git-svn.perlon commit git-svn: bugfix: allow SVN:: lib users to track the root of the repository (308906f)
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $SVN_URL $SVN_INFO $SVN_WC $SVN_UUID
   8                $GIT_SVN_INDEX $GIT_SVN
   9                $GIT_DIR $GIT_SVN_DIR $REVDB/;
  10$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  11$VERSION = '@@GIT_VERSION@@';
  12
  13use Cwd qw/abs_path/;
  14$GIT_DIR = abs_path($ENV{GIT_DIR} || '.git');
  15$ENV{GIT_DIR} = $GIT_DIR;
  16
  17my $LC_ALL = $ENV{LC_ALL};
  18my $TZ = $ENV{TZ};
  19# make sure the svn binary gives consistent output between locales and TZs:
  20$ENV{TZ} = 'UTC';
  21$ENV{LC_ALL} = 'C';
  22$| = 1; # unbuffer STDOUT
  23
  24# If SVN:: library support is added, please make the dependencies
  25# optional and preserve the capability to use the command-line client.
  26# use eval { require SVN::... } to make it lazy load
  27# We don't use any modules not in the standard Perl distribution:
  28use Carp qw/croak/;
  29use IO::File qw//;
  30use File::Basename qw/dirname basename/;
  31use File::Path qw/mkpath/;
  32use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  33use File::Spec qw//;
  34use File::Copy qw/copy/;
  35use POSIX qw/strftime/;
  36use IPC::Open3;
  37use Memoize;
  38memoize('revisions_eq');
  39memoize('cmt_metadata');
  40memoize('get_commit_time');
  41
  42my ($SVN_PATH, $SVN, $SVN_LOG, $_use_lib);
  43$_use_lib = 1 unless $ENV{GIT_SVN_NO_LIB};
  44libsvn_load();
  45my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  46my $sha1 = qr/[a-f\d]{40}/;
  47my $sha1_short = qr/[a-f\d]{4,40}/;
  48my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
  49        $_find_copies_harder, $_l, $_cp_similarity, $_cp_remote,
  50        $_repack, $_repack_nr, $_repack_flags, $_q,
  51        $_message, $_file, $_follow_parent, $_no_metadata,
  52        $_template, $_shared, $_no_default_regex, $_no_graft_copy,
  53        $_limit, $_verbose, $_incremental, $_oneline, $_l_fmt, $_show_commit,
  54        $_version, $_upgrade, $_authors, $_branch_all_refs, @_opt_m);
  55my (@_branch_from, %tree_map, %users, %rusers, %equiv);
  56my ($_svn_co_url_revs, $_svn_pg_peg_revs);
  57my @repo_path_split_cache;
  58
  59my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
  60                'branch|b=s' => \@_branch_from,
  61                'follow-parent|follow' => \$_follow_parent,
  62                'branch-all-refs|B' => \$_branch_all_refs,
  63                'authors-file|A=s' => \$_authors,
  64                'repack:i' => \$_repack,
  65                'no-metadata' => \$_no_metadata,
  66                'quiet|q' => \$_q,
  67                'repack-flags|repack-args|repack-opts=s' => \$_repack_flags);
  68
  69my ($_trunk, $_tags, $_branches);
  70my %multi_opts = ( 'trunk|T=s' => \$_trunk,
  71                'tags|t=s' => \$_tags,
  72                'branches|b=s' => \$_branches );
  73my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
  74my %cmt_opts = ( 'edit|e' => \$_edit,
  75                'rmdir' => \$_rmdir,
  76                'find-copies-harder' => \$_find_copies_harder,
  77                'l=i' => \$_l,
  78                'copy-similarity|C=i'=> \$_cp_similarity
  79);
  80
  81my %cmd = (
  82        fetch => [ \&fetch, "Download new revisions from SVN",
  83                        { 'revision|r=s' => \$_revision, %fc_opts } ],
  84        init => [ \&init, "Initialize a repo for tracking" .
  85                          " (requires URL argument)",
  86                          \%init_opts ],
  87        commit => [ \&commit, "Commit git revisions to SVN",
  88                        {       'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
  89        'show-ignore' => [ \&show_ignore, "Show svn:ignore listings",
  90                        { 'revision|r=i' => \$_revision } ],
  91        rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
  92                        { 'no-ignore-externals' => \$_no_ignore_ext,
  93                          'copy-remote|remote=s' => \$_cp_remote,
  94                          'upgrade' => \$_upgrade } ],
  95        'graft-branches' => [ \&graft_branches,
  96                        'Detect merges/branches from already imported history',
  97                        { 'merge-rx|m' => \@_opt_m,
  98                          'branch|b=s' => \@_branch_from,
  99                          'branch-all-refs|B' => \$_branch_all_refs,
 100                          'no-default-regex' => \$_no_default_regex,
 101                          'no-graft-copy' => \$_no_graft_copy } ],
 102        'multi-init' => [ \&multi_init,
 103                        'Initialize multiple trees (like git-svnimport)',
 104                        { %multi_opts, %fc_opts } ],
 105        'multi-fetch' => [ \&multi_fetch,
 106                        'Fetch multiple trees (like git-svnimport)',
 107                        \%fc_opts ],
 108        'log' => [ \&show_log, 'Show commit logs',
 109                        { 'limit=i' => \$_limit,
 110                          'revision|r=s' => \$_revision,
 111                          'verbose|v' => \$_verbose,
 112                          'incremental' => \$_incremental,
 113                          'oneline' => \$_oneline,
 114                          'show-commit' => \$_show_commit,
 115                          'authors-file|A=s' => \$_authors,
 116                        } ],
 117        'commit-diff' => [ \&commit_diff, 'Commit a diff between two trees',
 118                        { 'message|m=s' => \$_message,
 119                          'file|F=s' => \$_file,
 120                        %cmt_opts } ],
 121);
 122
 123my $cmd;
 124for (my $i = 0; $i < @ARGV; $i++) {
 125        if (defined $cmd{$ARGV[$i]}) {
 126                $cmd = $ARGV[$i];
 127                splice @ARGV, $i, 1;
 128                last;
 129        }
 130};
 131
 132my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 133
 134read_repo_config(\%opts);
 135my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
 136                                'version|V' => \$_version,
 137                                'id|i=s' => \$GIT_SVN);
 138exit 1 if (!$rv && $cmd ne 'log');
 139
 140set_default_vals();
 141usage(0) if $_help;
 142version() if $_version;
 143usage(1) unless defined $cmd;
 144init_vars();
 145load_authors() if $_authors;
 146load_all_refs() if $_branch_all_refs;
 147svn_compat_check() unless $_use_lib;
 148migration_check() unless $cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/;
 149$cmd{$cmd}->[0]->(@ARGV);
 150exit 0;
 151
 152####################### primary functions ######################
 153sub usage {
 154        my $exit = shift || 0;
 155        my $fd = $exit ? \*STDERR : \*STDOUT;
 156        print $fd <<"";
 157git-svn - bidirectional operations between a single Subversion tree and git
 158Usage: $0 <command> [options] [arguments]\n
 159
 160        print $fd "Available commands:\n" unless $cmd;
 161
 162        foreach (sort keys %cmd) {
 163                next if $cmd && $cmd ne $_;
 164                print $fd '  ',pack('A13',$_),$cmd{$_}->[1],"\n";
 165                foreach (keys %{$cmd{$_}->[2]}) {
 166                        # prints out arguments as they should be passed:
 167                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 168                        print $fd ' ' x 17, join(', ', map { length $_ > 1 ?
 169                                                        "--$_" : "-$_" }
 170                                                split /\|/,$_)," $x\n";
 171                }
 172        }
 173        print $fd <<"";
 174\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 175arbitrary identifier if you're tracking multiple SVN branches/repositories in
 176one git repository and want to keep them separate.  See git-svn(1) for more
 177information.
 178
 179        exit $exit;
 180}
 181
 182sub version {
 183        print "git-svn version $VERSION\n";
 184        exit 0;
 185}
 186
 187sub rebuild {
 188        if (quiet_run(qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0")) {
 189                copy_remote_ref();
 190        }
 191        $SVN_URL = shift or undef;
 192        my $newest_rev = 0;
 193        if ($_upgrade) {
 194                sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD");
 195        } else {
 196                check_upgrade_needed();
 197        }
 198
 199        my $pid = open(my $rev_list,'-|');
 200        defined $pid or croak $!;
 201        if ($pid == 0) {
 202                exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!;
 203        }
 204        my $latest;
 205        while (<$rev_list>) {
 206                chomp;
 207                my $c = $_;
 208                croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 209                my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
 210                next if (!@commit); # skip merges
 211                my ($url, $rev, $uuid) = extract_metadata($commit[$#commit]);
 212                if (!$rev || !$uuid) {
 213                        croak "Unable to extract revision or UUID from ",
 214                                "$c, $commit[$#commit]\n";
 215                }
 216
 217                # if we merged or otherwise started elsewhere, this is
 218                # how we break out of it
 219                next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
 220                next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
 221
 222                unless (defined $latest) {
 223                        if (!$SVN_URL && !$url) {
 224                                croak "SVN repository location required: $url\n";
 225                        }
 226                        $SVN_URL ||= $url;
 227                        $SVN_UUID ||= $uuid;
 228                        setup_git_svn();
 229                        $latest = $rev;
 230                }
 231                revdb_set($REVDB, $rev, $c);
 232                print "r$rev = $c\n";
 233                $newest_rev = $rev if ($rev > $newest_rev);
 234        }
 235        close $rev_list or croak $?;
 236
 237        goto out if $_use_lib;
 238        if (!chdir $SVN_WC) {
 239                svn_cmd_checkout($SVN_URL, $latest, $SVN_WC);
 240                chdir $SVN_WC or croak $!;
 241        }
 242
 243        $pid = fork;
 244        defined $pid or croak $!;
 245        if ($pid == 0) {
 246                my @svn_up = qw(svn up);
 247                push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 248                sys(@svn_up,"-r$newest_rev");
 249                $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 250                index_changes();
 251                exec('git-write-tree') or croak $!;
 252        }
 253        waitpid $pid, 0;
 254        croak $? if $?;
 255out:
 256        if ($_upgrade) {
 257                print STDERR <<"";
 258Keeping deprecated refs/head/$GIT_SVN-HEAD for now.  Please remove it
 259when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN
 260
 261        }
 262}
 263
 264sub init {
 265        my $url = shift or die "SVN repository location required " .
 266                                "as a command-line argument\n";
 267        $url =~ s!/+$!!; # strip trailing slash
 268
 269        if (my $repo_path = shift) {
 270                unless (-d $repo_path) {
 271                        mkpath([$repo_path]);
 272                }
 273                $GIT_DIR = $ENV{GIT_DIR} = $repo_path . "/.git";
 274                init_vars();
 275        }
 276
 277        $SVN_URL = $url;
 278        unless (-d $GIT_DIR) {
 279                my @init_db = ('git-init-db');
 280                push @init_db, "--template=$_template" if defined $_template;
 281                push @init_db, "--shared" if defined $_shared;
 282                sys(@init_db);
 283        }
 284        setup_git_svn();
 285}
 286
 287sub fetch {
 288        check_upgrade_needed();
 289        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 290        my $ret = $_use_lib ? fetch_lib(@_) : fetch_cmd(@_);
 291        if ($ret->{commit} && quiet_run(qw(git-rev-parse --verify
 292                                                refs/heads/master^0))) {
 293                sys(qw(git-update-ref refs/heads/master),$ret->{commit});
 294        }
 295        return $ret;
 296}
 297
 298sub fetch_cmd {
 299        my (@parents) = @_;
 300        my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 301        unless ($_revision) {
 302                $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 303        }
 304        push @log_args, "-r$_revision";
 305        push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 306
 307        my $svn_log = svn_log_raw(@log_args);
 308
 309        my $base = next_log_entry($svn_log) or croak "No base revision!\n";
 310        # don't need last_revision from grab_base_rev() because
 311        # user could've specified a different revision to skip (they
 312        # didn't want to import certain revisions into git for whatever
 313        # reason, so trust $base->{revision} instead.
 314        my (undef, $last_commit) = svn_grab_base_rev();
 315        unless (-d $SVN_WC) {
 316                svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC);
 317                chdir $SVN_WC or croak $!;
 318                read_uuid();
 319                $last_commit = git_commit($base, @parents);
 320                assert_tree($last_commit);
 321        } else {
 322                chdir $SVN_WC or croak $!;
 323                read_uuid();
 324                # looks like a user manually cp'd and svn switch'ed
 325                unless ($last_commit) {
 326                        sys(qw/svn revert -R ./);
 327                        assert_svn_wc_clean($base->{revision});
 328                        $last_commit = git_commit($base, @parents);
 329                        assert_tree($last_commit);
 330                }
 331        }
 332        my @svn_up = qw(svn up);
 333        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 334        my $last = $base;
 335        while (my $log_msg = next_log_entry($svn_log)) {
 336                if ($last->{revision} >= $log_msg->{revision}) {
 337                        croak "Out of order: last >= current: ",
 338                                "$last->{revision} >= $log_msg->{revision}\n";
 339                }
 340                # Revert is needed for cases like:
 341                # https://svn.musicpd.org/Jamming/trunk (r166:167), but
 342                # I can't seem to reproduce something like that on a test...
 343                sys(qw/svn revert -R ./);
 344                assert_svn_wc_clean($last->{revision});
 345                sys(@svn_up,"-r$log_msg->{revision}");
 346                $last_commit = git_commit($log_msg, $last_commit, @parents);
 347                $last = $log_msg;
 348        }
 349        close $svn_log->{fh};
 350        $last->{commit} = $last_commit;
 351        return $last;
 352}
 353
 354sub fetch_lib {
 355        my (@parents) = @_;
 356        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 357        my $repo;
 358        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 359        $SVN_LOG ||= libsvn_connect($repo);
 360        $SVN ||= libsvn_connect($repo);
 361        my ($last_rev, $last_commit) = svn_grab_base_rev();
 362        my ($base, $head) = libsvn_parse_revision($last_rev);
 363        if ($base > $head) {
 364                return { revision => $last_rev, commit => $last_commit }
 365        }
 366        my $index = set_index($GIT_SVN_INDEX);
 367
 368        # limit ourselves and also fork() since get_log won't release memory
 369        # after processing a revision and SVN stuff seems to leak
 370        my $inc = 1000;
 371        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
 372        read_uuid();
 373        if (defined $last_commit) {
 374                unless (-e $GIT_SVN_INDEX) {
 375                        sys(qw/git-read-tree/, $last_commit);
 376                }
 377                chomp (my $x = `git-write-tree`);
 378                my ($y) = (`git-cat-file commit $last_commit`
 379                                                        =~ /^tree ($sha1)/m);
 380                if ($y ne $x) {
 381                        unlink $GIT_SVN_INDEX or croak $!;
 382                        sys(qw/git-read-tree/, $last_commit);
 383                }
 384                chomp ($x = `git-write-tree`);
 385                if ($y ne $x) {
 386                        print STDERR "trees ($last_commit) $y != $x\n",
 387                                 "Something is seriously wrong...\n";
 388                }
 389        }
 390        while (1) {
 391                # fork, because using SVN::Pool with get_log() still doesn't
 392                # seem to help enough to keep memory usage down.
