn_log_raw()
1805        my $ret = undef;
1806        my $fh = $log->{fh};
1807
1808        while (<$fh>) {
1809                chomp;
1810                if (/^\-{72}$/) {
1811                        if ($log->{state} eq 'msg') {
1812                                if ($ret->{lines}) {
1813                                        $ret->{msg} .= $_."\n";
1814                                        unless(--$ret->{lines}) {
1815                                                $log->{state} = 'sep';
1816                                        }
1817                                } else {
1818                                        croak "Log parse error at: $_\n",
1819                                                $ret->{revision},
1820                                                "\n";
1821                                }
1822                                next;
1823                        }
1824                        if ($log->{state} ne 'sep') {
1825                                croak "Log parse error at: $_\n",
1826                                        "state: $log->{state}\n",
1827                                        $ret->{revision},
1828                                        "\n";
1829                        }
1830                        $log->{state} = 'rev';
1831
1832                        # if we have an empty log message, put something there:
1833                        if ($ret) {
1834                                $ret->{msg} ||= "\n";
1835                                delete $ret->{lines};
1836                                return $ret;
1837                        }
1838                        next;
1839                }
1840                if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
1841                        my $rev = $1;
1842                        my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
1843                        ($lines) = ($lines =~ /(\d+)/);
1844                        $date = '1970-01-01 00:00:00 +0000'
1845                                if ($_ignore_nodate && $date eq '(no date)');
1846                        my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1847                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1848                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1849                                         or croak "Failed to parse date: $date\n";
1850                        $ret = {        revision => $rev,
1851                                        date => "$tz $Y-$m-$d $H:$M:$S",
1852                                        author => $author,
1853                                        lines => $lines,
1854                                        msg => '' };
1855                        if (defined $_authors && ! defined $users{$author}) {
1856                                die "Author: $author not defined in ",
1857                                                "$_authors file\n";
1858                        }
1859                        $log->{state} = 'msg_start';
1860                        next;
1861                }
1862                # skip the first blank line of the message:
1863                if ($log->{state} eq 'msg_start' && /^$/) {
1864                        $log->{state} = 'msg';
1865                } elsif ($log->{state} eq 'msg') {
1866                        if ($ret->{lines}) {
1867                                $ret->{msg} .= $_."\n";
1868                                unless (--$ret->{lines}) {
1869                                        $log->{state} = 'sep';
1870                                }
1871                        } else {
1872                                croak "Log parse error at: $_\n",
1873                                        $ret->{revision},"\n";
1874                        }
1875                }
1876        }
1877        return $ret;
1878}
1879
1880sub svn_info {
1881        my $url = shift || $SVN_URL;
1882
1883        my $pid = open my $info_fh, '-|';
1884        defined $pid or croak $!;
1885
1886        if ($pid == 0) {
1887                exec(qw(svn info),$url) or croak $!;
1888        }
1889
1890        my $ret = {};
1891        # only single-lines seem to exist in svn info output
1892        while (<$info_fh>) {
1893                chomp $_;
1894                if (m#^([^:]+)\s*:\s*(\S.*)$#) {
1895                        $ret->{$1} = $2;
1896                        push @{$ret->{-order}}, $1;
1897                }
1898        }
1899        close $info_fh or croak $?;
1900        return $ret;
1901}
1902
1903sub sys { system(@_) == 0 or croak $? }
1904
1905sub do_update_index {
1906        my ($z_cmd, $cmd, $no_text_base) = @_;
1907
1908        my $z = open my $p, '-|';
1909        defined $z or croak $!;
1910        unless ($z) { exec @$z_cmd or croak $! }
1911
1912        my $pid = open my $ui, '|-';
1913        defined $pid or croak $!;
1914        unless ($pid) {
1915                exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!;
1916        }
1917        local $/ = "\0";
1918        while (my $x = <$p>) {
1919                chomp $x;
1920                if (!$no_text_base && lstat $x && ! -l _ &&
1921                                svn_propget_base('svn:keywords', $x)) {
1922                        my $mode = -x _ ? 0755 : 0644;
1923                        my ($v,$d,$f) = File::Spec->splitpath($x);
1924                        my $tb = File::Spec->catfile($d, '.svn', 'tmp',
1925                                                'text-base',"$f.svn-base");
1926                        $tb =~ s#^/##;
1927                        unless (-f $tb) {
1928                                $tb = File::Spec->catfile($d, '.svn',
1929                                                'text-base',"$f.svn-base");
1930                                $tb =~ s#^/##;
1931                        }
1932                        my @s = stat($x);
1933                        unlink $x or croak $!;
1934                        copy($tb, $x);
1935                        chmod(($mode &~ umask), $x) or croak $!;
1936                        utime $s[8], $s[9], $x;
1937                }
1938                print $ui $x,"\0";
1939        }
1940        close $ui or croak $?;
1941}
1942
1943sub index_changes {
1944        return if $_use_lib;
1945
1946        if (!-f "$GIT_SVN_DIR/info/exclude") {
1947                open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!;
1948                print $fd '.svn',"\n";
1949                close $fd or croak $!;
1950        }
1951        my $no_text_base = shift;
1952        do_update_index([qw/git-diff-files --name-only -z/],
1953                        'remove',
1954                        $no_text_base);
1955        do_update_index([qw/git-ls-files -z --others/,
1956                                "--exclude-from=$GIT_SVN_DIR/info/exclude"],
1957                        'add',
1958                        $no_text_base);
1959}
1960
1961sub s_to_file {
1962        my ($str, $file, $mode) = @_;
1963        open my $fd,'>',$file or croak $!;
1964        print $fd $str,"\n" or croak $!;
1965        close $fd or croak $!;
1966        chmod ($mode &~ umask, $file) if (defined $mode);
1967}
1968
1969sub file_to_s {
1970        my $file = shift;
1971        open my $fd,'<',$file or croak "$!: file: $file\n";
1972        local $/;
1973        my $ret = <$fd>;
1974        close $fd or croak $!;
1975        $ret =~ s/\s*$//s;
1976        return $ret;
1977}
1978
1979sub assert_revision_unknown {
1980        my $r = shift;
1981        if (my $c = revdb_get($REVDB, $r)) {
1982                croak "$r = $c already exists! Why are we refetching it?";
1983        }
1984}
1985
1986sub trees_eq {
1987        my ($x, $y) = @_;
1988        my @x = safe_qx('git-cat-file','commit',$x);
1989        my @y = safe_qx('git-cat-file','commit',$y);
1990        if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
1991                                || $y[0] !~ /^tree $sha1\n$/) {
1992                print STDERR "Trees not equal: $y[0] != $x[0]\n";
1993                return 0
1994        }
1995        return 1;
1996}
1997
1998sub git_commit {
1999        my ($log_msg, @parents) = @_;
2000        assert_revision_unknown($log_msg->{revision});
2001        map_tree_joins() if (@_branch_from && !%tree_map);
2002
2003        my (@tmp_parents, @exec_parents, %seen_parent);
2004        if (my $lparents = $log_msg->{parents}) {
2005                @tmp_parents = @$lparents
2006        }
2007        # commit parents can be conditionally bound to a particular
2008        # svn revision via: "svn_revno=commit_sha1", filter them out here:
2009        foreach my $p (@parents) {
2010                next unless defined $p;
2011                if ($p =~ /^(\d+)=($sha1_short)$/o) {
2012                        if ($1 == $log_msg->{revision}) {
2013                                push @tmp_parents, $2;
2014                        }
2015                } else {
2016                        push @tmp_parents, $p if $p =~ /$sha1_short/o;
2017                }
2018        }
2019        my $tree = $log_msg->{tree};
2020        if (!defined $tree) {
2021                my $index = set_index($GIT_SVN_INDEX);
2022                index_changes();
2023                chomp($tree = `git-write-tree`);
2024                croak $? if $?;
2025                restore_index($index);
2026        }
2027
2028        # just in case we clobber the existing ref, we still want that ref
2029        # as our parent:
2030        open my $null, '>', '/dev/null' or croak $!;
2031        open my $stderr, '>&', \*STDERR or croak $!;
2032        open STDERR, '>&', $null or croak $!;
2033        if (my $cur = eval { safe_qx('git-rev-parse',
2034                                     "refs/remotes/$GIT_SVN^0") }) {
2035                chomp $cur;
2036                push @tmp_parents, $cur;
2037        }
2038        open STDERR, '>&', $stderr or croak $!;
2039        close $stderr or croak $!;
2040        close $null or croak $!;
2041
2042        if (exists $tree_map{$tree}) {
2043                foreach my $p (@{$tree_map{$tree}}) {
2044                        my $skip;
2045                        foreach (@tmp_parents) {
2046                                # see if a common parent is found
2047                                my $mb = eval {
2048                                        safe_qx('git-merge-base', $_, $p)
2049                                };
2050                                next if ($@ || $?);
2051                                $skip = 1;
2052                                last;
2053                        }
2054                        next if $skip;
2055                        my ($url_p, $r_p, $uuid_p) = cmt_metadata($p);
2056                        next if (($SVN_UUID eq $uuid_p) &&
2057                                                ($log_msg->{revision} > $r_p));
2058                        next if (defined $url_p && defined $SVN_URL &&
2059                                                ($SVN_UUID eq $uuid_p) &&
2060                                                ($url_p eq $SVN_URL));
2061                        push @tmp_parents, $p;
2062                }
2063        }
2064        foreach (@tmp_parents) {
2065                next if $seen_parent{$_};
2066                $seen_parent{$_} = 1;
2067                push @exec_parents, $_;
2068                # MAXPARENT is defined to 16 in commit-tree.c:
2069                last if @exec_parents > 16;
2070        }
2071
2072        set_commit_env($log_msg);
2073        my @exec = ('git-commit-tree', $tree);
2074        push @exec, '-p', $_  foreach @exec_parents;
2075        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2076                                                                or croak $!;
2077        print $msg_fh $log_msg->{msg} or croak $!;
2078        unless ($_no_metadata) {
2079                print $msg_fh "\ngit-svn-id: $SVN_URL\@$log_msg->{revision}",
2080                                        " $SVN_UUID\n" or croak $!;
2081        }
2082        $msg_fh->flush == 0 or croak $!;
2083        close $msg_fh or croak $!;
2084        chomp(my $commit = do { local $/; <$out_fh> });
2085        close $out_fh or croak $!;
2086        waitpid $pid, 0;
2087        croak $? if $?;
2088        if ($commit !~ /^$sha1$/o) {
2089                die "Failed to commit, invalid sha1: $commit\n";
2090        }
2091        sys('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
2092        revdb_set($REVDB, $log_msg->{revision}, $commit);
2093
2094        # this output is read via pipe, do not change:
2095        print "r$log_msg->{revision} = $commit\n";
2096        check_repack();
2097        return $commit;
2098}
2099
2100sub check_repack {
2101        if ($_repack && (--$_repack_nr == 0)) {
2102                $_repack_nr = $_repack;
2103                sys("git repack $_repack_flags");
2104        }
2105}
2106
2107sub set_commit_env {
2108        my ($log_msg) = @_;
2109        my $author = $log_msg->{author};
2110        if (!defined $author || length $author == 0) {
2111                $author = '(no author)';
2112        }
2113        my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
2114                                : ($author,"$author\@$SVN_UUID");
2115        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
2116        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
