b3d3f479da727241034001491ad0e9b3b999335a
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $SVN_URL $SVN_INFO $SVN_WC $SVN_UUID
   8                $GIT_SVN_INDEX $GIT_SVN
   9                $GIT_DIR $GIT_SVN_DIR $REVDB/;
  10$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  11$VERSION = '1.1.1-broken';
  12
  13use Cwd qw/abs_path/;
  14$GIT_DIR = abs_path($ENV{GIT_DIR} || '.git');
  15$ENV{GIT_DIR} = $GIT_DIR;
  16
  17my $LC_ALL = $ENV{LC_ALL};
  18my $TZ = $ENV{TZ};
  19# make sure the svn binary gives consistent output between locales and TZs:
  20$ENV{TZ} = 'UTC';
  21$ENV{LC_ALL} = 'C';
  22$| = 1; # unbuffer STDOUT
  23
  24# If SVN:: library support is added, please make the dependencies
  25# optional and preserve the capability to use the command-line client.
  26# use eval { require SVN::... } to make it lazy load
  27# We don't use any modules not in the standard Perl distribution:
  28use Carp qw/croak/;
  29use IO::File qw//;
  30use File::Basename qw/dirname basename/;
  31use File::Path qw/mkpath/;
  32use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  33use File::Spec qw//;
  34use POSIX qw/strftime/;
  35use IPC::Open3;
  36use Memoize;
  37memoize('revisions_eq');
  38memoize('cmt_metadata');
  39memoize('get_commit_time');
  40
  41my ($SVN_PATH, $SVN, $SVN_LOG, $_use_lib);
  42$_use_lib = 1 unless $ENV{GIT_SVN_NO_LIB};
  43libsvn_load();
  44my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  45my $sha1 = qr/[a-f\d]{40}/;
  46my $sha1_short = qr/[a-f\d]{4,40}/;
  47my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
  48        $_find_copies_harder, $_l, $_cp_similarity, $_cp_remote,
  49        $_repack, $_repack_nr, $_repack_flags, $_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 $cmd =~ /^(?:init|rebuild|multi-init)$/;
 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        $SVN_URL = shift or die "SVN repository location required " .
 268                                "as a command-line argument\n";
 269        $SVN_URL =~ s!/+$!!; # strip trailing slash
 270        unless (-d $GIT_DIR) {
 271                my @init_db = ('git-init-db');
 272                push @init_db, "--template=$_template" if defined $_template;
 273                push @init_db, "--shared" if defined $_shared;
 274                sys(@init_db);
 275        }
 276        setup_git_svn();
 277}
 278
 279sub fetch {
 280        check_upgrade_needed();
 281        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 282        my $ret = $_use_lib ? fetch_lib(@_) : fetch_cmd(@_);
 283        if ($ret->{commit} && quiet_run(qw(git-rev-parse --verify
 284                                                refs/heads/master^0))) {
 285                sys(qw(git-update-ref refs/heads/master),$ret->{commit});
 286        }
 287        return $ret;
 288}
 289
 290sub fetch_cmd {
 291        my (@parents) = @_;
 292        my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 293        unless ($_revision) {
 294                $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 295        }
 296        push @log_args, "-r$_revision";
 297        push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 298
 299        my $svn_log = svn_log_raw(@log_args);
 300
 301        my $base = next_log_entry($svn_log) or croak "No base revision!\n";
 302        # don't need last_revision from grab_base_rev() because
 303        # user could've specified a different revision to skip (they
 304        # didn't want to import certain revisions into git for whatever
 305        # reason, so trust $base->{revision} instead.
 306        my (undef, $last_commit) = svn_grab_base_rev();
 307        unless (-d $SVN_WC) {
 308                svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC);
 309                chdir $SVN_WC or croak $!;
 310                read_uuid();
 311                $last_commit = git_commit($base, @parents);
 312                assert_tree($last_commit);
 313        } else {
 314                chdir $SVN_WC or croak $!;
 315                read_uuid();
 316                # looks like a user manually cp'd and svn switch'ed
 317                unless ($last_commit) {
 318                        sys(qw/svn revert -R ./);
 319                        assert_svn_wc_clean($base->{revision});
 320                        $last_commit = git_commit($base, @parents);
 321                        assert_tree($last_commit);
 322                }
 323        }
 324        my @svn_up = qw(svn up);
 325        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 326        my $last = $base;
 327        while (my $log_msg = next_log_entry($svn_log)) {
 328                if ($last->{revision} >= $log_msg->{revision}) {
 329                        croak "Out of order: last >= current: ",
 330                                "$last->{revision} >= $log_msg->{revision}\n";
 331                }
 332                # Revert is needed for cases like:
 333                # https://svn.musicpd.org/Jamming/trunk (r166:167), but
 334                # I can't seem to reproduce something like that on a test...
 335                sys(qw/svn revert -R ./);
 336                assert_svn_wc_clean($last->{revision});
 337                sys(@svn_up,"-r$log_msg->{revision}");
 338                $last_commit = git_commit($log_msg, $last_commit, @parents);
 339                $last = $log_msg;
 340        }
 341        close $svn_log->{fh};
 342        $last->{commit} = $last_commit;
 343        return $last;
 344}
 345
 346sub fetch_lib {
 347        my (@parents) = @_;
 348        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 349        my $repo;
 350        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 351        $SVN_LOG ||= libsvn_connect($repo);
 352        $SVN ||= libsvn_connect($repo);
 353        my ($last_rev, $last_commit) = svn_grab_base_rev();
 354        my ($base, $head) = libsvn_parse_revision($last_rev);
 355        if ($base > $head) {
 356                return { revision => $last_rev, commit => $last_commit }
 357        }
 358        my $index = set_index($GIT_SVN_INDEX);
 359
 360        # limit ourselves and also fork() since get_log won't release memory
 361        # after processing a revision and SVN stuff seems to leak
 362        my $inc = 1000;
 363        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
 364        read_uuid();
 365        if (defined $last_commit) {
 366                unless (-e $GIT_SVN_INDEX) {
 367                        sys(qw/git-read-tree/, $last_commit);
 368                }
 369                chomp (my $x = `git-write-tree`);
 370                my ($y) = (`git-cat-file commit $last_commit`
 371                                                        =~ /^tree ($sha1)/m);
 372                if ($y ne $x) {
 373                        unlink $GIT_SVN_INDEX or croak $!;
 374                        sys(qw/git-read-tree/, $last_commit);
 375                }
 376                chomp ($x = `git-write-tree`);
 377                if ($y ne $x) {
 378                        print STDERR "trees ($last_commit) $y != $x\n",
 379                                 "Something is seriously wrong...\n";
 380                }
 381        }
 382        while (1) {
 383                # fork, because using SVN::Pool with get_log() still doesn't
 384                # seem to help enough to keep memory usage down.
 385                defined(my $pid = fork) or croak $!;
 386                if (!$pid) {
 387                        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
 388
 389                        # Yes I'm perfectly aware that the fourth argument
 390                        # below is the limit revisions number.  Unfortunately
 391                        # performance sucks with it enabled, so it's much
 392                        # faster to fetch revision ranges instead of relying
 393                        # on the limiter.
 394                        libsvn_get_log($SVN_LOG, '/'.$SVN_PATH,
 395                                        $min, $max, 0, 1, 1,
 396                                sub {
 397                                        my $log_msg;
 398                                        if ($last_commit) {
 399                                                $log_msg = libsvn_fetch(
 400                                                        $last_commit, @_);
 401                                                $last_commit = git_commit(
 402                                                        $log_msg,
 403                                                        $last_commit,
 404                                                        @parents);
 405                                        } else {
 406                                                $log_msg = libsvn_new_tree(@_);
 407                                                $last_commit = git_commit(
 408                                                        $log_msg, @parents);
 409                                        }
 410                                });
 411                        exit 0;
 412                }
 413                waitpid $pid, 0;
 414                croak $? if $?;
 415                ($last_rev, $last_commit) = svn_grab_base_rev();
 416                last if ($max >= $head);
 417                $min = $max + 1;
 418                $max += $inc;
 419                $max = $head if ($max > $head);
 420        }
 421        restore_index($index);
 422        return { revision => $last_rev, commit => $last_commit };
 423}
 424
 425sub commit {
 426        my (@commits) = @_;
 427        check_upgrade_needed();
 428        if ($_stdin || !@commits) {
 429                print "Reading from stdin...\n";
 430                @commits = ();
 431                while (<STDIN>) {
 432                        if (/\b($sha1_short)\b/o) {
 433                                unshift @commits, $1;
 434                        }
 435                }
 436        }
 437        my @revs;
 438        foreach my $c (@commits) {
 439                chomp(my @tmp = safe_qx('git-rev-parse',$c));
 440                if (scalar @tmp == 1) {
 441                        push @revs, $tmp[0];
 442                } elsif (scalar @tmp > 1) {
 443                        push @revs, reverse (safe_qx('git-rev-list',@tmp));
 444                } else {
 445                        die "Failed to rev-parse $c\n";
 446                }
 447        }
 448        chomp @revs;
 449        $_use_lib ? commit_lib(@revs) : commit_cmd(@revs);
 450        print "Done committing ",scalar @revs," revisions to SVN\n";
 451}
 452
 453sub commit_cmd {
 454        my (@revs) = @_;
 455
 456        chdir $SVN_WC or croak "Unable to chdir $SVN_WC: $!\n";
 457        my $info = svn_info('.');
 458        my $fetched = fetch();
 459        if ($info->{Revision} != $fetched->{revision}) {
 460                print STDERR "There are new revisions that were fetched ",
 461                                "and need to be merged (or acknowledged) ",
 462                                "before committing.\n";
 463                exit 1;
 464        }
 465        $info = svn_info('.');
 466        read_uuid($info);
 467        my $last = $fetched;
 468        foreach my $c (@revs) {
 469                my $mods = svn_checkout_tree($last, $c);
 470                if (scalar @$mods == 0) {
 471                        print "Skipping, no changes detected\n";
 472                        next;
 473                }
 474                $last = svn_commit_tree($last, $c);
 475        }
 476}
 477
 478sub commit_lib {
 479        my (@revs) = @_;
 480        my ($r_last, $cmt_last) = svn_grab_base_rev();
 481        defined $r_last or die "Must have an existing revision to commit\n";
 482        my $fetched = fetch();
 483        if ($r_last != $fetched->{revision}) {
 484                print STDERR "There are new revisions that were fetched ",
 485                                "and need to be merged (or acknowledged) ",
 486                                "before committing.\n",
 487                                "last rev: $r_last\n",
 488                                " current: $fetched->{revision}\n";
 489                exit 1;
 490        }
 491        read_uuid();
 492        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 493        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
 494
 495        set_svn_commit_env();
 496        foreach my $c (@revs) {
 497                my $log_msg = get_commit_message($c, $commit_msg);
 498
 499                # fork for each commit because there's a memory leak I
 500                # can't track down... (it's probably in the SVN code)
 501                defined(my $pid = open my $fh, '-|') or croak $!;
 502                if (!$pid) {
 503                        my $ed = SVN::Git::Editor->new(
 504                                        {       r => $r_last,
 505                                                ra => $SVN,
 506                                                c => $c,
 507                                                svn_path => $SVN_PATH
 508                                        },
 509                                        $SVN->get_commit_editor(
 510                                                $log_msg->{msg},
 511                                                sub {
 512                                                        libsvn_commit_cb(
 513                                                                @_, $c,
 514                                                                $log_msg->{msg},
 515                                                                $r_last,
 516                                                                $cmt_last)
 517                                                },
 518                                                @lock)
 519                                        );
 520                        my $mods = libsvn_checkout_tree($cmt_last, $c, $ed);
 521                        if (@$mods == 0) {
 522                                print "No changes\nr$r_last = $cmt_last\n";
 523                                $ed->abort_edit;
 524                        } else {
 525                                $ed->close_edit;
 526                        }
 527                        exit 0;
 528                }
 529                my ($r_new, $cmt_new, $no);
 530                while (<$fh>) {
 531                        print $_;
 532                        chomp;
 533                        if (/^r(\d+) = ($sha1)$/o) {
 534                                ($r_new, $cmt_new) = ($1, $2);
 535                        } elsif ($_ eq 'No changes') {
 536                                $no = 1;
 537                        }
 538                }
 539                close $fh or croak $?;
 540                if (! defined $r_new && ! defined $cmt_new) {
 541                        unless ($no) {
 542                                die "Failed to parse revision information\n";
 543                        }
 544                } else {
 545                        ($r_last, $cmt_last) = ($r_new, $cmt_new);
 546                }
 547        }
 548        $ENV{LC_ALL} = 'C';
 549        unlink $commit_msg;
 550}
 551
 552sub show_ignore {
 553        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 554        $_use_lib ? show_ignore_lib() : show_ignore_cmd();
 555}
 556
 557sub show_ignore_cmd {
 558        require File::Find or die $!;
 559        if (defined $_revision) {
 560                die "-r/--revision option doesn't work unless the Perl SVN ",
 561                        "libraries are used\n";
 562        }
 563        chdir $SVN_WC or croak $!;
 564        my %ign;
 565        File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 566                s#^\./##;
 567                @{$ign{$_}} = svn_propget_base('svn:ignore', $_);
 568                }}, no_chdir=>1},'.');
 569
 570        print "\n# /\n";
 571        foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 572        delete $ign{'.'};
 573        foreach my $i (sort keys %ign) {
 574                print "\n# ",$i,"\n";
 575                foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 576        }
 577}
 578
 579sub show_ignore_lib {
 580        my $repo;
 581        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 582        $SVN ||= libsvn_connect($repo);
 583        my $r = defined $_revision ? $_revision : $SVN->get_latest_revnum;
 584        libsvn_traverse_ignore(\*STDOUT, $SVN_PATH, $r);
 585}
 586
 587sub graft_branches {
 588        my $gr_file = "$GIT_DIR/info/grafts";
 589        my ($grafts, $comments) = read_grafts($gr_file);
 590        my $gr_sha1;
 591
 592        if (%$grafts) {
 593                # temporarily disable our grafts file to make this idempotent
 594                chomp($gr_sha1 = safe_qx(qw/git-hash-object -w/,$gr_file));
 595                rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
 596        }
 597
 598        my $l_map = read_url_paths();
 599        my @re = map { qr/$_/is } @_opt_m if @_opt_m;
 600        unless ($_no_default_regex) {
 601                push @re, (qr/\b(?:merge|merging|merged)\s+with\s+([\w\.\-]+)/i,
 602                        qr/\b(?:merge|merging|merged)\s+([\w\.\-]+)/i,
 603                        qr/\b(?:from|of)\s+([\w\.\-]+)/i );
 604        }
 605        foreach my $u (keys %$l_map) {
 606                if (@re) {
 607                        foreach my $p (keys %{$l_map->{$u}}) {
 608                                graft_merge_msg($grafts,$l_map,$u,$p,@re);
 609                        }
 610                }
 611                unless ($_no_graft_copy) {
 612                        if ($_use_lib) {
 613                                graft_file_copy_lib($grafts,$l_map,$u);
 614                        } else {
 615                                graft_file_copy_cmd($grafts,$l_map,$u);
 616                        }
 617                }
 618        }
 619        graft_tree_joins($grafts);
 620
 621        write_grafts($grafts, $comments, $gr_file);
 622        unlink "$gr_file~$gr_sha1" if $gr_sha1;
 623}
 624
 625sub multi_init {
 626        my $url = shift;
 627        $_trunk ||= 'trunk';
 628        $_trunk =~ s#/+$##;
 629        $url =~ s#/+$## if $url;
 630        if ($_trunk !~ m#^[a-z\+]+://#) {
 631                $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#);
 632                unless ($url) {
 633                        print STDERR "E: '$_trunk' is not a complete URL ",
 634                                "and a separate URL is not specified\n";
 635                        exit 1;
 636                }
 637                $_trunk = $url . $_trunk;
 638        }
 639        if ($GIT_SVN eq 'git-svn') {
 640                print "GIT_SVN_ID set to 'trunk' for $_trunk\n";
 641                $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
 642        }
 643        init_vars();
 644        init($_trunk);
 645        complete_url_ls_init($url, $_branches, '--branches/-b', '');
 646        complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/');
 647}
 648
 649sub multi_fetch {
 650        # try to do trunk first, since branches/tags
 651        # may be descended from it.
