git-svn.perlon commit gitweb: Add author and contributor email to Atom feed (ab23c19)
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $SVN_URL $SVN_INFO $SVN_WC $SVN_UUID
   8                $GIT_SVN_INDEX $GIT_SVN
   9                $GIT_DIR $GIT_SVN_DIR $REVDB/;
  10$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  11$VERSION = '@@GIT_VERSION@@';
  12
  13use Cwd qw/abs_path/;
  14$GIT_DIR = abs_path($ENV{GIT_DIR} || '.git');
  15$ENV{GIT_DIR} = $GIT_DIR;
  16
  17my $LC_ALL = $ENV{LC_ALL};
  18my $TZ = $ENV{TZ};
  19# make sure the svn binary gives consistent output between locales and TZs:
  20$ENV{TZ} = 'UTC';
  21$ENV{LC_ALL} = 'C';
  22$| = 1; # unbuffer STDOUT
  23
  24sub fatal (@) { print STDERR $@; exit 1 }
  25# If SVN:: library support is added, please make the dependencies
  26# optional and preserve the capability to use the command-line client.
  27# use eval { require SVN::... } to make it lazy load
  28# We don't use any modules not in the standard Perl distribution:
  29use Carp qw/croak/;
  30use IO::File qw//;
  31use File::Basename qw/dirname basename/;
  32use File::Path qw/mkpath/;
  33use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  34use File::Spec qw//;
  35use File::Copy qw/copy/;
  36use POSIX qw/strftime/;
  37use IPC::Open3;
  38use Memoize;
  39memoize('revisions_eq');
  40memoize('cmt_metadata');
  41memoize('get_commit_time');
  42
  43my ($SVN, $_use_lib);
  44
  45sub nag_lib {
  46        print STDERR <<EOF;
  47! Please consider installing the SVN Perl libraries (version 1.1.0 or
  48! newer).  You will generally get better performance and fewer bugs,
  49! especially if you:
  50! 1) have a case-insensitive filesystem
  51! 2) replace symlinks with files (and vice-versa) in commits
  52
  53EOF
  54}
  55
  56$_use_lib = 1 unless $ENV{GIT_SVN_NO_LIB};
  57libsvn_load();
  58nag_lib() unless $_use_lib;
  59
  60my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  61my $sha1 = qr/[a-f\d]{40}/;
  62my $sha1_short = qr/[a-f\d]{4,40}/;
  63my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
  64        $_find_copies_harder, $_l, $_cp_similarity, $_cp_remote,
  65        $_repack, $_repack_nr, $_repack_flags, $_q,
  66        $_message, $_file, $_follow_parent, $_no_metadata,
  67        $_template, $_shared, $_no_default_regex, $_no_graft_copy,
  68        $_limit, $_verbose, $_incremental, $_oneline, $_l_fmt, $_show_commit,
  69        $_version, $_upgrade, $_authors, $_branch_all_refs, @_opt_m,
  70        $_merge, $_strategy, $_dry_run, $_ignore_nodate, $_non_recursive,
  71        $_username, $_config_dir, $_no_auth_cache);
  72my (@_branch_from, %tree_map, %users, %rusers, %equiv);
  73my ($_svn_co_url_revs, $_svn_pg_peg_revs);
  74my @repo_path_split_cache;
  75
  76my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
  77                'branch|b=s' => \@_branch_from,
  78                'follow-parent|follow' => \$_follow_parent,
  79                'branch-all-refs|B' => \$_branch_all_refs,
  80                'authors-file|A=s' => \$_authors,
  81                'repack:i' => \$_repack,
  82                'no-metadata' => \$_no_metadata,
  83                'quiet|q' => \$_q,
  84                'username=s' => \$_username,
  85                'config-dir=s' => \$_config_dir,
  86                'no-auth-cache' => \$_no_auth_cache,
  87                'ignore-nodate' => \$_ignore_nodate,
  88                'repack-flags|repack-args|repack-opts=s' => \$_repack_flags);
  89
  90my ($_trunk, $_tags, $_branches);
  91my %multi_opts = ( 'trunk|T=s' => \$_trunk,
  92                'tags|t=s' => \$_tags,
  93                'branches|b=s' => \$_branches );
  94my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
  95my %cmt_opts = ( 'edit|e' => \$_edit,
  96                'rmdir' => \$_rmdir,
  97                'find-copies-harder' => \$_find_copies_harder,
  98                'l=i' => \$_l,
  99                'copy-similarity|C=i'=> \$_cp_similarity
 100);
 101
 102my %cmd = (
 103        fetch => [ \&fetch, "Download new revisions from SVN",
 104                        { 'revision|r=s' => \$_revision, %fc_opts } ],
 105        init => [ \&init, "Initialize a repo for tracking" .
 106                          " (requires URL argument)",
 107                          \%init_opts ],
 108        commit => [ \&commit, "Commit git revisions to SVN",
 109                        {       'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 110        'show-ignore' => [ \&show_ignore, "Show svn:ignore listings",
 111                        { 'revision|r=i' => \$_revision } ],
 112        rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
 113                        { 'no-ignore-externals' => \$_no_ignore_ext,
 114                          'copy-remote|remote=s' => \$_cp_remote,
 115                          'upgrade' => \$_upgrade } ],
 116        'graft-branches' => [ \&graft_branches,
 117                        'Detect merges/branches from already imported history',
 118                        { 'merge-rx|m' => \@_opt_m,
 119                          'branch|b=s' => \@_branch_from,
 120                          'branch-all-refs|B' => \$_branch_all_refs,
 121                          'no-default-regex' => \$_no_default_regex,
 122                          'no-graft-copy' => \$_no_graft_copy } ],
 123        'multi-init' => [ \&multi_init,
 124                        'Initialize multiple trees (like git-svnimport)',
 125                        { %multi_opts, %fc_opts } ],
 126        'multi-fetch' => [ \&multi_fetch,
 127                        'Fetch multiple trees (like git-svnimport)',
 128                        \%fc_opts ],
 129        'log' => [ \&show_log, 'Show commit logs',
 130                        { 'limit=i' => \$_limit,
 131                          'revision|r=s' => \$_revision,
 132                          'verbose|v' => \$_verbose,
 133                          'incremental' => \$_incremental,
 134                          'oneline' => \$_oneline,
 135                          'show-commit' => \$_show_commit,
 136                          'non-recursive' => \$_non_recursive,
 137                          'authors-file|A=s' => \$_authors,
 138                        } ],
 139        'commit-diff' => [ \&commit_diff, 'Commit a diff between two trees',
 140                        { 'message|m=s' => \$_message,
 141                          'file|F=s' => \$_file,
 142                          'revision|r=s' => \$_revision,
 143                        %cmt_opts } ],
 144        dcommit => [ \&dcommit, 'Commit several diffs to merge with upstream',
 145                        { 'merge|m|M' => \$_merge,
 146                          'strategy|s=s' => \$_strategy,
 147                          'dry-run|n' => \$_dry_run,
 148                        %cmt_opts } ],
 149);
 150
 151my $cmd;
 152for (my $i = 0; $i < @ARGV; $i++) {
 153        if (defined $cmd{$ARGV[$i]}) {
 154                $cmd = $ARGV[$i];
 155                splice @ARGV, $i, 1;
 156                last;
 157        }
 158};
 159
 160my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 161
 162read_repo_config(\%opts);
 163my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
 164                                'version|V' => \$_version,
 165                                'id|i=s' => \$GIT_SVN);
 166exit 1 if (!$rv && $cmd ne 'log');
 167
 168set_default_vals();
 169usage(0) if $_help;
 170version() if $_version;
 171usage(1) unless defined $cmd;
 172init_vars();
 173load_authors() if $_authors;
 174load_all_refs() if $_branch_all_refs;
 175svn_compat_check() unless $_use_lib;
 176migration_check() unless $cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/;
 177$cmd{$cmd}->[0]->(@ARGV);
 178exit 0;
 179
 180####################### primary functions ######################
 181sub usage {
 182        my $exit = shift || 0;
 183        my $fd = $exit ? \*STDERR : \*STDOUT;
 184        print $fd <<"";
 185git-svn - bidirectional operations between a single Subversion tree and git
 186Usage: $0 <command> [options] [arguments]\n
 187
 188        print $fd "Available commands:\n" unless $cmd;
 189
 190        foreach (sort keys %cmd) {
 191                next if $cmd && $cmd ne $_;
 192                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 193                foreach (keys %{$cmd{$_}->[2]}) {
 194                        # prints out arguments as they should be passed:
 195                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 196                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 197                                                        "--$_" : "-$_" }
 198                                                split /\|/,$_)," $x\n";
 199                }
 200        }
 201        print $fd <<"";
 202\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 203arbitrary identifier if you're tracking multiple SVN branches/repositories in
 204one git repository and want to keep them separate.  See git-svn(1) for more
 205information.
 206
 207        exit $exit;
 208}
 209
 210sub version {
 211        print "git-svn version $VERSION\n";
 212        exit 0;
 213}
 214
 215sub rebuild {
 216        if (quiet_run(qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0")) {
 217                copy_remote_ref();
 218        }
 219        $SVN_URL = shift or undef;
 220        my $newest_rev = 0;
 221        if ($_upgrade) {
 222                sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD");
 223        } else {
 224                check_upgrade_needed();
 225        }
 226
 227        my $pid = open(my $rev_list,'-|');
 228        defined $pid or croak $!;
 229        if ($pid == 0) {
 230                exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!;
 231        }
 232        my $latest;
 233        while (<$rev_list>) {
 234                chomp;
 235                my $c = $_;
 236                croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 237                my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
 238                next if (!@commit); # skip merges
 239                my ($url, $rev, $uuid) = extract_metadata($commit[$#commit]);
 240                if (!defined $rev || !$uuid) {
 241                        croak "Unable to extract revision or UUID from ",
 242                                "$c, $commit[$#commit]\n";
 243                }
 244
 245                # if we merged or otherwise started elsewhere, this is
 246                # how we break out of it
 247                next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
 248                next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
 249
 250                unless (defined $latest) {
 251                        if (!$SVN_URL && !$url) {
 252                                croak "SVN repository location required: $url\n";
 253                        }
 254                        $SVN_URL ||= $url;
 255                        $SVN_UUID ||= $uuid;
 256                        setup_git_svn();
 257                        $latest = $rev;
 258                }
 259                revdb_set($REVDB, $rev, $c);
 260                print "r$rev = $c\n";
 261                $newest_rev = $rev if ($rev > $newest_rev);
 262        }
 263        close $rev_list or croak $?;
 264
 265        goto out if $_use_lib;
 266        if (!chdir $SVN_WC) {
 267                svn_cmd_checkout($SVN_URL, $latest, $SVN_WC);
 268                chdir $SVN_WC or croak $!;
 269        }
 270
 271        $pid = fork;
 272        defined $pid or croak $!;
 273        if ($pid == 0) {
 274                my @svn_up = qw(svn up);
 275                push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 276                sys(@svn_up,"-r$newest_rev");
 277                $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 278                index_changes();
 279                exec('git-write-tree') or croak $!;
 280        }
 281        waitpid $pid, 0;
 282        croak $? if $?;
 283out:
 284        if ($_upgrade) {
 285                print STDERR <<"";
 286Keeping deprecated refs/head/$GIT_SVN-HEAD for now.  Please remove it
 287when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN
 288
 289        }
 290}
 291
 292sub init {
 293        my $url = shift or die "SVN repository location required " .
 294                                "as a command-line argument\n";
 295        $url =~ s!/+$!!; # strip trailing slash
 296
 297        if (my $repo_path = shift) {
 298                unless (-d $repo_path) {
 299                        mkpath([$repo_path]);
 300                }
 301                $GIT_DIR = $ENV{GIT_DIR} = $repo_path . "/.git";
 302                init_vars();
 303        }
 304
 305        $SVN_URL = $url;
 306        unless (-d $GIT_DIR) {
 307                my @init_db = ('git-init-db');
 308                push @init_db, "--template=$_template" if defined $_template;
 309                push @init_db, "--shared" if defined $_shared;
 310                sys(@init_db);
 311        }
 312        setup_git_svn();
 313}
 314
 315sub fetch {
 316        check_upgrade_needed();
 317        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 318        my $ret = $_use_lib ? fetch_lib(@_) : fetch_cmd(@_);
 319        if ($ret->{commit} && quiet_run(qw(git-rev-parse --verify
 320                                                refs/heads/master^0))) {
 321                sys(qw(git-update-ref refs/heads/master),$ret->{commit});
 322        }
 323        return $ret;
 324}
 325
 326sub fetch_cmd {
 327        my (@parents) = @_;
 328        my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 329        unless ($_revision) {
 330                $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 331        }
 332        push @log_args, "-r$_revision";
 333        push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 334
 335        my $svn_log = svn_log_raw(@log_args);
 336
 337        my $base = next_log_entry($svn_log) or croak "No base revision!\n";
 338        # don't need last_revision from grab_base_rev() because
 339        # user could've specified a different revision to skip (they
 340        # didn't want to import certain revisions into git for whatever
 341        # reason, so trust $base->{revision} instead.
 342        my (undef, $last_commit) = svn_grab_base_rev();
 343        unless (-d $SVN_WC) {
 344                svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC);
 345                chdir $SVN_WC or croak $!;
 346                read_uuid();
 347                $last_commit = git_commit($base, @parents);
 348                assert_tree($last_commit);
 349        } else {
 350                chdir $SVN_WC or croak $!;
 351                read_uuid();
 352                # looks like a user manually cp'd and svn switch'ed
 353                unless ($last_commit) {
 354                        sys(qw/svn revert -R ./);
 355                        assert_svn_wc_clean($base->{revision});
 356                        $last_commit = git_commit($base, @parents);
 357                        assert_tree($last_commit);
 358                }
 359        }
 360        my @svn_up = qw(svn up);
 361        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 362        my $last = $base;
 363        while (my $log_msg = next_log_entry($svn_log)) {
 364                if ($last->{revision} >= $log_msg->{revision}) {
 365                        croak "Out of order: last >= current: ",
 366                                "$last->{revision} >= $log_msg->{revision}\n";
 367                }
 368                # Revert is needed for cases like:
 369                # https://svn.musicpd.org/Jamming/trunk (r166:167), but
 370                # I can't seem to reproduce something like that on a test...
 371                sys(qw/svn revert -R ./);
 372                assert_svn_wc_clean($last->{revision});
 373                sys(@svn_up,"-r$log_msg->{revision}");
 374                $last_commit = git_commit($log_msg, $last_commit, @parents);
 375                $last = $log_msg;
 376        }
 377        close $svn_log->{fh};
 378        $last->{commit} = $last_commit;
 379        return $last;
 380}
 381
 382sub fetch_lib {
 383        my (@parents) = @_;
 384        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 385        $SVN ||= libsvn_connect($SVN_URL);
 386        my ($last_rev, $last_commit) = svn_grab_base_rev();
 387        my ($base, $head) = libsvn_parse_revision($last_rev);
 388        if ($base > $head) {
 389                return { revision => $last_rev, commit => $last_commit }
 390        }
 391        my $index = set_index($GIT_SVN_INDEX);
 392
 393        # limit ourselves and also fork() since get_log won't release memory
 394        # after processing a revision and SVN stuff seems to leak
 395        my $inc = 1000;
 396        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
 397        read_uuid();
 398        if (defined $last_commit) {
 399                unless (-e $GIT_SVN_INDEX) {
 400                        sys(qw/git-read-tree/, $last_commit);
 401                }
 402                chomp (my $x = `git-write-tree`);
 403                my ($y) = (`git-cat-file commit $last_commit`
 404                                                        =~ /^tree ($sha1)/m);
 405                if ($y ne $x) {
 406                        unlink $GIT_SVN_INDEX or croak $!;
 407                        sys(qw/git-read-tree/, $last_commit);
 408                }
 409                chomp ($x = `git-write-tree`);
 410                if ($y ne $x) {
 411                        print STDERR "trees ($last_commit) $y != $x\n",
 412                                 "Something is seriously wrong...\n";
 413                }
 414        }
 415        while (1) {
 416                # fork, because using SVN::Pool with get_log() still doesn't
 417                # seem to help enough to keep memory usage down.
 418                defined(my $pid = fork) or croak $!;
 419                if (!$pid) {
 420                        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
 421
 422                        # Yes I'm perfectly aware that the fourth argument
 423                        # below is the limit revisions number.  Unfortunately
 424                        # performance sucks with it enabled, so it's much
 425                        # faster to fetch revision ranges instead of relying
 426                        # on the limiter.