2117        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
2118}
2119
2120sub apply_mod_line_blob {
2121        my $m = shift;
2122        if ($m->{mode_b} =~ /^120/) {
2123                blob_to_symlink($m->{sha1_b}, $m->{file_b});
2124        } else {
2125                blob_to_file($m->{sha1_b}, $m->{file_b});
2126        }
2127}
2128
2129sub blob_to_symlink {
2130        my ($blob, $link) = @_;
2131        defined $link or croak "\$link not defined!\n";
2132        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2133        if (-l $link || -f _) {
2134                unlink $link or croak $!;
2135        }
2136
2137        my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
2138        symlink $dest, $link or croak $!;
2139}
2140
2141sub blob_to_file {
2142        my ($blob, $file) = @_;
2143        defined $file or croak "\$file not defined!\n";
2144        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2145        if (-l $file || -f _) {
2146                unlink $file or croak $!;
2147        }
2148
2149        open my $blob_fh, '>', $file or croak "$!: $file\n";
2150        my $pid = fork;
2151        defined $pid or croak $!;
2152
2153        if ($pid == 0) {
2154                open STDOUT, '>&', $blob_fh or croak $!;
2155                exec('git-cat-file','blob',$blob) or croak $!;
2156        }
2157        waitpid $pid, 0;
2158        croak $? if $?;
2159
2160        close $blob_fh or croak $!;
2161}
2162
2163sub safe_qx {
2164        my $pid = open my $child, '-|';
2165        defined $pid or croak $!;
2166        if ($pid == 0) {
2167                exec(@_) or croak $!;
2168        }
2169        my @ret = (<$child>);
2170        close $child or croak $?;
2171        die $? if $?; # just in case close didn't error out
2172        return wantarray ? @ret : join('',@ret);
2173}
2174
2175sub svn_compat_check {
2176        if ($_follow_parent) {
2177                print STDERR 'E: --follow-parent functionality is only ',
2178                                "available when SVN libraries are used\n";
2179                exit 1;
2180        }
2181        my @co_help = safe_qx(qw(svn co -h));
2182        unless (grep /ignore-externals/,@co_help) {
2183                print STDERR "W: Installed svn version does not support ",
2184                                "--ignore-externals\n";
2185                $_no_ignore_ext = 1;
2186        }
2187        if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
2188                $_svn_co_url_revs = 1;
2189        }
2190        if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) {
2191                $_svn_pg_peg_revs = 1;
2192        }
2193
2194        # I really, really hope nobody hits this...
2195        unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
2196                print STDERR <<'';
2197W: The installed svn version does not support the --stop-on-copy flag in
2198   the log command.
2199   Lets hope the directory you're tracking is not a branch or tag
2200   and was never moved within the repository...
2201
2202                $_no_stop_copy = 1;
2203        }
2204}
2205
2206# *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
2207# (and they won't honor URL@<rev> without -r<rev>, too!)
2208sub svn_cmd_checkout {
2209        my ($url, $rev, $dir) = @_;
2210        my @cmd = ('svn','co', "-r$rev");
2211        push @cmd, '--ignore-externals' unless $_no_ignore_ext;
2212        $url .= "\@$rev" if $_svn_co_url_revs;
2213        sys(@cmd, $url, $dir);
2214}
2215
2216sub check_upgrade_needed {
2217        if (!-r $REVDB) {
2218                -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2219                open my $fh, '>>',$REVDB or croak $!;
2220                close $fh;
2221        }
2222        my $old = eval {
2223                my $pid = open my $child, '-|';
2224                defined $pid or croak $!;
2225                if ($pid == 0) {
2226                        close STDERR;
2227                        exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $!;
2228                }
2229                my @ret = (<$child>);
2230                close $child or croak $?;
2231                die $? if $?; # just in case close didn't error out
2232                return wantarray ? @ret : join('',@ret);
2233        };
2234        return unless $old;
2235        my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
2236        if ($@ || !$head) {
2237                print STDERR "Please run: $0 rebuild --upgrade\n";
2238                exit 1;
2239        }
2240}
2241
2242# fills %tree_map with a reverse mapping of trees to commits.  Useful
2243# for finding parents to commit on.
2244sub map_tree_joins {
2245        my %seen;
2246        foreach my $br (@_branch_from) {
2247                my $pid = open my $pipe, '-|';
2248                defined $pid or croak $!;
2249                if ($pid == 0) {
2250                        exec(qw(git-rev-list --topo-order --pretty=raw), $br)
2251                                                                or croak $!;
2252                }
2253                while (<$pipe>) {
2254                        if (/^commit ($sha1)$/o) {
2255                                my $commit = $1;
2256
2257                                # if we've seen a commit,
2258                                # we've seen its parents
2259                                last if $seen{$commit};
2260                                my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
2261                                unless (defined $tree) {
2262                                        die "Failed to parse commit $commit\n";
2263                                }
2264                                push @{$tree_map{$tree}}, $commit;
2265                                $seen{$commit} = 1;
2266                        }
2267                }
2268                close $pipe; # we could be breaking the pipe early
2269        }
2270}
2271
2272sub load_all_refs {
2273        if (@_branch_from) {
2274                print STDERR '--branch|-b parameters are ignored when ',
2275                        "--branch-all-refs|-B is passed\n";
2276        }
2277
2278        # don't worry about rev-list on non-commit objects/tags,
2279        # it shouldn't blow up if a ref is a blob or tree...
2280        chomp(@_branch_from = `git-rev-parse --symbolic --all`);
2281}
2282
2283# '<svn username> = real-name <email address>' mapping based on git-svnimport:
2284sub load_authors {
2285        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2286        while (<$authors>) {
2287                chomp;
2288                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
2289                my ($user, $name, $email) = ($1, $2, $3);
2290                $users{$user} = [$name, $email];
2291        }
2292        close $authors or croak $!;
2293}
2294
2295sub rload_authors {
2296        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2297        while (<$authors>) {
2298                chomp;
2299                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2300                my ($user, $name, $email) = ($1, $2, $3);
2301                $rusers{"$name <$email>"} = $user;
2302        }
2303        close $authors or croak $!;
2304}
2305
2306sub svn_propget_base {
2307        my ($p, $f) = @_;
2308        $f .= '@BASE' if $_svn_pg_peg_revs;
2309        return safe_qx(qw/svn propget/, $p, $f);
2310}
2311
2312sub git_svn_each {
2313        my $sub = shift;
2314        foreach (`git-rev-parse --symbolic --all`) {
2315                next unless s#^refs/remotes/##;
2316                chomp $_;
2317                next unless -f "$GIT_DIR/svn/$_/info/url";
2318                &$sub($_);
2319        }
2320}
2321
2322sub migrate_revdb {
2323        git_svn_each(sub {
2324                my $id = shift;
2325                defined(my $pid = fork) or croak $!;
2326                if (!$pid) {
2327                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2328                        init_vars();
2329                        exit 0 if -r $REVDB;
2330                        print "Upgrading svn => git mapping...\n";
2331                        -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2332                        open my $fh, '>>',$REVDB or croak $!;
2333                        close $fh;
2334                        rebuild();
2335                        print "Done upgrading. You may now delete the ",
2336                                "deprecated $GIT_SVN_DIR/revs directory\n";
2337                        exit 0;
2338                }
2339                waitpid $pid, 0;
2340                croak $? if $?;
2341        });
2342}
2343
2344sub migration_check {
2345        migrate_revdb() unless (-e $REVDB);
2346        return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
2347        print "Upgrading repository...\n";
2348        unless (-d "$GIT_DIR/svn") {
2349                mkdir "$GIT_DIR/svn" or croak $!;
2350        }
2351        print "Data from a previous version of git-svn exists, but\n\t",
2352                                "$GIT_SVN_DIR\n\t(required for this version ",
2353                                "($VERSION) of git-svn) does not.\n";
2354
2355        foreach my $x (`git-rev-parse --symbolic --all`) {
2356                next unless $x =~ s#^refs/remotes/##;
2357                chomp $x;
2358                next unless -f "$GIT_DIR/$x/info/url";
2359                my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
2360                next unless $u;
2361                my $dn = dirname("$GIT_DIR/svn/$x");
2362                mkpath([$dn]) unless -d $dn;
2363                rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
2364        }
2365        migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
2366        print "Done upgrading.\n";
2367}
2368
2369sub find_rev_before {
2370        my ($r, $id, $eq_ok) = @_;
2371        my $f = "$GIT_DIR/svn/$id/.rev_db";
2372        return (undef,undef) unless -r $f;
2373        --$r unless $eq_ok;
2374        while ($r > 0) {
2375                if (my $c = revdb_get($f, $r)) {
2376                        return ($r, $c);
2377                }
2378                --$r;
2379        }
2380        return (undef, undef);
2381}
2382
2383sub init_vars {
2384        $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
2385        $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
2386        $REVDB = "$GIT_SVN_DIR/.rev_db";
2387        $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
2388        $SVN_URL = undef;
2389        $SVN_WC = "$GIT_SVN_DIR/tree";
2390        %tree_map = ();
2391}
2392
2393# convert GetOpt::Long specs for use by git-repo-config
2394sub read_repo_config {
2395        return unless -d $GIT_DIR;
2396        my $opts = shift;
2397        foreach my $o (keys %$opts) {
2398                my $v = $opts->{$o};
2399                my ($key) = ($o =~ /^([a-z\-]+)/);
2400                $key =~ s/-//g;
2401                my $arg = 'git-repo-config';
2402                $arg .= ' --int' if ($o =~ /[:=]i$/);
2403                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
2404                if (ref $v eq 'ARRAY') {
2405                        chomp(my @tmp = `$arg --get-all svn.$key`);
2406                        @$v = @tmp if @tmp;
2407                } else {
2408                        chomp(my $tmp = `$arg --get svn.$key`);
2409                        if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
2410                                $$v = $tmp;
2411                        }
2412                }
2413        }
2414}
2415
2416sub set_default_vals {
2417        if (defined $_repack) {
2418                $_repack = 1000 if ($_repack <= 0);
2419                $_repack_nr = $_repack;
2420                $_repack_flags ||= '-d';
2421        }
2422}
2423
2424sub read_grafts {
2425        my $gr_file = shift;
2426        my ($grafts, $comments) = ({}, {});
2427        if (open my $fh, '<', $gr_file) {
2428                my @tmp;
2429                while (<$fh>) {
2430                        if (/^($sha1)\s+/) {
2431                                my $c = $1;
2432                                if (@tmp) {
2433                                        @{$comments->{$c}} = @tmp;
2434                                        @tmp = ();
2435                                }
2436                                foreach my $p (split /\s+/, $_) {
2437                                        $grafts->{$c}->{$p} = 1;
2438                                }
2439                        } else {
2440                                push @tmp, $_;
2441                        }
2442                }
2443                close $fh or croak $!;
2444                @{$comments->{'END'}} = @tmp if @tmp;
2445        }
2446        return ($grafts, $comments);
2447}
2448
2449sub write_grafts {
2450        my ($grafts, $comments, $gr_file) = @_;
2451
2452        open my $fh, '>', $gr_file or croak $!;
2453        foreach my $c (sort keys %$grafts) {
2454                if ($comments->{$c}) {
2455                        print $fh $_ foreach @{$comments->{$c}};
2456                }
2457                my $p = $grafts->{$c};
2458                my %x; # real parents
2459                delete $p->{$c}; # commits are not self-reproducing...