 393                defined(my $pid = fork) or croak $!;
 394                if (!$pid) {
 395                        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
 396
 397                        # Yes I'm perfectly aware that the fourth argument
 398                        # below is the limit revisions number.  Unfortunately
 399                        # performance sucks with it enabled, so it's much
 400                        # faster to fetch revision ranges instead of relying
 401                        # on the limiter.
 402                        libsvn_get_log($SVN_LOG, '/'.$SVN_PATH,
 403                                        $min, $max, 0, 1, 1,
 404                                sub {
 405                                        my $log_msg;
 406                                        if ($last_commit) {
 407                                                $log_msg = libsvn_fetch(
 408                                                        $last_commit, @_);
 409                                                $last_commit = git_commit(
 410                                                        $log_msg,
 411                                                        $last_commit,
 412                                                        @parents);
 413                                        } else {
 414                                                $log_msg = libsvn_new_tree(@_);
 415                                                $last_commit = git_commit(
 416                                                        $log_msg, @parents);
 417                                        }
 418                                });
 419                        exit 0;
 420                }
 421                waitpid $pid, 0;
 422                croak $? if $?;
 423                ($last_rev, $last_commit) = svn_grab_base_rev();
 424                last if ($max >= $head);
 425                $min = $max + 1;
 426                $max += $inc;
 427                $max = $head if ($max > $head);
 428        }
 429        restore_index($index);
 430        return { revision => $last_rev, commit => $last_commit };
 431}
 432
 433sub commit {
 434        my (@commits) = @_;
 435        check_upgrade_needed();
 436        if ($_stdin || !@commits) {
 437                print "Reading from stdin...\n";
 438                @commits = ();
 439                while (<STDIN>) {
 440                        if (/\b($sha1_short)\b/o) {
 441                                unshift @commits, $1;
 442                        }
 443                }
 444        }
 445        my @revs;
 446        foreach my $c (@commits) {
 447                chomp(my @tmp = safe_qx('git-rev-parse',$c));
 448                if (scalar @tmp == 1) {
 449                        push @revs, $tmp[0];
 450                } elsif (scalar @tmp > 1) {
 451                        push @revs, reverse (safe_qx('git-rev-list',@tmp));
 452                } else {
 453                        die "Failed to rev-parse $c\n";
 454                }
 455        }
 456        chomp @revs;
 457        $_use_lib ? commit_lib(@revs) : commit_cmd(@revs);
 458        print "Done committing ",scalar @revs," revisions to SVN\n";
 459}
 460
 461sub commit_cmd {
 462        my (@revs) = @_;
 463
 464        chdir $SVN_WC or croak "Unable to chdir $SVN_WC: $!\n";
 465        my $info = svn_info('.');
 466        my $fetched = fetch();
 467        if ($info->{Revision} != $fetched->{revision}) {
 468                print STDERR "There are new revisions that were fetched ",
 469                                "and need to be merged (or acknowledged) ",
 470                                "before committing.\n";
 471                exit 1;
 472        }
 473        $info = svn_info('.');
 474        read_uuid($info);
 475        my $last = $fetched;
 476        foreach my $c (@revs) {
 477                my $mods = svn_checkout_tree($last, $c);
 478                if (scalar @$mods == 0) {
 479                        print "Skipping, no changes detected\n";
 480                        next;
 481                }
 482                $last = svn_commit_tree($last, $c);
 483        }
 484}
 485
 486sub commit_lib {
 487        my (@revs) = @_;
 488        my ($r_last, $cmt_last) = svn_grab_base_rev();
 489        defined $r_last or die "Must have an existing revision to commit\n";
 490        my $fetched = fetch();
 491        if ($r_last != $fetched->{revision}) {
 492                print STDERR "There are new revisions that were fetched ",
 493                                "and need to be merged (or acknowledged) ",
 494                                "before committing.\n",
 495                                "last rev: $r_last\n",
 496                                " current: $fetched->{revision}\n";
 497                exit 1;
 498        }
 499        read_uuid();
 500        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 501        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
 502
 503        set_svn_commit_env();
 504        foreach my $c (@revs) {
 505                my $log_msg = get_commit_message($c, $commit_msg);
 506
 507                # fork for each commit because there's a memory leak I
 508                # can't track down... (it's probably in the SVN code)
 509                defined(my $pid = open my $fh, '-|') or croak $!;
 510                if (!$pid) {
 511                        my $ed = SVN::Git::Editor->new(
 512                                        {       r => $r_last,
 513                                                ra => $SVN,
 514                                                c => $c,
 515                                                svn_path => $SVN_PATH
 516                                        },
 517                                        $SVN->get_commit_editor(
 518                                                $log_msg->{msg},
 519                                                sub {
 520                                                        libsvn_commit_cb(
 521                                                                @_, $c,
 522                                                                $log_msg->{msg},
 523                                                                $r_last,
 524                                                                $cmt_last)
 525                                                },
 526                                                @lock)
 527                                        );
 528                        my $mods = libsvn_checkout_tree($cmt_last, $c, $ed);
 529                        if (@$mods == 0) {
 530                                print "No changes\nr$r_last = $cmt_last\n";
 531                                $ed->abort_edit;
 532                        } else {
 533                                $ed->close_edit;
 534                        }
 535                        exit 0;
 536                }
 537                my ($r_new, $cmt_new, $no);
 538                while (<$fh>) {
 539                        print $_;
 540                        chomp;
 541                        if (/^r(\d+) = ($sha1)$/o) {
 542                                ($r_new, $cmt_new) = ($1, $2);
 543                        } elsif ($_ eq 'No changes') {
 544                                $no = 1;
 545                        }
 546                }
 547                close $fh or croak $?;
 548                if (! defined $r_new && ! defined $cmt_new) {
 549                        unless ($no) {
 550                                die "Failed to parse revision information\n";
 551                        }
 552                } else {
 553                        ($r_last, $cmt_last) = ($r_new, $cmt_new);
 554                }
 555        }
 556        $ENV{LC_ALL} = 'C';
 557        unlink $commit_msg;
 558}
 559
 560sub show_ignore {
 561        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 562        $_use_lib ? show_ignore_lib() : show_ignore_cmd();
 563}
 564
 565sub show_ignore_cmd {
 566        require File::Find or die $!;
 567        if (defined $_revision) {
 568                die "-r/--revision option doesn't work unless the Perl SVN ",
 569                        "libraries are used\n";
 570        }
 571        chdir $SVN_WC or croak $!;
 572        my %ign;
 573        File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 574                s#^\./##;
 575                @{$ign{$_}} = svn_propget_base('svn:ignore', $_);
 576                }}, no_chdir=>1},'.');
 577
 578        print "\n# /\n";
 579        foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 580        delete $ign{'.'};
 581        foreach my $i (sort keys %ign) {
 582                print "\n# ",$i,"\n";
 583                foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 584        }
 585}
 586
 587sub show_ignore_lib {
 588        my $repo;
 589        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 590        $SVN ||= libsvn_connect($repo);
 591        my $r = defined $_revision ? $_revision : $SVN->get_latest_revnum;
 592        libsvn_traverse_ignore(\*STDOUT, $SVN_PATH, $r);
 593}
 594
 595sub graft_branches {
 596        my $gr_file = "$GIT_DIR/info/grafts";
 597        my ($grafts, $comments) = read_grafts($gr_file);
 598        my $gr_sha1;
 599
 600        if (%$grafts) {
 601                # temporarily disable our grafts file to make this idempotent
 602                chomp($gr_sha1 = safe_qx(qw/git-hash-object -w/,$gr_file));
 603                rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
 604        }
 605
 606        my $l_map = read_url_paths();
 607        my @re = map { qr/$_/is } @_opt_m if @_opt_m;
 608        unless ($_no_default_regex) {
 609                push @re, (qr/\b(?:merge|merging|merged)\s+with\s+([\w\.\-]+)/i,
 610                        qr/\b(?:merge|merging|merged)\s+([\w\.\-]+)/i,
 611                        qr/\b(?:from|of)\s+([\w\.\-]+)/i );
 612        }
 613        foreach my $u (keys %$l_map) {
 614                if (@re) {
 615                        foreach my $p (keys %{$l_map->{$u}}) {
 616                                graft_merge_msg($grafts,$l_map,$u,$p,@re);
 617                        }
 618                }
 619                unless ($_no_graft_copy) {
 620                        if ($_use_lib) {
 621                                graft_file_copy_lib($grafts,$l_map,$u);
 622                        } else {
 623                                graft_file_copy_cmd($grafts,$l_map,$u);
 624                        }
 625                }
 626        }
 627        graft_tree_joins($grafts);
 628
 629        write_grafts($grafts, $comments, $gr_file);
 630        unlink "$gr_file~$gr_sha1" if $gr_sha1;
 631}
 632
 633sub multi_init {
 634        my $url = shift;
 635        $_trunk ||= 'trunk';
 636        $_trunk =~ s#/+$##;
 637        $url =~ s#/+$## if $url;
 638        if ($_trunk !~ m#^[a-z\+]+://#) {
 639                $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#);
 640                unless ($url) {
 641                        print STDERR "E: '$_trunk' is not a complete URL ",
 642                                "and a separate URL is not specified\n";
 643                        exit 1;
 644                }
 645                $_trunk = $url . $_trunk;
 646        }
 647        if ($GIT_SVN eq 'git-svn') {
 648                print "GIT_SVN_ID set to 'trunk' for $_trunk\n";
 649                $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
 650        }
 651        init_vars();
 652        init($_trunk);
 653        complete_url_ls_init($url, $_branches, '--branches/-b', '');
 654        complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/');
 655}
 656
 657sub multi_fetch {
 658        # try to do trunk first, since branches/tags
 659        # may be descended from it.
 660        if (-e "$GIT_DIR/svn/trunk/info/url") {
 661                fetch_child_id('trunk', @_);
 662        }
 663        rec_fetch('', "$GIT_DIR/svn", @_);
 664}
 665
 666sub show_log {
 667        my (@args) = @_;
 668        my ($r_min, $r_max);
 669        my $r_last = -1; # prevent dupes
 670        rload_authors() if $_authors;
 671        if (defined $TZ) {
 672                $ENV{TZ} = $TZ;
 673        } else {
 674                delete $ENV{TZ};
 675        }
 676        if (defined $_revision) {
 677                if ($_revision =~ /^(\d+):(\d+)$/) {
 678                        ($r_min, $r_max) = ($1, $2);
 679                } elsif ($_revision =~ /^\d+$/) {
 680                        $r_min = $r_max = $_revision;
 681                } else {
 682                        print STDERR "-r$_revision is not supported, use ",
 683                                "standard \'git log\' arguments instead\n";
 684                        exit 1;
 685                }
 686        }
 687
 688        my $pid = open(my $log,'-|');
 689        defined $pid or croak $!;
 690        if (!$pid) {
 691                exec(git_svn_log_cmd($r_min,$r_max), @args) or croak $!;
 692        }
 693        setup_pager();
 694        my (@k, $c, $d);
 695
 696        while (<$log>) {
 697                if (/^commit ($sha1_short)/o) {
 698                        my $cmt = $1;
 699                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
 700                                $r_last = $c->{r};
 701                                process_commit($c, $r_min, $r_max, \@k) or
 702                                                                goto out;
 703                        }
 704                        $d = undef;
 705                        $c = { c => $cmt };
 706                } elsif (/^author (.+) (\d+) ([\-\+]?\d+)$/) {
 707                        get_author_info($c, $1, $2, $3);
 708                } elsif (/^(?:tree|parent|committer) /) {
 709                        # ignore
 710                } elsif (/^:\d{6} \d{6} $sha1_short/o) {
 711                        push @{$c->{raw}}, $_;
 712                } elsif (/^diff /) {
 713                        $d = 1;
 714                        push @{$c->{diff}}, $_;
 715                } elsif ($d) {
 716                        push @{$c->{diff}}, $_;
 717                } elsif (/^    (git-svn-id:.+)$/) {
 718                        (undef, $c->{r}, undef) = extract_metadata($1);
 719                } elsif (s/^    //) {
 720                        push @{$c->{l}}, $_;
 721                }
 722        }
 723        if ($c && defined $c->{r} && $c->{r} != $r_last) {
 724                $r_last = $c->{r};
 725                process_commit($c, $r_min, $r_max, \@k);
 726        }
 727        if (@k) {
 728                my $swap = $r_max;
 729                $r_max = $r_min;
 730                $r_min = $swap;
 731                process_commit($_, $r_min, $r_max) foreach reverse @k;
 732        }
 733out:
 734        close $log;
 735        print '-' x72,"\n" unless $_incremental || $_oneline;
 736}
 737
 738sub commit_diff_usage {
 739        print STDERR "Usage: $0 commit-diff <tree-ish> <tree-ish> [<URL>]\n";
 740        exit 1
 741}
 742
 743sub commit_diff {
 744        if (!$_use_lib) {
 745                print STDERR "commit-diff must be used with SVN libraries\n";
 746                exit 1;
 747        }
 748        my $ta = shift or commit_diff_usage();
 749        my $tb = shift or commit_diff_usage();
 750        if (!eval { $SVN_URL = shift || file_to_s("$GIT_SVN_DIR/info/url") }) {
 751                print STDERR "Needed URL or usable git-svn id command-line\n";
 752                commit_diff_usage();
 753        }
 754        if (defined $_message && defined $_file) {
 755                print STDERR "Both --message/-m and --file/-F specified ",
 756                                "for the commit message.\n",
 757                                "I have no idea what you mean\n";
 758                exit 1;
 759        }
 760        if (defined $_file) {
 761                $_message = file_to_s($_file);
 762        } else {
 763                $_message ||= get_commit_message($tb,
 764                                        "$GIT_DIR/.svn-commit.tmp.$$")->{msg};
 765        }
 766        my $repo;
 767        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 768        $SVN_LOG ||= libsvn_connect($repo);
 769        $SVN ||= libsvn_connect($repo);
 770        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 771        my $ed = SVN::Git::Editor->new({        r => $SVN->get_latest_revnum,
 772                                                ra => $SVN, c => $tb,
 773                                                svn_path => $SVN_PATH
 774                                        },
 775                                $SVN->get_commit_editor($_message,
 776                                        sub {print "Committed $_[0]\n"},@lock)
 777                                );
 778        my $mods = libsvn_checkout_tree($ta, $tb, $ed);
 779        if (@$mods == 0) {
 780                print "No changes\n$ta == $tb\n";
 781                $ed->abort_edit;
 782        } else {
 783                $ed->close_edit;
 784        }
 785}
 786
 787########################### utility functions #########################
 788
 789sub cmt_showable {
 790        my ($c) = @_;
 791        return 1 if defined $c->{r};
 792        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
 793                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
 794                my @msg = safe_qx(qw/git-cat-file commit/, $c->{c});
 795                shift @msg while ($msg[0] ne "\n");
 796                shift @msg;
 797                @{$c->{l}} = grep !/^git-svn-id: /, @msg;
 798
 799                (undef, $c->{r}, undef) = extract_metadata(
 800                                (grep(/^git-svn-id: /, @msg))[-1]);
 801        }