 652        if (-e "$GIT_DIR/svn/trunk/info/url") {
 653                fetch_child_id('trunk', @_);
 654        }
 655        rec_fetch('', "$GIT_DIR/svn", @_);
 656}
 657
 658sub show_log {
 659        my (@args) = @_;
 660        my ($r_min, $r_max);
 661        my $r_last = -1; # prevent dupes
 662        rload_authors() if $_authors;
 663        if (defined $TZ) {
 664                $ENV{TZ} = $TZ;
 665        } else {
 666                delete $ENV{TZ};
 667        }
 668        if (defined $_revision) {
 669                if ($_revision =~ /^(\d+):(\d+)$/) {
 670                        ($r_min, $r_max) = ($1, $2);
 671                } elsif ($_revision =~ /^\d+$/) {
 672                        $r_min = $r_max = $_revision;
 673                } else {
 674                        print STDERR "-r$_revision is not supported, use ",
 675                                "standard \'git log\' arguments instead\n";
 676                        exit 1;
 677                }
 678        }
 679
 680        my $pid = open(my $log,'-|');
 681        defined $pid or croak $!;
 682        if (!$pid) {
 683                exec(git_svn_log_cmd($r_min,$r_max), @args) or croak $!;
 684        }
 685        setup_pager();
 686        my (@k, $c, $d);
 687
 688        while (<$log>) {
 689                if (/^commit ($sha1_short)/o) {
 690                        my $cmt = $1;
 691                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
 692                                $r_last = $c->{r};
 693                                process_commit($c, $r_min, $r_max, \@k) or
 694                                                                goto out;
 695                        }
 696                        $d = undef;
 697                        $c = { c => $cmt };
 698                } elsif (/^author (.+) (\d+) ([\-\+]?\d+)$/) {
 699                        get_author_info($c, $1, $2, $3);
 700                } elsif (/^(?:tree|parent|committer) /) {
 701                        # ignore
 702                } elsif (/^:\d{6} \d{6} $sha1_short/o) {
 703                        push @{$c->{raw}}, $_;
 704                } elsif (/^diff /) {
 705                        $d = 1;
 706                        push @{$c->{diff}}, $_;
 707                } elsif ($d) {
 708                        push @{$c->{diff}}, $_;
 709                } elsif (/^    (git-svn-id:.+)$/) {
 710                        (undef, $c->{r}, undef) = extract_metadata($1);
 711                } elsif (s/^    //) {
 712                        push @{$c->{l}}, $_;
 713                }
 714        }
 715        if ($c && defined $c->{r} && $c->{r} != $r_last) {
 716                $r_last = $c->{r};
 717                process_commit($c, $r_min, $r_max, \@k);
 718        }
 719        if (@k) {
 720                my $swap = $r_max;
 721                $r_max = $r_min;
 722                $r_min = $swap;
 723                process_commit($_, $r_min, $r_max) foreach reverse @k;
 724        }
 725out:
 726        close $log;
 727        print '-' x72,"\n" unless $_incremental || $_oneline;
 728}
 729
 730sub commit_diff_usage {
 731        print STDERR "Usage: $0 commit-diff <tree-ish> <tree-ish> [<URL>]\n";
 732        exit 1
 733}
 734
 735sub commit_diff {
 736        if (!$_use_lib) {
 737                print STDERR "commit-diff must be used with SVN libraries\n";
 738                exit 1;
 739        }
 740        my $ta = shift or commit_diff_usage();
 741        my $tb = shift or commit_diff_usage();
 742        if (!eval { $SVN_URL = shift || file_to_s("$GIT_SVN_DIR/info/url") }) {
 743                print STDERR "Needed URL or usable git-svn id command-line\n";
 744                commit_diff_usage();
 745        }
 746        if (defined $_message && defined $_file) {
 747                print STDERR "Both --message/-m and --file/-F specified ",
 748                                "for the commit message.\n",
 749                                "I have no idea what you mean\n";
 750                exit 1;
 751        }
 752        if (defined $_file) {
 753                $_message = file_to_s($_message);
 754        } else {
 755                $_message ||= get_commit_message($tb,
 756                                        "$GIT_DIR/.svn-commit.tmp.$$")->{msg};
 757        }
 758        my $repo;
 759        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 760        $SVN_LOG ||= libsvn_connect($repo);
 761        $SVN ||= libsvn_connect($repo);
 762        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 763        my $ed = SVN::Git::Editor->new({        r => $SVN->get_latest_revnum,
 764                                                ra => $SVN, c => $tb,
 765                                                svn_path => $SVN_PATH
 766                                        },
 767                                $SVN->get_commit_editor($_message,
 768                                        sub {print "Committed $_[0]\n"},@lock)
 769                                );
 770        my $mods = libsvn_checkout_tree($ta, $tb, $ed);
 771        if (@$mods == 0) {
 772                print "No changes\n$ta == $tb\n";
 773                $ed->abort_edit;
 774        } else {
 775                $ed->close_edit;
 776        }
 777}
 778
 779########################### utility functions #########################
 780
 781sub cmt_showable {
 782        my ($c) = @_;
 783        return 1 if defined $c->{r};
 784        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
 785                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
 786                my @msg = safe_qx(qw/git-cat-file commit/, $c->{c});
 787                shift @msg while ($msg[0] ne "\n");
 788                shift @msg;
 789                @{$c->{l}} = grep !/^git-svn-id: /, @msg;
 790
 791                (undef, $c->{r}, undef) = extract_metadata(
 792                                (grep(/^git-svn-id: /, @msg))[-1]);
 793        }
 794        return defined $c->{r};
 795}
 796
 797sub git_svn_log_cmd {
 798        my ($r_min, $r_max) = @_;
 799        my @cmd = (qw/git-log --abbrev-commit --pretty=raw
 800                        --default/, "refs/remotes/$GIT_SVN");
 801        push @cmd, '--summary' if $_verbose;
 802        return @cmd unless defined $r_max;
 803        if ($r_max == $r_min) {
 804                push @cmd, '--max-count=1';
 805                if (my $c = revdb_get($REVDB, $r_max)) {
 806                        push @cmd, $c;
 807                }
 808        } else {
 809                my ($c_min, $c_max);
 810                $c_max = revdb_get($REVDB, $r_max);
 811                $c_min = revdb_get($REVDB, $r_min);
 812                if ($c_min && $c_max) {
 813                        if ($r_max > $r_max) {
 814                                push @cmd, "$c_min..$c_max";
 815                        } else {
 816                                push @cmd, "$c_max..$c_min";
 817                        }
 818                } elsif ($r_max > $r_min) {
 819                        push @cmd, $c_max;
 820                } else {
 821                        push @cmd, $c_min;
 822                }
 823        }
 824        return @cmd;
 825}
 826
 827sub fetch_child_id {
 828        my $id = shift;
 829        print "Fetching $id\n";
 830        my $ref = "$GIT_DIR/refs/remotes/$id";
 831        defined(my $pid = open my $fh, '-|') or croak $!;
 832        if (!$pid) {
 833                $_repack = undef;
 834                $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 835                init_vars();
 836                fetch(@_);
 837                exit 0;
 838        }
 839        while (<$fh>) {
 840                print $_;
 841                check_repack() if (/^r\d+ = $sha1/);
 842        }
 843        close $fh or croak $?;
 844}
 845
 846sub rec_fetch {
 847        my ($pfx, $p, @args) = @_;
 848        my @dir;
 849        foreach (sort <$p/*>) {
 850                if (-r "$_/info/url") {
 851                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
 852                        my $id = $pfx . basename $_;
 853                        next if $id eq 'trunk';
 854                        fetch_child_id($id, @args);
 855                } elsif (-d $_) {
 856                        push @dir, $_;
 857                }
 858        }
 859        foreach (@dir) {
 860                my $x = $_;
 861                $x =~ s!^\Q$GIT_DIR\E/svn/!!;
 862                rec_fetch($x, $_);
 863        }
 864}
 865
 866sub complete_url_ls_init {
 867        my ($url, $var, $switch, $pfx) = @_;
 868        unless ($var) {
 869                print STDERR "W: $switch not specified\n";
 870                return;
 871        }
 872        $var =~ s#/+$##;
 873        if ($var !~ m#^[a-z\+]+://#) {
 874                $var = '/' . $var if ($var !~ m#^/#);
 875                unless ($url) {
 876                        print STDERR "E: '$var' is not a complete URL ",
 877                                "and a separate URL is not specified\n";
 878                        exit 1;
 879                }
 880                $var = $url . $var;
 881        }
 882        chomp(my @ls = $_use_lib ? libsvn_ls_fullurl($var)
 883                                : safe_qx(qw/svn ls --non-interactive/, $var));
 884        my $old = $GIT_SVN;
 885        defined(my $pid = fork) or croak $!;
 886        if (!$pid) {
 887                foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
 888                        $u =~ s#/+$##;
 889                        if ($u !~ m!\Q$var\E/(.+)$!) {
 890                                print STDERR "W: Unrecognized URL: $u\n";
 891                                die "This should never happen\n";
 892                        }
 893                        my $id = $pfx.$1;
 894                        print "init $u => $id\n";
 895                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 896                        init_vars();
 897                        init($u);
 898                }
 899                exit 0;
 900        }
 901        waitpid $pid, 0;
 902        croak $? if $?;
 903}
 904
 905sub common_prefix {
 906        my $paths = shift;
 907        my %common;
 908        foreach (@$paths) {
 909                my @tmp = split m#/#, $_;
 910                my $p = '';
 911                while (my $x = shift @tmp) {
 912                        $p .= "/$x";
 913                        $common{$p} ||= 0;
 914                        $common{$p}++;
 915                }
 916        }
 917        foreach (sort {length $b <=> length $a} keys %common) {
 918                if ($common{$_} == @$paths) {
 919                        return $_;
 920                }
 921        }
 922        return '';
 923}
 924
 925# grafts set here are 'stronger' in that they're based on actual tree
 926# matches, and won't be deleted from merge-base checking in write_grafts()
 927sub graft_tree_joins {
 928        my $grafts = shift;
 929        map_tree_joins() if (@_branch_from && !%tree_map);
 930        return unless %tree_map;
 931
 932        git_svn_each(sub {
 933                my $i = shift;
 934                defined(my $pid = open my $fh, '-|') or croak $!;
 935                if (!$pid) {
 936                        exec qw/git-rev-list --pretty=raw/,
 937                                        "refs/remotes/$i" or croak $!;
 938                }
 939                while (<$fh>) {
 940                        next unless /^commit ($sha1)$/o;
 941                        my $c = $1;
 942                        my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
 943                        next unless $tree_map{$t};
 944
 945                        my $l;
 946                        do {
 947                                $l = readline $fh;
 948                        } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
 949
 950                        my ($s, $tz) = ($1, $2);
 951                        if ($tz =~ s/^\+//) {
 952                                $s += tz_to_s_offset($tz);
 953                        } elsif ($tz =~ s/^\-//) {
 954                                $s -= tz_to_s_offset($tz);
 955                        }
 956
 957                        my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
 958
 959                        foreach my $p (@{$tree_map{$t}}) {
 960                                next if $p eq $c;
 961                                my $mb = eval {
 962                                        safe_qx('git-merge-base', $c, $p)
 963                                };
 964                                next unless ($@ || $?);
 965                                if (defined $r_a) {
 966                                        # see if SVN says it's a relative
 967                                        my ($url_b, $r_b, $uuid_b) =
 968                                                        cmt_metadata($p);
 969                                        next if (defined $url_b &&
 970                                                        defined $url_a &&
 971                                                        ($url_a eq $url_b) &&
 972                                                        ($uuid_a eq $uuid_b));
 973                                        if ($uuid_a eq $uuid_b) {
 974                                                if ($r_b < $r_a) {
 975                                                        $grafts->{$c}->{$p} = 2;
 976                                                        next;
 977                                                } elsif ($r_b > $r_a) {
 978                                                        $grafts->{$p}->{$c} = 2;
 979                                                        next;
 980                                                }
 981                                        }
 982                                }
 983                                my $ct = get_commit_time($p);
 984                                if ($ct < $s) {
 985                                        $grafts->{$c}->{$p} = 2;
 986                                } elsif ($ct > $s) {
 987                                        $grafts->{$p}->{$c} = 2;
 988                                }
 989                                # what should we do when $ct == $s ?