2460                my $pid = open my $ch, '-|';
2461                defined $pid or croak $!;
2462                if (!$pid) {
2463                        exec(qw/git-cat-file commit/, $c) or croak $!;
2464                }
2465                while (<$ch>) {
2466                        if (/^parent ($sha1)/) {
2467                                $x{$1} = $p->{$1} = 1;
2468                        } else {
2469                                last unless /^\S/;
2470                        }
2471                }
2472                close $ch; # breaking the pipe
2473
2474                # if real parents are the only ones in the grafts, drop it
2475                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2476
2477                my (@ip, @jp, $mb);
2478                my %del = %x;
2479                @ip = @jp = keys %$p;
2480                foreach my $i (@ip) {
2481                        next if $del{$i} || $p->{$i} == 2;
2482                        foreach my $j (@jp) {
2483                                next if $i eq $j || $del{$j} || $p->{$j} == 2;
2484                                $mb = eval { safe_qx('git-merge-base',$i,$j) };
2485                                next unless $mb;
2486                                chomp $mb;
2487                                next if $x{$mb};
2488                                if ($mb eq $j) {
2489                                        delete $p->{$i};
2490                                        $del{$i} = 1;
2491                                } elsif ($mb eq $i) {
2492                                        delete $p->{$j};
2493                                        $del{$j} = 1;
2494                                }
2495                        }
2496                }
2497
2498                # if real parents are the only ones in the grafts, drop it
2499                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2500
2501                print $fh $c, ' ', join(' ', sort keys %$p),"\n";
2502        }
2503        if ($comments->{'END'}) {
2504                print $fh $_ foreach @{$comments->{'END'}};
2505        }
2506        close $fh or croak $!;
2507}
2508
2509sub read_url_paths_all {
2510        my ($l_map, $pfx, $p) = @_;
2511        my @dir;
2512        foreach (<$p/*>) {
2513                if (-r "$_/info/url") {
2514                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2515                        my $id = $pfx . basename $_;
2516                        my $url = file_to_s("$_/info/url");
2517                        my ($u, $p) = repo_path_split($url);
2518                        $l_map->{$u}->{$p} = $id;
2519                } elsif (-d $_) {
2520                        push @dir, $_;
2521                }
2522        }
2523        foreach (@dir) {
2524                my $x = $_;
2525                $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
2526                read_url_paths_all($l_map, $x, $_);
2527        }
2528}
2529
2530# this one only gets ids that have been imported, not new ones
2531sub read_url_paths {
2532        my $l_map = {};
2533        git_svn_each(sub { my $x = shift;
2534                        my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
2535                        my ($u, $p) = repo_path_split($url);
2536                        $l_map->{$u}->{$p} = $x;
2537                        });
2538        return $l_map;
2539}
2540
2541sub extract_metadata {
2542        my $id = shift or return (undef, undef, undef);
2543        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
2544                                                        \s([a-f\d\-]+)$/x);
2545        if (!defined $rev || !$uuid || !$url) {
2546                # some of the original repositories I made had
2547                # identifiers like this:
2548                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
2549        }
2550        return ($url, $rev, $uuid);
2551}
2552
2553sub cmt_metadata {
2554        return extract_metadata((grep(/^git-svn-id: /,
2555                safe_qx(qw/git-cat-file commit/, shift)))[-1]);
2556}
2557
2558sub get_commit_time {
2559        my $cmt = shift;
2560        defined(my $pid = open my $fh, '-|') or croak $!;
2561        if (!$pid) {
2562                exec qw/git-rev-list --pretty=raw -n1/, $cmt or croak $!;
2563        }
2564        while (<$fh>) {
2565                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
2566                my ($s, $tz) = ($1, $2);
2567                if ($tz =~ s/^\+//) {
2568                        $s += tz_to_s_offset($tz);
2569                } elsif ($tz =~ s/^\-//) {
2570                        $s -= tz_to_s_offset($tz);
2571                }
2572                close $fh;
2573                return $s;
2574        }
2575        die "Can't get commit time for commit: $cmt\n";
2576}
2577
2578sub tz_to_s_offset {
2579        my ($tz) = @_;
2580        $tz =~ s/(\d\d)$//;
2581        return ($1 * 60) + ($tz * 3600);
2582}
2583
2584# adapted from pager.c
2585sub config_pager {
2586        $_pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2587        if (!defined $_pager) {
2588                $_pager = 'less';
2589        } elsif (length $_pager == 0 || $_pager eq 'cat') {
2590                $_pager = undef;
2591        }
2592}
2593
2594sub run_pager {
2595        return unless -t *STDOUT;
2596        pipe my $rfd, my $wfd or return;
2597        defined(my $pid = fork) or croak $!;
2598        if (!$pid) {
2599                open STDOUT, '>&', $wfd or croak $!;
2600                return;
2601        }
2602        open STDIN, '<&', $rfd or croak $!;
2603        $ENV{LESS} ||= 'FRSX';
2604        exec $_pager or croak "Can't run pager: $! ($_pager)\n";
2605}
2606
2607sub get_author_info {
2608        my ($dest, $author, $t, $tz) = @_;
2609        $author =~ s/(?:^\s*|\s*$)//g;
2610        $dest->{a_raw} = $author;
2611        my $_a;
2612        if ($_authors) {
2613                $_a = $rusers{$author} || undef;
2614        }
2615        if (!$_a) {
2616                ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
2617        }
2618        $dest->{t} = $t;
2619        $dest->{tz} = $tz;
2620        $dest->{a} = $_a;
2621        # Date::Parse isn't in the standard Perl distro :(
2622        if ($tz =~ s/^\+//) {
2623                $t += tz_to_s_offset($tz);
2624        } elsif ($tz =~ s/^\-//) {
2625                $t -= tz_to_s_offset($tz);
2626        }
2627        $dest->{t_utc} = $t;
2628}
2629
2630sub process_commit {
2631        my ($c, $r_min, $r_max, $defer) = @_;
2632        if (defined $r_min && defined $r_max) {
2633                if ($r_min == $c->{r} && $r_min == $r_max) {
2634                        show_commit($c);
2635                        return 0;
2636                }
2637                return 1 if $r_min == $r_max;
2638                if ($r_min < $r_max) {
2639                        # we need to reverse the print order
2640                        return 0 if (defined $_limit && --$_limit < 0);
2641                        push @$defer, $c;
2642                        return 1;
2643                }
2644                if ($r_min != $r_max) {
2645                        return 1 if ($r_min < $c->{r});
2646                        return 1 if ($r_max > $c->{r});
2647                }
2648        }
2649        return 0 if (defined $_limit && --$_limit < 0);
2650        show_commit($c);
2651        return 1;
2652}
2653
2654sub show_commit {
2655        my $c = shift;
2656        if ($_oneline) {
2657                my $x = "\n";
2658                if (my $l = $c->{l}) {
2659                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2660                        $x = $l->[0];
2661                }
2662                $_l_fmt ||= 'A' . length($c->{r});
2663                print 'r',pack($_l_fmt, $c->{r}),' | ';
2664                print "$c->{c} | " if $_show_commit;
2665                print $x;
2666        } else {
2667                show_commit_normal($c);
2668        }
2669}
2670
2671sub show_commit_changed_paths {
2672        my ($c) = @_;
2673        return unless $c->{changed};
2674        print "Changed paths:\n", @{$c->{changed}};
2675}
2676
2677sub show_commit_normal {
2678        my ($c) = @_;
2679        print '-' x72, "\nr$c->{r} | ";
2680        print "$c->{c} | " if $_show_commit;
2681        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2682                                 localtime($c->{t_utc})), ' | ';
2683        my $nr_line = 0;
2684
2685        if (my $l = $c->{l}) {
2686                while ($l->[$#$l] eq "\n" && $#$l > 0
2687                                          && $l->[($#$l - 1)] eq "\n") {
2688                        pop @$l;
2689                }
2690                $nr_line = scalar @$l;
2691                if (!$nr_line) {
2692                        print "1 line\n\n\n";
2693                } else {
2694                        if ($nr_line == 1) {
2695                                $nr_line = '1 line';
2696                        } else {
2697                                $nr_line .= ' lines';
2698                        }
2699                        print $nr_line, "\n";
2700                        show_commit_changed_paths($c);
2701                        print "\n";
2702                        print $_ foreach @$l;
2703                }
2704        } else {
2705                print "1 line\n";
2706                show_commit_changed_paths($c);
2707                print "\n";
2708
2709        }
2710        foreach my $x (qw/raw diff/) {
2711                if ($c->{$x}) {
2712                        print "\n";
2713                        print $_ foreach @{$c->{$x}}
2714                }
2715        }
2716}
2717
2718sub libsvn_load {
2719        return unless $_use_lib;
2720        $_use_lib = eval {
2721                require SVN::Core;
2722                if ($SVN::Core::VERSION lt '1.1.0') {
2723                        die "Need SVN::Core 1.1.0 or better ",
2724                                        "(got $SVN::Core::VERSION) ",
2725                                        "Falling back to command-line svn\n";
2726                }
2727                require SVN::Ra;
2728                require SVN::Delta;
2729                push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
2730                push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
2731                *SVN::Git::Fetcher::process_rm = *process_rm;
2732                *SVN::Git::Fetcher::safe_qx = *safe_qx;
2733                my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2734                                        $SVN::Node::dir.$SVN::Node::unknown.
