git-svn.perlon commit commit walkers: setup_ident() to record correct committer in ref-log. (19c4588)
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $SVN_URL $SVN_INFO $SVN_WC $SVN_UUID
   8                $GIT_SVN_INDEX $GIT_SVN
   9                $GIT_DIR $GIT_SVN_DIR $REVDB/;
  10$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  11$VERSION = '@@GIT_VERSION@@';
  12
  13use Cwd qw/abs_path/;
  14$GIT_DIR = abs_path($ENV{GIT_DIR} || '.git');
  15$ENV{GIT_DIR} = $GIT_DIR;
  16
  17my $LC_ALL = $ENV{LC_ALL};
  18my $TZ = $ENV{TZ};
  19# make sure the svn binary gives consistent output between locales and TZs:
  20$ENV{TZ} = 'UTC';
  21$ENV{LC_ALL} = 'C';
  22$| = 1; # unbuffer STDOUT
  23
  24# If SVN:: library support is added, please make the dependencies
  25# optional and preserve the capability to use the command-line client.
  26# use eval { require SVN::... } to make it lazy load
  27# We don't use any modules not in the standard Perl distribution:
  28use Carp qw/croak/;
  29use IO::File qw//;
  30use File::Basename qw/dirname basename/;
  31use File::Path qw/mkpath/;
  32use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  33use File::Spec qw//;
  34use POSIX qw/strftime/;
  35use IPC::Open3;
  36use Memoize;
  37memoize('revisions_eq');
  38memoize('cmt_metadata');
  39memoize('get_commit_time');
  40
  41my ($SVN_PATH, $SVN, $SVN_LOG, $_use_lib);
  42$_use_lib = 1 unless $ENV{GIT_SVN_NO_LIB};
  43libsvn_load();
  44my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  45my $sha1 = qr/[a-f\d]{40}/;
  46my $sha1_short = qr/[a-f\d]{4,40}/;
  47my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
  48        $_find_copies_harder, $_l, $_cp_similarity, $_cp_remote,
  49        $_repack, $_repack_nr, $_repack_flags, $_q,
  50        $_message, $_file, $_follow_parent, $_no_metadata,
  51        $_template, $_shared, $_no_default_regex, $_no_graft_copy,
  52        $_limit, $_verbose, $_incremental, $_oneline, $_l_fmt, $_show_commit,
  53        $_version, $_upgrade, $_authors, $_branch_all_refs, @_opt_m);
  54my (@_branch_from, %tree_map, %users, %rusers, %equiv);
  55my ($_svn_co_url_revs, $_svn_pg_peg_revs);
  56my @repo_path_split_cache;
  57
  58my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
  59                'branch|b=s' => \@_branch_from,
  60                'follow-parent|follow' => \$_follow_parent,
  61                'branch-all-refs|B' => \$_branch_all_refs,
  62                'authors-file|A=s' => \$_authors,
  63                'repack:i' => \$_repack,
  64                'no-metadata' => \$_no_metadata,
  65                'quiet|q' => \$_q,
  66                'repack-flags|repack-args|repack-opts=s' => \$_repack_flags);
  67
  68my ($_trunk, $_tags, $_branches);
  69my %multi_opts = ( 'trunk|T=s' => \$_trunk,
  70                'tags|t=s' => \$_tags,
  71                'branches|b=s' => \$_branches );
  72my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
  73my %cmt_opts = ( 'edit|e' => \$_edit,
  74                'rmdir' => \$_rmdir,
  75                'find-copies-harder' => \$_find_copies_harder,
  76                'l=i' => \$_l,
  77                'copy-similarity|C=i'=> \$_cp_similarity
  78);
  79
  80# yes, 'native' sets "\n".  Patches to fix this for non-*nix systems welcome:
  81my %EOL = ( CR => "\015", LF => "\012", CRLF => "\015\012", native => "\012" );
  82
  83my %cmd = (
  84        fetch => [ \&fetch, "Download new revisions from SVN",
  85                        { 'revision|r=s' => \$_revision, %fc_opts } ],
  86        init => [ \&init, "Initialize a repo for tracking" .
  87                          " (requires URL argument)",
  88                          \%init_opts ],
  89        commit => [ \&commit, "Commit git revisions to SVN",
  90                        {       'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
  91        'show-ignore' => [ \&show_ignore, "Show svn:ignore listings",
  92                        { 'revision|r=i' => \$_revision } ],
  93        rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
  94                        { 'no-ignore-externals' => \$_no_ignore_ext,
  95                          'copy-remote|remote=s' => \$_cp_remote,
  96                          'upgrade' => \$_upgrade } ],
  97        'graft-branches' => [ \&graft_branches,
  98                        'Detect merges/branches from already imported history',
  99                        { 'merge-rx|m' => \@_opt_m,
 100                          'branch|b=s' => \@_branch_from,
 101                          'branch-all-refs|B' => \$_branch_all_refs,
 102                          'no-default-regex' => \$_no_default_regex,
 103                          'no-graft-copy' => \$_no_graft_copy } ],
 104        'multi-init' => [ \&multi_init,
 105                        'Initialize multiple trees (like git-svnimport)',
 106                        { %multi_opts, %fc_opts } ],
 107        'multi-fetch' => [ \&multi_fetch,
 108                        'Fetch multiple trees (like git-svnimport)',
 109                        \%fc_opts ],
 110        'log' => [ \&show_log, 'Show commit logs',
 111                        { 'limit=i' => \$_limit,
 112                          'revision|r=s' => \$_revision,
 113                          'verbose|v' => \$_verbose,
 114                          'incremental' => \$_incremental,
 115                          'oneline' => \$_oneline,
 116                          'show-commit' => \$_show_commit,
 117                          'authors-file|A=s' => \$_authors,
 118                        } ],
 119        'commit-diff' => [ \&commit_diff, 'Commit a diff between two trees',
 120                        { 'message|m=s' => \$_message,
 121                          'file|F=s' => \$_file,
 122                        %cmt_opts } ],
 123);
 124
 125my $cmd;
 126for (my $i = 0; $i < @ARGV; $i++) {
 127        if (defined $cmd{$ARGV[$i]}) {
 128                $cmd = $ARGV[$i];
 129                splice @ARGV, $i, 1;
 130                last;
 131        }
 132};
 133
 134my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 135
 136read_repo_config(\%opts);
 137my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
 138                                'version|V' => \$_version,
 139                                'id|i=s' => \$GIT_SVN);
 140exit 1 if (!$rv && $cmd ne 'log');
 141
 142set_default_vals();
 143usage(0) if $_help;
 144version() if $_version;
 145usage(1) unless defined $cmd;
 146init_vars();
 147load_authors() if $_authors;
 148load_all_refs() if $_branch_all_refs;
 149svn_compat_check() unless $_use_lib;
 150migration_check() unless $cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/;
 151$cmd{$cmd}->[0]->(@ARGV);
 152exit 0;
 153
 154####################### primary functions ######################
 155sub usage {
 156        my $exit = shift || 0;
 157        my $fd = $exit ? \*STDERR : \*STDOUT;
 158        print $fd <<"";
 159git-svn - bidirectional operations between a single Subversion tree and git
 160Usage: $0 <command> [options] [arguments]\n
 161
 162        print $fd "Available commands:\n" unless $cmd;
 163
 164        foreach (sort keys %cmd) {
 165                next if $cmd && $cmd ne $_;
 166                print $fd '  ',pack('A13',$_),$cmd{$_}->[1],"\n";
 167                foreach (keys %{$cmd{$_}->[2]}) {
 168                        # prints out arguments as they should be passed:
 169                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 170                        print $fd ' ' x 17, join(', ', map { length $_ > 1 ?
 171                                                        "--$_" : "-$_" }
 172                                                split /\|/,$_)," $x\n";
 173                }
 174        }
 175        print $fd <<"";
 176\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 177arbitrary identifier if you're tracking multiple SVN branches/repositories in
 178one git repository and want to keep them separate.  See git-svn(1) for more
 179information.
 180
 181        exit $exit;
 182}
 183
 184sub version {
 185        print "git-svn version $VERSION\n";
 186        exit 0;
 187}
 188
 189sub rebuild {
 190        if (quiet_run(qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0")) {
 191                copy_remote_ref();
 192        }
 193        $SVN_URL = shift or undef;
 194        my $newest_rev = 0;
 195        if ($_upgrade) {
 196                sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD");
 197        } else {
 198                check_upgrade_needed();
 199        }
 200
 201        my $pid = open(my $rev_list,'-|');
 202        defined $pid or croak $!;
 203        if ($pid == 0) {
 204                exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!;
 205        }
 206        my $latest;
 207        while (<$rev_list>) {
 208                chomp;
 209                my $c = $_;
 210                croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 211                my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
 212                next if (!@commit); # skip merges
 213                my ($url, $rev, $uuid) = extract_metadata($commit[$#commit]);
 214                if (!$rev || !$uuid) {
 215                        croak "Unable to extract revision or UUID from ",
 216                                "$c, $commit[$#commit]\n";
 217                }
 218
 219                # if we merged or otherwise started elsewhere, this is
 220                # how we break out of it
 221                next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
 222                next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
 223
 224                unless (defined $latest) {
 225                        if (!$SVN_URL && !$url) {
 226                                croak "SVN repository location required: $url\n";
 227                        }
 228                        $SVN_URL ||= $url;
 229                        $SVN_UUID ||= $uuid;
 230                        setup_git_svn();
 231                        $latest = $rev;
 232                }
 233                revdb_set($REVDB, $rev, $c);
 234                print "r$rev = $c\n";
 235                $newest_rev = $rev if ($rev > $newest_rev);
 236        }
 237        close $rev_list or croak $?;
 238
 239        goto out if $_use_lib;
 240        if (!chdir $SVN_WC) {
 241                svn_cmd_checkout($SVN_URL, $latest, $SVN_WC);
 242                chdir $SVN_WC or croak $!;
 243        }
 244
 245        $pid = fork;
 246        defined $pid or croak $!;
 247        if ($pid == 0) {
 248                my @svn_up = qw(svn up);
 249                push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 250                sys(@svn_up,"-r$newest_rev");
 251                $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 252                index_changes();
 253                exec('git-write-tree') or croak $!;
 254        }
 255        waitpid $pid, 0;
 256        croak $? if $?;
 257out:
 258        if ($_upgrade) {
 259                print STDERR <<"";
 260Keeping deprecated refs/head/$GIT_SVN-HEAD for now.  Please remove it
 261when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN
 262
 263        }
 264}
 265
 266sub init {
 267        my $url = shift or die "SVN repository location required " .
 268                                "as a command-line argument\n";
 269        $url =~ s!/+$!!; # strip trailing slash
 270
 271        if (my $repo_path = shift) {
 272                unless (-d $repo_path) {
 273                        mkpath([$repo_path]);
 274                }
 275                $GIT_DIR = $ENV{GIT_DIR} = $repo_path . "/.git";
 276                init_vars();
 277        }
 278
 279        $SVN_URL = $url;
 280        unless (-d $GIT_DIR) {
 281                my @init_db = ('git-init-db');
 282                push @init_db, "--template=$_template" if defined $_template;
 283                push @init_db, "--shared" if defined $_shared;
 284                sys(@init_db);
 285        }
 286        setup_git_svn();
 287}
 288
 289sub fetch {
 290        check_upgrade_needed();
 291        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 292        my $ret = $_use_lib ? fetch_lib(@_) : fetch_cmd(@_);
 293        if ($ret->{commit} && quiet_run(qw(git-rev-parse --verify
 294                                                refs/heads/master^0))) {
 295                sys(qw(git-update-ref refs/heads/master),$ret->{commit});
 296        }
 297        return $ret;
 298}
 299
 300sub fetch_cmd {
 301        my (@parents) = @_;
 302        my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 303        unless ($_revision) {
 304                $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 305        }
 306        push @log_args, "-r$_revision";
 307        push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 308
 309        my $svn_log = svn_log_raw(@log_args);
 310
 311        my $base = next_log_entry($svn_log) or croak "No base revision!\n";
 312        # don't need last_revision from grab_base_rev() because
 313        # user could've specified a different revision to skip (they
 314        # didn't want to import certain revisions into git for whatever
 315        # reason, so trust $base->{revision} instead.
 316        my (undef, $last_commit) = svn_grab_base_rev();
 317        unless (-d $SVN_WC) {
 318                svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC);
 319                chdir $SVN_WC or croak $!;
 320                read_uuid();
 321                $last_commit = git_commit($base, @parents);
 322                assert_tree($last_commit);
 323        } else {
 324                chdir $SVN_WC or croak $!;
 325                read_uuid();
 326                # looks like a user manually cp'd and svn switch'ed
 327                unless ($last_commit) {
 328                        sys(qw/svn revert -R ./);
 329                        assert_svn_wc_clean($base->{revision});
 330                        $last_commit = git_commit($base, @parents);
 331                        assert_tree($last_commit);
 332                }
 333        }
 334        my @svn_up = qw(svn up);
 335        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 336        my $last = $base;
 337        while (my $log_msg = next_log_entry($svn_log)) {
 338                if ($last->{revision} >= $log_msg->{revision}) {
 339                        croak "Out of order: last >= current: ",
 340                                "$last->{revision} >= $log_msg->{revision}\n";
 341                }
 342                # Revert is needed for cases like:
 343                # https://svn.musicpd.org/Jamming/trunk (r166:167), but
 344                # I can't seem to reproduce something like that on a test...
 345                sys(qw/svn revert -R ./);
 346                assert_svn_wc_clean($last->{revision});
 347                sys(@svn_up,"-r$log_msg->{revision}");
 348                $last_commit = git_commit($log_msg, $last_commit, @parents);
 349                $last = $log_msg;
 350        }
 351        close $svn_log->{fh};
 352        $last->{commit} = $last_commit;
 353        return $last;
 354}
 355
 356sub fetch_lib {
 357        my (@parents) = @_;
 358        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 359        my $repo;
 360        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 361        $SVN_LOG ||= libsvn_connect($repo);
 362        $SVN ||= libsvn_connect($repo);
 363        my ($last_rev, $last_commit) = svn_grab_base_rev();
 364        my ($base, $head) = libsvn_parse_revision($last_rev);
 365        if ($base > $head) {
 366                return { revision => $last_rev, commit => $last_commit }
 367        }
 368        my $index = set_index($GIT_SVN_INDEX);
 369
 370        # limit ourselves and also fork() since get_log won't release memory
 371        # after processing a revision and SVN stuff seems to leak
 372        my $inc = 1000;
 373        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
 374        read_uuid();
 375        if (defined $last_commit) {
 376                unless (-e $GIT_SVN_INDEX) {
 377                        sys(qw/git-read-tree/, $last_commit);
 378                }
 379                chomp (my $x = `git-write-tree`);
 380                my ($y) = (`git-cat-file commit $last_commit`
 381                                                        =~ /^tree ($sha1)/m);
 382                if ($y ne $x) {
 383                        unlink $GIT_SVN_INDEX or croak $!;
 384                        sys(qw/git-read-tree/, $last_commit);
 385                }
 386                chomp ($x = `git-write-tree`);
 387                if ($y ne $x) {
 388                        print STDERR "trees ($last_commit) $y != $x\n",
 389                                 "Something is seriously wrong...\n";
 390                }
 391        }
 392        while (1) {
 393                # fork, because using SVN::Pool with get_log() still doesn't
 394                # seem to help enough to keep memory usage down.