 427                        libsvn_get_log(libsvn_dup_ra($SVN), [''],
 428                                        $min, $max, 0, 1, 1,
 429                                sub {
 430                                        my $log_msg;
 431                                        if ($last_commit) {
 432                                                $log_msg = libsvn_fetch(
 433                                                        $last_commit, @_);
 434                                                $last_commit = git_commit(
 435                                                        $log_msg,
 436                                                        $last_commit,
 437                                                        @parents);
 438                                        } else {
 439                                                $log_msg = libsvn_new_tree(@_);
 440                                                $last_commit = git_commit(
 441                                                        $log_msg, @parents);
 442                                        }
 443                                });
 444                        exit 0;
 445                }
 446                waitpid $pid, 0;
 447                croak $? if $?;
 448                ($last_rev, $last_commit) = svn_grab_base_rev();
 449                last if ($max >= $head);
 450                $min = $max + 1;
 451                $max += $inc;
 452                $max = $head if ($max > $head);
 453        }
 454        restore_index($index);
 455        return { revision => $last_rev, commit => $last_commit };
 456}
 457
 458sub commit {
 459        my (@commits) = @_;
 460        check_upgrade_needed();
 461        if ($_stdin || !@commits) {
 462                print "Reading from stdin...\n";
 463                @commits = ();
 464                while (<STDIN>) {
 465                        if (/\b($sha1_short)\b/o) {
 466                                unshift @commits, $1;
 467                        }
 468                }
 469        }
 470        my @revs;
 471        foreach my $c (@commits) {
 472                chomp(my @tmp = safe_qx('git-rev-parse',$c));
 473                if (scalar @tmp == 1) {
 474                        push @revs, $tmp[0];
 475                } elsif (scalar @tmp > 1) {
 476                        push @revs, reverse (safe_qx('git-rev-list',@tmp));
 477                } else {
 478                        die "Failed to rev-parse $c\n";
 479                }
 480        }
 481        chomp @revs;
 482        $_use_lib ? commit_lib(@revs) : commit_cmd(@revs);
 483        print "Done committing ",scalar @revs," revisions to SVN\n";
 484}
 485
 486sub commit_cmd {
 487        my (@revs) = @_;
 488
 489        chdir $SVN_WC or croak "Unable to chdir $SVN_WC: $!\n";
 490        my $info = svn_info('.');
 491        my $fetched = fetch();
 492        if ($info->{Revision} != $fetched->{revision}) {
 493                print STDERR "There are new revisions that were fetched ",
 494                                "and need to be merged (or acknowledged) ",
 495                                "before committing.\n";
 496                exit 1;
 497        }
 498        $info = svn_info('.');
 499        read_uuid($info);
 500        my $last = $fetched;
 501        foreach my $c (@revs) {
 502                my $mods = svn_checkout_tree($last, $c);
 503                if (scalar @$mods == 0) {
 504                        print "Skipping, no changes detected\n";
 505                        next;
 506                }
 507                $last = svn_commit_tree($last, $c);
 508        }
 509}
 510
 511sub commit_lib {
 512        my (@revs) = @_;
 513        my ($r_last, $cmt_last) = svn_grab_base_rev();
 514        defined $r_last or die "Must have an existing revision to commit\n";
 515        my $fetched = fetch();
 516        if ($r_last != $fetched->{revision}) {
 517                print STDERR "There are new revisions that were fetched ",
 518                                "and need to be merged (or acknowledged) ",
 519                                "before committing.\n",
 520                                "last rev: $r_last\n",
 521                                " current: $fetched->{revision}\n";
 522                exit 1;
 523        }
 524        read_uuid();
 525        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 526        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
 527
 528        my $repo;
 529        set_svn_commit_env();
 530        foreach my $c (@revs) {
 531                my $log_msg = get_commit_message($c, $commit_msg);
 532
 533                # fork for each commit because there's a memory leak I
 534                # can't track down... (it's probably in the SVN code)
 535                defined(my $pid = open my $fh, '-|') or croak $!;
 536                if (!$pid) {
 537                        my $ed = SVN::Git::Editor->new(
 538                                        {       r => $r_last,
 539                                                ra => libsvn_dup_ra($SVN),
 540                                                c => $c,
 541                                                svn_path => $SVN->{svn_path},
 542                                        },
 543                                        $SVN->get_commit_editor(
 544                                                $log_msg->{msg},
 545                                                sub {
 546                                                        libsvn_commit_cb(
 547                                                                @_, $c,
 548                                                                $log_msg->{msg},
 549                                                                $r_last,
 550                                                                $cmt_last)
 551                                                },
 552                                                @lock)
 553                                        );
 554                        my $mods = libsvn_checkout_tree($cmt_last, $c, $ed);
 555                        if (@$mods == 0) {
 556                                print "No changes\nr$r_last = $cmt_last\n";
 557                                $ed->abort_edit;
 558                        } else {
 559                                $ed->close_edit;
 560                        }
 561                        exit 0;
 562                }
 563                my ($r_new, $cmt_new, $no);
 564                while (<$fh>) {
 565                        print $_;
 566                        chomp;
 567                        if (/^r(\d+) = ($sha1)$/o) {
 568                                ($r_new, $cmt_new) = ($1, $2);
 569                        } elsif ($_ eq 'No changes') {
 570                                $no = 1;
 571                        }
 572                }
 573                close $fh or exit 1;
 574                if (! defined $r_new && ! defined $cmt_new) {
 575                        unless ($no) {
 576                                die "Failed to parse revision information\n";
 577                        }
 578                } else {
 579                        ($r_last, $cmt_last) = ($r_new, $cmt_new);
 580                }
 581        }
 582        $ENV{LC_ALL} = 'C';
 583        unlink $commit_msg;
 584}
 585
 586sub dcommit {
 587        my $gs = "refs/remotes/$GIT_SVN";
 588        chomp(my @refs = safe_qx(qw/git-rev-list --no-merges/, "$gs..HEAD"));
 589        my $last_rev;
 590        foreach my $d (reverse @refs) {
 591                if (quiet_run('git-rev-parse','--verify',"$d~1") != 0) {
 592                        die "Commit $d\n",
 593                            "has no parent commit, and therefore ",
 594                            "nothing to diff against.\n",
 595                            "You should be working from a repository ",
 596                            "originally created by git-svn\n";
 597                }
 598                unless (defined $last_rev) {
 599                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 600                        unless (defined $last_rev) {
 601                                die "Unable to extract revision information ",
 602                                    "from commit $d~1\n";
 603                        }
 604                }
 605                if ($_dry_run) {
 606                        print "diff-tree $d~1 $d\n";
 607                } else {
 608                        if (my $r = commit_diff("$d~1", $d, undef, $last_rev)) {
 609                                $last_rev = $r;
 610                        } # else: no changes, same $last_rev
 611                }
 612        }
 613        return if $_dry_run;
 614        fetch();
 615        my @diff = safe_qx(qw/git-diff-tree HEAD/, $gs);
 616        my @finish;
 617        if (@diff) {
 618                @finish = qw/rebase/;
 619                push @finish, qw/--merge/ if $_merge;
 620                push @finish, "--strategy=$_strategy" if $_strategy;
 621                print STDERR "W: HEAD and $gs differ, using @finish:\n", @diff;
 622        } else {
 623                print "No changes between current HEAD and $gs\n",
 624                      "Hard resetting to the latest $gs\n";
 625                @finish = qw/reset --mixed/;
 626        }
 627        sys('git', @finish, $gs);
 628}
 629
 630sub show_ignore {
 631        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 632        $_use_lib ? show_ignore_lib() : show_ignore_cmd();
 633}
 634
 635sub show_ignore_cmd {
 636        require File::Find or die $!;
 637        if (defined $_revision) {
 638                die "-r/--revision option doesn't work unless the Perl SVN ",
 639                        "libraries are used\n";
 640        }
 641        chdir $SVN_WC or croak $!;
 642        my %ign;
 643        File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 644                s#^\./##;
 645                @{$ign{$_}} = svn_propget_base('svn:ignore', $_);
 646                }}, no_chdir=>1},'.');
 647
 648        print "\n# /\n";
 649        foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 650        delete $ign{'.'};
 651        foreach my $i (sort keys %ign) {
 652                print "\n# ",$i,"\n";
 653                foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 654        }
 655}
 656
 657sub show_ignore_lib {
 658        my $repo;
 659        $SVN ||= libsvn_connect($SVN_URL);
 660        my $r = defined $_revision ? $_revision : $SVN->get_latest_revnum;
 661        libsvn_traverse_ignore(\*STDOUT, $SVN->{svn_path}, $r);
 662}
 663
 664sub graft_branches {
 665        my $gr_file = "$GIT_DIR/info/grafts";
 666        my ($grafts, $comments) = read_grafts($gr_file);
 667        my $gr_sha1;
 668
 669        if (%$grafts) {
 670                # temporarily disable our grafts file to make this idempotent
 671                chomp($gr_sha1 = safe_qx(qw/git-hash-object -w/,$gr_file));
 672                rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
 673        }
 674
 675        my $l_map = read_url_paths();
 676        my @re = map { qr/$_/is } @_opt_m if @_opt_m;
 677        unless ($_no_default_regex) {
 678                push @re, (qr/\b(?:merge|merging|merged)\s+with\s+([\w\.\-]+)/i,
 679                        qr/\b(?:merge|merging|merged)\s+([\w\.\-]+)/i,
 680                        qr/\b(?:from|of)\s+([\w\.\-]+)/i );
 681        }
 682        foreach my $u (keys %$l_map) {
 683                if (@re) {
 684                        foreach my $p (keys %{$l_map->{$u}}) {
 685                                graft_merge_msg($grafts,$l_map,$u,$p,@re);
 686                        }
 687                }
 688                unless ($_no_graft_copy) {
 689                        if ($_use_lib) {
 690                                graft_file_copy_lib($grafts,$l_map,$u);
 691                        } else {
 692                                graft_file_copy_cmd($grafts,$l_map,$u);
 693                        }
 694                }
 695        }
 696        graft_tree_joins($grafts);
 697
 698        write_grafts($grafts, $comments, $gr_file);
 699        unlink "$gr_file~$gr_sha1" if $gr_sha1;
 700}
 701
 702sub multi_init {
 703        my $url = shift;
 704        $_trunk ||= 'trunk';
 705        $_trunk =~ s#/+$##;
 706        $url =~ s#/+$## if $url;
 707        if ($_trunk !~ m#^[a-z\+]+://#) {
 708                $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#);
 709                unless ($url) {
 710                        print STDERR "E: '$_trunk' is not a complete URL ",
 711                                "and a separate URL is not specified\n";
 712                        exit 1;
 713                }
 714                $_trunk = $url . $_trunk;
 715        }
 716        my $ch_id;
 717        if ($GIT_SVN eq 'git-svn') {
 718                $ch_id = 1;
 719                $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
 720        }
 721        init_vars();
 722        unless (-d $GIT_SVN_DIR) {
 723                print "GIT_SVN_ID set to 'trunk' for $_trunk\n" if $ch_id;
 724                init($_trunk);
 725                sys('git-repo-config', 'svn.trunk', $_trunk);
 726        }
 727        complete_url_ls_init($url, $_branches, '--branches/-b', '');
 728        complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/');
 729}
 730
 731sub multi_fetch {
 732        # try to do trunk first, since branches/tags
 733        # may be descended from it.
 734        if (-e "$GIT_DIR/svn/trunk/info/url") {
 735                fetch_child_id('trunk', @_);
 736        }
 737        rec_fetch('', "$GIT_DIR/svn", @_);
 738}
 739
 740sub show_log {
 741        my (@args) = @_;
 742        my ($r_min, $r_max);
 743        my $r_last = -1; # prevent dupes
 744        rload_authors() if $_authors;
 745        if (defined $TZ) {
 746                $ENV{TZ} = $TZ;
 747        } else {
 748                delete $ENV{TZ};
 749        }
 750        if (defined $_revision) {
 751                if ($_revision =~ /^(\d+):(\d+)$/) {
 752                        ($r_min, $r_max) = ($1, $2);
 753                } elsif ($_revision =~ /^\d+$/) {
 754                        $r_min = $r_max = $_revision;
 755                } else {
 756                        print STDERR "-r$_revision is not supported, use ",
 757                                "standard \'git log\' arguments instead\n";
 758                        exit 1;
 759                }
 760        }
 761
 762        my $pid = open(my $log,'-|');
 763        defined $pid or croak $!;
 764        if (!$pid) {
 765                exec(git_svn_log_cmd($r_min,$r_max), @args) or croak $!;
 766        }
 767        setup_pager();
 768        my (@k, $c, $d);
 769
 770        while (<$log>) {
 771                if (/^commit ($sha1_short)/o) {
 772                        my $cmt = $1;
 773                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
 774                                $r_last = $c->{r};
 775                                process_commit($c, $r_min, $r_max, \@k) or
 776                                                                goto out;
 777                        }
 778                        $d = undef;
 779                        $c = { c => $cmt };
 780                } elsif (/^author (.+) (\d+) ([\-\+]?\d+)$/) {
 781                        get_author_info($c, $1, $2, $3);
 782                } elsif (/^(?:tree|parent|committer) /) {
 783                        # ignore
 784                } elsif (/^:\d{6} \d{6} $sha1_short/o) {
 785                        push @{$c->{raw}}, $_;
 786                } elsif (/^[ACRMDT]\t/) {
 787                        # we could add $SVN->{svn_path} here, but that requires
 788                        # remote access at the moment (repo_path_split)...