 990                        }
 991                }
 992                close $fh or croak $?;
 993        });
 994}
 995
 996# this isn't funky-filename safe, but good enough for now...
 997sub graft_file_copy_cmd {
 998        my ($grafts, $l_map, $u) = @_;
 999        my $paths = $l_map->{$u};
1000        my $pfx = common_prefix([keys %$paths]);
1001        $SVN_URL ||= $u.$pfx;
1002        my $pid = open my $fh, '-|';
1003        defined $pid or croak $!;
1004        unless ($pid) {
1005                my @exec = qw/svn log -v/;
1006                push @exec, "-r$_revision" if defined $_revision;
1007                exec @exec, $u.$pfx or croak $!;
1008        }
1009        my ($r, $mp) = (undef, undef);
1010        while (<$fh>) {
1011                chomp;
1012                if (/^\-{72}$/) {
1013                        $mp = $r = undef;
1014                } elsif (/^r(\d+) \| /) {
1015                        $r = $1 unless defined $r;
1016                } elsif (/^Changed paths:/) {
1017                        $mp = 1;
1018                } elsif ($mp && m#^   [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) {
1019                        my ($p1, $p0, $r0) = ($1, $2, $3);
1020                        my $c = find_graft_path_commit($paths, $p1, $r);
1021                        next unless $c;
1022                        find_graft_path_parents($grafts, $paths, $c, $p0, $r0);
1023                }
1024        }
1025}
1026
1027sub graft_file_copy_lib {
1028        my ($grafts, $l_map, $u) = @_;
1029        my $tree_paths = $l_map->{$u};
1030        my $pfx = common_prefix([keys %$tree_paths]);
1031        my ($repo, $path) = repo_path_split($u.$pfx);
1032        $SVN_LOG ||= libsvn_connect($repo);
1033        $SVN ||= libsvn_connect($repo);
1034
1035        my ($base, $head) = libsvn_parse_revision();
1036        my $inc = 1000;
1037        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
1038        my $eh = $SVN::Error::handler;
1039        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
1040        while (1) {
1041                my $pool = SVN::Pool->new;
1042                libsvn_get_log($SVN_LOG, "/$path", $min, $max, 0, 1, 1,
1043                        sub {
1044                                libsvn_graft_file_copies($grafts, $tree_paths,
1045                                                        $path, @_);
1046                        }, $pool);
1047                $pool->clear;
1048                last if ($max >= $head);
1049                $min = $max + 1;
1050                $max += $inc;
1051                $max = $head if ($max > $head);
1052        }
1053        $SVN::Error::handler = $eh;
1054}
1055
1056sub process_merge_msg_matches {
1057        my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
1058        my (@strong, @weak);
1059        foreach (@matches) {
1060                # merging with ourselves is not interesting
1061                next if $_ eq $p;
1062                if ($l_map->{$u}->{$_}) {
1063                        push @strong, $_;
1064                } else {
1065                        push @weak, $_;
1066                }
1067        }
1068        foreach my $w (@weak) {
1069                last if @strong;
1070                # no exact match, use branch name as regexp.
1071                my $re = qr/\Q$w\E/i;
1072                foreach (keys %{$l_map->{$u}}) {
1073                        if (/$re/) {
1074                                push @strong, $l_map->{$u}->{$_};
1075                                last;
1076                        }
1077                }
1078                last if @strong;
1079                $w = basename($w);
1080                $re = qr/\Q$w\E/i;
1081                foreach (keys %{$l_map->{$u}}) {
1082                        if (/$re/) {
1083                                push @strong, $l_map->{$u}->{$_};
1084                                last;
1085                        }
1086                }
1087        }
1088        my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
1089                                        \s(?:[a-f\d\-]+)$/xsm);
1090        unless (defined $rev) {
1091                ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
1092                                        \@(?:[a-f\d\-]+)/xsm);
1093                return unless defined $rev;
1094        }
1095        foreach my $m (@strong) {
1096                my ($r0, $s0) = find_rev_before($rev, $m, 1);
1097                $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
1098        }
1099}
1100
1101sub graft_merge_msg {
1102        my ($grafts, $l_map, $u, $p, @re) = @_;
1103
1104        my $x = $l_map->{$u}->{$p};
1105        my $rl = rev_list_raw($x);
1106        while (my $c = next_rev_list_entry($rl)) {
1107                foreach my $re (@re) {
1108                        my (@br) = ($c->{m} =~ /$re/g);
1109                        next unless @br;
1110                        process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
1111                }
1112        }
1113}
1114
1115sub read_uuid {
1116        return if $SVN_UUID;
1117        if ($_use_lib) {
1118                my $pool = SVN::Pool->new;
1119                $SVN_UUID = $SVN->get_uuid($pool);
1120                $pool->clear;
1121        } else {
1122                my $info = shift || svn_info('.');
1123                $SVN_UUID = $info->{'Repository UUID'} or
1124                                        croak "Repository UUID unreadable\n";
1125        }
1126}
1127
1128sub quiet_run {
1129        my $pid = fork;
1130        defined $pid or croak $!;
1131        if (!$pid) {
1132                open my $null, '>', '/dev/null' or croak $!;
1133                open STDERR, '>&', $null or croak $!;
1134                open STDOUT, '>&', $null or croak $!;
1135                exec @_ or croak $!;
1136        }
1137        waitpid $pid, 0;
1138        return $?;
1139}
1140
1141sub repo_path_split {
1142        my $full_url = shift;
1143        $full_url =~ s#/+$##;
1144
1145        foreach (@repo_path_split_cache) {
1146                if ($full_url =~ s#$_##) {
1147                        my $u = $1;
1148                        $full_url =~ s#^/+##;
1149                        return ($u, $full_url);
1150                }
1151        }
1152
1153        my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i);
1154        $path =~ s#^/+##;
1155        my @paths = split(m#/+#, $path);
1156
1157        if ($_use_lib) {
1158                while (1) {
1159                        $SVN = libsvn_connect($url);
1160                        last if (defined $SVN &&
1161                                defined eval { $SVN->get_latest_revnum });
1162                        my $n = shift @paths || last;
1163                        $url .= "/$n";
1164                }
1165        } else {
1166                while (quiet_run(qw/svn ls --non-interactive/, $url)) {
1167                        my $n = shift @paths || last;
1168                        $url .= "/$n";
1169                }
1170        }
1171        push @repo_path_split_cache, qr/^(\Q$url\E)/;
1172        $path = join('/',@paths);
1173        return ($url, $path);
1174}
1175
1176sub setup_git_svn {
1177        defined $SVN_URL or croak "SVN repository location required\n";
1178        unless (-d $GIT_DIR) {
1179                croak "GIT_DIR=$GIT_DIR does not exist!\n";
1180        }
1181        mkpath([$GIT_SVN_DIR]);
1182        mkpath(["$GIT_SVN_DIR/info"]);
1183        open my $fh, '>>',$REVDB or croak $!;
1184        close $fh;
1185        s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1186
1187}
1188
1189sub assert_svn_wc_clean {
1190        return if $_use_lib;
1191        my ($svn_rev) = @_;
1192        croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
1193        my $lcr = svn_info('.')->{'Last Changed Rev'};
1194        if ($svn_rev != $lcr) {
1195                print STDERR "Checking for copy-tree ... ";
1196                my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
1197                                                "-r$lcr:$svn_rev")));
1198                if (@diff) {
1199                        croak "Nope!  Expected r$svn_rev, got r$lcr\n";
1200                } else {
1201                        print STDERR "OK!\n";
1202                }
1203        }
1204        my @status = grep(!/^Performing status on external/,(`svn status`));
1205        @status = grep(!/^\s*$/,@status);
1206        if (scalar @status) {
1207                print STDERR "Tree ($SVN_WC) is not clean:\n";
1208                print STDERR $_ foreach @status;
1209                croak;
1210        }
1211}
1212
1213sub get_tree_from_treeish {
1214        my ($treeish) = @_;
1215        croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1216        chomp(my $type = `git-cat-file -t $treeish`);
1217        my $expected;
1218        while ($type eq 'tag') {
1219                chomp(($treeish, $type) = `git-cat-file tag $treeish`);
1220        }
1221        if ($type eq 'commit') {
1222                $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
1223                ($expected) = ($expected =~ /^tree ($sha1)$/);
1224                die "Unable to get tree from $treeish\n" unless $expected;
1225        } elsif ($type eq 'tree') {
1226                $expected = $treeish;
1227        } else {
1228                die "$treeish is a $type, expected tree, tag or commit\n";
1229        }
1230        return $expected;
1231}
1232
1233sub assert_tree {
1234        return if $_use_lib;
1235        my ($treeish) = @_;
1236        my $expected = get_tree_from_treeish($treeish);
1237
1238        my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
1239        if (-e $tmpindex) {
1240                unlink $tmpindex or croak $!;
1241        }
1242        my $old_index = set_index($tmpindex);
1243        index_changes(1);
1244        chomp(my $tree = `git-write-tree`);
1245        restore_index($old_index);
1246        if ($tree ne $expected) {
1247                croak "Tree mismatch, Got: $tree, Expected: $expected\n";
1248        }
1249        unlink $tmpindex;
1250}
1251
1252sub parse_diff_tree {
1253        my $diff_fh = shift;
1254        local $/ = "\0";
1255        my $state = 'meta';
1256        my @mods;
1257        while (<$diff_fh>) {
1258                chomp $_; # this gets rid of the trailing "\0"
1259                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1260                                        $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1261                        push @mods, {   mode_a => $1, mode_b => $2,
1262                                        sha1_b => $3, chg => $4 };
1263                        if ($4 =~ /^(?:C|R)$/) {
1264                                $state = 'file_a';
1265                        } else {
1266                                $state = 'file_b';
1267                        }
1268                } elsif ($state eq 'file_a') {
1269                        my $x = $mods[$#mods] or croak "Empty array\n";
1270                        if ($x->{chg} !~ /^(?:C|R)$/) {
1271                                croak "Error parsing $_, $x->{chg}\n";
1272                        }
1273                        $x->{file_a} = $_;
1274                        $state = 'file_b';
1275                } elsif ($state eq 'file_b') {
1276                        my $x = $mods[$#mods] or croak "Empty array\n";
1277                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1278                                croak "Error parsing $_, $x->{chg}\n";
1279                        }
1280                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1281                                croak "Error parsing $_, $x->{chg}\n";
1282                        }
1283                        $x->{file_b} = $_;
1284                        $state = 'meta';
1285                } else {
1286                        croak "Error parsing $_\n";
1287                }
1288        }
1289        close $diff_fh or croak $?;
1290
1291        return \@mods;
1292}
1293
1294sub svn_check_prop_executable {
1295        my $m = shift;
1296        return if -l $m->{file_b};
1297        if ($m->{mode_b} =~ /755$/) {
1298                chmod((0755 &~ umask),$m->{file_b}) or croak $!;
1299                if ($m->{mode_a} !~ /755$/) {
1300                        sys(qw(svn propset svn:executable 1), $m->{file_b});
1301                }
1302                -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
1303        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
1304                sys(qw(svn propdel svn:executable), $m->{file_b});
1305                chmod((0644 &~ umask),$m->{file_b}) or croak $!;
1306                -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
1307        }
1308}
1309
1310sub svn_ensure_parent_path {
1311        my $dir_b = dirname(shift);
1312        svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
1313        mkpath([$dir_b]) unless (-d $dir_b);
1314        sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
1315}
1316
1317sub precommit_check {
1318        my $mods = shift;
1319        my (%rm_file, %rmdir_check, %added_check);
1320
1321        my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
1322        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1323                if ($m->{chg} eq 'R') {
1324                        if (-d $m->{file_b}) {
1325                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1326                        }
1327                        # dir/$file => dir/file/$file
1328                        my $dirname = dirname($m->{file_b});
1329                        while ($dirname ne File::Spec->curdir) {
1330                                if ($dirname ne $m->{file_a}) {
1331                                        $dirname = dirname($dirname);
1332                                        next;
1333                                }
1334                                err_file_to_dir("$m->{file_a} => $m->{file_b}");
1335                        }
1336                        # baz/zzz => baz (baz is a file)
1337                        $dirname = dirname($m->{file_a});
1338                        while ($dirname ne File::Spec->curdir) {
1339                                if ($dirname ne $m->{file_b}) {
1340                                        $dirname = dirname($dirname);
1341                                        next;
1342                                }
1343                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1344                        }
1345                }
1346                if ($m->{chg} =~ /^(D|R)$/) {
1347                        my $t = $1 eq 'D' ? 'file_b' : 'file_a';
1348                        $rm_file{ $m->{$t} } = 1;
1349                        my $dirname = dirname( $m->{$t} );
1350                        my $basename = basename( $m->{$t} );
1351                        $rmdir_check{$dirname}->{$basename} = 1;
1352                } elsif ($m->{chg} =~ /^(?:A|C)$/) {
1353                        if (-d $m->{file_b}) {
1354                                err_dir_to_file($m->{file_b});
1355                        }
1356                        my $dirname = dirname( $m->{file_b} );
1357                        my $basename = basename( $m->{file_b} );
1358                        $added_check{$dirname}->{$basename} = 1;
1359                        while ($dirname ne File::Spec->curdir) {
1360                                if ($rm_file{$dirname}) {
1361                                        err_file_to_dir($m->{file_b});
1362                                }
1363                                $dirname = dirname $dirname;
1364                        }
1365                }
1366        }
1367        return (\%rmdir_check, \%added_check);
1368
1369        sub err_dir_to_file {
1370                my $file = shift;
1371                print STDERR "Node change from directory to file ",
1372                                "is not supported by Subversion: ",$file,"\n";