 395                defined(my $pid = fork) or croak $!;
 396                if (!$pid) {
 397                        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
 398
 399                        # Yes I'm perfectly aware that the fourth argument
 400                        # below is the limit revisions number.  Unfortunately
 401                        # performance sucks with it enabled, so it's much
 402                        # faster to fetch revision ranges instead of relying
 403                        # on the limiter.
 404                        libsvn_get_log($SVN_LOG, '/'.$SVN_PATH,
 405                                        $min, $max, 0, 1, 1,
 406                                sub {
 407                                        my $log_msg;
 408                                        if ($last_commit) {
 409                                                $log_msg = libsvn_fetch(
 410                                                        $last_commit, @_);
 411                                                $last_commit = git_commit(
 412                                                        $log_msg,
 413                                                        $last_commit,
 414                                                        @parents);
 415                                        } else {
 416                                                $log_msg = libsvn_new_tree(@_);
 417                                                $last_commit = git_commit(
 418                                                        $log_msg, @parents);
 419                                        }
 420                                });
 421                        exit 0;
 422                }
 423                waitpid $pid, 0;
 424                croak $? if $?;
 425                ($last_rev, $last_commit) = svn_grab_base_rev();
 426                last if ($max >= $head);
 427                $min = $max + 1;
 428                $max += $inc;
 429                $max = $head if ($max > $head);
 430        }
 431        restore_index($index);
 432        return { revision => $last_rev, commit => $last_commit };
 433}
 434
 435sub commit {
 436        my (@commits) = @_;
 437        check_upgrade_needed();
 438        if ($_stdin || !@commits) {
 439                print "Reading from stdin...\n";
 440                @commits = ();
 441                while (<STDIN>) {
 442                        if (/\b($sha1_short)\b/o) {
 443                                unshift @commits, $1;
 444                        }
 445                }
 446        }
 447        my @revs;
 448        foreach my $c (@commits) {
 449                chomp(my @tmp = safe_qx('git-rev-parse',$c));
 450                if (scalar @tmp == 1) {
 451                        push @revs, $tmp[0];
 452                } elsif (scalar @tmp > 1) {
 453                        push @revs, reverse (safe_qx('git-rev-list',@tmp));
 454                } else {
 455                        die "Failed to rev-parse $c\n";
 456                }
 457        }
 458        chomp @revs;
 459        $_use_lib ? commit_lib(@revs) : commit_cmd(@revs);
 460        print "Done committing ",scalar @revs," revisions to SVN\n";
 461}
 462
 463sub commit_cmd {
 464        my (@revs) = @_;
 465
 466        chdir $SVN_WC or croak "Unable to chdir $SVN_WC: $!\n";
 467        my $info = svn_info('.');
 468        my $fetched = fetch();
 469        if ($info->{Revision} != $fetched->{revision}) {
 470                print STDERR "There are new revisions that were fetched ",
 471                                "and need to be merged (or acknowledged) ",
 472                                "before committing.\n";
 473                exit 1;
 474        }
 475        $info = svn_info('.');
 476        read_uuid($info);
 477        my $last = $fetched;
 478        foreach my $c (@revs) {
 479                my $mods = svn_checkout_tree($last, $c);
 480                if (scalar @$mods == 0) {
 481                        print "Skipping, no changes detected\n";
 482                        next;
 483                }
 484                $last = svn_commit_tree($last, $c);
 485        }
 486}
 487
 488sub commit_lib {
 489        my (@revs) = @_;
 490        my ($r_last, $cmt_last) = svn_grab_base_rev();
 491        defined $r_last or die "Must have an existing revision to commit\n";
 492        my $fetched = fetch();
 493        if ($r_last != $fetched->{revision}) {
 494                print STDERR "There are new revisions that were fetched ",
 495                                "and need to be merged (or acknowledged) ",
 496                                "before committing.\n",
 497                                "last rev: $r_last\n",
 498                                " current: $fetched->{revision}\n";
 499                exit 1;
 500        }
 501        read_uuid();
 502        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 503        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
 504
 505        set_svn_commit_env();
 506        foreach my $c (@revs) {
 507                my $log_msg = get_commit_message($c, $commit_msg);
 508
 509                # fork for each commit because there's a memory leak I
 510                # can't track down... (it's probably in the SVN code)
 511                defined(my $pid = open my $fh, '-|') or croak $!;
 512                if (!$pid) {
 513                        my $ed = SVN::Git::Editor->new(
 514                                        {       r => $r_last,
 515                                                ra => $SVN,
 516                                                c => $c,
 517                                                svn_path => $SVN_PATH
 518                                        },
 519                                        $SVN->get_commit_editor(
 520                                                $log_msg->{msg},
 521                                                sub {
 522                                                        libsvn_commit_cb(
 523                                                                @_, $c,
 524                                                                $log_msg->{msg},
 525                                                                $r_last,
 526                                                                $cmt_last)
 527                                                },
 528                                                @lock)
 529                                        );
 530                        my $mods = libsvn_checkout_tree($cmt_last, $c, $ed);
 531                        if (@$mods == 0) {
 532                                print "No changes\nr$r_last = $cmt_last\n";
 533                                $ed->abort_edit;
 534                        } else {
 535                                $ed->close_edit;
 536                        }
 537                        exit 0;
 538                }
 539                my ($r_new, $cmt_new, $no);
 540                while (<$fh>) {
 541                        print $_;
 542                        chomp;
 543                        if (/^r(\d+) = ($sha1)$/o) {
 544                                ($r_new, $cmt_new) = ($1, $2);
 545                        } elsif ($_ eq 'No changes') {
 546                                $no = 1;
 547                        }
 548                }
 549                close $fh or croak $?;
 550                if (! defined $r_new && ! defined $cmt_new) {
 551                        unless ($no) {
 552                                die "Failed to parse revision information\n";
 553                        }
 554                } else {
 555                        ($r_last, $cmt_last) = ($r_new, $cmt_new);
 556                }
 557        }
 558        $ENV{LC_ALL} = 'C';
 559        unlink $commit_msg;
 560}
 561
 562sub show_ignore {
 563        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 564        $_use_lib ? show_ignore_lib() : show_ignore_cmd();
 565}
 566
 567sub show_ignore_cmd {
 568        require File::Find or die $!;
 569        if (defined $_revision) {
 570                die "-r/--revision option doesn't work unless the Perl SVN ",
 571                        "libraries are used\n";
 572        }
 573        chdir $SVN_WC or croak $!;
 574        my %ign;
 575        File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 576                s#^\./##;
 577                @{$ign{$_}} = svn_propget_base('svn:ignore', $_);
 578                }}, no_chdir=>1},'.');
 579
 580        print "\n# /\n";
 581        foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 582        delete $ign{'.'};
 583        foreach my $i (sort keys %ign) {
 584                print "\n# ",$i,"\n";
 585                foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 586        }
 587}
 588
 589sub show_ignore_lib {
 590        my $repo;
 591        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 592        $SVN ||= libsvn_connect($repo);
 593        my $r = defined $_revision ? $_revision : $SVN->get_latest_revnum;
 594        libsvn_traverse_ignore(\*STDOUT, $SVN_PATH, $r);
 595}
 596
 597sub graft_branches {
 598        my $gr_file = "$GIT_DIR/info/grafts";
 599        my ($grafts, $comments) = read_grafts($gr_file);
 600        my $gr_sha1;
 601
 602        if (%$grafts) {
 603                # temporarily disable our grafts file to make this idempotent
 604                chomp($gr_sha1 = safe_qx(qw/git-hash-object -w/,$gr_file));
 605                rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
 606        }
 607
 608        my $l_map = read_url_paths();
 609        my @re = map { qr/$_/is } @_opt_m if @_opt_m;
 610        unless ($_no_default_regex) {
 611                push @re, (qr/\b(?:merge|merging|merged)\s+with\s+([\w\.\-]+)/i,
 612                        qr/\b(?:merge|merging|merged)\s+([\w\.\-]+)/i,
 613                        qr/\b(?:from|of)\s+([\w\.\-]+)/i );
 614        }
 615        foreach my $u (keys %$l_map) {
 616                if (@re) {
 617                        foreach my $p (keys %{$l_map->{$u}}) {
 618                                graft_merge_msg($grafts,$l_map,$u,$p,@re);
 619                        }
 620                }
 621                unless ($_no_graft_copy) {
 622                        if ($_use_lib) {
 623                                graft_file_copy_lib($grafts,$l_map,$u);
 624                        } else {
 625                                graft_file_copy_cmd($grafts,$l_map,$u);
 626                        }
 627                }
 628        }
 629        graft_tree_joins($grafts);
 630
 631        write_grafts($grafts, $comments, $gr_file);
 632        unlink "$gr_file~$gr_sha1" if $gr_sha1;
 633}
 634
 635sub multi_init {
 636        my $url = shift;
 637        $_trunk ||= 'trunk';
 638        $_trunk =~ s#/+$##;
 639        $url =~ s#/+$## if $url;
 640        if ($_trunk !~ m#^[a-z\+]+://#) {
 641                $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#);
 642                unless ($url) {
 643                        print STDERR "E: '$_trunk' is not a complete URL ",
 644                                "and a separate URL is not specified\n";
 645                        exit 1;
 646                }
 647                $_trunk = $url . $_trunk;
 648        }
 649        if ($GIT_SVN eq 'git-svn') {
 650                print "GIT_SVN_ID set to 'trunk' for $_trunk\n";
 651                $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
 652        }
 653        init_vars();
 654        init($_trunk);
 655        complete_url_ls_init($url, $_branches, '--branches/-b', '');
 656        complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/');
 657}
 658
 659sub multi_fetch {
 660        # try to do trunk first, since branches/tags
 661        # may be descended from it.
 662        if (-e "$GIT_DIR/svn/trunk/info/url") {
 663                fetch_child_id('trunk', @_);
 664        }
 665        rec_fetch('', "$GIT_DIR/svn", @_);
 666}
 667
 668sub show_log {
 669        my (@args) = @_;
 670        my ($r_min, $r_max);
 671        my $r_last = -1; # prevent dupes
 672        rload_authors() if $_authors;
 673        if (defined $TZ) {
 674                $ENV{TZ} = $TZ;
 675        } else {
 676                delete $ENV{TZ};
 677        }
 678        if (defined $_revision) {
 679                if ($_revision =~ /^(\d+):(\d+)$/) {
 680                        ($r_min, $r_max) = ($1, $2);
 681                } elsif ($_revision =~ /^\d+$/) {
 682                        $r_min = $r_max = $_revision;
 683                } else {
 684                        print STDERR "-r$_revision is not supported, use ",
 685                                "standard \'git log\' arguments instead\n";
 686                        exit 1;
 687                }
 688        }
 689
 690        my $pid = open(my $log,'-|');
 691        defined $pid or croak $!;
 692        if (!$pid) {
 693                exec(git_svn_log_cmd($r_min,$r_max), @args) or croak $!;
 694        }
 695        setup_pager();
 696        my (@k, $c, $d);
 697
 698        while (<$log>) {
 699                if (/^commit ($sha1_short)/o) {
 700                        my $cmt = $1;
 701                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
 702                                $r_last = $c->{r};
 703                                process_commit($c, $r_min, $r_max, \@k) or
 704                                                                goto out;
 705                        }
 706                        $d = undef;
 707                        $c = { c => $cmt };
 708                } elsif (/^author (.+) (\d+) ([\-\+]?\d+)$/) {
 709                        get_author_info($c, $1, $2, $3);
 710                } elsif (/^(?:tree|parent|committer) /) {
 711                        # ignore
 712                } elsif (/^:\d{6} \d{6} $sha1_short/o) {
 713                        push @{$c->{raw}}, $_;
 714                } elsif (/^diff /) {
 715                        $d = 1;
 716                        push @{$c->{diff}}, $_;
 717                } elsif ($d) {
 718                        push @{$c->{diff}}, $_;
 719                } elsif (/^    (git-svn-id:.+)$/) {
 720                        (undef, $c->{r}, undef) = extract_metadata($1);
 721                } elsif (s/^    //) {
 722                        push @{$c->{l}}, $_;
 723                }
 724        }
 725        if ($c && defined $c->{r} && $c->{r} != $r_last) {
 726                $r_last = $c->{r};
 727                process_commit($c, $r_min, $r_max, \@k);
 728        }
 729        if (@k) {
 730                my $swap = $r_max;
 731                $r_max = $r_min;
 732                $r_min = $swap;
 733                process_commit($_, $r_min, $r_max) foreach reverse @k;
 734        }
 735out:
 736        close $log;
 737        print '-' x72,"\n" unless $_incremental || $_oneline;
 738}
 739
 740sub commit_diff_usage {
 741        print STDERR "Usage: $0 commit-diff <tree-ish> <tree-ish> [<URL>]\n";
 742        exit 1
 743}
 744
 745sub commit_diff {
 746        if (!$_use_lib) {
 747                print STDERR "commit-diff must be used with SVN libraries\n";
 748                exit 1;
 749        }
 750        my $ta = shift or commit_diff_usage();
 751        my $tb = shift or commit_diff_usage();
 752        if (!eval { $SVN_URL = shift || file_to_s("$GIT_SVN_DIR/info/url") }) {
 753                print STDERR "Needed URL or usable git-svn id command-line\n";
 754                commit_diff_usage();
 755        }
 756        if (defined $_message && defined $_file) {
 757                print STDERR "Both --message/-m and --file/-F specified ",
 758                                "for the commit message.\n",
 759                                "I have no idea what you mean\n";
 760                exit 1;
 761        }
 762        if (defined $_file) {
 763                $_message = file_to_s($_file);
 764        } else {
 765                $_message ||= get_commit_message($tb,
 766                                        "$GIT_DIR/.svn-commit.tmp.$$")->{msg};
 767        }
 768        my $repo;
 769        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 770        $SVN_LOG ||= libsvn_connect($repo);
 771        $SVN ||= libsvn_connect($repo);
 772        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 773        my $ed = SVN::Git::Editor->new({        r => $SVN->get_latest_revnum,
 774                                                ra => $SVN, c => $tb,
 775                                                svn_path => $SVN_PATH
 776                                        },
 777                                $SVN->get_commit_editor($_message,
 778                                        sub {print "Committed $_[0]\n"},@lock)
 779                                );
 780        my $mods = libsvn_checkout_tree($ta, $tb, $ed);
 781        if (@$mods == 0) {
 782                print "No changes\n$ta == $tb\n";
 783                $ed->abort_edit;
 784        } else {
 785                $ed->close_edit;
 786        }
 787}
 788
 789########################### utility functions #########################
 790
 791sub cmt_showable {
 792        my ($c) = @_;
 793        return 1 if defined $c->{r};
 794        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
 795                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
 796                my @msg = safe_qx(qw/git-cat-file commit/, $c->{c});
 797                shift @msg while ($msg[0] ne "\n");
 798                shift @msg;
 799                @{$c->{l}} = grep !/^git-svn-id: /, @msg;
 800
 801                (undef, $c->{r}, undef) = extract_metadata(
 802                                (grep(/^git-svn-id: /, @msg))[-1]);
 803        }