2735                                        $SVN::Node::none.$SVN::Node::file.
2736                                        $SVN::Node::dir.$SVN::Node::unknown.
2737                                        $SVN::Auth::SSL::CNMISMATCH.
2738                                        $SVN::Auth::SSL::NOTYETVALID.
2739                                        $SVN::Auth::SSL::EXPIRED.
2740                                        $SVN::Auth::SSL::UNKNOWNCA.
2741                                        $SVN::Auth::SSL::OTHER;
2742                1;
2743        };
2744}
2745
2746sub _simple_prompt {
2747        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2748        $may_save = undef if $_no_auth_cache;
2749        $default_username = $_username if defined $_username;
2750        if (defined $default_username && length $default_username) {
2751                if (defined $realm && length $realm) {
2752                        print "Authentication realm: $realm\n";
2753                }
2754                $cred->username($default_username);
2755        } else {
2756                _username_prompt($cred, $realm, $may_save, $pool);
2757        }
2758        $cred->password(_read_password("Password for '" .
2759                                       $cred->username . "': ", $realm));
2760        $cred->may_save($may_save);
2761        $SVN::_Core::SVN_NO_ERROR;
2762}
2763
2764sub _ssl_server_trust_prompt {
2765        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2766        $may_save = undef if $_no_auth_cache;
2767        print "Error validating server certificate for '$realm':\n";
2768        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2769                print " - The certificate is not issued by a trusted ",
2770                      "authority. Use the\n",
2771                      "   fingerprint to validate the certificate manually!\n";
2772        }
2773        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2774                print " - The certificate hostname does not match.\n";
2775        }
2776        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2777                print " - The certificate is not yet valid.\n";
2778        }
2779        if ($failures & $SVN::Auth::SSL::EXPIRED) {
2780                print " - The certificate has expired.\n";
2781        }
2782        if ($failures & $SVN::Auth::SSL::OTHER) {
2783                print " - The certificate has an unknown error.\n";
2784        }
2785        printf( "Certificate information:\n".
2786                " - Hostname: %s\n".
2787                " - Valid: from %s until %s\n".
2788                " - Issuer: %s\n".
2789                " - Fingerprint: %s\n",
2790                map $cert_info->$_, qw(hostname valid_from valid_until
2791                                       issuer_dname fingerprint) );
2792        my $choice;
2793prompt:
2794        print $may_save ?
2795              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2796              "(R)eject or accept (t)emporarily? ";
2797        $choice = lc(substr(<STDIN> || 'R', 0, 1));
2798        if ($choice =~ /^t$/i) {
2799                $cred->may_save(undef);
2800        } elsif ($choice =~ /^r$/i) {
2801                return -1;
2802        } elsif ($may_save && $choice =~ /^p$/i) {
2803                $cred->may_save($may_save);
2804        } else {
2805                goto prompt;
2806        }
2807        $cred->accepted_failures($failures);
2808        $SVN::_Core::SVN_NO_ERROR;
2809}
2810
2811sub _ssl_client_cert_prompt {
2812        my ($cred, $realm, $may_save, $pool) = @_;
2813        $may_save = undef if $_no_auth_cache;
2814        print "Client certificate filename: ";
2815        chomp(my $filename = <STDIN>);
2816        $cred->cert_file($filename);
2817        $cred->may_save($may_save);
2818        $SVN::_Core::SVN_NO_ERROR;
2819}
2820
2821sub _ssl_client_cert_pw_prompt {
2822        my ($cred, $realm, $may_save, $pool) = @_;
2823        $may_save = undef if $_no_auth_cache;
2824        $cred->password(_read_password("Password: ", $realm));
2825        $cred->may_save($may_save);
2826        $SVN::_Core::SVN_NO_ERROR;
2827}
2828
2829sub _username_prompt {
2830        my ($cred, $realm, $may_save, $pool) = @_;
2831        $may_save = undef if $_no_auth_cache;
2832        if (defined $realm && length $realm) {
2833                print "Authentication realm: $realm\n";
2834        }
2835        my $username;
2836        if (defined $_username) {
2837                $username = $_username;
2838        } else {
2839                print "Username: ";
2840                chomp($username = <STDIN>);
2841        }
2842        $cred->username($username);
2843        $cred->may_save($may_save);
2844        $SVN::_Core::SVN_NO_ERROR;
2845}
2846
2847sub _read_password {
2848        my ($prompt, $realm) = @_;
2849        print $prompt;
2850        require Term::ReadKey;
2851        Term::ReadKey::ReadMode('noecho');
2852        my $password = '';
2853        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2854                last if $key =~ /[\012\015]/; # \n\r
2855                $password .= $key;
2856        }
2857        Term::ReadKey::ReadMode('restore');
2858        print "\n";
2859        $password;
2860}
2861
2862sub libsvn_connect {
2863        my ($url) = @_;
2864        SVN::_Core::svn_config_ensure($_config_dir, undef);
2865        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2866            SVN::Client::get_simple_provider(),
2867            SVN::Client::get_ssl_server_trust_file_provider(),
2868            SVN::Client::get_simple_prompt_provider(
2869              \&_simple_prompt, 2),
2870            SVN::Client::get_ssl_client_cert_prompt_provider(
2871              \&_ssl_client_cert_prompt, 2),
2872            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2873              \&_ssl_client_cert_pw_prompt, 2),
2874            SVN::Client::get_username_provider(),
2875            SVN::Client::get_ssl_server_trust_prompt_provider(
2876              \&_ssl_server_trust_prompt),
2877            SVN::Client::get_username_prompt_provider(
2878              \&_username_prompt, 2),
2879          ]);
2880        my $config = SVN::Core::config_get_config($_config_dir);
2881        my $ra = SVN::Ra->new(url => $url, auth => $baton,
2882                              config => $config,
2883                              pool => SVN::Pool->new,
2884                              auth_provider_callbacks => $callbacks);
2885
2886        my $df = $ENV{GIT_SVN_DELTA_FETCH};
2887        if (defined $df) {
2888                $_xfer_delta = $df;
2889        } else {
2890                $_xfer_delta = ($url =~ m#^file://#) ? undef : 1;
2891        }
2892        $ra->{svn_path} = $url;
2893        $ra->{repos_root} = $ra->get_repos_root;
2894        $ra->{svn_path} =~ s#^\Q$ra->{repos_root}\E/*##;
2895        push @repo_path_split_cache, qr/^(\Q$ra->{repos_root}\E)/;
2896        return $ra;
2897}
2898
2899sub libsvn_can_do_switch {
2900        unless (defined $_svn_can_do_switch) {
2901                my $pool = SVN::Pool->new;
2902                my $rep = eval {
2903                        $SVN->do_switch(1, '', 0, $SVN->{url},
2904                                        SVN::Delta::Editor->new, $pool);
2905                };
2906                if ($@) {
2907                        $_svn_can_do_switch = 0;
2908                } else {
2909                        $rep->abort_report($pool);
2910                        $_svn_can_do_switch = 1;
2911                }
2912                $pool->clear;
2913        }
2914        $_svn_can_do_switch;
2915}
2916
2917sub libsvn_dup_ra {
2918        my ($ra) = @_;
2919        SVN::Ra->new(map { $_ => $ra->{$_} } qw/config url
2920                     auth auth_provider_callbacks repos_root svn_path/);
2921}
2922
2923sub libsvn_get_file {
2924        my ($gui, $f, $rev, $chg, $untracked) = @_;
2925        $f =~ s#^/##;
2926        print "\t$chg\t$f\n" unless $_q;
2927
2928        my ($hash, $pid, $in, $out);
2929        my $pool = SVN::Pool->new;
2930        defined($pid = open3($in, $out, '>&STDERR',
2931                                qw/git-hash-object -w --stdin/)) or croak $!;
2932        # redirect STDOUT for SVN 1.1.x compatibility
2933        open my $stdout, '>&', \*STDOUT or croak $!;
2934        open STDOUT, '>&', $in or croak $!;
2935        my ($r, $props) = $SVN->get_file($f, $rev, \*STDOUT, $pool);
2936        $in->flush == 0 or croak $!;
2937        open STDOUT, '>&', $stdout or croak $!;
2938        close $in or croak $!;
2939        close $stdout or croak $!;
2940        $pool->clear;
2941        chomp($hash = do { local $/; <$out> });
2942        close $out or croak $!;
2943        waitpid $pid, 0;
2944        $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2945
2946        my $mode = exists $props->{'svn:executable'} ? '100755' : '100644';
2947        if (exists $props->{'svn:special'}) {
2948                $mode = '120000';
2949                my $link = `git-cat-file blob $hash`;
2950                $link =~ s/^link // or die "svn:special file with contents: <",
2951                                                $link, "> is not understood\n";
2952                defined($pid = open3($in, $out, '>&STDERR',
2953                                qw/git-hash-object -w --stdin/)) or croak $!;
2954                print $in $link;
2955                $in->flush == 0 or croak $!;
2956                close $in or croak $!;
2957                chomp($hash = do { local $/; <$out> });
2958                close $out or croak $!;
2959                waitpid $pid, 0;
2960                $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2961        }
2962        %{$untracked->{file_prop}->{$f}} = %$props;
2963        print $gui $mode,' ',$hash,"\t",$f,"\0" or croak $!;
2964}
2965
2966sub uri_encode {
2967        my ($f) = @_;
2968        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2969        $f
2970}
2971
2972sub uri_decode {
2973        my ($f) = @_;
2974        $f =~ tr/+/ /;
2975        $f =~ s/%([A-F0-9]{2})/chr hex($1)/ge;
2976        $f
2977}
2978
2979sub libsvn_log_entry {
2980        my ($rev, $author, $date, $msg, $parents, $untracked) = @_;
2981        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2982                                         (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
2983                                or die "Unable to parse date: $date\n";
2984        if (defined $_authors && ! defined $users{$author}) {
2985                die "Author: $author not defined in $_authors file\n";
2986        }
2987        $msg = '' if ($rev == 0 && !defined $msg);
2988
2989        open my $un, '>>', "$GIT_SVN_DIR/unhandled.log" or croak $!;
2990        my $h;
2991        print $un "r$rev\n" or croak $!;
2992        $h = $untracked->{empty};
2993        foreach (sort keys %$h) {
2994                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2995                print $un "  $act: ", uri_encode($_), "\n" or croak $!;
2996                warn "W: $act: $_\n";
2997        }
2998        foreach my $t (qw/dir_prop file_prop/) {
2999                $h = $untracked->{$t} or next;