 802        return defined $c->{r};
 803}
 804
 805sub git_svn_log_cmd {
 806        my ($r_min, $r_max) = @_;
 807        my @cmd = (qw/git-log --abbrev-commit --pretty=raw
 808                        --default/, "refs/remotes/$GIT_SVN");
 809        push @cmd, '--summary' if $_verbose;
 810        return @cmd unless defined $r_max;
 811        if ($r_max == $r_min) {
 812                push @cmd, '--max-count=1';
 813                if (my $c = revdb_get($REVDB, $r_max)) {
 814                        push @cmd, $c;
 815                }
 816        } else {
 817                my ($c_min, $c_max);
 818                $c_max = revdb_get($REVDB, $r_max);
 819                $c_min = revdb_get($REVDB, $r_min);
 820                if ($c_min && $c_max) {
 821                        if ($r_max > $r_max) {
 822                                push @cmd, "$c_min..$c_max";
 823                        } else {
 824                                push @cmd, "$c_max..$c_min";
 825                        }
 826                } elsif ($r_max > $r_min) {
 827                        push @cmd, $c_max;
 828                } else {
 829                        push @cmd, $c_min;
 830                }
 831        }
 832        return @cmd;
 833}
 834
 835sub fetch_child_id {
 836        my $id = shift;
 837        print "Fetching $id\n";
 838        my $ref = "$GIT_DIR/refs/remotes/$id";
 839        defined(my $pid = open my $fh, '-|') or croak $!;
 840        if (!$pid) {
 841                $_repack = undef;
 842                $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 843                init_vars();
 844                fetch(@_);
 845                exit 0;
 846        }
 847        while (<$fh>) {
 848                print $_;
 849                check_repack() if (/^r\d+ = $sha1/);
 850        }
 851        close $fh or croak $?;
 852}
 853
 854sub rec_fetch {
 855        my ($pfx, $p, @args) = @_;
 856        my @dir;
 857        foreach (sort <$p/*>) {
 858                if (-r "$_/info/url") {
 859                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
 860                        my $id = $pfx . basename $_;
 861                        next if $id eq 'trunk';
 862                        fetch_child_id($id, @args);
 863                } elsif (-d $_) {
 864                        push @dir, $_;
 865                }
 866        }
 867        foreach (@dir) {
 868                my $x = $_;
 869                $x =~ s!^\Q$GIT_DIR\E/svn/!!;
 870                rec_fetch($x, $_);
 871        }
 872}
 873
 874sub complete_url_ls_init {
 875        my ($url, $var, $switch, $pfx) = @_;
 876        unless ($var) {
 877                print STDERR "W: $switch not specified\n";
 878                return;
 879        }
 880        $var =~ s#/+$##;
 881        if ($var !~ m#^[a-z\+]+://#) {
 882                $var = '/' . $var if ($var !~ m#^/#);
 883                unless ($url) {
 884                        print STDERR "E: '$var' is not a complete URL ",
 885                                "and a separate URL is not specified\n";
 886                        exit 1;
 887                }
 888                $var = $url . $var;
 889        }
 890        chomp(my @ls = $_use_lib ? libsvn_ls_fullurl($var)
 891                                : safe_qx(qw/svn ls --non-interactive/, $var));
 892        my $old = $GIT_SVN;
 893        defined(my $pid = fork) or croak $!;
 894        if (!$pid) {
 895                foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
 896                        $u =~ s#/+$##;
 897                        if ($u !~ m!\Q$var\E/(.+)$!) {
 898                                print STDERR "W: Unrecognized URL: $u\n";
 899                                die "This should never happen\n";
 900                        }
 901                        my $id = $pfx.$1;
 902                        print "init $u => $id\n";
 903                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 904                        init_vars();
 905                        init($u);
 906                }
 907                exit 0;
 908        }
 909        waitpid $pid, 0;
 910        croak $? if $?;
 911}
 912
 913sub common_prefix {
 914        my $paths = shift;
 915        my %common;
 916        foreach (@$paths) {
 917                my @tmp = split m#/#, $_;
 918                my $p = '';
 919                while (my $x = shift @tmp) {
 920                        $p .= "/$x";
 921                        $common{$p} ||= 0;
 922                        $common{$p}++;
 923                }
 924        }
 925        foreach (sort {length $b <=> length $a} keys %common) {
 926                if ($common{$_} == @$paths) {
 927                        return $_;
 928                }
 929        }
 930        return '';
 931}
 932
 933# grafts set here are 'stronger' in that they're based on actual tree
 934# matches, and won't be deleted from merge-base checking in write_grafts()
 935sub graft_tree_joins {
 936        my $grafts = shift;
 937        map_tree_joins() if (@_branch_from && !%tree_map);
 938        return unless %tree_map;
 939
 940        git_svn_each(sub {
 941                my $i = shift;
 942                defined(my $pid = open my $fh, '-|') or croak $!;
 943                if (!$pid) {
 944                        exec qw/git-rev-list --pretty=raw/,
 945                                        "refs/remotes/$i" or croak $!;
 946                }
 947                while (<$fh>) {
 948                        next unless /^commit ($sha1)$/o;
 949                        my $c = $1;
 950                        my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
 951                        next unless $tree_map{$t};
 952
 953                        my $l;
 954                        do {
 955                                $l = readline $fh;
 956                        } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
 957
 958                        my ($s, $tz) = ($1, $2);
 959                        if ($tz =~ s/^\+//) {
 960                                $s += tz_to_s_offset($tz);
 961                        } elsif ($tz =~ s/^\-//) {
 962                                $s -= tz_to_s_offset($tz);
 963                        }
 964
 965                        my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
 966
 967                        foreach my $p (@{$tree_map{$t}}) {
 968                                next if $p eq $c;
 969                                my $mb = eval {
 970                                        safe_qx('git-merge-base', $c, $p)
 971                                };
 972                                next unless ($@ || $?);
 973                                if (defined $r_a) {
 974                                        # see if SVN says it's a relative
 975                                        my ($url_b, $r_b, $uuid_b) =
 976                                                        cmt_metadata($p);
 977                                        next if (defined $url_b &&
 978                                                        defined $url_a &&
 979                                                        ($url_a eq $url_b) &&
 980                                                        ($uuid_a eq $uuid_b));
 981                                        if ($uuid_a eq $uuid_b) {
 982                                                if ($r_b < $r_a) {
 983                                                        $grafts->{$c}->{$p} = 2;
 984                                                        next;
 985                                                } elsif ($r_b > $r_a) {
 986                                                        $grafts->{$p}->{$c} = 2;
 987                                                        next;
 988                                                }
 989                                        }
 990                                }
 991                                my $ct = get_commit_time($p);
 992                                if ($ct < $s) {
 993                                        $grafts->{$c}->{$p} = 2;
 994                                } elsif ($ct > $s) {
 995                                        $grafts->{$p}->{$c} = 2;
 996                                }
 997                                # what should we do when $ct == $s ?
 998                        }
 999                }
1000                close $fh or croak $?;
1001        });
1002}
1003
1004# this isn't funky-filename safe, but good enough for now...
1005sub graft_file_copy_cmd {
1006        my ($grafts, $l_map, $u) = @_;
1007        my $paths = $l_map->{$u};
1008        my $pfx = common_prefix([keys %$paths]);
1009        $SVN_URL ||= $u.$pfx;
1010        my $pid = open my $fh, '-|';
1011        defined $pid or croak $!;
1012        unless ($pid) {
1013                my @exec = qw/svn log -v/;
1014                push @exec, "-r$_revision" if defined $_revision;
1015                exec @exec, $u.$pfx or croak $!;
1016        }
1017        my ($r, $mp) = (undef, undef);
1018        while (<$fh>) {
1019                chomp;
1020                if (/^\-{72}$/) {
1021                        $mp = $r = undef;
1022                } elsif (/^r(\d+) \| /) {
1023                        $r = $1 unless defined $r;
1024                } elsif (/^Changed paths:/) {
1025                        $mp = 1;
1026                } elsif ($mp && m#^   [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) {
1027                        my ($p1, $p0, $r0) = ($1, $2, $3);
1028                        my $c = find_graft_path_commit($paths, $p1, $r);
1029                        next unless $c;
1030                        find_graft_path_parents($grafts, $paths, $c, $p0, $r0);
1031                }
1032        }
1033}
1034
1035sub graft_file_copy_lib {
1036        my ($grafts, $l_map, $u) = @_;
1037        my $tree_paths = $l_map->{$u};
1038        my $pfx = common_prefix([keys %$tree_paths]);
1039        my ($repo, $path) = repo_path_split($u.$pfx);
1040        $SVN_LOG ||= libsvn_connect($repo);
1041        $SVN ||= libsvn_connect($repo);
1042
1043        my ($base, $head) = libsvn_parse_revision();
1044        my $inc = 1000;
1045        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
1046        my $eh = $SVN::Error::handler;
1047        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
1048        while (1) {
1049                my $pool = SVN::Pool->new;
1050                libsvn_get_log($SVN_LOG, "/$path", $min, $max, 0, 1, 1,
1051                        sub {
1052                                libsvn_graft_file_copies($grafts, $tree_paths,
1053                                                        $path, @_);
1054                        }, $pool);
1055                $pool->clear;
1056                last if ($max >= $head);
1057                $min = $max + 1;
1058                $max += $inc;
1059                $max = $head if ($max > $head);
1060        }
1061        $SVN::Error::handler = $eh;
1062}
1063
1064sub process_merge_msg_matches {
1065        my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
1066        my (@strong, @weak);
1067        foreach (@matches) {
1068                # merging with ourselves is not interesting
1069                next if $_ eq $p;
1070                if ($l_map->{$u}->{$_}) {
1071                        push @strong, $_;
1072                } else {
1073                        push @weak, $_;
1074                }
1075        }
1076        foreach my $w (@weak) {
1077                last if @strong;
1078                # no exact match, use branch name as regexp.
1079                my $re = qr/\Q$w\E/i;
1080                foreach (keys %{$l_map->{$u}}) {
1081                        if (/$re/) {
1082                                push @strong, $l_map->{$u}->{$_};
1083                                last;
1084                        }
1085                }
1086                last if @strong;
1087                $w = basename($w);
1088                $re = qr/\Q$w\E/i;
1089                foreach (keys %{$l_map->{$u}}) {
1090                        if (/$re/) {
1091                                push @strong, $l_map->{$u}->{$_};
1092                                last;
1093                        }
1094                }
1095        }
1096        my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
1097                                        \s(?:[a-f\d\-]+)$/xsm);
1098        unless (defined $rev) {
1099                ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
1100                                        \@(?:[a-f\d\-]+)/xsm);
1101                return unless defined $rev;
1102        }
1103        foreach my $m (@strong) {
1104                my ($r0, $s0) = find_rev_before($rev, $m, 1);
1105                $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
1106        }
1107}
1108
1109sub graft_merge_msg {
1110        my ($grafts, $l_map, $u, $p, @re) = @_;
1111
1112        my $x = $l_map->{$u}->{$p};
1113        my $rl = rev_list_raw($x);
1114        while (my $c = next_rev_list_entry($rl)) {
1115                foreach my $re (@re) {
1116                        my (@br) = ($c->{m} =~ /$re/g);
1117                        next unless @br;
1118                        process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
1119                }
1120        }
1121}
1122
1123sub read_uuid {
1124        return if $SVN_UUID;
1125        if ($_use_lib) {
1126                my $pool = SVN::Pool->new;
1127                $SVN_UUID = $SVN->get_uuid($pool);
1128                $pool->clear;
1129        } else {
1130                my $info = shift || svn_info('.');
1131                $SVN_UUID = $info->{'Repository UUID'} or
1132                                        croak "Repository UUID unreadable\n";
1133        }
1134}
1135
1136sub quiet_run {
1137        my $pid = fork;
1138        defined $pid or croak $!;
1139        if (!$pid) {
1140                open my $null, '>', '/dev/null' or croak $!;
1141                open STDERR, '>&', $null or croak $!;
1142                open STDOUT, '>&', $null or croak $!;
1143                exec @_ or croak $!;
1144        }
1145        waitpid $pid, 0;
1146        return $?;
1147}
1148
1149sub repo_path_split {
1150        my $full_url = shift;
1151        $full_url =~ s#/+$##;
1152
1153        foreach (@repo_path_split_cache) {
1154                if ($full_url =~ s#$_##) {
1155                        my $u = $1;
1156                        $full_url =~ s#^/+##;
1157                        return ($u, $full_url);
1158                }
1159        }
1160
1161        my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i);
1162        $path =~ s#^/+##;
1163        my @paths = split(m#/+#, $path);
1164
1165        if ($_use_lib) {
1166                while (1) {
1167                        $SVN = libsvn_connect($url);
1168                        last if (defined $SVN &&
1169                                defined eval { $SVN->get_latest_revnum });
1170                        my $n = shift @paths || last;
1171                        $url .= "/$n";
1172                }
1173        } else {
1174                while (quiet_run(qw/svn ls --non-interactive/, $url)) {
1175                        my $n = shift @paths || last;
1176                        $url .= "/$n";
1177                }
1178        }
1179        push @repo_path_split_cache, qr/^(\Q$url\E)/;
1180        $path = join('/',@paths);
1181        return ($url, $path);
1182}
1183
1184sub setup_git_svn {
1185        defined $SVN_URL or croak "SVN repository location required\n";
1186        unless (-d $GIT_DIR) {
1187                croak "GIT_DIR=$GIT_DIR does not exist!\n";
1188        }
1189        mkpath([$GIT_SVN_DIR]);
1190        mkpath(["$GIT_SVN_DIR/info"]);
1191        open my $fh, '>>',$REVDB or croak $!;
1192        close $fh;
1193        s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1194
1195}
1196
1197sub assert_svn_wc_clean {
1198        return if $_use_lib;
1199        my ($svn_rev) = @_;
1200        croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
1201        my $lcr = svn_info('.')->{'Last Changed Rev'};
1202        if ($svn_rev != $lcr) {
1203                print STDERR "Checking for copy-tree ... ";
1204                my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
1205                                                "-r$lcr:$svn_rev")));
1206                if (@diff) {
1207                        croak "Nope!  Expected r$svn_rev, got r$lcr\n";
1208                } else {
1209                        print STDERR "OK!\n";
1210                }
1211        }
1212        my @status = grep(!/^Performing status on external/,(`svn status`));
1213        @status = grep(!/^\s*$/,@status);
1214        if (scalar @status) {
1215                print STDERR "Tree ($SVN_WC) is not clean:\n";
1216                print STDERR $_ foreach @status;
1217                croak;
1218        }
1219}
1220
1221sub get_tree_from_treeish {
1222        my ($treeish) = @_;
1223        croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1224        chomp(my $type = `git-cat-file -t $treeish`);
1225        my $expected;
1226        while ($type eq 'tag') {
1227                chomp(($treeish, $type) = `git-cat-file tag $treeish`);
1228        }
1229        if ($type eq 'commit') {
1230                $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
1231                ($expected) = ($expected =~ /^tree ($sha1)$/);
1232                die "Unable to get tree from $treeish\n" unless $expected;
1233        } elsif ($type eq 'tree') {
1234                $expected = $treeish;
1235        } else {
1236                die "$treeish is a $type, expected tree, tag or commit\n";
1237        }
1238        return $expected;
1239}
1240
