git-svn.perlon commit git-svn: establish new connections on commit after fork (5a990e4)
   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        my $repo;
 504        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 505        set_svn_commit_env();
 506        foreach my $c (@revs) {
 507                my $log_msg = get_commit_message($c, $commit_msg);
 508
 509                # fork for each commit because there's a memory leak I
 510                # can't track down... (it's probably in the SVN code)
 511                defined(my $pid = open my $fh, '-|') or croak $!;
 512                if (!$pid) {
 513                        $SVN_LOG = libsvn_connect($repo);
 514                        $SVN = libsvn_connect($repo);
 515                        my $ed = SVN::Git::Editor->new(
 516                                        {       r => $r_last,
 517                                                ra => $SVN,
 518                                                c => $c,
 519                                                svn_path => $SVN_PATH
 520                                        },
 521                                        $SVN->get_commit_editor(
 522                                                $log_msg->{msg},
 523                                                sub {
 524                                                        libsvn_commit_cb(
 525                                                                @_, $c,
 526                                                                $log_msg->{msg},
 527                                                                $r_last,
 528                                                                $cmt_last)
 529                                                },
 530                                                @lock)
 531                                        );
 532                        my $mods = libsvn_checkout_tree($cmt_last, $c, $ed);
 533                        if (@$mods == 0) {
 534                                print "No changes\nr$r_last = $cmt_last\n";
 535                                $ed->abort_edit;
 536                        } else {
 537                                $ed->close_edit;
 538                        }
 539                        exit 0;
 540                }
 541                my ($r_new, $cmt_new, $no);
 542                while (<$fh>) {
 543                        print $_;
 544                        chomp;
 545                        if (/^r(\d+) = ($sha1)$/o) {
 546                                ($r_new, $cmt_new) = ($1, $2);
 547                        } elsif ($_ eq 'No changes') {
 548                                $no = 1;
 549                        }
 550                }
 551                close $fh or croak $?;
 552                if (! defined $r_new && ! defined $cmt_new) {
 553                        unless ($no) {
 554                                die "Failed to parse revision information\n";
 555                        }
 556                } else {
 557                        ($r_last, $cmt_last) = ($r_new, $cmt_new);
 558                }
 559        }
 560        $ENV{LC_ALL} = 'C';
 561        unlink $commit_msg;
 562}
 563
 564sub show_ignore {
 565        $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 566        $_use_lib ? show_ignore_lib() : show_ignore_cmd();
 567}
 568
 569sub show_ignore_cmd {
 570        require File::Find or die $!;
 571        if (defined $_revision) {
 572                die "-r/--revision option doesn't work unless the Perl SVN ",
 573                        "libraries are used\n";
 574        }
 575        chdir $SVN_WC or croak $!;
 576        my %ign;
 577        File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 578                s#^\./##;
 579                @{$ign{$_}} = svn_propget_base('svn:ignore', $_);
 580                }}, no_chdir=>1},'.');
 581
 582        print "\n# /\n";
 583        foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 584        delete $ign{'.'};
 585        foreach my $i (sort keys %ign) {
 586                print "\n# ",$i,"\n";
 587                foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 588        }
 589}
 590
 591sub show_ignore_lib {
 592        my $repo;
 593        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 594        $SVN ||= libsvn_connect($repo);
 595        my $r = defined $_revision ? $_revision : $SVN->get_latest_revnum;
 596        libsvn_traverse_ignore(\*STDOUT, $SVN_PATH, $r);
 597}
 598
 599sub graft_branches {
 600        my $gr_file = "$GIT_DIR/info/grafts";
 601        my ($grafts, $comments) = read_grafts($gr_file);
 602        my $gr_sha1;
 603
 604        if (%$grafts) {
 605                # temporarily disable our grafts file to make this idempotent
 606                chomp($gr_sha1 = safe_qx(qw/git-hash-object -w/,$gr_file));
 607                rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
 608        }
 609
 610        my $l_map = read_url_paths();
 611        my @re = map { qr/$_/is } @_opt_m if @_opt_m;
 612        unless ($_no_default_regex) {
 613                push @re, (qr/\b(?:merge|merging|merged)\s+with\s+([\w\.\-]+)/i,
 614                        qr/\b(?:merge|merging|merged)\s+([\w\.\-]+)/i,
 615                        qr/\b(?:from|of)\s+([\w\.\-]+)/i );
 616        }
 617        foreach my $u (keys %$l_map) {
 618                if (@re) {
 619                        foreach my $p (keys %{$l_map->{$u}}) {
 620                                graft_merge_msg($grafts,$l_map,$u,$p,@re);
 621                        }
 622                }
 623                unless ($_no_graft_copy) {
 624                        if ($_use_lib) {
 625                                graft_file_copy_lib($grafts,$l_map,$u);
 626                        } else {
 627                                graft_file_copy_cmd($grafts,$l_map,$u);
 628                        }
 629                }
 630        }
 631        graft_tree_joins($grafts);
 632
 633        write_grafts($grafts, $comments, $gr_file);
 634        unlink "$gr_file~$gr_sha1" if $gr_sha1;
 635}
 636
 637sub multi_init {
 638        my $url = shift;
 639        $_trunk ||= 'trunk';
 640        $_trunk =~ s#/+$##;
 641        $url =~ s#/+$## if $url;
 642        if ($_trunk !~ m#^[a-z\+]+://#) {
 643                $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#);
 644                unless ($url) {
 645                        print STDERR "E: '$_trunk' is not a complete URL ",
 646                                "and a separate URL is not specified\n";
 647                        exit 1;
 648                }
 649                $_trunk = $url . $_trunk;
 650        }
 651        if ($GIT_SVN eq 'git-svn') {
 652                print "GIT_SVN_ID set to 'trunk' for $_trunk\n";
 653                $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
 654        }
 655        init_vars();
 656        init($_trunk);
 657        complete_url_ls_init($url, $_branches, '--branches/-b', '');
 658        complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/');
 659}
 660
 661sub multi_fetch {
 662        # try to do trunk first, since branches/tags
 663        # may be descended from it.
 664        if (-e "$GIT_DIR/svn/trunk/info/url") {
 665                fetch_child_id('trunk', @_);
 666        }
 667        rec_fetch('', "$GIT_DIR/svn", @_);
 668}
 669
 670sub show_log {
 671        my (@args) = @_;
 672        my ($r_min, $r_max);
 673        my $r_last = -1; # prevent dupes
 674        rload_authors() if $_authors;
 675        if (defined $TZ) {
 676                $ENV{TZ} = $TZ;
 677        } else {
 678                delete $ENV{TZ};
 679        }
 680        if (defined $_revision) {
 681                if ($_revision =~ /^(\d+):(\d+)$/) {
 682                        ($r_min, $r_max) = ($1, $2);
 683                } elsif ($_revision =~ /^\d+$/) {
 684                        $r_min = $r_max = $_revision;
 685                } else {
 686                        print STDERR "-r$_revision is not supported, use ",
 687                                "standard \'git log\' arguments instead\n";
 688                        exit 1;
 689                }
 690        }
 691
 692        my $pid = open(my $log,'-|');
 693        defined $pid or croak $!;
 694        if (!$pid) {
 695                exec(git_svn_log_cmd($r_min,$r_max), @args) or croak $!;
 696        }
 697        setup_pager();
 698        my (@k, $c, $d);
 699
 700        while (<$log>) {
 701                if (/^commit ($sha1_short)/o) {
 702                        my $cmt = $1;
 703                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
 704                                $r_last = $c->{r};
 705                                process_commit($c, $r_min, $r_max, \@k) or
 706                                                                goto out;
 707                        }
 708                        $d = undef;
 709                        $c = { c => $cmt };
 710                } elsif (/^author (.+) (\d+) ([\-\+]?\d+)$/) {
 711                        get_author_info($c, $1, $2, $3);
 712                } elsif (/^(?:tree|parent|committer) /) {
 713                        # ignore
 714                } elsif (/^:\d{6} \d{6} $sha1_short/o) {
 715                        push @{$c->{raw}}, $_;
 716                } elsif (/^diff /) {
 717                        $d = 1;
 718                        push @{$c->{diff}}, $_;
 719                } elsif ($d) {
 720                        push @{$c->{diff}}, $_;
 721                } elsif (/^    (git-svn-id:.+)$/) {
 722                        (undef, $c->{r}, undef) = extract_metadata($1);
 723                } elsif (s/^    //) {
 724                        push @{$c->{l}}, $_;
 725                }
 726        }
 727        if ($c && defined $c->{r} && $c->{r} != $r_last) {
 728                $r_last = $c->{r};
 729                process_commit($c, $r_min, $r_max, \@k);
 730        }
 731        if (@k) {
 732                my $swap = $r_max;
 733                $r_max = $r_min;
 734                $r_min = $swap;
 735                process_commit($_, $r_min, $r_max) foreach reverse @k;
 736        }
 737out:
 738        close $log;
 739        print '-' x72,"\n" unless $_incremental || $_oneline;
 740}
 741
 742sub commit_diff_usage {
 743        print STDERR "Usage: $0 commit-diff <tree-ish> <tree-ish> [<URL>]\n";
 744        exit 1
 745}
 746
 747sub commit_diff {
 748        if (!$_use_lib) {
 749                print STDERR "commit-diff must be used with SVN libraries\n";
 750                exit 1;
 751        }
 752        my $ta = shift or commit_diff_usage();
 753        my $tb = shift or commit_diff_usage();
 754        if (!eval { $SVN_URL = shift || file_to_s("$GIT_SVN_DIR/info/url") }) {
 755                print STDERR "Needed URL or usable git-svn id command-line\n";
 756                commit_diff_usage();
 757        }
 758        if (defined $_message && defined $_file) {
 759                print STDERR "Both --message/-m and --file/-F specified ",
 760                                "for the commit message.\n",
 761                                "I have no idea what you mean\n";
 762                exit 1;
 763        }
 764        if (defined $_file) {
 765                $_message = file_to_s($_file);
 766        } else {
 767                $_message ||= get_commit_message($tb,
 768                                        "$GIT_DIR/.svn-commit.tmp.$$")->{msg};
 769        }
 770        my $repo;
 771        ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
 772        $SVN_LOG ||= libsvn_connect($repo);
 773        $SVN ||= libsvn_connect($repo);
 774        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
 775        my $ed = SVN::Git::Editor->new({        r => $SVN->get_latest_revnum,
 776                                                ra => $SVN, c => $tb,
 777                                                svn_path => $SVN_PATH
 778                                        },
 779                                $SVN->get_commit_editor($_message,
 780                                        sub {print "Committed $_[0]\n"},@lock)
 781                                );
 782        my $mods = libsvn_checkout_tree($ta, $tb, $ed);
 783        if (@$mods == 0) {
 784                print "No changes\n$ta == $tb\n";
 785                $ed->abort_edit;