1373                exit 1;
1374        }
1375        sub err_file_to_dir {
1376                my $file = shift;
1377                print STDERR "Node change from file to directory ",
1378                                "is not supported by Subversion: ",$file,"\n";
1379                exit 1;
1380        }
1381}
1382
1383
1384sub get_diff {
1385        my ($from, $treeish) = @_;
1386        assert_tree($from);
1387        print "diff-tree $from $treeish\n";
1388        my $pid = open my $diff_fh, '-|';
1389        defined $pid or croak $!;
1390        if ($pid == 0) {
1391                my @diff_tree = qw(git-diff-tree -z -r);
1392                if ($_cp_similarity) {
1393                        push @diff_tree, "-C$_cp_similarity";
1394                } else {
1395                        push @diff_tree, '-C';
1396                }
1397                push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1398                push @diff_tree, "-l$_l" if defined $_l;
1399                exec(@diff_tree, $from, $treeish) or croak $!;
1400        }
1401        return parse_diff_tree($diff_fh);
1402}
1403
1404sub svn_checkout_tree {
1405        my ($from, $treeish) = @_;
1406        my $mods = get_diff($from->{commit}, $treeish);
1407        return $mods unless (scalar @$mods);
1408        my ($rm, $add) = precommit_check($mods);
1409
1410        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1411        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1412                if ($m->{chg} eq 'C') {
1413                        svn_ensure_parent_path( $m->{file_b} );
1414                        sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
1415                        apply_mod_line_blob($m);
1416                        svn_check_prop_executable($m);
1417                } elsif ($m->{chg} eq 'D') {
1418                        sys(qw(svn rm --force), $m->{file_b});
1419                } elsif ($m->{chg} eq 'R') {
1420                        svn_ensure_parent_path( $m->{file_b} );
1421                        sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
1422                        apply_mod_line_blob($m);
1423                        svn_check_prop_executable($m);
1424                } elsif ($m->{chg} eq 'M') {
1425                        apply_mod_line_blob($m);
1426                        svn_check_prop_executable($m);
1427                } elsif ($m->{chg} eq 'T') {
1428                        sys(qw(svn rm --force),$m->{file_b});
1429                        apply_mod_line_blob($m);
1430                        sys(qw(svn add), $m->{file_b});
1431                        svn_check_prop_executable($m);
1432                } elsif ($m->{chg} eq 'A') {
1433                        svn_ensure_parent_path( $m->{file_b} );
1434                        apply_mod_line_blob($m);
1435                        sys(qw(svn add), $m->{file_b});
1436                        svn_check_prop_executable($m);
1437                } else {
1438                        croak "Invalid chg: $m->{chg}\n";
1439                }
1440        }
1441
1442        assert_tree($treeish);
1443        if ($_rmdir) { # remove empty directories
1444                handle_rmdir($rm, $add);
1445        }
1446        assert_tree($treeish);
1447        return $mods;
1448}
1449
1450sub libsvn_checkout_tree {
1451        my ($from, $treeish, $ed) = @_;
1452        my $mods = get_diff($from, $treeish);
1453        return $mods unless (scalar @$mods);
1454        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1455        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1456                my $f = $m->{chg};
1457                if (defined $o{$f}) {
1458                        $ed->$f($m, $_q);
1459                } else {
1460                        croak "Invalid change type: $f\n";
1461                }
1462        }
1463        $ed->rmdirs($_q) if $_rmdir;
1464        return $mods;
1465}
1466
1467# svn ls doesn't work with respect to the current working tree, but what's
1468# in the repository.  There's not even an option for it... *sigh*
1469# (added files don't show up and removed files remain in the ls listing)
1470sub svn_ls_current {
1471        my ($dir, $rm, $add) = @_;
1472        chomp(my @ls = safe_qx('svn','ls',$dir));
1473        my @ret = ();
1474        foreach (@ls) {
1475                s#/$##; # trailing slashes are evil
1476                push @ret, $_ unless $rm->{$dir}->{$_};
1477        }
1478        if (exists $add->{$dir}) {
1479                push @ret, keys %{$add->{$dir}};
1480        }
1481        return \@ret;
1482}
1483
1484sub handle_rmdir {
1485        my ($rm, $add) = @_;
1486
1487        foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
1488                my $ls = svn_ls_current($dir, $rm, $add);
1489                next if (scalar @$ls);
1490                sys(qw(svn rm --force),$dir);
1491
1492                my $dn = dirname $dir;
1493                $rm->{ $dn }->{ basename $dir } = 1;
1494                $ls = svn_ls_current($dn, $rm, $add);
1495                while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
1496                        sys(qw(svn rm --force),$dn);
1497                        $dir = basename $dn;
1498                        $dn = dirname $dn;
1499                        $rm->{ $dn }->{ $dir } = 1;
1500                        $ls = svn_ls_current($dn, $rm, $add);
1501                }
1502        }
1503}
1504
1505sub get_commit_message {
1506        my ($commit, $commit_msg) = (@_);
1507        my %log_msg = ( msg => '' );
1508        open my $msg, '>', $commit_msg or croak $!;
1509
1510        chomp(my $type = `git-cat-file -t $commit`);
1511        if ($type eq 'commit') {
1512                my $pid = open my $msg_fh, '-|';
1513                defined $pid or croak $!;
1514
1515                if ($pid == 0) {
1516                        exec(qw(git-cat-file commit), $commit) or croak $!;
1517                }
1518                my $in_msg = 0;
1519                while (<$msg_fh>) {
1520                        if (!$in_msg) {
1521                                $in_msg = 1 if (/^\s*$/);
1522                        } elsif (/^git-svn-id: /) {
1523                                # skip this, we regenerate the correct one
1524                                # on re-fetch anyways
1525                        } else {
1526                                print $msg $_ or croak $!;
1527                        }
1528                }
1529                close $msg_fh or croak $?;
1530        }
1531        close $msg or croak $!;
1532
1533        if ($_edit || ($type eq 'tree')) {
1534                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1535                system($editor, $commit_msg);
1536        }
1537
1538        # file_to_s removes all trailing newlines, so just use chomp() here:
1539        open $msg, '<', $commit_msg or croak $!;
1540        { local $/; chomp($log_msg{msg} = <$msg>); }
1541        close $msg or croak $!;
1542
1543        return \%log_msg;
1544}
1545
1546sub set_svn_commit_env {
1547        if (defined $LC_ALL) {
1548                $ENV{LC_ALL} = $LC_ALL;
1549        } else {
1550                delete $ENV{LC_ALL};
1551        }
1552}
1553
1554sub svn_commit_tree {
1555        my ($last, $commit) = @_;
1556        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
1557        my $log_msg = get_commit_message($commit, $commit_msg);
1558        my ($oneline) = ($log_msg->{msg} =~ /([^\n\r]+)/);
1559        print "Committing $commit: $oneline\n";
1560
1561        set_svn_commit_env();
1562        my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
1563        $ENV{LC_ALL} = 'C';
1564        unlink $commit_msg;
1565        my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/);
1566        if (!defined $committed) {
1567                my $out = join("\n",@ci_output);
1568                print STDERR "W: Trouble parsing \`svn commit' output:\n\n",
1569                                $out, "\n\nAssuming English locale...";
1570                ($committed) = ($out =~ /^Committed revision \d+\./sm);
1571                defined $committed or die " FAILED!\n",
1572                        "Commit output failed to parse committed revision!\n",
1573                print STDERR " OK\n";
1574        }
1575
1576        my @svn_up = qw(svn up);
1577        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
1578        if ($_optimize_commits && ($committed == ($last->{revision} + 1))) {
1579                push @svn_up, "-r$committed";
1580                sys(@svn_up);
1581                my $info = svn_info('.');
1582                my $date = $info->{'Last Changed Date'} or die "Missing date\n";
1583                if ($info->{'Last Changed Rev'} != $committed) {
1584                        croak "$info->{'Last Changed Rev'} != $committed\n"
1585                }
1586                my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1587                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1588                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1589                                         or croak "Failed to parse date: $date\n";
1590                $log_msg->{date} = "$tz $Y-$m-$d $H:$M:$S";
1591                $log_msg->{author} = $info->{'Last Changed Author'};
1592                $log_msg->{revision} = $committed;
1593                $log_msg->{msg} .= "\n";
1594                $log_msg->{parents} = [ $last->{commit} ];
1595                $log_msg->{commit} = git_commit($log_msg, $commit);
1596                return $log_msg;
1597        }
1598        # resync immediately
1599        push @svn_up, "-r$last->{revision}";
1600        sys(@svn_up);
1601        return fetch("$committed=$commit");
1602}
1603
1604sub rev_list_raw {
1605        my (@args) = @_;
1606        my $pid = open my $fh, '-|';
1607        defined $pid or croak $!;
1608        if (!$pid) {
1609                exec(qw/git-rev-list --pretty=raw/, @args) or croak $!;
1610        }
1611        return { fh => $fh, t => { } };
1612}
1613
1614sub next_rev_list_entry {
1615        my $rl = shift;
1616        my $fh = $rl->{fh};
1617        my $x = $rl->{t};
1618        while (<$fh>) {
1619                if (/^commit ($sha1)$/o) {
1620                        if ($x->{c}) {
1621                                $rl->{t} = { c => $1 };
1622                                return $x;
1623                        } else {
1624                                $x->{c} = $1;
1625                        }
1626                } elsif (/^parent ($sha1)$/o) {
1627                        $x->{p}->{$1} = 1;
1628                } elsif (s/^    //) {
1629                        $x->{m} ||= '';
1630                        $x->{m} .= $_;
1631                }
1632        }
1633        return ($x != $rl->{t}) ? $x : undef;
1634}
1635
1636# read the entire log into a temporary file (which is removed ASAP)
1637# and store the file handle + parser state
1638sub svn_log_raw {
1639        my (@log_args) = @_;
1640        my $log_fh = IO::File->new_tmpfile or croak $!;
1641        my $pid = fork;
1642        defined $pid or croak $!;
1643        if (!$pid) {
1644                open STDOUT, '>&', $log_fh or croak $!;
1645                exec (qw(svn log), @log_args) or croak $!
1646        }
1647        waitpid $pid, 0;
1648        croak $? if $?;
1649        seek $log_fh, 0, 0 or croak $!;
1650        return { state => 'sep', fh => $log_fh };
1651}
1652
1653sub next_log_entry {
1654        my $log = shift; # retval of svn_log_raw()
1655        my $ret = undef;
1656        my $fh = $log->{fh};
1657
1658        while (<$fh>) {
1659                chomp;
1660                if (/^\-{72}$/) {
1661                        if ($log->{state} eq 'msg') {
1662                                if ($ret->{lines}) {
1663                                        $ret->{msg} .= $_."\n";
1664                                        unless(--$ret->{lines}) {
1665                                                $log->{state} = 'sep';
1666                                        }
1667                                } else {
1668                                        croak "Log parse error at: $_\n",
1669                                                $ret->{revision},
1670                                                "\n";
1671                                }
1672                                next;
1673                        }
1674                        if ($log->{state} ne 'sep') {
1675                                croak "Log parse error at: $_\n",
1676                                        "state: $log->{state}\n",
1677                                        $ret->{revision},
1678                                        "\n";
1679                        }
1680                        $log->{state} = 'rev';
1681
1682                        # if we have an empty log message, put something there:
1683                        if ($ret) {
1684                                $ret->{msg} ||= "\n";
1685                                delete $ret->{lines};
1686                                return $ret;
1687                        }
1688                        next;
1689                }
1690                if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
1691                        my $rev = $1;
1692                        my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
1693                        ($lines) = ($lines =~ /(\d+)/);
1694                        my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1695                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1696                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1697                                         or croak "Failed to parse date: $date\n";
1698                        $ret = {        revision => $rev,
1699                                        date => "$tz $Y-$m-$d $H:$M:$S",
1700                                        author => $author,
1701                                        lines => $lines,
1702                                        msg => '' };
1703                        if (defined $_authors && ! defined $users{$author}) {
1704                                die "Author: $author not defined in ",
1705                                                "$_authors file\n";
1706                        }
1707                        $log->{state} = 'msg_start';
1708                        next;
1709                }
1710                # skip the first blank line of the message:
1711                if ($log->{state} eq 'msg_start' && /^$/) {
1712                        $log->{state} = 'msg';
1713                } elsif ($log->{state} eq 'msg') {
1714                        if ($ret->{lines}) {
1715                                $ret->{msg} .= $_."\n";
1716                                unless (--$ret->{lines}) {
1717                                        $log->{state} = 'sep';
1718                                }
1719                        } else {
1720                                croak "Log parse error at: $_\n",
1721                                        $ret->{revision},"\n";
1722                        }
1723                }
1724        }
1725        return $ret;
1726}
1727
1728sub svn_info {
1729        my $url = shift || $SVN_URL;
1730
1731        my $pid = open my $info_fh, '-|';
1732        defined $pid or croak $!;
1733
1734        if ($pid == 0) {
1735                exec(qw(svn info),$url) or croak $!;
1736        }
1737
1738        my $ret = {};
1739        # only single-lines seem to exist in svn info output
1740        while (<$info_fh>) {
1741                chomp $_;
1742                if (m#^([^:]+)\s*:\s*(\S.*)$#) {
1743                        $ret->{$1} = $2;
1744                        push @{$ret->{-order}}, $1;
1745                }
1746        }
1747        close $info_fh or croak $?;
1748        return $ret;
1749}
1750
1751sub sys { system(@_) == 0 or croak $? }