 789                        s#^([ACRMDT])\t#   $1 #;
 790                        push @{$c->{changed}}, $_;
 791                } elsif (/^diff /) {
 792                        $d = 1;
 793                        push @{$c->{diff}}, $_;
 794                } elsif ($d) {
 795                        push @{$c->{diff}}, $_;
 796                } elsif (/^    (git-svn-id:.+)$/) {
 797                        ($c->{url}, $c->{r}, undef) = extract_metadata($1);
 798                } elsif (s/^    //) {
 799                        push @{$c->{l}}, $_;
 800                }
 801        }
 802        if ($c && defined $c->{r} && $c->{r} != $r_last) {
 803                $r_last = $c->{r};
 804                process_commit($c, $r_min, $r_max, \@k);
 805        }
 806        if (@k) {
 807                my $swap = $r_max;
 808                $r_max = $r_min;
 809                $r_min = $swap;
 810                process_commit($_, $r_min, $r_max) foreach reverse @k;
 811        }
 812out:
 813        close $log;
 814        print '-' x72,"\n" unless $_incremental || $_oneline;
 815}
 816
 817sub commit_diff_usage {
 818        print STDERR "Usage: $0 commit-diff <tree-ish> <tree-ish> [<URL>]\n";
 819        exit 1
 820}
 821
 822sub commit_diff {
 823        if (!$_use_lib) {
 824                print STDERR "commit-diff must be used with SVN libraries\n";
 825                exit 1;
 826        }
 827        my $ta = shift or commit_diff_usage();
 828        my $tb = shift or commit_diff_usage();
 829        if (!eval { $SVN_URL = shift || file_to_s("$GIT_SVN_DIR/info/url") }) {
 830                print STDERR "Needed URL or usable git-svn id command-line\n";
 831                commit_diff_usage();
 832        }
 833        my $r = shift;
 834        unless (defined $r) {
 835                if (defined $_revision) {
 836                        $r = $_revision
 837                } else {
 838                        die "-r|--revision is a required argument\n";
 839                }
 840        }
 841        if (defined $_message && defined $_file) {
 842                print STDERR "Both --message/-m and --file/-F specified ",
 843                                "for the commit message.\n",
 844                                "I have no idea what you mean\n";
 845                exit 1;
 846        }
 847        if (defined $_file) {
 848                $_message = file_to_s($_file);
 849        } else {
 850                $_message ||= get_commit_message($tb,
 851                                        "$GIT_DIR/.svn-commit.tmp.$$")->{msg};
 852        }
 853        $SVN ||= libsvn_connect($SVN_URL);
 854        if ($r eq 'HEAD') {
 855                $r = $SVN->get_latest_revnum;
 856        } elsif ($r !~ /^\d+$/) {
 857                die "revision argument: $r not understood by git-svn\n";
 858        }
 859        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 860        my $rev_committed;
 861        my $ed = SVN::Git::Editor->new({        r => $r,
 862                                                ra => libsvn_dup_ra($SVN),
 863                                                c => $tb,
 864                                                svn_path => $SVN->{svn_path}
 865                                        },
 866                                $SVN->get_commit_editor($_message,
 867                                        sub {
 868                                                $rev_committed = $_[0];
 869                                                print "Committed $_[0]\n";
 870                                        }, @lock)
 871                                );
 872        eval {
 873                my $mods = libsvn_checkout_tree($ta, $tb, $ed);
 874                if (@$mods == 0) {
 875                        print "No changes\n$ta == $tb\n";
 876                        $ed->abort_edit;
 877                } else {
 878                        $ed->close_edit;
 879                }
 880        };
 881        fatal "$@\n" if $@;
 882        $_message = $_file = undef;
 883        return $rev_committed;
 884}
 885
 886########################### utility functions #########################
 887
 888sub cmt_showable {
 889        my ($c) = @_;
 890        return 1 if defined $c->{r};
 891        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
 892                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
 893                my @msg = safe_qx(qw/git-cat-file commit/, $c->{c});
 894                shift @msg while ($msg[0] ne "\n");
 895                shift @msg;
 896                @{$c->{l}} = grep !/^git-svn-id: /, @msg;
 897
 898                (undef, $c->{r}, undef) = extract_metadata(
 899                                (grep(/^git-svn-id: /, @msg))[-1]);
 900        }
 901        return defined $c->{r};
 902}
 903
 904sub git_svn_log_cmd {
 905        my ($r_min, $r_max) = @_;
 906        my @cmd = (qw/git-log --abbrev-commit --pretty=raw
 907                        --default/, "refs/remotes/$GIT_SVN");
 908        push @cmd, '-r' unless $_non_recursive;
 909        push @cmd, qw/--raw --name-status/ if $_verbose;
 910        return @cmd unless defined $r_max;
 911        if ($r_max == $r_min) {
 912                push @cmd, '--max-count=1';
 913                if (my $c = revdb_get($REVDB, $r_max)) {
 914                        push @cmd, $c;
 915                }
 916        } else {
 917                my ($c_min, $c_max);
 918                $c_max = revdb_get($REVDB, $r_max);
 919                $c_min = revdb_get($REVDB, $r_min);
 920                if (defined $c_min && defined $c_max) {
 921                        if ($r_max > $r_max) {
 922                                push @cmd, "$c_min..$c_max";
 923                        } else {
 924                                push @cmd, "$c_max..$c_min";
 925                        }
 926                } elsif ($r_max > $r_min) {
 927                        push @cmd, $c_max;
 928                } else {
 929                        push @cmd, $c_min;
 930                }
 931        }
 932        return @cmd;
 933}
 934
 935sub fetch_child_id {
 936        my $id = shift;
 937        print "Fetching $id\n";
 938        my $ref = "$GIT_DIR/refs/remotes/$id";
 939        defined(my $pid = open my $fh, '-|') or croak $!;
 940        if (!$pid) {
 941                $_repack = undef;
 942                $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 943                init_vars();
 944                fetch(@_);
 945                exit 0;
 946        }
 947        while (<$fh>) {
 948                print $_;
 949                check_repack() if (/^r\d+ = $sha1/);
 950        }
 951        close $fh or croak $?;
 952}
 953
 954sub rec_fetch {
 955        my ($pfx, $p, @args) = @_;
 956        my @dir;
 957        foreach (sort <$p/*>) {
 958                if (-r "$_/info/url") {
 959                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
 960                        my $id = $pfx . basename $_;
 961                        next if $id eq 'trunk';
 962                        fetch_child_id($id, @args);
 963                } elsif (-d $_) {
 964                        push @dir, $_;
 965                }
 966        }
 967        foreach (@dir) {
 968                my $x = $_;
 969                $x =~ s!^\Q$GIT_DIR\E/svn/!!;
 970                rec_fetch($x, $_);
 971        }
 972}
 973
 974sub complete_url_ls_init {
 975        my ($url, $var, $switch, $pfx) = @_;
 976        unless ($var) {
 977                print STDERR "W: $switch not specified\n";
 978                return;
 979        }
 980        $var =~ s#/+$##;
 981        if ($var !~ m#^[a-z\+]+://#) {
 982                $var = '/' . $var if ($var !~ m#^/#);
 983                unless ($url) {
 984                        print STDERR "E: '$var' is not a complete URL ",
 985                                "and a separate URL is not specified\n";
 986                        exit 1;
 987                }
 988                $var = $url . $var;
 989        }
 990        chomp(my @ls = $_use_lib ? libsvn_ls_fullurl($var)
 991                                : safe_qx(qw/svn ls --non-interactive/, $var));
 992        my $old = $GIT_SVN;
 993        defined(my $pid = fork) or croak $!;
 994        if (!$pid) {
 995                foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
 996                        $u =~ s#/+$##;
 997                        if ($u !~ m!\Q$var\E/(.+)$!) {
 998                                print STDERR "W: Unrecognized URL: $u\n";
 999                                die "This should never happen\n";
1000                        }
1001                        # don't try to init already existing refs
1002                        my $id = $pfx.$1;
1003                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
1004                        init_vars();
1005                        unless (-d $GIT_SVN_DIR) {
1006                                print "init $u => $id\n";
1007                                init($u);
1008                        }
1009                }
1010                exit 0;
1011        }
1012        waitpid $pid, 0;
1013        croak $? if $?;
1014        my ($n) = ($switch =~ /^--(\w+)/);
1015        sys('git-repo-config', "svn.$n", $var);
1016}
1017
1018sub common_prefix {
1019        my $paths = shift;
1020        my %common;
1021        foreach (@$paths) {
1022                my @tmp = split m#/#, $_;
1023                my $p = '';
1024                while (my $x = shift @tmp) {
1025                        $p .= "/$x";
1026                        $common{$p} ||= 0;
1027                        $common{$p}++;
1028                }
1029        }
1030        foreach (sort {length $b <=> length $a} keys %common) {
1031                if ($common{$_} == @$paths) {
1032                        return $_;
1033                }
1034        }
1035        return '';
1036}
1037
1038# grafts set here are 'stronger' in that they're based on actual tree
1039# matches, and won't be deleted from merge-base checking in write_grafts()
1040sub graft_tree_joins {
1041        my $grafts = shift;
1042        map_tree_joins() if (@_branch_from && !%tree_map);
1043        return unless %tree_map;
1044
1045        git_svn_each(sub {
1046                my $i = shift;
1047                defined(my $pid = open my $fh, '-|') or croak $!;
1048                if (!$pid) {
1049                        exec qw/git-rev-list --pretty=raw/,
1050                                        "refs/remotes/$i" or croak $!;
1051                }
1052                while (<$fh>) {
1053                        next unless /^commit ($sha1)$/o;
1054                        my $c = $1;
1055                        my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
1056                        next unless $tree_map{$t};
1057
1058                        my $l;
1059                        do {
1060                                $l = readline $fh;
1061                        } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
1062
1063                        my ($s, $tz) = ($1, $2);
1064                        if ($tz =~ s/^\+//) {
1065                                $s += tz_to_s_offset($tz);
1066                        } elsif ($tz =~ s/^\-//) {
1067                                $s -= tz_to_s_offset($tz);
1068                        }
1069
1070                        my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
1071
1072                        foreach my $p (@{$tree_map{$t}}) {
1073                                next if $p eq $c;
1074                                my $mb = eval {
1075                                        safe_qx('git-merge-base', $c, $p)
1076                                };
1077                                next unless ($@ || $?);
1078                                if (defined $r_a) {
1079                                        # see if SVN says it's a relative
1080                                        my ($url_b, $r_b, $uuid_b) =
1081                                                        cmt_metadata($p);
1082                                        next if (defined $url_b &&
1083                                                        defined $url_a &&
1084                                                        ($url_a eq $url_b) &&
1085                                                        ($uuid_a eq $uuid_b));
1086                                        if ($uuid_a eq $uuid_b) {
1087                                                if ($r_b < $r_a) {
1088                                                        $grafts->{$c}->{$p} = 2;
1089                                                        next;
1090                                                } elsif ($r_b > $r_a) {
1091                                                        $grafts->{$p}->{$c} = 2;
1092                                                        next;
1093                                                }
1094                                        }
1095                                }
1096                                my $ct = get_commit_time($p);
1097                                if ($ct < $s) {
1098                                        $grafts->{$c}->{$p} = 2;
1099                                } elsif ($ct > $s) {
1100                                        $grafts->{$p}->{$c} = 2;
1101                                }
1102                                # what should we do when $ct == $s ?
1103                        }
1104                }
1105                close $fh or croak $?;
1106        });
1107}
1108
1109# this isn't funky-filename safe, but good enough for now...
1110sub graft_file_copy_cmd {
1111        my ($grafts, $l_map, $u) = @_;
1112        my $paths = $l_map->{$u};
1113        my $pfx = common_prefix([keys %$paths]);
1114        $SVN_URL ||= $u.$pfx;
1115        my $pid = open my $fh, '-|';
1116        defined $pid or croak $!;
1117        unless ($pid) {
1118                my @exec = qw/svn log -v/;
1119                push @exec, "-r$_revision" if defined $_revision;
1120                exec @exec, $u.$pfx or croak $!;
1121        }
1122        my ($r, $mp) = (undef, undef);
1123        while (<$fh>) {
1124                chomp;
1125                if (/^\-{72}$/) {
1126                        $mp = $r = undef;
1127                } elsif (/^r(\d+) \| /) {
1128                        $r = $1 unless defined $r;
1129                } elsif (/^Changed paths:/) {
1130                        $mp = 1;
1131                } elsif ($mp && m#^   [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) {
1132                        my ($p1, $p0, $r0) = ($1, $2, $3);
1133                        my $c = find_graft_path_commit($paths, $p1, $r);
1134                        next unless $c;
1135                        find_graft_path_parents($grafts, $paths, $c, $p0, $r0);
1136                }
1137        }
1138}
1139
1140sub graft_file_copy_lib {
1141        my ($grafts, $l_map, $u) = @_;
1142        my $tree_paths = $l_map->{$u};
1143        my $pfx = common_prefix([keys %$tree_paths]);
1144        my ($repo, $path) = repo_path_split($u.$pfx);
1145        $SVN = libsvn_connect($repo);
1146
1147        my ($base, $head) = libsvn_parse_revision();
1148        my $inc = 1000;
1149        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
1150        my $eh = $SVN::Error::handler;
1151        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
1152        while (1) {
1153                my $pool = SVN::Pool->new;
1154                libsvn_get_log(libsvn_dup_ra($SVN), [$path],
1155                               $min, $max, 0, 1, 1,
1156                        sub {
1157                                libsvn_graft_file_copies($grafts, $tree_paths,
1158                                                        $path, @_);
1159                        }, $pool);
1160                $pool->clear;
1161                last if ($max >= $head);
1162                $min = $max + 1;
1163                $max += $inc;
1164                $max = $head if ($max > $head);
1165        }
1166        $SVN::Error::handler = $eh;
1167}
1168
1169sub process_merge_msg_matches {
1170        my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
1171        my (@strong, @weak);
1172        foreach (@matches) {
1173                # merging with ourselves is not interesting
1174                next if $_ eq $p;
1175                if ($l_map->{$u}->{$_}) {
1176                        push @strong, $_;
1177                } else {
1178                        push @weak, $_;
1179                }
1180        }
1181        foreach my $w (@weak) {
1182                last if @strong;
1183                # no exact match, use branch name as regexp.