 786        } else {
 787                $ed->close_edit;
 788        }
 789}
 790
 791########################### utility functions #########################
 792
 793sub cmt_showable {
 794        my ($c) = @_;
 795        return 1 if defined $c->{r};
 796        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
 797                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
 798                my @msg = safe_qx(qw/git-cat-file commit/, $c->{c});
 799                shift @msg while ($msg[0] ne "\n");
 800                shift @msg;
 801                @{$c->{l}} = grep !/^git-svn-id: /, @msg;
 802
 803                (undef, $c->{r}, undef) = extract_metadata(
 804                                (grep(/^git-svn-id: /, @msg))[-1]);
 805        }
 806        return defined $c->{r};
 807}
 808
 809sub git_svn_log_cmd {
 810        my ($r_min, $r_max) = @_;
 811        my @cmd = (qw/git-log --abbrev-commit --pretty=raw
 812                        --default/, "refs/remotes/$GIT_SVN");
 813        push @cmd, '--summary' if $_verbose;
 814        return @cmd unless defined $r_max;
 815        if ($r_max == $r_min) {
 816                push @cmd, '--max-count=1';
 817                if (my $c = revdb_get($REVDB, $r_max)) {
 818                        push @cmd, $c;
 819                }
 820        } else {
 821                my ($c_min, $c_max);
 822                $c_max = revdb_get($REVDB, $r_max);
 823                $c_min = revdb_get($REVDB, $r_min);
 824                if ($c_min && $c_max) {
 825                        if ($r_max > $r_max) {
 826                                push @cmd, "$c_min..$c_max";
 827                        } else {
 828                                push @cmd, "$c_max..$c_min";
 829                        }
 830                } elsif ($r_max > $r_min) {
 831                        push @cmd, $c_max;
 832                } else {
 833                        push @cmd, $c_min;
 834                }
 835        }
 836        return @cmd;
 837}
 838
 839sub fetch_child_id {
 840        my $id = shift;
 841        print "Fetching $id\n";
 842        my $ref = "$GIT_DIR/refs/remotes/$id";
 843        defined(my $pid = open my $fh, '-|') or croak $!;
 844        if (!$pid) {
 845                $_repack = undef;
 846                $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 847                init_vars();
 848                fetch(@_);
 849                exit 0;
 850        }
 851        while (<$fh>) {
 852                print $_;
 853                check_repack() if (/^r\d+ = $sha1/);
 854        }
 855        close $fh or croak $?;
 856}
 857
 858sub rec_fetch {
 859        my ($pfx, $p, @args) = @_;
 860        my @dir;
 861        foreach (sort <$p/*>) {
 862                if (-r "$_/info/url") {
 863                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
 864                        my $id = $pfx . basename $_;
 865                        next if $id eq 'trunk';
 866                        fetch_child_id($id, @args);
 867                } elsif (-d $_) {
 868                        push @dir, $_;
 869                }
 870        }
 871        foreach (@dir) {
 872                my $x = $_;
 873                $x =~ s!^\Q$GIT_DIR\E/svn/!!;
 874                rec_fetch($x, $_);
 875        }
 876}
 877
 878sub complete_url_ls_init {
 879        my ($url, $var, $switch, $pfx) = @_;
 880        unless ($var) {
 881                print STDERR "W: $switch not specified\n";
 882                return;
 883        }
 884        $var =~ s#/+$##;
 885        if ($var !~ m#^[a-z\+]+://#) {
 886                $var = '/' . $var if ($var !~ m#^/#);
 887                unless ($url) {
 888                        print STDERR "E: '$var' is not a complete URL ",
 889                                "and a separate URL is not specified\n";
 890                        exit 1;
 891                }
 892                $var = $url . $var;
 893        }
 894        chomp(my @ls = $_use_lib ? libsvn_ls_fullurl($var)
 895                                : safe_qx(qw/svn ls --non-interactive/, $var));
 896        my $old = $GIT_SVN;
 897        defined(my $pid = fork) or croak $!;
 898        if (!$pid) {
 899                foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
 900                        $u =~ s#/+$##;
 901                        if ($u !~ m!\Q$var\E/(.+)$!) {
 902                                print STDERR "W: Unrecognized URL: $u\n";
 903                                die "This should never happen\n";
 904                        }
 905                        my $id = $pfx.$1;
 906                        print "init $u => $id\n";
 907                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 908                        init_vars();
 909                        init($u);
 910                }
 911                exit 0;
 912        }
 913        waitpid $pid, 0;
 914        croak $? if $?;
 915}
 916
 917sub common_prefix {
 918        my $paths = shift;
 919        my %common;
 920        foreach (@$paths) {
 921                my @tmp = split m#/#, $_;
 922                my $p = '';
 923                while (my $x = shift @tmp) {
 924                        $p .= "/$x";
 925                        $common{$p} ||= 0;
 926                        $common{$p}++;
 927                }
 928        }
 929        foreach (sort {length $b <=> length $a} keys %common) {
 930                if ($common{$_} == @$paths) {
 931                        return $_;
 932                }
 933        }
 934        return '';
 935}
 936
 937# grafts set here are 'stronger' in that they're based on actual tree
 938# matches, and won't be deleted from merge-base checking in write_grafts()
 939sub graft_tree_joins {
 940        my $grafts = shift;
 941        map_tree_joins() if (@_branch_from && !%tree_map);
 942        return unless %tree_map;
 943
 944        git_svn_each(sub {
 945                my $i = shift;
 946                defined(my $pid = open my $fh, '-|') or croak $!;
 947                if (!$pid) {
 948                        exec qw/git-rev-list --pretty=raw/,
 949                                        "refs/remotes/$i" or croak $!;
 950                }
 951                while (<$fh>) {
 952                        next unless /^commit ($sha1)$/o;
 953                        my $c = $1;
 954                        my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
 955                        next unless $tree_map{$t};
 956
 957                        my $l;
 958                        do {
 959                                $l = readline $fh;
 960                        } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
 961
 962                        my ($s, $tz) = ($1, $2);
 963                        if ($tz =~ s/^\+//) {
 964                                $s += tz_to_s_offset($tz);
 965                        } elsif ($tz =~ s/^\-//) {
 966                                $s -= tz_to_s_offset($tz);
 967                        }
 968
 969                        my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
 970
 971                        foreach my $p (@{$tree_map{$t}}) {
 972                                next if $p eq $c;
 973                                my $mb = eval {
 974                                        safe_qx('git-merge-base', $c, $p)
 975                                };
 976                                next unless ($@ || $?);
 977                                if (defined $r_a) {
 978                                        # see if SVN says it's a relative
 979                                        my ($url_b, $r_b, $uuid_b) =
 980                                                        cmt_metadata($p);
 981                                        next if (defined $url_b &&
 982                                                        defined $url_a &&
 983                                                        ($url_a eq $url_b) &&
 984                                                        ($uuid_a eq $uuid_b));
 985                                        if ($uuid_a eq $uuid_b) {
 986                                                if ($r_b < $r_a) {
 987                                                        $grafts->{$c}->{$p} = 2;
 988                                                        next;
 989                                                } elsif ($r_b > $r_a) {
 990                                                        $grafts->{$p}->{$c} = 2;
 991                                                        next;
 992                                                }
 993                                        }
 994                                }
 995                                my $ct = get_commit_time($p);
 996                                if ($ct < $s) {
 997                                        $grafts->{$c}->{$p} = 2;
 998                                } elsif ($ct > $s) {
 999                                        $grafts->{$p}->{$c} = 2;
1000                                }
1001                                # what should we do when $ct == $s ?
1002                        }
1003                }
1004                close $fh or croak $?;
1005        });
1006}
1007
1008# this isn't funky-filename safe, but good enough for now...
1009sub graft_file_copy_cmd {
1010        my ($grafts, $l_map, $u) = @_;
1011        my $paths = $l_map->{$u};
1012        my $pfx = common_prefix([keys %$paths]);
1013        $SVN_URL ||= $u.$pfx;
1014        my $pid = open my $fh, '-|';
1015        defined $pid or croak $!;
1016        unless ($pid) {
1017                my @exec = qw/svn log -v/;
1018                push @exec, "-r$_revision" if defined $_revision;
1019                exec @exec, $u.$pfx or croak $!;
1020        }
1021        my ($r, $mp) = (undef, undef);
1022        while (<$fh>) {
1023                chomp;
1024                if (/^\-{72}$/) {
1025                        $mp = $r = undef;
1026                } elsif (/^r(\d+) \| /) {
1027                        $r = $1 unless defined $r;
1028                } elsif (/^Changed paths:/) {
1029                        $mp = 1;
1030                } elsif ($mp && m#^   [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) {
1031                        my ($p1, $p0, $r0) = ($1, $2, $3);
1032                        my $c = find_graft_path_commit($paths, $p1, $r);
1033                        next unless $c;
1034                        find_graft_path_parents($grafts, $paths, $c, $p0, $r0);
1035                }
1036        }
1037}
1038
1039sub graft_file_copy_lib {
1040        my ($grafts, $l_map, $u) = @_;
1041        my $tree_paths = $l_map->{$u};
1042        my $pfx = common_prefix([keys %$tree_paths]);
1043        my ($repo, $path) = repo_path_split($u.$pfx);
1044        $SVN_LOG ||= libsvn_connect($repo);
1045        $SVN ||= libsvn_connect($repo);
1046
1047        my ($base, $head) = libsvn_parse_revision();
1048        my $inc = 1000;
1049        my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
1050        my $eh = $SVN::Error::handler;
1051        $SVN::Error::handler = \&libsvn_skip_unknown_revs;
1052        while (1) {
1053                my $pool = SVN::Pool->new;
1054                libsvn_get_log($SVN_LOG, "/$path", $min, $max, 0, 1, 1,
1055                        sub {
1056                                libsvn_graft_file_copies($grafts, $tree_paths,
1057                                                        $path, @_);
1058                        }, $pool);
1059                $pool->clear;
1060                last if ($max >= $head);
1061                $min = $max + 1;
1062                $max += $inc;
1063                $max = $head if ($max > $head);
1064        }
1065        $SVN::Error::handler = $eh;
1066}
1067
1068sub process_merge_msg_matches {
1069        my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
1070        my (@strong, @weak);
1071        foreach (@matches) {
1072                # merging with ourselves is not interesting
1073                next if $_ eq $p;
1074                if ($l_map->{$u}->{$_}) {
1075                        push @strong, $_;
1076                } else {
1077                        push @weak, $_;
1078                }
1079        }
1080        foreach my $w (@weak) {
1081                last if @strong;
1082                # no exact match, use branch name as regexp.