3000                foreach my $path (sort keys %$h) {
3001                        my $ppath = $path eq '' ? '.' : $path;
3002                        foreach my $prop (sort keys %{$h->{$path}}) {
3003                                next if $SKIP{$prop};
3004                                my $v = $h->{$path}->{$prop};
3005                                if (defined $v) {
3006                                        print $un "  +$t: ",
3007                                                  uri_encode($ppath), ' ',
3008                                                  uri_encode($prop), ' ',
3009                                                  uri_encode($v), "\n"
3010                                                  or croak $!;
3011                                } else {
3012                                        print $un "  -$t: ",
3013                                                  uri_encode($ppath), ' ',
3014                                                  uri_encode($prop), "\n"
3015                                                  or croak $!;
3016                                }
3017                        }
3018                }
3019        }
3020        foreach my $t (qw/absent_file absent_directory/) {
3021                $h = $untracked->{$t} or next;
3022                foreach my $parent (sort keys %$h) {
3023                        foreach my $path (sort @{$h->{$parent}}) {
3024                                print $un "  $t: ",
3025                                      uri_encode("$parent/$path"), "\n"
3026                                      or croak $!;
3027                                warn "W: $t: $parent/$path ",
3028                                     "Insufficient permissions?\n";
3029                        }
3030                }
3031        }
3032
3033        # revprops (make this optional? it's an extra network trip...)
3034        my $pool = SVN::Pool->new;
3035        my $rp = $SVN->rev_proplist($rev, $pool);
3036        foreach (sort keys %$rp) {
3037                next if /^svn:(?:author|date|log)$/;
3038                print $un "  rev_prop: ", uri_encode($_), ' ',
3039                          uri_encode($rp->{$_}), "\n";
3040        }
3041        $pool->clear;
3042        close $un or croak $!;
3043
3044        { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
3045          author => $author, msg => $msg."\n", parents => $parents || [],
3046          revprops => $rp }
3047}
3048
3049sub process_rm {
3050        my ($gui, $last_commit, $f, $q) = @_;
3051        # remove entire directories.
3052        if (safe_qx('git-ls-tree',$last_commit,'--',$f) =~ /^040000 tree/) {
3053                defined(my $pid = open my $ls, '-|') or croak $!;
3054                if (!$pid) {
3055                        exec(qw/git-ls-tree -r --name-only -z/,
3056                                $last_commit,'--',$f) or croak $!;
3057                }
3058                local $/ = "\0";
3059                while (<$ls>) {
3060                        print $gui '0 ',0 x 40,"\t",$_ or croak $!;
3061                        print "\tD\t$_\n" unless $q;
3062                }
3063                print "\tD\t$f/\n" unless $q;
3064                close $ls or croak $?;
3065                return $SVN::Node::dir;
3066        } else {
3067                print $gui '0 ',0 x 40,"\t",$f,"\0" or croak $!;
3068                print "\tD\t$f\n" unless $q;
3069                return $SVN::Node::file;
3070        }
3071}
3072
3073sub libsvn_fetch {
3074        $_xfer_delta ? libsvn_fetch_delta(@_) : libsvn_fetch_full(@_);
3075}
3076
3077sub libsvn_fetch_delta {
3078        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
3079        my $pool = SVN::Pool->new;
3080        my $ed = SVN::Git::Fetcher->new({ c => $last_commit, q => $_q });
3081        my $reporter = $SVN->do_update($rev, '', 1, $ed, $pool);
3082        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3083        my (undef, $last_rev, undef) = cmt_metadata($last_commit);
3084        $reporter->set_path('', $last_rev, 0, @lock, $pool);
3085        $reporter->finish_report($pool);
3086        $pool->clear;
3087        unless ($ed->{git_commit_ok}) {
3088                die "SVN connection failed somewhere...\n";
3089        }
3090        libsvn_log_entry($rev, $author, $date, $msg, [$last_commit], $ed);
3091}
3092
3093sub libsvn_fetch_full {
3094        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
3095        open my $gui, '| git-update-index -z --index-info' or croak $!;
3096        my %amr;
3097        my $ut = { empty => {}, dir_prop => {}, file_prop => {} };
3098        my $p = $SVN->{svn_path};
3099        foreach my $f (keys %$paths) {
3100                my $m = $paths->{$f}->action();
3101                if (length $p) {
3102                        $f =~ s#^/\Q$p\E/##;
3103                        next if $f =~ m#^/#;
3104                } else {
3105                        $f =~ s#^/##;
3106                }
3107                if ($m =~ /^[DR]$/) {
3108                        my $t = process_rm($gui, $last_commit, $f, $_q);
3109                        if ($m eq 'D') {
3110                                $ut->{empty}->{$f} = 0 if $t == $SVN::Node::dir;
3111                                next;
3112                        }
3113                        # 'R' can be file replacements, too, right?
3114                }
3115                my $pool = SVN::Pool->new;
3116                my $t = $SVN->check_path($f, $rev, $pool);
3117                if ($t == $SVN::Node::file) {
3118                        if ($m =~ /^[AMR]$/) {
3119                                $amr{$f} = $m;
3120                        } else {
3121                                die "Unrecognized action: $m, ($f r$rev)\n";
3122                        }
3123                } elsif ($t == $SVN::Node::dir && $m =~ /^[AR]$/) {
3124                        my @traversed = ();
3125                        libsvn_traverse($gui, '', $f, $rev, \@traversed, $ut);
3126                        if (@traversed) {
3127                                foreach (@traversed) {
3128                                        $amr{$_} = $m;
3129                                }
3130                        } else {
3131                                my ($dir, $file) = ($f =~ m#^(.*?)/?([^/]+)$#);
3132                                delete $ut->{empty}->{$dir};
3133                                $ut->{empty}->{$f} = 1;
3134                        }
3135                }
3136                $pool->clear;
3137        }
3138        foreach (keys %amr) {
3139                libsvn_get_file($gui, $_, $rev, $amr{$_}, $ut);
3140                my ($d) = ($_ =~ m#^(.*?)/?(?:[^/]+)$#);
3141                delete $ut->{empty}->{$d};
3142        }
3143        unless (exists $ut->{dir_prop}->{''}) {
3144                my $pool = SVN::Pool->new;
3145                my (undef, undef, $props) = $SVN->get_dir('', $rev, $pool);
3146                %{$ut->{dir_prop}->{''}} = %$props;
3147                $pool->clear;
3148        }
3149        close $gui or croak $?;
3150        libsvn_log_entry($rev, $author, $date, $msg, [$last_commit], $ut);
3151}
3152
3153sub svn_grab_base_rev {
3154        defined(my $pid = open my $fh, '-|') or croak $!;
3155        if (!$pid) {
3156                open my $null, '>', '/dev/null' or croak $!;
3157                open STDERR, '>&', $null or croak $!;
3158                exec qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0"
3159                                                                or croak $!;
3160        }
3161        chomp(my $c = do { local $/; <$fh> });
3162        close $fh;
3163        if (defined $c && length $c) {
3164                my ($url, $rev, $uuid) = cmt_metadata($c);
3165                return ($rev, $c) if defined $rev;
3166        }
3167        if ($_no_metadata) {
3168                my $offset = -41; # from tail
3169                my $rl;
3170                open my $fh, '<', $REVDB or
3171                        die "--no-metadata specified and $REVDB not readable\n";
3172                seek $fh, $offset, 2;
3173                $rl = readline $fh;
3174                defined $rl or return (undef, undef);
3175                chomp $rl;
3176                while ($c ne $rl && tell $fh != 0) {
3177                        $offset -= 41;
3178                        seek $fh, $offset, 2;
3179                        $rl = readline $fh;
3180                        defined $rl or return (undef, undef);
3181                        chomp $rl;
3182                }
3183                my $rev = tell $fh;
3184                croak $! if ($rev < -1);
3185                $rev =  ($rev - 41) / 41;
3186                close $fh or croak $!;
3187                return ($rev, $c);
3188        }
3189        return (undef, undef);
3190}
3191
3192sub libsvn_parse_revision {
3193        my $base = shift;
3194        my $head = $SVN->get_latest_revnum();
3195        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
3196                return ($base + 1, $head) if (defined $base);
3197                return (0, $head);
3198        }
3199        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
3200        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
3201        if ($_revision =~ /^BASE:(\d+)$/) {
3202                return ($base + 1, $1) if (defined $base);
3203                return (0, $head);
3204        }
3205        return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
3206        die "revision argument: $_revision not understood by git-svn\n",
3207                "Try using the command-line svn client instead\n";
3208}
3209
3210sub libsvn_traverse {
3211        my ($gui, $pfx, $path, $rev, $files, $untracked) = @_;
3212        my $cwd = length $pfx ? "$pfx/$path" : $path;
3213        my $pool = SVN::Pool->new;
3214        $cwd =~ s#^\Q$SVN->{svn_path}\E##;
3215        my $nr = 0;
3216        my ($dirent, $r, $props) = $SVN->get_dir($cwd, $rev, $pool);
3217        %{$untracked->{dir_prop}->{$cwd}} = %$props;
3218        foreach my $d (keys %$dirent) {
3219                my $t = $dirent->{$d}->kind;
3220                if ($t == $SVN::Node::dir) {
3221                        my $i = libsvn_traverse($gui, $cwd, $d, $rev,
3222                                                $files, $untracked);
3223                        if ($i) {
3224                                $nr += $i;
3225                        } else {
3226                                $untracked->{empty}->{"$cwd/$d"} = 1;
3227                        }
3228                } elsif ($t == $SVN::Node::file) {
3229                        $nr++;
3230                        my $file = "$cwd/$d";
3231                        if (defined $files) {
3232                                push @$files, $file;
3233                        } else {
3234                                libsvn_get_file($gui, $file, $rev, 'A',
3235                                                $untracked);
3236                                my ($dir) = ($file =~ m#^(.*?)/?(?:[^/]+)$#);
3237                                delete $untracked->{empty}->{$dir};
3238                        }