1184                my $re = qr/\Q$w\E/i;
1185                foreach (keys %{$l_map->{$u}}) {
1186                        if (/$re/) {
1187                                push @strong, $l_map->{$u}->{$_};
1188                                last;
1189                        }
1190                }
1191                last if @strong;
1192                $w = basename($w);
1193                $re = qr/\Q$w\E/i;
1194                foreach (keys %{$l_map->{$u}}) {
1195                        if (/$re/) {
1196                                push @strong, $l_map->{$u}->{$_};
1197                                last;
1198                        }
1199                }
1200        }
1201        my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
1202                                        \s(?:[a-f\d\-]+)$/xsm);
1203        unless (defined $rev) {
1204                ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
1205                                        \@(?:[a-f\d\-]+)/xsm);
1206                return unless defined $rev;
1207        }
1208        foreach my $m (@strong) {
1209                my ($r0, $s0) = find_rev_before($rev, $m, 1);
1210                $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
1211        }
1212}
1213
1214sub graft_merge_msg {
1215        my ($grafts, $l_map, $u, $p, @re) = @_;
1216
1217        my $x = $l_map->{$u}->{$p};
1218        my $rl = rev_list_raw($x);
1219        while (my $c = next_rev_list_entry($rl)) {
1220                foreach my $re (@re) {
1221                        my (@br) = ($c->{m} =~ /$re/g);
1222                        next unless @br;
1223                        process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
1224                }
1225        }
1226}
1227
1228sub read_uuid {
1229        return if $SVN_UUID;
1230        if ($_use_lib) {
1231                my $pool = SVN::Pool->new;
1232                $SVN_UUID = $SVN->get_uuid($pool);
1233                $pool->clear;
1234        } else {
1235                my $info = shift || svn_info('.');
1236                $SVN_UUID = $info->{'Repository UUID'} or
1237                                        croak "Repository UUID unreadable\n";
1238        }
1239}
1240
1241sub quiet_run {
1242        my $pid = fork;
1243        defined $pid or croak $!;
1244        if (!$pid) {
1245                open my $null, '>', '/dev/null' or croak $!;
1246                open STDERR, '>&', $null or croak $!;
1247                open STDOUT, '>&', $null or croak $!;
1248                exec @_ or croak $!;
1249        }
1250        waitpid $pid, 0;
1251        return $?;
1252}
1253
1254sub repo_path_split {
1255        my $full_url = shift;
1256        $full_url =~ s#/+$##;
1257
1258        foreach (@repo_path_split_cache) {
1259                if ($full_url =~ s#$_##) {
1260                        my $u = $1;
1261                        $full_url =~ s#^/+##;
1262                        return ($u, $full_url);
1263                }
1264        }
1265        if ($_use_lib) {
1266                my $tmp = libsvn_connect($full_url);
1267                return ($tmp->{repos_root}, $tmp->{svn_path});
1268        } else {
1269                my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i);
1270                $path =~ s#^/+##;
1271                my @paths = split(m#/+#, $path);
1272                while (quiet_run(qw/svn ls --non-interactive/, $url)) {
1273                        my $n = shift @paths || last;
1274                        $url .= "/$n";
1275                }
1276                push @repo_path_split_cache, qr/^(\Q$url\E)/;
1277                $path = join('/',@paths);
1278                return ($url, $path);
1279        }
1280}
1281
1282sub setup_git_svn {
1283        defined $SVN_URL or croak "SVN repository location required\n";
1284        unless (-d $GIT_DIR) {
1285                croak "GIT_DIR=$GIT_DIR does not exist!\n";
1286        }
1287        mkpath([$GIT_SVN_DIR]);
1288        mkpath(["$GIT_SVN_DIR/info"]);
1289        open my $fh, '>>',$REVDB or croak $!;
1290        close $fh;
1291        s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1292
1293}
1294
1295sub assert_svn_wc_clean {
1296        return if $_use_lib;
1297        my ($svn_rev) = @_;
1298        croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
1299        my $lcr = svn_info('.')->{'Last Changed Rev'};
1300        if ($svn_rev != $lcr) {
1301                print STDERR "Checking for copy-tree ... ";
1302                my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
1303                                                "-r$lcr:$svn_rev")));
1304                if (@diff) {
1305                        croak "Nope!  Expected r$svn_rev, got r$lcr\n";
1306                } else {
1307                        print STDERR "OK!\n";
1308                }
1309        }
1310        my @status = grep(!/^Performing status on external/,(`svn status`));
1311        @status = grep(!/^\s*$/,@status);
1312        @status = grep(!/^X/,@status) if $_no_ignore_ext;
1313        if (scalar @status) {
1314                print STDERR "Tree ($SVN_WC) is not clean:\n";
1315                print STDERR $_ foreach @status;
1316                croak;
1317        }
1318}
1319
1320sub get_tree_from_treeish {
1321        my ($treeish) = @_;
1322        croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1323        chomp(my $type = `git-cat-file -t $treeish`);
1324        my $expected;
1325        while ($type eq 'tag') {
1326                chomp(($treeish, $type) = `git-cat-file tag $treeish`);
1327        }
1328        if ($type eq 'commit') {
1329                $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
1330                ($expected) = ($expected =~ /^tree ($sha1)$/);
1331                die "Unable to get tree from $treeish\n" unless $expected;
1332        } elsif ($type eq 'tree') {
1333                $expected = $treeish;
1334        } else {
1335                die "$treeish is a $type, expected tree, tag or commit\n";
1336        }
1337        return $expected;
1338}
1339
1340sub assert_tree {
1341        return if $_use_lib;
1342        my ($treeish) = @_;
1343        my $expected = get_tree_from_treeish($treeish);
1344
1345        my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
1346        if (-e $tmpindex) {
1347                unlink $tmpindex or croak $!;
1348        }
1349        my $old_index = set_index($tmpindex);
1350        index_changes(1);
1351        chomp(my $tree = `git-write-tree`);
1352        restore_index($old_index);
1353        if ($tree ne $expected) {
1354                croak "Tree mismatch, Got: $tree, Expected: $expected\n";
1355        }
1356        unlink $tmpindex;
1357}
1358
1359sub parse_diff_tree {
1360        my $diff_fh = shift;
1361        local $/ = "\0";
1362        my $state = 'meta';
1363        my @mods;
1364        while (<$diff_fh>) {
1365                chomp $_; # this gets rid of the trailing "\0"
1366                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1367                                        $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1368                        push @mods, {   mode_a => $1, mode_b => $2,
1369                                        sha1_b => $3, chg => $4 };
1370                        if ($4 =~ /^(?:C|R)$/) {
1371                                $state = 'file_a';
1372                        } else {
1373                                $state = 'file_b';
1374                        }
1375                } elsif ($state eq 'file_a') {
1376                        my $x = $mods[$#mods] or croak "Empty array\n";
1377                        if ($x->{chg} !~ /^(?:C|R)$/) {
1378                                croak "Error parsing $_, $x->{chg}\n";
1379                        }
1380                        $x->{file_a} = $_;
1381                        $state = 'file_b';
1382                } elsif ($state eq 'file_b') {
1383                        my $x = $mods[$#mods] or croak "Empty array\n";
1384                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1385                                croak "Error parsing $_, $x->{chg}\n";
1386                        }
1387                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1388                                croak "Error parsing $_, $x->{chg}\n";
1389                        }
1390                        $x->{file_b} = $_;
1391                        $state = 'meta';
1392                } else {
1393                        croak "Error parsing $_\n";
1394                }
1395        }
1396        close $diff_fh or croak $?;
1397
1398        return \@mods;
1399}
1400
1401sub svn_check_prop_executable {
1402        my $m = shift;
1403        return if -l $m->{file_b};
1404        if ($m->{mode_b} =~ /755$/) {
1405                chmod((0755 &~ umask),$m->{file_b}) or croak $!;
1406                if ($m->{mode_a} !~ /755$/) {
1407                        sys(qw(svn propset svn:executable 1), $m->{file_b});
1408                }
1409                -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
1410        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
1411                sys(qw(svn propdel svn:executable), $m->{file_b});
1412                chmod((0644 &~ umask),$m->{file_b}) or croak $!;
1413                -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
1414        }
1415}
1416
1417sub svn_ensure_parent_path {
1418        my $dir_b = dirname(shift);
1419        svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
1420        mkpath([$dir_b]) unless (-d $dir_b);
1421        sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
1422}
1423
1424sub precommit_check {
1425        my $mods = shift;
1426        my (%rm_file, %rmdir_check, %added_check);
1427
1428        my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
1429        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1430                if ($m->{chg} eq 'R') {
1431                        if (-d $m->{file_b}) {
1432                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1433                        }
1434                        # dir/$file => dir/file/$file
1435                        my $dirname = dirname($m->{file_b});
1436                        while ($dirname ne File::Spec->curdir) {
1437                                if ($dirname ne $m->{file_a}) {
1438                                        $dirname = dirname($dirname);
1439                                        next;
1440                                }
1441                                err_file_to_dir("$m->{file_a} => $m->{file_b}");
1442                        }
1443                        # baz/zzz => baz (baz is a file)
1444                        $dirname = dirname($m->{file_a});
1445                        while ($dirname ne File::Spec->curdir) {
1446                                if ($dirname ne $m->{file_b}) {
1447                                        $dirname = dirname($dirname);
1448                                        next;
1449                                }
1450                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1451                        }
1452                }
1453                if ($m->{chg} =~ /^(D|R)$/) {
1454                        my $t = $1 eq 'D' ? 'file_b' : 'file_a';
1455                        $rm_file{ $m->{$t} } = 1;
1456                        my $dirname = dirname( $m->{$t} );
1457                        my $basename = basename( $m->{$t} );
1458                        $rmdir_check{$dirname}->{$basename} = 1;
1459                } elsif ($m->{chg} =~ /^(?:A|C)$/) {
1460                        if (-d $m->{file_b}) {
1461                                err_dir_to_file($m->{file_b});
1462                        }
1463                        my $dirname = dirname( $m->{file_b} );
1464                        my $basename = basename( $m->{file_b} );
1465                        $added_check{$dirname}->{$basename} = 1;
1466                        while ($dirname ne File::Spec->curdir) {
1467                                if ($rm_file{$dirname}) {
1468                                        err_file_to_dir($m->{file_b});
1469                                }
1470                                $dirname = dirname $dirname;
1471                        }
1472                }
1473        }
1474        return (\%rmdir_check, \%added_check);
1475
1476        sub err_dir_to_file {
1477                my $file = shift;
1478                print STDERR "Node change from directory to file ",
1479                                "is not supported by Subversion: ",$file,"\n";
1480                exit 1;
1481        }
1482        sub err_file_to_dir {
1483                my $file = shift;
1484                print STDERR "Node change from file to directory ",
1485                                "is not supported by Subversion: ",$file,"\n";
1486                exit 1;
1487        }
1488}
1489
1490
1491sub get_diff {
1492        my ($from, $treeish) = @_;
1493        assert_tree($from);
1494        print "diff-tree $from $treeish\n";
1495        my $pid = open my $diff_fh, '-|';
1496        defined $pid or croak $!;
1497        if ($pid == 0) {
1498                my @diff_tree = qw(git-diff-tree -z -r);
1499                if ($_cp_similarity) {
1500                        push @diff_tree, "-C$_cp_similarity";
1501                } else {
1502                        push @diff_tree, '-C';
1503                }
1504                push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1505                push @diff_tree, "-l$_l" if defined $_l;
1506                exec(@diff_tree, $from, $treeish) or croak $!;
1507        }
1508        return parse_diff_tree($diff_fh);
1509}
1510
1511sub svn_checkout_tree {
1512        my ($from, $treeish) = @_;
1513        my $mods = get_diff($from->{commit}, $treeish);
1514        return $mods unless (scalar @$mods);
1515        my ($rm, $add) = precommit_check($mods);
1516
1517        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1518        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1519                if ($m->{chg} eq 'C') {
1520                        svn_ensure_parent_path( $m->{file_b} );
1521                        sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
1522                        apply_mod_line_blob($m);
1523                        svn_check_prop_executable($m);
1524                } elsif ($m->{chg} eq 'D') {
1525                        sys(qw(svn rm --force), $m->{file_b});
1526                } elsif ($m->{chg} eq 'R') {
1527                        svn_ensure_parent_path( $m->{file_b} );
1528                        sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
1529                        apply_mod_line_blob($m);
1530                        svn_check_prop_executable($m);
1531                } elsif ($m->{chg} eq 'M') {
1532                        apply_mod_line_blob($m);
1533                        svn_check_prop_executable($m);
1534                } elsif ($m->{chg} eq 'T') {
1535                        svn_check_prop_executable($m);
1536                        apply_mod_line_blob($m);
1537                        if ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
1538                                sys(qw(svn propdel svn:special), $m->{file_b});
1539                        } else {
1540                                sys(qw(svn propset svn:special *),$m->{file_b});
1541                        }
1542                } elsif ($m->{chg} eq 'A') {
1543                        svn_ensure_parent_path( $m->{file_b} );
1544                        apply_mod_line_blob($m);
1545                        sys(qw(svn add), $m->{file_b});
1546                        svn_check_prop_executable($m);
1547                } else {
1548                        croak "Invalid chg: $m->{chg}\n";
1549                }
1550        }
1551
1552        assert_tree($treeish);
1553        if ($_rmdir) { # remove empty directories
1554                handle_rmdir($rm, $add);
1555        }
1556        assert_tree($treeish);
1557        return $mods;
1558}
1559
1560sub libsvn_checkout_tree {
1561        my ($from, $treeish, $ed) = @_;
1562        my $mods = get_diff($from, $treeish);
1563        return $mods unless (scalar @$mods);
1564        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1565        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1566                my $f = $m->{chg};
1567                if (defined $o{$f}) {
1568                        $ed->$f($m, $_q);
1569                } else {
1570                        croak "Invalid change type: $f\n";
1571                }
1572        }
1573        $ed->rmdirs($_q) if $_rmdir;
1574        return $mods;
1575}
1576
1577# svn ls doesn't work with respect to the current working tree, but what's
1578# in the repository.  There's not even an option for it... *sigh*
1579# (added files don't show up and removed files remain in the ls listing)
1580sub svn_ls_current {
1581        my ($dir, $rm, $add) = @_;
1582        chomp(my @ls = safe_qx('svn','ls',$dir));
1583        my @ret = ();
1584        foreach (@ls) {
1585                s#/$##; # trailing slashes are evil
1586                push @ret, $_ unless $rm->{$dir}->{$_};
1587        }
1588        if (exists $add->{$dir}) {
1589                push @ret, keys %{$add->{$dir}};
1590        }
1591        return \@ret;
1592}
1593
1594sub handle_rmdir {
1595        my ($rm, $add) = @_;
1596
1597        foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
1598                my $ls = svn_ls_current($dir, $rm, $add);
1599                next if (scalar @$ls);
1600                sys(qw(svn rm --force),$dir);
1601
1602                my $dn = dirname $dir;
1603                $rm->{ $dn }->{ basename $dir } = 1;
1604                $ls = svn_ls_current($dn, $rm, $add);
1605                while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
1606                        sys(qw(svn rm --force),$dn);
1607                        $dir = basename $dn;
1608                        $dn = dirname $dn;
1609                        $rm->{ $dn }->{ $dir } = 1;
1610                        $ls = svn_ls_current($dn, $rm, $add);
1611                }
1612        }
1613}
1614
1615sub get_commit_message {
1616        my ($commit, $commit_msg) = (@_);
1617        my %log_msg = ( msg => '' );
1618        open my $msg, '>', $commit_msg or croak $!;
1619
1620        chomp(my $type = `git-cat-file -t $commit`);
1621        if ($type eq 'commit' || $type eq 'tag') {
1622                my $pid = open my $msg_fh, '-|';
1623                defined $pid or croak $!;
1624
1625                if ($pid == 0) {
1626                        exec('git-cat-file', $type, $commit) or croak $!;
1627                }
1628                my $in_msg = 0;
1629                while (<$msg_fh>) {
1630                        if (!$in_msg) {
1631                                $in_msg = 1 if (/^\s*$/);
1632                        } elsif (/^git-svn-id: /) {
1633                                # skip this, we regenerate the correct one
1634                                # on re-fetch anyways
1635                        } else {
1636                                print $msg $_ or croak $!;
1637                        }
1638                }
1639                close $msg_fh or croak $?;
1640        }
1641        close $msg or croak $!;
1642
1643        if ($_edit || ($type eq 'tree')) {
1644                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1645                system($editor, $commit_msg);
1646        }
1647
1648        # file_to_s removes all trailing newlines, so just use chomp() here:
1649        open $msg, '<', $commit_msg or croak $!;
1650        { local $/; chomp($log_msg{msg} = <$msg>); }
1651        close $msg or croak $!;
1652
1653        return \%log_msg;
1654}
1655
1656sub set_svn_commit_env {
1657        if (defined $LC_ALL) {
1658                $ENV{LC_ALL} = $LC_ALL;
1659        } else {
1660                delete $ENV{LC_ALL};
1661        }
1662}
1663
1664sub svn_commit_tree {
1665        my ($last, $commit) = @_;
1666        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
1667        my $log_msg = get_commit_message($commit, $commit_msg);
1668        my ($oneline) = ($log_msg->{msg} =~ /([^\n\r]+)/);
1669        print "Committing $commit: $oneline\n";
1670
1671        set_svn_commit_env();
1672        my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
1673        $ENV{LC_ALL} = 'C';
1674        unlink $commit_msg;
1675        my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/);
1676        if (!defined $committed) {
1677                my $out = join("\n",@ci_output);
1678                print STDERR "W: Trouble parsing \`svn commit' output:\n\n",
1679                                $out, "\n\nAssuming English locale...";
1680                ($committed) = ($out =~ /^Committed revision \d+\./sm);
1681                defined $committed or die " FAILED!\n",
1682                        "Commit output failed to parse committed revision!\n",
1683                print STDERR " OK\n";
1684        }
1685
1686        my @svn_up = qw(svn up);
1687        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
1688        if ($_optimize_commits && ($committed == ($last->{revision} + 1))) {
1689                push @svn_up, "-r$committed";
1690                sys(@svn_up);
1691                my $info = svn_info('.');
1692                my $date = $info->{'Last Changed Date'} or die "Missing date\n";
1693                if ($info->{'Last Changed Rev'} != $committed) {
1694                        croak "$info->{'Last Changed Rev'} != $committed\n"
1695                }
1696                my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1697                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1698                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1699                                         or croak "Failed to parse date: $date\n";
1700                $log_msg->{date} = "$tz $Y-$m-$d $H:$M:$S";
1701                $log_msg->{author} = $info->{'Last Changed Author'};
1702                $log_msg->{revision} = $committed;
1703                $log_msg->{msg} .= "\n";
1704                $log_msg->{parents} = [ $last->{commit} ];
1705                $log_msg->{commit} = git_commit($log_msg, $commit);
1706                return $log_msg;
1707        }
1708        # resync immediately
1709        push @svn_up, "-r$last->{revision}";
1710        sys(@svn_up);
1711        return fetch("$committed=$commit");
1712}
1713
1714sub rev_list_raw {
1715        my (@args) = @_;
1716        my $pid = open my $fh, '-|';
1717        defined $pid or croak $!;
1718        if (!$pid) {
1719                exec(qw/git-rev-list --pretty=raw/, @args) or croak $!;
1720        }
1721        return { fh => $fh, t => { } };
1722}
1723
1724sub next_rev_list_entry {
1725        my $rl = shift;
1726        my $fh = $rl->{fh};
1727        my $x = $rl->{t};
1728        while (<$fh>) {
1729                if (/^commit ($sha1)$/o) {
1730                        if ($x->{c}) {
1731                                $rl->{t} = { c => $1 };
1732                                return $x;
1733                        } else {
1734                                $x->{c} = $1;
1735                        }
1736                } elsif (/^parent ($sha1)$/o) {
1737                        $x->{p}->{$1} = 1;
1738                } elsif (s/^    //) {
1739                        $x->{m} ||= '';
1740                        $x->{m} .= $_;
1741                }
1742        }
1743        return ($x != $rl->{t}) ? $x : undef;
1744}
1745
1746# read the entire log into a temporary file (which is removed ASAP)
1747# and store the file handle + parser state
1748sub svn_log_raw {
1749        my (@log_args) = @_;
1750        my $log_fh = IO::File->new_tmpfile or croak $!;
1751        my $pid = fork;
1752        defined $pid or croak $!;
1753        if (!$pid) {
1754                open STDOUT, '>&', $log_fh or croak $!;
1755                exec (qw(svn log), @log_args) or croak $!