1083                my $re = qr/\Q$w\E/i;
1084                foreach (keys %{$l_map->{$u}}) {
1085                        if (/$re/) {
1086                                push @strong, $l_map->{$u}->{$_};
1087                                last;
1088                        }
1089                }
1090                last if @strong;
1091                $w = basename($w);
1092                $re = qr/\Q$w\E/i;
1093                foreach (keys %{$l_map->{$u}}) {
1094                        if (/$re/) {
1095                                push @strong, $l_map->{$u}->{$_};
1096                                last;
1097                        }
1098                }
1099        }
1100        my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
1101                                        \s(?:[a-f\d\-]+)$/xsm);
1102        unless (defined $rev) {
1103                ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
1104                                        \@(?:[a-f\d\-]+)/xsm);
1105                return unless defined $rev;
1106        }
1107        foreach my $m (@strong) {
1108                my ($r0, $s0) = find_rev_before($rev, $m, 1);
1109                $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
1110        }
1111}
1112
1113sub graft_merge_msg {
1114        my ($grafts, $l_map, $u, $p, @re) = @_;
1115
1116        my $x = $l_map->{$u}->{$p};
1117        my $rl = rev_list_raw($x);
1118        while (my $c = next_rev_list_entry($rl)) {
1119                foreach my $re (@re) {
1120                        my (@br) = ($c->{m} =~ /$re/g);
1121                        next unless @br;
1122                        process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
1123                }
1124        }
1125}
1126
1127sub read_uuid {
1128        return if $SVN_UUID;
1129        if ($_use_lib) {
1130                my $pool = SVN::Pool->new;
1131                $SVN_UUID = $SVN->get_uuid($pool);
1132                $pool->clear;
1133        } else {
1134                my $info = shift || svn_info('.');
1135                $SVN_UUID = $info->{'Repository UUID'} or
1136                                        croak "Repository UUID unreadable\n";
1137        }
1138}
1139
1140sub quiet_run {
1141        my $pid = fork;
1142        defined $pid or croak $!;
1143        if (!$pid) {
1144                open my $null, '>', '/dev/null' or croak $!;
1145                open STDERR, '>&', $null or croak $!;
1146                open STDOUT, '>&', $null or croak $!;
1147                exec @_ or croak $!;
1148        }
1149        waitpid $pid, 0;
1150        return $?;
1151}
1152
1153sub repo_path_split {
1154        my $full_url = shift;
1155        $full_url =~ s#/+$##;
1156
1157        foreach (@repo_path_split_cache) {
1158                if ($full_url =~ s#$_##) {
1159                        my $u = $1;
1160                        $full_url =~ s#^/+##;
1161                        return ($u, $full_url);
1162                }
1163        }
1164
1165        if ($_use_lib) {
1166                my $tmp = libsvn_connect($full_url);
1167                my $url = $tmp->get_repos_root;
1168                $full_url =~ s#^\Q$url\E/*##;
1169                push @repo_path_split_cache, qr/^(\Q$url\E)/;
1170                return ($url, $full_url);
1171        } else {
1172                my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i);
1173                $path =~ s#^/+##;
1174                my @paths = split(m#/+#, $path);
1175                while (quiet_run(qw/svn ls --non-interactive/, $url)) {
1176                        my $n = shift @paths || last;
1177                        $url .= "/$n";
1178                }
1179                push @repo_path_split_cache, qr/^(\Q$url\E)/;
1180                $path = join('/',@paths);
1181                return ($url, $path);
1182        }
1183}
1184
1185sub setup_git_svn {
1186        defined $SVN_URL or croak "SVN repository location required\n";
1187        unless (-d $GIT_DIR) {
1188                croak "GIT_DIR=$GIT_DIR does not exist!\n";
1189        }
1190        mkpath([$GIT_SVN_DIR]);
1191        mkpath(["$GIT_SVN_DIR/info"]);
1192        open my $fh, '>>',$REVDB or croak $!;
1193        close $fh;
1194        s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1195
1196}
1197
1198sub assert_svn_wc_clean {
1199        return if $_use_lib;
1200        my ($svn_rev) = @_;
1201        croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
1202        my $lcr = svn_info('.')->{'Last Changed Rev'};
1203        if ($svn_rev != $lcr) {
1204                print STDERR "Checking for copy-tree ... ";
1205                my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
1206                                                "-r$lcr:$svn_rev")));
1207                if (@diff) {
1208                        croak "Nope!  Expected r$svn_rev, got r$lcr\n";
1209                } else {
1210                        print STDERR "OK!\n";
1211                }
1212        }
1213        my @status = grep(!/^Performing status on external/,(`svn status`));
1214        @status = grep(!/^\s*$/,@status);
1215        if (scalar @status) {
1216                print STDERR "Tree ($SVN_WC) is not clean:\n";
1217                print STDERR $_ foreach @status;
1218                croak;
1219        }
1220}
1221
1222sub get_tree_from_treeish {
1223        my ($treeish) = @_;
1224        croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1225        chomp(my $type = `git-cat-file -t $treeish`);
1226        my $expected;
1227        while ($type eq 'tag') {
1228                chomp(($treeish, $type) = `git-cat-file tag $treeish`);
1229        }
1230        if ($type eq 'commit') {
1231                $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
1232                ($expected) = ($expected =~ /^tree ($sha1)$/);
1233                die "Unable to get tree from $treeish\n" unless $expected;
1234        } elsif ($type eq 'tree') {
1235                $expected = $treeish;
1236        } else {
1237                die "$treeish is a $type, expected tree, tag or commit\n";
1238        }
1239        return $expected;
1240}
1241
1242sub assert_tree {
1243        return if $_use_lib;
1244        my ($treeish) = @_;
1245        my $expected = get_tree_from_treeish($treeish);
1246
1247        my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
1248        if (-e $tmpindex) {
1249                unlink $tmpindex or croak $!;
1250        }
1251        my $old_index = set_index($tmpindex);
1252        index_changes(1);
1253        chomp(my $tree = `git-write-tree`);
1254        restore_index($old_index);
1255        if ($tree ne $expected) {
1256                croak "Tree mismatch, Got: $tree, Expected: $expected\n";
1257        }
1258        unlink $tmpindex;
1259}
1260
1261sub parse_diff_tree {
1262        my $diff_fh = shift;
1263        local $/ = "\0";
1264        my $state = 'meta';
1265        my @mods;
1266        while (<$diff_fh>) {
1267                chomp $_; # this gets rid of the trailing "\0"
1268                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1269                                        $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1270                        push @mods, {   mode_a => $1, mode_b => $2,
1271                                        sha1_b => $3, chg => $4 };
1272                        if ($4 =~ /^(?:C|R)$/) {
1273                                $state = 'file_a';
1274                        } else {
1275                                $state = 'file_b';
1276                        }
1277                } elsif ($state eq 'file_a') {
1278                        my $x = $mods[$#mods] or croak "Empty array\n";
1279                        if ($x->{chg} !~ /^(?:C|R)$/) {
1280                                croak "Error parsing $_, $x->{chg}\n";
1281                        }
1282                        $x->{file_a} = $_;
1283                        $state = 'file_b';
1284                } elsif ($state eq 'file_b') {
1285                        my $x = $mods[$#mods] or croak "Empty array\n";
1286                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1287                                croak "Error parsing $_, $x->{chg}\n";
1288                        }
1289                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1290                                croak "Error parsing $_, $x->{chg}\n";
1291                        }
1292                        $x->{file_b} = $_;
1293                        $state = 'meta';
1294                } else {
1295                        croak "Error parsing $_\n";
1296                }
1297        }
1298        close $diff_fh or croak $?;
1299
1300        return \@mods;
1301}
1302
1303sub svn_check_prop_executable {
1304        my $m = shift;
1305        return if -l $m->{file_b};
1306        if ($m->{mode_b} =~ /755$/) {
1307                chmod((0755 &~ umask),$m->{file_b}) or croak $!;
1308                if ($m->{mode_a} !~ /755$/) {
1309                        sys(qw(svn propset svn:executable 1), $m->{file_b});
1310                }
1311                -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
1312        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
1313                sys(qw(svn propdel svn:executable), $m->{file_b});
1314                chmod((0644 &~ umask),$m->{file_b}) or croak $!;
1315                -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
1316        }
1317}
1318
1319sub svn_ensure_parent_path {
1320        my $dir_b = dirname(shift);
1321        svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
1322        mkpath([$dir_b]) unless (-d $dir_b);
1323        sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
1324}
1325
1326sub precommit_check {
1327        my $mods = shift;
1328        my (%rm_file, %rmdir_check, %added_check);
1329
1330        my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
1331        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1332                if ($m->{chg} eq 'R') {
1333                        if (-d $m->{file_b}) {
1334                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1335                        }
1336                        # dir/$file => dir/file/$file
1337                        my $dirname = dirname($m->{file_b});
1338                        while ($dirname ne File::Spec->curdir) {
1339                                if ($dirname ne $m->{file_a}) {
1340                                        $dirname = dirname($dirname);
1341                                        next;
1342                                }
1343                                err_file_to_dir("$m->{file_a} => $m->{file_b}");
1344                        }
1345                        # baz/zzz => baz (baz is a file)
1346                        $dirname = dirname($m->{file_a});
1347                        while ($dirname ne File::Spec->curdir) {
1348                                if ($dirname ne $m->{file_b}) {
1349                                        $dirname = dirname($dirname);
1350                                        next;
1351                                }
1352                                err_dir_to_file("$m->{file_a} => $m->{file_b}");
1353                        }
1354                }
1355                if ($m->{chg} =~ /^(D|R)$/) {
1356                        my $t = $1 eq 'D' ? 'file_b' : 'file_a';
1357                        $rm_file{ $m->{$t} } = 1;
1358                        my $dirname = dirname( $m->{$t} );
1359                        my $basename = basename( $m->{$t} );
1360                        $rmdir_check{$dirname}->{$basename} = 1;
1361                } elsif ($m->{chg} =~ /^(?:A|C)$/) {
1362                        if (-d $m->{file_b}) {
1363                                err_dir_to_file($m->{file_b});
1364                        }
1365                        my $dirname = dirname( $m->{file_b} );
1366                        my $basename = basename( $m->{file_b} );
1367                        $added_check{$dirname}->{$basename} = 1;
1368                        while ($dirname ne File::Spec->curdir) {
1369                                if ($rm_file{$dirname}) {
1370                                        err_file_to_dir($m->{file_b});
1371                                }
1372                                $dirname = dirname $dirname;
1373                        }
1374                }
1375        }
1376        return (\%rmdir_check, \%added_check);
1377
1378        sub err_dir_to_file {
1379                my $file = shift;
1380                print STDERR "Node change from directory to file ",
1381                                "is not supported by Subversion: ",$file,"\n";
1382                exit 1;
1383        }
1384        sub err_file_to_dir {
1385                my $file = shift;
1386                print STDERR "Node change from file to directory ",
1387                                "is not supported by Subversion: ",$file,"\n";
1388                exit 1;
1389        }
1390}
1391
1392
1393sub get_diff {
1394        my ($from, $treeish) = @_;
1395        assert_tree($from);
1396        print "diff-tree $from $treeish\n";
1397        my $pid = open my $diff_fh, '-|';
1398        defined $pid or croak $!;
1399        if ($pid == 0) {
1400                my @diff_tree = qw(git-diff-tree -z -r);
1401                if ($_cp_similarity) {
1402                        push @diff_tree, "-C$_cp_similarity";
1403                } else {
1404                        push @diff_tree, '-C';
1405                }
1406                push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1407                push @diff_tree, "-l$_l" if defined $_l;
1408                exec(@diff_tree, $from, $treeish) or croak $!;
1409        }
1410        return parse_diff_tree($diff_fh);
1411}
1412
1413sub svn_checkout_tree {
1414        my ($from, $treeish) = @_;
1415        my $mods = get_diff($from->{commit}, $treeish);
1416        return $mods unless (scalar @$mods);
1417        my ($rm, $add) = precommit_check($mods);
1418
1419        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1420        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1421                if ($m->{chg} eq 'C') {
1422                        svn_ensure_parent_path( $m->{file_b} );
1423                        sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
1424                        apply_mod_line_blob($m);
1425                        svn_check_prop_executable($m);