3239                }
3240        }
3241        $pool->clear;
3242        $nr;
3243}
3244
3245sub libsvn_traverse_ignore {
3246        my ($fh, $path, $r) = @_;
3247        $path =~ s#^/+##g;
3248        my $pool = SVN::Pool->new;
3249        my ($dirent, undef, $props) = $SVN->get_dir($path, $r, $pool);
3250        my $p = $path;
3251        $p =~ s#^\Q$SVN->{svn_path}\E/##;
3252        print $fh length $p ? "\n# $p\n" : "\n# /\n";
3253        if (my $s = $props->{'svn:ignore'}) {
3254                $s =~ s/[\r\n]+/\n/g;
3255                chomp $s;
3256                if (length $p == 0) {
3257                        $s =~ s#\n#\n/$p#g;
3258                        print $fh "/$s\n";
3259                } else {
3260                        $s =~ s#\n#\n/$p/#g;
3261                        print $fh "/$p/$s\n";
3262                }
3263        }
3264        foreach (sort keys %$dirent) {
3265                next if $dirent->{$_}->kind != $SVN::Node::dir;
3266                libsvn_traverse_ignore($fh, "$path/$_", $r);
3267        }
3268        $pool->clear;
3269}
3270
3271sub revisions_eq {
3272        my ($path, $r0, $r1) = @_;
3273        return 1 if $r0 == $r1;
3274        my $nr = 0;
3275        if ($_use_lib) {
3276                # should be OK to use Pool here (r1 - r0) should be small
3277                my $pool = SVN::Pool->new;
3278                libsvn_get_log($SVN, [$path], $r0, $r1,
3279                                0, 0, 1, sub {$nr++}, $pool);
3280                $pool->clear;
3281        } else {
3282                my ($url, undef) = repo_path_split($SVN_URL);
3283                my $svn_log = svn_log_raw("$url/$path","-r$r0:$r1");
3284                while (next_log_entry($svn_log)) { $nr++ }
3285                close $svn_log->{fh};
3286        }
3287        return 0 if ($nr > 1);
3288        return 1;
3289}
3290
3291sub libsvn_find_parent_branch {
3292        my ($paths, $rev, $author, $date, $msg) = @_;
3293        my $svn_path = '/'.$SVN->{svn_path};
3294
3295        # look for a parent from another branch:
3296        my $i = $paths->{$svn_path} or return;
3297        my $branch_from = $i->copyfrom_path or return;
3298        my $r = $i->copyfrom_rev;
3299        print STDERR  "Found possible branch point: ",
3300                                "$branch_from => $svn_path, $r\n";
3301        $branch_from =~ s#^/##;
3302        my $l_map = {};
3303        read_url_paths_all($l_map, '', "$GIT_DIR/svn");
3304        my $url = $SVN->{repos_root};
3305        defined $l_map->{$url} or return;
3306        my $id = $l_map->{$url}->{$branch_from};
3307        if (!defined $id && $_follow_parent) {
3308                print STDERR "Following parent: $branch_from\@$r\n";
3309                # auto create a new branch and follow it
3310                $id = basename($branch_from);
3311                $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
3312                while (-r "$GIT_DIR/svn/$id") {
3313                        # just grow a tail if we're not unique enough :x
3314                        $id .= '-';
3315                }
3316        }
3317        return unless defined $id;
3318
3319        my ($r0, $parent) = find_rev_before($r,$id,1);
3320        if ($_follow_parent && (!defined $r0 || !defined $parent)) {
3321                defined(my $pid = fork) or croak $!;
3322                if (!$pid) {
3323                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
3324                        init_vars();
3325                        $SVN_URL = "$url/$branch_from";
3326                        $SVN = undef;
3327                        setup_git_svn();
3328                        # we can't assume SVN_URL exists at r+1:
3329                        $_revision = "0:$r";
3330                        fetch_lib();
3331                        exit 0;
3332                }
3333                waitpid $pid, 0;
3334                croak $? if $?;
3335                ($r0, $parent) = find_rev_before($r,$id,1);
3336        }
3337        return unless (defined $r0 && defined $parent);
3338        if (revisions_eq($branch_from, $r0, $r)) {
3339                unlink $GIT_SVN_INDEX;
3340                print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
3341                sys(qw/git-read-tree/, $parent);
3342                unless (libsvn_can_do_switch()) {
3343                        return libsvn_fetch_full($parent, $paths, $rev,
3344                                                $author, $date, $msg);
3345                }
3346                # do_switch works with svn/trunk >= r22312, but that is not
3347                # included with SVN 1.4.2 (the latest version at the moment),
3348                # so we can't rely on it.
3349                my $ra = libsvn_connect("$url/$branch_from");
3350                my $ed = SVN::Git::Fetcher->new({c => $parent, q => $_q});
3351                my $pool = SVN::Pool->new;
3352                my $reporter = $ra->do_switch($rev, '', 1, $SVN->{url},
3353                                              $ed, $pool);
3354                my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3355                $reporter->set_path('', $r0, 0, @lock, $pool);
3356                $reporter->finish_report($pool);
3357                $pool->clear;
3358                unless ($ed->{git_commit_ok}) {
3359                        die "SVN connection failed somewhere...\n";
3360                }
3361                return libsvn_log_entry($rev, $author, $date, $msg, [$parent]);
3362        }
3363        print STDERR "Nope, branch point not imported or unknown\n";
3364        return undef;
3365}
3366
3367sub libsvn_get_log {
3368        my ($ra, @args) = @_;
3369        $args[4]-- if $args[4] && $_xfer_delta && ! $_follow_parent;
3370        if ($SVN::Core::VERSION le '1.2.0') {
3371                splice(@args, 3, 1);
3372        }
3373        $ra->get_log(@args);
3374}
3375
3376sub libsvn_new_tree {
3377        if (my $log_entry = libsvn_find_parent_branch(@_)) {
3378                return $log_entry;
3379        }
3380        my ($paths, $rev, $author, $date, $msg) = @_;
3381        my $ut;
3382        if ($_xfer_delta) {
3383                my $pool = SVN::Pool->new;
3384                my $ed = SVN::Git::Fetcher->new({q => $_q});
3385                my $reporter = $SVN->do_update($rev, '', 1, $ed, $pool);
3386                my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3387                $reporter->set_path('', $rev, 1, @lock, $pool);
3388                $reporter->finish_report($pool);
3389                $pool->clear;
3390                unless ($ed->{git_commit_ok}) {
3391                        die "SVN connection failed somewhere...\n";
3392                }
3393                $ut = $ed;
3394        } else {
3395                $ut = { empty => {}, dir_prop => {}, file_prop => {} };
3396                open my $gui, '| git-update-index -z --index-info' or croak $!;
3397                libsvn_traverse($gui, '', $SVN->{svn_path}, $rev, undef, $ut);
3398                close $gui or croak $?;
3399        }
3400        libsvn_log_entry($rev, $author, $date, $msg, [], $ut);
3401}
3402
3403sub find_graft_path_commit {
3404        my ($tree_paths, $p1, $r1) = @_;
3405        foreach my $x (keys %$tree_paths) {
3406                next unless ($p1 =~ /^\Q$x\E/);
3407                my $i = $tree_paths->{$x};
3408                my ($r0, $parent) = find_rev_before($r1,$i,1);
3409                return $parent if (defined $r0 && $r0 == $r1);
3410                print STDERR "r$r1 of $i not imported\n";
3411                next;
3412        }
3413        return undef;
3414}
3415
3416sub find_graft_path_parents {
3417        my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
3418        foreach my $x (keys %$tree_paths) {
3419                next unless ($p0 =~ /^\Q$x\E/);
3420                my $i = $tree_paths->{$x};
3421                my ($r, $parent) = find_rev_before($r0, $i, 1);
3422                if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
3423                        my ($url_b, undef, $uuid_b) = cmt_metadata($c);
3424                        my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
3425                        next if ($url_a && $url_b && $url_a eq $url_b &&
3426                                                        $uuid_b eq $uuid_a);
3427                        $grafts->{$c}->{$parent} = 1;
3428                }
3429        }
3430}
3431
3432sub libsvn_graft_file_copies {
3433        my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
3434        foreach (keys %$paths) {
3435                my $i = $paths->{$_};
3436                my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
3437                                        $i->copyfrom_rev);
3438                next unless (defined $p0 && defined $r0);
3439
3440                my $p1 = $_;
3441                $p1 =~ s#^/##;
3442                $p0 =~ s#^/##;
3443                my $c = find_graft_path_commit($tree_paths, $p1, $rev);
3444                next unless $c;
3445                find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
3446        }
3447}
3448
3449sub set_index {
3450        my $old = $ENV{GIT_INDEX_FILE};
3451        $ENV{GIT_INDEX_FILE} = shift;
3452        return $old;
3453}
3454
3455sub restore_index {
3456        my ($old) = @_;
3457        if (defined $old) {
3458                $ENV{GIT_INDEX_FILE} = $old;
3459        } else {
3460                delete $ENV{GIT_INDEX_FILE};
3461        }
3462}
3463
3464sub libsvn_commit_cb {
3465        my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
3466        if ($_optimize_commits && $rev == ($r_last + 1)) {
3467                my $log = libsvn_log_entry($rev,$committer,$date,$msg);
3468                $log->{tree} = get_tree_from_treeish($c);
3469                my $cmt = git_commit($log, $cmt_last, $c);
3470                my @diff = safe_qx('git-diff-tree', $cmt, $c);
3471                if (@diff) {
3472                        print STDERR "Trees differ: $cmt $c\n",
3473                                        join('',@diff),"\n";
3474                        exit 1;
3475                }
3476        } else {
3477                fetch("$rev=$c");
3478        }
3479}
3480
3481sub libsvn_ls_fullurl {
3482        my $fullurl = shift;
3483        my $ra = libsvn_connect($fullurl);
3484        my @ret;
3485        my $pool = SVN::Pool->new;
3486        my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
3487        my ($dirent, undef, undef) = $ra->get_dir('', $r, $pool);
3488        foreach my $d (keys %$dirent) {
3489                if ($dirent->{$d}->kind == $SVN::Node::dir) {
3490                        push @ret, "$d/"; # add '/' for compat with cli svn
3491                }
3492        }
3493        $pool->clear;
3494        return @ret;
3495}
3496
3497
3498sub libsvn_skip_unknown_revs {
3499        my $err = shift;
3500        my $errno = $err->apr_err();
3501        # Maybe the branch we're tracking didn't
3502        # exist when the repo started, so it's
3503        # not an error if it doesn't, just continue
3504        #
3505        # Wonderfully consistent library, eh?