1756        }
1757        waitpid $pid, 0;
1758        croak $? if $?;
1759        seek $log_fh, 0, 0 or croak $!;
1760        return { state => 'sep', fh => $log_fh };
1761}
1762
1763sub next_log_entry {
1764        my $log = shift; # retval of svn_log_raw()
1765        my $ret = undef;
1766        my $fh = $log->{fh};
1767
1768        while (<$fh>) {
1769                chomp;
1770                if (/^\-{72}$/) {
1771                        if ($log->{state} eq 'msg') {
1772                                if ($ret->{lines}) {
1773                                        $ret->{msg} .= $_."\n";
1774                                        unless(--$ret->{lines}) {
1775                                                $log->{state} = 'sep';
1776                                        }
1777                                } else {
1778                                        croak "Log parse error at: $_\n",
1779                                                $ret->{revision},
1780                                                "\n";
1781                                }
1782                                next;
1783                        }
1784                        if ($log->{state} ne 'sep') {
1785                                croak "Log parse error at: $_\n",
1786                                        "state: $log->{state}\n",
1787                                        $ret->{revision},
1788                                        "\n";
1789                        }
1790                        $log->{state} = 'rev';
1791
1792                        # if we have an empty log message, put something there:
1793                        if ($ret) {
1794                                $ret->{msg} ||= "\n";
1795                                delete $ret->{lines};
1796                                return $ret;
1797                        }
1798                        next;
1799                }
1800                if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
1801                        my $rev = $1;
1802                        my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
1803                        ($lines) = ($lines =~ /(\d+)/);
1804                        $date = '1970-01-01 00:00:00 +0000'
1805                                if ($_ignore_nodate && $date eq '(no date)');
1806                        my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1807                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1808                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1809                                         or croak "Failed to parse date: $date\n";
1810                        $ret = {        revision => $rev,
1811                                        date => "$tz $Y-$m-$d $H:$M:$S",
1812                                        author => $author,
1813                                        lines => $lines,
1814                                        msg => '' };
1815                        if (defined $_authors && ! defined $users{$author}) {
1816                                die "Author: $author not defined in ",
1817                                                "$_authors file\n";
1818                        }
1819                        $log->{state} = 'msg_start';
1820                        next;
1821                }
1822                # skip the first blank line of the message:
1823                if ($log->{state} eq 'msg_start' && /^$/) {
1824                        $log->{state} = 'msg';
1825                } elsif ($log->{state} eq 'msg') {
1826                        if ($ret->{lines}) {
1827                                $ret->{msg} .= $_."\n";
1828                                unless (--$ret->{lines}) {
1829                                        $log->{state} = 'sep';
1830                                }
1831                        } else {
1832                                croak "Log parse error at: $_\n",
1833                                        $ret->{revision},"\n";
1834                        }
1835                }
1836        }
1837        return $ret;
1838}
1839
1840sub svn_info {
1841        my $url = shift || $SVN_URL;
1842
1843        my $pid = open my $info_fh, '-|';
1844        defined $pid or croak $!;
1845
1846        if ($pid == 0) {
1847                exec(qw(svn info),$url) or croak $!;
1848        }
1849
1850        my $ret = {};
1851        # only single-lines seem to exist in svn info output
1852        while (<$info_fh>) {
1853                chomp $_;
1854                if (m#^([^:]+)\s*:\s*(\S.*)$#) {
1855                        $ret->{$1} = $2;
1856                        push @{$ret->{-order}}, $1;
1857                }
1858        }
1859        close $info_fh or croak $?;
1860        return $ret;
1861}
1862
1863sub sys { system(@_) == 0 or croak $? }
1864
1865sub do_update_index {
1866        my ($z_cmd, $cmd, $no_text_base) = @_;
1867
1868        my $z = open my $p, '-|';
1869        defined $z or croak $!;
1870        unless ($z) { exec @$z_cmd or croak $! }
1871
1872        my $pid = open my $ui, '|-';
1873        defined $pid or croak $!;
1874        unless ($pid) {
1875                exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!;
1876        }
1877        local $/ = "\0";
1878        while (my $x = <$p>) {
1879                chomp $x;
1880                if (!$no_text_base && lstat $x && ! -l _ &&
1881                                svn_propget_base('svn:keywords', $x)) {
1882                        my $mode = -x _ ? 0755 : 0644;
1883                        my ($v,$d,$f) = File::Spec->splitpath($x);
1884                        my $tb = File::Spec->catfile($d, '.svn', 'tmp',
1885                                                'text-base',"$f.svn-base");
1886                        $tb =~ s#^/##;
1887                        unless (-f $tb) {
1888                                $tb = File::Spec->catfile($d, '.svn',
1889                                                'text-base',"$f.svn-base");
1890                                $tb =~ s#^/##;
1891                        }
1892                        my @s = stat($x);
1893                        unlink $x or croak $!;
1894                        copy($tb, $x);
1895                        chmod(($mode &~ umask), $x) or croak $!;
1896                        utime $s[8], $s[9], $x;
1897                }
1898                print $ui $x,"\0";
1899        }
1900        close $ui or croak $?;
1901}
1902
1903sub index_changes {
1904        return if $_use_lib;
1905
1906        if (!-f "$GIT_SVN_DIR/info/exclude") {
1907                open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!;
1908                print $fd '.svn',"\n";
1909                close $fd or croak $!;
1910        }
1911        my $no_text_base = shift;
1912        do_update_index([qw/git-diff-files --name-only -z/],
1913                        'remove',
1914                        $no_text_base);
1915        do_update_index([qw/git-ls-files -z --others/,
1916                                "--exclude-from=$GIT_SVN_DIR/info/exclude"],
1917                        'add',
1918                        $no_text_base);
1919}
1920
1921sub s_to_file {
1922        my ($str, $file, $mode) = @_;
1923        open my $fd,'>',$file or croak $!;
1924        print $fd $str,"\n" or croak $!;
1925        close $fd or croak $!;
1926        chmod ($mode &~ umask, $file) if (defined $mode);
1927}
1928
1929sub file_to_s {
1930        my $file = shift;
1931        open my $fd,'<',$file or croak "$!: file: $file\n";
1932        local $/;
1933        my $ret = <$fd>;
1934        close $fd or croak $!;
1935        $ret =~ s/\s*$//s;
1936        return $ret;
1937}
1938
1939sub assert_revision_unknown {
1940        my $r = shift;
1941        if (my $c = revdb_get($REVDB, $r)) {
1942                croak "$r = $c already exists! Why are we refetching it?";
1943        }
1944}
1945
1946sub trees_eq {
1947        my ($x, $y) = @_;
1948        my @x = safe_qx('git-cat-file','commit',$x);
1949        my @y = safe_qx('git-cat-file','commit',$y);
1950        if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
1951                                || $y[0] !~ /^tree $sha1\n$/) {
1952                print STDERR "Trees not equal: $y[0] != $x[0]\n";
1953                return 0
1954        }
1955        return 1;
1956}
1957
1958sub git_commit {
1959        my ($log_msg, @parents) = @_;
1960        assert_revision_unknown($log_msg->{revision});
1961        map_tree_joins() if (@_branch_from && !%tree_map);
1962
1963        my (@tmp_parents, @exec_parents, %seen_parent);
1964        if (my $lparents = $log_msg->{parents}) {
1965                @tmp_parents = @$lparents
1966        }
1967        # commit parents can be conditionally bound to a particular
1968        # svn revision via: "svn_revno=commit_sha1", filter them out here:
1969        foreach my $p (@parents) {
1970                next unless defined $p;
1971                if ($p =~ /^(\d+)=($sha1_short)$/o) {
1972                        if ($1 == $log_msg->{revision}) {
1973                                push @tmp_parents, $2;
1974                        }
1975                } else {
1976                        push @tmp_parents, $p if $p =~ /$sha1_short/o;
1977                }
1978        }
1979        my $tree = $log_msg->{tree};
1980        if (!defined $tree) {
1981                my $index = set_index($GIT_SVN_INDEX);
1982                index_changes();
1983                chomp($tree = `git-write-tree`);
1984                croak $? if $?;
1985                restore_index($index);
1986        }
1987
1988        # just in case we clobber the existing ref, we still want that ref
1989        # as our parent:
1990        if (my $cur = eval { file_to_s("$GIT_DIR/refs/remotes/$GIT_SVN") }) {
1991                push @tmp_parents, $cur;
1992        }
1993
1994        if (exists $tree_map{$tree}) {
1995                foreach my $p (@{$tree_map{$tree}}) {
1996                        my $skip;
1997                        foreach (@tmp_parents) {
1998                                # see if a common parent is found
1999                                my $mb = eval {
2000                                        safe_qx('git-merge-base', $_, $p)
2001                                };
2002                                next if ($@ || $?);
2003                                $skip = 1;
2004                                last;
2005                        }
2006                        next if $skip;
2007                        my ($url_p, $r_p, $uuid_p) = cmt_metadata($p);
2008                        next if (($SVN_UUID eq $uuid_p) &&
2009                                                ($log_msg->{revision} > $r_p));
2010                        next if (defined $url_p && defined $SVN_URL &&
2011                                                ($SVN_UUID eq $uuid_p) &&
2012                                                ($url_p eq $SVN_URL));
2013                        push @tmp_parents, $p;
2014                }
2015        }
2016        foreach (@tmp_parents) {
2017                next if $seen_parent{$_};
2018                $seen_parent{$_} = 1;
2019                push @exec_parents, $_;
2020                # MAXPARENT is defined to 16 in commit-tree.c:
2021                last if @exec_parents > 16;
2022        }
2023
2024        set_commit_env($log_msg);
2025        my @exec = ('git-commit-tree', $tree);
2026        push @exec, '-p', $_  foreach @exec_parents;
2027        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2028                                                                or croak $!;
2029        print $msg_fh $log_msg->{msg} or croak $!;
2030        unless ($_no_metadata) {
2031                print $msg_fh "\ngit-svn-id: $SVN_URL\@$log_msg->{revision}",
2032                                        " $SVN_UUID\n" or croak $!;
2033        }
2034        $msg_fh->flush == 0 or croak $!;
2035        close $msg_fh or croak $!;
2036        chomp(my $commit = do { local $/; <$out_fh> });
2037        close $out_fh or croak $!;
2038        waitpid $pid, 0;
2039        croak $? if $?;
2040        if ($commit !~ /^$sha1$/o) {
2041                die "Failed to commit, invalid sha1: $commit\n";
2042        }
2043        sys('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
2044        revdb_set($REVDB, $log_msg->{revision}, $commit);
2045
2046        # this output is read via pipe, do not change:
2047        print "r$log_msg->{revision} = $commit\n";
2048        check_repack();
2049        return $commit;
2050}
2051
2052sub check_repack {
2053        if ($_repack && (--$_repack_nr == 0)) {
2054                $_repack_nr = $_repack;
2055                sys("git repack $_repack_flags");
2056        }
2057}
2058
2059sub set_commit_env {
2060        my ($log_msg) = @_;
2061        my $author = $log_msg->{author};
2062        if (!defined $author || length $author == 0) {
2063                $author = '(no author)';
2064        }
2065        my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
2066                                : ($author,"$author\@$SVN_UUID");
2067        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
2068        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
2069        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
2070}
2071
2072sub apply_mod_line_blob {
2073        my $m = shift;
2074        if ($m->{mode_b} =~ /^120/) {
2075                blob_to_symlink($m->{sha1_b}, $m->{file_b});
2076        } else {
2077                blob_to_file($m->{sha1_b}, $m->{file_b});
2078        }
2079}
2080
2081sub blob_to_symlink {
2082        my ($blob, $link) = @_;
2083        defined $link or croak "\$link not defined!\n";
2084        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2085        if (-l $link || -f _) {
2086                unlink $link or croak $!;
2087        }
2088
2089        my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
2090        symlink $dest, $link or croak $!;
2091}
2092
2093sub blob_to_file {
2094        my ($blob, $file) = @_;
2095        defined $file or croak "\$file not defined!\n";
2096        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2097        if (-l $file || -f _) {
2098                unlink $file or croak $!;
2099        }
2100
2101        open my $blob_fh, '>', $file or croak "$!: $file\n";
2102        my $pid = fork;
2103        defined $pid or croak $!;
2104
2105        if ($pid == 0) {
2106                open STDOUT, '>&', $blob_fh or croak $!;
2107                exec('git-cat-file','blob',$blob) or croak $!;
2108        }
2109        waitpid $pid, 0;
2110        croak $? if $?;
2111
2112        close $blob_fh or croak $!;
2113}
2114
2115sub safe_qx {
2116        my $pid = open my $child, '-|';
2117        defined $pid or croak $!;
2118        if ($pid == 0) {
2119                exec(@_) or croak $!;
2120        }
2121        my @ret = (<$child>);
2122        close $child or croak $?;
2123        die $? if $?; # just in case close didn't error out
2124        return wantarray ? @ret : join('',@ret);
2125}
2126
2127sub svn_compat_check {
2128        if ($_follow_parent) {
2129                print STDERR 'E: --follow-parent functionality is only ',
2130                                "available when SVN libraries are used\n";
2131                exit 1;
2132        }
2133        my @co_help = safe_qx(qw(svn co -h));
2134        unless (grep /ignore-externals/,@co_help) {
2135                print STDERR "W: Installed svn version does not support ",
2136                                "--ignore-externals\n";
2137                $_no_ignore_ext = 1;
2138        }
2139        if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
2140                $_svn_co_url_revs = 1;
2141        }
2142        if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) {
2143                $_svn_pg_peg_revs = 1;
2144        }
2145
2146        # I really, really hope nobody hits this...
2147        unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
2148                print STDERR <<'';
2149W: The installed svn version does not support the --stop-on-copy flag in
2150   the log command.
2151   Lets hope the directory you're tracking is not a branch or tag
2152   and was never moved within the repository...
2153
2154                $_no_stop_copy = 1;
2155        }
2156}
2157
2158# *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
2159# (and they won't honor URL@<rev> without -r<rev>, too!)
2160sub svn_cmd_checkout {
2161        my ($url, $rev, $dir) = @_;
2162        my @cmd = ('svn','co', "-r$rev");
2163        push @cmd, '--ignore-externals' unless $_no_ignore_ext;
2164        $url .= "\@$rev" if $_svn_co_url_revs;
2165        sys(@cmd, $url, $dir);
2166}
2167
2168sub check_upgrade_needed {
2169        if (!-r $REVDB) {
2170                -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2171                open my $fh, '>>',$REVDB or croak $!;
2172                close $fh;
2173        }
2174        my $old = eval {
2175                my $pid = open my $child, '-|';
2176                defined $pid or croak $!;
2177                if ($pid == 0) {
2178                        close STDERR;
2179                        exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $!;
2180                }
2181                my @ret = (<$child>);
2182                close $child or croak $?;
2183                die $? if $?; # just in case close didn't error out
2184                return wantarray ? @ret : join('',@ret);
2185        };
2186        return unless $old;
2187        my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
2188        if ($@ || !$head) {
2189                print STDERR "Please run: $0 rebuild --upgrade\n";
2190                exit 1;
2191        }
2192}
2193
2194# fills %tree_map with a reverse mapping of trees to commits.  Useful
2195# for finding parents to commit on.
2196sub map_tree_joins {
2197        my %seen;
2198        foreach my $br (@_branch_from) {
2199                my $pid = open my $pipe, '-|';
2200                defined $pid or croak $!;
2201                if ($pid == 0) {
2202                        exec(qw(git-rev-list --topo-order --pretty=raw), $br)
2203                                                                or croak $!;
2204                }
2205                while (<$pipe>) {
2206                        if (/^commit ($sha1)$/o) {
2207                                my $commit = $1;
2208
2209                                # if we've seen a commit,
2210                                # we've seen its parents
2211                                last if $seen{$commit};
2212                                my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
2213                                unless (defined $tree) {
2214                                        die "Failed to parse commit $commit\n";
2215                                }
2216                                push @{$tree_map{$tree}}, $commit;
2217                                $seen{$commit} = 1;
2218                        }
2219                }
2220                close $pipe; # we could be breaking the pipe early
2221        }
2222}
2223
2224sub load_all_refs {
2225        if (@_branch_from) {
2226                print STDERR '--branch|-b parameters are ignored when ',
2227                        "--branch-all-refs|-B is passed\n";
2228        }
2229
2230        # don't worry about rev-list on non-commit objects/tags,
2231        # it shouldn't blow up if a ref is a blob or tree...