1752
1753sub eol_cp {
1754        my ($from, $to) = @_;
1755        my $es = svn_propget_base('svn:eol-style', $to);
1756        open my $rfd, '<', $from or croak $!;
1757        binmode $rfd or croak $!;
1758        open my $wfd, '>', $to or croak $!;
1759        binmode $wfd or croak $!;
1760        eol_cp_fd($rfd, $wfd, $es);
1761        close $rfd or croak $!;
1762        close $wfd or croak $!;
1763}
1764
1765sub eol_cp_fd {
1766        my ($rfd, $wfd, $es) = @_;
1767        my $eol = defined $es ? $EOL{$es} : undef;
1768        my $buf;
1769        use bytes;
1770        while (1) {
1771                my ($r, $w, $t);
1772                defined($r = sysread($rfd, $buf, 4096)) or croak $!;
1773                return unless $r;
1774                if ($eol) {
1775                        if ($buf =~ /\015$/) {
1776                                my $c;
1777                                defined($r = sysread($rfd,$c,1)) or croak $!;
1778                                $buf .= $c if $r > 0;
1779                        }
1780                        $buf =~ s/(?:\015\012|\015|\012)/$eol/gs;
1781                        $r = length($buf);
1782                }
1783                for ($w = 0; $w < $r; $w += $t) {
1784                        $t = syswrite($wfd, $buf, $r - $w, $w) or croak $!;
1785                }
1786        }
1787        no bytes;
1788}
1789
1790sub do_update_index {
1791        my ($z_cmd, $cmd, $no_text_base) = @_;
1792
1793        my $z = open my $p, '-|';
1794        defined $z or croak $!;
1795        unless ($z) { exec @$z_cmd or croak $! }
1796
1797        my $pid = open my $ui, '|-';
1798        defined $pid or croak $!;
1799        unless ($pid) {
1800                exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!;
1801        }
1802        local $/ = "\0";
1803        while (my $x = <$p>) {
1804                chomp $x;
1805                if (!$no_text_base && lstat $x && ! -l _ &&
1806                                svn_propget_base('svn:keywords', $x)) {
1807                        my $mode = -x _ ? 0755 : 0644;
1808                        my ($v,$d,$f) = File::Spec->splitpath($x);
1809                        my $tb = File::Spec->catfile($d, '.svn', 'tmp',
1810                                                'text-base',"$f.svn-base");
1811                        $tb =~ s#^/##;
1812                        unless (-f $tb) {
1813                                $tb = File::Spec->catfile($d, '.svn',
1814                                                'text-base',"$f.svn-base");
1815                                $tb =~ s#^/##;
1816                        }
1817                        unlink $x or croak $!;
1818                        eol_cp($tb, $x);
1819                        chmod(($mode &~ umask), $x) or croak $!;
1820                }
1821                print $ui $x,"\0";
1822        }
1823        close $ui or croak $?;
1824}
1825
1826sub index_changes {
1827        return if $_use_lib;
1828
1829        if (!-f "$GIT_SVN_DIR/info/exclude") {
1830                open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!;
1831                print $fd '.svn',"\n";
1832                close $fd or croak $!;
1833        }
1834        my $no_text_base = shift;
1835        do_update_index([qw/git-diff-files --name-only -z/],
1836                        'remove',
1837                        $no_text_base);
1838        do_update_index([qw/git-ls-files -z --others/,
1839                                "--exclude-from=$GIT_SVN_DIR/info/exclude"],
1840                        'add',
1841                        $no_text_base);
1842}
1843
1844sub s_to_file {
1845        my ($str, $file, $mode) = @_;
1846        open my $fd,'>',$file or croak $!;
1847        print $fd $str,"\n" or croak $!;
1848        close $fd or croak $!;
1849        chmod ($mode &~ umask, $file) if (defined $mode);
1850}
1851
1852sub file_to_s {
1853        my $file = shift;
1854        open my $fd,'<',$file or croak "$!: file: $file\n";
1855        local $/;
1856        my $ret = <$fd>;
1857        close $fd or croak $!;
1858        $ret =~ s/\s*$//s;
1859        return $ret;
1860}
1861
1862sub assert_revision_unknown {
1863        my $r = shift;
1864        if (my $c = revdb_get($REVDB, $r)) {
1865                croak "$r = $c already exists! Why are we refetching it?";
1866        }
1867}
1868
1869sub trees_eq {
1870        my ($x, $y) = @_;
1871        my @x = safe_qx('git-cat-file','commit',$x);
1872        my @y = safe_qx('git-cat-file','commit',$y);
1873        if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
1874                                || $y[0] !~ /^tree $sha1\n$/) {
1875                print STDERR "Trees not equal: $y[0] != $x[0]\n";
1876                return 0
1877        }
1878        return 1;
1879}
1880
1881sub git_commit {
1882        my ($log_msg, @parents) = @_;
1883        assert_revision_unknown($log_msg->{revision});
1884        map_tree_joins() if (@_branch_from && !%tree_map);
1885
1886        my (@tmp_parents, @exec_parents, %seen_parent);
1887        if (my $lparents = $log_msg->{parents}) {
1888                @tmp_parents = @$lparents
1889        }
1890        # commit parents can be conditionally bound to a particular
1891        # svn revision via: "svn_revno=commit_sha1", filter them out here:
1892        foreach my $p (@parents) {
1893                next unless defined $p;
1894                if ($p =~ /^(\d+)=($sha1_short)$/o) {
1895                        if ($1 == $log_msg->{revision}) {
1896                                push @tmp_parents, $2;
1897                        }
1898                } else {
1899                        push @tmp_parents, $p if $p =~ /$sha1_short/o;
1900                }
1901        }
1902        my $tree = $log_msg->{tree};
1903        if (!defined $tree) {
1904                my $index = set_index($GIT_SVN_INDEX);
1905                index_changes();
1906                chomp($tree = `git-write-tree`);
1907                croak $? if $?;
1908                restore_index($index);
1909        }
1910
1911        # just in case we clobber the existing ref, we still want that ref
1912        # as our parent:
1913        if (my $cur = eval { file_to_s("$GIT_DIR/refs/remotes/$GIT_SVN") }) {
1914                push @tmp_parents, $cur;
1915        }
1916
1917        if (exists $tree_map{$tree}) {
1918                foreach my $p (@{$tree_map{$tree}}) {
1919                        my $skip;
1920                        foreach (@tmp_parents) {
1921                                # see if a common parent is found
1922                                my $mb = eval {
1923                                        safe_qx('git-merge-base', $_, $p)
1924                                };
1925                                next if ($@ || $?);
1926                                $skip = 1;
1927                                last;
1928                        }
1929                        next if $skip;
1930                        my ($url_p, $r_p, $uuid_p) = cmt_metadata($p);
1931                        next if (($SVN_UUID eq $uuid_p) &&
1932                                                ($log_msg->{revision} > $r_p));
1933                        next if (defined $url_p && defined $SVN_URL &&
1934                                                ($SVN_UUID eq $uuid_p) &&
1935                                                ($url_p eq $SVN_URL));
1936                        push @tmp_parents, $p;
1937                }
1938        }
1939        foreach (@tmp_parents) {
1940                next if $seen_parent{$_};
1941                $seen_parent{$_} = 1;
1942                push @exec_parents, $_;
1943                # MAXPARENT is defined to 16 in commit-tree.c:
1944                last if @exec_parents > 16;
1945        }
1946
1947        set_commit_env($log_msg);
1948        my @exec = ('git-commit-tree', $tree);
1949        push @exec, '-p', $_  foreach @exec_parents;
1950        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1951                                                                or croak $!;
1952        print $msg_fh $log_msg->{msg} or croak $!;
1953        unless ($_no_metadata) {
1954                print $msg_fh "\ngit-svn-id: $SVN_URL\@$log_msg->{revision}",
1955                                        " $SVN_UUID\n" or croak $!;
1956        }
1957        $msg_fh->flush == 0 or croak $!;
1958        close $msg_fh or croak $!;
1959        chomp(my $commit = do { local $/; <$out_fh> });
1960        close $out_fh or croak $!;
1961        waitpid $pid, 0;
1962        croak $? if $?;
1963        if ($commit !~ /^$sha1$/o) {
1964                die "Failed to commit, invalid sha1: $commit\n";
1965        }
1966        sys('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
1967        revdb_set($REVDB, $log_msg->{revision}, $commit);
1968
1969        # this output is read via pipe, do not change:
1970        print "r$log_msg->{revision} = $commit\n";
1971        check_repack();
1972        return $commit;
1973}
1974
1975sub check_repack {
1976        if ($_repack && (--$_repack_nr == 0)) {
1977                $_repack_nr = $_repack;
1978                sys("git repack $_repack_flags");
1979        }
1980}
1981
1982sub set_commit_env {
1983        my ($log_msg) = @_;
1984        my $author = $log_msg->{author};
1985        if (!defined $author || length $author == 0) {
1986                $author = '(no author)';
1987        }
1988        my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
1989                                : ($author,"$author\@$SVN_UUID");
1990        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1991        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1992        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
1993}
1994
1995sub apply_mod_line_blob {
1996        my $m = shift;
1997        if ($m->{mode_b} =~ /^120/) {
1998                blob_to_symlink($m->{sha1_b}, $m->{file_b});
1999        } else {
2000                blob_to_file($m->{sha1_b}, $m->{file_b});
2001        }
2002}
2003
2004sub blob_to_symlink {
2005        my ($blob, $link) = @_;
2006        defined $link or croak "\$link not defined!\n";
2007        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2008        if (-l $link || -f _) {
2009                unlink $link or croak $!;
2010        }
2011
2012        my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
2013        symlink $dest, $link or croak $!;
2014}
2015
2016sub blob_to_file {
2017        my ($blob, $file) = @_;
2018        defined $file or croak "\$file not defined!\n";
2019        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2020        if (-l $file || -f _) {
2021                unlink $file or croak $!;
2022        }
2023
2024        open my $blob_fh, '>', $file or croak "$!: $file\n";
2025        my $pid = fork;
2026        defined $pid or croak $!;
2027
2028        if ($pid == 0) {
2029                open STDOUT, '>&', $blob_fh or croak $!;
2030                exec('git-cat-file','blob',$blob) or croak $!;
2031        }
2032        waitpid $pid, 0;
2033        croak $? if $?;
2034
2035        close $blob_fh or croak $!;
2036}
2037
2038sub safe_qx {
2039        my $pid = open my $child, '-|';
2040        defined $pid or croak $!;
2041        if ($pid == 0) {
2042                exec(@_) or croak $!;
2043        }
2044        my @ret = (<$child>);
2045        close $child or croak $?;
2046        die $? if $?; # just in case close didn't error out
2047        return wantarray ? @ret : join('',@ret);
2048}
2049
2050sub svn_compat_check {
2051        if ($_follow_parent) {
2052                print STDERR 'E: --follow-parent functionality is only ',
2053                                "available when SVN libraries are used\n";
2054                exit 1;
2055        }
2056        my @co_help = safe_qx(qw(svn co -h));
2057        unless (grep /ignore-externals/,@co_help) {
2058                print STDERR "W: Installed svn version does not support ",
2059                                "--ignore-externals\n";
2060                $_no_ignore_ext = 1;
2061        }
2062        if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
2063                $_svn_co_url_revs = 1;
2064        }
2065        if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) {
2066                $_svn_pg_peg_revs = 1;
2067        }
2068
2069        # I really, really hope nobody hits this...
2070        unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
2071                print STDERR <<'';
2072W: The installed svn version does not support the --stop-on-copy flag in
2073   the log command.
2074   Lets hope the directory you're tracking is not a branch or tag
2075   and was never moved within the repository...
2076
2077                $_no_stop_copy = 1;
2078        }
2079}
2080
2081# *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
2082# (and they won't honor URL@<rev> without -r<rev>, too!)
2083sub svn_cmd_checkout {
2084        my ($url, $rev, $dir) = @_;
2085        my @cmd = ('svn','co', "-r$rev");
2086        push @cmd, '--ignore-externals' unless $_no_ignore_ext;
2087        $url .= "\@$rev" if $_svn_co_url_revs;
2088        sys(@cmd, $url, $dir);
2089}
2090
2091sub check_upgrade_needed {
2092        if (!-r $REVDB) {
2093                -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2094                open my $fh, '>>',$REVDB or croak $!;
2095                close $fh;
2096        }
2097        my $old = eval {
2098                my $pid = open my $child, '-|';
2099                defined $pid or croak $!;
2100                if ($pid == 0) {
2101                        close STDERR;
2102                        exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $!;
2103                }
2104                my @ret = (<$child>);
2105                close $child or croak $?;
2106                die $? if $?; # just in case close didn't error out
2107                return wantarray ? @ret : join('',@ret);
2108        };
2109        return unless $old;
2110        my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
2111        if ($@ || !$head) {
2112                print STDERR "Please run: $0 rebuild --upgrade\n";
2113                exit 1;
2114        }
2115}
2116
2117# fills %tree_map with a reverse mapping of trees to commits.  Useful
2118# for finding parents to commit on.
2119sub map_tree_joins {
2120        my %seen;
2121        foreach my $br (@_branch_from) {
2122                my $pid = open my $pipe, '-|';
2123                defined $pid or croak $!;
2124                if ($pid == 0) {
2125                        exec(qw(git-rev-list --topo-order --pretty=raw), $br)
2126                                                                or croak $!;
2127                }
2128                while (<$pipe>) {
2129                        if (/^commit ($sha1)$/o) {
2130                                my $commit = $1;
2131
2132                                # if we've seen a commit,
2133                                # we've seen its parents
2134                                last if $seen{$commit};
2135                                my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
2136                                unless (defined $tree) {
2137                                        die "Failed to parse commit $commit\n";
2138                                }
2139                                push @{$tree_map{$tree}}, $commit;
2140                                $seen{$commit} = 1;
2141                        }
2142                }
2143                close $pipe; # we could be breaking the pipe early
2144        }
2145}
2146
2147sub load_all_refs {
2148        if (@_branch_from) {
2149                print STDERR '--branch|-b parameters are ignored when ',
2150                        "--branch-all-refs|-B is passed\n";
2151        }
2152
2153        # don't worry about rev-list on non-commit objects/tags,
2154        # it shouldn't blow up if a ref is a blob or tree...