1241sub assert_tree {
1242        return if $_use_lib;
1243        my ($treeish) = @_;
1244        my $expected = get_tree_from_treeish($treeish);
1245
1246        my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
1247        if (-e $tmpindex) {
1248                unlink $tmpindex or croak $!;
1249        }
1250        my $old_index = set_index($tmpindex);
1251        index_changes(1);
1252        chomp(my $tree = `git-write-tree`);
1253        restore_index($old_index);
1254        if ($tree ne $expected) {
1255                croak "Tree mismatch, Got: $tree, Expected: $expected\n";
1256        }
1257        unlink $tmpindex;
1258}
1259
1260sub parse_diff_tree {
1261        my $diff_fh = shift;
1262        local $/ = "\0";
1263        my $state = 'meta';
1264        my @mods;
1265        while (<$diff_fh>) {
1266                chomp $_; # this gets rid of the trailing "\0"
1267                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1268                                        $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1269                        push @mods, {   mode_a => $1, mode_b => $2,
1270                                        sha1_b => $3, chg => $4 };
1271                        if ($4 =~ /^(?:C|R)$/) {
1272                                $state = 'file_a';
1273                        } else {
1274                                $state = 'file_b';
1275                        }
1276                } elsif ($state eq 'file_a') {
1277                        my $x = $mods[$#mods] or croak "Empty array\n";
1278                        if ($x->{chg} !~ /^(?:C|R)$/) {
1279                                croak "Error parsing $_, $x->{chg}\n";
1280                        }
1281                        $x->{file_a} = $_;
1282                        $state = 'file_b';
1283                } elsif ($state eq 'file_b') {
1284                        my $x = $mods[$#mods] or croak "Empty array\n";
1285                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1286                                croak "Error parsing $_, $x->{chg}\n";
1287                        }
1288                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1289                                croak "Error parsing $_, $x->{chg}\n";
1290                        }
1291                        $x->{file_b} = $_;
1292                        $state = 'meta';
1293                } else {
1294                        croak "Error parsing $_\n";
1295                }
1296        }
1297        close $diff_fh or croak $?;
1298
1299        return \@mods;
1300}
1301
1302sub svn_check_prop_executable {
1303        my $m = shift;
1304        return if -l $m->{file_b};
1305        if ($m->{mode_b} =~ /755$/) {
1306                chmod((0755 &~ umask),$m->{file_b}) or croak $!;
1307                if ($m->{mode_a} !~ /755$/) {
1308                        sys(qw(svn propset svn:executable 1), $m->{file_b});
1309                }
1310                -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
1311        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
1312                sys(qw(svn propdel svn:executable), $m->{file_b});
1313                chmod((0644 &~ umask),$m->{file_b}) or croak $!;
1314                -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
1315        }
1316}
1317
1318sub svn_ensure_parent_path {
1319        my $dir_b = dirname(shift);
1320        svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
1321        mkpath([$dir_b]) unless (-d $dir_b);
1322        sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
1323}
1324
1325sub precommit_check {
1326        my $mods = shift;
1327        my (%rm_file, %rmdir_check, %added_check);
1328
1329        my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
1330        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1331                if ($m->{chg} eq 'R') {
1332                        if (-d $m->{file_b}) {
1333                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1334                        }
1335                        # dir/$file => dir/file/$file
1336                        my $dirname = dirname($m->{file_b});
1337                        while ($dirname ne File::Spec->curdir) {
1338                                if ($dirname ne $m->{file_a}) {
1339                                        $dirname = dirname($dirname);
1340                                        next;
1341                                }
1342                                err_file_to_dir("$m->{file_a} => $m->{file_b}");
1343                        }
1344                        # baz/zzz => baz (baz is a file)
1345                        $dirname = dirname($m->{file_a});
1346                        while ($dirname ne File::Spec->curdir) {
1347                                if ($dirname ne $m->{file_b}) {
1348                                        $dirname = dirname($dirname);
1349                                        next;
1350                                }
1351                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1352                        }
1353                }
1354                if ($m->{chg} =~ /^(D|R)$/) {
1355                        my $t = $1 eq 'D' ? 'file_b' : 'file_a';
1356                        $rm_file{ $m->{$t} } = 1;
1357                        my $dirname = dirname( $m->{$t} );
1358                        my $basename = basename( $m->{$t} );
1359                        $rmdir_check{$dirname}->{$basename} = 1;
1360                } elsif ($m->{chg} =~ /^(?:A|C)$/) {
1361                        if (-d $m->{file_b}) {
1362                                err_dir_to_file($m->{file_b});
1363                        }
1364                        my $dirname = dirname( $m->{file_b} );
1365                        my $basename = basename( $m->{file_b} );
1366                        $added_check{$dirname}->{$basename} = 1;
1367                        while ($dirname ne File::Spec->curdir) {
1368                                if ($rm_file{$dirname}) {
1369                                        err_file_to_dir($m->{file_b});
1370                                }
1371                                $dirname = dirname $dirname;
1372                        }
1373                }
1374        }
1375        return (\%rmdir_check, \%added_check);
1376
1377        sub err_dir_to_file {
1378                my $file = shift;
1379                print STDERR "Node change from directory to file ",
1380                                "is not supported by Subversion: ",$file,"\n";
1381                exit 1;
1382        }
1383        sub err_file_to_dir {
1384                my $file = shift;
1385                print STDERR "Node change from file to directory ",
1386                                "is not supported by Subversion: ",$file,"\n";
1387                exit 1;
1388        }
1389}
1390
1391
1392sub get_diff {
1393        my ($from, $treeish) = @_;
1394        assert_tree($from);
1395        print "diff-tree $from $treeish\n";
1396        my $pid = open my $diff_fh, '-|';
1397        defined $pid or croak $!;
1398        if ($pid == 0) {
1399                my @diff_tree = qw(git-diff-tree -z -r);
1400                if ($_cp_similarity) {
1401                        push @diff_tree, "-C$_cp_similarity";
1402                } else {
1403                        push @diff_tree, '-C';
1404                }
1405                push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1406                push @diff_tree, "-l$_l" if defined $_l;
1407                exec(@diff_tree, $from, $treeish) or croak $!;
1408        }
1409        return parse_diff_tree($diff_fh);
1410}
1411
1412sub svn_checkout_tree {
1413        my ($from, $treeish) = @_;
1414        my $mods = get_diff($from->{commit}, $treeish);
1415        return $mods unless (scalar @$mods);
1416        my ($rm, $add) = precommit_check($mods);
1417
1418        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1419        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1420                if ($m->{chg} eq 'C') {
1421                        svn_ensure_parent_path( $m->{file_b} );
1422                        sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
1423                        apply_mod_line_blob($m);
1424                        svn_check_prop_executable($m);
1425                } elsif ($m->{chg} eq 'D') {
1426                        sys(qw(svn rm --force), $m->{file_b});
1427                } elsif ($m->{chg} eq 'R') {
1428                        svn_ensure_parent_path( $m->{file_b} );
1429                        sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
1430                        apply_mod_line_blob($m);
1431                        svn_check_prop_executable($m);
1432                } elsif ($m->{chg} eq 'M') {
1433                        apply_mod_line_blob($m);
1434                        svn_check_prop_executable($m);
1435                } elsif ($m->{chg} eq 'T') {
1436                        sys(qw(svn rm --force),$m->{file_b});
1437                        apply_mod_line_blob($m);
1438                        sys(qw(svn add), $m->{file_b});
1439                        svn_check_prop_executable($m);
1440                } elsif ($m->{chg} eq 'A') {
1441                        svn_ensure_parent_path( $m->{file_b} );
1442                        apply_mod_line_blob($m);
1443                        sys(qw(svn add), $m->{file_b});
1444                        svn_check_prop_executable($m);
1445                } else {
1446                        croak "Invalid chg: $m->{chg}\n";
1447                }
1448        }
1449
1450        assert_tree($treeish);
1451        if ($_rmdir) { # remove empty directories
1452                handle_rmdir($rm, $add);
1453        }
1454        assert_tree($treeish);
1455        return $mods;
1456}
1457
1458sub libsvn_checkout_tree {
1459        my ($from, $treeish, $ed) = @_;
1460        my $mods = get_diff($from, $treeish);
1461        return $mods unless (scalar @$mods);
1462        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1463        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1464                my $f = $m->{chg};
1465                if (defined $o{$f}) {
1466                        $ed->$f($m, $_q);
1467                } else {
1468                        croak "Invalid change type: $f\n";
1469                }
1470        }
1471        $ed->rmdirs($_q) if $_rmdir;
1472        return $mods;
1473}
1474
1475# svn ls doesn't work with respect to the current working tree, but what's
1476# in the repository.  There's not even an option for it... *sigh*
1477# (added files don't show up and removed files remain in the ls listing)
1478sub svn_ls_current {
1479        my ($dir, $rm, $add) = @_;
1480        chomp(my @ls = safe_qx('svn','ls',$dir));
1481        my @ret = ();
1482        foreach (@ls) {
1483                s#/$##; # trailing slashes are evil
1484                push @ret, $_ unless $rm->{$dir}->{$_};
1485        }
1486        if (exists $add->{$dir}) {
1487                push @ret, keys %{$add->{$dir}};
1488        }
1489        return \@ret;
1490}
1491
1492sub handle_rmdir {
1493        my ($rm, $add) = @_;
1494
1495        foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
1496                my $ls = svn_ls_current($dir, $rm, $add);
1497                next if (scalar @$ls);
1498                sys(qw(svn rm --force),$dir);
1499
1500                my $dn = dirname $dir;
1501                $rm->{ $dn }->{ basename $dir } = 1;
1502                $ls = svn_ls_current($dn, $rm, $add);
1503                while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
1504                        sys(qw(svn rm --force),$dn);
1505                        $dir = basename $dn;
1506                        $dn = dirname $dn;
1507                        $rm->{ $dn }->{ $dir } = 1;
1508                        $ls = svn_ls_current($dn, $rm, $add);
1509                }
1510        }
1511}
1512
1513sub get_commit_message {
1514        my ($commit, $commit_msg) = (@_);
1515        my %log_msg = ( msg => '' );
1516        open my $msg, '>', $commit_msg or croak $!;
1517
1518        chomp(my $type = `git-cat-file -t $commit`);
1519        if ($type eq 'commit' || $type eq 'tag') {
1520                my $pid = open my $msg_fh, '-|';
1521                defined $pid or croak $!;
1522
1523                if ($pid == 0) {
1524                        exec('git-cat-file', $type, $commit) or croak $!;
1525                }
1526                my $in_msg = 0;
1527                while (<$msg_fh>) {
1528                        if (!$in_msg) {
1529                                $in_msg = 1 if (/^\s*$/);
1530                        } elsif (/^git-svn-id: /) {
1531                                # skip this, we regenerate the correct one
1532                                # on re-fetch anyways
1533                        } else {
1534                                print $msg $_ or croak $!;
1535                        }
1536                }
1537                close $msg_fh or croak $?;
1538        }
1539        close $msg or croak $!;
1540
1541        if ($_edit || ($type eq 'tree')) {
1542                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1543                system($editor, $commit_msg);
1544        }
1545
1546        # file_to_s removes all trailing newlines, so just use chomp() here:
1547        open $msg, '<', $commit_msg or croak $!;
1548        { local $/; chomp($log_msg{msg} = <$msg>); }
1549        close $msg or croak $!;
1550
1551        return \%log_msg;
1552}
1553
1554sub set_svn_commit_env {
1555        if (defined $LC_ALL) {
1556                $ENV{LC_ALL} = $LC_ALL;
1557        } else {
1558                delete $ENV{LC_ALL};
1559        }
1560}
1561
1562sub svn_commit_tree {
1563        my ($last, $commit) = @_;
1564        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
1565        my $log_msg = get_commit_message($commit, $commit_msg);
1566        my ($oneline) = ($log_msg->{msg} =~ /([^\n\r]+)/);
1567        print "Committing $commit: $oneline\n";
1568
1569        set_svn_commit_env();
1570        my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
1571        $ENV{LC_ALL} = 'C';
1572        unlink $commit_msg;
1573        my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/);
1574        if (!defined $committed) {
1575                my $out = join("\n",@ci_output);
1576                print STDERR "W: Trouble parsing \`svn commit' output:\n\n",
1577                                $out, "\n\nAssuming English locale...";
1578                ($committed) = ($out =~ /^Committed revision \d+\./sm);
1579                defined $committed or die " FAILED!\n",
1580                        "Commit output failed to parse committed revision!\n",
1581                print STDERR " OK\n";
1582        }
1583
1584        my @svn_up = qw(svn up);
1585        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
1586        if ($_optimize_commits && ($committed == ($last->{revision} + 1))) {
1587                push @svn_up, "-r$committed";
1588                sys(@svn_up);
1589                my $info = svn_info('.');
1590                my $date = $info->{'Last Changed Date'} or die "Missing date\n";
1591                if ($info->{'Last Changed Rev'} != $committed) {
1592                        croak "$info->{'Last Changed Rev'} != $committed\n"
1593                }
1594                my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1595                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1596                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1597                                         or croak "Failed to parse date: $date\n";
1598                $log_msg->{date} = "$tz $Y-$m-$d $H:$M:$S";
1599                $log_msg->{author} = $info->{'Last Changed Author'};
1600                $log_msg->{revision} = $committed;
1601                $log_msg->{msg} .= "\n";
1602                $log_msg->{parents} = [ $last->{commit} ];
1603                $log_msg->{commit} = git_commit($log_msg, $commit);
1604                return $log_msg;
1605        }
1606        # resync immediately
1607        push @svn_up, "-r$last->{revision}";
1608        sys(@svn_up);
1609        return fetch("$committed=$commit");
1610}
1611
1612sub rev_list_raw {
1613        my (@args) = @_;
1614        my $pid = open my $fh, '-|';
1615        defined $pid or croak $!;
1616        if (!$pid) {
1617                exec(qw/git-rev-list --pretty=raw/, @args) or croak $!;
1618        }
1619        return { fh => $fh, t => { } };
1620}
1621
1622sub next_rev_list_entry {
1623        my $rl = shift;
1624        my $fh = $rl->{fh};
1625        my $x = $rl->{t};
1626        while (<$fh>) {
1627                if (/^commit ($sha1)$/o) {
1628                        if ($x->{c}) {
1629                                $rl->{t} = { c => $1 };
1630                                return $x;
1631                        } else {
1632                                $x->{c} = $1;
1633                        }
1634                } elsif (/^parent ($sha1)$/o) {
1635                        $x->{p}->{$1} = 1;
1636                } elsif (s/^    //) {
1637                        $x->{m} ||= '';
1638                        $x->{m} .= $_;
1639                }
1640        }
1641        return ($x != $rl->{t}) ? $x : undef;
1642}
1643
1644# read the entire log into a temporary file (which is removed ASAP)
1645# and store the file handle + parser state
1646sub svn_log_raw {
1647        my (@log_args) = @_;
1648        my $log_fh = IO::File->new_tmpfile or croak $!;
1649        my $pid = fork;
1650        defined $pid or croak $!;
1651        if (!$pid) {
1652                open STDOUT, '>&', $log_fh or croak $!;
1653                exec (qw(svn log), @log_args) or croak $!