 804        return defined $c->{r};
 805}
 806
 807sub git_svn_log_cmd {
 808        my ($r_min, $r_max) = @_;
 809        my @cmd = (qw/git-log --abbrev-commit --pretty=raw
 810                        --default/, "refs/remotes/$GIT_SVN");
 811        push @cmd, '--summary' if $_verbose;
 812        return @cmd unless defined $r_max;
 813        if ($r_max == $r_min) {
 814                push @cmd, '--max-count=1';
 815                if (my $c = revdb_get($REVDB, $r_max)) {
 816                        push @cmd, $c;
 817                }
 818        } else {
 819                my ($c_min, $c_max);
 820                $c_max = revdb_get($REVDB, $r_max);
 821                $c_min = revdb_get($REVDB, $r_min);
 822                if ($c_min && $c_max) {
 823                        if ($r_max > $r_max) {
 824                                push @cmd, "$c_min..$c_max";
 825                        } else {
 826                                push @cmd, "$c_max..$c_min";
 827                        }
 828                } elsif ($r_max > $r_min) {
 829                        push @cmd, $c_max;
 830                } else {
 831                        push @cmd, $c_min;
 832                }
 833        }
 834        return @cmd;
 835}
 836
 837sub fetch_child_id {
 838        my $id = shift;
 839        print "Fetching $id\n";
 840        my $ref = "$GIT_DIR/refs/remotes/$id";
 841        defined(my $pid = open my $fh, '-|') or croak $!;
 842        if (!$pid) {
 843                $_repack = undef;
 844                $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 845                init_vars();
 846                fetch(@_);
 847                exit 0;
 848        }
 849        while (<$fh>) {
 850                print $_;
 851                check_repack() if (/^r\d+ = $sha1/);
 852        }
 853        close $fh or croak $?;
 854}
 855
 856sub rec_fetch {
 857        my ($pfx, $p, @args) = @_;
 858        my @dir;
 859        foreach (sort <$p/*>) {
 860                if (-r "$_/info/url") {
 861                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
 862                        my $id = $pfx . basename $_;
 863                        next if $id eq 'trunk';
 864                        fetch_child_id($id, @args);
 865                } elsif (-d $_) {
 866                        push @dir, $_;
 867                }
 868        }
 869        foreach (@dir) {
 870                my $x = $_;
 871                $x =~ s!^\Q$GIT_DIR\E/svn/!!;
 872                rec_fetch($x, $_);
 873        }
 874}
 875
 876sub complete_url_ls_init {
 877        my ($url, $var, $switch, $pfx) = @_;
 878        unless ($var) {
 879                print STDERR "W: $switch not specified\n";
 880                return;
 881        }
 882        $var =~ s#/+$##;
 883        if ($var !~ m#^[a-z\+]+://#) {
 884                $var = '/' . $var if ($var !~ m#^/#);
 885                unless ($url) {
 886                        print STDERR "E: '$var' is not a complete URL ",
 887                                "and a separate URL is not specified\n";
 888                        exit 1;
 889                }
 890                $var = $url . $var;
 891        }
 892        chomp(my @ls = $_use_lib ? libsvn_ls_fullurl($var)
 893                                : safe_qx(qw/svn ls --non-interactive/, $var));
 894        my $old = $GIT_SVN;
 895        defined(my $pid = fork) or croak $!;
 896        if (!$pid) {
 897                foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
 898                        $u =~ s#/+$##;
 899                        if ($u !~ m!\Q$var\E/(.+)$!) {
 900                                print STDERR "W: Unrecognized URL: $u\n";
 901                                die "This should never happen\n";
 902                        }
 903                        my $id = $pfx.$1;
 904                        print "init $u => $id\n";
 905                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 906                        init_vars();
 907                        init($u);
 908                }
 909                exit 0;
 910        }
 911        waitpid $pid, 0;
 912        croak $? if $?;
 913}
 914
 915sub common_prefix {
 916        my $paths = shift;
 917        my %common;
 918        foreach (@$paths) {
 919                my @tmp = split m#/#, $_;
 920                my $p = '';
 921                while (my $x = shift @tmp) {
 922                        $p .= "/$x";
 923                        $common{$p} ||= 0;
 924                        $common{$p}++;
 925                }
 926        }
 927        foreach (sort {length $b <=> length $a} keys %common) {
 928                if ($common{$_} == @$paths) {
 929                        return $_;
 930                }
 931        }
 932        return '';
 933}
 934
 935# grafts set here are 'stronger' in that they're based on actual tree
 936# matches, and won't be deleted from merge-base checking in write_grafts()
 937sub graft_tree_joins {
 938        my $grafts = shift;
 939        map_tree_joins() if (@_branch_from && !%tree_map);
 940        return unless %tree_map;
 941
 942        git_svn_each(sub {
 943                my $i = shift;
 944                defined(my $pid = open my $fh, '-|') or croak $!;
 945                if (!$pid) {
 946                        exec qw/git-rev-list --pretty=raw/,
 947                                        "refs/remotes/$i" or croak $!;
 948                }
 949                while (<$fh>) {
 950                        next unless /^commit ($sha1)$/o;
 951                        my $c = $1;
 952                        my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
 953                        next unless $tree_map{$t};
 954
 955                        my $l;
 956                        do {
 957                                $l = readline $fh;
 958                        } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
 959
 960                        my ($s, $tz) = ($1, $2);
 961                        if ($tz =~ s/^\+//) {
 962                                $s += tz_to_s_offset($tz);
 963                        } elsif ($tz =~ s/^\-//) {
 964                                $s -= tz_to_s_offset($tz);
 965                        }
 966
 967                        my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
 968
 969                        foreach my $p (@{$tree_map{$t}}) {
 970                                next if $p eq $c;
 971                                my $mb = eval {
 972                                        safe_qx('git-merge-base', $c, $p)
 973                                };
 974                                next unless ($@ || $?);
 975                                if (defined $r_a) {
 976                                        # see if SVN says it's a relative
 977                                        my ($url_b, $r_b, $uuid_b) =
 978                                                        cmt_metadata($p);
 979                                        next if (defined $url_b &&
 980                                                        defined $url_a &&
 981                                                        ($url_a eq $url_b) &&
 982                                                        ($uuid_a eq $uuid_b));
 983                                        if ($uuid_a eq $uuid_b) {
 984                                                if ($r_b < $r_a) {
 985                                                        $grafts->{$c}->{$p} = 2;
 986                                                        next;
 987                                                } elsif ($r_b > $r_a) {
 988                                                        $grafts->{$p}->{$c} = 2;
 989                                                        next;
 990                                                }
 991                                        }
 992                                }
 993                                my $ct = get_commit_time($p);
 994                                if ($ct < $s) {
 995                                        $grafts->{$c}->{$p} = 2;
 996                                } elsif ($ct > $s) {
 997                                        $grafts->{$p}->{$c} = 2;
 998                                }
 999                                # what should we do when $ct == $s ?
1000                        }
1001                }
1002                close $fh or croak $?;
1003        });
1004}
1005
1006# this isn't funky-filename safe, but good enough for now...
1007sub graft_file_copy_cmd {
1008        my ($grafts, $l_map, $u) = @_;
1009        my $paths = $l_map->{$u};
1010        my $pfx = common_prefix([keys %$paths]);
1011        $SVN_URL ||= $u.$pfx;
1012        my $pid = open my $fh, '-|';
1013        defined $pid or croak $!;
1014        unless ($pid) {
1015                my @exec = qw/svn log -v/;
1016                push @exec, "-r$_revision" if defined $_revision;
1017                exec @exec, $u.$pfx or croak $!;
1018        }
1019        my ($r, $mp) = (undef, undef);
1020        while (<$fh>) {
1021                chomp;
1022                if (/^\-{72}$/) {
1023                        $mp = $r = undef;
1024                } elsif (/^r(\d+) \| /) {
1025                        $r = $1 unless defined $r;
1026                } elsif (/^Changed paths:/) {
1027                        $mp = 1;
1028                } elsif ($mp && m#^   [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) {
1029                        my ($p1, $p0, $r0) = ($1, $2, $3);
1030                        my $c = find_graft_path_commit($paths, $p1, $r);
1031                        next unless $c;
1032                        find_graft_path_parents($grafts, $paths, $c, $p0, $r0);
1033                }
1034        }
1035}
1036
1037sub graft_file_copy_lib {
1038        my ($grafts, $l_map, $u) = @_;
1039        my $tree_paths = $l_map->{$u};
1040        my $pfx = common_prefix([keys %$tree_paths]);
1041        my ($repo, $path) = repo_path_split($u.$pfx);
1042        $SVN_LOG ||= libsvn_connect($repo);
1043        $SVN ||= libsvn_connect($repo);
1044
1045        my ($base, $head) = libsvn_parse_revision();
1046        my $inc = 1000;
1047        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
1048        my $eh = $SVN::Error::handler;
1049        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
1050        while (1) {
1051                my $pool = SVN::Pool->new;
1052                libsvn_get_log($SVN_LOG, "/$path", $min, $max, 0, 1, 1,
1053                        sub {
1054                                libsvn_graft_file_copies($grafts, $tree_paths,
1055                                                        $path, @_);
1056                        }, $pool);
1057                $pool->clear;
1058                last if ($max >= $head);
1059                $min = $max + 1;
1060                $max += $inc;
1061                $max = $head if ($max > $head);
1062        }
1063        $SVN::Error::handler = $eh;
1064}
1065
1066sub process_merge_msg_matches {
1067        my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
1068        my (@strong, @weak);
1069        foreach (@matches) {
1070                # merging with ourselves is not interesting
1071                next if $_ eq $p;
1072                if ($l_map->{$u}->{$_}) {
1073                        push @strong, $_;
1074                } else {
1075                        push @weak, $_;
1076                }
1077        }
1078        foreach my $w (@weak) {
1079                last if @strong;
1080                # no exact match, use branch name as regexp.
1081                my $re = qr/\Q$w\E/i;
1082                foreach (keys %{$l_map->{$u}}) {
1083                        if (/$re/) {
1084                                push @strong, $l_map->{$u}->{$_};
1085                                last;
1086                        }
1087                }
1088                last if @strong;
1089                $w = basename($w);
1090                $re = qr/\Q$w\E/i;
1091                foreach (keys %{$l_map->{$u}}) {
1092                        if (/$re/) {
1093                                push @strong, $l_map->{$u}->{$_};
1094                                last;
1095                        }
1096                }
1097        }
1098        my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
1099                                        \s(?:[a-f\d\-]+)$/xsm);
1100        unless (defined $rev) {
1101                ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
1102                                        \@(?:[a-f\d\-]+)/xsm);
1103                return unless defined $rev;
1104        }
1105        foreach my $m (@strong) {
1106                my ($r0, $s0) = find_rev_before($rev, $m, 1);
1107                $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
1108        }
1109}
1110
1111sub graft_merge_msg {
1112        my ($grafts, $l_map, $u, $p, @re) = @_;
1113
1114        my $x = $l_map->{$u}->{$p};
1115        my $rl = rev_list_raw($x);
1116        while (my $c = next_rev_list_entry($rl)) {
1117                foreach my $re (@re) {
1118                        my (@br) = ($c->{m} =~ /$re/g);
1119                        next unless @br;
1120                        process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
1121                }
1122        }
1123}
1124
1125sub read_uuid {
1126        return if $SVN_UUID;
1127        if ($_use_lib) {
1128                my $pool = SVN::Pool->new;
1129                $SVN_UUID = $SVN->get_uuid($pool);
1130                $pool->clear;
1131        } else {
1132                my $info = shift || svn_info('.');
1133                $SVN_UUID = $info->{'Repository UUID'} or
1134                                        croak "Repository UUID unreadable\n";
1135        }
1136}
1137
1138sub quiet_run {
1139        my $pid = fork;
1140        defined $pid or croak $!;
1141        if (!$pid) {
1142                open my $null, '>', '/dev/null' or croak $!;
1143                open STDERR, '>&', $null or croak $!;
1144                open STDOUT, '>&', $null or croak $!;
1145                exec @_ or croak $!;
1146        }
1147        waitpid $pid, 0;
1148        return $?;
1149}
1150
1151sub repo_path_split {
1152        my $full_url = shift;
1153        $full_url =~ s#/+$##;
1154
1155        foreach (@repo_path_split_cache) {
1156                if ($full_url =~ s#$_##) {
1157                        my $u = $1;
1158                        $full_url =~ s#^/+##;
1159                        return ($u, $full_url);
1160                }
1161        }
1162
1163        my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i);
1164        $path =~ s#^/+##;
1165        my @paths = split(m#/+#, $path);
1166
1167        if ($_use_lib) {
1168                while (1) {
1169                        $SVN = libsvn_connect($url);
1170                        last if (defined $SVN &&
1171                                defined eval { $SVN->get_latest_revnum });
1172                        my $n = shift @paths || last;
1173                        $url .= "/$n";
1174                }
1175        } else {
1176                while (quiet_run(qw/svn ls --non-interactive/, $url)) {
1177                        my $n = shift @paths || last;
1178                        $url .= "/$n";
1179                }
1180        }
1181        push @repo_path_split_cache, qr/^(\Q$url\E)/;
1182        $path = join('/',@paths);
1183        return ($url, $path);
1184}
1185
1186sub setup_git_svn {
1187        defined $SVN_URL or croak "SVN repository location required\n";
1188        unless (-d $GIT_DIR) {
1189                croak "GIT_DIR=$GIT_DIR does not exist!\n";
1190        }
1191        mkpath([$GIT_SVN_DIR]);
1192        mkpath(["$GIT_SVN_DIR/info"]);
1193        open my $fh, '>>',$REVDB or croak $!;
1194        close $fh;
1195        s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1196
1197}
1198
1199sub assert_svn_wc_clean {
1200        return if $_use_lib;
1201        my ($svn_rev) = @_;
1202        croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
1203        my $lcr = svn_info('.')->{'Last Changed Rev'};
1204        if ($svn_rev != $lcr) {
1205                print STDERR "Checking for copy-tree ... ";
1206                my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
1207                                                "-r$lcr:$svn_rev")));
1208                if (@diff) {
1209                        croak "Nope!  Expected r$svn_rev, got r$lcr\n";
1210                } else {
1211                        print STDERR "OK!\n";
1212                }
1213        }
1214        my @status = grep(!/^Performing status on external/,(`svn status`));
1215        @status = grep(!/^\s*$/,@status);
1216        if (scalar @status) {
1217                print STDERR "Tree ($SVN_WC) is not clean:\n";
1218                print STDERR $_ foreach @status;
1219                croak;
1220        }
1221}
1222
1223sub get_tree_from_treeish {
1224        my ($treeish) = @_;
1225        croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1226        chomp(my $type = `git-cat-file -t $treeish`);
1227        my $expected;
1228        while ($type eq 'tag') {
1229                chomp(($treeish, $type) = `git-cat-file tag $treeish`);
1230        }
1231        if ($type eq 'commit') {
1232                $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
1233                ($expected) = ($expected =~ /^tree ($sha1)$/);
1234                die "Unable to get tree from $treeish\n" unless $expected;
1235        } elsif ($type eq 'tree') {
1236                $expected = $treeish;
1237        } else {
1238                die "$treeish is a $type, expected tree, tag or commit\n";
1239        }
1240        return $expected;
1241}
1242