2232        chomp(@_branch_from = `git-rev-parse --symbolic --all`);
2233}
2234
2235# '<svn username> = real-name <email address>' mapping based on git-svnimport:
2236sub load_authors {
2237        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2238        while (<$authors>) {
2239                chomp;
2240                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
2241                my ($user, $name, $email) = ($1, $2, $3);
2242                $users{$user} = [$name, $email];
2243        }
2244        close $authors or croak $!;
2245}
2246
2247sub rload_authors {
2248        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2249        while (<$authors>) {
2250                chomp;
2251                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2252                my ($user, $name, $email) = ($1, $2, $3);
2253                $rusers{"$name <$email>"} = $user;
2254        }
2255        close $authors or croak $!;
2256}
2257
2258sub svn_propget_base {
2259        my ($p, $f) = @_;
2260        $f .= '@BASE' if $_svn_pg_peg_revs;
2261        return safe_qx(qw/svn propget/, $p, $f);
2262}
2263
2264sub git_svn_each {
2265        my $sub = shift;
2266        foreach (`git-rev-parse --symbolic --all`) {
2267                next unless s#^refs/remotes/##;
2268                chomp $_;
2269                next unless -f "$GIT_DIR/svn/$_/info/url";
2270                &$sub($_);
2271        }
2272}
2273
2274sub migrate_revdb {
2275        git_svn_each(sub {
2276                my $id = shift;
2277                defined(my $pid = fork) or croak $!;
2278                if (!$pid) {
2279                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2280                        init_vars();
2281                        exit 0 if -r $REVDB;
2282                        print "Upgrading svn => git mapping...\n";
2283                        -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2284                        open my $fh, '>>',$REVDB or croak $!;
2285                        close $fh;
2286                        rebuild();
2287                        print "Done upgrading. You may now delete the ",
2288                                "deprecated $GIT_SVN_DIR/revs directory\n";
2289                        exit 0;
2290                }
2291                waitpid $pid, 0;
2292                croak $? if $?;
2293        });
2294}
2295
2296sub migration_check {
2297        migrate_revdb() unless (-e $REVDB);
2298        return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
2299        print "Upgrading repository...\n";
2300        unless (-d "$GIT_DIR/svn") {
2301                mkdir "$GIT_DIR/svn" or croak $!;
2302        }
2303        print "Data from a previous version of git-svn exists, but\n\t",
2304                                "$GIT_SVN_DIR\n\t(required for this version ",
2305                                "($VERSION) of git-svn) does not.\n";
2306
2307        foreach my $x (`git-rev-parse --symbolic --all`) {
2308                next unless $x =~ s#^refs/remotes/##;
2309                chomp $x;
2310                next unless -f "$GIT_DIR/$x/info/url";
2311                my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
2312                next unless $u;
2313                my $dn = dirname("$GIT_DIR/svn/$x");
2314                mkpath([$dn]) unless -d $dn;
2315                rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
2316        }
2317        migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
2318        print "Done upgrading.\n";
2319}
2320
2321sub find_rev_before {
2322        my ($r, $id, $eq_ok) = @_;
2323        my $f = "$GIT_DIR/svn/$id/.rev_db";
2324        return (undef,undef) unless -r $f;
2325        --$r unless $eq_ok;
2326        while ($r > 0) {
2327                if (my $c = revdb_get($f, $r)) {
2328                        return ($r, $c);
2329                }
2330                --$r;
2331        }
2332        return (undef, undef);
2333}
2334
2335sub init_vars {
2336        $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
2337        $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
2338        $REVDB = "$GIT_SVN_DIR/.rev_db";
2339        $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
2340        $SVN_URL = undef;
2341        $SVN_WC = "$GIT_SVN_DIR/tree";
2342        %tree_map = ();
2343}
2344
2345# convert GetOpt::Long specs for use by git-repo-config
2346sub read_repo_config {
2347        return unless -d $GIT_DIR;
2348        my $opts = shift;
2349        foreach my $o (keys %$opts) {
2350                my $v = $opts->{$o};
2351                my ($key) = ($o =~ /^([a-z\-]+)/);
2352                $key =~ s/-//g;
2353                my $arg = 'git-repo-config';
2354                $arg .= ' --int' if ($o =~ /[:=]i$/);
2355                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
2356                if (ref $v eq 'ARRAY') {
2357                        chomp(my @tmp = `$arg --get-all svn.$key`);
2358                        @$v = @tmp if @tmp;
2359                } else {
2360                        chomp(my $tmp = `$arg --get svn.$key`);
2361                        if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
2362                                $$v = $tmp;
2363                        }
2364                }
2365        }
2366}
2367
2368sub set_default_vals {
2369        if (defined $_repack) {
2370                $_repack = 1000 if ($_repack <= 0);
2371                $_repack_nr = $_repack;
2372                $_repack_flags ||= '-d';
2373        }
2374}
2375
2376sub read_grafts {
2377        my $gr_file = shift;
2378        my ($grafts, $comments) = ({}, {});
2379        if (open my $fh, '<', $gr_file) {
2380                my @tmp;
2381                while (<$fh>) {
2382                        if (/^($sha1)\s+/) {
2383                                my $c = $1;
2384                                if (@tmp) {
2385                                        @{$comments->{$c}} = @tmp;
2386                                        @tmp = ();
2387                                }
2388                                foreach my $p (split /\s+/, $_) {
2389                                        $grafts->{$c}->{$p} = 1;
2390                                }
2391                        } else {
2392                                push @tmp, $_;
2393                        }
2394                }
2395                close $fh or croak $!;
2396                @{$comments->{'END'}} = @tmp if @tmp;
2397        }
2398        return ($grafts, $comments);
2399}
2400
2401sub write_grafts {
2402        my ($grafts, $comments, $gr_file) = @_;
2403
2404        open my $fh, '>', $gr_file or croak $!;
2405        foreach my $c (sort keys %$grafts) {
2406                if ($comments->{$c}) {
2407                        print $fh $_ foreach @{$comments->{$c}};
2408                }
2409                my $p = $grafts->{$c};
2410                my %x; # real parents
2411                delete $p->{$c}; # commits are not self-reproducing...
2412                my $pid = open my $ch, '-|';
2413                defined $pid or croak $!;
2414                if (!$pid) {
2415                        exec(qw/git-cat-file commit/, $c) or croak $!;
2416                }
2417                while (<$ch>) {
2418                        if (/^parent ($sha1)/) {
2419                                $x{$1} = $p->{$1} = 1;
2420                        } else {
2421                                last unless /^\S/;
2422                        }
2423                }
2424                close $ch; # breaking the pipe
2425
2426                # if real parents are the only ones in the grafts, drop it
2427                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2428
2429                my (@ip, @jp, $mb);
2430                my %del = %x;
2431                @ip = @jp = keys %$p;
2432                foreach my $i (@ip) {
2433                        next if $del{$i} || $p->{$i} == 2;
2434                        foreach my $j (@jp) {
2435                                next if $i eq $j || $del{$j} || $p->{$j} == 2;
2436                                $mb = eval { safe_qx('git-merge-base',$i,$j) };
2437                                next unless $mb;
2438                                chomp $mb;
2439                                next if $x{$mb};
2440                                if ($mb eq $j) {
2441                                        delete $p->{$i};
2442                                        $del{$i} = 1;
2443                                } elsif ($mb eq $i) {
2444                                        delete $p->{$j};
2445                                        $del{$j} = 1;
2446                                }
2447                        }
2448                }
2449
2450                # if real parents are the only ones in the grafts, drop it
2451                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2452
2453                print $fh $c, ' ', join(' ', sort keys %$p),"\n";
2454        }
2455        if ($comments->{'END'}) {
2456                print $fh $_ foreach @{$comments->{'END'}};
2457        }
2458        close $fh or croak $!;
2459}
2460
2461sub read_url_paths_all {
2462        my ($l_map, $pfx, $p) = @_;
2463        my @dir;
2464        foreach (<$p/*>) {
2465                if (-r "$_/info/url") {
2466                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2467                        my $id = $pfx . basename $_;
2468                        my $url = file_to_s("$_/info/url");
2469                        my ($u, $p) = repo_path_split($url);
2470                        $l_map->{$u}->{$p} = $id;
2471                } elsif (-d $_) {
2472                        push @dir, $_;
2473                }
2474        }
2475        foreach (@dir) {
2476                my $x = $_;
2477                $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
2478                read_url_paths_all($l_map, $x, $_);
2479        }
2480}
2481
2482# this one only gets ids that have been imported, not new ones
2483sub read_url_paths {
2484        my $l_map = {};
2485        git_svn_each(sub { my $x = shift;
2486                        my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
2487                        my ($u, $p) = repo_path_split($url);
2488                        $l_map->{$u}->{$p} = $x;
2489                        });
2490        return $l_map;
2491}
2492
2493sub extract_metadata {
2494        my $id = shift or return (undef, undef, undef);
2495        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
2496                                                        \s([a-f\d\-]+)$/x);
2497        if (!defined $rev || !$uuid || !$url) {
2498                # some of the original repositories I made had
2499                # identifiers like this:
2500                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
2501        }
2502        return ($url, $rev, $uuid);
2503}
2504
2505sub cmt_metadata {
2506        return extract_metadata((grep(/^git-svn-id: /,
2507                safe_qx(qw/git-cat-file commit/, shift)))[-1]);
2508}
2509
2510sub get_commit_time {
2511        my $cmt = shift;
2512        defined(my $pid = open my $fh, '-|') or croak $!;
2513        if (!$pid) {
2514                exec qw/git-rev-list --pretty=raw -n1/, $cmt or croak $!;
2515        }
2516        while (<$fh>) {
2517                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
2518                my ($s, $tz) = ($1, $2);
2519                if ($tz =~ s/^\+//) {
2520                        $s += tz_to_s_offset($tz);
2521                } elsif ($tz =~ s/^\-//) {
2522                        $s -= tz_to_s_offset($tz);
2523                }
2524                close $fh;
2525                return $s;
2526        }
2527        die "Can't get commit time for commit: $cmt\n";
2528}
2529
2530sub tz_to_s_offset {
2531        my ($tz) = @_;
2532        $tz =~ s/(\d\d)$//;
2533        return ($1 * 60) + ($tz * 3600);
2534}
2535
2536sub setup_pager { # translated to Perl from pager.c
2537        return unless (-t *STDOUT);
2538        my $pager = $ENV{PAGER};
2539        if (!defined $pager) {
2540                $pager = 'less';
2541        } elsif (length $pager == 0 || $pager eq 'cat') {
2542                return;
2543        }
2544        pipe my $rfd, my $wfd or return;
2545        defined(my $pid = fork) or croak $!;
2546        if (!$pid) {
2547                open STDOUT, '>&', $wfd or croak $!;
2548                return;
2549        }
2550        open STDIN, '<&', $rfd or croak $!;
2551        $ENV{LESS} ||= '-S';
2552        exec $pager or croak "Can't run pager: $!\n";;
2553}
2554
2555sub get_author_info {
2556        my ($dest, $author, $t, $tz) = @_;
2557        $author =~ s/(?:^\s*|\s*$)//g;
2558        $dest->{a_raw} = $author;
2559        my $_a;
2560        if ($_authors) {
2561                $_a = $rusers{$author} || undef;
2562        }
2563        if (!$_a) {
2564                ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
2565        }
2566        $dest->{t} = $t;
2567        $dest->{tz} = $tz;
2568        $dest->{a} = $_a;
2569        # Date::Parse isn't in the standard Perl distro :(
2570        if ($tz =~ s/^\+//) {
2571                $t += tz_to_s_offset($tz);
2572        } elsif ($tz =~ s/^\-//) {
2573                $t -= tz_to_s_offset($tz);
2574        }
2575        $dest->{t_utc} = $t;
2576}
2577
2578sub process_commit {
2579        my ($c, $r_min, $r_max, $defer) = @_;
2580        if (defined $r_min && defined $r_max) {
2581                if ($r_min == $c->{r} && $r_min == $r_max) {
2582                        show_commit($c);
2583                        return 0;
2584                }
2585                return 1 if $r_min == $r_max;
2586                if ($r_min < $r_max) {
2587                        # we need to reverse the print order
2588                        return 0 if (defined $_limit && --$_limit < 0);
2589                        push @$defer, $c;
2590                        return 1;
2591                }
2592                if ($r_min != $r_max) {
2593                        return 1 if ($r_min < $c->{r});
2594                        return 1 if ($r_max > $c->{r});
2595                }
2596        }
2597        return 0 if (defined $_limit && --$_limit < 0);
2598        show_commit($c);
2599        return 1;
2600}
2601
2602sub show_commit {
2603        my $c = shift;
2604        if ($_oneline) {
2605                my $x = "\n";
2606                if (my $l = $c->{l}) {
2607                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2608                        $x = $l->[0];
2609                }
2610                $_l_fmt ||= 'A' . length($c->{r});
2611                print 'r',pack($_l_fmt, $c->{r}),' | ';
2612                print "$c->{c} | " if $_show_commit;
2613                print $x;
2614        } else {
2615                show_commit_normal($c);
2616        }
2617}
2618
2619sub show_commit_changed_paths {
2620        my ($c) = @_;
2621        return unless $c->{changed};
2622        print "Changed paths:\n", @{$c->{changed}};
2623}
2624
2625sub show_commit_normal {
2626        my ($c) = @_;
2627        print '-' x72, "\nr$c->{r} | ";
2628        print "$c->{c} | " if $_show_commit;
2629        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2630                                 localtime($c->{t_utc})), ' | ';
2631        my $nr_line = 0;
2632
2633        if (my $l = $c->{l}) {
2634                while ($l->[$#$l] eq "\n" && $#$l > 0
2635                                          && $l->[($#$l - 1)] eq "\n") {
2636                        pop @$l;
2637                }
2638                $nr_line = scalar @$l;
2639                if (!$nr_line) {
2640                        print "1 line\n\n\n";
2641                } else {
2642                        if ($nr_line == 1) {
2643                                $nr_line = '1 line';
2644                        } else {
2645                                $nr_line .= ' lines';
2646                        }
2647                        print $nr_line, "\n";
2648                        show_commit_changed_paths($c);
2649                        print "\n";
2650                        print $_ foreach @$l;
2651                }
2652        } else {
2653                print "1 line\n";
2654                show_commit_changed_paths($c);
2655                print "\n";
2656
2657        }
2658        foreach my $x (qw/raw diff/) {
2659                if ($c->{$x}) {
2660                        print "\n";
2661                        print $_ foreach @{$c->{$x}}
2662                }
2663        }
2664}
2665
2666sub libsvn_load {
2667        return unless $_use_lib;
2668        $_use_lib = eval {
2669                require SVN::Core;
2670                if ($SVN::Core::VERSION lt '1.1.0') {
2671                        die "Need SVN::Core 1.1.0 or better ",
2672                                        "(got $SVN::Core::VERSION) ",
2673                                        "Falling back to command-line svn\n";
2674                }
2675                require SVN::Ra;
2676                require SVN::Delta;
2677                push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
2678                my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2679                                        $SVN::Node::dir.$SVN::Node::unknown.
2680                                        $SVN::Node::none.$SVN::Node::file.
2681                                        $SVN::Node::dir.$SVN::Node::unknown.
2682                                        $SVN::Auth::SSL::CNMISMATCH.
2683                                        $SVN::Auth::SSL::NOTYETVALID.
2684                                        $SVN::Auth::SSL::EXPIRED.
2685                                        $SVN::Auth::SSL::UNKNOWNCA.
2686                                        $SVN::Auth::SSL::OTHER;
2687                1;
2688        };
2689}
2690
2691sub _simple_prompt {
2692        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2693        $may_save = undef if $_no_auth_cache;
2694        $default_username = $_username if defined $_username;
2695        if (defined $default_username && length $default_username) {
2696                if (defined $realm && length $realm) {
2697                        print "Authentication realm: $realm\n";
2698                }
2699                $cred->username($default_username);
2700        } else {
2701                _username_prompt($cred, $realm, $may_save, $pool);
2702        }
2703        $cred->password(_read_password("Password for '" .
2704                                       $cred->username . "': ", $realm));
2705        $cred->may_save($may_save);
2706        $SVN::_Core::SVN_NO_ERROR;
2707}
2708
2709sub _ssl_server_trust_prompt {
2710        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2711        $may_save = undef if $_no_auth_cache;
2712        print "Error validating server certificate for '$realm':\n";
2713        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2714                print " - The certificate is not issued by a trusted ",
2715                      "authority. Use the\n",
2716                      "   fingerprint to validate the certificate manually!\n";
2717        }
2718        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2719                print " - The certificate hostname does not match.\n";
2720        }
2721        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2722                print " - The certificate is not yet valid.\n";
2723        }
2724        if ($failures & $SVN::Auth::SSL::EXPIRED) {
2725                print " - The certificate has expired.\n";
2726        }
2727        if ($failures & $SVN::Auth::SSL::OTHER) {
2728                print " - The certificate has an unknown error.\n";
2729        }
2730        printf( "Certificate information:\n".