1654        }
1655        waitpid $pid, 0;
1656        croak $? if $?;
1657        seek $log_fh, 0, 0 or croak $!;
1658        return { state => 'sep', fh => $log_fh };
1659}
1660
1661sub next_log_entry {
1662        my $log = shift; # retval of svn_log_raw()
1663        my $ret = undef;
1664        my $fh = $log->{fh};
1665
1666        while (<$fh>) {
1667                chomp;
1668                if (/^\-{72}$/) {
1669                        if ($log->{state} eq 'msg') {
1670                                if ($ret->{lines}) {
1671                                        $ret->{msg} .= $_."\n";
1672                                        unless(--$ret->{lines}) {
1673                                                $log->{state} = 'sep';
1674                                        }
1675                                } else {
1676                                        croak "Log parse error at: $_\n",
1677                                                $ret->{revision},
1678                                                "\n";
1679                                }
1680                                next;
1681                        }
1682                        if ($log->{state} ne 'sep') {
1683                                croak "Log parse error at: $_\n",
1684                                        "state: $log->{state}\n",
1685                                        $ret->{revision},
1686                                        "\n";
1687                        }
1688                        $log->{state} = 'rev';
1689
1690                        # if we have an empty log message, put something there:
1691                        if ($ret) {
1692                                $ret->{msg} ||= "\n";
1693                                delete $ret->{lines};
1694                                return $ret;
1695                        }
1696                        next;
1697                }
1698                if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
1699                        my $rev = $1;
1700                        my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
1701                        ($lines) = ($lines =~ /(\d+)/);
1702                        my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1703                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1704                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1705                                         or croak "Failed to parse date: $date\n";
1706                        $ret = {        revision => $rev,
1707                                        date => "$tz $Y-$m-$d $H:$M:$S",
1708                                        author => $author,
1709                                        lines => $lines,
1710                                        msg => '' };
1711                        if (defined $_authors && ! defined $users{$author}) {
1712                                die "Author: $author not defined in ",
1713                                                "$_authors file\n";
1714                        }
1715                        $log->{state} = 'msg_start';
1716                        next;
1717                }
1718                # skip the first blank line of the message:
1719                if ($log->{state} eq 'msg_start' && /^$/) {
1720                        $log->{state} = 'msg';
1721                } elsif ($log->{state} eq 'msg') {
1722                        if ($ret->{lines}) {
1723                                $ret->{msg} .= $_."\n";
1724                                unless (--$ret->{lines}) {
1725                                        $log->{state} = 'sep';
1726                                }
1727                        } else {
1728                                croak "Log parse error at: $_\n",
1729                                        $ret->{revision},"\n";
1730                        }
1731                }
1732        }
1733        return $ret;
1734}
1735
1736sub svn_info {
1737        my $url = shift || $SVN_URL;
1738
1739        my $pid = open my $info_fh, '-|';
1740        defined $pid or croak $!;
1741
1742        if ($pid == 0) {
1743                exec(qw(svn info),$url) or croak $!;
1744        }
1745
1746        my $ret = {};
1747        # only single-lines seem to exist in svn info output
1748        while (<$info_fh>) {
1749                chomp $_;
1750                if (m#^([^:]+)\s*:\s*(\S.*)$#) {
1751                        $ret->{$1} = $2;
1752                        push @{$ret->{-order}}, $1;
1753                }
1754        }
1755        close $info_fh or croak $?;
1756        return $ret;
1757}
1758
1759sub sys { system(@_) == 0 or croak $? }
1760
1761sub do_update_index {
1762        my ($z_cmd, $cmd, $no_text_base) = @_;
1763
1764        my $z = open my $p, '-|';
1765        defined $z or croak $!;
1766        unless ($z) { exec @$z_cmd or croak $! }
1767
1768        my $pid = open my $ui, '|-';
1769        defined $pid or croak $!;
1770        unless ($pid) {
1771                exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!;
1772        }
1773        local $/ = "\0";
1774        while (my $x = <$p>) {
1775                chomp $x;
1776                if (!$no_text_base && lstat $x && ! -l _ &&
1777                                svn_propget_base('svn:keywords', $x)) {
1778                        my $mode = -x _ ? 0755 : 0644;
1779                        my ($v,$d,$f) = File::Spec->splitpath($x);
1780                        my $tb = File::Spec->catfile($d, '.svn', 'tmp',
1781                                                'text-base',"$f.svn-base");
1782                        $tb =~ s#^/##;
1783                        unless (-f $tb) {
1784                                $tb = File::Spec->catfile($d, '.svn',
1785                                                'text-base',"$f.svn-base");
1786                                $tb =~ s#^/##;
1787                        }
1788                        my @s = stat($x);
1789                        unlink $x or croak $!;
1790                        copy($tb, $x);
1791                        chmod(($mode &~ umask), $x) or croak $!;
1792                        utime $s[8], $s[9], $x;
1793                }
1794                print $ui $x,"\0";
1795        }
1796        close $ui or croak $?;
1797}
1798
1799sub index_changes {
1800        return if $_use_lib;
1801
1802        if (!-f "$GIT_SVN_DIR/info/exclude") {
1803                open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!;
1804                print $fd '.svn',"\n";
1805                close $fd or croak $!;
1806        }
1807        my $no_text_base = shift;
1808        do_update_index([qw/git-diff-files --name-only -z/],
1809                        'remove',
1810                        $no_text_base);
1811        do_update_index([qw/git-ls-files -z --others/,
1812                                "--exclude-from=$GIT_SVN_DIR/info/exclude"],
1813                        'add',
1814                        $no_text_base);
1815}
1816
1817sub s_to_file {
1818        my ($str, $file, $mode) = @_;
1819        open my $fd,'>',$file or croak $!;
1820        print $fd $str,"\n" or croak $!;
1821        close $fd or croak $!;
1822        chmod ($mode &~ umask, $file) if (defined $mode);
1823}
1824
1825sub file_to_s {
1826        my $file = shift;
1827        open my $fd,'<',$file or croak "$!: file: $file\n";
1828        local $/;
1829        my $ret = <$fd>;
1830        close $fd or croak $!;
1831        $ret =~ s/\s*$//s;
1832        return $ret;
1833}
1834
1835sub assert_revision_unknown {
1836        my $r = shift;
1837        if (my $c = revdb_get($REVDB, $r)) {
1838                croak "$r = $c already exists! Why are we refetching it?";
1839        }
1840}
1841
1842sub trees_eq {
1843        my ($x, $y) = @_;
1844        my @x = safe_qx('git-cat-file','commit',$x);
1845        my @y = safe_qx('git-cat-file','commit',$y);
1846        if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
1847                                || $y[0] !~ /^tree $sha1\n$/) {
1848                print STDERR "Trees not equal: $y[0] != $x[0]\n";
1849                return 0
1850        }
1851        return 1;
1852}
1853
1854sub git_commit {
1855        my ($log_msg, @parents) = @_;
1856        assert_revision_unknown($log_msg->{revision});
1857        map_tree_joins() if (@_branch_from && !%tree_map);
1858
1859        my (@tmp_parents, @exec_parents, %seen_parent);
1860        if (my $lparents = $log_msg->{parents}) {
1861                @tmp_parents = @$lparents
1862        }
1863        # commit parents can be conditionally bound to a particular
1864        # svn revision via: "svn_revno=commit_sha1", filter them out here:
1865        foreach my $p (@parents) {
1866                next unless defined $p;
1867                if ($p =~ /^(\d+)=($sha1_short)$/o) {
1868                        if ($1 == $log_msg->{revision}) {
1869                                push @tmp_parents, $2;
1870                        }
1871                } else {
1872                        push @tmp_parents, $p if $p =~ /$sha1_short/o;
1873                }
1874        }
1875        my $tree = $log_msg->{tree};
1876        if (!defined $tree) {
1877                my $index = set_index($GIT_SVN_INDEX);
1878                index_changes();
1879                chomp($tree = `git-write-tree`);
1880                croak $? if $?;
1881                restore_index($index);
1882        }
1883
1884        # just in case we clobber the existing ref, we still want that ref
1885        # as our parent:
1886        if (my $cur = eval { file_to_s("$GIT_DIR/refs/remotes/$GIT_SVN") }) {
1887                push @tmp_parents, $cur;
1888        }
1889
1890        if (exists $tree_map{$tree}) {
1891                foreach my $p (@{$tree_map{$tree}}) {
1892                        my $skip;
1893                        foreach (@tmp_parents) {
1894                                # see if a common parent is found
1895                                my $mb = eval {
1896                                        safe_qx('git-merge-base', $_, $p)
1897                                };
1898                                next if ($@ || $?);
1899                                $skip = 1;
1900                                last;
1901                        }
1902                        next if $skip;
1903                        my ($url_p, $r_p, $uuid_p) = cmt_metadata($p);
1904                        next if (($SVN_UUID eq $uuid_p) &&
1905                                                ($log_msg->{revision} > $r_p));
1906                        next if (defined $url_p && defined $SVN_URL &&
1907                                                ($SVN_UUID eq $uuid_p) &&
1908                                                ($url_p eq $SVN_URL));
1909                        push @tmp_parents, $p;
1910                }
1911        }
1912        foreach (@tmp_parents) {
1913                next if $seen_parent{$_};
1914                $seen_parent{$_} = 1;
1915                push @exec_parents, $_;
1916                # MAXPARENT is defined to 16 in commit-tree.c:
1917                last if @exec_parents > 16;
1918        }
1919
1920        set_commit_env($log_msg);
1921        my @exec = ('git-commit-tree', $tree);
1922        push @exec, '-p', $_  foreach @exec_parents;
1923        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1924                                                                or croak $!;
1925        print $msg_fh $log_msg->{msg} or croak $!;
1926        unless ($_no_metadata) {
1927                print $msg_fh "\ngit-svn-id: $SVN_URL\@$log_msg->{revision}",
1928                                        " $SVN_UUID\n" or croak $!;
1929        }
1930        $msg_fh->flush == 0 or croak $!;
1931        close $msg_fh or croak $!;
1932        chomp(my $commit = do { local $/; <$out_fh> });
1933        close $out_fh or croak $!;
1934        waitpid $pid, 0;
1935        croak $? if $?;
1936        if ($commit !~ /^$sha1$/o) {
1937                die "Failed to commit, invalid sha1: $commit\n";
1938        }
1939        sys('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
1940        revdb_set($REVDB, $log_msg->{revision}, $commit);
1941
1942        # this output is read via pipe, do not change:
1943        print "r$log_msg->{revision} = $commit\n";
1944        check_repack();
1945        return $commit;
1946}
1947
1948sub check_repack {
1949        if ($_repack && (--$_repack_nr == 0)) {
1950                $_repack_nr = $_repack;
1951                sys("git repack $_repack_flags");
1952        }
1953}
1954
1955sub set_commit_env {
1956        my ($log_msg) = @_;
1957        my $author = $log_msg->{author};
1958        if (!defined $author || length $author == 0) {
1959                $author = '(no author)';
1960        }
1961        my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
1962                                : ($author,"$author\@$SVN_UUID");
1963        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1964        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1965        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
1966}
1967
1968sub apply_mod_line_blob {
1969        my $m = shift;
1970        if ($m->{mode_b} =~ /^120/) {
1971                blob_to_symlink($m->{sha1_b}, $m->{file_b});
1972        } else {
1973                blob_to_file($m->{sha1_b}, $m->{file_b});
1974        }
1975}
1976
1977sub blob_to_symlink {
1978        my ($blob, $link) = @_;
1979        defined $link or croak "\$link not defined!\n";
1980        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
1981        if (-l $link || -f _) {
1982                unlink $link or croak $!;
1983        }
1984
1985        my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
1986        symlink $dest, $link or croak $!;
1987}
1988
1989sub blob_to_file {
1990        my ($blob, $file) = @_;
1991        defined $file or croak "\$file not defined!\n";
1992        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
1993        if (-l $file || -f _) {
1994                unlink $file or croak $!;
1995        }
1996
1997        open my $blob_fh, '>', $file or croak "$!: $file\n";
1998        my $pid = fork;
1999        defined $pid or croak $!;
2000
2001        if ($pid == 0) {
2002                open STDOUT, '>&', $blob_fh or croak $!;
2003                exec('git-cat-file','blob',$blob) or croak $!;
2004        }
2005        waitpid $pid, 0;
2006        croak $? if $?;
2007
2008        close $blob_fh or croak $!;
2009}
2010
2011sub safe_qx {
2012        my $pid = open my $child, '-|';
2013        defined $pid or croak $!;
2014        if ($pid == 0) {
2015                exec(@_) or croak $!;
2016        }
2017        my @ret = (<$child>);
2018        close $child or croak $?;
2019        die $? if $?; # just in case close didn't error out
2020        return wantarray ? @ret : join('',@ret);
2021}
2022
2023sub svn_compat_check {
2024        if ($_follow_parent) {
2025                print STDERR 'E: --follow-parent functionality is only ',
2026                                "available when SVN libraries are used\n";
2027                exit 1;
2028        }
2029        my @co_help = safe_qx(qw(svn co -h));
2030        unless (grep /ignore-externals/,@co_help) {
2031                print STDERR "W: Installed svn version does not support ",
2032                                "--ignore-externals\n";
2033                $_no_ignore_ext = 1;
2034        }
2035        if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
2036                $_svn_co_url_revs = 1;
2037        }
2038        if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) {
2039                $_svn_pg_peg_revs = 1;
2040        }
2041
2042        # I really, really hope nobody hits this...
2043        unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
2044                print STDERR <<'';
2045W: The installed svn version does not support the --stop-on-copy flag in
2046   the log command.
2047   Lets hope the directory you're tracking is not a branch or tag
2048   and was never moved within the repository...
2049
2050                $_no_stop_copy = 1;
2051        }
2052}
2053
2054# *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
2055# (and they won't honor URL@<rev> without -r<rev>, too!)
2056sub svn_cmd_checkout {
2057        my ($url, $rev, $dir) = @_;
2058        my @cmd = ('svn','co', "-r$rev");
2059        push @cmd, '--ignore-externals' unless $_no_ignore_ext;
2060        $url .= "\@$rev" if $_svn_co_url_revs;
2061        sys(@cmd, $url, $dir);
2062}
2063
2064sub check_upgrade_needed {
2065        if (!-r $REVDB) {
2066                -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2067                open my $fh, '>>',$REVDB or croak $!;
2068                close $fh;
2069        }
2070        my $old = eval {
2071                my $pid = open my $child, '-|';
2072                defined $pid or croak $!;
2073                if ($pid == 0) {
2074                        close STDERR;
2075                        exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $!;
2076                }
2077                my @ret = (<$child>);
2078                close $child or croak $?;
2079                die $? if $?; # just in case close didn't error out
2080                return wantarray ? @ret : join('',@ret);
2081        };
2082        return unless $old;
2083        my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
2084        if ($@ || !$head) {
2085                print STDERR "Please run: $0 rebuild --upgrade\n";
2086                exit 1;
2087        }
2088}
2089
2090# fills %tree_map with a reverse mapping of trees to commits.  Useful
2091# for finding parents to commit on.