3506        # 160013 - svn:// and file://
3507        # 175002 - http(s)://
3508        # 175007 - http(s):// (this repo required authorization, too...)
3509        #   More codes may be discovered later...
3510        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3511                return;
3512        }
3513        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3514};
3515
3516# Tie::File seems to be prone to offset errors if revisions get sparse,
3517# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3518# one of my favorite modules is out :<  Next up would be one of the DBM
3519# modules, but I'm not sure which is most portable...  So I'll just
3520# go with something that's plain-text, but still capable of
3521# being randomly accessed.  So here's my ultra-simple fixed-width
3522# database.  All records are 40 characters + "\n", so it's easy to seek
3523# to a revision: (41 * rev) is the byte offset.
3524# A record of 40 0s denotes an empty revision.
3525# And yes, it's still pretty fast (faster than Tie::File).
3526sub revdb_set {
3527        my ($file, $rev, $commit) = @_;
3528        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
3529        open my $fh, '+<', $file or croak $!;
3530        my $offset = $rev * 41;
3531        # assume that append is the common case:
3532        seek $fh, 0, 2 or croak $!;
3533        my $pos = tell $fh;
3534        if ($pos < $offset) {
3535                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
3536        }
3537        seek $fh, $offset, 0 or croak $!;
3538        print $fh $commit,"\n";
3539        close $fh or croak $!;
3540}
3541
3542sub revdb_get {
3543        my ($file, $rev) = @_;
3544        my $ret;
3545        my $offset = $rev * 41;
3546        open my $fh, '<', $file or croak $!;
3547        seek $fh, $offset, 0;
3548        if (tell $fh == $offset) {
3549                $ret = readline $fh;
3550                if (defined $ret) {
3551                        chomp $ret;
3552                        $ret = undef if ($ret =~ /^0{40}$/);
3553                }
3554        }
3555        close $fh or croak $!;
3556        return $ret;
3557}
3558
3559sub copy_remote_ref {
3560        my $origin = $_cp_remote ? $_cp_remote : 'origin';
3561        my $ref = "refs/remotes/$GIT_SVN";
3562        if (safe_qx('git-ls-remote', $origin, $ref)) {
3563                sys(qw/git fetch/, $origin, "$ref:$ref");
3564        } elsif ($_cp_remote && !$_upgrade) {
3565                die "Unable to find remote reference: ",
3566                                "refs/remotes/$GIT_SVN on $origin\n";
3567        }
3568}
3569package SVN::Git::Fetcher;
3570use vars qw/@ISA/;
3571use strict;
3572use warnings;
3573use Carp qw/croak/;
3574use IO::File qw//;
3575
3576# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3577sub new {
3578        my ($class, $git_svn) = @_;
3579        my $self = SVN::Delta::Editor->new;
3580        bless $self, $class;
3581        open my $gui, '| git-update-index -z --index-info' or croak $!;
3582        $self->{gui} = $gui;
3583        $self->{c} = $git_svn->{c} if exists $git_svn->{c};
3584        $self->{q} = $git_svn->{q};
3585        $self->{empty} = {};
3586        $self->{dir_prop} = {};
3587        $self->{file_prop} = {};
3588        $self->{absent_dir} = {};
3589        $self->{absent_file} = {};
3590        require Digest::MD5;
3591        $self;
3592}
3593
3594sub open_root {
3595        { path => '' };
3596}
3597
3598sub open_directory {
3599        my ($self, $path, $pb, $rev) = @_;
3600        { path => $path };
3601}
3602
3603sub delete_entry {
3604        my ($self, $path, $rev, $pb) = @_;
3605        my $t = process_rm($self->{gui}, $self->{c}, $path, $self->{q});
3606        $self->{empty}->{$path} = 0 if $t == $SVN::Node::dir;
3607        undef;
3608}
3609
3610sub open_file {
3611        my ($self, $path, $pb, $rev) = @_;
3612        my ($mode, $blob) = (safe_qx('git-ls-tree',$self->{c},'--',$path)
3613                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3614        unless (defined $mode && defined $blob) {
3615                die "$path was not found in commit $self->{c} (r$rev)\n";
3616        }
3617        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3618          pool => SVN::Pool->new, action => 'M' };
3619}
3620
3621sub add_file {
3622        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3623        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3624        delete $self->{empty}->{$dir};
3625        { path => $path, mode_a => 100644, mode_b => 100644,
3626          pool => SVN::Pool->new, action => 'A' };
3627}
3628
3629sub add_directory {
3630        my ($self, $path, $cp_path, $cp_rev) = @_;
3631        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3632        delete $self->{empty}->{$dir};
3633        $self->{empty}->{$path} = 1;
3634        { path => $path };
3635}
3636
3637sub change_dir_prop {
3638        my ($self, $db, $prop, $value) = @_;
3639        $self->{dir_prop}->{$db->{path}} ||= {};
3640        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3641        undef;
3642}
3643
3644sub absent_directory {
3645        my ($self, $path, $pb) = @_;
3646        $self->{absent_dir}->{$pb->{path}} ||= [];
3647        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3648        undef;
3649}
3650
3651sub absent_file {
3652        my ($self, $path, $pb) = @_;
3653        $self->{absent_file}->{$pb->{path}} ||= [];
3654        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3655        undef;
3656}
3657
3658sub change_file_prop {
3659        my ($self, $fb, $prop, $value) = @_;
3660        if ($prop eq 'svn:executable') {
3661                if ($fb->{mode_b} != 120000) {
3662                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3663                }
3664        } elsif ($prop eq 'svn:special') {
3665                $fb->{mode_b} = defined $value ? 120000 : 100644;
3666        } else {
3667                $self->{file_prop}->{$fb->{path}} ||= {};
3668                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3669        }
3670        undef;
3671}
3672
3673sub apply_textdelta {
3674        my ($self, $fb, $exp) = @_;
3675        my $fh = IO::File->new_tmpfile;
3676        $fh->autoflush(1);
3677        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3678        # (but $base does not,) so dup() it for reading in close_file
3679        open my $dup, '<&', $fh or croak $!;
3680        my $base = IO::File->new_tmpfile;
3681        $base->autoflush(1);
3682        if ($fb->{blob}) {
3683                defined (my $pid = fork) or croak $!;
3684                if (!$pid) {
3685                        open STDOUT, '>&', $base or croak $!;
3686                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
3687                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
3688                }
3689                waitpid $pid, 0;
3690                croak $? if $?;
3691
3692                if (defined $exp) {
3693                        seek $base, 0, 0 or croak $!;
3694                        my $md5 = Digest::MD5->new;
3695                        $md5->addfile($base);
3696                        my $got = $md5->hexdigest;
3697                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3698                            "expected: $exp\n",
3699                            "     got: $got\n" if ($got ne $exp);
3700                }
3701        }
3702        seek $base, 0, 0 or croak $!;
3703        $fb->{fh} = $dup;
3704        $fb->{base} = $base;
3705        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
3706}
3707
3708sub close_file {
3709        my ($self, $fb, $exp) = @_;
3710        my $hash;
3711        my $path = $fb->{path};
3712        if (my $fh = $fb->{fh}) {
3713                seek($fh, 0, 0) or croak $!;
3714                my $md5 = Digest::MD5->new;
3715                $md5->addfile($fh);
3716                my $got = $md5->hexdigest;
3717                die "Checksum mismatch: $path\n",
3718                    "expected: $exp\n    got: $got\n" if ($got ne $exp);
3719                seek($fh, 0, 0) or croak $!;
3720                if ($fb->{mode_b} == 120000) {
3721                        read($fh, my $buf, 5) == 5 or croak $!;
3722                        $buf eq 'link ' or die "$path has mode 120000",
3723                                               "but is not a link\n";
3724                }
3725                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
3726                if (!$pid) {
3727                        open STDIN, '<&', $fh or croak $!;
3728                        exec qw/git-hash-object -w --stdin/ or croak $!;
3729                }
3730                chomp($hash = do { local $/; <$out> });
3731                close $out or croak $!;
3732                close $fh or croak $!;
3733                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3734                close $fb->{base} or croak $!;
3735        } else {
3736                $hash = $fb->{blob} or die "no blob information\n";
3737        }
3738        $fb->{pool}->clear;
3739        my $gui = $self->{gui};
3740        print $gui "$fb->{mode_b} $hash\t$path\0" or croak $!;
3741        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
3742        undef;
3743}
3744
3745sub abort_edit {
3746        my $self = shift;
3747        close $self->{gui};
3748        $self->SUPER::abort_edit(@_);
3749}
3750
3751sub close_edit {
3752        my $self = shift;
3753        close $self->{gui} or croak $!;
3754        $self->{git_commit_ok} = 1;
3755        $self->SUPER::close_edit(@_);
3756}
3757
3758package SVN::Git::Editor;
3759use vars qw/@ISA/;
3760use strict;
3761use warnings;
3762use Carp qw/croak/;
3763use IO::File;
3764
3765sub new {
3766        my $class = shift;
3767        my $git_svn = shift;
3768        my $self = SVN::Delta::Editor->new(@_);
3769        bless $self, $class;
3770        foreach (qw/svn_path c r ra /) {
3771                die "$_ required!\n" unless (defined $git_svn->{$_});