2731                " - Hostname: %s\n".
2732                " - Valid: from %s until %s\n".
2733                " - Issuer: %s\n".
2734                " - Fingerprint: %s\n",
2735                map $cert_info->$_, qw(hostname valid_from valid_until
2736                                       issuer_dname fingerprint) );
2737        my $choice;
2738prompt:
2739        print $may_save ?
2740              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2741              "(R)eject or accept (t)emporarily? ";
2742        $choice = lc(substr(<STDIN> || 'R', 0, 1));
2743        if ($choice =~ /^t$/i) {
2744                $cred->may_save(undef);
2745        } elsif ($choice =~ /^r$/i) {
2746                return -1;
2747        } elsif ($may_save && $choice =~ /^p$/i) {
2748                $cred->may_save($may_save);
2749        } else {
2750                goto prompt;
2751        }
2752        $cred->accepted_failures($failures);
2753        $SVN::_Core::SVN_NO_ERROR;
2754}
2755
2756sub _ssl_client_cert_prompt {
2757        my ($cred, $realm, $may_save, $pool) = @_;
2758        $may_save = undef if $_no_auth_cache;
2759        print "Client certificate filename: ";
2760        chomp(my $filename = <STDIN>);
2761        $cred->cert_file($filename);
2762        $cred->may_save($may_save);
2763        $SVN::_Core::SVN_NO_ERROR;
2764}
2765
2766sub _ssl_client_cert_pw_prompt {
2767        my ($cred, $realm, $may_save, $pool) = @_;
2768        $may_save = undef if $_no_auth_cache;
2769        $cred->password(_read_password("Password: ", $realm));
2770        $cred->may_save($may_save);
2771        $SVN::_Core::SVN_NO_ERROR;
2772}
2773
2774sub _username_prompt {
2775        my ($cred, $realm, $may_save, $pool) = @_;
2776        $may_save = undef if $_no_auth_cache;
2777        if (defined $realm && length $realm) {
2778                print "Authentication realm: $realm\n";
2779        }
2780        my $username;
2781        if (defined $_username) {
2782                $username = $_username;
2783        } else {
2784                print "Username: ";
2785                chomp($username = <STDIN>);
2786        }
2787        $cred->username($username);
2788        $cred->may_save($may_save);
2789        $SVN::_Core::SVN_NO_ERROR;
2790}
2791
2792sub _read_password {
2793        my ($prompt, $realm) = @_;
2794        print $prompt;
2795        require Term::ReadKey;
2796        Term::ReadKey::ReadMode('noecho');
2797        my $password = '';
2798        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2799                last if $key =~ /[\012\015]/; # \n\r
2800                $password .= $key;
2801        }
2802        Term::ReadKey::ReadMode('restore');
2803        print "\n";
2804        $password;
2805}
2806
2807sub libsvn_connect {
2808        my ($url) = @_;
2809        SVN::_Core::svn_config_ensure($_config_dir, undef);
2810        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2811            SVN::Client::get_simple_provider(),
2812            SVN::Client::get_ssl_server_trust_file_provider(),
2813            SVN::Client::get_simple_prompt_provider(
2814              \&_simple_prompt, 2),
2815            SVN::Client::get_ssl_client_cert_prompt_provider(
2816              \&_ssl_client_cert_prompt, 2),
2817            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2818              \&_ssl_client_cert_pw_prompt, 2),
2819            SVN::Client::get_username_provider(),
2820            SVN::Client::get_ssl_server_trust_prompt_provider(
2821              \&_ssl_server_trust_prompt),
2822            SVN::Client::get_username_prompt_provider(
2823              \&_username_prompt, 2),
2824          ]);
2825        my $ra = SVN::Ra->new(url => $url, auth => $baton,
2826                              pool => SVN::Pool->new,
2827                              auth_provider_callbacks => $callbacks);
2828        $ra->{svn_path} = $url;
2829        $ra->{repos_root} = $ra->get_repos_root;
2830        $ra->{svn_path} =~ s#^\Q$ra->{repos_root}\E/*##;
2831        push @repo_path_split_cache, qr/^(\Q$ra->{repos_root}\E)/;
2832        return $ra;
2833}
2834
2835sub libsvn_dup_ra {
2836        my ($ra) = @_;
2837        SVN::Ra->new(map { $_ => $ra->{$_} }
2838                     qw/url auth auth_provider_callbacks repos_root svn_path/);
2839}
2840
2841sub libsvn_get_file {
2842        my ($gui, $f, $rev, $chg) = @_;
2843        $f =~ s#^/##;
2844        print "\t$chg\t$f\n" unless $_q;
2845
2846        my ($hash, $pid, $in, $out);
2847        my $pool = SVN::Pool->new;
2848        defined($pid = open3($in, $out, '>&STDERR',
2849                                qw/git-hash-object -w --stdin/)) or croak $!;
2850        # redirect STDOUT for SVN 1.1.x compatibility
2851        open my $stdout, '>&', \*STDOUT or croak $!;
2852        open STDOUT, '>&', $in or croak $!;
2853        my ($r, $props) = $SVN->get_file($f, $rev, \*STDOUT, $pool);
2854        $in->flush == 0 or croak $!;
2855        open STDOUT, '>&', $stdout or croak $!;
2856        close $in or croak $!;
2857        close $stdout or croak $!;
2858        $pool->clear;
2859        chomp($hash = do { local $/; <$out> });
2860        close $out or croak $!;
2861        waitpid $pid, 0;
2862        $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2863
2864        my $mode = exists $props->{'svn:executable'} ? '100755' : '100644';
2865        if (exists $props->{'svn:special'}) {
2866                $mode = '120000';
2867                my $link = `git-cat-file blob $hash`;
2868                $link =~ s/^link // or die "svn:special file with contents: <",
2869                                                $link, "> is not understood\n";
2870                defined($pid = open3($in, $out, '>&STDERR',
2871                                qw/git-hash-object -w --stdin/)) or croak $!;
2872                print $in $link;
2873                $in->flush == 0 or croak $!;
2874                close $in or croak $!;
2875                chomp($hash = do { local $/; <$out> });
2876                close $out or croak $!;
2877                waitpid $pid, 0;
2878                $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2879        }
2880        print $gui $mode,' ',$hash,"\t",$f,"\0" or croak $!;
2881}
2882
2883sub libsvn_log_entry {
2884        my ($rev, $author, $date, $msg, $parents) = @_;
2885        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2886                                         (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
2887                                or die "Unable to parse date: $date\n";
2888        if (defined $_authors && ! defined $users{$author}) {
2889                die "Author: $author not defined in $_authors file\n";
2890        }
2891        $msg = '' if ($rev == 0 && !defined $msg);
2892        return { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
2893                author => $author, msg => $msg."\n", parents => $parents || [] }
2894}
2895
2896sub process_rm {
2897        my ($gui, $last_commit, $f) = @_;
2898        # remove entire directories.
2899        if (safe_qx('git-ls-tree',$last_commit,'--',$f) =~ /^040000 tree/) {
2900                defined(my $pid = open my $ls, '-|') or croak $!;
2901                if (!$pid) {
2902                        exec(qw/git-ls-tree -r --name-only -z/,
2903                                $last_commit,'--',$f) or croak $!;
2904                }
2905                local $/ = "\0";
2906                while (<$ls>) {
2907                        print $gui '0 ',0 x 40,"\t",$_ or croak $!;
2908                }
2909                close $ls or croak $?;
2910        } else {
2911                print $gui '0 ',0 x 40,"\t",$f,"\0" or croak $!;
2912        }
2913}
2914
2915sub libsvn_fetch {
2916        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
2917        open my $gui, '| git-update-index -z --index-info' or croak $!;
2918        my @amr;
2919        my $p = $SVN->{svn_path};
2920        foreach my $f (keys %$paths) {
2921                my $m = $paths->{$f}->action();
2922                $f =~ s#^/\Q$p\E/##;
2923                next if $f =~ m#^/#;
2924                if ($m =~ /^[DR]$/) {
2925                        print "\t$m\t$f\n" unless $_q;
2926                        process_rm($gui, $last_commit, $f);
2927                        next if $m eq 'D';
2928                        # 'R' can be file replacements, too, right?
2929                }
2930                my $pool = SVN::Pool->new;
2931                my $t = $SVN->check_path($f, $rev, $pool);
2932                if ($t == $SVN::Node::file) {
2933                        if ($m =~ /^[AMR]$/) {
2934                                push @amr, [ $m, $f ];
2935                        } else {
2936                                die "Unrecognized action: $m, ($f r$rev)\n";
2937                        }
2938                } elsif ($t == $SVN::Node::dir && $m =~ /^[AR]$/) {
2939                        my @traversed = ();
2940                        libsvn_traverse($gui, '', $f, $rev, \@traversed);
2941                        foreach (@traversed) {
2942                                push @amr, [ $m, $_ ]
2943                        }
2944                }
2945                $pool->clear;
2946        }
2947        foreach (@amr) {
2948                libsvn_get_file($gui, $_->[1], $rev, $_->[0]);
2949        }
2950        close $gui or croak $?;
2951        return libsvn_log_entry($rev, $author, $date, $msg, [$last_commit]);
2952}
2953
2954sub svn_grab_base_rev {
2955        defined(my $pid = open my $fh, '-|') or croak $!;
2956        if (!$pid) {
2957                open my $null, '>', '/dev/null' or croak $!;
2958                open STDERR, '>&', $null or croak $!;
2959                exec qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0"
2960                                                                or croak $!;
2961        }
2962        chomp(my $c = do { local $/; <$fh> });
2963        close $fh;
2964        if (defined $c && length $c) {
2965                my ($url, $rev, $uuid) = cmt_metadata($c);
2966                return ($rev, $c) if defined $rev;
2967        }
2968        if ($_no_metadata) {
2969                my $offset = -41; # from tail
2970                my $rl;
2971                open my $fh, '<', $REVDB or
2972                        die "--no-metadata specified and $REVDB not readable\n";
2973                seek $fh, $offset, 2;
2974                $rl = readline $fh;
2975                defined $rl or return (undef, undef);
2976                chomp $rl;
2977                while ($c ne $rl && tell $fh != 0) {
2978                        $offset -= 41;
2979                        seek $fh, $offset, 2;
2980                        $rl = readline $fh;
2981                        defined $rl or return (undef, undef);
2982                        chomp $rl;
2983                }
2984                my $rev = tell $fh;
2985                croak $! if ($rev < -1);
2986                $rev =  ($rev - 41) / 41;
2987                close $fh or croak $!;
2988                return ($rev, $c);
2989        }
2990        return (undef, undef);
2991}
2992
2993sub libsvn_parse_revision {
2994        my $base = shift;
2995        my $head = $SVN->get_latest_revnum();
2996        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
2997                return ($base + 1, $head) if (defined $base);
2998                return (0, $head);
2999        }
3000        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
3001        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
3002        if ($_revision =~ /^BASE:(\d+)$/) {
3003                return ($base + 1, $1) if (defined $base);
3004                return (0, $head);
3005        }
3006        return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
3007        die "revision argument: $_revision not understood by git-svn\n",
3008                "Try using the command-line svn client instead\n";
3009}
3010
3011sub libsvn_traverse {
3012        my ($gui, $pfx, $path, $rev, $files) = @_;
3013        my $cwd = length $pfx ? "$pfx/$path" : $path;
3014        my $pool = SVN::Pool->new;
3015        $cwd =~ s#^\Q$SVN->{svn_path}\E##;
3016        my ($dirent, $r, $props) = $SVN->get_dir($cwd, $rev, $pool);
3017        foreach my $d (keys %$dirent) {
3018                my $t = $dirent->{$d}->kind;
3019                if ($t == $SVN::Node::dir) {
3020                        libsvn_traverse($gui, $cwd, $d, $rev, $files);
3021                } elsif ($t == $SVN::Node::file) {
3022                        my $file = "$cwd/$d";
3023                        if (defined $files) {
3024                                push @$files, $file;
3025                        } else {
3026                                libsvn_get_file($gui, $file, $rev, 'A');
3027                        }
3028                }
3029        }
3030        $pool->clear;
3031}
3032
3033sub libsvn_traverse_ignore {
3034        my ($fh, $path, $r) = @_;
3035        $path =~ s#^/+##g;
3036        my $pool = SVN::Pool->new;
3037        my ($dirent, undef, $props) = $SVN->get_dir($path, $r, $pool);
3038        my $p = $path;
3039        $p =~ s#^\Q$SVN->{svn_path}\E/##;
3040        print $fh length $p ? "\n# $p\n" : "\n# /\n";
3041        if (my $s = $props->{'svn:ignore'}) {
3042                $s =~ s/[\r\n]+/\n/g;
3043                chomp $s;
3044                if (length $p == 0) {
3045                        $s =~ s#\n#\n/$p#g;
3046                        print $fh "/$s\n";
3047                } else {
3048                        $s =~ s#\n#\n/$p/#g;
3049                        print $fh "/$p/$s\n";
3050                }
3051        }
3052        foreach (sort keys %$dirent) {
3053                next if $dirent->{$_}->kind != $SVN::Node::dir;
3054                libsvn_traverse_ignore($fh, "$path/$_", $r);
3055        }
3056        $pool->clear;
3057}
3058
3059sub revisions_eq {
3060        my ($path, $r0, $r1) = @_;
3061        return 1 if $r0 == $r1;
3062        my $nr = 0;
3063        if ($_use_lib) {
3064                # should be OK to use Pool here (r1 - r0) should be small
3065                my $pool = SVN::Pool->new;
3066                libsvn_get_log($SVN, [$path], $r0, $r1,
3067                                0, 1, 1, sub {$nr++}, $pool);
3068                $pool->clear;
3069        } else {
3070                my ($url, undef) = repo_path_split($SVN_URL);
3071                my $svn_log = svn_log_raw("$url/$path","-r$r0:$r1");
3072                while (next_log_entry($svn_log)) { $nr++ }
3073                close $svn_log->{fh};
3074        }
3075        return 0 if ($nr > 1);
3076        return 1;
3077}
3078
3079sub libsvn_find_parent_branch {
3080        my ($paths, $rev, $author, $date, $msg) = @_;
3081        my $svn_path = '/'.$SVN->{svn_path};
3082
3083        # look for a parent from another branch:
3084        my $i = $paths->{$svn_path} or return;
3085        my $branch_from = $i->copyfrom_path or return;
3086        my $r = $i->copyfrom_rev;
3087        print STDERR  "Found possible branch point: ",
3088                                "$branch_from => $svn_path, $r\n";
3089        $branch_from =~ s#^/##;
3090        my $l_map = {};
3091        read_url_paths_all($l_map, '', "$GIT_DIR/svn");
3092        my $url = $SVN->{repos_root};
3093        defined $l_map->{$url} or return;
3094        my $id = $l_map->{$url}->{$branch_from};
3095        if (!defined $id && $_follow_parent) {
3096                print STDERR "Following parent: $branch_from\@$r\n";
3097                # auto create a new branch and follow it
3098                $id = basename($branch_from);
3099                $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
3100                while (-r "$GIT_DIR/svn/$id") {
3101                        # just grow a tail if we're not unique enough :x
3102                        $id .= '-';
3103                }
3104        }
3105        return unless defined $id;
3106
3107        my ($r0, $parent) = find_rev_before($r,$id,1);
3108        if ($_follow_parent && (!defined $r0 || !defined $parent)) {
3109                defined(my $pid = fork) or croak $!;
3110                if (!$pid) {
3111                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
3112                        init_vars();
3113                        $SVN_URL = "$url/$branch_from";
3114                        $SVN = undef;
3115                        setup_git_svn();
3116                        # we can't assume SVN_URL exists at r+1:
3117                        $_revision = "0:$r";
3118                        fetch_lib();
3119                        exit 0;
3120                }
3121                waitpid $pid, 0;
3122                croak $? if $?;
3123                ($r0, $parent) = find_rev_before($r,$id,1);
3124        }
3125        return unless (defined $r0 && defined $parent);
3126        if (revisions_eq($branch_from, $r0, $r)) {
3127                unlink $GIT_SVN_INDEX;
3128                print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
3129                sys(qw/git-read-tree/, $parent);
3130                return libsvn_fetch($parent, $paths, $rev,
3131                                        $author, $date, $msg);
3132        }
3133        print STDERR "Nope, branch point not imported or unknown\n";
3134        return undef;
3135}
3136
3137sub libsvn_get_log {
3138        my ($ra, @args) = @_;
3139        if ($SVN::Core::VERSION le '1.2.0') {
3140                splice(@args, 3, 1);
3141        }
3142        $ra->get_log(@args);
3143}
3144
3145sub libsvn_new_tree {
3146        if (my $log_entry = libsvn_find_parent_branch(@_)) {
3147                return $log_entry;
3148        }
3149        my ($paths, $rev, $author, $date, $msg) = @_;
3150        open my $gui, '| git-update-index -z --index-info' or croak $!;
3151        libsvn_traverse($gui, '', $SVN->{svn_path}, $rev);
3152        close $gui or croak $?;
3153        return libsvn_log_entry($rev, $author, $date, $msg);
3154}
3155
3156sub find_graft_path_commit {
3157        my ($tree_paths, $p1, $r1) = @_;
3158        foreach my $x (keys %$tree_paths) {
3159                next unless ($p1 =~ /^\Q$x\E/);
3160                my $i = $tree_paths->{$x};
3161                my ($r0, $parent) = find_rev_before($r1,$i,1);
3162                return $parent if (defined $r0 && $r0 == $r1);
3163                print STDERR "r$r1 of $i not imported\n";
3164                next;
3165        }
3166        return undef;
3167}
3168
3169sub find_graft_path_parents {
3170        my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
3171        foreach my $x (keys %$tree_paths) {
3172                next unless ($p0 =~ /^\Q$x\E/);
3173                my $i = $tree_paths->{$x};
3174                my ($r, $parent) = find_rev_before($r0, $i, 1);
3175                if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
3176                        my ($url_b, undef, $uuid_b) = cmt_metadata($c);
3177                        my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
3178                        next if ($url_a && $url_b && $url_a eq $url_b &&
3179                                                        $uuid_b eq $uuid_a);
3180                        $grafts->{$c}->{$parent} = 1;
3181                }
3182        }
3183}
3184
3185sub libsvn_graft_file_copies {
3186        my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
3187        foreach (keys %$paths) {
3188                my $i = $paths->{$_};
3189                my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
3190                                        $i->copyfrom_rev);
3191                next unless (defined $p0 && defined $r0);
3192
3193                my $p1 = $_;
3194                $p1 =~ s#^/##;
3195                $p0 =~ s#^/##;
3196                my $c = find_graft_path_commit($tree_paths, $p1, $rev);
3197                next unless $c;
3198                find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
3199        }
3200}
3201
3202sub set_index {
3203        my $old = $ENV{GIT_INDEX_FILE};
3204        $ENV{GIT_INDEX_FILE} = shift;
3205        return $old;
3206}
3207
3208sub restore_index {
3209        my ($old) = @_;
3210        if (defined $old) {
3211                $ENV{GIT_INDEX_FILE} = $old;
3212        } else {
3213                delete $ENV{GIT_INDEX_FILE};
3214        }
3215}
3216
3217sub libsvn_commit_cb {
3218        my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
3219        if ($_optimize_commits && $rev == ($r_last + 1)) {
3220                my $log = libsvn_log_entry($rev,$committer,$date,$msg);
3221                $log->{tree} = get_tree_from_treeish($c);
3222                my $cmt = git_commit($log, $cmt_last, $c);
3223                my @diff = safe_qx('git-diff-tree', $cmt, $c);
3224                if (@diff) {
3225                        print STDERR "Trees differ: $cmt $c\n",
3226                                        join('',@diff),"\n";
3227                        exit 1;
3228                }
3229        } else {
3230                fetch("$rev=$c");
3231        }
3232}
3233
3234sub libsvn_ls_fullurl {
3235        my $fullurl = shift;
3236        $SVN ||= libsvn_connect($fullurl);
3237        my @ret;
3238        my $pool = SVN::Pool->new;
3239        my ($dirent, undef, undef) = $SVN->get_dir($SVN->{svn_path},
3240                                                $SVN->get_latest_revnum, $pool);
3241        foreach my $d (keys %$dirent) {
3242                if ($dirent->{$d}->kind == $SVN::Node::dir) {
3243                        push @ret, "$d/"; # add '/' for compat with cli svn
3244                }
3245        }
3246        $pool->clear;
3247        return @ret;
3248}
3249
3250
3251sub libsvn_skip_unknown_revs {
3252        my $err = shift;
3253        my $errno = $err->apr_err();
3254        # Maybe the branch we're tracking didn't
3255        # exist when the repo started, so it's
3256        # not an error if it doesn't, just continue
3257        #
3258        # Wonderfully consistent library, eh?