2092sub map_tree_joins {
2093        my %seen;
2094        foreach my $br (@_branch_from) {
2095                my $pid = open my $pipe, '-|';
2096                defined $pid or croak $!;
2097                if ($pid == 0) {
2098                        exec(qw(git-rev-list --topo-order --pretty=raw), $br)
2099                                                                or croak $!;
2100                }
2101                while (<$pipe>) {
2102                        if (/^commit ($sha1)$/o) {
2103                                my $commit = $1;
2104
2105                                # if we've seen a commit,
2106                                # we've seen its parents
2107                                last if $seen{$commit};
2108                                my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
2109                                unless (defined $tree) {
2110                                        die "Failed to parse commit $commit\n";
2111                                }
2112                                push @{$tree_map{$tree}}, $commit;
2113                                $seen{$commit} = 1;
2114                        }
2115                }
2116                close $pipe; # we could be breaking the pipe early
2117        }
2118}
2119
2120sub load_all_refs {
2121        if (@_branch_from) {
2122                print STDERR '--branch|-b parameters are ignored when ',
2123                        "--branch-all-refs|-B is passed\n";
2124        }
2125
2126        # don't worry about rev-list on non-commit objects/tags,
2127        # it shouldn't blow up if a ref is a blob or tree...
2128        chomp(@_branch_from = `git-rev-parse --symbolic --all`);
2129}
2130
2131# '<svn username> = real-name <email address>' mapping based on git-svnimport:
2132sub load_authors {
2133        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2134        while (<$authors>) {
2135                chomp;
2136                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2137                my ($user, $name, $email) = ($1, $2, $3);
2138                $users{$user} = [$name, $email];
2139        }
2140        close $authors or croak $!;
2141}
2142
2143sub rload_authors {
2144        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2145        while (<$authors>) {
2146                chomp;
2147                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2148                my ($user, $name, $email) = ($1, $2, $3);
2149                $rusers{"$name <$email>"} = $user;
2150        }
2151        close $authors or croak $!;
2152}
2153
2154sub svn_propget_base {
2155        my ($p, $f) = @_;
2156        $f .= '@BASE' if $_svn_pg_peg_revs;
2157        return safe_qx(qw/svn propget/, $p, $f);
2158}
2159
2160sub git_svn_each {
2161        my $sub = shift;
2162        foreach (`git-rev-parse --symbolic --all`) {
2163                next unless s#^refs/remotes/##;
2164                chomp $_;
2165                next unless -f "$GIT_DIR/svn/$_/info/url";
2166                &$sub($_);
2167        }
2168}
2169
2170sub migrate_revdb {
2171        git_svn_each(sub {
2172                my $id = shift;
2173                defined(my $pid = fork) or croak $!;
2174                if (!$pid) {
2175                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2176                        init_vars();
2177                        exit 0 if -r $REVDB;
2178                        print "Upgrading svn => git mapping...\n";
2179                        -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2180                        open my $fh, '>>',$REVDB or croak $!;
2181                        close $fh;
2182                        rebuild();
2183                        print "Done upgrading. You may now delete the ",
2184                                "deprecated $GIT_SVN_DIR/revs directory\n";
2185                        exit 0;
2186                }
2187                waitpid $pid, 0;
2188                croak $? if $?;
2189        });
2190}
2191
2192sub migration_check {
2193        migrate_revdb() unless (-e $REVDB);
2194        return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
2195        print "Upgrading repository...\n";
2196        unless (-d "$GIT_DIR/svn") {
2197                mkdir "$GIT_DIR/svn" or croak $!;
2198        }
2199        print "Data from a previous version of git-svn exists, but\n\t",
2200                                "$GIT_SVN_DIR\n\t(required for this version ",
2201                                "($VERSION) of git-svn) does not.\n";
2202
2203        foreach my $x (`git-rev-parse --symbolic --all`) {
2204                next unless $x =~ s#^refs/remotes/##;
2205                chomp $x;
2206                next unless -f "$GIT_DIR/$x/info/url";
2207                my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
2208                next unless $u;
2209                my $dn = dirname("$GIT_DIR/svn/$x");
2210                mkpath([$dn]) unless -d $dn;
2211                rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
2212        }
2213        migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
2214        print "Done upgrading.\n";
2215}
2216
2217sub find_rev_before {
2218        my ($r, $id, $eq_ok) = @_;
2219        my $f = "$GIT_DIR/svn/$id/.rev_db";
2220        return (undef,undef) unless -r $f;
2221        --$r unless $eq_ok;
2222        while ($r > 0) {
2223                if (my $c = revdb_get($f, $r)) {
2224                        return ($r, $c);
2225                }
2226                --$r;
2227        }
2228        return (undef, undef);
2229}
2230
2231sub init_vars {
2232        $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
2233        $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
2234        $REVDB = "$GIT_SVN_DIR/.rev_db";
2235        $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
2236        $SVN_URL = undef;
2237        $SVN_WC = "$GIT_SVN_DIR/tree";
2238        %tree_map = ();
2239}
2240
2241# convert GetOpt::Long specs for use by git-repo-config
2242sub read_repo_config {
2243        return unless -d $GIT_DIR;
2244        my $opts = shift;
2245        foreach my $o (keys %$opts) {
2246                my $v = $opts->{$o};
2247                my ($key) = ($o =~ /^([a-z\-]+)/);
2248                $key =~ s/-//g;
2249                my $arg = 'git-repo-config';
2250                $arg .= ' --int' if ($o =~ /[:=]i$/);
2251                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
2252                if (ref $v eq 'ARRAY') {
2253                        chomp(my @tmp = `$arg --get-all svn.$key`);
2254                        @$v = @tmp if @tmp;
2255                } else {
2256                        chomp(my $tmp = `$arg --get svn.$key`);
2257                        if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
2258                                $$v = $tmp;
2259                        }
2260                }
2261        }
2262}
2263
2264sub set_default_vals {
2265        if (defined $_repack) {
2266                $_repack = 1000 if ($_repack <= 0);
2267                $_repack_nr = $_repack;
2268                $_repack_flags ||= '-d';
2269        }
2270}
2271
2272sub read_grafts {
2273        my $gr_file = shift;
2274        my ($grafts, $comments) = ({}, {});
2275        if (open my $fh, '<', $gr_file) {
2276                my @tmp;
2277                while (<$fh>) {
2278                        if (/^($sha1)\s+/) {
2279                                my $c = $1;
2280                                if (@tmp) {
2281                                        @{$comments->{$c}} = @tmp;
2282                                        @tmp = ();
2283                                }
2284                                foreach my $p (split /\s+/, $_) {
2285                                        $grafts->{$c}->{$p} = 1;
2286                                }
2287                        } else {
2288                                push @tmp, $_;
2289                        }
2290                }
2291                close $fh or croak $!;
2292                @{$comments->{'END'}} = @tmp if @tmp;
2293        }
2294        return ($grafts, $comments);
2295}
2296
2297sub write_grafts {
2298        my ($grafts, $comments, $gr_file) = @_;
2299
2300        open my $fh, '>', $gr_file or croak $!;
2301        foreach my $c (sort keys %$grafts) {
2302                if ($comments->{$c}) {
2303                        print $fh $_ foreach @{$comments->{$c}};
2304                }
2305                my $p = $grafts->{$c};
2306                my %x; # real parents
2307                delete $p->{$c}; # commits are not self-reproducing...
2308                my $pid = open my $ch, '-|';
2309                defined $pid or croak $!;
2310                if (!$pid) {
2311                        exec(qw/git-cat-file commit/, $c) or croak $!;
2312                }
2313                while (<$ch>) {
2314                        if (/^parent ($sha1)/) {
2315                                $x{$1} = $p->{$1} = 1;
2316                        } else {
2317                                last unless /^\S/;
2318                        }
2319                }
2320                close $ch; # breaking the pipe
2321
2322                # if real parents are the only ones in the grafts, drop it
2323                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2324
2325                my (@ip, @jp, $mb);
2326                my %del = %x;
2327                @ip = @jp = keys %$p;
2328                foreach my $i (@ip) {
2329                        next if $del{$i} || $p->{$i} == 2;
2330                        foreach my $j (@jp) {
2331                                next if $i eq $j || $del{$j} || $p->{$j} == 2;
2332                                $mb = eval { safe_qx('git-merge-base',$i,$j) };
2333                                next unless $mb;
2334                                chomp $mb;
2335                                next if $x{$mb};
2336                                if ($mb eq $j) {
2337                                        delete $p->{$i};
2338                                        $del{$i} = 1;
2339                                } elsif ($mb eq $i) {
2340                                        delete $p->{$j};
2341                                        $del{$j} = 1;
2342                                }
2343                        }
2344                }
2345
2346                # if real parents are the only ones in the grafts, drop it
2347                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2348
2349                print $fh $c, ' ', join(' ', sort keys %$p),"\n";
2350        }
2351        if ($comments->{'END'}) {
2352                print $fh $_ foreach @{$comments->{'END'}};
2353        }
2354        close $fh or croak $!;
2355}
2356
2357sub read_url_paths_all {
2358        my ($l_map, $pfx, $p) = @_;
2359        my @dir;
2360        foreach (<$p/*>) {
2361                if (-r "$_/info/url") {
2362                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2363                        my $id = $pfx . basename $_;
2364                        my $url = file_to_s("$_/info/url");
2365                        my ($u, $p) = repo_path_split($url);
2366                        $l_map->{$u}->{$p} = $id;
2367                } elsif (-d $_) {
2368                        push @dir, $_;
2369                }
2370        }
2371        foreach (@dir) {
2372                my $x = $_;
2373                $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
2374                read_url_paths_all($l_map, $x, $_);
2375        }
2376}
2377
2378# this one only gets ids that have been imported, not new ones
2379sub read_url_paths {
2380        my $l_map = {};
2381        git_svn_each(sub { my $x = shift;
2382                        my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
2383                        my ($u, $p) = repo_path_split($url);
2384                        $l_map->{$u}->{$p} = $x;
2385                        });
2386        return $l_map;
2387}
2388
2389sub extract_metadata {
2390        my $id = shift or return (undef, undef, undef);
2391        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
2392                                                        \s([a-f\d\-]+)$/x);
2393        if (!$rev || !$uuid || !$url) {
2394                # some of the original repositories I made had
2395                # identifiers like this:
2396                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
2397        }
2398        return ($url, $rev, $uuid);
2399}
2400
2401sub cmt_metadata {
2402        return extract_metadata((grep(/^git-svn-id: /,
2403                safe_qx(qw/git-cat-file commit/, shift)))[-1]);
2404}
2405
2406sub get_commit_time {
2407        my $cmt = shift;
2408        defined(my $pid = open my $fh, '-|') or croak $!;
2409        if (!$pid) {
2410                exec qw/git-rev-list --pretty=raw -n1/, $cmt or croak $!;
2411        }
2412        while (<$fh>) {
2413                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
2414                my ($s, $tz) = ($1, $2);
2415                if ($tz =~ s/^\+//) {
2416                        $s += tz_to_s_offset($tz);
2417                } elsif ($tz =~ s/^\-//) {
2418                        $s -= tz_to_s_offset($tz);
2419                }
2420                close $fh;
2421                return $s;
2422        }
2423        die "Can't get commit time for commit: $cmt\n";
2424}
2425
2426sub tz_to_s_offset {
2427        my ($tz) = @_;
2428        $tz =~ s/(\d\d)$//;
2429        return ($1 * 60) + ($tz * 3600);
2430}
2431
2432sub setup_pager { # translated to Perl from pager.c
2433        return unless (-t *STDOUT);
2434        my $pager = $ENV{PAGER};
2435        if (!defined $pager) {
2436                $pager = 'less';
2437        } elsif (length $pager == 0 || $pager eq 'cat') {
2438                return;
2439        }
2440        pipe my $rfd, my $wfd or return;
2441        defined(my $pid = fork) or croak $!;
2442        if (!$pid) {
2443                open STDOUT, '>&', $wfd or croak $!;
2444                return;
2445        }
2446        open STDIN, '<&', $rfd or croak $!;
2447        $ENV{LESS} ||= '-S';
2448        exec $pager or croak "Can't run pager: $!\n";;
2449}
2450
2451sub get_author_info {
2452        my ($dest, $author, $t, $tz) = @_;
2453        $author =~ s/(?:^\s*|\s*$)//g;
2454        $dest->{a_raw} = $author;
2455        my $_a;
2456        if ($_authors) {
2457                $_a = $rusers{$author} || undef;
2458        }
2459        if (!$_a) {
2460                ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
2461        }
2462        $dest->{t} = $t;
2463        $dest->{tz} = $tz;
2464        $dest->{a} = $_a;
2465        # Date::Parse isn't in the standard Perl distro :(
2466        if ($tz =~ s/^\+//) {
2467                $t += tz_to_s_offset($tz);
2468        } elsif ($tz =~ s/^\-//) {
2469                $t -= tz_to_s_offset($tz);
2470        }
2471        $dest->{t_utc} = $t;
2472}
2473
2474sub process_commit {
2475        my ($c, $r_min, $r_max, $defer) = @_;
2476        if (defined $r_min && defined $r_max) {
2477                if ($r_min == $c->{r} && $r_min == $r_max) {
2478                        show_commit($c);
2479                        return 0;
2480                }
2481                return 1 if $r_min == $r_max;
2482                if ($r_min < $r_max) {
2483                        # we need to reverse the print order
2484                        return 0 if (defined $_limit && --$_limit < 0);
2485                        push @$defer, $c;
2486                        return 1;
2487                }
2488                if ($r_min != $r_max) {
2489                        return 1 if ($r_min < $c->{r});
2490                        return 1 if ($r_max > $c->{r});
2491                }
2492        }
2493        return 0 if (defined $_limit && --$_limit < 0);
2494        show_commit($c);
2495        return 1;
2496}
2497
2498sub show_commit {
2499        my $c = shift;
2500        if ($_oneline) {
2501                my $x = "\n";
2502                if (my $l = $c->{l}) {
2503                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2504                        $x = $l->[0];
2505                }
2506                $_l_fmt ||= 'A' . length($c->{r});
2507                print 'r',pack($_l_fmt, $c->{r}),' | ';
2508                print "$c->{c} | " if $_show_commit;
2509                print $x;
2510        } else {
2511                show_commit_normal($c);
2512        }
2513}
2514
2515sub show_commit_normal {
2516        my ($c) = @_;
2517        print '-' x72, "\nr$c->{r} | ";
2518        print "$c->{c} | " if $_show_commit;
2519        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2520                                 localtime($c->{t_utc})), ' | ';
2521        my $nr_line = 0;
2522
2523        if (my $l = $c->{l}) {
2524                while ($l->[$#$l] eq "\n" && $l->[($#$l - 1)] eq "\n") {
2525                        pop @$l;
2526                }
2527                $nr_line = scalar @$l;
2528                if (!$nr_line) {
2529                        print "1 line\n\n\n";
2530                } else {
2531                        if ($nr_line == 1) {
2532                                $nr_line = '1 line';
2533                        } else {
2534                                $nr_line .= ' lines';
2535                        }
2536                        print $nr_line, "\n\n";
2537                        print $_ foreach @$l;
2538                }
2539        } else {
2540                print "1 line\n\n";
2541
2542        }
2543        foreach my $x (qw/raw diff/) {
2544                if ($c->{$x}) {
2545                        print "\n";
2546                        print $_ foreach @{$c->{$x}}
2547                }
2548        }
2549}
2550
2551sub libsvn_load {
2552        return unless $_use_lib;
2553        $_use_lib = eval {
2554                require SVN::Core;
2555                if ($SVN::Core::VERSION lt '1.1.0') {
2556                        die "Need SVN::Core 1.1.0 or better ",
2557                                        "(got $SVN::Core::VERSION) ",
2558                                        "Falling back to command-line svn\n";
2559                }
2560                require SVN::Ra;
2561                require SVN::Delta;
2562                push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
2563                my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2564                                        $SVN::Node::dir.$SVN::Node::unknown.