1243sub assert_tree {
1244        return if $_use_lib;
1245        my ($treeish) = @_;
1246        my $expected = get_tree_from_treeish($treeish);
1247
1248        my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
1249        if (-e $tmpindex) {
1250                unlink $tmpindex or croak $!;
1251        }
1252        my $old_index = set_index($tmpindex);
1253        index_changes(1);
1254        chomp(my $tree = `git-write-tree`);
1255        restore_index($old_index);
1256        if ($tree ne $expected) {
1257                croak "Tree mismatch, Got: $tree, Expected: $expected\n";
1258        }
1259        unlink $tmpindex;
1260}
1261
1262sub parse_diff_tree {
1263        my $diff_fh = shift;
1264        local $/ = "\0";
1265        my $state = 'meta';
1266        my @mods;
1267        while (<$diff_fh>) {
1268                chomp $_; # this gets rid of the trailing "\0"
1269                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1270                                        $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1271                        push @mods, {   mode_a => $1, mode_b => $2,
1272                                        sha1_b => $3, chg => $4 };
1273                        if ($4 =~ /^(?:C|R)$/) {
1274                                $state = 'file_a';
1275                        } else {
1276                                $state = 'file_b';
1277                        }
1278                } elsif ($state eq 'file_a') {
1279                        my $x = $mods[$#mods] or croak "Empty array\n";
1280                        if ($x->{chg} !~ /^(?:C|R)$/) {
1281                                croak "Error parsing $_, $x->{chg}\n";
1282                        }
1283                        $x->{file_a} = $_;
1284                        $state = 'file_b';
1285                } elsif ($state eq 'file_b') {
1286                        my $x = $mods[$#mods] or croak "Empty array\n";
1287                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1288                                croak "Error parsing $_, $x->{chg}\n";
1289                        }
1290                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1291                                croak "Error parsing $_, $x->{chg}\n";
1292                        }
1293                        $x->{file_b} = $_;
1294                        $state = 'meta';
1295                } else {
1296                        croak "Error parsing $_\n";
1297                }
1298        }
1299        close $diff_fh or croak $?;
1300
1301        return \@mods;
1302}
1303
1304sub svn_check_prop_executable {
1305        my $m = shift;
1306        return if -l $m->{file_b};
1307        if ($m->{mode_b} =~ /755$/) {
1308                chmod((0755 &~ umask),$m->{file_b}) or croak $!;
1309                if ($m->{mode_a} !~ /755$/) {
1310                        sys(qw(svn propset svn:executable 1), $m->{file_b});
1311                }
1312                -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
1313        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
1314                sys(qw(svn propdel svn:executable), $m->{file_b});
1315                chmod((0644 &~ umask),$m->{file_b}) or croak $!;
1316                -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
1317        }
1318}
1319
1320sub svn_ensure_parent_path {
1321        my $dir_b = dirname(shift);
1322        svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
1323        mkpath([$dir_b]) unless (-d $dir_b);
1324        sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
1325}
1326
1327sub precommit_check {
1328        my $mods = shift;
1329        my (%rm_file, %rmdir_check, %added_check);
1330
1331        my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
1332        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1333                if ($m->{chg} eq 'R') {
1334                        if (-d $m->{file_b}) {
1335                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1336                        }
1337                        # dir/$file => dir/file/$file
1338                        my $dirname = dirname($m->{file_b});
1339                        while ($dirname ne File::Spec->curdir) {
1340                                if ($dirname ne $m->{file_a}) {
1341                                        $dirname = dirname($dirname);
1342                                        next;
1343                                }
1344                                err_file_to_dir("$m->{file_a} => $m->{file_b}");
1345                        }
1346                        # baz/zzz => baz (baz is a file)
1347                        $dirname = dirname($m->{file_a});
1348                        while ($dirname ne File::Spec->curdir) {
1349                                if ($dirname ne $m->{file_b}) {
1350                                        $dirname = dirname($dirname);
1351                                        next;
1352                                }
1353                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1354                        }
1355                }
1356                if ($m->{chg} =~ /^(D|R)$/) {
1357                        my $t = $1 eq 'D' ? 'file_b' : 'file_a';
1358                        $rm_file{ $m->{$t} } = 1;
1359                        my $dirname = dirname( $m->{$t} );
1360                        my $basename = basename( $m->{$t} );
1361                        $rmdir_check{$dirname}->{$basename} = 1;
1362                } elsif ($m->{chg} =~ /^(?:A|C)$/) {
1363                        if (-d $m->{file_b}) {
1364                                err_dir_to_file($m->{file_b});
1365                        }
1366                        my $dirname = dirname( $m->{file_b} );
1367                        my $basename = basename( $m->{file_b} );
1368                        $added_check{$dirname}->{$basename} = 1;
1369                        while ($dirname ne File::Spec->curdir) {
1370                                if ($rm_file{$dirname}) {
1371                                        err_file_to_dir($m->{file_b});
1372                                }
1373                                $dirname = dirname $dirname;
1374                        }
1375                }
1376        }
1377        return (\%rmdir_check, \%added_check);
1378
1379        sub err_dir_to_file {
1380                my $file = shift;
1381                print STDERR "Node change from directory to file ",
1382                                "is not supported by Subversion: ",$file,"\n";
1383                exit 1;
1384        }
1385        sub err_file_to_dir {
1386                my $file = shift;
1387                print STDERR "Node change from file to directory ",
1388                                "is not supported by Subversion: ",$file,"\n";
1389                exit 1;
1390        }
1391}
1392
1393
1394sub get_diff {
1395        my ($from, $treeish) = @_;
1396        assert_tree($from);
1397        print "diff-tree $from $treeish\n";
1398        my $pid = open my $diff_fh, '-|';
1399        defined $pid or croak $!;
1400        if ($pid == 0) {
1401                my @diff_tree = qw(git-diff-tree -z -r);
1402                if ($_cp_similarity) {
1403                        push @diff_tree, "-C$_cp_similarity";
1404                } else {
1405                        push @diff_tree, '-C';
1406                }
1407                push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1408                push @diff_tree, "-l$_l" if defined $_l;
1409                exec(@diff_tree, $from, $treeish) or croak $!;
1410        }
1411        return parse_diff_tree($diff_fh);
1412}
1413
1414sub svn_checkout_tree {
1415        my ($from, $treeish) = @_;
1416        my $mods = get_diff($from->{commit}, $treeish);
1417        return $mods unless (scalar @$mods);
1418        my ($rm, $add) = precommit_check($mods);
1419
1420        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1421        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1422                if ($m->{chg} eq 'C') {
1423                        svn_ensure_parent_path( $m->{file_b} );
1424                        sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
1425                        apply_mod_line_blob($m);
1426                        svn_check_prop_executable($m);
1427                } elsif ($m->{chg} eq 'D') {
1428                        sys(qw(svn rm --force), $m->{file_b});
1429                } elsif ($m->{chg} eq 'R') {
1430                        svn_ensure_parent_path( $m->{file_b} );
1431                        sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
1432                        apply_mod_line_blob($m);
1433                        svn_check_prop_executable($m);
1434                } elsif ($m->{chg} eq 'M') {
1435                        apply_mod_line_blob($m);
1436                        svn_check_prop_executable($m);
1437                } elsif ($m->{chg} eq 'T') {
1438                        sys(qw(svn rm --force),$m->{file_b});
1439                        apply_mod_line_blob($m);
1440                        sys(qw(svn add), $m->{file_b});
1441                        svn_check_prop_executable($m);
1442                } elsif ($m->{chg} eq 'A') {
1443                        svn_ensure_parent_path( $m->{file_b} );
1444                        apply_mod_line_blob($m);
1445                        sys(qw(svn add), $m->{file_b});
1446                        svn_check_prop_executable($m);
1447                } else {
1448                        croak "Invalid chg: $m->{chg}\n";
1449                }
1450        }
1451
1452        assert_tree($treeish);
1453        if ($_rmdir) { # remove empty directories
1454                handle_rmdir($rm, $add);
1455        }
1456        assert_tree($treeish);
1457        return $mods;
1458}
1459
1460sub libsvn_checkout_tree {
1461        my ($from, $treeish, $ed) = @_;
1462        my $mods = get_diff($from, $treeish);
1463        return $mods unless (scalar @$mods);
1464        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1465        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1466                my $f = $m->{chg};
1467                if (defined $o{$f}) {
1468                        $ed->$f($m, $_q);
1469                } else {
1470                        croak "Invalid change type: $f\n";
1471                }
1472        }
1473        $ed->rmdirs($_q) if $_rmdir;
1474        return $mods;
1475}
1476
1477# svn ls doesn't work with respect to the current working tree, but what's
1478# in the repository.  There's not even an option for it... *sigh*
1479# (added files don't show up and removed files remain in the ls listing)
1480sub svn_ls_current {
1481        my ($dir, $rm, $add) = @_;
1482        chomp(my @ls = safe_qx('svn','ls',$dir));
1483        my @ret = ();
1484        foreach (@ls) {
1485                s#/$##; # trailing slashes are evil
1486                push @ret, $_ unless $rm->{$dir}->{$_};
1487        }
1488        if (exists $add->{$dir}) {
1489                push @ret, keys %{$add->{$dir}};
1490        }
1491        return \@ret;
1492}
1493
1494sub handle_rmdir {
1495        my ($rm, $add) = @_;
1496
1497        foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
1498                my $ls = svn_ls_current($dir, $rm, $add);
1499                next if (scalar @$ls);
1500                sys(qw(svn rm --force),$dir);
1501
1502                my $dn = dirname $dir;
1503                $rm->{ $dn }->{ basename $dir } = 1;
1504                $ls = svn_ls_current($dn, $rm, $add);
1505                while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
1506                        sys(qw(svn rm --force),$dn);
1507                        $dir = basename $dn;
1508                        $dn = dirname $dn;
1509                        $rm->{ $dn }->{ $dir } = 1;
1510                        $ls = svn_ls_current($dn, $rm, $add);
1511                }
1512        }
1513}
1514
1515sub get_commit_message {
1516        my ($commit, $commit_msg) = (@_);
1517        my %log_msg = ( msg => '' );
1518        open my $msg, '>', $commit_msg or croak $!;
1519
1520        chomp(my $type = `git-cat-file -t $commit`);
1521        if ($type eq 'commit' || $type eq 'tag') {
1522                my $pid = open my $msg_fh, '-|';
1523                defined $pid or croak $!;
1524
1525                if ($pid == 0) {
1526                        exec('git-cat-file', $type, $commit) or croak $!;
1527                }
1528                my $in_msg = 0;
1529                while (<$msg_fh>) {
1530                        if (!$in_msg) {
1531                                $in_msg = 1 if (/^\s*$/);
1532                        } elsif (/^git-svn-id: /) {
1533                                # skip this, we regenerate the correct one
1534                                # on re-fetch anyways
1535                        } else {
1536                                print $msg $_ or croak $!;
1537                        }
1538                }
1539                close $msg_fh or croak $?;
1540        }
1541        close $msg or croak $!;
1542
1543        if ($_edit || ($type eq 'tree')) {
1544                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1545                system($editor, $commit_msg);
1546        }
1547
1548        # file_to_s removes all trailing newlines, so just use chomp() here:
1549        open $msg, '<', $commit_msg or croak $!;
1550        { local $/; chomp($log_msg{msg} = <$msg>); }
1551        close $msg or croak $!;
1552
1553        return \%log_msg;
1554}
1555
1556sub set_svn_commit_env {
1557        if (defined $LC_ALL) {
1558                $ENV{LC_ALL} = $LC_ALL;
1559        } else {
1560                delete $ENV{LC_ALL};
1561        }
1562}
1563
1564sub svn_commit_tree {
1565        my ($last, $commit) = @_;
1566        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
1567        my $log_msg = get_commit_message($commit, $commit_msg);
1568        my ($oneline) = ($log_msg->{msg} =~ /([^\n\r]+)/);
1569        print "Committing $commit: $oneline\n";
1570
1571        set_svn_commit_env();
1572        my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
1573        $ENV{LC_ALL} = 'C';
1574        unlink $commit_msg;
1575        my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/);
1576        if (!defined $committed) {
1577                my $out = join("\n",@ci_output);
1578                print STDERR "W: Trouble parsing \`svn commit' output:\n\n",
1579                                $out, "\n\nAssuming English locale...";
1580                ($committed) = ($out =~ /^Committed revision \d+\./sm);
1581                defined $committed or die " FAILED!\n",
1582                        "Commit output failed to parse committed revision!\n",
1583                print STDERR " OK\n";
1584        }
1585
1586        my @svn_up = qw(svn up);
1587        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
1588        if ($_optimize_commits && ($committed == ($last->{revision} + 1))) {
1589                push @svn_up, "-r$committed";
1590                sys(@svn_up);
1591                my $info = svn_info('.');
1592                my $date = $info->{'Last Changed Date'} or die "Missing date\n";
1593                if ($info->{'Last Changed Rev'} != $committed) {
1594                        croak "$info->{'Last Changed Rev'} != $committed\n"
1595                }
1596                my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1597                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1598                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1599                                         or croak "Failed to parse date: $date\n";
1600                $log_msg->{date} = "$tz $Y-$m-$d $H:$M:$S";
1601                $log_msg->{author} = $info->{'Last Changed Author'};
1602                $log_msg->{revision} = $committed;
1603                $log_msg->{msg} .= "\n";
1604                $log_msg->{parents} = [ $last->{commit} ];
1605                $log_msg->{commit} = git_commit($log_msg, $commit);
1606                return $log_msg;
1607        }
1608        # resync immediately
1609        push @svn_up, "-r$last->{revision}";
1610        sys(@svn_up);
1611        return fetch("$committed=$commit");
1612}
1613
1614sub rev_list_raw {
1615        my (@args) = @_;
1616        my $pid = open my $fh, '-|';
1617        defined $pid or croak $!;
1618        if (!$pid) {
1619                exec(qw/git-rev-list --pretty=raw/, @args) or croak $!;
1620        }
1621        return { fh => $fh, t => { } };
1622}
1623
1624sub next_rev_list_entry {
1625        my $rl = shift;
1626        my $fh = $rl->{fh};
1627        my $x = $rl->{t};
1628        while (<$fh>) {
1629                if (/^commit ($sha1)$/o) {
1630                        if ($x->{c}) {
1631                                $rl->{t} = { c => $1 };
1632                                return $x;
1633                        } else {
1634                                $x->{c} = $1;
1635                        }
1636                } elsif (/^parent ($sha1)$/o) {
1637                        $x->{p}->{$1} = 1;
1638                } elsif (s/^    //) {
1639                        $x->{m} ||= '';
1640                        $x->{m} .= $_;
1641                }
1642        }
1643        return ($x != $rl->{t}) ? $x : undef;
1644}
1645
1646# read the entire log into a temporary file (which is removed ASAP)
1647# and store the file handle + parser state
1648sub svn_log_raw {
1649        my (@log_args) = @_;
1650        my $log_fh = IO::File->new_tmpfile or croak $!;
1651        my $pid = fork;
1652        defined $pid or croak $!;
1653        if (!$pid) {
1654                open STDOUT, '>&', $log_fh or croak $!;
1655                exec (qw(svn log), @log_args) or croak $!