1426                } elsif ($m->{chg} eq 'D') {
1427                        sys(qw(svn rm --force), $m->{file_b});
1428                } elsif ($m->{chg} eq 'R') {
1429                        svn_ensure_parent_path( $m->{file_b} );
1430                        sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
1431                        apply_mod_line_blob($m);
1432                        svn_check_prop_executable($m);
1433                } elsif ($m->{chg} eq 'M') {
1434                        apply_mod_line_blob($m);
1435                        svn_check_prop_executable($m);
1436                } elsif ($m->{chg} eq 'T') {
1437                        sys(qw(svn rm --force),$m->{file_b});
1438                        apply_mod_line_blob($m);
1439                        sys(qw(svn add), $m->{file_b});
1440                        svn_check_prop_executable($m);
1441                } elsif ($m->{chg} eq 'A') {
1442                        svn_ensure_parent_path( $m->{file_b} );
1443                        apply_mod_line_blob($m);
1444                        sys(qw(svn add), $m->{file_b});
1445                        svn_check_prop_executable($m);
1446                } else {
1447                        croak "Invalid chg: $m->{chg}\n";
1448                }
1449        }
1450
1451        assert_tree($treeish);
1452        if ($_rmdir) { # remove empty directories
1453                handle_rmdir($rm, $add);
1454        }
1455        assert_tree($treeish);
1456        return $mods;
1457}
1458
1459sub libsvn_checkout_tree {
1460        my ($from, $treeish, $ed) = @_;
1461        my $mods = get_diff($from, $treeish);
1462        return $mods unless (scalar @$mods);
1463        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1464        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1465                my $f = $m->{chg};
1466                if (defined $o{$f}) {
1467                        $ed->$f($m, $_q);
1468                } else {
1469                        croak "Invalid change type: $f\n";
1470                }
1471        }
1472        $ed->rmdirs($_q) if $_rmdir;
1473        return $mods;
1474}
1475
1476# svn ls doesn't work with respect to the current working tree, but what's
1477# in the repository.  There's not even an option for it... *sigh*
1478# (added files don't show up and removed files remain in the ls listing)
1479sub svn_ls_current {
1480        my ($dir, $rm, $add) = @_;
1481        chomp(my @ls = safe_qx('svn','ls',$dir));
1482        my @ret = ();
1483        foreach (@ls) {
1484                s#/$##; # trailing slashes are evil
1485                push @ret, $_ unless $rm->{$dir}->{$_};
1486        }
1487        if (exists $add->{$dir}) {
1488                push @ret, keys %{$add->{$dir}};
1489        }
1490        return \@ret;
1491}
1492
1493sub handle_rmdir {
1494        my ($rm, $add) = @_;
1495
1496        foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
1497                my $ls = svn_ls_current($dir, $rm, $add);
1498                next if (scalar @$ls);
1499                sys(qw(svn rm --force),$dir);
1500
1501                my $dn = dirname $dir;
1502                $rm->{ $dn }->{ basename $dir } = 1;
1503                $ls = svn_ls_current($dn, $rm, $add);
1504                while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
1505                        sys(qw(svn rm --force),$dn);
1506                        $dir = basename $dn;
1507                        $dn = dirname $dn;
1508                        $rm->{ $dn }->{ $dir } = 1;
1509                        $ls = svn_ls_current($dn, $rm, $add);
1510                }
1511        }
1512}
1513
1514sub get_commit_message {
1515        my ($commit, $commit_msg) = (@_);
1516        my %log_msg = ( msg => '' );
1517        open my $msg, '>', $commit_msg or croak $!;
1518
1519        chomp(my $type = `git-cat-file -t $commit`);
1520        if ($type eq 'commit' || $type eq 'tag') {
1521                my $pid = open my $msg_fh, '-|';
1522                defined $pid or croak $!;
1523
1524                if ($pid == 0) {
1525                        exec('git-cat-file', $type, $commit) or croak $!;
1526                }
1527                my $in_msg = 0;
1528                while (<$msg_fh>) {
1529                        if (!$in_msg) {
1530                                $in_msg = 1 if (/^\s*$/);
1531                        } elsif (/^git-svn-id: /) {
1532                                # skip this, we regenerate the correct one
1533                                # on re-fetch anyways
1534                        } else {
1535                                print $msg $_ or croak $!;
1536                        }
1537                }
1538                close $msg_fh or croak $?;
1539        }
1540        close $msg or croak $!;
1541
1542        if ($_edit || ($type eq 'tree')) {
1543                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1544                system($editor, $commit_msg);
1545        }
1546
1547        # file_to_s removes all trailing newlines, so just use chomp() here:
1548        open $msg, '<', $commit_msg or croak $!;
1549        { local $/; chomp($log_msg{msg} = <$msg>); }
1550        close $msg or croak $!;
1551
1552        return \%log_msg;
1553}
1554
1555sub set_svn_commit_env {
1556        if (defined $LC_ALL) {
1557                $ENV{LC_ALL} = $LC_ALL;
1558        } else {
1559                delete $ENV{LC_ALL};
1560        }
1561}
1562
1563sub svn_commit_tree {
1564        my ($last, $commit) = @_;
1565        my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
1566        my $log_msg = get_commit_message($commit, $commit_msg);
1567        my ($oneline) = ($log_msg->{msg} =~ /([^\n\r]+)/);
1568        print "Committing $commit: $oneline\n";
1569
1570        set_svn_commit_env();
1571        my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
1572        $ENV{LC_ALL} = 'C';
1573        unlink $commit_msg;
1574        my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/);
1575        if (!defined $committed) {
1576                my $out = join("\n",@ci_output);
1577                print STDERR "W: Trouble parsing \`svn commit' output:\n\n",
1578                                $out, "\n\nAssuming English locale...";
1579                ($committed) = ($out =~ /^Committed revision \d+\./sm);
1580                defined $committed or die " FAILED!\n",
1581                        "Commit output failed to parse committed revision!\n",
1582                print STDERR " OK\n";
1583        }
1584
1585        my @svn_up = qw(svn up);
1586        push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
1587        if ($_optimize_commits && ($committed == ($last->{revision} + 1))) {
1588                push @svn_up, "-r$committed";
1589                sys(@svn_up);
1590                my $info = svn_info('.');
1591                my $date = $info->{'Last Changed Date'} or die "Missing date\n";
1592                if ($info->{'Last Changed Rev'} != $committed) {
1593                        croak "$info->{'Last Changed Rev'} != $committed\n"
1594                }
1595                my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1596                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1597                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1598                                         or croak "Failed to parse date: $date\n";
1599                $log_msg->{date} = "$tz $Y-$m-$d $H:$M:$S";
1600                $log_msg->{author} = $info->{'Last Changed Author'};
1601                $log_msg->{revision} = $committed;
1602                $log_msg->{msg} .= "\n";
1603                $log_msg->{parents} = [ $last->{commit} ];
1604                $log_msg->{commit} = git_commit($log_msg, $commit);
1605                return $log_msg;
1606        }
1607        # resync immediately
1608        push @svn_up, "-r$last->{revision}";
1609        sys(@svn_up);
1610        return fetch("$committed=$commit");
1611}
1612
1613sub rev_list_raw {
1614        my (@args) = @_;
1615        my $pid = open my $fh, '-|';
1616        defined $pid or croak $!;
1617        if (!$pid) {
1618                exec(qw/git-rev-list --pretty=raw/, @args) or croak $!;
1619        }
1620        return { fh => $fh, t => { } };
1621}
1622
1623sub next_rev_list_entry {
1624        my $rl = shift;
1625        my $fh = $rl->{fh};
1626        my $x = $rl->{t};
1627        while (<$fh>) {
1628                if (/^commit ($sha1)$/o) {
1629                        if ($x->{c}) {
1630                                $rl->{t} = { c => $1 };
1631                                return $x;
1632                        } else {
1633                                $x->{c} = $1;
1634                        }
1635                } elsif (/^parent ($sha1)$/o) {
1636                        $x->{p}->{$1} = 1;
1637                } elsif (s/^    //) {
1638                        $x->{m} ||= '';
1639                        $x->{m} .= $_;
1640                }
1641        }
1642        return ($x != $rl->{t}) ? $x : undef;
1643}
1644
1645# read the entire log into a temporary file (which is removed ASAP)
1646# and store the file handle + parser state
1647sub svn_log_raw {
1648        my (@log_args) = @_;
1649        my $log_fh = IO::File->new_tmpfile or croak $!;
1650        my $pid = fork;
1651        defined $pid or croak $!;
1652        if (!$pid) {
1653                open STDOUT, '>&', $log_fh or croak $!;
1654                exec (qw(svn log), @log_args) or croak $!
1655        }
1656        waitpid $pid, 0;
1657        croak $? if $?;
1658        seek $log_fh, 0, 0 or croak $!;
1659        return { state => 'sep', fh => $log_fh };
1660}
1661
1662sub next_log_entry {
1663        my $log = shift; # retval of svn_log_raw()
1664        my $ret = undef;
1665        my $fh = $log->{fh};
1666
1667        while (<$fh>) {
1668                chomp;
1669                if (/^\-{72}$/) {
1670                        if ($log->{state} eq 'msg') {
1671                                if ($ret->{lines}) {
1672                                        $ret->{msg} .= $_."\n";
1673                                        unless(--$ret->{lines}) {
1674                                                $log->{state} = 'sep';
1675                                        }
1676                                } else {
1677                                        croak "Log parse error at: $_\n",
1678                                                $ret->{revision},
1679                                                "\n";
1680                                }
1681                                next;
1682                        }
1683                        if ($log->{state} ne 'sep') {
1684                                croak "Log parse error at: $_\n",
1685                                        "state: $log->{state}\n",
1686                                        $ret->{revision},
1687                                        "\n";
1688                        }
1689                        $log->{state} = 'rev';
1690
1691                        # if we have an empty log message, put something there:
1692                        if ($ret) {
1693                                $ret->{msg} ||= "\n";
1694                                delete $ret->{lines};
1695                                return $ret;
1696                        }
1697                        next;
1698                }
1699                if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
1700                        my $rev = $1;
1701                        my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
1702                        ($lines) = ($lines =~ /(\d+)/);
1703                        my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1704                                        /(\d{4})\-(\d\d)\-(\d\d)\s
1705                                         (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1706                                         or croak "Failed to parse date: $date\n";
1707                        $ret = {        revision => $rev,
1708                                        date => "$tz $Y-$m-$d $H:$M:$S",
1709                                        author => $author,
1710                                        lines => $lines,
1711                                        msg => '' };
1712                        if (defined $_authors && ! defined $users{$author}) {
1713                                die "Author: $author not defined in ",
1714                                                "$_authors file\n";
1715                        }
1716                        $log->{state} = 'msg_start';
1717                        next;
1718                }
1719                # skip the first blank line of the message:
1720                if ($log->{state} eq 'msg_start' && /^$/) {
1721                        $log->{state} = 'msg';
1722                } elsif ($log->{state} eq 'msg') {
1723                        if ($ret->{lines}) {
1724                                $ret->{msg} .= $_."\n";
1725                                unless (--$ret->{lines}) {
1726                                        $log->{state} = 'sep';
1727                                }
1728                        } else {
1729                                croak "Log parse error at: $_\n",
1730                                        $ret->{revision},"\n";
1731                        }
1732                }
1733        }
1734        return $ret;
1735}
1736
1737sub svn_info {
1738        my $url = shift || $SVN_URL;
1739
1740        my $pid = open my $info_fh, '-|';
1741        defined $pid or croak $!;
1742
1743        if ($pid == 0) {
1744                exec(qw(svn info),$url) or croak $!;
1745        }
1746
1747        my $ret = {};
1748        # only single-lines seem to exist in svn info output
1749        while (<$info_fh>) {
1750                chomp $_;
1751                if (m#^([^:]+)\s*:\s*(\S.*)$#) {
1752                        $ret->{$1} = $2;
1753                        push @{$ret->{-order}}, $1;
1754                }
1755        }
1756        close $info_fh or croak $?;
1757        return $ret;
1758}
1759
1760sub sys { system(@_) == 0 or croak $? }
1761
1762sub do_update_index {
1763        my ($z_cmd, $cmd, $no_text_base) = @_;
1764
1765        my $z = open my $p, '-|';
1766        defined $z or croak $!;
1767        unless ($z) { exec @$z_cmd or croak $! }
1768