3772                $self->{$_} = $git_svn->{$_};
3773        }
3774        $self->{pool} = SVN::Pool->new;
3775        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3776        $self->{rm} = { };
3777        require Digest::MD5;
3778        return $self;
3779}
3780
3781sub split_path {
3782        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3783}
3784
3785sub repo_path {
3786        (defined $_[1] && length $_[1]) ? $_[1] : ''
3787}
3788
3789sub url_path {
3790        my ($self, $path) = @_;
3791        $self->{ra}->{url} . '/' . $self->repo_path($path);
3792}
3793
3794sub rmdirs {
3795        my ($self, $q) = @_;
3796        my $rm = $self->{rm};
3797        delete $rm->{''}; # we never delete the url we're tracking
3798        return unless %$rm;
3799
3800        foreach (keys %$rm) {
3801                my @d = split m#/#, $_;
3802                my $c = shift @d;
3803                $rm->{$c} = 1;
3804                while (@d) {
3805                        $c .= '/' . shift @d;
3806                        $rm->{$c} = 1;
3807                }
3808        }
3809        delete $rm->{$self->{svn_path}};
3810        delete $rm->{''}; # we never delete the url we're tracking
3811        return unless %$rm;
3812
3813        defined(my $pid = open my $fh,'-|') or croak $!;
3814        if (!$pid) {
3815                exec qw/git-ls-tree --name-only -r -z/, $self->{c} or croak $!;
3816        }
3817        local $/ = "\0";
3818        while (<$fh>) {
3819                chomp;
3820                my @dn = split m#/#, $_;
3821                while (pop @dn) {
3822                        delete $rm->{join '/', @dn};
3823                }
3824                unless (%$rm) {
3825                        close $fh;
3826                        return;
3827                }
3828        }
3829        close $fh;
3830
3831        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3832        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3833                $self->close_directory($bat->{$d}, $p);
3834                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3835                print "\tD+\t/$d/\n" unless $q;
3836                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3837                delete $bat->{$d};
3838        }
3839}
3840
3841sub open_or_add_dir {
3842        my ($self, $full_path, $baton) = @_;
3843        my $p = SVN::Pool->new;
3844        my $t = $self->{ra}->check_path($full_path, $self->{r}, $p);
3845        $p->clear;
3846        if ($t == $SVN::Node::none) {
3847                return $self->add_directory($full_path, $baton,
3848                                                undef, -1, $self->{pool});
3849        } elsif ($t == $SVN::Node::dir) {
3850                return $self->open_directory($full_path, $baton,
3851                                                $self->{r}, $self->{pool});
3852        }
3853        print STDERR "$full_path already exists in repository at ",
3854                "r$self->{r} and it is not a directory (",
3855                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3856        exit 1;
3857}
3858
3859sub ensure_path {
3860        my ($self, $path) = @_;
3861        my $bat = $self->{bat};
3862        $path = $self->repo_path($path);
3863        return $bat->{''} unless (length $path);
3864        my @p = split m#/+#, $path;
3865        my $c = shift @p;
3866        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3867        while (@p) {
3868                my $c0 = $c;
3869                $c .= '/' . shift @p;
3870                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3871        }
3872        return $bat->{$c};
3873}
3874
3875sub A {
3876        my ($self, $m, $q) = @_;
3877        my ($dir, $file) = split_path($m->{file_b});
3878        my $pbat = $self->ensure_path($dir);
3879        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3880                                        undef, -1);
3881        print "\tA\t$m->{file_b}\n" unless $q;
3882        $self->chg_file($fbat, $m);
3883        $self->close_file($fbat,undef,$self->{pool});
3884}
3885
3886sub C {
3887        my ($self, $m, $q) = @_;
3888        my ($dir, $file) = split_path($m->{file_b});
3889        my $pbat = $self->ensure_path($dir);
3890        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3891                                $self->url_path($m->{file_a}), $self->{r});
3892        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $q;
3893        $self->chg_file($fbat, $m);
3894        $self->close_file($fbat,undef,$self->{pool});
3895}
3896
3897sub delete_entry {
3898        my ($self, $path, $pbat) = @_;
3899        my $rpath = $self->repo_path($path);
3900        my ($dir, $file) = split_path($rpath);
3901        $self->{rm}->{$dir} = 1;
3902        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3903}
3904
3905sub R {
3906        my ($self, $m, $q) = @_;
3907        my ($dir, $file) = split_path($m->{file_b});
3908        my $pbat = $self->ensure_path($dir);
3909        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3910                                $self->url_path($m->{file_a}), $self->{r});
3911        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $q;
3912        $self->chg_file($fbat, $m);
3913        $self->close_file($fbat,undef,$self->{pool});
3914
3915        ($dir, $file) = split_path($m->{file_a});
3916        $pbat = $self->ensure_path($dir);
3917        $self->delete_entry($m->{file_a}, $pbat);
3918}
3919
3920sub M {
3921        my ($self, $m, $q) = @_;
3922        my ($dir, $file) = split_path($m->{file_b});
3923        my $pbat = $self->ensure_path($dir);
3924        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3925                                $pbat,$self->{r},$self->{pool});
3926        print "\t$m->{chg}\t$m->{file_b}\n" unless $q;
3927        $self->chg_file($fbat, $m);
3928        $self->close_file($fbat,undef,$self->{pool});
3929}
3930
3931sub T { shift->M(@_) }
3932
3933sub change_file_prop {
3934        my ($self, $fbat, $pname, $pval) = @_;
3935        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3936}
3937
3938sub chg_file {
3939        my ($self, $fbat, $m) = @_;
3940        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3941                $self->change_file_prop($fbat,'svn:executable','*');
3942        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3943                $self->change_file_prop($fbat,'svn:executable',undef);
3944        }
3945        my $fh = IO::File->new_tmpfile or croak $!;
3946        if ($m->{mode_b} =~ /^120/) {
3947                print $fh 'link ' or croak $!;
3948                $self->change_file_prop($fbat,'svn:special','*');
3949        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3950                $self->change_file_prop($fbat,'svn:special',undef);
3951        }
3952        defined(my $pid = fork) or croak $!;
3953        if (!$pid) {
3954                open STDOUT, '>&', $fh or croak $!;
3955                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3956        }
3957        waitpid $pid, 0;
3958        croak $? if $?;
3959        $fh->flush == 0 or croak $!;
3960        seek $fh, 0, 0 or croak $!;
3961
3962        my $md5 = Digest::MD5->new;
3963        $md5->addfile($fh) or croak $!;
3964        seek $fh, 0, 0 or croak $!;
3965
3966        my $exp = $md5->hexdigest;
3967        my $pool = SVN::Pool->new;
3968        my $atd = $self->apply_textdelta($fbat, undef, $pool);
3969        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3970        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3971        $pool->clear;
3972
3973        close $fh or croak $!;
3974}
3975
3976sub D {
3977        my ($self, $m, $q) = @_;
3978        my ($dir, $file) = split_path($m->{file_b});
3979        my $pbat = $self->ensure_path($dir);
3980        print "\tD\t$m->{file_b}\n" unless $q;
3981        $self->delete_entry($m->{file_b}, $pbat);
3982}
3983
3984sub close_edit {
3985        my ($self) = @_;
3986        my ($p,$bat) = ($self->{pool}, $self->{bat});
3987        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3988                $self->close_directory($bat->{$_}, $p);
3989        }
3990        $self->SUPER::close_edit($p);
3991        $p->clear;
3992}
3993
3994sub abort_edit {
3995        my ($self) = @_;
3996        $self->SUPER::abort_edit($self->{pool});
3997        $self->{pool}->clear;
3998}
3999
4000__END__
4001
4002Data structures:
4003
4004$svn_log hashref (as returned by svn_log_raw)
4005{
4006        fh => file handle of the log file,
4007        state => state of the log file parser (sep/msg/rev/msg_start...)
4008}
4009
4010$log_msg hashref as returned by next_log_entry($svn_log)
4011{
4012        msg => 'whitespace-formatted log entry
4013',                                              # trailing newline is preserved
4014        revision => '8',                        # integer
4015        date => '2004-02-24T17:01:44.108345Z',  # commit date
4016        author => 'committer name'
4017};
4018
4019
4020@mods = array of diff-index line hashes, each element represents one line
4021        of diff-index output
4022
4023diff-index line ($m hash)
4024{
4025        mode_a => first column of diff-index output, no leading ':',
4026        mode_b => second column of diff-index output,
4027        sha1_b => sha1sum of the final blob,
4028        chg => change type [MCRADT],
4029        file_a => original file name of a file (iff chg is 'C' or 'R')
4030        file_b => new/current file name of a file (any chg)
4031}
4032;
4033
4034# retval of read_url_paths{,_all}();
4035$l_map = {
4036        # repository root url
4037        'https://svn.musicpd.org' => {
4038                # repository path               # GIT_SVN_ID
4039                'mpd/trunk'             =>      'trunk',
4040                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
4041        },
4042}
4043
4044Notes:
4045        I don't trust the each() function on unless I created %hash myself
4046        because the internal iterator may not have started at base.