1656        }
1657        waitpid $pid, 0;
1658        croak $? if $?;
1659        seek $log_fh, 0, 0 or croak $!;
1660        return { state => 'sep', fh => $log_fh };
1661}
1662
1663sub next_log_entry {
1664        my $log = shift; # retval of svn_log_raw()
1665        my $ret = undef;
1666        my $fh = $log->{fh};
1667
1668        while (<$fh>) {
1669                chomp;
1670                if (/^\-{72}$/) {
1671                        if ($log->{state} eq 'msg') {
1672                                if ($ret->{lines}) {
1673                                        $ret->{msg} .= $_."\n";
1674                                        unless(--$ret->{lines}) {
1675                                                $log->{state} = 'sep';
1676                                        }
1677                                } else {
1678                                        croak "Log parse error at: $_\n",
1679                                                $ret->{revision},
1680                                                "\n";
1681                                }
1682                                next;
1683                        }
1684                        if ($log->{state} ne 'sep') {
1685                                croak "Log parse error at: $_\n",
1686                                        "state: $log->{state}\n",
1687                                        $ret->{revision},
1688                                        "\n";
1689                        }
1690                        $log->{state} = 'rev';
1691
1692                        # if we have an empty log message, put something there:
1693                        if ($ret) {
1694                                $ret->{msg} ||= "\n";
1695                                delete $ret->{lines};
1696                                return $ret;
1697                        }
1698                        next;
1699                }
1700                if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
1701                        my $rev = $1;
1702                        my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
1703                        ($lines) = ($lines =~ /(\d+)/);
1704                        my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1705                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1706                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1707                                         or croak "Failed to parse date: $date\n";
1708                        $ret = {        revision => $rev,
1709                                        date => "$tz $Y-$m-$d $H:$M:$S",
1710                                        author => $author,
1711                                        lines => $lines,
1712                                        msg => '' };
1713                        if (defined $_authors && ! defined $users{$author}) {
1714                                die "Author: $author not defined in ",
1715                                                "$_authors file\n";
1716                        }
1717                        $log->{state} = 'msg_start';
1718                        next;
1719                }
1720                # skip the first blank line of the message:
1721                if ($log->{state} eq 'msg_start' && /^$/) {
1722                        $log->{state} = 'msg';
1723                } elsif ($log->{state} eq 'msg') {
1724                        if ($ret->{lines}) {
1725                                $ret->{msg} .= $_."\n";
1726                                unless (--$ret->{lines}) {
1727                                        $log->{state} = 'sep';
1728                                }
1729                        } else {
1730                                croak "Log parse error at: $_\n",
1731                                        $ret->{revision},"\n";
1732                        }
1733                }
1734        }
1735        return $ret;
1736}
1737
1738sub svn_info {
1739        my $url = shift || $SVN_URL;
1740
1741        my $pid = open my $info_fh, '-|';
1742        defined $pid or croak $!;
1743
1744        if ($pid == 0) {
1745                exec(qw(svn info),$url) or croak $!;
1746        }
1747
1748        my $ret = {};
1749        # only single-lines seem to exist in svn info output
1750        while (<$info_fh>) {
1751                chomp $_;
1752                if (m#^([^:]+)\s*:\s*(\S.*)$#) {
1753                        $ret->{$1} = $2;
1754                        push @{$ret->{-order}}, $1;
1755                }
1756        }
1757        close $info_fh or croak $?;
1758        return $ret;
1759}
1760
1761sub sys { system(@_) == 0 or croak $? }
1762
1763sub eol_cp {
1764        my ($from, $to) = @_;
1765        my $es = svn_propget_base('svn:eol-style', $to);
1766        open my $rfd, '<', $from or croak $!;
1767        binmode $rfd or croak $!;
1768        open my $wfd, '>', $to or croak $!;
1769        binmode $wfd or croak $!;
1770        eol_cp_fd($rfd, $wfd, $es);
1771        close $rfd or croak $!;
1772        close $wfd or croak $!;
1773}
1774
1775sub eol_cp_fd {
1776        my ($rfd, $wfd, $es) = @_;
1777        my $eol = defined $es ? $EOL{$es} : undef;
1778        my $buf;
1779        use bytes;
1780        while (1) {
1781                my ($r, $w, $t);
1782                defined($r = sysread($rfd, $buf, 4096)) or croak $!;
1783                return unless $r;
1784                if ($eol) {
1785                        if ($buf =~ /\015$/) {
1786                                my $c;
1787                                defined($r = sysread($rfd,$c,1)) or croak $!;
1788                                $buf .= $c if $r > 0;
1789                        }
1790                        $buf =~ s/(?:\015\012|\015|\012)/$eol/gs;
1791                        $r = length($buf);
1792                }
1793                for ($w = 0; $w < $r; $w += $t) {
1794                        $t = syswrite($wfd, $buf, $r - $w, $w) or croak $!;
1795                }
1796        }
1797        no bytes;
1798}
1799
1800sub do_update_index {
1801        my ($z_cmd, $cmd, $no_text_base) = @_;
1802
1803        my $z = open my $p, '-|';
1804        defined $z or croak $!;
1805        unless ($z) { exec @$z_cmd or croak $! }
1806
1807        my $pid = open my $ui, '|-';
1808        defined $pid or croak $!;
1809        unless ($pid) {
1810                exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!;
1811        }
1812        local $/ = "\0";
1813        while (my $x = <$p>) {
1814                chomp $x;
1815                if (!$no_text_base && lstat $x && ! -l _ &&
1816                                svn_propget_base('svn:keywords', $x)) {
1817                        my $mode = -x _ ? 0755 : 0644;
1818                        my ($v,$d,$f) = File::Spec->splitpath($x);
1819                        my $tb = File::Spec->catfile($d, '.svn', 'tmp',
1820                                                'text-base',"$f.svn-base");
1821                        $tb =~ s#^/##;
1822                        unless (-f $tb) {
1823                                $tb = File::Spec->catfile($d, '.svn',
1824                                                'text-base',"$f.svn-base");
1825                                $tb =~ s#^/##;
1826                        }
1827                        unlink $x or croak $!;
1828                        eol_cp($tb, $x);
1829                        chmod(($mode &~ umask), $x) or croak $!;
1830                }
1831                print $ui $x,"\0";
1832        }
1833        close $ui or croak $?;
1834}
1835
1836sub index_changes {
1837        return if $_use_lib;
1838
1839        if (!-f "$GIT_SVN_DIR/info/exclude") {
1840                open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!;
1841                print $fd '.svn',"\n";
1842                close $fd or croak $!;
1843        }
1844        my $no_text_base = shift;
1845        do_update_index([qw/git-diff-files --name-only -z/],
1846                        'remove',
1847                        $no_text_base);
1848        do_update_index([qw/git-ls-files -z --others/,
1849                                "--exclude-from=$GIT_SVN_DIR/info/exclude"],
1850                        'add',
1851                        $no_text_base);
1852}
1853
1854sub s_to_file {
1855        my ($str, $file, $mode) = @_;
1856        open my $fd,'>',$file or croak $!;
1857        print $fd $str,"\n" or croak $!;
1858        close $fd or croak $!;
1859        chmod ($mode &~ umask, $file) if (defined $mode);
1860}
1861
1862sub file_to_s {
1863        my $file = shift;
1864        open my $fd,'<',$file or croak "$!: file: $file\n";
1865        local $/;
1866        my $ret = <$fd>;
1867        close $fd or croak $!;
1868        $ret =~ s/\s*$//s;
1869        return $ret;
1870}
1871
1872sub assert_revision_unknown {
1873        my $r = shift;
1874        if (my $c = revdb_get($REVDB, $r)) {
1875                croak "$r = $c already exists! Why are we refetching it?";
1876        }
1877}
1878
1879sub trees_eq {
1880        my ($x, $y) = @_;
1881        my @x = safe_qx('git-cat-file','commit',$x);
1882        my @y = safe_qx('git-cat-file','commit',$y);
1883        if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
1884                                || $y[0] !~ /^tree $sha1\n$/) {
1885                print STDERR "Trees not equal: $y[0] != $x[0]\n";
1886                return 0
1887        }
1888        return 1;
1889}
1890
1891sub git_commit {
1892        my ($log_msg, @parents) = @_;
1893        assert_revision_unknown($log_msg->{revision});
1894        map_tree_joins() if (@_branch_from && !%tree_map);
1895
1896        my (@tmp_parents, @exec_parents, %seen_parent);
1897        if (my $lparents = $log_msg->{parents}) {
1898                @tmp_parents = @$lparents
1899        }
1900        # commit parents can be conditionally bound to a particular
1901        # svn revision via: "svn_revno=commit_sha1", filter them out here:
1902        foreach my $p (@parents) {
1903                next unless defined $p;
1904                if ($p =~ /^(\d+)=($sha1_short)$/o) {
1905                        if ($1 == $log_msg->{revision}) {
1906                                push @tmp_parents, $2;
1907                        }
1908                } else {
1909                        push @tmp_parents, $p if $p =~ /$sha1_short/o;
1910                }
1911        }
1912        my $tree = $log_msg->{tree};
1913        if (!defined $tree) {
1914                my $index = set_index($GIT_SVN_INDEX);
1915                index_changes();
1916                chomp($tree = `git-write-tree`);
1917                croak $? if $?;
1918                restore_index($index);
1919        }
1920
1921        # just in case we clobber the existing ref, we still want that ref
1922        # as our parent:
1923        if (my $cur = eval { file_to_s("$GIT_DIR/refs/remotes/$GIT_SVN") }) {
1924                push @tmp_parents, $cur;
1925        }
1926
1927        if (exists $tree_map{$tree}) {
1928                foreach my $p (@{$tree_map{$tree}}) {
1929                        my $skip;
1930                        foreach (@tmp_parents) {
1931                                # see if a common parent is found
1932                                my $mb = eval {
1933                                        safe_qx('git-merge-base', $_, $p)
1934                                };
1935                                next if ($@ || $?);
1936                                $skip = 1;
1937                                last;
1938                        }
1939                        next if $skip;
1940                        my ($url_p, $r_p, $uuid_p) = cmt_metadata($p);
1941                        next if (($SVN_UUID eq $uuid_p) &&
1942                                                ($log_msg->{revision} > $r_p));
1943                        next if (defined $url_p && defined $SVN_URL &&
1944                                                ($SVN_UUID eq $uuid_p) &&
1945                                                ($url_p eq $SVN_URL));
1946                        push @tmp_parents, $p;
1947                }
1948        }
1949        foreach (@tmp_parents) {
1950                next if $seen_parent{$_};
1951                $seen_parent{$_} = 1;
1952                push @exec_parents, $_;
1953                # MAXPARENT is defined to 16 in commit-tree.c:
1954                last if @exec_parents > 16;
1955        }
1956
1957        set_commit_env($log_msg);
1958        my @exec = ('git-commit-tree', $tree);
1959        push @exec, '-p', $_  foreach @exec_parents;
1960        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1961                                                                or croak $!;
1962        print $msg_fh $log_msg->{msg} or croak $!;
1963        unless ($_no_metadata) {
1964                print $msg_fh "\ngit-svn-id: $SVN_URL\@$log_msg->{revision}",
1965                                        " $SVN_UUID\n" or croak $!;
1966        }
1967        $msg_fh->flush == 0 or croak $!;
1968        close $msg_fh or croak $!;
1969        chomp(my $commit = do { local $/; <$out_fh> });
1970        close $out_fh or croak $!;
1971        waitpid $pid, 0;
1972        croak $? if $?;
1973        if ($commit !~ /^$sha1$/o) {
1974                die "Failed to commit, invalid sha1: $commit\n";
1975        }
1976        sys('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
1977        revdb_set($REVDB, $log_msg->{revision}, $commit);
1978
1979        # this output is read via pipe, do not change:
1980        print "r$log_msg->{revision} = $commit\n";
1981        check_repack();
1982        return $commit;
1983}
1984
1985sub check_repack {
1986        if ($_repack && (--$_repack_nr == 0)) {
1987                $_repack_nr = $_repack;
1988                sys("git repack $_repack_flags");
1989        }
1990}
1991
1992sub set_commit_env {
1993        my ($log_msg) = @_;
1994        my $author = $log_msg->{author};
1995        if (!defined $author || length $author == 0) {
1996                $author = '(no author)';
1997        }
1998        my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
1999                                : ($author,"$author\@$SVN_UUID");
2000        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
2001        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
2002        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
2003}
2004
2005sub apply_mod_line_blob {
2006        my $m = shift;
2007        if ($m->{mode_b} =~ /^120/) {
2008                blob_to_symlink($m->{sha1_b}, $m->{file_b});
2009        } else {
2010                blob_to_file($m->{sha1_b}, $m->{file_b});
2011        }
2012}
2013
2014sub blob_to_symlink {
2015        my ($blob, $link) = @_;
2016        defined $link or croak "\$link not defined!\n";
2017        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2018        if (-l $link || -f _) {
2019                unlink $link or croak $!;
2020        }
2021
2022        my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
2023        symlink $dest, $link or croak $!;
2024}
2025
2026sub blob_to_file {
2027        my ($blob, $file) = @_;
2028        defined $file or croak "\$file not defined!\n";
2029        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2030        if (-l $file || -f _) {
2031                unlink $file or croak $!;
2032        }
2033
2034        open my $blob_fh, '>', $file or croak "$!: $file\n";
2035        my $pid = fork;
2036        defined $pid or croak $!;
2037
2038        if ($pid == 0) {
2039                open STDOUT, '>&', $blob_fh or croak $!;
2040                exec('git-cat-file','blob',$blob) or croak $!;
2041        }
2042        waitpid $pid, 0;
2043        croak $? if $?;
2044
2045        close $blob_fh or croak $!;
2046}
2047
2048sub safe_qx {
2049        my $pid = open my $child, '-|';
2050        defined $pid or croak $!;
2051        if ($pid == 0) {
2052                exec(@_) or croak $!;
2053        }
2054        my @ret = (<$child>);
2055        close $child or croak $?;
2056        die $? if $?; # just in case close didn't error out
2057        return wantarray ? @ret : join('',@ret);
2058}
2059
2060sub svn_compat_check {
2061        if ($_follow_parent) {
2062                print STDERR 'E: --follow-parent functionality is only ',
2063                                "available when SVN libraries are used\n";
2064                exit 1;
2065        }
2066        my @co_help = safe_qx(qw(svn co -h));
2067        unless (grep /ignore-externals/,@co_help) {
2068                print STDERR "W: Installed svn version does not support ",
2069                                "--ignore-externals\n";
2070                $_no_ignore_ext = 1;
2071        }
2072        if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
2073                $_svn_co_url_revs = 1;
2074        }
2075        if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) {
2076                $_svn_pg_peg_revs = 1;
2077        }
2078
2079        # I really, really hope nobody hits this...
2080        unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
2081                print STDERR <<'';
2082W: The installed svn version does not support the --stop-on-copy flag in
2083   the log command.
2084   Lets hope the directory you're tracking is not a branch or tag
2085   and was never moved within the repository...
2086
2087                $_no_stop_copy = 1;
2088        }
2089}
2090
2091# *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
2092# (and they won't honor URL@<rev> without -r<rev>, too!)