2565                                        $SVN::Node::none.$SVN::Node::file.
2566                                        $SVN::Node::dir.$SVN::Node::unknown;
2567                1;
2568        };
2569}
2570
2571sub libsvn_connect {
2572        my ($url) = @_;
2573        my $auth = SVN::Core::auth_open([SVN::Client::get_simple_provider(),
2574                          SVN::Client::get_ssl_server_trust_file_provider(),
2575                          SVN::Client::get_username_provider()]);
2576        my $s = eval { SVN::Ra->new(url => $url, auth => $auth) };
2577        return $s;
2578}
2579
2580sub libsvn_get_file {
2581        my ($gui, $f, $rev) = @_;
2582        my $p = $f;
2583        if (length $SVN_PATH > 0) {
2584                return unless ($p =~ s#^\Q$SVN_PATH\E/##);
2585        }
2586
2587        my ($hash, $pid, $in, $out);
2588        my $pool = SVN::Pool->new;
2589        defined($pid = open3($in, $out, '>&STDERR',
2590                                qw/git-hash-object -w --stdin/)) or croak $!;
2591        # redirect STDOUT for SVN 1.1.x compatibility
2592        open my $stdout, '>&', \*STDOUT or croak $!;
2593        open STDOUT, '>&', $in or croak $!;
2594        my ($r, $props) = $SVN->get_file($f, $rev, \*STDOUT, $pool);
2595        $in->flush == 0 or croak $!;
2596        open STDOUT, '>&', $stdout or croak $!;
2597        close $in or croak $!;
2598        close $stdout or croak $!;
2599        $pool->clear;
2600        chomp($hash = do { local $/; <$out> });
2601        close $out or croak $!;
2602        waitpid $pid, 0;
2603        $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2604
2605        my $mode = exists $props->{'svn:executable'} ? '100755' : '100644';
2606        if (exists $props->{'svn:special'}) {
2607                $mode = '120000';
2608                my $link = `git-cat-file blob $hash`;
2609                $link =~ s/^link // or die "svn:special file with contents: <",
2610                                                $link, "> is not understood\n";
2611                defined($pid = open3($in, $out, '>&STDERR',
2612                                qw/git-hash-object -w --stdin/)) or croak $!;
2613                print $in $link;
2614                $in->flush == 0 or croak $!;
2615                close $in or croak $!;
2616                chomp($hash = do { local $/; <$out> });
2617                close $out or croak $!;
2618                waitpid $pid, 0;
2619                $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2620        }
2621        print $gui $mode,' ',$hash,"\t",$p,"\0" or croak $!;
2622}
2623
2624sub libsvn_log_entry {
2625        my ($rev, $author, $date, $msg, $parents) = @_;
2626        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2627                                         (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
2628                                or die "Unable to parse date: $date\n";
2629        if (defined $_authors && ! defined $users{$author}) {
2630                die "Author: $author not defined in $_authors file\n";
2631        }
2632        $msg = '' if ($rev == 0 && !defined $msg);
2633        return { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
2634                author => $author, msg => $msg."\n", parents => $parents || [] }
2635}
2636
2637sub process_rm {
2638        my ($gui, $last_commit, $f) = @_;
2639        $f =~ s#^\Q$SVN_PATH\E/?## or return;
2640        # remove entire directories.
2641        if (safe_qx('git-ls-tree',$last_commit,'--',$f) =~ /^040000 tree/) {
2642                defined(my $pid = open my $ls, '-|') or croak $!;
2643                if (!$pid) {
2644                        exec(qw/git-ls-tree -r --name-only -z/,
2645                                $last_commit,'--',$f) or croak $!;
2646                }
2647                local $/ = "\0";
2648                while (<$ls>) {
2649                        print $gui '0 ',0 x 40,"\t",$_ or croak $!;
2650                }
2651                close $ls or croak $?;
2652        } else {
2653                print $gui '0 ',0 x 40,"\t",$f,"\0" or croak $!;
2654        }
2655}
2656
2657sub libsvn_fetch {
2658        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
2659        open my $gui, '| git-update-index -z --index-info' or croak $!;
2660        my @amr;
2661        foreach my $f (keys %$paths) {
2662                my $m = $paths->{$f}->action();
2663                $f =~ s#^/+##;
2664                if ($m =~ /^[DR]$/) {
2665                        print "\t$m\t$f\n" unless $_q;
2666                        process_rm($gui, $last_commit, $f);
2667                        next if $m eq 'D';
2668                        # 'R' can be file replacements, too, right?
2669                }
2670                my $pool = SVN::Pool->new;
2671                my $t = $SVN->check_path($f, $rev, $pool);
2672                if ($t == $SVN::Node::file) {
2673                        if ($m =~ /^[AMR]$/) {
2674                                push @amr, [ $m, $f ];
2675                        } else {
2676                                die "Unrecognized action: $m, ($f r$rev)\n";
2677                        }
2678                } elsif ($t == $SVN::Node::dir && $m =~ /^[AR]$/) {
2679                        my @traversed = ();
2680                        libsvn_traverse($gui, '', $f, $rev, \@traversed);
2681                        foreach (@traversed) {
2682                                push @amr, [ $m, $_ ]
2683                        }
2684                }
2685                $pool->clear;
2686        }
2687        foreach (@amr) {
2688                print "\t$_->[0]\t$_->[1]\n" unless $_q;
2689                libsvn_get_file($gui, $_->[1], $rev)
2690        }
2691        close $gui or croak $?;
2692        return libsvn_log_entry($rev, $author, $date, $msg, [$last_commit]);
2693}
2694
2695sub svn_grab_base_rev {
2696        defined(my $pid = open my $fh, '-|') or croak $!;
2697        if (!$pid) {
2698                open my $null, '>', '/dev/null' or croak $!;
2699                open STDERR, '>&', $null or croak $!;
2700                exec qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0"
2701                                                                or croak $!;
2702        }
2703        chomp(my $c = do { local $/; <$fh> });
2704        close $fh;
2705        if (defined $c && length $c) {
2706                my ($url, $rev, $uuid) = cmt_metadata($c);
2707                return ($rev, $c) if defined $rev;
2708        }
2709        if ($_no_metadata) {
2710                my $offset = -41; # from tail
2711                my $rl;
2712                open my $fh, '<', $REVDB or
2713                        die "--no-metadata specified and $REVDB not readable\n";
2714                seek $fh, $offset, 2;
2715                $rl = readline $fh;
2716                defined $rl or return (undef, undef);
2717                chomp $rl;
2718                while ($c ne $rl && tell $fh != 0) {
2719                        $offset -= 41;
2720                        seek $fh, $offset, 2;
2721                        $rl = readline $fh;
2722                        defined $rl or return (undef, undef);
2723                        chomp $rl;
2724                }
2725                my $rev = tell $fh;
2726                croak $! if ($rev < -1);
2727                $rev =  ($rev - 41) / 41;
2728                close $fh or croak $!;
2729                return ($rev, $c);
2730        }
2731        return (undef, undef);
2732}
2733
2734sub libsvn_parse_revision {
2735        my $base = shift;
2736        my $head = $SVN->get_latest_revnum();
2737        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
2738                return ($base + 1, $head) if (defined $base);
2739                return (0, $head);
2740        }
2741        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
2742        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
2743        if ($_revision =~ /^BASE:(\d+)$/) {
2744                return ($base + 1, $1) if (defined $base);
2745                return (0, $head);
2746        }
2747        return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
2748        die "revision argument: $_revision not understood by git-svn\n",
2749                "Try using the command-line svn client instead\n";
2750}
2751
2752sub libsvn_traverse {
2753        my ($gui, $pfx, $path, $rev, $files) = @_;
2754        my $cwd = "$pfx/$path";
2755        my $pool = SVN::Pool->new;
2756        $cwd =~ s#^/+##g;
2757        my ($dirent, $r, $props) = $SVN->get_dir($cwd, $rev, $pool);
2758        foreach my $d (keys %$dirent) {
2759                my $t = $dirent->{$d}->kind;
2760                if ($t == $SVN::Node::dir) {
2761                        libsvn_traverse($gui, $cwd, $d, $rev, $files);
2762                } elsif ($t == $SVN::Node::file) {
2763                        my $file = "$cwd/$d";
2764                        if (defined $files) {
2765                                push @$files, $file;
2766                        } else {
2767                                print "\tA\t$file\n" unless $_q;
2768                                libsvn_get_file($gui, $file, $rev);
2769                        }
2770                }
2771        }
2772        $pool->clear;
2773}
2774
2775sub libsvn_traverse_ignore {
2776        my ($fh, $path, $r) = @_;
2777        $path =~ s#^/+##g;
2778        my $pool = SVN::Pool->new;
2779        my ($dirent, undef, $props) = $SVN->get_dir($path, $r, $pool);
2780        my $p = $path;
2781        $p =~ s#^\Q$SVN_PATH\E/?##;
2782        print $fh length $p ? "\n# $p\n" : "\n# /\n";
2783        if (my $s = $props->{'svn:ignore'}) {
2784                $s =~ s/[\r\n]+/\n/g;
2785                chomp $s;
2786                if (length $p == 0) {
2787                        $s =~ s#\n#\n/$p#g;
2788                        print $fh "/$s\n";
2789                } else {
2790                        $s =~ s#\n#\n/$p/#g;
2791                        print $fh "/$p/$s\n";
2792                }
2793        }
2794        foreach (sort keys %$dirent) {
2795                next if $dirent->{$_}->kind != $SVN::Node::dir;
2796                libsvn_traverse_ignore($fh, "$path/$_", $r);
2797        }
2798        $pool->clear;
2799}
2800
2801sub revisions_eq {
2802        my ($path, $r0, $r1) = @_;
2803        return 1 if $r0 == $r1;
2804        my $nr = 0;
2805        if ($_use_lib) {
2806                # should be OK to use Pool here (r1 - r0) should be small
2807                my $pool = SVN::Pool->new;
2808                libsvn_get_log($SVN, "/$path", $r0, $r1,
2809                                0, 1, 1, sub {$nr++}, $pool);
2810                $pool->clear;
2811        } else {
2812                my ($url, undef) = repo_path_split($SVN_URL);
2813                my $svn_log = svn_log_raw("$url/$path","-r$r0:$r1");
2814                while (next_log_entry($svn_log)) { $nr++ }
2815                close $svn_log->{fh};
2816        }
2817        return 0 if ($nr > 1);
2818        return 1;
2819}
2820
2821sub libsvn_find_parent_branch {
2822        my ($paths, $rev, $author, $date, $msg) = @_;
2823        my $svn_path = '/'.$SVN_PATH;
2824
2825        # look for a parent from another branch:
2826        my $i = $paths->{$svn_path} or return;
2827        my $branch_from = $i->copyfrom_path or return;
2828        my $r = $i->copyfrom_rev;
2829        print STDERR  "Found possible branch point: ",
2830                                "$branch_from => $svn_path, $r\n";
2831        $branch_from =~ s#^/##;
2832        my $l_map = {};
2833        read_url_paths_all($l_map, '', "$GIT_DIR/svn");
2834        my $url = $SVN->{url};
2835        defined $l_map->{$url} or return;
2836        my $id = $l_map->{$url}->{$branch_from};
2837        if (!defined $id && $_follow_parent) {
2838                print STDERR "Following parent: $branch_from\@$r\n";
2839                # auto create a new branch and follow it
2840                $id = basename($branch_from);
2841                $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
2842                while (-r "$GIT_DIR/svn/$id") {
2843                        # just grow a tail if we're not unique enough :x
2844                        $id .= '-';
2845                }
2846        }
2847        return unless defined $id;
2848
2849        my ($r0, $parent) = find_rev_before($r,$id,1);
2850        if ($_follow_parent && (!defined $r0 || !defined $parent)) {
2851                defined(my $pid = fork) or croak $!;
2852                if (!$pid) {
2853                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2854                        init_vars();
2855                        $SVN_URL = "$url/$branch_from";
2856                        $SVN_LOG = $SVN = undef;
2857                        setup_git_svn();
2858                        # we can't assume SVN_URL exists at r+1:
2859                        $_revision = "0:$r";
2860                        fetch_lib();
2861                        exit 0;
2862                }
2863                waitpid $pid, 0;
2864                croak $? if $?;
2865                ($r0, $parent) = find_rev_before($r,$id,1);
2866        }
2867        return unless (defined $r0 && defined $parent);
2868        if (revisions_eq($branch_from, $r0, $r)) {
2869                unlink $GIT_SVN_INDEX;
2870                print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
2871                sys(qw/git-read-tree/, $parent);
2872                return libsvn_fetch($parent, $paths, $rev,
2873                                        $author, $date, $msg);
2874        }
2875        print STDERR "Nope, branch point not imported or unknown\n";
2876        return undef;
2877}
2878
2879sub libsvn_get_log {
2880        my ($ra, @args) = @_;
2881        if ($SVN::Core::VERSION le '1.2.0') {
2882                splice(@args, 3, 1);
2883        }
2884        $ra->get_log(@args);
2885}
2886
2887sub libsvn_new_tree {
2888        if (my $log_entry = libsvn_find_parent_branch(@_)) {
2889                return $log_entry;
2890        }
2891        my ($paths, $rev, $author, $date, $msg) = @_;
2892        open my $gui, '| git-update-index -z --index-info' or croak $!;
2893        libsvn_traverse($gui, '', $SVN_PATH, $rev);
2894        close $gui or croak $?;
2895        return libsvn_log_entry($rev, $author, $date, $msg);
2896}
2897
2898sub find_graft_path_commit {
2899        my ($tree_paths, $p1, $r1) = @_;
2900        foreach my $x (keys %$tree_paths) {
2901                next unless ($p1 =~ /^\Q$x\E/);
2902                my $i = $tree_paths->{$x};
2903                my ($r0, $parent) = find_rev_before($r1,$i,1);
2904                return $parent if (defined $r0 && $r0 == $r1);
2905                print STDERR "r$r1 of $i not imported\n";
2906                next;
2907        }
2908        return undef;
2909}
2910
2911sub find_graft_path_parents {
2912        my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
2913        foreach my $x (keys %$tree_paths) {
2914                next unless ($p0 =~ /^\Q$x\E/);
2915                my $i = $tree_paths->{$x};
2916                my ($r, $parent) = find_rev_before($r0, $i, 1);
2917                if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
2918                        my ($url_b, undef, $uuid_b) = cmt_metadata($c);
2919                        my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
2920                        next if ($url_a && $url_b && $url_a eq $url_b &&
2921                                                        $uuid_b eq $uuid_a);
2922                        $grafts->{$c}->{$parent} = 1;
2923                }
2924        }
2925}
2926
2927sub libsvn_graft_file_copies {
2928        my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
2929        foreach (keys %$paths) {
2930                my $i = $paths->{$_};
2931                my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
2932                                        $i->copyfrom_rev);
2933                next unless (defined $p0 && defined $r0);
2934
2935                my $p1 = $_;
2936                $p1 =~ s#^/##;
2937                $p0 =~ s#^/##;
2938                my $c = find_graft_path_commit($tree_paths, $p1, $rev);
2939                next unless $c;
2940                find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
2941        }
2942}
2943
2944sub set_index {
2945        my $old = $ENV{GIT_INDEX_FILE};
2946        $ENV{GIT_INDEX_FILE} = shift;
2947        return $old;
2948}
2949
2950sub restore_index {
2951        my ($old) = @_;
2952        if (defined $old) {
2953                $ENV{GIT_INDEX_FILE} = $old;
2954        } else {
2955                delete $ENV{GIT_INDEX_FILE};
2956        }
2957}
2958
2959sub libsvn_commit_cb {
2960        my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
2961        if ($_optimize_commits && $rev == ($r_last + 1)) {
2962                my $log = libsvn_log_entry($rev,$committer,$date,$msg);
2963                $log->{tree} = get_tree_from_treeish($c);
2964                my $cmt = git_commit($log, $cmt_last, $c);
2965                my @diff = safe_qx('git-diff-tree', $cmt, $c);
2966                if (@diff) {
2967                        print STDERR "Trees differ: $cmt $c\n",
2968                                        join('',@diff),"\n";
2969                        exit 1;
2970                }
2971        } else {
2972                fetch("$rev=$c");
2973        }
2974}
2975
2976sub libsvn_ls_fullurl {
2977        my $fullurl = shift;
2978        my ($repo, $path) = repo_path_split($fullurl);
2979        $SVN ||= libsvn_connect($repo);
2980        my @ret;
2981        my $pool = SVN::Pool->new;
2982        my ($dirent, undef, undef) = $SVN->get_dir($path,
2983                                                $SVN->get_latest_revnum, $pool);
2984        foreach my $d (keys %$dirent) {
2985                if ($dirent->{$d}->kind == $SVN::Node::dir) {
2986                        push @ret, "$d/"; # add '/' for compat with cli svn
2987                }
2988        }
2989        $pool->clear;
2990        return @ret;
2991}
2992
2993
2994sub libsvn_skip_unknown_revs {
2995        my $err = shift;
2996        my $errno = $err->apr_err();
2997        # Maybe the branch we're tracking didn't
2998        # exist when the repo started, so it's
2999        # not an error if it doesn't, just continue
3000        #
3001        # Wonderfully consistent library, eh?