2155        chomp(@_branch_from = `git-rev-parse --symbolic --all`);
2156}
2157
2158# '<svn username> = real-name <email address>' mapping based on git-svnimport:
2159sub load_authors {
2160        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2161        while (<$authors>) {
2162                chomp;
2163                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2164                my ($user, $name, $email) = ($1, $2, $3);
2165                $users{$user} = [$name, $email];
2166        }
2167        close $authors or croak $!;
2168}
2169
2170sub rload_authors {
2171        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2172        while (<$authors>) {
2173                chomp;
2174                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2175                my ($user, $name, $email) = ($1, $2, $3);
2176                $rusers{"$name <$email>"} = $user;
2177        }
2178        close $authors or croak $!;
2179}
2180
2181sub svn_propget_base {
2182        my ($p, $f) = @_;
2183        $f .= '@BASE' if $_svn_pg_peg_revs;
2184        return safe_qx(qw/svn propget/, $p, $f);
2185}
2186
2187sub git_svn_each {
2188        my $sub = shift;
2189        foreach (`git-rev-parse --symbolic --all`) {
2190                next unless s#^refs/remotes/##;
2191                chomp $_;
2192                next unless -f "$GIT_DIR/svn/$_/info/url";
2193                &$sub($_);
2194        }
2195}
2196
2197sub migrate_revdb {
2198        git_svn_each(sub {
2199                my $id = shift;
2200                defined(my $pid = fork) or croak $!;
2201                if (!$pid) {
2202                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2203                        init_vars();
2204                        exit 0 if -r $REVDB;
2205                        print "Upgrading svn => git mapping...\n";
2206                        -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2207                        open my $fh, '>>',$REVDB or croak $!;
2208                        close $fh;
2209                        rebuild();
2210                        print "Done upgrading. You may now delete the ",
2211                                "deprecated $GIT_SVN_DIR/revs directory\n";
2212                        exit 0;
2213                }
2214                waitpid $pid, 0;
2215                croak $? if $?;
2216        });
2217}
2218
2219sub migration_check {
2220        migrate_revdb() unless (-e $REVDB);
2221        return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
2222        print "Upgrading repository...\n";
2223        unless (-d "$GIT_DIR/svn") {
2224                mkdir "$GIT_DIR/svn" or croak $!;
2225        }
2226        print "Data from a previous version of git-svn exists, but\n\t",
2227                                "$GIT_SVN_DIR\n\t(required for this version ",
2228                                "($VERSION) of git-svn) does not.\n";
2229
2230        foreach my $x (`git-rev-parse --symbolic --all`) {
2231                next unless $x =~ s#^refs/remotes/##;
2232                chomp $x;
2233                next unless -f "$GIT_DIR/$x/info/url";
2234                my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
2235                next unless $u;
2236                my $dn = dirname("$GIT_DIR/svn/$x");
2237                mkpath([$dn]) unless -d $dn;
2238                rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
2239        }
2240        migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
2241        print "Done upgrading.\n";
2242}
2243
2244sub find_rev_before {
2245        my ($r, $id, $eq_ok) = @_;
2246        my $f = "$GIT_DIR/svn/$id/.rev_db";
2247        return (undef,undef) unless -r $f;
2248        --$r unless $eq_ok;
2249        while ($r > 0) {
2250                if (my $c = revdb_get($f, $r)) {
2251                        return ($r, $c);
2252                }
2253                --$r;
2254        }
2255        return (undef, undef);
2256}
2257
2258sub init_vars {
2259        $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
2260        $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
2261        $REVDB = "$GIT_SVN_DIR/.rev_db";
2262        $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
2263        $SVN_URL = undef;
2264        $SVN_WC = "$GIT_SVN_DIR/tree";
2265        %tree_map = ();
2266}
2267
2268# convert GetOpt::Long specs for use by git-repo-config
2269sub read_repo_config {
2270        return unless -d $GIT_DIR;
2271        my $opts = shift;
2272        foreach my $o (keys %$opts) {
2273                my $v = $opts->{$o};
2274                my ($key) = ($o =~ /^([a-z\-]+)/);
2275                $key =~ s/-//g;
2276                my $arg = 'git-repo-config';
2277                $arg .= ' --int' if ($o =~ /[:=]i$/);
2278                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
2279                if (ref $v eq 'ARRAY') {
2280                        chomp(my @tmp = `$arg --get-all svn.$key`);
2281                        @$v = @tmp if @tmp;
2282                } else {
2283                        chomp(my $tmp = `$arg --get svn.$key`);
2284                        if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
2285                                $$v = $tmp;
2286                        }
2287                }
2288        }
2289}
2290
2291sub set_default_vals {
2292        if (defined $_repack) {
2293                $_repack = 1000 if ($_repack <= 0);
2294                $_repack_nr = $_repack;
2295                $_repack_flags ||= '-d';
2296        }
2297}
2298
2299sub read_grafts {
2300        my $gr_file = shift;
2301        my ($grafts, $comments) = ({}, {});
2302        if (open my $fh, '<', $gr_file) {
2303                my @tmp;
2304                while (<$fh>) {
2305                        if (/^($sha1)\s+/) {
2306                                my $c = $1;
2307                                if (@tmp) {
2308                                        @{$comments->{$c}} = @tmp;
2309                                        @tmp = ();
2310                                }
2311                                foreach my $p (split /\s+/, $_) {
2312                                        $grafts->{$c}->{$p} = 1;
2313                                }
2314                        } else {
2315                                push @tmp, $_;
2316                        }
2317                }
2318                close $fh or croak $!;
2319                @{$comments->{'END'}} = @tmp if @tmp;
2320        }
2321        return ($grafts, $comments);
2322}
2323
2324sub write_grafts {
2325        my ($grafts, $comments, $gr_file) = @_;
2326
2327        open my $fh, '>', $gr_file or croak $!;
2328        foreach my $c (sort keys %$grafts) {
2329                if ($comments->{$c}) {
2330                        print $fh $_ foreach @{$comments->{$c}};
2331                }
2332                my $p = $grafts->{$c};
2333                my %x; # real parents
2334                delete $p->{$c}; # commits are not self-reproducing...
2335                my $pid = open my $ch, '-|';
2336                defined $pid or croak $!;
2337                if (!$pid) {
2338                        exec(qw/git-cat-file commit/, $c) or croak $!;
2339                }
2340                while (<$ch>) {
2341                        if (/^parent ($sha1)/) {
2342                                $x{$1} = $p->{$1} = 1;
2343                        } else {
2344                                last unless /^\S/;
2345                        }
2346                }
2347                close $ch; # breaking the pipe
2348
2349                # if real parents are the only ones in the grafts, drop it
2350                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2351
2352                my (@ip, @jp, $mb);
2353                my %del = %x;
2354                @ip = @jp = keys %$p;
2355                foreach my $i (@ip) {
2356                        next if $del{$i} || $p->{$i} == 2;
2357                        foreach my $j (@jp) {
2358                                next if $i eq $j || $del{$j} || $p->{$j} == 2;
2359                                $mb = eval { safe_qx('git-merge-base',$i,$j) };
2360                                next unless $mb;
2361                                chomp $mb;
2362                                next if $x{$mb};
2363                                if ($mb eq $j) {
2364                                        delete $p->{$i};
2365                                        $del{$i} = 1;
2366                                } elsif ($mb eq $i) {
2367                                        delete $p->{$j};
2368                                        $del{$j} = 1;
2369                                }
2370                        }
2371                }
2372
2373                # if real parents are the only ones in the grafts, drop it
2374                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2375
2376                print $fh $c, ' ', join(' ', sort keys %$p),"\n";
2377        }
2378        if ($comments->{'END'}) {
2379                print $fh $_ foreach @{$comments->{'END'}};
2380        }
2381        close $fh or croak $!;
2382}
2383
2384sub read_url_paths_all {
2385        my ($l_map, $pfx, $p) = @_;
2386        my @dir;
2387        foreach (<$p/*>) {
2388                if (-r "$_/info/url") {
2389                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2390                        my $id = $pfx . basename $_;
2391                        my $url = file_to_s("$_/info/url");
2392                        my ($u, $p) = repo_path_split($url);
2393                        $l_map->{$u}->{$p} = $id;
2394                } elsif (-d $_) {
2395                        push @dir, $_;
2396                }
2397        }
2398        foreach (@dir) {
2399                my $x = $_;
2400                $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
2401                read_url_paths_all($l_map, $x, $_);
2402        }
2403}
2404
2405# this one only gets ids that have been imported, not new ones
2406sub read_url_paths {
2407        my $l_map = {};
2408        git_svn_each(sub { my $x = shift;
2409                        my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
2410                        my ($u, $p) = repo_path_split($url);
2411                        $l_map->{$u}->{$p} = $x;
2412                        });
2413        return $l_map;
2414}
2415
2416sub extract_metadata {
2417        my $id = shift or return (undef, undef, undef);
2418        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
2419                                                        \s([a-f\d\-]+)$/x);
2420        if (!$rev || !$uuid || !$url) {
2421                # some of the original repositories I made had
2422                # indentifiers like this:
2423                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
2424        }
2425        return ($url, $rev, $uuid);
2426}
2427
2428sub cmt_metadata {
2429        return extract_metadata((grep(/^git-svn-id: /,
2430                safe_qx(qw/git-cat-file commit/, shift)))[-1]);
2431}
2432
2433sub get_commit_time {
2434        my $cmt = shift;
2435        defined(my $pid = open my $fh, '-|') or croak $!;
2436        if (!$pid) {
2437                exec qw/git-rev-list --pretty=raw -n1/, $cmt or croak $!;
2438        }
2439        while (<$fh>) {
2440                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
2441                my ($s, $tz) = ($1, $2);
2442                if ($tz =~ s/^\+//) {
2443                        $s += tz_to_s_offset($tz);
2444                } elsif ($tz =~ s/^\-//) {
2445                        $s -= tz_to_s_offset($tz);
2446                }
2447                close $fh;
2448                return $s;
2449        }
2450        die "Can't get commit time for commit: $cmt\n";
2451}
2452
2453sub tz_to_s_offset {
2454        my ($tz) = @_;
2455        $tz =~ s/(\d\d)$//;
2456        return ($1 * 60) + ($tz * 3600);
2457}
2458
2459sub setup_pager { # translated to Perl from pager.c
2460        return unless (-t *STDOUT);
2461        my $pager = $ENV{PAGER};
2462        if (!defined $pager) {
2463                $pager = 'less';
2464        } elsif (length $pager == 0 || $pager eq 'cat') {
2465                return;
2466        }
2467        pipe my $rfd, my $wfd or return;
2468        defined(my $pid = fork) or croak $!;
2469        if (!$pid) {
2470                open STDOUT, '>&', $wfd or croak $!;
2471                return;
2472        }
2473        open STDIN, '<&', $rfd or croak $!;
2474        $ENV{LESS} ||= '-S';
2475        exec $pager or croak "Can't run pager: $!\n";;
2476}
2477
2478sub get_author_info {
2479        my ($dest, $author, $t, $tz) = @_;
2480        $author =~ s/(?:^\s*|\s*$)//g;
2481        $dest->{a_raw} = $author;
2482        my $_a;
2483        if ($_authors) {
2484                $_a = $rusers{$author} || undef;
2485        }
2486        if (!$_a) {
2487                ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
2488        }
2489        $dest->{t} = $t;
2490        $dest->{tz} = $tz;
2491        $dest->{a} = $_a;
2492        # Date::Parse isn't in the standard Perl distro :(
2493        if ($tz =~ s/^\+//) {
2494                $t += tz_to_s_offset($tz);
2495        } elsif ($tz =~ s/^\-//) {
2496                $t -= tz_to_s_offset($tz);
2497        }
2498        $dest->{t_utc} = $t;
2499}
2500
2501sub process_commit {
2502        my ($c, $r_min, $r_max, $defer) = @_;
2503        if (defined $r_min && defined $r_max) {
2504                if ($r_min == $c->{r} && $r_min == $r_max) {
2505                        show_commit($c);
2506                        return 0;
2507                }
2508                return 1 if $r_min == $r_max;
2509                if ($r_min < $r_max) {
2510                        # we need to reverse the print order
2511                        return 0 if (defined $_limit && --$_limit < 0);
2512                        push @$defer, $c;
2513                        return 1;
2514                }
2515                if ($r_min != $r_max) {
2516                        return 1 if ($r_min < $c->{r});
2517                        return 1 if ($r_max > $c->{r});
2518                }
2519        }
2520        return 0 if (defined $_limit && --$_limit < 0);
2521        show_commit($c);
2522        return 1;
2523}
2524
2525sub show_commit {
2526        my $c = shift;
2527        if ($_oneline) {
2528                my $x = "\n";
2529                if (my $l = $c->{l}) {
2530                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2531                        $x = $l->[0];
2532                }
2533                $_l_fmt ||= 'A' . length($c->{r});
2534                print 'r',pack($_l_fmt, $c->{r}),' | ';
2535                print "$c->{c} | " if $_show_commit;
2536                print $x;
2537        } else {
2538                show_commit_normal($c);
2539        }
2540}
2541
2542sub show_commit_normal {
2543        my ($c) = @_;
2544        print '-' x72, "\nr$c->{r} | ";
2545        print "$c->{c} | " if $_show_commit;
2546        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2547                                 localtime($c->{t_utc})), ' | ';
2548        my $nr_line = 0;
2549
2550        if (my $l = $c->{l}) {
2551                while ($l->[$#$l] eq "\n" && $l->[($#$l - 1)] eq "\n") {
2552                        pop @$l;
2553                }
2554                $nr_line = scalar @$l;
2555                if (!$nr_line) {
2556                        print "1 line\n\n\n";
2557                } else {
2558                        if ($nr_line == 1) {
2559                                $nr_line = '1 line';
2560                        } else {
2561                                $nr_line .= ' lines';
2562                        }
2563                        print $nr_line, "\n\n";
2564                        print $_ foreach @$l;
2565                }
2566        } else {
2567                print "1 line\n\n";
2568
2569        }
2570        foreach my $x (qw/raw diff/) {
2571                if ($c->{$x}) {
2572                        print "\n";
2573                        print $_ foreach @{$c->{$x}}
2574                }
2575        }
2576}
2577
2578sub libsvn_load {
2579        return unless $_use_lib;
2580        $_use_lib = eval {
2581                require SVN::Core;
2582                if ($SVN::Core::VERSION lt '1.1.0') {
2583                        die "Need SVN::Core 1.1.0 or better ",
2584                                        "(got $SVN::Core::VERSION) ",
2585                                        "Falling back to command-line svn\n";
2586                }
2587                require SVN::Ra;
2588                require SVN::Delta;
2589                push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
2590                my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2591                                        $SVN::Node::dir.$SVN::Node::unknown.
2592                                        $SVN::Node::none.$SVN::Node::file.