2093sub svn_cmd_checkout {
2094        my ($url, $rev, $dir) = @_;
2095        my @cmd = ('svn','co', "-r$rev");
2096        push @cmd, '--ignore-externals' unless $_no_ignore_ext;
2097        $url .= "\@$rev" if $_svn_co_url_revs;
2098        sys(@cmd, $url, $dir);
2099}
2100
2101sub check_upgrade_needed {
2102        if (!-r $REVDB) {
2103                -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2104                open my $fh, '>>',$REVDB or croak $!;
2105                close $fh;
2106        }
2107        my $old = eval {
2108                my $pid = open my $child, '-|';
2109                defined $pid or croak $!;
2110                if ($pid == 0) {
2111                        close STDERR;
2112                        exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $!;
2113                }
2114                my @ret = (<$child>);
2115                close $child or croak $?;
2116                die $? if $?; # just in case close didn't error out
2117                return wantarray ? @ret : join('',@ret);
2118        };
2119        return unless $old;
2120        my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
2121        if ($@ || !$head) {
2122                print STDERR "Please run: $0 rebuild --upgrade\n";
2123                exit 1;
2124        }
2125}
2126
2127# fills %tree_map with a reverse mapping of trees to commits.  Useful
2128# for finding parents to commit on.
2129sub map_tree_joins {
2130        my %seen;
2131        foreach my $br (@_branch_from) {
2132                my $pid = open my $pipe, '-|';
2133                defined $pid or croak $!;
2134                if ($pid == 0) {
2135                        exec(qw(git-rev-list --topo-order --pretty=raw), $br)
2136                                                                or croak $!;
2137                }
2138                while (<$pipe>) {
2139                        if (/^commit ($sha1)$/o) {
2140                                my $commit = $1;
2141
2142                                # if we've seen a commit,
2143                                # we've seen its parents
2144                                last if $seen{$commit};
2145                                my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
2146                                unless (defined $tree) {
2147                                        die "Failed to parse commit $commit\n";
2148                                }
2149                                push @{$tree_map{$tree}}, $commit;
2150                                $seen{$commit} = 1;
2151                        }
2152                }
2153                close $pipe; # we could be breaking the pipe early
2154        }
2155}
2156
2157sub load_all_refs {
2158        if (@_branch_from) {
2159                print STDERR '--branch|-b parameters are ignored when ',
2160                        "--branch-all-refs|-B is passed\n";
2161        }
2162
2163        # don't worry about rev-list on non-commit objects/tags,
2164        # it shouldn't blow up if a ref is a blob or tree...
2165        chomp(@_branch_from = `git-rev-parse --symbolic --all`);
2166}
2167
2168# '<svn username> = real-name <email address>' mapping based on git-svnimport:
2169sub load_authors {
2170        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2171        while (<$authors>) {
2172                chomp;
2173                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2174                my ($user, $name, $email) = ($1, $2, $3);
2175                $users{$user} = [$name, $email];
2176        }
2177        close $authors or croak $!;
2178}
2179
2180sub rload_authors {
2181        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2182        while (<$authors>) {
2183                chomp;
2184                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2185                my ($user, $name, $email) = ($1, $2, $3);
2186                $rusers{"$name <$email>"} = $user;
2187        }
2188        close $authors or croak $!;
2189}
2190
2191sub svn_propget_base {
2192        my ($p, $f) = @_;
2193        $f .= '@BASE' if $_svn_pg_peg_revs;
2194        return safe_qx(qw/svn propget/, $p, $f);
2195}
2196
2197sub git_svn_each {
2198        my $sub = shift;
2199        foreach (`git-rev-parse --symbolic --all`) {
2200                next unless s#^refs/remotes/##;
2201                chomp $_;
2202                next unless -f "$GIT_DIR/svn/$_/info/url";
2203                &$sub($_);
2204        }
2205}
2206
2207sub migrate_revdb {
2208        git_svn_each(sub {
2209                my $id = shift;
2210                defined(my $pid = fork) or croak $!;
2211                if (!$pid) {
2212                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2213                        init_vars();
2214                        exit 0 if -r $REVDB;
2215                        print "Upgrading svn => git mapping...\n";
2216                        -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2217                        open my $fh, '>>',$REVDB or croak $!;
2218                        close $fh;
2219                        rebuild();
2220                        print "Done upgrading. You may now delete the ",
2221                                "deprecated $GIT_SVN_DIR/revs directory\n";
2222                        exit 0;
2223                }
2224                waitpid $pid, 0;
2225                croak $? if $?;
2226        });
2227}
2228
2229sub migration_check {
2230        migrate_revdb() unless (-e $REVDB);
2231        return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
2232        print "Upgrading repository...\n";
2233        unless (-d "$GIT_DIR/svn") {
2234                mkdir "$GIT_DIR/svn" or croak $!;
2235        }
2236        print "Data from a previous version of git-svn exists, but\n\t",
2237                                "$GIT_SVN_DIR\n\t(required for this version ",
2238                                "($VERSION) of git-svn) does not.\n";
2239
2240        foreach my $x (`git-rev-parse --symbolic --all`) {
2241                next unless $x =~ s#^refs/remotes/##;
2242                chomp $x;
2243                next unless -f "$GIT_DIR/$x/info/url";
2244                my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
2245                next unless $u;
2246                my $dn = dirname("$GIT_DIR/svn/$x");
2247                mkpath([$dn]) unless -d $dn;
2248                rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
2249        }
2250        migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
2251        print "Done upgrading.\n";
2252}
2253
2254sub find_rev_before {
2255        my ($r, $id, $eq_ok) = @_;
2256        my $f = "$GIT_DIR/svn/$id/.rev_db";
2257        return (undef,undef) unless -r $f;
2258        --$r unless $eq_ok;
2259        while ($r > 0) {
2260                if (my $c = revdb_get($f, $r)) {
2261                        return ($r, $c);
2262                }
2263                --$r;
2264        }
2265        return (undef, undef);
2266}
2267
2268sub init_vars {
2269        $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
2270        $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
2271        $REVDB = "$GIT_SVN_DIR/.rev_db";
2272        $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
2273        $SVN_URL = undef;
2274        $SVN_WC = "$GIT_SVN_DIR/tree";
2275        %tree_map = ();
2276}
2277
2278# convert GetOpt::Long specs for use by git-repo-config
2279sub read_repo_config {
2280        return unless -d $GIT_DIR;
2281        my $opts = shift;
2282        foreach my $o (keys %$opts) {
2283                my $v = $opts->{$o};
2284                my ($key) = ($o =~ /^([a-z\-]+)/);
2285                $key =~ s/-//g;
2286                my $arg = 'git-repo-config';
2287                $arg .= ' --int' if ($o =~ /[:=]i$/);
2288                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
2289                if (ref $v eq 'ARRAY') {
2290                        chomp(my @tmp = `$arg --get-all svn.$key`);
2291                        @$v = @tmp if @tmp;
2292                } else {
2293                        chomp(my $tmp = `$arg --get svn.$key`);
2294                        if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
2295                                $$v = $tmp;
2296                        }
2297                }
2298        }
2299}
2300
2301sub set_default_vals {
2302        if (defined $_repack) {
2303                $_repack = 1000 if ($_repack <= 0);
2304                $_repack_nr = $_repack;
2305                $_repack_flags ||= '-d';
2306        }
2307}
2308
2309sub read_grafts {
2310        my $gr_file = shift;
2311        my ($grafts, $comments) = ({}, {});
2312        if (open my $fh, '<', $gr_file) {
2313                my @tmp;
2314                while (<$fh>) {
2315                        if (/^($sha1)\s+/) {
2316                                my $c = $1;
2317                                if (@tmp) {
2318                                        @{$comments->{$c}} = @tmp;
2319                                        @tmp = ();
2320                                }
2321                                foreach my $p (split /\s+/, $_) {
2322                                        $grafts->{$c}->{$p} = 1;
2323                                }
2324                        } else {
2325                                push @tmp, $_;
2326                        }
2327                }
2328                close $fh or croak $!;
2329                @{$comments->{'END'}} = @tmp if @tmp;
2330        }
2331        return ($grafts, $comments);
2332}
2333
2334sub write_grafts {
2335        my ($grafts, $comments, $gr_file) = @_;
2336
2337        open my $fh, '>', $gr_file or croak $!;
2338        foreach my $c (sort keys %$grafts) {
2339                if ($comments->{$c}) {
2340                        print $fh $_ foreach @{$comments->{$c}};
2341                }
2342                my $p = $grafts->{$c};
2343                my %x; # real parents
2344                delete $p->{$c}; # commits are not self-reproducing...
2345                my $pid = open my $ch, '-|';
2346                defined $pid or croak $!;
2347                if (!$pid) {
2348                        exec(qw/git-cat-file commit/, $c) or croak $!;
2349                }
2350                while (<$ch>) {
2351                        if (/^parent ($sha1)/) {
2352                                $x{$1} = $p->{$1} = 1;
2353                        } else {
2354                                last unless /^\S/;
2355                        }
2356                }
2357                close $ch; # breaking the pipe
2358
2359                # if real parents are the only ones in the grafts, drop it
2360                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2361
2362                my (@ip, @jp, $mb);
2363                my %del = %x;
2364                @ip = @jp = keys %$p;
2365                foreach my $i (@ip) {
2366                        next if $del{$i} || $p->{$i} == 2;
2367                        foreach my $j (@jp) {
2368                                next if $i eq $j || $del{$j} || $p->{$j} == 2;
2369                                $mb = eval { safe_qx('git-merge-base',$i,$j) };
2370                                next unless $mb;
2371                                chomp $mb;
2372                                next if $x{$mb};
2373                                if ($mb eq $j) {
2374                                        delete $p->{$i};
2375                                        $del{$i} = 1;
2376                                } elsif ($mb eq $i) {
2377                                        delete $p->{$j};
2378                                        $del{$j} = 1;
2379                                }
2380                        }
2381                }
2382
2383                # if real parents are the only ones in the grafts, drop it
2384                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2385
2386                print $fh $c, ' ', join(' ', sort keys %$p),"\n";
2387        }
2388        if ($comments->{'END'}) {
2389                print $fh $_ foreach @{$comments->{'END'}};
2390        }
2391        close $fh or croak $!;
2392}
2393
2394sub read_url_paths_all {
2395        my ($l_map, $pfx, $p) = @_;
2396        my @dir;
2397        foreach (<$p/*>) {
2398                if (-r "$_/info/url") {
2399                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2400                        my $id = $pfx . basename $_;
2401                        my $url = file_to_s("$_/info/url");
2402                        my ($u, $p) = repo_path_split($url);
2403                        $l_map->{$u}->{$p} = $id;
2404                } elsif (-d $_) {
2405                        push @dir, $_;
2406                }
2407        }
2408        foreach (@dir) {
2409                my $x = $_;
2410                $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
2411                read_url_paths_all($l_map, $x, $_);
2412        }
2413}
2414
2415# this one only gets ids that have been imported, not new ones
2416sub read_url_paths {
2417        my $l_map = {};
2418        git_svn_each(sub { my $x = shift;
2419                        my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
2420                        my ($u, $p) = repo_path_split($url);
2421                        $l_map->{$u}->{$p} = $x;
2422                        });
2423        return $l_map;
2424}
2425
2426sub extract_metadata {
2427        my $id = shift or return (undef, undef, undef);
2428        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
2429                                                        \s([a-f\d\-]+)$/x);
2430        if (!$rev || !$uuid || !$url) {
2431                # some of the original repositories I made had
2432                # identifiers like this:
2433                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
2434        }
2435        return ($url, $rev, $uuid);
2436}
2437
2438sub cmt_metadata {
2439        return extract_metadata((grep(/^git-svn-id: /,
2440                safe_qx(qw/git-cat-file commit/, shift)))[-1]);
2441}
2442
2443sub get_commit_time {
2444        my $cmt = shift;
2445        defined(my $pid = open my $fh, '-|') or croak $!;
2446        if (!$pid) {
2447                exec qw/git-rev-list --pretty=raw -n1/, $cmt or croak $!;
2448        }
2449        while (<$fh>) {
2450                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
2451                my ($s, $tz) = ($1, $2);
2452                if ($tz =~ s/^\+//) {
2453                        $s += tz_to_s_offset($tz);
2454                } elsif ($tz =~ s/^\-//) {
2455                        $s -= tz_to_s_offset($tz);
2456                }
2457                close $fh;
2458                return $s;
2459        }
2460        die "Can't get commit time for commit: $cmt\n";
2461}
2462
2463sub tz_to_s_offset {
2464        my ($tz) = @_;
2465        $tz =~ s/(\d\d)$//;
2466        return ($1 * 60) + ($tz * 3600);
2467}
2468
2469sub setup_pager { # translated to Perl from pager.c
2470        return unless (-t *STDOUT);
2471        my $pager = $ENV{PAGER};
2472        if (!defined $pager) {
2473                $pager = 'less';
2474        } elsif (length $pager == 0 || $pager eq 'cat') {
2475                return;
2476        }
2477        pipe my $rfd, my $wfd or return;
2478        defined(my $pid = fork) or croak $!;
2479        if (!$pid) {
2480                open STDOUT, '>&', $wfd or croak $!;
2481                return;
2482        }
2483        open STDIN, '<&', $rfd or croak $!;
2484        $ENV{LESS} ||= '-S';
2485        exec $pager or croak "Can't run pager: $!\n";;
2486}
2487
2488sub get_author_info {
2489        my ($dest, $author, $t, $tz) = @_;
2490        $author =~ s/(?:^\s*|\s*$)//g;
2491        $dest->{a_raw} = $author;
2492        my $_a;
2493        if ($_authors) {
2494                $_a = $rusers{$author} || undef;
2495        }
2496        if (!$_a) {
2497                ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
2498        }
2499        $dest->{t} = $t;
2500        $dest->{tz} = $tz;
2501        $dest->{a} = $_a;
2502        # Date::Parse isn't in the standard Perl distro :(
2503        if ($tz =~ s/^\+//) {
2504                $t += tz_to_s_offset($tz);
2505        } elsif ($tz =~ s/^\-//) {
2506                $t -= tz_to_s_offset($tz);
2507        }
2508        $dest->{t_utc} = $t;
2509}
2510
2511sub process_commit {
2512        my ($c, $r_min, $r_max, $defer) = @_;
2513        if (defined $r_min && defined $r_max) {
2514                if ($r_min == $c->{r} && $r_min == $r_max) {
2515                        show_commit($c);
2516                        return 0;
2517                }
2518                return 1 if $r_min == $r_max;
2519                if ($r_min < $r_max) {
2520                        # we need to reverse the print order
2521                        return 0 if (defined $_limit && --$_limit < 0);
2522                        push @$defer, $c;
2523                        return 1;
2524                }
2525                if ($r_min != $r_max) {
2526                        return 1 if ($r_min < $c->{r});
2527                        return 1 if ($r_max > $c->{r});
2528                }
2529        }
2530        return 0 if (defined $_limit && --$_limit < 0);
2531        show_commit($c);
2532        return 1;
2533}
2534
2535sub show_commit {
2536        my $c = shift;
2537        if ($_oneline) {
2538                my $x = "\n";
2539                if (my $l = $c->{l}) {
2540                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2541                        $x = $l->[0];
2542                }
2543                $_l_fmt ||= 'A' . length($c->{r});
2544                print 'r',pack($_l_fmt, $c->{r}),' | ';
2545                print "$c->{c} | " if $_show_commit;
2546                print $x;
2547        } else {
2548                show_commit_normal($c);
2549        }
2550}
2551
2552sub show_commit_normal {
2553        my ($c) = @_;
2554        print '-' x72, "\nr$c->{r} | ";
2555        print "$c->{c} | " if $_show_commit;
2556        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2557                                 localtime($c->{t_utc})), ' | ';
2558        my $nr_line = 0;
2559
2560        if (my $l = $c->{l}) {
2561                while ($l->[$#$l] eq "\n" && $l->[($#$l - 1)] eq "\n") {
2562                        pop @$l;
2563                }
2564                $nr_line = scalar @$l;
2565                if (!$nr_line) {
2566                        print "1 line\n\n\n";
2567                } else {
2568                        if ($nr_line == 1) {
2569                                $nr_line = '1 line';
2570                        } else {
2571                                $nr_line .= ' lines';
2572                        }
2573                        print $nr_line, "\n\n";
2574                        print $_ foreach @$l;
2575                }
2576        } else {
2577                print "1 line\n\n";
2578
2579        }
2580        foreach my $x (qw/raw diff/) {
2581                if ($c->{$x}) {
2582                        print "\n";
2583                        print $_ foreach @{$c->{$x}}
2584                }
2585        }
2586}
2587
2588sub libsvn_load {
2589        return unless $_use_lib;
2590        $_use_lib = eval {
2591                require SVN::Core;
2592                if ($SVN::Core::VERSION lt '1.1.0') {
2593                        die "Need SVN::Core 1.1.0 or better ",
2594                                        "(got $SVN::Core::VERSION) ",
2595                                        "Falling back to command-line svn\n";
2596                }
2597                require SVN::Ra;
2598                require SVN::Delta;
2599                push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
2600                my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2601                                        $SVN::Node::dir.$SVN::Node::unknown.