1769        my $pid = open my $ui, '|-';
1770        defined $pid or croak $!;
1771        unless ($pid) {
1772                exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!;
1773        }
1774        local $/ = "\0";
1775        while (my $x = <$p>) {
1776                chomp $x;
1777                if (!$no_text_base && lstat $x && ! -l _ &&
1778                                svn_propget_base('svn:keywords', $x)) {
1779                        my $mode = -x _ ? 0755 : 0644;
1780                        my ($v,$d,$f) = File::Spec->splitpath($x);
1781                        my $tb = File::Spec->catfile($d, '.svn', 'tmp',
1782                                                'text-base',"$f.svn-base");
1783                        $tb =~ s#^/##;
1784                        unless (-f $tb) {
1785                                $tb = File::Spec->catfile($d, '.svn',
1786                                                'text-base',"$f.svn-base");
1787                                $tb =~ s#^/##;
1788                        }
1789                        my @s = stat($x);
1790                        unlink $x or croak $!;
1791                        copy($tb, $x);
1792                        chmod(($mode &~ umask), $x) or croak $!;
1793                        utime $s[8], $s[9], $x;
1794                }
1795                print $ui $x,"\0";
1796        }
1797        close $ui or croak $?;
1798}
1799
1800sub index_changes {
1801        return if $_use_lib;
1802
1803        if (!-f "$GIT_SVN_DIR/info/exclude") {
1804                open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!;
1805                print $fd '.svn',"\n";
1806                close $fd or croak $!;
1807        }
1808        my $no_text_base = shift;
1809        do_update_index([qw/git-diff-files --name-only -z/],
1810                        'remove',
1811                        $no_text_base);
1812        do_update_index([qw/git-ls-files -z --others/,
1813                                "--exclude-from=$GIT_SVN_DIR/info/exclude"],
1814                        'add',
1815                        $no_text_base);
1816}
1817
1818sub s_to_file {
1819        my ($str, $file, $mode) = @_;
1820        open my $fd,'>',$file or croak $!;
1821        print $fd $str,"\n" or croak $!;
1822        close $fd or croak $!;
1823        chmod ($mode &~ umask, $file) if (defined $mode);
1824}
1825
1826sub file_to_s {
1827        my $file = shift;
1828        open my $fd,'<',$file or croak "$!: file: $file\n";
1829        local $/;
1830        my $ret = <$fd>;
1831        close $fd or croak $!;
1832        $ret =~ s/\s*$//s;
1833        return $ret;
1834}
1835
1836sub assert_revision_unknown {
1837        my $r = shift;
1838        if (my $c = revdb_get($REVDB, $r)) {
1839                croak "$r = $c already exists! Why are we refetching it?";
1840        }
1841}
1842
1843sub trees_eq {
1844        my ($x, $y) = @_;
1845        my @x = safe_qx('git-cat-file','commit',$x);
1846        my @y = safe_qx('git-cat-file','commit',$y);
1847        if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
1848                                || $y[0] !~ /^tree $sha1\n$/) {
1849                print STDERR "Trees not equal: $y[0] != $x[0]\n";
1850                return 0
1851        }
1852        return 1;
1853}
1854
1855sub git_commit {
1856        my ($log_msg, @parents) = @_;
1857        assert_revision_unknown($log_msg->{revision});
1858        map_tree_joins() if (@_branch_from && !%tree_map);
1859
1860        my (@tmp_parents, @exec_parents, %seen_parent);
1861        if (my $lparents = $log_msg->{parents}) {
1862                @tmp_parents = @$lparents
1863        }
1864        # commit parents can be conditionally bound to a particular
1865        # svn revision via: "svn_revno=commit_sha1", filter them out here:
1866        foreach my $p (@parents) {
1867                next unless defined $p;
1868                if ($p =~ /^(\d+)=($sha1_short)$/o) {
1869                        if ($1 == $log_msg->{revision}) {
1870                                push @tmp_parents, $2;
1871                        }
1872                } else {
1873                        push @tmp_parents, $p if $p =~ /$sha1_short/o;
1874                }
1875        }
1876        my $tree = $log_msg->{tree};
1877        if (!defined $tree) {
1878                my $index = set_index($GIT_SVN_INDEX);
1879                index_changes();
1880                chomp($tree = `git-write-tree`);
1881                croak $? if $?;
1882                restore_index($index);
1883        }
1884
1885        # just in case we clobber the existing ref, we still want that ref
1886        # as our parent:
1887        if (my $cur = eval { file_to_s("$GIT_DIR/refs/remotes/$GIT_SVN") }) {
1888                push @tmp_parents, $cur;
1889        }
1890
1891        if (exists $tree_map{$tree}) {
1892                foreach my $p (@{$tree_map{$tree}}) {
1893                        my $skip;
1894                        foreach (@tmp_parents) {
1895                                # see if a common parent is found
1896                                my $mb = eval {
1897                                        safe_qx('git-merge-base', $_, $p)
1898                                };
1899                                next if ($@ || $?);
1900                                $skip = 1;
1901                                last;
1902                        }
1903                        next if $skip;
1904                        my ($url_p, $r_p, $uuid_p) = cmt_metadata($p);
1905                        next if (($SVN_UUID eq $uuid_p) &&
1906                                                ($log_msg->{revision} > $r_p));
1907                        next if (defined $url_p && defined $SVN_URL &&
1908                                                ($SVN_UUID eq $uuid_p) &&
1909                                                ($url_p eq $SVN_URL));
1910                        push @tmp_parents, $p;
1911                }
1912        }
1913        foreach (@tmp_parents) {
1914                next if $seen_parent{$_};
1915                $seen_parent{$_} = 1;
1916                push @exec_parents, $_;
1917                # MAXPARENT is defined to 16 in commit-tree.c:
1918                last if @exec_parents > 16;
1919        }
1920
1921        set_commit_env($log_msg);
1922        my @exec = ('git-commit-tree', $tree);
1923        push @exec, '-p', $_  foreach @exec_parents;
1924        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1925                                                                or croak $!;
1926        print $msg_fh $log_msg->{msg} or croak $!;
1927        unless ($_no_metadata) {
1928                print $msg_fh "\ngit-svn-id: $SVN_URL\@$log_msg->{revision}",
1929                                        " $SVN_UUID\n" or croak $!;
1930        }
1931        $msg_fh->flush == 0 or croak $!;
1932        close $msg_fh or croak $!;
1933        chomp(my $commit = do { local $/; <$out_fh> });
1934        close $out_fh or croak $!;
1935        waitpid $pid, 0;
1936        croak $? if $?;
1937        if ($commit !~ /^$sha1$/o) {
1938                die "Failed to commit, invalid sha1: $commit\n";
1939        }
1940        sys('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
1941        revdb_set($REVDB, $log_msg->{revision}, $commit);
1942
1943        # this output is read via pipe, do not change:
1944        print "r$log_msg->{revision} = $commit\n";
1945        check_repack();
1946        return $commit;
1947}
1948
1949sub check_repack {
1950        if ($_repack && (--$_repack_nr == 0)) {
1951                $_repack_nr = $_repack;
1952                sys("git repack $_repack_flags");
1953        }
1954}
1955
1956sub set_commit_env {
1957        my ($log_msg) = @_;
1958        my $author = $log_msg->{author};
1959        if (!defined $author || length $author == 0) {
1960                $author = '(no author)';
1961        }
1962        my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
1963                                : ($author,"$author\@$SVN_UUID");
1964        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1965        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1966        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
1967}
1968
1969sub apply_mod_line_blob {
1970        my $m = shift;
1971        if ($m->{mode_b} =~ /^120/) {
1972                blob_to_symlink($m->{sha1_b}, $m->{file_b});
1973        } else {
1974                blob_to_file($m->{sha1_b}, $m->{file_b});
1975        }
1976}
1977
1978sub blob_to_symlink {
1979        my ($blob, $link) = @_;
1980        defined $link or croak "\$link not defined!\n";
1981        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
1982        if (-l $link || -f _) {
1983                unlink $link or croak $!;
1984        }
1985
1986        my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
1987        symlink $dest, $link or croak $!;
1988}
1989
1990sub blob_to_file {
1991        my ($blob, $file) = @_;
1992        defined $file or croak "\$file not defined!\n";
1993        croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
1994        if (-l $file || -f _) {
1995                unlink $file or croak $!;
1996        }
1997
1998        open my $blob_fh, '>', $file or croak "$!: $file\n";
1999        my $pid = fork;
2000        defined $pid or croak $!;
2001
2002        if ($pid == 0) {
2003                open STDOUT, '>&', $blob_fh or croak $!;
2004                exec('git-cat-file','blob',$blob) or croak $!;
2005        }
2006        waitpid $pid, 0;
2007        croak $? if $?;
2008
2009        close $blob_fh or croak $!;
2010}
2011
2012sub safe_qx {
2013        my $pid = open my $child, '-|';
2014        defined $pid or croak $!;
2015        if ($pid == 0) {
2016                exec(@_) or croak $!;
2017        }
2018        my @ret = (<$child>);
2019        close $child or croak $?;
2020        die $? if $?; # just in case close didn't error out
2021        return wantarray ? @ret : join('',@ret);
2022}
2023
2024sub svn_compat_check {
2025        if ($_follow_parent) {
2026                print STDERR 'E: --follow-parent functionality is only ',
2027                                "available when SVN libraries are used\n";
2028                exit 1;
2029        }
2030        my @co_help = safe_qx(qw(svn co -h));
2031        unless (grep /ignore-externals/,@co_help) {
2032                print STDERR "W: Installed svn version does not support ",
2033                                "--ignore-externals\n";
2034                $_no_ignore_ext = 1;
2035        }
2036        if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
2037                $_svn_co_url_revs = 1;
2038        }
2039        if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) {
2040                $_svn_pg_peg_revs = 1;
2041        }
2042
2043        # I really, really hope nobody hits this...
2044        unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
2045                print STDERR <<'';
2046W: The installed svn version does not support the --stop-on-copy flag in
2047   the log command.
2048   Lets hope the directory you're tracking is not a branch or tag
2049   and was never moved within the repository...
2050
2051                $_no_stop_copy = 1;
2052        }
2053}
2054
2055# *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
2056# (and they won't honor URL@<rev> without -r<rev>, too!)
2057sub svn_cmd_checkout {
2058        my ($url, $rev, $dir) = @_;
2059        my @cmd = ('svn','co', "-r$rev");
2060        push @cmd, '--ignore-externals' unless $_no_ignore_ext;
2061        $url .= "\@$rev" if $_svn_co_url_revs;
2062        sys(@cmd, $url, $dir);
2063}
2064
2065sub check_upgrade_needed {
2066        if (!-r $REVDB) {
2067                -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2068                open my $fh, '>>',$REVDB or croak $!;
2069                close $fh;
2070        }
2071        my $old = eval {
2072                my $pid = open my $child, '-|';
2073                defined $pid or croak $!;
2074                if ($pid == 0) {
2075                        close STDERR;
2076                        exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $!;
2077                }
2078                my @ret = (<$child>);
2079                close $child or croak $?;
2080                die $? if $?; # just in case close didn't error out
2081                return wantarray ? @ret : join('',@ret);
2082        };
2083        return unless $old;
2084        my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
2085        if ($@ || !$head) {
2086                print STDERR "Please run: $0 rebuild --upgrade\n";
2087                exit 1;
2088        }
2089}
2090
2091# fills %tree_map with a reverse mapping of trees to commits.  Useful
2092# for finding parents to commit on.
2093sub map_tree_joins {
2094        my %seen;
2095        foreach my $br (@_branch_from) {
2096                my $pid = open my $pipe, '-|';
2097                defined $pid or croak $!;
2098                if ($pid == 0) {
2099                        exec(qw(git-rev-list --topo-order --pretty=raw), $br)
2100                                                                or croak $!;
2101                }
2102                while (<$pipe>) {
2103                        if (/^commit ($sha1)$/o) {
2104                                my $commit = $1;
2105
2106                                # if we've seen a commit,
2107                                # we've seen its parents
2108                                last if $seen{$commit};
2109                                my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
2110                                unless (defined $tree) {
2111                                        die "Failed to parse commit $commit\n";
2112                                }
2113                                push @{$tree_map{$tree}}, $commit;
2114                                $seen{$commit} = 1;
2115                        }
2116                }
2117                close $pipe; # we could be breaking the pipe early
2118        }
2119}
2120
2121sub load_all_refs {
2122        if (@_branch_from) {
2123                print STDERR '--branch|-b parameters are ignored when ',
2124                        "--branch-all-refs|-B is passed\n";
2125        }
2126
2127        # don't worry about rev-list on non-commit objects/tags,
2128        # it shouldn't blow up if a ref is a blob or tree...