3259        # 160013 - svn:// and file://
3260        # 175002 - http(s)://
3261        # 175007 - http(s):// (this repo required authorization, too...)
3262        #   More codes may be discovered later...
3263        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3264                return;
3265        }
3266        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3267};
3268
3269# Tie::File seems to be prone to offset errors if revisions get sparse,
3270# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3271# one of my favorite modules is out :<  Next up would be one of the DBM
3272# modules, but I'm not sure which is most portable...  So I'll just
3273# go with something that's plain-text, but still capable of
3274# being randomly accessed.  So here's my ultra-simple fixed-width
3275# database.  All records are 40 characters + "\n", so it's easy to seek
3276# to a revision: (41 * rev) is the byte offset.
3277# A record of 40 0s denotes an empty revision.
3278# And yes, it's still pretty fast (faster than Tie::File).
3279sub revdb_set {
3280        my ($file, $rev, $commit) = @_;
3281        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
3282        open my $fh, '+<', $file or croak $!;
3283        my $offset = $rev * 41;
3284        # assume that append is the common case:
3285        seek $fh, 0, 2 or croak $!;
3286        my $pos = tell $fh;
3287        if ($pos < $offset) {
3288                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
3289        }
3290        seek $fh, $offset, 0 or croak $!;
3291        print $fh $commit,"\n";
3292        close $fh or croak $!;
3293}
3294
3295sub revdb_get {
3296        my ($file, $rev) = @_;
3297        my $ret;
3298        my $offset = $rev * 41;
3299        open my $fh, '<', $file or croak $!;
3300        seek $fh, $offset, 0;
3301        if (tell $fh == $offset) {
3302                $ret = readline $fh;
3303                if (defined $ret) {
3304                        chomp $ret;
3305                        $ret = undef if ($ret =~ /^0{40}$/);
3306                }
3307        }
3308        close $fh or croak $!;
3309        return $ret;
3310}
3311
3312sub copy_remote_ref {
3313        my $origin = $_cp_remote ? $_cp_remote : 'origin';
3314        my $ref = "refs/remotes/$GIT_SVN";
3315        if (safe_qx('git-ls-remote', $origin, $ref)) {
3316                sys(qw/git fetch/, $origin, "$ref:$ref");
3317        } elsif ($_cp_remote && !$_upgrade) {
3318                die "Unable to find remote reference: ",
3319                                "refs/remotes/$GIT_SVN on $origin\n";
3320        }
3321}
3322
3323package SVN::Git::Editor;
3324use vars qw/@ISA/;
3325use strict;
3326use warnings;
3327use Carp qw/croak/;
3328use IO::File;
3329
3330sub new {
3331        my $class = shift;
3332        my $git_svn = shift;
3333        my $self = SVN::Delta::Editor->new(@_);
3334        bless $self, $class;
3335        foreach (qw/svn_path c r ra /) {
3336                die "$_ required!\n" unless (defined $git_svn->{$_});
3337                $self->{$_} = $git_svn->{$_};
3338        }
3339        $self->{pool} = SVN::Pool->new;
3340        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3341        $self->{rm} = { };
3342        require Digest::MD5;
3343        return $self;
3344}
3345
3346sub split_path {
3347        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3348}
3349
3350sub repo_path {
3351        (defined $_[1] && length $_[1]) ? $_[1] : ''
3352}
3353
3354sub url_path {
3355        my ($self, $path) = @_;
3356        $self->{ra}->{url} . '/' . $self->repo_path($path);
3357}
3358
3359sub rmdirs {
3360        my ($self, $q) = @_;
3361        my $rm = $self->{rm};
3362        delete $rm->{''}; # we never delete the url we're tracking
3363        return unless %$rm;
3364
3365        foreach (keys %$rm) {
3366                my @d = split m#/#, $_;
3367                my $c = shift @d;
3368                $rm->{$c} = 1;
3369                while (@d) {
3370                        $c .= '/' . shift @d;
3371                        $rm->{$c} = 1;
3372                }
3373        }
3374        delete $rm->{$self->{svn_path}};
3375        delete $rm->{''}; # we never delete the url we're tracking
3376        return unless %$rm;
3377
3378        defined(my $pid = open my $fh,'-|') or croak $!;
3379        if (!$pid) {
3380                exec qw/git-ls-tree --name-only -r -z/, $self->{c} or croak $!;
3381        }
3382        local $/ = "\0";
3383        while (<$fh>) {
3384                chomp;
3385                my @dn = split m#/#, $_;
3386                while (pop @dn) {
3387                        delete $rm->{join '/', @dn};
3388                }
3389                unless (%$rm) {
3390                        close $fh;
3391                        return;
3392                }
3393        }
3394        close $fh;
3395
3396        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3397        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3398                $self->close_directory($bat->{$d}, $p);
3399                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3400                print "\tD+\t/$d/\n" unless $q;
3401                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3402                delete $bat->{$d};
3403        }
3404}
3405
3406sub open_or_add_dir {
3407        my ($self, $full_path, $baton) = @_;
3408        my $p = SVN::Pool->new;
3409        my $t = $self->{ra}->check_path($full_path, $self->{r}, $p);
3410        $p->clear;
3411        if ($t == $SVN::Node::none) {
3412                return $self->add_directory($full_path, $baton,
3413                                                undef, -1, $self->{pool});
3414        } elsif ($t == $SVN::Node::dir) {
3415                return $self->open_directory($full_path, $baton,
3416                                                $self->{r}, $self->{pool});
3417        }
3418        print STDERR "$full_path already exists in repository at ",
3419                "r$self->{r} and it is not a directory (",
3420                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3421        exit 1;
3422}
3423
3424sub ensure_path {
3425        my ($self, $path) = @_;
3426        my $bat = $self->{bat};
3427        $path = $self->repo_path($path);
3428        return $bat->{''} unless (length $path);
3429        my @p = split m#/+#, $path;
3430        my $c = shift @p;
3431        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3432        while (@p) {
3433                my $c0 = $c;
3434                $c .= '/' . shift @p;
3435                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3436        }
3437        return $bat->{$c};
3438}
3439
3440sub A {
3441        my ($self, $m, $q) = @_;
3442        my ($dir, $file) = split_path($m->{file_b});
3443        my $pbat = $self->ensure_path($dir);
3444        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3445                                        undef, -1);
3446        print "\tA\t$m->{file_b}\n" unless $q;
3447        $self->chg_file($fbat, $m);
3448        $self->close_file($fbat,undef,$self->{pool});
3449}
3450
3451sub C {
3452        my ($self, $m, $q) = @_;
3453        my ($dir, $file) = split_path($m->{file_b});
3454        my $pbat = $self->ensure_path($dir);
3455        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3456                                $self->url_path($m->{file_a}), $self->{r});
3457        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $q;
3458        $self->chg_file($fbat, $m);
3459        $self->close_file($fbat,undef,$self->{pool});
3460}
3461
3462sub delete_entry {
3463        my ($self, $path, $pbat) = @_;
3464        my $rpath = $self->repo_path($path);
3465        my ($dir, $file) = split_path($rpath);
3466        $self->{rm}->{$dir} = 1;
3467        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3468}
3469
3470sub R {
3471        my ($self, $m, $q) = @_;
3472        my ($dir, $file) = split_path($m->{file_b});
3473        my $pbat = $self->ensure_path($dir);
3474        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3475                                $self->url_path($m->{file_a}), $self->{r});
3476        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $q;
3477        $self->chg_file($fbat, $m);
3478        $self->close_file($fbat,undef,$self->{pool});
3479
3480        ($dir, $file) = split_path($m->{file_a});
3481        $pbat = $self->ensure_path($dir);
3482        $self->delete_entry($m->{file_a}, $pbat);
3483}
3484
3485sub M {
3486        my ($self, $m, $q) = @_;
3487        my ($dir, $file) = split_path($m->{file_b});
3488        my $pbat = $self->ensure_path($dir);
3489        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3490                                $pbat,$self->{r},$self->{pool});
3491        print "\t$m->{chg}\t$m->{file_b}\n" unless $q;
3492        $self->chg_file($fbat, $m);
3493        $self->close_file($fbat,undef,$self->{pool});
3494}
3495
3496sub T { shift->M(@_) }
3497
3498sub change_file_prop {
3499        my ($self, $fbat, $pname, $pval) = @_;
3500        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3501}
3502
3503sub chg_file {
3504        my ($self, $fbat, $m) = @_;
3505        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3506                $self->change_file_prop($fbat,'svn:executable','*');
3507        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3508                $self->change_file_prop($fbat,'svn:executable',undef);
3509        }
3510        my $fh = IO::File->new_tmpfile or croak $!;
3511        if ($m->{mode_b} =~ /^120/) {
3512                print $fh 'link ' or croak $!;
3513                $self->change_file_prop($fbat,'svn:special','*');
3514        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3515                $self->change_file_prop($fbat,'svn:special',undef);
3516        }
3517        defined(my $pid = fork) or croak $!;
3518        if (!$pid) {
3519                open STDOUT, '>&', $fh or croak $!;
3520                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3521        }
3522        waitpid $pid, 0;
3523        croak $? if $?;
3524        $fh->flush == 0 or croak $!;
3525        seek $fh, 0, 0 or croak $!;
3526
3527        my $md5 = Digest::MD5->new;
3528        $md5->addfile($fh) or croak $!;
3529        seek $fh, 0, 0 or croak $!;
3530
3531        my $exp = $md5->hexdigest;
3532        my $pool = SVN::Pool->new;
3533        my $atd = $self->apply_textdelta($fbat, undef, $pool);
3534        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3535        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3536        $pool->clear;
3537
3538        close $fh or croak $!;
3539}
3540
3541sub D {
3542        my ($self, $m, $q) = @_;
3543        my ($dir, $file) = split_path($m->{file_b});
3544        my $pbat = $self->ensure_path($dir);
3545        print "\tD\t$m->{file_b}\n" unless $q;
3546        $self->delete_entry($m->{file_b}, $pbat);
3547}
3548
3549sub close_edit {
3550        my ($self) = @_;
3551        my ($p,$bat) = ($self->{pool}, $self->{bat});
3552        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3553                $self->close_directory($bat->{$_}, $p);
3554        }
3555        $self->SUPER::close_edit($p);
3556        $p->clear;
3557}
3558
3559sub abort_edit {
3560        my ($self) = @_;
3561        $self->SUPER::abort_edit($self->{pool});
3562        $self->{pool}->clear;
3563}
3564
3565__END__
3566
3567Data structures:
3568
3569$svn_log hashref (as returned by svn_log_raw)
3570{
3571        fh => file handle of the log file,
3572        state => state of the log file parser (sep/msg/rev/msg_start...)
3573}
3574
3575$log_msg hashref as returned by next_log_entry($svn_log)
3576{
3577        msg => 'whitespace-formatted log entry
3578',                                              # trailing newline is preserved
3579        revision => '8',                        # integer
3580        date => '2004-02-24T17:01:44.108345Z',  # commit date
3581        author => 'committer name'
3582};
3583
3584
3585@mods = array of diff-index line hashes, each element represents one line
3586        of diff-index output
3587
3588diff-index line ($m hash)
3589{
3590        mode_a => first column of diff-index output, no leading ':',
3591        mode_b => second column of diff-index output,
3592        sha1_b => sha1sum of the final blob,
3593        chg => change type [MCRADT],
3594        file_a => original file name of a file (iff chg is 'C' or 'R')
3595        file_b => new/current file name of a file (any chg)
3596}
3597;
3598
3599# retval of read_url_paths{,_all}();
3600$l_map = {
3601        # repository root url
3602        'https://svn.musicpd.org' => {
3603                # repository path               # GIT_SVN_ID
3604                'mpd/trunk'             =>      'trunk',
3605                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
3606        },
3607}
3608
3609Notes:
3610        I don't trust the each() function on unless I created %hash myself
3611        because the internal iterator may not have started at base.