2602                                        $SVN::Node::none.$SVN::Node::file.
2603                                        $SVN::Node::dir.$SVN::Node::unknown;
2604                1;
2605        };
2606}
2607
2608sub libsvn_connect {
2609        my ($url) = @_;
2610        my $auth = SVN::Core::auth_open([SVN::Client::get_simple_provider(),
2611                          SVN::Client::get_ssl_server_trust_file_provider(),
2612                          SVN::Client::get_username_provider()]);
2613        my $s = eval { SVN::Ra->new(url => $url, auth => $auth) };
2614        return $s;
2615}
2616
2617sub libsvn_get_file {
2618        my ($gui, $f, $rev) = @_;
2619        my $p = $f;
2620        return unless ($p =~ s#^\Q$SVN_PATH\E/##);
2621
2622        my ($hash, $pid, $in, $out);
2623        my $pool = SVN::Pool->new;
2624        defined($pid = open3($in, $out, '>&STDERR',
2625                                qw/git-hash-object -w --stdin/)) or croak $!;
2626        # redirect STDOUT for SVN 1.1.x compatibility
2627        open my $stdout, '>&', \*STDOUT or croak $!;
2628        open STDOUT, '>&', $in or croak $!;
2629        my ($r, $props) = $SVN->get_file($f, $rev, \*STDOUT, $pool);
2630        $in->flush == 0 or croak $!;
2631        open STDOUT, '>&', $stdout or croak $!;
2632        close $in or croak $!;
2633        close $stdout or croak $!;
2634        $pool->clear;
2635        chomp($hash = do { local $/; <$out> });
2636        close $out or croak $!;
2637        waitpid $pid, 0;
2638        $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2639
2640        my $mode = exists $props->{'svn:executable'} ? '100755' : '100644';
2641        if (exists $props->{'svn:special'}) {
2642                $mode = '120000';
2643                my $link = `git-cat-file blob $hash`;
2644                $link =~ s/^link // or die "svn:special file with contents: <",
2645                                                $link, "> is not understood\n";
2646                defined($pid = open3($in, $out, '>&STDERR',
2647                                qw/git-hash-object -w --stdin/)) or croak $!;
2648                print $in $link;
2649                $in->flush == 0 or croak $!;
2650                close $in or croak $!;
2651                chomp($hash = do { local $/; <$out> });
2652                close $out or croak $!;
2653                waitpid $pid, 0;
2654                $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2655        }
2656        print $gui $mode,' ',$hash,"\t",$p,"\0" or croak $!;
2657}
2658
2659sub libsvn_log_entry {
2660        my ($rev, $author, $date, $msg, $parents) = @_;
2661        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2662                                         (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
2663                                or die "Unable to parse date: $date\n";
2664        if (defined $_authors && ! defined $users{$author}) {
2665                die "Author: $author not defined in $_authors file\n";
2666        }
2667        return { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
2668                author => $author, msg => $msg."\n", parents => $parents || [] }
2669}
2670
2671sub process_rm {
2672        my ($gui, $last_commit, $f) = @_;
2673        $f =~ s#^\Q$SVN_PATH\E/?## or return;
2674        # remove entire directories.
2675        if (safe_qx('git-ls-tree',$last_commit,'--',$f) =~ /^040000 tree/) {
2676                defined(my $pid = open my $ls, '-|') or croak $!;
2677                if (!$pid) {
2678                        exec(qw/git-ls-tree -r --name-only -z/,
2679                                $last_commit,'--',$f) or croak $!;
2680                }
2681                local $/ = "\0";
2682                while (<$ls>) {
2683                        print $gui '0 ',0 x 40,"\t",$_ or croak $!;
2684                }
2685                close $ls or croak $?;
2686        } else {
2687                print $gui '0 ',0 x 40,"\t",$f,"\0" or croak $!;
2688        }
2689}
2690
2691sub libsvn_fetch {
2692        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
2693        open my $gui, '| git-update-index -z --index-info' or croak $!;
2694        my @amr;
2695        foreach my $f (keys %$paths) {
2696                my $m = $paths->{$f}->action();
2697                $f =~ s#^/+##;
2698                if ($m =~ /^[DR]$/) {
2699                        print "\t$m\t$f\n" unless $_q;
2700                        process_rm($gui, $last_commit, $f);
2701                        next if $m eq 'D';
2702                        # 'R' can be file replacements, too, right?
2703                }
2704                my $pool = SVN::Pool->new;
2705                my $t = $SVN->check_path($f, $rev, $pool);
2706                if ($t == $SVN::Node::file) {
2707                        if ($m =~ /^[AMR]$/) {
2708                                push @amr, [ $m, $f ];
2709                        } else {
2710                                die "Unrecognized action: $m, ($f r$rev)\n";
2711                        }
2712                } elsif ($t == $SVN::Node::dir && $m =~ /^[AR]$/) {
2713                        my @traversed = ();
2714                        libsvn_traverse($gui, '', $f, $rev, \@traversed);
2715                        foreach (@traversed) {
2716                                push @amr, [ $m, $_ ]
2717                        }
2718                }
2719                $pool->clear;
2720        }
2721        foreach (@amr) {
2722                print "\t$_->[0]\t$_->[1]\n" unless $_q;
2723                libsvn_get_file($gui, $_->[1], $rev)
2724        }
2725        close $gui or croak $?;
2726        return libsvn_log_entry($rev, $author, $date, $msg, [$last_commit]);
2727}
2728
2729sub svn_grab_base_rev {
2730        defined(my $pid = open my $fh, '-|') or croak $!;
2731        if (!$pid) {
2732                open my $null, '>', '/dev/null' or croak $!;
2733                open STDERR, '>&', $null or croak $!;
2734                exec qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0"
2735                                                                or croak $!;
2736        }
2737        chomp(my $c = do { local $/; <$fh> });
2738        close $fh;
2739        if (defined $c && length $c) {
2740                my ($url, $rev, $uuid) = cmt_metadata($c);
2741                return ($rev, $c) if defined $rev;
2742        }
2743        if ($_no_metadata) {
2744                my $offset = -41; # from tail
2745                my $rl;
2746                open my $fh, '<', $REVDB or
2747                        die "--no-metadata specified and $REVDB not readable\n";
2748                seek $fh, $offset, 2;
2749                $rl = readline $fh;
2750                defined $rl or return (undef, undef);
2751                chomp $rl;
2752                while ($c ne $rl && tell $fh != 0) {
2753                        $offset -= 41;
2754                        seek $fh, $offset, 2;
2755                        $rl = readline $fh;
2756                        defined $rl or return (undef, undef);
2757                        chomp $rl;
2758                }
2759                my $rev = tell $fh;
2760                croak $! if ($rev < -1);
2761                $rev =  ($rev - 41) / 41;
2762                close $fh or croak $!;
2763                return ($rev, $c);
2764        }
2765        return (undef, undef);
2766}
2767
2768sub libsvn_parse_revision {
2769        my $base = shift;
2770        my $head = $SVN->get_latest_revnum();
2771        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
2772                return ($base + 1, $head) if (defined $base);
2773                return (0, $head);
2774        }
2775        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
2776        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
2777        if ($_revision =~ /^BASE:(\d+)$/) {
2778                return ($base + 1, $1) if (defined $base);
2779                return (0, $head);
2780        }
2781        return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
2782        die "revision argument: $_revision not understood by git-svn\n",
2783                "Try using the command-line svn client instead\n";
2784}
2785
2786sub libsvn_traverse {
2787        my ($gui, $pfx, $path, $rev, $files) = @_;
2788        my $cwd = "$pfx/$path";
2789        my $pool = SVN::Pool->new;
2790        $cwd =~ s#^/+##g;
2791        my ($dirent, $r, $props) = $SVN->get_dir($cwd, $rev, $pool);
2792        foreach my $d (keys %$dirent) {
2793                my $t = $dirent->{$d}->kind;
2794                if ($t == $SVN::Node::dir) {
2795                        libsvn_traverse($gui, $cwd, $d, $rev, $files);
2796                } elsif ($t == $SVN::Node::file) {
2797                        my $file = "$cwd/$d";
2798                        if (defined $files) {
2799                                push @$files, $file;
2800                        } else {
2801                                print "\tA\t$file\n" unless $_q;
2802                                libsvn_get_file($gui, $file, $rev);
2803                        }
2804                }
2805        }
2806        $pool->clear;
2807}
2808
2809sub libsvn_traverse_ignore {
2810        my ($fh, $path, $r) = @_;
2811        $path =~ s#^/+##g;
2812        my $pool = SVN::Pool->new;
2813        my ($dirent, undef, $props) = $SVN->get_dir($path, $r, $pool);
2814        my $p = $path;
2815        $p =~ s#^\Q$SVN_PATH\E/?##;
2816        print $fh length $p ? "\n# $p\n" : "\n# /\n";
2817        if (my $s = $props->{'svn:ignore'}) {
2818                $s =~ s/[\r\n]+/\n/g;
2819                chomp $s;
2820                if (length $p == 0) {
2821                        $s =~ s#\n#\n/$p#g;
2822                        print $fh "/$s\n";
2823                } else {
2824                        $s =~ s#\n#\n/$p/#g;
2825                        print $fh "/$p/$s\n";
2826                }
2827        }
2828        foreach (sort keys %$dirent) {
2829                next if $dirent->{$_}->kind != $SVN::Node::dir;
2830                libsvn_traverse_ignore($fh, "$path/$_", $r);
2831        }
2832        $pool->clear;
2833}
2834
2835sub revisions_eq {
2836        my ($path, $r0, $r1) = @_;
2837        return 1 if $r0 == $r1;
2838        my $nr = 0;
2839        if ($_use_lib) {
2840                # should be OK to use Pool here (r1 - r0) should be small
2841                my $pool = SVN::Pool->new;
2842                libsvn_get_log($SVN, "/$path", $r0, $r1,
2843                                0, 1, 1, sub {$nr++}, $pool);
2844                $pool->clear;
2845        } else {
2846                my ($url, undef) = repo_path_split($SVN_URL);
2847                my $svn_log = svn_log_raw("$url/$path","-r$r0:$r1");
2848                while (next_log_entry($svn_log)) { $nr++ }
2849                close $svn_log->{fh};
2850        }
2851        return 0 if ($nr > 1);
2852        return 1;
2853}
2854
2855sub libsvn_find_parent_branch {
2856        my ($paths, $rev, $author, $date, $msg) = @_;
2857        my $svn_path = '/'.$SVN_PATH;
2858
2859        # look for a parent from another branch:
2860        my $i = $paths->{$svn_path} or return;
2861        my $branch_from = $i->copyfrom_path or return;
2862        my $r = $i->copyfrom_rev;
2863        print STDERR  "Found possible branch point: ",
2864                                "$branch_from => $svn_path, $r\n";
2865        $branch_from =~ s#^/##;
2866        my $l_map = {};
2867        read_url_paths_all($l_map, '', "$GIT_DIR/svn");
2868        my $url = $SVN->{url};
2869        defined $l_map->{$url} or return;
2870        my $id = $l_map->{$url}->{$branch_from};
2871        if (!defined $id && $_follow_parent) {
2872                print STDERR "Following parent: $branch_from\@$r\n";
2873                # auto create a new branch and follow it
2874                $id = basename($branch_from);
2875                $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
2876                while (-r "$GIT_DIR/svn/$id") {
2877                        # just grow a tail if we're not unique enough :x
2878                        $id .= '-';
2879                }
2880        }
2881        return unless defined $id;
2882
2883        my ($r0, $parent) = find_rev_before($r,$id,1);
2884        if ($_follow_parent && (!defined $r0 || !defined $parent)) {
2885                defined(my $pid = fork) or croak $!;
2886                if (!$pid) {
2887                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2888                        init_vars();
2889                        $SVN_URL = "$url/$branch_from";
2890                        $SVN_LOG = $SVN = undef;
2891                        setup_git_svn();
2892                        # we can't assume SVN_URL exists at r+1:
2893                        $_revision = "0:$r";
2894                        fetch_lib();
2895                        exit 0;
2896                }
2897                waitpid $pid, 0;
2898                croak $? if $?;
2899                ($r0, $parent) = find_rev_before($r,$id,1);
2900        }
2901        return unless (defined $r0 && defined $parent);
2902        if (revisions_eq($branch_from, $r0, $r)) {
2903                unlink $GIT_SVN_INDEX;
2904                print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
2905                sys(qw/git-read-tree/, $parent);
2906                return libsvn_fetch($parent, $paths, $rev,
2907                                        $author, $date, $msg);
2908        }
2909        print STDERR "Nope, branch point not imported or unknown\n";
2910        return undef;
2911}
2912
2913sub libsvn_get_log {
2914        my ($ra, @args) = @_;
2915        if ($SVN::Core::VERSION le '1.2.0') {
2916                splice(@args, 3, 1);
2917        }
2918        $ra->get_log(@args);
2919}
2920
2921sub libsvn_new_tree {
2922        if (my $log_entry = libsvn_find_parent_branch(@_)) {
2923                return $log_entry;
2924        }
2925        my ($paths, $rev, $author, $date, $msg) = @_;
2926        open my $gui, '| git-update-index -z --index-info' or croak $!;
2927        libsvn_traverse($gui, '', $SVN_PATH, $rev);
2928        close $gui or croak $?;
2929        return libsvn_log_entry($rev, $author, $date, $msg);
2930}
2931
2932sub find_graft_path_commit {
2933        my ($tree_paths, $p1, $r1) = @_;
2934        foreach my $x (keys %$tree_paths) {
2935                next unless ($p1 =~ /^\Q$x\E/);
2936                my $i = $tree_paths->{$x};
2937                my ($r0, $parent) = find_rev_before($r1,$i,1);
2938                return $parent if (defined $r0 && $r0 == $r1);
2939                print STDERR "r$r1 of $i not imported\n";
2940                next;
2941        }
2942        return undef;
2943}
2944
2945sub find_graft_path_parents {
2946        my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
2947        foreach my $x (keys %$tree_paths) {
2948                next unless ($p0 =~ /^\Q$x\E/);
2949                my $i = $tree_paths->{$x};
2950                my ($r, $parent) = find_rev_before($r0, $i, 1);
2951                if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
2952                        my ($url_b, undef, $uuid_b) = cmt_metadata($c);
2953                        my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
2954                        next if ($url_a && $url_b && $url_a eq $url_b &&
2955                                                        $uuid_b eq $uuid_a);
2956                        $grafts->{$c}->{$parent} = 1;
2957                }
2958        }
2959}
2960
2961sub libsvn_graft_file_copies {
2962        my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
2963        foreach (keys %$paths) {
2964                my $i = $paths->{$_};
2965                my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
2966                                        $i->copyfrom_rev);
2967                next unless (defined $p0 && defined $r0);
2968
2969                my $p1 = $_;
2970                $p1 =~ s#^/##;
2971                $p0 =~ s#^/##;
2972                my $c = find_graft_path_commit($tree_paths, $p1, $rev);
2973                next unless $c;
2974                find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
2975        }
2976}
2977
2978sub set_index {
2979        my $old = $ENV{GIT_INDEX_FILE};
2980        $ENV{GIT_INDEX_FILE} = shift;
2981        return $old;
2982}
2983
2984sub restore_index {
2985        my ($old) = @_;
2986        if (defined $old) {
2987                $ENV{GIT_INDEX_FILE} = $old;
2988        } else {
2989                delete $ENV{GIT_INDEX_FILE};
2990        }
2991}
2992
2993sub libsvn_commit_cb {
2994        my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
2995        if ($_optimize_commits && $rev == ($r_last + 1)) {
2996                my $log = libsvn_log_entry($rev,$committer,$date,$msg);
2997                $log->{tree} = get_tree_from_treeish($c);
2998                my $cmt = git_commit($log, $cmt_last, $c);
2999                my @diff = safe_qx('git-diff-tree', $cmt, $c);
3000                if (@diff) {
3001                        print STDERR "Trees differ: $cmt $c\n",
3002                                        join('',@diff),"\n";
3003                        exit 1;
3004                }
3005        } else {
3006                fetch("$rev=$c");
3007        }
3008}
3009
3010sub libsvn_ls_fullurl {
3011        my $fullurl = shift;
3012        my ($repo, $path) = repo_path_split($fullurl);
3013        $SVN ||= libsvn_connect($repo);
3014        my @ret;
3015        my $pool = SVN::Pool->new;
3016        my ($dirent, undef, undef) = $SVN->get_dir($path,
3017                                                $SVN->get_latest_revnum, $pool);
3018        foreach my $d (keys %$dirent) {
3019                if ($dirent->{$d}->kind == $SVN::Node::dir) {
3020                        push @ret, "$d/"; # add '/' for compat with cli svn
3021                }
3022        }
3023        $pool->clear;
3024        return @ret;
3025}
3026
3027
3028sub libsvn_skip_unknown_revs {
3029        my $err = shift;
3030        my $errno = $err->apr_err();
3031        # Maybe the branch we're tracking didn't
3032        # exist when the repo started, so it's
3033        # not an error if it doesn't, just continue
3034        #
3035        # Wonderfully consistent library, eh?