2129        chomp(@_branch_from = `git-rev-parse --symbolic --all`);
2130}
2131
2132# '<svn username> = real-name <email address>' mapping based on git-svnimport:
2133sub load_authors {
2134        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2135        while (<$authors>) {
2136                chomp;
2137                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2138                my ($user, $name, $email) = ($1, $2, $3);
2139                $users{$user} = [$name, $email];
2140        }
2141        close $authors or croak $!;
2142}
2143
2144sub rload_authors {
2145        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2146        while (<$authors>) {
2147                chomp;
2148                next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2149                my ($user, $name, $email) = ($1, $2, $3);
2150                $rusers{"$name <$email>"} = $user;
2151        }
2152        close $authors or croak $!;
2153}
2154
2155sub svn_propget_base {
2156        my ($p, $f) = @_;
2157        $f .= '@BASE' if $_svn_pg_peg_revs;
2158        return safe_qx(qw/svn propget/, $p, $f);
2159}
2160
2161sub git_svn_each {
2162        my $sub = shift;
2163        foreach (`git-rev-parse --symbolic --all`) {
2164                next unless s#^refs/remotes/##;
2165                chomp $_;
2166                next unless -f "$GIT_DIR/svn/$_/info/url";
2167                &$sub($_);
2168        }
2169}
2170
2171sub migrate_revdb {
2172        git_svn_each(sub {
2173                my $id = shift;
2174                defined(my $pid = fork) or croak $!;
2175                if (!$pid) {
2176                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2177                        init_vars();
2178                        exit 0 if -r $REVDB;
2179                        print "Upgrading svn => git mapping...\n";
2180                        -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2181                        open my $fh, '>>',$REVDB or croak $!;
2182                        close $fh;
2183                        rebuild();
2184                        print "Done upgrading. You may now delete the ",
2185                                "deprecated $GIT_SVN_DIR/revs directory\n";
2186                        exit 0;
2187                }
2188                waitpid $pid, 0;
2189                croak $? if $?;
2190        });
2191}
2192
2193sub migration_check {
2194        migrate_revdb() unless (-e $REVDB);
2195        return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
2196        print "Upgrading repository...\n";
2197        unless (-d "$GIT_DIR/svn") {
2198                mkdir "$GIT_DIR/svn" or croak $!;
2199        }
2200        print "Data from a previous version of git-svn exists, but\n\t",
2201                                "$GIT_SVN_DIR\n\t(required for this version ",
2202                                "($VERSION) of git-svn) does not.\n";
2203
2204        foreach my $x (`git-rev-parse --symbolic --all`) {
2205                next unless $x =~ s#^refs/remotes/##;
2206                chomp $x;
2207                next unless -f "$GIT_DIR/$x/info/url";
2208                my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
2209                next unless $u;
2210                my $dn = dirname("$GIT_DIR/svn/$x");
2211                mkpath([$dn]) unless -d $dn;
2212                rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
2213        }
2214        migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
2215        print "Done upgrading.\n";
2216}
2217
2218sub find_rev_before {
2219        my ($r, $id, $eq_ok) = @_;
2220        my $f = "$GIT_DIR/svn/$id/.rev_db";
2221        return (undef,undef) unless -r $f;
2222        --$r unless $eq_ok;
2223        while ($r > 0) {
2224                if (my $c = revdb_get($f, $r)) {
2225                        return ($r, $c);
2226                }
2227                --$r;
2228        }
2229        return (undef, undef);
2230}
2231
2232sub init_vars {
2233        $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
2234        $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
2235        $REVDB = "$GIT_SVN_DIR/.rev_db";
2236        $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
2237        $SVN_URL = undef;
2238        $SVN_WC = "$GIT_SVN_DIR/tree";
2239        %tree_map = ();
2240}
2241
2242# convert GetOpt::Long specs for use by git-repo-config
2243sub read_repo_config {
2244        return unless -d $GIT_DIR;
2245        my $opts = shift;
2246        foreach my $o (keys %$opts) {
2247                my $v = $opts->{$o};
2248                my ($key) = ($o =~ /^([a-z\-]+)/);
2249                $key =~ s/-//g;
2250                my $arg = 'git-repo-config';
2251                $arg .= ' --int' if ($o =~ /[:=]i$/);
2252                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
2253                if (ref $v eq 'ARRAY') {
2254                        chomp(my @tmp = `$arg --get-all svn.$key`);
2255                        @$v = @tmp if @tmp;
2256                } else {
2257                        chomp(my $tmp = `$arg --get svn.$key`);
2258                        if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
2259                                $$v = $tmp;
2260                        }
2261                }
2262        }
2263}
2264
2265sub set_default_vals {
2266        if (defined $_repack) {
2267                $_repack = 1000 if ($_repack <= 0);
2268                $_repack_nr = $_repack;
2269                $_repack_flags ||= '-d';
2270        }
2271}
2272
2273sub read_grafts {
2274        my $gr_file = shift;
2275        my ($grafts, $comments) = ({}, {});
2276        if (open my $fh, '<', $gr_file) {
2277                my @tmp;
2278                while (<$fh>) {
2279                        if (/^($sha1)\s+/) {
2280                                my $c = $1;
2281                                if (@tmp) {
2282                                        @{$comments->{$c}} = @tmp;
2283                                        @tmp = ();
2284                                }
2285                                foreach my $p (split /\s+/, $_) {
2286                                        $grafts->{$c}->{$p} = 1;
2287                                }
2288                        } else {
2289                                push @tmp, $_;
2290                        }
2291                }
2292                close $fh or croak $!;
2293                @{$comments->{'END'}} = @tmp if @tmp;
2294        }
2295        return ($grafts, $comments);
2296}
2297
2298sub write_grafts {
2299        my ($grafts, $comments, $gr_file) = @_;
2300
2301        open my $fh, '>', $gr_file or croak $!;
2302        foreach my $c (sort keys %$grafts) {
2303                if ($comments->{$c}) {
2304                        print $fh $_ foreach @{$comments->{$c}};
2305                }
2306                my $p = $grafts->{$c};
2307                my %x; # real parents
2308                delete $p->{$c}; # commits are not self-reproducing...
2309                my $pid = open my $ch, '-|';
2310                defined $pid or croak $!;
2311                if (!$pid) {
2312                        exec(qw/git-cat-file commit/, $c) or croak $!;
2313                }
2314                while (<$ch>) {
2315                        if (/^parent ($sha1)/) {
2316                                $x{$1} = $p->{$1} = 1;
2317                        } else {
2318                                last unless /^\S/;
2319                        }
2320                }
2321                close $ch; # breaking the pipe
2322
2323                # if real parents are the only ones in the grafts, drop it
2324                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2325
2326                my (@ip, @jp, $mb);
2327                my %del = %x;
2328                @ip = @jp = keys %$p;
2329                foreach my $i (@ip) {
2330                        next if $del{$i} || $p->{$i} == 2;
2331                        foreach my $j (@jp) {
2332                                next if $i eq $j || $del{$j} || $p->{$j} == 2;
2333                                $mb = eval { safe_qx('git-merge-base',$i,$j) };
2334                                next unless $mb;
2335                                chomp $mb;
2336                                next if $x{$mb};
2337                                if ($mb eq $j) {
2338                                        delete $p->{$i};
2339                                        $del{$i} = 1;
2340                                } elsif ($mb eq $i) {
2341                                        delete $p->{$j};
2342                                        $del{$j} = 1;
2343                                }
2344                        }
2345                }
2346
2347                # if real parents are the only ones in the grafts, drop it
2348                next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2349
2350                print $fh $c, ' ', join(' ', sort keys %$p),"\n";
2351        }
2352        if ($comments->{'END'}) {
2353                print $fh $_ foreach @{$comments->{'END'}};
2354        }
2355        close $fh or croak $!;
2356}
2357
2358sub read_url_paths_all {
2359        my ($l_map, $pfx, $p) = @_;
2360        my @dir;
2361        foreach (<$p/*>) {
2362                if (-r "$_/info/url") {
2363                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2364                        my $id = $pfx . basename $_;
2365                        my $url = file_to_s("$_/info/url");
2366                        my ($u, $p) = repo_path_split($url);
2367                        $l_map->{$u}->{$p} = $id;
2368                } elsif (-d $_) {
2369                        push @dir, $_;
2370                }
2371        }
2372        foreach (@dir) {
2373                my $x = $_;
2374                $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
2375                read_url_paths_all($l_map, $x, $_);
2376        }
2377}
2378
2379# this one only gets ids that have been imported, not new ones
2380sub read_url_paths {
2381        my $l_map = {};
2382        git_svn_each(sub { my $x = shift;
2383                        my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
2384                        my ($u, $p) = repo_path_split($url);
2385                        $l_map->{$u}->{$p} = $x;
2386                        });
2387        return $l_map;
2388}
2389
2390sub extract_metadata {
2391        my $id = shift or return (undef, undef, undef);
2392        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
2393                                                        \s([a-f\d\-]+)$/x);
2394        if (!$rev || !$uuid || !$url) {
2395                # some of the original repositories I made had
2396                # identifiers like this:
2397                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
2398        }
2399        return ($url, $rev, $uuid);
2400}
2401
2402sub cmt_metadata {
2403        return extract_metadata((grep(/^git-svn-id: /,
2404                safe_qx(qw/git-cat-file commit/, shift)))[-1]);
2405}
2406
2407sub get_commit_time {
2408        my $cmt = shift;
2409        defined(my $pid = open my $fh, '-|') or croak $!;
2410        if (!$pid) {
2411                exec qw/git-rev-list --pretty=raw -n1/, $cmt or croak $!;
2412        }
2413        while (<$fh>) {
2414                /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
2415                my ($s, $tz) = ($1, $2);
2416                if ($tz =~ s/^\+//) {
2417                        $s += tz_to_s_offset($tz);
2418                } elsif ($tz =~ s/^\-//) {
2419                        $s -= tz_to_s_offset($tz);
2420                }
2421                close $fh;
2422                return $s;
2423        }
2424        die "Can't get commit time for commit: $cmt\n";
2425}
2426
2427sub tz_to_s_offset {
2428        my ($tz) = @_;
2429        $tz =~ s/(\d\d)$//;
2430        return ($1 * 60) + ($tz * 3600);
2431}
2432
2433sub setup_pager { # translated to Perl from pager.c
2434        return unless (-t *STDOUT);
2435        my $pager = $ENV{PAGER};
2436        if (!defined $pager) {
2437                $pager = 'less';
2438        } elsif (length $pager == 0 || $pager eq 'cat') {
2439                return;
2440        }
2441        pipe my $rfd, my $wfd or return;
2442        defined(my $pid = fork) or croak $!;
2443        if (!$pid) {
2444                open STDOUT, '>&', $wfd or croak $!;
2445                return;
2446        }
2447        open STDIN, '<&', $rfd or croak $!;
2448        $ENV{LESS} ||= '-S';
2449        exec $pager or croak "Can't run pager: $!\n";;
2450}
2451
2452sub get_author_info {
2453        my ($dest, $author, $t, $tz) = @_;
2454        $author =~ s/(?:^\s*|\s*$)//g;
2455        $dest->{a_raw} = $author;
2456        my $_a;
2457        if ($_authors) {
2458                $_a = $rusers{$author} || undef;
2459        }
2460        if (!$_a) {
2461                ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
2462        }
2463        $dest->{t} = $t;
2464        $dest->{tz} = $tz;
2465        $dest->{a} = $_a;
2466        # Date::Parse isn't in the standard Perl distro :(
2467        if ($tz =~ s/^\+//) {
2468                $t += tz_to_s_offset($tz);
2469        } elsif ($tz =~ s/^\-//) {
2470                $t -= tz_to_s_offset($tz);
2471        }
2472        $dest->{t_utc} = $t;
2473}
2474
2475sub process_commit {
2476        my ($c, $r_min, $r_max, $defer) = @_;
2477        if (defined $r_min && defined $r_max) {
2478                if ($r_min == $c->{r} && $r_min == $r_max) {
2479                        show_commit($c);
2480                        return 0;
2481                }
2482                return 1 if $r_min == $r_max;
2483                if ($r_min < $r_max) {
2484                        # we need to reverse the print order
2485                        return 0 if (defined $_limit && --$_limit < 0);
2486                        push @$defer, $c;
2487                        return 1;
2488                }
2489                if ($r_min != $r_max) {
2490                        return 1 if ($r_min < $c->{r});
2491                        return 1 if ($r_max > $c->{r});
2492                }
2493        }
2494        return 0 if (defined $_limit && --$_limit < 0);
2495        show_commit($c);
2496        return 1;
2497}
2498
2499sub show_commit {
2500        my $c = shift;
2501        if ($_oneline) {
2502                my $x = "\n";
2503                if (my $l = $c->{l}) {
2504                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2505                        $x = $l->[0];
2506                }
2507                $_l_fmt ||= 'A' . length($c->{r});
2508                print 'r',pack($_l_fmt, $c->{r}),' | ';
2509                print "$c->{c} | " if $_show_commit;
2510                print $x;
2511        } else {
2512                show_commit_normal($c);
2513        }
2514}
2515
2516sub show_commit_normal {
2517        my ($c) = @_;
2518        print '-' x72, "\nr$c->{r} | ";
2519        print "$c->{c} | " if $_show_commit;
2520        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2521                                 localtime($c->{t_utc})), ' | ';
2522        my $nr_line = 0;
2523
2524        if (my $l = $c->{l}) {
2525                while ($l->[$#$l] eq "\n" && $l->[($#$l - 1)] eq "\n") {
2526                        pop @$l;
2527                }
2528                $nr_line = scalar @$l;
2529                if (!$nr_line) {
2530                        print "1 line\n\n\n";
2531                } else {
2532                        if ($nr_line == 1) {
2533                                $nr_line = '1 line';
2534                        } else {
2535                                $nr_line .= ' lines';
2536                        }
2537                        print $nr_line, "\n\n";
2538                        print $_ foreach @$l;
2539                }
2540        } else {
2541                print "1 line\n\n";
2542
2543        }
2544        foreach my $x (qw/raw diff/) {
2545                if ($c->{$x}) {
2546                        print "\n";
2547                        print $_ foreach @{$c->{$x}}
2548                }
2549        }
2550}
2551
2552sub libsvn_load {
2553        return unless $_use_lib;
2554        $_use_lib = eval {
2555                require SVN::Core;
2556                if ($SVN::Core::VERSION lt '1.1.0') {
2557                        die "Need SVN::Core 1.1.0 or better ",
2558                                        "(got $SVN::Core::VERSION) ",
2559                                        "Falling back to command-line svn\n";
2560                }
2561                require SVN::Ra;
2562                require SVN::Delta;
2563                push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
2564                my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2565                                        $SVN::Node::dir.$SVN::Node::unknown.
2566                                        $SVN::Node::none.$SVN::Node::file.