2593                                        $SVN::Node::dir.$SVN::Node::unknown;
2594                1;
2595        };
2596}
2597
2598sub libsvn_connect {
2599        my ($url) = @_;
2600        my $auth = SVN::Core::auth_open([SVN::Client::get_simple_provider(),
2601                          SVN::Client::get_ssl_server_trust_file_provider(),
2602                          SVN::Client::get_username_provider()]);
2603        my $s = eval { SVN::Ra->new(url => $url, auth => $auth) };
2604        return $s;
2605}
2606
2607sub libsvn_get_file {
2608        my ($gui, $f, $rev) = @_;
2609        my $p = $f;
2610        return unless ($p =~ s#^\Q$SVN_PATH\E/?##);
2611
2612        my ($hash, $pid, $in, $out);
2613        my $pool = SVN::Pool->new;
2614        defined($pid = open3($in, $out, '>&STDERR',
2615                                qw/git-hash-object -w --stdin/)) or croak $!;
2616        # redirect STDOUT for SVN 1.1.x compatibility
2617        open my $stdout, '>&', \*STDOUT or croak $!;
2618        open STDOUT, '>&', $in or croak $!;
2619        my ($r, $props) = $SVN->get_file($f, $rev, \*STDOUT, $pool);
2620        $in->flush == 0 or croak $!;
2621        open STDOUT, '>&', $stdout or croak $!;
2622        close $in or croak $!;
2623        close $stdout or croak $!;
2624        $pool->clear;
2625        chomp($hash = do { local $/; <$out> });
2626        close $out or croak $!;
2627        waitpid $pid, 0;
2628        $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2629
2630        my $mode = exists $props->{'svn:executable'} ? '100755' : '100644';
2631        if (exists $props->{'svn:special'}) {
2632                $mode = '120000';
2633                my $link = `git-cat-file blob $hash`;
2634                $link =~ s/^link // or die "svn:special file with contents: <",
2635                                                $link, "> is not understood\n";
2636                defined($pid = open3($in, $out, '>&STDERR',
2637                                qw/git-hash-object -w --stdin/)) or croak $!;
2638                print $in $link;
2639                $in->flush == 0 or croak $!;
2640                close $in or croak $!;
2641                chomp($hash = do { local $/; <$out> });
2642                close $out or croak $!;
2643                waitpid $pid, 0;
2644                $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2645        }
2646        print $gui $mode,' ',$hash,"\t",$p,"\0" or croak $!;
2647}
2648
2649sub libsvn_log_entry {
2650        my ($rev, $author, $date, $msg, $parents) = @_;
2651        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2652                                         (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
2653                                or die "Unable to parse date: $date\n";
2654        if (defined $_authors && ! defined $users{$author}) {
2655                die "Author: $author not defined in $_authors file\n";
2656        }
2657        return { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
2658                author => $author, msg => $msg."\n", parents => $parents || [] }
2659}
2660
2661sub process_rm {
2662        my ($gui, $last_commit, $f) = @_;
2663        $f =~ s#^\Q$SVN_PATH\E/?## or return;
2664        # remove entire directories.
2665        if (safe_qx('git-ls-tree',$last_commit,'--',$f) =~ /^040000 tree/) {
2666                defined(my $pid = open my $ls, '-|') or croak $!;
2667                if (!$pid) {
2668                        exec(qw/git-ls-tree -r --name-only -z/,
2669                                $last_commit,'--',$f) or croak $!;
2670                }
2671                local $/ = "\0";
2672                while (<$ls>) {
2673                        print $gui '0 ',0 x 40,"\t",$_ or croak $!;
2674                }
2675                close $ls or croak $?;
2676        } else {
2677                print $gui '0 ',0 x 40,"\t",$f,"\0" or croak $!;
2678        }
2679}
2680
2681sub libsvn_fetch {
2682        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
2683        open my $gui, '| git-update-index -z --index-info' or croak $!;
2684        my @amr;
2685        foreach my $f (keys %$paths) {
2686                my $m = $paths->{$f}->action();
2687                $f =~ s#^/+##;
2688                if ($m =~ /^[DR]$/) {
2689                        print "\t$m\t$f\n" unless $_q;
2690                        process_rm($gui, $last_commit, $f);
2691                        next if $m eq 'D';
2692                        # 'R' can be file replacements, too, right?
2693                }
2694                my $pool = SVN::Pool->new;
2695                my $t = $SVN->check_path($f, $rev, $pool);
2696                if ($t == $SVN::Node::file) {
2697                        if ($m =~ /^[AMR]$/) {
2698                                push @amr, [ $m, $f ];
2699                        } else {
2700                                die "Unrecognized action: $m, ($f r$rev)\n";
2701                        }
2702                }
2703                $pool->clear;
2704        }
2705        foreach (@amr) {
2706                print "\t$_->[0]\t$_->[1]\n" unless $_q;
2707                libsvn_get_file($gui, $_->[1], $rev)
2708        }
2709        close $gui or croak $?;
2710        return libsvn_log_entry($rev, $author, $date, $msg, [$last_commit]);
2711}
2712
2713sub svn_grab_base_rev {
2714        defined(my $pid = open my $fh, '-|') or croak $!;
2715        if (!$pid) {
2716                open my $null, '>', '/dev/null' or croak $!;
2717                open STDERR, '>&', $null or croak $!;
2718                exec qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0"
2719                                                                or croak $!;
2720        }
2721        chomp(my $c = do { local $/; <$fh> });
2722        close $fh;
2723        if (defined $c && length $c) {
2724                my ($url, $rev, $uuid) = cmt_metadata($c);
2725                return ($rev, $c) if defined $rev;
2726        }
2727        if ($_no_metadata) {
2728                my $offset = -41; # from tail
2729                my $rl;
2730                open my $fh, '<', $REVDB or
2731                        die "--no-metadata specified and $REVDB not readable\n";
2732                seek $fh, $offset, 2;
2733                $rl = readline $fh;
2734                defined $rl or return (undef, undef);
2735                chomp $rl;
2736                while ($c ne $rl && tell $fh != 0) {
2737                        $offset -= 41;
2738                        seek $fh, $offset, 2;
2739                        $rl = readline $fh;
2740                        defined $rl or return (undef, undef);
2741                        chomp $rl;
2742                }
2743                my $rev = tell $fh;
2744                croak $! if ($rev < -1);
2745                $rev =  ($rev - 41) / 41;
2746                close $fh or croak $!;
2747                return ($rev, $c);
2748        }
2749        return (undef, undef);
2750}
2751
2752sub libsvn_parse_revision {
2753        my $base = shift;
2754        my $head = $SVN->get_latest_revnum();
2755        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
2756                return ($base + 1, $head) if (defined $base);
2757                return (0, $head);
2758        }
2759        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
2760        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
2761        if ($_revision =~ /^BASE:(\d+)$/) {
2762                return ($base + 1, $1) if (defined $base);
2763                return (0, $head);
2764        }
2765        return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
2766        die "revision argument: $_revision not understood by git-svn\n",
2767                "Try using the command-line svn client instead\n";
2768}
2769
2770sub libsvn_traverse {
2771        my ($gui, $pfx, $path, $rev) = @_;
2772        my $cwd = "$pfx/$path";
2773        my $pool = SVN::Pool->new;
2774        $cwd =~ s#^/+##g;
2775        my ($dirent, $r, $props) = $SVN->get_dir($cwd, $rev, $pool);
2776        foreach my $d (keys %$dirent) {
2777                my $t = $dirent->{$d}->kind;
2778                if ($t == $SVN::Node::dir) {
2779                        libsvn_traverse($gui, $cwd, $d, $rev);
2780                } elsif ($t == $SVN::Node::file) {
2781                        print "\tA\t$cwd/$d\n" unless $_q;
2782                        libsvn_get_file($gui, "$cwd/$d", $rev);
2783                }
2784        }
2785        $pool->clear;
2786}
2787
2788sub libsvn_traverse_ignore {
2789        my ($fh, $path, $r) = @_;
2790        $path =~ s#^/+##g;
2791        my $pool = SVN::Pool->new;
2792        my ($dirent, undef, $props) = $SVN->get_dir($path, $r, $pool);
2793        my $p = $path;
2794        $p =~ s#^\Q$SVN_PATH\E/?##;
2795        print $fh length $p ? "\n# $p\n" : "\n# /\n";
2796        if (my $s = $props->{'svn:ignore'}) {
2797                $s =~ s/[\r\n]+/\n/g;
2798                chomp $s;
2799                if (length $p == 0) {
2800                        $s =~ s#\n#\n/$p#g;
2801                        print $fh "/$s\n";
2802                } else {
2803                        $s =~ s#\n#\n/$p/#g;
2804                        print $fh "/$p/$s\n";
2805                }
2806        }
2807        foreach (sort keys %$dirent) {
2808                next if $dirent->{$_}->kind != $SVN::Node::dir;
2809                libsvn_traverse_ignore($fh, "$path/$_", $r);
2810        }
2811        $pool->clear;
2812}
2813
2814sub revisions_eq {
2815        my ($path, $r0, $r1) = @_;
2816        return 1 if $r0 == $r1;
2817        my $nr = 0;
2818        if ($_use_lib) {
2819                # should be OK to use Pool here (r1 - r0) should be small
2820                my $pool = SVN::Pool->new;
2821                libsvn_get_log($SVN, "/$path", $r0, $r1,
2822                                0, 1, 1, sub {$nr++}, $pool);
2823                $pool->clear;
2824        } else {
2825                my ($url, undef) = repo_path_split($SVN_URL);
2826                my $svn_log = svn_log_raw("$url/$path","-r$r0:$r1");
2827                while (next_log_entry($svn_log)) { $nr++ }
2828                close $svn_log->{fh};
2829        }
2830        return 0 if ($nr > 1);
2831        return 1;
2832}
2833
2834sub libsvn_find_parent_branch {
2835        my ($paths, $rev, $author, $date, $msg) = @_;
2836        my $svn_path = '/'.$SVN_PATH;
2837
2838        # look for a parent from another branch:
2839        my $i = $paths->{$svn_path} or return;
2840        my $branch_from = $i->copyfrom_path or return;
2841        my $r = $i->copyfrom_rev;
2842        print STDERR  "Found possible branch point: ",
2843                                "$branch_from => $svn_path, $r\n";
2844        $branch_from =~ s#^/##;
2845        my $l_map = {};
2846        read_url_paths_all($l_map, '', "$GIT_DIR/svn");
2847        my $url = $SVN->{url};
2848        defined $l_map->{$url} or return;
2849        my $id = $l_map->{$url}->{$branch_from};
2850        if (!defined $id && $_follow_parent) {
2851                print STDERR "Following parent: $branch_from\@$r\n";
2852                # auto create a new branch and follow it
2853                $id = basename($branch_from);
2854                $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
2855                while (-r "$GIT_DIR/svn/$id") {
2856                        # just grow a tail if we're not unique enough :x
2857                        $id .= '-';
2858                }
2859        }
2860        return unless defined $id;
2861
2862        my ($r0, $parent) = find_rev_before($r,$id,1);
2863        if ($_follow_parent && (!defined $r0 || !defined $parent)) {
2864                defined(my $pid = fork) or croak $!;
2865                if (!$pid) {
2866                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2867                        init_vars();
2868                        $SVN_URL = "$url/$branch_from";
2869                        $SVN_LOG = $SVN = undef;
2870                        setup_git_svn();
2871                        # we can't assume SVN_URL exists at r+1:
2872                        $_revision = "0:$r";
2873                        fetch_lib();
2874                        exit 0;
2875                }
2876                waitpid $pid, 0;
2877                croak $? if $?;
2878                ($r0, $parent) = find_rev_before($r,$id,1);
2879        }
2880        return unless (defined $r0 && defined $parent);
2881        if (revisions_eq($branch_from, $r0, $r)) {
2882                unlink $GIT_SVN_INDEX;
2883                print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
2884                sys(qw/git-read-tree/, $parent);
2885                return libsvn_fetch($parent, $paths, $rev,
2886                                        $author, $date, $msg);
2887        }
2888        print STDERR "Nope, branch point not imported or unknown\n";
2889        return undef;
2890}
2891
2892sub libsvn_get_log {
2893        my ($ra, @args) = @_;
2894        if ($SVN::Core::VERSION le '1.2.0') {
2895                splice(@args, 3, 1);
2896        }
2897        $ra->get_log(@args);
2898}
2899
2900sub libsvn_new_tree {
2901        if (my $log_entry = libsvn_find_parent_branch(@_)) {
2902                return $log_entry;
2903        }
2904        my ($paths, $rev, $author, $date, $msg) = @_;
2905        open my $gui, '| git-update-index -z --index-info' or croak $!;
2906        my $pool = SVN::Pool->new;
2907        libsvn_traverse($gui, '', $SVN_PATH, $rev, $pool);
2908        $pool->clear;
2909        close $gui or croak $?;
2910        return libsvn_log_entry($rev, $author, $date, $msg);
2911}
2912
2913sub find_graft_path_commit {
2914        my ($tree_paths, $p1, $r1) = @_;
2915        foreach my $x (keys %$tree_paths) {
2916                next unless ($p1 =~ /^\Q$x\E/);
2917                my $i = $tree_paths->{$x};
2918                my ($r0, $parent) = find_rev_before($r1,$i,1);
2919                return $parent if (defined $r0 && $r0 == $r1);
2920                print STDERR "r$r1 of $i not imported\n";
2921                next;
2922        }
2923        return undef;
2924}
2925
2926sub find_graft_path_parents {
2927        my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
2928        foreach my $x (keys %$tree_paths) {
2929                next unless ($p0 =~ /^\Q$x\E/);
2930                my $i = $tree_paths->{$x};
2931                my ($r, $parent) = find_rev_before($r0, $i, 1);
2932                if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
2933                        my ($url_b, undef, $uuid_b) = cmt_metadata($c);
2934                        my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
2935                        next if ($url_a && $url_b && $url_a eq $url_b &&
2936                                                        $uuid_b eq $uuid_a);
2937                        $grafts->{$c}->{$parent} = 1;
2938                }
2939        }
2940}
2941
2942sub libsvn_graft_file_copies {
2943        my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
2944        foreach (keys %$paths) {
2945                my $i = $paths->{$_};
2946                my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
2947                                        $i->copyfrom_rev);
2948                next unless (defined $p0 && defined $r0);
2949
2950                my $p1 = $_;
2951                $p1 =~ s#^/##;
2952                $p0 =~ s#^/##;
2953                my $c = find_graft_path_commit($tree_paths, $p1, $rev);
2954                next unless $c;
2955                find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
2956        }
2957}
2958
2959sub set_index {
2960        my $old = $ENV{GIT_INDEX_FILE};
2961        $ENV{GIT_INDEX_FILE} = shift;
2962        return $old;
2963}
2964
2965sub restore_index {
2966        my ($old) = @_;
2967        if (defined $old) {
2968                $ENV{GIT_INDEX_FILE} = $old;
2969        } else {
2970                delete $ENV{GIT_INDEX_FILE};
2971        }
2972}
2973
2974sub libsvn_commit_cb {
2975        my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
2976        if ($_optimize_commits && $rev == ($r_last + 1)) {
2977                my $log = libsvn_log_entry($rev,$committer,$date,$msg);
2978                $log->{tree} = get_tree_from_treeish($c);
2979                my $cmt = git_commit($log, $cmt_last, $c);
2980                my @diff = safe_qx('git-diff-tree', $cmt, $c);
2981                if (@diff) {
2982                        print STDERR "Trees differ: $cmt $c\n",
2983                                        join('',@diff),"\n";
2984                        exit 1;
2985                }
2986        } else {
2987                fetch("$rev=$c");
2988        }
2989}
2990
2991sub libsvn_ls_fullurl {
2992        my $fullurl = shift;
2993        my ($repo, $path) = repo_path_split($fullurl);
2994        $SVN ||= libsvn_connect($repo);
2995        my @ret;
2996        my $pool = SVN::Pool->new;
2997        my ($dirent, undef, undef) = $SVN->get_dir($path,
2998                                                $SVN->get_latest_revnum, $pool);
2999        foreach my $d (keys %$dirent) {
3000                if ($dirent->{$d}->kind == $SVN::Node::dir) {
3001                        push @ret, "$d/"; # add '/' for compat with cli svn
3002                }
3003        }
3004        $pool->clear;
3005        return @ret;
3006}
3007
3008
3009sub libsvn_skip_unknown_revs {
3010        my $err = shift;
3011        my $errno = $err->apr_err();
3012        # Maybe the branch we're tracking didn't
3013        # exist when the repo started, so it's
3014        # not an error if it doesn't, just continue
3015        #
3016        # Wonderfully consistent library, eh?