3036        # 160013 - svn:// and file://
3037        # 175002 - http(s)://
3038        #   More codes may be discovered later...
3039        if ($errno == 175002 || $errno == 160013) {
3040                return;
3041        }
3042        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3043};
3044
3045# Tie::File seems to be prone to offset errors if revisions get sparse,
3046# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3047# one of my favorite modules is out :<  Next up would be one of the DBM
3048# modules, but I'm not sure which is most portable...  So I'll just
3049# go with something that's plain-text, but still capable of
3050# being randomly accessed.  So here's my ultra-simple fixed-width
3051# database.  All records are 40 characters + "\n", so it's easy to seek
3052# to a revision: (41 * rev) is the byte offset.
3053# A record of 40 0s denotes an empty revision.
3054# And yes, it's still pretty fast (faster than Tie::File).
3055sub revdb_set {
3056        my ($file, $rev, $commit) = @_;
3057        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
3058        open my $fh, '+<', $file or croak $!;
3059        my $offset = $rev * 41;
3060        # assume that append is the common case:
3061        seek $fh, 0, 2 or croak $!;
3062        my $pos = tell $fh;
3063        if ($pos < $offset) {
3064                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
3065        }
3066        seek $fh, $offset, 0 or croak $!;
3067        print $fh $commit,"\n";
3068        close $fh or croak $!;
3069}
3070
3071sub revdb_get {
3072        my ($file, $rev) = @_;
3073        my $ret;
3074        my $offset = $rev * 41;
3075        open my $fh, '<', $file or croak $!;
3076        seek $fh, $offset, 0;
3077        if (tell $fh == $offset) {
3078                $ret = readline $fh;
3079                if (defined $ret) {
3080                        chomp $ret;
3081                        $ret = undef if ($ret =~ /^0{40}$/);
3082                }
3083        }
3084        close $fh or croak $!;
3085        return $ret;
3086}
3087
3088sub copy_remote_ref {
3089        my $origin = $_cp_remote ? $_cp_remote : 'origin';
3090        my $ref = "refs/remotes/$GIT_SVN";
3091        if (safe_qx('git-ls-remote', $origin, $ref)) {
3092                sys(qw/git fetch/, $origin, "$ref:$ref");
3093        } else {
3094                die "Unable to find remote reference: ",
3095                                "refs/remotes/$GIT_SVN on $origin\n";
3096        }
3097}
3098
3099package SVN::Git::Editor;
3100use vars qw/@ISA/;
3101use strict;
3102use warnings;
3103use Carp qw/croak/;
3104use IO::File;
3105
3106sub new {
3107        my $class = shift;
3108        my $git_svn = shift;
3109        my $self = SVN::Delta::Editor->new(@_);
3110        bless $self, $class;
3111        foreach (qw/svn_path c r ra /) {
3112                die "$_ required!\n" unless (defined $git_svn->{$_});
3113                $self->{$_} = $git_svn->{$_};
3114        }
3115        $self->{pool} = SVN::Pool->new;
3116        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3117        $self->{rm} = { };
3118        require Digest::MD5;
3119        return $self;
3120}
3121
3122sub split_path {
3123        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3124}
3125
3126sub repo_path {
3127        (defined $_[1] && length $_[1]) ? "$_[0]->{svn_path}/$_[1]"
3128                                        : $_[0]->{svn_path}
3129}
3130
3131sub url_path {
3132        my ($self, $path) = @_;
3133        $self->{ra}->{url} . '/' . $self->repo_path($path);
3134}
3135
3136sub rmdirs {
3137        my ($self, $q) = @_;
3138        my $rm = $self->{rm};
3139        delete $rm->{''}; # we never delete the url we're tracking
3140        return unless %$rm;
3141
3142        foreach (keys %$rm) {
3143                my @d = split m#/#, $_;
3144                my $c = shift @d;
3145                $rm->{$c} = 1;
3146                while (@d) {
3147                        $c .= '/' . shift @d;
3148                        $rm->{$c} = 1;
3149                }
3150        }
3151        delete $rm->{$self->{svn_path}};
3152        delete $rm->{''}; # we never delete the url we're tracking
3153        return unless %$rm;
3154
3155        defined(my $pid = open my $fh,'-|') or croak $!;
3156        if (!$pid) {
3157                exec qw/git-ls-tree --name-only -r -z/, $self->{c} or croak $!;
3158        }
3159        local $/ = "\0";
3160        my @svn_path = split m#/#, $self->{svn_path};
3161        while (<$fh>) {
3162                chomp;
3163                my @dn = (@svn_path, (split m#/#, $_));
3164                while (pop @dn) {
3165                        delete $rm->{join '/', @dn};
3166                }
3167                unless (%$rm) {
3168                        close $fh;
3169                        return;
3170                }
3171        }
3172        close $fh;
3173
3174        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3175        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3176                $self->close_directory($bat->{$d}, $p);
3177                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3178                print "\tD+\t/$d/\n" unless $q;
3179                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3180                delete $bat->{$d};
3181        }
3182}
3183
3184sub open_or_add_dir {
3185        my ($self, $full_path, $baton) = @_;
3186        my $p = SVN::Pool->new;
3187        my $t = $self->{ra}->check_path($full_path, $self->{r}, $p);
3188        $p->clear;
3189        if ($t == $SVN::Node::none) {
3190                return $self->add_directory($full_path, $baton,
3191                                                undef, -1, $self->{pool});
3192        } elsif ($t == $SVN::Node::dir) {
3193                return $self->open_directory($full_path, $baton,
3194                                                $self->{r}, $self->{pool});
3195        }
3196        print STDERR "$full_path already exists in repository at ",
3197                "r$self->{r} and it is not a directory (",
3198                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3199        exit 1;
3200}
3201
3202sub ensure_path {
3203        my ($self, $path) = @_;
3204        my $bat = $self->{bat};
3205        $path = $self->repo_path($path);
3206        return $bat->{''} unless (length $path);
3207        my @p = split m#/+#, $path;
3208        my $c = shift @p;
3209        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3210        while (@p) {
3211                my $c0 = $c;
3212                $c .= '/' . shift @p;
3213                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3214        }
3215        return $bat->{$c};
3216}
3217
3218sub A {
3219        my ($self, $m, $q) = @_;
3220        my ($dir, $file) = split_path($m->{file_b});
3221        my $pbat = $self->ensure_path($dir);
3222        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3223                                        undef, -1);
3224        print "\tA\t$m->{file_b}\n" unless $q;
3225        $self->chg_file($fbat, $m);
3226        $self->close_file($fbat,undef,$self->{pool});
3227}
3228
3229sub C {
3230        my ($self, $m, $q) = @_;
3231        my ($dir, $file) = split_path($m->{file_b});
3232        my $pbat = $self->ensure_path($dir);
3233        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3234                                $self->url_path($m->{file_a}), $self->{r});
3235        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $q;
3236        $self->chg_file($fbat, $m);
3237        $self->close_file($fbat,undef,$self->{pool});
3238}
3239
3240sub delete_entry {
3241        my ($self, $path, $pbat) = @_;
3242        my $rpath = $self->repo_path($path);
3243        my ($dir, $file) = split_path($rpath);
3244        $self->{rm}->{$dir} = 1;
3245        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3246}
3247
3248sub R {
3249        my ($self, $m, $q) = @_;
3250        my ($dir, $file) = split_path($m->{file_b});
3251        my $pbat = $self->ensure_path($dir);
3252        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3253                                $self->url_path($m->{file_a}), $self->{r});
3254        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $q;
3255        $self->chg_file($fbat, $m);
3256        $self->close_file($fbat,undef,$self->{pool});
3257
3258        ($dir, $file) = split_path($m->{file_a});
3259        $pbat = $self->ensure_path($dir);
3260        $self->delete_entry($m->{file_a}, $pbat);
3261}
3262
3263sub M {
3264        my ($self, $m, $q) = @_;
3265        my ($dir, $file) = split_path($m->{file_b});
3266        my $pbat = $self->ensure_path($dir);
3267        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3268                                $pbat,$self->{r},$self->{pool});
3269        print "\t$m->{chg}\t$m->{file_b}\n" unless $q;
3270        $self->chg_file($fbat, $m);
3271        $self->close_file($fbat,undef,$self->{pool});
3272}
3273
3274sub T { shift->M(@_) }
3275
3276sub change_file_prop {
3277        my ($self, $fbat, $pname, $pval) = @_;
3278        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3279}
3280
3281sub chg_file {
3282        my ($self, $fbat, $m) = @_;
3283        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3284                $self->change_file_prop($fbat,'svn:executable','*');
3285        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3286                $self->change_file_prop($fbat,'svn:executable',undef);
3287        }
3288        my $fh = IO::File->new_tmpfile or croak $!;
3289        if ($m->{mode_b} =~ /^120/) {
3290                print $fh 'link ' or croak $!;
3291                $self->change_file_prop($fbat,'svn:special','*');
3292        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3293                $self->change_file_prop($fbat,'svn:special',undef);
3294        }
3295        defined(my $pid = fork) or croak $!;
3296        if (!$pid) {
3297                open STDOUT, '>&', $fh or croak $!;
3298                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3299        }
3300        waitpid $pid, 0;
3301        croak $? if $?;
3302        $fh->flush == 0 or croak $!;
3303        seek $fh, 0, 0 or croak $!;
3304
3305        my $md5 = Digest::MD5->new;
3306        $md5->addfile($fh) or croak $!;
3307        seek $fh, 0, 0 or croak $!;
3308
3309        my $exp = $md5->hexdigest;
3310        my $atd = $self->apply_textdelta($fbat, undef, $self->{pool});
3311        my $got = SVN::TxDelta::send_stream($fh, @$atd, $self->{pool});
3312        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3313
3314        close $fh or croak $!;
3315}
3316
3317sub D {
3318        my ($self, $m, $q) = @_;
3319        my ($dir, $file) = split_path($m->{file_b});
3320        my $pbat = $self->ensure_path($dir);
3321        print "\tD\t$m->{file_b}\n" unless $q;
3322        $self->delete_entry($m->{file_b}, $pbat);
3323}
3324
3325sub close_edit {
3326        my ($self) = @_;
3327        my ($p,$bat) = ($self->{pool}, $self->{bat});
3328        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3329                $self->close_directory($bat->{$_}, $p);
3330        }
3331        $self->SUPER::close_edit($p);
3332        $p->clear;
3333}
3334
3335sub abort_edit {
3336        my ($self) = @_;
3337        $self->SUPER::abort_edit($self->{pool});
3338        $self->{pool}->clear;
3339}
3340
3341__END__
3342
3343Data structures:
3344
3345$svn_log hashref (as returned by svn_log_raw)
3346{
3347        fh => file handle of the log file,
3348        state => state of the log file parser (sep/msg/rev/msg_start...)
3349}
3350
3351$log_msg hashref as returned by next_log_entry($svn_log)
3352{
3353        msg => 'whitespace-formatted log entry
3354',                                              # trailing newline is preserved
3355        revision => '8',                        # integer
3356        date => '2004-02-24T17:01:44.108345Z',  # commit date
3357        author => 'committer name'
3358};
3359
3360
3361@mods = array of diff-index line hashes, each element represents one line
3362        of diff-index output
3363
3364diff-index line ($m hash)
3365{
3366        mode_a => first column of diff-index output, no leading ':',
3367        mode_b => second column of diff-index output,
3368        sha1_b => sha1sum of the final blob,
3369        chg => change type [MCRADT],
3370        file_a => original file name of a file (iff chg is 'C' or 'R')
3371        file_b => new/current file name of a file (any chg)
3372}
3373;
3374
3375# retval of read_url_paths{,_all}();
3376$l_map = {
3377        # repository root url
3378        'https://svn.musicpd.org' => {
3379                # repository path               # GIT_SVN_ID
3380                'mpd/trunk'             =>      'trunk',
3381                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
3382        },
3383}
3384
3385Notes:
3386        I don't trust the each() function on unless I created %hash myself
3387        because the internal iterator may not have started at base.