2567                                        $SVN::Node::dir.$SVN::Node::unknown;
2568                1;
2569        };
2570}
2571
2572sub libsvn_connect {
2573        my ($url) = @_;
2574        my $auth = SVN::Core::auth_open([SVN::Client::get_simple_provider(),
2575                          SVN::Client::get_ssl_server_trust_file_provider(),
2576                          SVN::Client::get_username_provider()]);
2577        my $s = eval { SVN::Ra->new(url => $url, auth => $auth) };
2578        return $s;
2579}
2580
2581sub libsvn_get_file {
2582        my ($gui, $f, $rev) = @_;
2583        my $p = $f;
2584        if (length $SVN_PATH > 0) {
2585                return unless ($p =~ s#^\Q$SVN_PATH\E/##);
2586        }
2587
2588        my ($hash, $pid, $in, $out);
2589        my $pool = SVN::Pool->new;
2590        defined($pid = open3($in, $out, '>&STDERR',
2591                                qw/git-hash-object -w --stdin/)) or croak $!;
2592        # redirect STDOUT for SVN 1.1.x compatibility
2593        open my $stdout, '>&', \*STDOUT or croak $!;
2594        open STDOUT, '>&', $in or croak $!;
2595        my ($r, $props) = $SVN->get_file($f, $rev, \*STDOUT, $pool);
2596        $in->flush == 0 or croak $!;
2597        open STDOUT, '>&', $stdout or croak $!;
2598        close $in or croak $!;
2599        close $stdout or croak $!;
2600        $pool->clear;
2601        chomp($hash = do { local $/; <$out> });
2602        close $out or croak $!;
2603        waitpid $pid, 0;
2604        $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2605
2606        my $mode = exists $props->{'svn:executable'} ? '100755' : '100644';
2607        if (exists $props->{'svn:special'}) {
2608                $mode = '120000';
2609                my $link = `git-cat-file blob $hash`;
2610                $link =~ s/^link // or die "svn:special file with contents: <",
2611                                                $link, "> is not understood\n";
2612                defined($pid = open3($in, $out, '>&STDERR',
2613                                qw/git-hash-object -w --stdin/)) or croak $!;
2614                print $in $link;
2615                $in->flush == 0 or croak $!;
2616                close $in or croak $!;
2617                chomp($hash = do { local $/; <$out> });
2618                close $out or croak $!;
2619                waitpid $pid, 0;
2620                $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2621        }
2622        print $gui $mode,' ',$hash,"\t",$p,"\0" or croak $!;
2623}
2624
2625sub libsvn_log_entry {
2626        my ($rev, $author, $date, $msg, $parents) = @_;
2627        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2628                                         (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
2629                                or die "Unable to parse date: $date\n";
2630        if (defined $_authors && ! defined $users{$author}) {
2631                die "Author: $author not defined in $_authors file\n";
2632        }
2633        $msg = '' if ($rev == 0 && !defined $msg);
2634        return { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
2635                author => $author, msg => $msg."\n", parents => $parents || [] }
2636}
2637
2638sub process_rm {
2639        my ($gui, $last_commit, $f) = @_;
2640        $f =~ s#^\Q$SVN_PATH\E/?## or return;
2641        # remove entire directories.
2642        if (safe_qx('git-ls-tree',$last_commit,'--',$f) =~ /^040000 tree/) {
2643                defined(my $pid = open my $ls, '-|') or croak $!;
2644                if (!$pid) {
2645                        exec(qw/git-ls-tree -r --name-only -z/,
2646                                $last_commit,'--',$f) or croak $!;
2647                }
2648                local $/ = "\0";
2649                while (<$ls>) {
2650                        print $gui '0 ',0 x 40,"\t",$_ or croak $!;
2651                }
2652                close $ls or croak $?;
2653        } else {
2654                print $gui '0 ',0 x 40,"\t",$f,"\0" or croak $!;
2655        }
2656}
2657
2658sub libsvn_fetch {
2659        my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
2660        open my $gui, '| git-update-index -z --index-info' or croak $!;
2661        my @amr;
2662        foreach my $f (keys %$paths) {
2663                my $m = $paths->{$f}->action();
2664                $f =~ s#^/+##;
2665                if ($m =~ /^[DR]$/) {
2666                        print "\t$m\t$f\n" unless $_q;
2667                        process_rm($gui, $last_commit, $f);
2668                        next if $m eq 'D';
2669                        # 'R' can be file replacements, too, right?
2670                }
2671                my $pool = SVN::Pool->new;
2672                my $t = $SVN->check_path($f, $rev, $pool);
2673                if ($t == $SVN::Node::file) {
2674                        if ($m =~ /^[AMR]$/) {
2675                                push @amr, [ $m, $f ];
2676                        } else {
2677                                die "Unrecognized action: $m, ($f r$rev)\n";
2678                        }
2679                } elsif ($t == $SVN::Node::dir && $m =~ /^[AR]$/) {
2680                        my @traversed = ();
2681                        libsvn_traverse($gui, '', $f, $rev, \@traversed);
2682                        foreach (@traversed) {
2683                                push @amr, [ $m, $_ ]
2684                        }
2685                }
2686                $pool->clear;
2687        }
2688        foreach (@amr) {
2689                print "\t$_->[0]\t$_->[1]\n" unless $_q;
2690                libsvn_get_file($gui, $_->[1], $rev)
2691        }
2692        close $gui or croak $?;
2693        return libsvn_log_entry($rev, $author, $date, $msg, [$last_commit]);
2694}
2695
2696sub svn_grab_base_rev {
2697        defined(my $pid = open my $fh, '-|') or croak $!;
2698        if (!$pid) {
2699                open my $null, '>', '/dev/null' or croak $!;
2700                open STDERR, '>&', $null or croak $!;
2701                exec qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0"
2702                                                                or croak $!;
2703        }
2704        chomp(my $c = do { local $/; <$fh> });
2705        close $fh;
2706        if (defined $c && length $c) {
2707                my ($url, $rev, $uuid) = cmt_metadata($c);
2708                return ($rev, $c) if defined $rev;
2709        }
2710        if ($_no_metadata) {
2711                my $offset = -41; # from tail
2712                my $rl;
2713                open my $fh, '<', $REVDB or
2714                        die "--no-metadata specified and $REVDB not readable\n";
2715                seek $fh, $offset, 2;
2716                $rl = readline $fh;
2717                defined $rl or return (undef, undef);
2718                chomp $rl;
2719                while ($c ne $rl && tell $fh != 0) {
2720                        $offset -= 41;
2721                        seek $fh, $offset, 2;
2722                        $rl = readline $fh;
2723                        defined $rl or return (undef, undef);
2724                        chomp $rl;
2725                }
2726                my $rev = tell $fh;
2727                croak $! if ($rev < -1);
2728                $rev =  ($rev - 41) / 41;
2729                close $fh or croak $!;
2730                return ($rev, $c);
2731        }
2732        return (undef, undef);
2733}
2734
2735sub libsvn_parse_revision {
2736        my $base = shift;
2737        my $head = $SVN->get_latest_revnum();
2738        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
2739                return ($base + 1, $head) if (defined $base);
2740                return (0, $head);
2741        }
2742        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
2743        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
2744        if ($_revision =~ /^BASE:(\d+)$/) {
2745                return ($base + 1, $1) if (defined $base);
2746                return (0, $head);
2747        }
2748        return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
2749        die "revision argument: $_revision not understood by git-svn\n",
2750                "Try using the command-line svn client instead\n";
2751}
2752
2753sub libsvn_traverse {
2754        my ($gui, $pfx, $path, $rev, $files) = @_;
2755        my $cwd = "$pfx/$path";
2756        my $pool = SVN::Pool->new;
2757        $cwd =~ s#^/+##g;
2758        my ($dirent, $r, $props) = $SVN->get_dir($cwd, $rev, $pool);
2759        foreach my $d (keys %$dirent) {
2760                my $t = $dirent->{$d}->kind;
2761                if ($t == $SVN::Node::dir) {
2762                        libsvn_traverse($gui, $cwd, $d, $rev, $files);
2763                } elsif ($t == $SVN::Node::file) {
2764                        my $file = "$cwd/$d";
2765                        if (defined $files) {
2766                                push @$files, $file;
2767                        } else {
2768                                print "\tA\t$file\n" unless $_q;
2769                                libsvn_get_file($gui, $file, $rev);
2770                        }
2771                }
2772        }
2773        $pool->clear;
2774}
2775
2776sub libsvn_traverse_ignore {
2777        my ($fh, $path, $r) = @_;
2778        $path =~ s#^/+##g;
2779        my $pool = SVN::Pool->new;
2780        my ($dirent, undef, $props) = $SVN->get_dir($path, $r, $pool);
2781        my $p = $path;
2782        $p =~ s#^\Q$SVN_PATH\E/?##;
2783        print $fh length $p ? "\n# $p\n" : "\n# /\n";
2784        if (my $s = $props->{'svn:ignore'}) {
2785                $s =~ s/[\r\n]+/\n/g;
2786                chomp $s;
2787                if (length $p == 0) {
2788                        $s =~ s#\n#\n/$p#g;
2789                        print $fh "/$s\n";
2790                } else {
2791                        $s =~ s#\n#\n/$p/#g;
2792                        print $fh "/$p/$s\n";
2793                }
2794        }
2795        foreach (sort keys %$dirent) {
2796                next if $dirent->{$_}->kind != $SVN::Node::dir;
2797                libsvn_traverse_ignore($fh, "$path/$_", $r);
2798        }
2799        $pool->clear;
2800}
2801
2802sub revisions_eq {
2803        my ($path, $r0, $r1) = @_;
2804        return 1 if $r0 == $r1;
2805        my $nr = 0;
2806        if ($_use_lib) {
2807                # should be OK to use Pool here (r1 - r0) should be small
2808                my $pool = SVN::Pool->new;
2809                libsvn_get_log($SVN, "/$path", $r0, $r1,
2810                                0, 1, 1, sub {$nr++}, $pool);
2811                $pool->clear;
2812        } else {
2813                my ($url, undef) = repo_path_split($SVN_URL);
2814                my $svn_log = svn_log_raw("$url/$path","-r$r0:$r1");
2815                while (next_log_entry($svn_log)) { $nr++ }
2816                close $svn_log->{fh};
2817        }
2818        return 0 if ($nr > 1);
2819        return 1;
2820}
2821
2822sub libsvn_find_parent_branch {
2823        my ($paths, $rev, $author, $date, $msg) = @_;
2824        my $svn_path = '/'.$SVN_PATH;
2825
2826        # look for a parent from another branch:
2827        my $i = $paths->{$svn_path} or return;
2828        my $branch_from = $i->copyfrom_path or return;
2829        my $r = $i->copyfrom_rev;
2830        print STDERR  "Found possible branch point: ",
2831                                "$branch_from => $svn_path, $r\n";
2832        $branch_from =~ s#^/##;
2833        my $l_map = {};
2834        read_url_paths_all($l_map, '', "$GIT_DIR/svn");
2835        my $url = $SVN->{url};
2836        defined $l_map->{$url} or return;
2837        my $id = $l_map->{$url}->{$branch_from};
2838        if (!defined $id && $_follow_parent) {
2839                print STDERR "Following parent: $branch_from\@$r\n";
2840                # auto create a new branch and follow it
2841                $id = basename($branch_from);
2842                $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
2843                while (-r "$GIT_DIR/svn/$id") {
2844                        # just grow a tail if we're not unique enough :x
2845                        $id .= '-';
2846                }
2847        }
2848        return unless defined $id;
2849
2850        my ($r0, $parent) = find_rev_before($r,$id,1);
2851        if ($_follow_parent && (!defined $r0 || !defined $parent)) {
2852                defined(my $pid = fork) or croak $!;
2853                if (!$pid) {
2854                        $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2855                        init_vars();
2856                        $SVN_URL = "$url/$branch_from";
2857                        $SVN_LOG = $SVN = undef;
2858                        setup_git_svn();
2859                        # we can't assume SVN_URL exists at r+1:
2860                        $_revision = "0:$r";
2861                        fetch_lib();
2862                        exit 0;
2863                }
2864                waitpid $pid, 0;
2865                croak $? if $?;
2866                ($r0, $parent) = find_rev_before($r,$id,1);
2867        }
2868        return unless (defined $r0 && defined $parent);
2869        if (revisions_eq($branch_from, $r0, $r)) {
2870                unlink $GIT_SVN_INDEX;
2871                print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
2872                sys(qw/git-read-tree/, $parent);
2873                return libsvn_fetch($parent, $paths, $rev,
2874                                        $author, $date, $msg);
2875        }