3017        # 160013 - svn:// and file://
3018        # 175002 - http(s)://
3019        #   More codes may be discovered later...
3020        if ($errno == 175002 || $errno == 160013) {
3021                return;
3022        }
3023        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3024};
3025
3026# Tie::File seems to be prone to offset errors if revisions get sparse,
3027# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3028# one of my favorite modules is out :<  Next up would be one of the DBM
3029# modules, but I'm not sure which is most portable...  So I'll just
3030# go with something that's plain-text, but still capable of
3031# being randomly accessed.  So here's my ultra-simple fixed-width
3032# database.  All records are 40 characters + "\n", so it's easy to seek
3033# to a revision: (41 * rev) is the byte offset.
3034# A record of 40 0s denotes an empty revision.
3035# And yes, it's still pretty fast (faster than Tie::File).
3036sub revdb_set {
3037        my ($file, $rev, $commit) = @_;
3038        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
3039        open my $fh, '+<', $file or croak $!;
3040        my $offset = $rev * 41;
3041        # assume that append is the common case:
3042        seek $fh, 0, 2 or croak $!;
3043        my $pos = tell $fh;
3044        if ($pos < $offset) {
3045                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
3046        }
3047        seek $fh, $offset, 0 or croak $!;
3048        print $fh $commit,"\n";
3049        close $fh or croak $!;
3050}
3051
3052sub revdb_get {
3053        my ($file, $rev) = @_;
3054        my $ret;
3055        my $offset = $rev * 41;
3056        open my $fh, '<', $file or croak $!;
3057        seek $fh, $offset, 0;
3058        if (tell $fh == $offset) {
3059                $ret = readline $fh;
3060                if (defined $ret) {
3061                        chomp $ret;
3062                        $ret = undef if ($ret =~ /^0{40}$/);
3063                }
3064        }
3065        close $fh or croak $!;
3066        return $ret;
3067}
3068
3069sub copy_remote_ref {
3070        my $origin = $_cp_remote ? $_cp_remote : 'origin';
3071        my $ref = "refs/remotes/$GIT_SVN";
3072        if (safe_qx('git-ls-remote', $origin, $ref)) {
3073                sys(qw/git fetch/, $origin, "$ref:$ref");
3074        } else {
3075                die "Unable to find remote reference: ",
3076                                "refs/remotes/$GIT_SVN on $origin\n";
3077        }
3078}
3079
3080package SVN::Git::Editor;
3081use vars qw/@ISA/;
3082use strict;
3083use warnings;
3084use Carp qw/croak/;
3085use IO::File;
3086
3087sub new {
3088        my $class = shift;
3089        my $git_svn = shift;
3090        my $self = SVN::Delta::Editor->new(@_);
3091        bless $self, $class;
3092        foreach (qw/svn_path c r ra /) {
3093                die "$_ required!\n" unless (defined $git_svn->{$_});
3094                $self->{$_} = $git_svn->{$_};
3095        }
3096        $self->{pool} = SVN::Pool->new;
3097        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3098        $self->{rm} = { };
3099        require Digest::MD5;
3100        return $self;
3101}
3102
3103sub split_path {
3104        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3105}
3106
3107sub repo_path {
3108        (defined $_[1] && length $_[1]) ? "$_[0]->{svn_path}/$_[1]"
3109                                        : $_[0]->{svn_path}
3110}
3111
3112sub url_path {
3113        my ($self, $path) = @_;
3114        $self->{ra}->{url} . '/' . $self->repo_path($path);
3115}
3116
3117sub rmdirs {
3118        my ($self, $q) = @_;
3119        my $rm = $self->{rm};
3120        delete $rm->{''}; # we never delete the url we're tracking
3121        return unless %$rm;
3122
3123        foreach (keys %$rm) {
3124                my @d = split m#/#, $_;
3125                my $c = shift @d;
3126                $rm->{$c} = 1;
3127                while (@d) {
3128                        $c .= '/' . shift @d;
3129                        $rm->{$c} = 1;
3130                }
3131        }
3132        delete $rm->{$self->{svn_path}};
3133        delete $rm->{''}; # we never delete the url we're tracking
3134        return unless %$rm;
3135
3136        defined(my $pid = open my $fh,'-|') or croak $!;
3137        if (!$pid) {
3138                exec qw/git-ls-tree --name-only -r -z/, $self->{c} or croak $!;
3139        }
3140        local $/ = "\0";
3141        my @svn_path = split m#/#, $self->{svn_path};
3142        while (<$fh>) {
3143                chomp;
3144                my @dn = (@svn_path, (split m#/#, $_));
3145                while (pop @dn) {
3146                        delete $rm->{join '/', @dn};
3147                }
3148                unless (%$rm) {
3149                        close $fh;
3150                        return;
3151                }
3152        }
3153        close $fh;
3154
3155        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3156        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3157                $self->close_directory($bat->{$d}, $p);
3158                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3159                print "\tD+\t/$d/\n" unless $q;
3160                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3161                delete $bat->{$d};
3162        }
3163}
3164
3165sub open_or_add_dir {
3166        my ($self, $full_path, $baton) = @_;
3167        my $p = SVN::Pool->new;
3168        my $t = $self->{ra}->check_path($full_path, $self->{r}, $p);
3169        $p->clear;
3170        if ($t == $SVN::Node::none) {
3171                return $self->add_directory($full_path, $baton,
3172                                                undef, -1, $self->{pool});
3173        } elsif ($t == $SVN::Node::dir) {
3174                return $self->open_directory($full_path, $baton,
3175                                                $self->{r}, $self->{pool});
3176        }
3177        print STDERR "$full_path already exists in repository at ",
3178                "r$self->{r} and it is not a directory (",
3179                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3180        exit 1;
3181}
3182
3183sub ensure_path {
3184        my ($self, $path) = @_;
3185        my $bat = $self->{bat};
3186        $path = $self->repo_path($path);
3187        return $bat->{''} unless (length $path);
3188        my @p = split m#/+#, $path;
3189        my $c = shift @p;
3190        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3191        while (@p) {
3192                my $c0 = $c;
3193                $c .= '/' . shift @p;
3194                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3195        }
3196        return $bat->{$c};
3197}
3198
3199sub A {
3200        my ($self, $m, $q) = @_;
3201        my ($dir, $file) = split_path($m->{file_b});
3202        my $pbat = $self->ensure_path($dir);
3203        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3204                                        undef, -1);
3205        print "\tA\t$m->{file_b}\n" unless $q;
3206        $self->chg_file($fbat, $m);
3207        $self->close_file($fbat,undef,$self->{pool});
3208}
3209
3210sub C {
3211        my ($self, $m, $q) = @_;
3212        my ($dir, $file) = split_path($m->{file_b});
3213        my $pbat = $self->ensure_path($dir);
3214        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3215                                $self->url_path($m->{file_a}), $self->{r});
3216        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $q;
3217        $self->chg_file($fbat, $m);
3218        $self->close_file($fbat,undef,$self->{pool});
3219}
3220
3221sub delete_entry {
3222        my ($self, $path, $pbat) = @_;
3223        my $rpath = $self->repo_path($path);
3224        my ($dir, $file) = split_path($rpath);
3225        $self->{rm}->{$dir} = 1;
3226        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3227}
3228
3229sub R {
3230        my ($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}), $self->{r});
3235        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $q;
3236        $self->chg_file($fbat, $m);
3237        $self->close_file($fbat,undef,$self->{pool});
3238
3239        ($dir, $file) = split_path($m->{file_a});
3240        $pbat = $self->ensure_path($dir);
3241        $self->delete_entry($m->{file_a}, $pbat);
3242}
3243
3244sub M {
3245        my ($self, $m, $q) = @_;
3246        my ($dir, $file) = split_path($m->{file_b});
3247        my $pbat = $self->ensure_path($dir);
3248        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3249                                $pbat,$self->{r},$self->{pool});
3250        print "\t$m->{chg}\t$m->{file_b}\n" unless $q;
3251        $self->chg_file($fbat, $m);
3252        $self->close_file($fbat,undef,$self->{pool});
3253}
3254
3255sub T { shift->M(@_) }
3256
3257sub change_file_prop {
3258        my ($self, $fbat, $pname, $pval) = @_;
3259        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3260}
3261
3262sub chg_file {
3263        my ($self, $fbat, $m) = @_;
3264        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3265                $self->change_file_prop($fbat,'svn:executable','*');
3266        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3267                $self->change_file_prop($fbat,'svn:executable',undef);
3268        }
3269        my $fh = IO::File->new_tmpfile or croak $!;
3270        if ($m->{mode_b} =~ /^120/) {
3271                print $fh 'link ' or croak $!;
3272                $self->change_file_prop($fbat,'svn:special','*');
3273        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3274                $self->change_file_prop($fbat,'svn:special',undef);
3275        }
3276        defined(my $pid = fork) or croak $!;
3277        if (!$pid) {
3278                open STDOUT, '>&', $fh or croak $!;
3279                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3280        }
3281        waitpid $pid, 0;
3282        croak $? if $?;
3283        $fh->flush == 0 or croak $!;
3284        seek $fh, 0, 0 or croak $!;
3285
3286        my $md5 = Digest::MD5->new;
3287        $md5->addfile($fh) or croak $!;
3288        seek $fh, 0, 0 or croak $!;
3289
3290        my $exp = $md5->hexdigest;
3291        my $atd = $self->apply_textdelta($fbat, undef, $self->{pool});
3292        my $got = SVN::TxDelta::send_stream($fh, @$atd, $self->{pool});
3293        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3294
3295        close $fh or croak $!;
3296}
3297
3298sub D {
3299        my ($self, $m, $q) = @_;
3300        my ($dir, $file) = split_path($m->{file_b});
3301        my $pbat = $self->ensure_path($dir);
3302        print "\tD\t$m->{file_b}\n" unless $q;
3303        $self->delete_entry($m->{file_b}, $pbat);
3304}
3305
3306sub close_edit {
3307        my ($self) = @_;
3308        my ($p,$bat) = ($self->{pool}, $self->{bat});
3309        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3310                $self->close_directory($bat->{$_}, $p);
3311        }
3312        $self->SUPER::close_edit($p);
3313        $p->clear;
3314}
3315
3316sub abort_edit {
3317        my ($self) = @_;
3318        $self->SUPER::abort_edit($self->{pool});
3319        $self->{pool}->clear;
3320}
3321
3322__END__
3323
3324Data structures:
3325
3326$svn_log hashref (as returned by svn_log_raw)
3327{
3328        fh => file handle of the log file,
3329        state => state of the log file parser (sep/msg/rev/msg_start...)
3330}
3331
3332$log_msg hashref as returned by next_log_entry($svn_log)
3333{
3334        msg => 'whitespace-formatted log entry
3335',                                              # trailing newline is preserved
3336        revision => '8',                        # integer
3337        date => '2004-02-24T17:01:44.108345Z',  # commit date
3338        author => 'committer name'
3339};
3340
3341
3342@mods = array of diff-index line hashes, each element represents one line
3343        of diff-index output
3344
3345diff-index line ($m hash)
3346{
3347        mode_a => first column of diff-index output, no leading ':',
3348        mode_b => second column of diff-index output,
3349        sha1_b => sha1sum of the final blob,
3350        chg => change type [MCRADT],
3351        file_a => original file name of a file (iff chg is 'C' or 'R')
3352        file_b => new/current file name of a file (any chg)
3353}
3354;
3355
3356# retval of read_url_paths{,_all}();
3357$l_map = {
3358        # repository root url
3359        'https://svn.musicpd.org' => {
3360                # repository path               # GIT_SVN_ID
3361                'mpd/trunk'             =>      'trunk',
3362                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
3363        },
3364}
3365
3366Notes:
3367        I don't trust the each() function on unless I created %hash myself
3368        because the internal iterator may not have started at base.