3002        # 160013 - svn:// and file://
3003        # 175002 - http(s)://
3004        #   More codes may be discovered later...
3005        if ($errno == 175002 || $errno == 160013) {
3006                return;
3007        }
3008        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3009};
3010
3011# Tie::File seems to be prone to offset errors if revisions get sparse,
3012# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3013# one of my favorite modules is out :<  Next up would be one of the DBM
3014# modules, but I'm not sure which is most portable...  So I'll just
3015# go with something that's plain-text, but still capable of
3016# being randomly accessed.  So here's my ultra-simple fixed-width
3017# database.  All records are 40 characters + "\n", so it's easy to seek
3018# to a revision: (41 * rev) is the byte offset.
3019# A record of 40 0s denotes an empty revision.
3020# And yes, it's still pretty fast (faster than Tie::File).
3021sub revdb_set {
3022        my ($file, $rev, $commit) = @_;
3023        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
3024        open my $fh, '+<', $file or croak $!;
3025        my $offset = $rev * 41;
3026        # assume that append is the common case:
3027        seek $fh, 0, 2 or croak $!;
3028        my $pos = tell $fh;
3029        if ($pos < $offset) {
3030                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
3031        }
3032        seek $fh, $offset, 0 or croak $!;
3033        print $fh $commit,"\n";
3034        close $fh or croak $!;
3035}
3036
3037sub revdb_get {
3038        my ($file, $rev) = @_;
3039        my $ret;
3040        my $offset = $rev * 41;
3041        open my $fh, '<', $file or croak $!;
3042        seek $fh, $offset, 0;
3043        if (tell $fh == $offset) {
3044                $ret = readline $fh;
3045                if (defined $ret) {
3046                        chomp $ret;
3047                        $ret = undef if ($ret =~ /^0{40}$/);
3048                }
3049        }
3050        close $fh or croak $!;
3051        return $ret;
3052}
3053
3054sub copy_remote_ref {
3055        my $origin = $_cp_remote ? $_cp_remote : 'origin';
3056        my $ref = "refs/remotes/$GIT_SVN";
3057        if (safe_qx('git-ls-remote', $origin, $ref)) {
3058                sys(qw/git fetch/, $origin, "$ref:$ref");
3059        } else {
3060                die "Unable to find remote reference: ",
3061                                "refs/remotes/$GIT_SVN on $origin\n";
3062        }
3063}
3064
3065package SVN::Git::Editor;
3066use vars qw/@ISA/;
3067use strict;
3068use warnings;
3069use Carp qw/croak/;
3070use IO::File;
3071
3072sub new {
3073        my $class = shift;
3074        my $git_svn = shift;
3075        my $self = SVN::Delta::Editor->new(@_);
3076        bless $self, $class;
3077        foreach (qw/svn_path c r ra /) {
3078                die "$_ required!\n" unless (defined $git_svn->{$_});
3079                $self->{$_} = $git_svn->{$_};
3080        }
3081        $self->{pool} = SVN::Pool->new;
3082        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3083        $self->{rm} = { };
3084        require Digest::MD5;
3085        return $self;
3086}
3087
3088sub split_path {
3089        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3090}
3091
3092sub repo_path {
3093        (defined $_[1] && length $_[1]) ? "$_[0]->{svn_path}/$_[1]"
3094                                        : $_[0]->{svn_path}
3095}
3096
3097sub url_path {
3098        my ($self, $path) = @_;
3099        $self->{ra}->{url} . '/' . $self->repo_path($path);
3100}
3101
3102sub rmdirs {
3103        my ($self, $q) = @_;
3104        my $rm = $self->{rm};
3105        delete $rm->{''}; # we never delete the url we're tracking
3106        return unless %$rm;
3107
3108        foreach (keys %$rm) {
3109                my @d = split m#/#, $_;
3110                my $c = shift @d;
3111                $rm->{$c} = 1;
3112                while (@d) {
3113                        $c .= '/' . shift @d;
3114                        $rm->{$c} = 1;
3115                }
3116        }
3117        delete $rm->{$self->{svn_path}};
3118        delete $rm->{''}; # we never delete the url we're tracking
3119        return unless %$rm;
3120
3121        defined(my $pid = open my $fh,'-|') or croak $!;
3122        if (!$pid) {
3123                exec qw/git-ls-tree --name-only -r -z/, $self->{c} or croak $!;
3124        }
3125        local $/ = "\0";
3126        my @svn_path = split m#/#, $self->{svn_path};
3127        while (<$fh>) {
3128                chomp;
3129                my @dn = (@svn_path, (split m#/#, $_));
3130                while (pop @dn) {
3131                        delete $rm->{join '/', @dn};
3132                }
3133                unless (%$rm) {
3134                        close $fh;
3135                        return;
3136                }
3137        }
3138        close $fh;
3139
3140        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3141        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3142                $self->close_directory($bat->{$d}, $p);
3143                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3144                print "\tD+\t/$d/\n" unless $q;
3145                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3146                delete $bat->{$d};
3147        }
3148}
3149
3150sub open_or_add_dir {
3151        my ($self, $full_path, $baton) = @_;
3152        my $p = SVN::Pool->new;
3153        my $t = $self->{ra}->check_path($full_path, $self->{r}, $p);
3154        $p->clear;
3155        if ($t == $SVN::Node::none) {
3156                return $self->add_directory($full_path, $baton,
3157                                                undef, -1, $self->{pool});
3158        } elsif ($t == $SVN::Node::dir) {
3159                return $self->open_directory($full_path, $baton,
3160                                                $self->{r}, $self->{pool});
3161        }
3162        print STDERR "$full_path already exists in repository at ",
3163                "r$self->{r} and it is not a directory (",
3164                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3165        exit 1;
3166}
3167
3168sub ensure_path {
3169        my ($self, $path) = @_;
3170        my $bat = $self->{bat};
3171        $path = $self->repo_path($path);
3172        return $bat->{''} unless (length $path);
3173        my @p = split m#/+#, $path;
3174        my $c = shift @p;
3175        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3176        while (@p) {
3177                my $c0 = $c;
3178                $c .= '/' . shift @p;
3179                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3180        }
3181        return $bat->{$c};
3182}
3183
3184sub A {
3185        my ($self, $m, $q) = @_;
3186        my ($dir, $file) = split_path($m->{file_b});
3187        my $pbat = $self->ensure_path($dir);
3188        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3189                                        undef, -1);
3190        print "\tA\t$m->{file_b}\n" unless $q;
3191        $self->chg_file($fbat, $m);
3192        $self->close_file($fbat,undef,$self->{pool});
3193}
3194
3195sub C {
3196        my ($self, $m, $q) = @_;
3197        my ($dir, $file) = split_path($m->{file_b});
3198        my $pbat = $self->ensure_path($dir);
3199        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3200                                $self->url_path($m->{file_a}), $self->{r});
3201        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $q;
3202        $self->chg_file($fbat, $m);
3203        $self->close_file($fbat,undef,$self->{pool});
3204}
3205
3206sub delete_entry {
3207        my ($self, $path, $pbat) = @_;
3208        my $rpath = $self->repo_path($path);
3209        my ($dir, $file) = split_path($rpath);
3210        $self->{rm}->{$dir} = 1;
3211        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3212}
3213
3214sub R {
3215        my ($self, $m, $q) = @_;
3216        my ($dir, $file) = split_path($m->{file_b});
3217        my $pbat = $self->ensure_path($dir);
3218        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3219                                $self->url_path($m->{file_a}), $self->{r});
3220        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $q;
3221        $self->chg_file($fbat, $m);
3222        $self->close_file($fbat,undef,$self->{pool});
3223
3224        ($dir, $file) = split_path($m->{file_a});
3225        $pbat = $self->ensure_path($dir);
3226        $self->delete_entry($m->{file_a}, $pbat);
3227}
3228
3229sub M {
3230        my ($self, $m, $q) = @_;
3231        my ($dir, $file) = split_path($m->{file_b});
3232        my $pbat = $self->ensure_path($dir);
3233        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3234                                $pbat,$self->{r},$self->{pool});
3235        print "\t$m->{chg}\t$m->{file_b}\n" unless $q;
3236        $self->chg_file($fbat, $m);
3237        $self->close_file($fbat,undef,$self->{pool});
3238}
3239
3240sub T { shift->M(@_) }
3241
3242sub change_file_prop {
3243        my ($self, $fbat, $pname, $pval) = @_;
3244        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3245}
3246
3247sub chg_file {
3248        my ($self, $fbat, $m) = @_;
3249        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3250                $self->change_file_prop($fbat,'svn:executable','*');
3251        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3252                $self->change_file_prop($fbat,'svn:executable',undef);
3253        }
3254        my $fh = IO::File->new_tmpfile or croak $!;
3255        if ($m->{mode_b} =~ /^120/) {
3256                print $fh 'link ' or croak $!;
3257                $self->change_file_prop($fbat,'svn:special','*');
3258        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3259                $self->change_file_prop($fbat,'svn:special',undef);
3260        }
3261        defined(my $pid = fork) or croak $!;
3262        if (!$pid) {
3263                open STDOUT, '>&', $fh or croak $!;
3264                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3265        }
3266        waitpid $pid, 0;
3267        croak $? if $?;
3268        $fh->flush == 0 or croak $!;
3269        seek $fh, 0, 0 or croak $!;
3270
3271        my $md5 = Digest::MD5->new;
3272        $md5->addfile($fh) or croak $!;
3273        seek $fh, 0, 0 or croak $!;
3274
3275        my $exp = $md5->hexdigest;
3276        my $atd = $self->apply_textdelta($fbat, undef, $self->{pool});
3277        my $got = SVN::TxDelta::send_stream($fh, @$atd, $self->{pool});
3278        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3279
3280        close $fh or croak $!;
3281}
3282
3283sub D {
3284        my ($self, $m, $q) = @_;
3285        my ($dir, $file) = split_path($m->{file_b});
3286        my $pbat = $self->ensure_path($dir);
3287        print "\tD\t$m->{file_b}\n" unless $q;
3288        $self->delete_entry($m->{file_b}, $pbat);
3289}
3290
3291sub close_edit {
3292        my ($self) = @_;
3293        my ($p,$bat) = ($self->{pool}, $self->{bat});
3294        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3295                $self->close_directory($bat->{$_}, $p);
3296        }
3297        $self->SUPER::close_edit($p);
3298        $p->clear;
3299}
3300
3301sub abort_edit {
3302        my ($self) = @_;
3303        $self->SUPER::abort_edit($self->{pool});
3304        $self->{pool}->clear;
3305}
3306
3307__END__
3308
3309Data structures:
3310
3311$svn_log hashref (as returned by svn_log_raw)
3312{
3313        fh => file handle of the log file,
3314        state => state of the log file parser (sep/msg/rev/msg_start...)
3315}
3316
3317$log_msg hashref as returned by next_log_entry($svn_log)
3318{
3319        msg => 'whitespace-formatted log entry
3320',                                              # trailing newline is preserved
3321        revision => '8',                        # integer
3322        date => '2004-02-24T17:01:44.108345Z',  # commit date
3323        author => 'committer name'
3324};
3325
3326
3327@mods = array of diff-index line hashes, each element represents one line
3328        of diff-index output
3329
3330diff-index line ($m hash)
3331{
3332        mode_a => first column of diff-index output, no leading ':',
3333        mode_b => second column of diff-index output,
3334        sha1_b => sha1sum of the final blob,
3335        chg => change type [MCRADT],
3336        file_a => original file name of a file (iff chg is 'C' or 'R')
3337        file_b => new/current file name of a file (any chg)
3338}
3339;
3340
3341# retval of read_url_paths{,_all}();
3342$l_map = {
3343        # repository root url
3344        'https://svn.musicpd.org' => {
3345                # repository path               # GIT_SVN_ID
3346                'mpd/trunk'             =>      'trunk',
3347                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
3348        },
3349}
3350
3351Notes:
3352        I don't trust the each() function on unless I created %hash myself
3353        because the internal iterator may not have started at base.