2876        print STDERR "Nope, branch point not imported or unknown\n";
2877        return undef;
2878}
2879
2880sub libsvn_get_log {
2881        my ($ra, @args) = @_;
2882        if ($SVN::Core::VERSION le '1.2.0') {
2883                splice(@args, 3, 1);
2884        }
2885        $ra->get_log(@args);
2886}
2887
2888sub libsvn_new_tree {
2889        if (my $log_entry = libsvn_find_parent_branch(@_)) {
2890                return $log_entry;
2891        }
2892        my ($paths, $rev, $author, $date, $msg) = @_;
2893        open my $gui, '| git-update-index -z --index-info' or croak $!;
2894        libsvn_traverse($gui, '', $SVN_PATH, $rev);
2895        close $gui or croak $?;
2896        return libsvn_log_entry($rev, $author, $date, $msg);
2897}
2898
2899sub find_graft_path_commit {
2900        my ($tree_paths, $p1, $r1) = @_;
2901        foreach my $x (keys %$tree_paths) {
2902                next unless ($p1 =~ /^\Q$x\E/);
2903                my $i = $tree_paths->{$x};
2904                my ($r0, $parent) = find_rev_before($r1,$i,1);
2905                return $parent if (defined $r0 && $r0 == $r1);
2906                print STDERR "r$r1 of $i not imported\n";
2907                next;
2908        }
2909        return undef;
2910}
2911
2912sub find_graft_path_parents {
2913        my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
2914        foreach my $x (keys %$tree_paths) {
2915                next unless ($p0 =~ /^\Q$x\E/);
2916                my $i = $tree_paths->{$x};
2917                my ($r, $parent) = find_rev_before($r0, $i, 1);
2918                if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
2919                        my ($url_b, undef, $uuid_b) = cmt_metadata($c);
2920                        my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
2921                        next if ($url_a && $url_b && $url_a eq $url_b &&
2922                                                        $uuid_b eq $uuid_a);
2923                        $grafts->{$c}->{$parent} = 1;
2924                }
2925        }
2926}
2927
2928sub libsvn_graft_file_copies {
2929        my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
2930        foreach (keys %$paths) {
2931                my $i = $paths->{$_};
2932                my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
2933                                        $i->copyfrom_rev);
2934                next unless (defined $p0 && defined $r0);
2935
2936                my $p1 = $_;
2937                $p1 =~ s#^/##;
2938                $p0 =~ s#^/##;
2939                my $c = find_graft_path_commit($tree_paths, $p1, $rev);
2940                next unless $c;
2941                find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
2942        }
2943}
2944
2945sub set_index {
2946        my $old = $ENV{GIT_INDEX_FILE};
2947        $ENV{GIT_INDEX_FILE} = shift;
2948        return $old;
2949}
2950
2951sub restore_index {
2952        my ($old) = @_;
2953        if (defined $old) {
2954                $ENV{GIT_INDEX_FILE} = $old;
2955        } else {
2956                delete $ENV{GIT_INDEX_FILE};
2957        }
2958}
2959
2960sub libsvn_commit_cb {
2961        my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
2962        if ($_optimize_commits && $rev == ($r_last + 1)) {
2963                my $log = libsvn_log_entry($rev,$committer,$date,$msg);
2964                $log->{tree} = get_tree_from_treeish($c);
2965                my $cmt = git_commit($log, $cmt_last, $c);
2966                my @diff = safe_qx('git-diff-tree', $cmt, $c);
2967                if (@diff) {
2968                        print STDERR "Trees differ: $cmt $c\n",
2969                                        join('',@diff),"\n";
2970                        exit 1;
2971                }
2972        } else {
2973                fetch("$rev=$c");
2974        }
2975}
2976
2977sub libsvn_ls_fullurl {
2978        my $fullurl = shift;
2979        my ($repo, $path) = repo_path_split($fullurl);
2980        $SVN ||= libsvn_connect($repo);
2981        my @ret;
2982        my $pool = SVN::Pool->new;
2983        my ($dirent, undef, undef) = $SVN->get_dir($path,
2984                                                $SVN->get_latest_revnum, $pool);
2985        foreach my $d (keys %$dirent) {
2986                if ($dirent->{$d}->kind == $SVN::Node::dir) {
2987                        push @ret, "$d/"; # add '/' for compat with cli svn
2988                }
2989        }
2990        $pool->clear;
2991        return @ret;
2992}
2993
2994
2995sub libsvn_skip_unknown_revs {
2996        my $err = shift;
2997        my $errno = $err->apr_err();
2998        # Maybe the branch we're tracking didn't
2999        # exist when the repo started, so it's
3000        # not an error if it doesn't, just continue
3001        #
3002        # Wonderfully consistent library, eh?
3003        # 160013 - svn:// and file://
3004        # 175002 - http(s)://
3005        #   More codes may be discovered later...
3006        if ($errno == 175002 || $errno == 160013) {
3007                return;
3008        }
3009        croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3010};
3011
3012# Tie::File seems to be prone to offset errors if revisions get sparse,
3013# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3014# one of my favorite modules is out :<  Next up would be one of the DBM
3015# modules, but I'm not sure which is most portable...  So I'll just
3016# go with something that's plain-text, but still capable of
3017# being randomly accessed.  So here's my ultra-simple fixed-width
3018# database.  All records are 40 characters + "\n", so it's easy to seek
3019# to a revision: (41 * rev) is the byte offset.
3020# A record of 40 0s denotes an empty revision.
3021# And yes, it's still pretty fast (faster than Tie::File).
3022sub revdb_set {
3023        my ($file, $rev, $commit) = @_;
3024        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
3025        open my $fh, '+<', $file or croak $!;
3026        my $offset = $rev * 41;
3027        # assume that append is the common case:
3028        seek $fh, 0, 2 or croak $!;
3029        my $pos = tell $fh;
3030        if ($pos < $offset) {
3031                print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
3032        }
3033        seek $fh, $offset, 0 or croak $!;
3034        print $fh $commit,"\n";
3035        close $fh or croak $!;
3036}
3037
3038sub revdb_get {
3039        my ($file, $rev) = @_;
3040        my $ret;
3041        my $offset = $rev * 41;
3042        open my $fh, '<', $file or croak $!;
3043        seek $fh, $offset, 0;
3044        if (tell $fh == $offset) {
3045                $ret = readline $fh;
3046                if (defined $ret) {
3047                        chomp $ret;
3048                        $ret = undef if ($ret =~ /^0{40}$/);
3049                }
3050        }
3051        close $fh or croak $!;
3052        return $ret;
3053}
3054
3055sub copy_remote_ref {
3056        my $origin = $_cp_remote ? $_cp_remote : 'origin';
3057        my $ref = "refs/remotes/$GIT_SVN";
3058        if (safe_qx('git-ls-remote', $origin, $ref)) {
3059                sys(qw/git fetch/, $origin, "$ref:$ref");
3060        } else {
3061                die "Unable to find remote reference: ",
3062                                "refs/remotes/$GIT_SVN on $origin\n";
3063        }
3064}
3065
3066package SVN::Git::Editor;
3067use vars qw/@ISA/;
3068use strict;
3069use warnings;
3070use Carp qw/croak/;
3071use IO::File;
3072
3073sub new {
3074        my $class = shift;
3075        my $git_svn = shift;
3076        my $self = SVN::Delta::Editor->new(@_);
3077        bless $self, $class;
3078        foreach (qw/svn_path c r ra /) {
3079                die "$_ required!\n" unless (defined $git_svn->{$_});
3080                $self->{$_} = $git_svn->{$_};
3081        }
3082        $self->{pool} = SVN::Pool->new;
3083        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3084        $self->{rm} = { };
3085        require Digest::MD5;
3086        return $self;
3087}
3088
3089sub split_path {
3090        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3091}
3092
3093sub repo_path {
3094        (defined $_[1] && length $_[1]) ? "$_[0]->{svn_path}/$_[1]"
3095                                        : $_[0]->{svn_path}
3096}
3097
3098sub url_path {
3099        my ($self, $path) = @_;
3100        $self->{ra}->{url} . '/' . $self->repo_path($path);
3101}
3102
3103sub rmdirs {
3104        my ($self, $q) = @_;
3105        my $rm = $self->{rm};
3106        delete $rm->{''}; # we never delete the url we're tracking
3107        return unless %$rm;
3108
3109        foreach (keys %$rm) {
3110                my @d = split m#/#, $_;
3111                my $c = shift @d;
3112                $rm->{$c} = 1;
3113                while (@d) {
3114                        $c .= '/' . shift @d;
3115                        $rm->{$c} = 1;
3116                }
3117        }
3118        delete $rm->{$self->{svn_path}};
3119        delete $rm->{''}; # we never delete the url we're tracking
3120        return unless %$rm;
3121
3122        defined(my $pid = open my $fh,'-|') or croak $!;
3123        if (!$pid) {
3124                exec qw/git-ls-tree --name-only -r -z/, $self->{c} or croak $!;
3125        }
3126        local $/ = "\0";
3127        my @svn_path = split m#/#, $self->{svn_path};
3128        while (<$fh>) {
3129                chomp;
3130                my @dn = (@svn_path, (split m#/#, $_));
3131                while (pop @dn) {
3132                        delete $rm->{join '/', @dn};
3133                }
3134                unless (%$rm) {
3135                        close $fh;
3136                        return;
3137                }
3138        }
3139        close $fh;
3140
3141        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3142        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3143                $self->close_directory($bat->{$d}, $p);
3144                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3145                print "\tD+\t/$d/\n" unless $q;
3146                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3147                delete $bat->{$d};
3148        }
3149}
3150
3151sub open_or_add_dir {
3152        my ($self, $full_path, $baton) = @_;
3153        my $p = SVN::Pool->new;
3154        my $t = $self->{ra}->check_path($full_path, $self->{r}, $p);
3155        $p->clear;
3156        if ($t == $SVN::Node::none) {
3157                return $self->add_directory($full_path, $baton,
3158                                                undef, -1, $self->{pool});
3159        } elsif ($t == $SVN::Node::dir) {
3160                return $self->open_directory($full_path, $baton,
3161                                                $self->{r}, $self->{pool});
3162        }
3163        print STDERR "$full_path already exists in repository at ",
3164                "r$self->{r} and it is not a directory (",
3165                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3166        exit 1;
3167}
3168
3169sub ensure_path {
3170        my ($self, $path) = @_;
3171        my $bat = $self->{bat};
3172        $path = $self->repo_path($path);
3173        return $bat->{''} unless (length $path);
3174        my @p = split m#/+#, $path;
3175        my $c = shift @p;
3176        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3177        while (@p) {
3178                my $c0 = $c;
3179                $c .= '/' . shift @p;
3180                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3181        }
3182        return $bat->{$c};
3183}
3184
3185sub A {
3186        my ($self, $m, $q) = @_;
3187        my ($dir, $file) = split_path($m->{file_b});
3188        my $pbat = $self->ensure_path($dir);
3189        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3190                                        undef, -1);
3191        print "\tA\t$m->{file_b}\n" unless $q;
3192        $self->chg_file($fbat, $m);
3193        $self->close_file($fbat,undef,$self->{pool});
3194}
3195
3196sub C {
3197        my ($self, $m, $q) = @_;
3198        my ($dir, $file) = split_path($m->{file_b});
3199        my $pbat = $self->ensure_path($dir);
3200        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3201                                $self->url_path($m->{file_a}), $self->{r});
3202        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $q;
3203        $self->chg_file($fbat, $m);
3204        $self->close_file($fbat,undef,$self->{pool});
3205}
3206
3207sub delete_entry {
3208        my ($self, $path, $pbat) = @_;
3209        my $rpath = $self->repo_path($path);
3210        my ($dir, $file) = split_path($rpath);
3211        $self->{rm}->{$dir} = 1;
3212        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3213}
3214
3215sub R {
3216        my ($self, $m, $q) = @_;
3217        my ($dir, $file) = split_path($m->{file_b});
3218        my $pbat = $self->ensure_path($dir);
3219        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3220                                $self->url_path($m->{file_a}), $self->{r});
3221        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $q;
3222        $self->chg_file($fbat, $m);
3223        $self->close_file($fbat,undef,$self->{pool});
3224
3225        ($dir, $file) = split_path($m->{file_a});
3226        $pbat = $self->ensure_path($dir);
3227        $self->delete_entry($m->{file_a}, $pbat);
3228}
3229
3230sub M {
3231        my ($self, $m, $q) = @_;
3232        my ($dir, $file) = split_path($m->{file_b});
3233        my $pbat = $self->ensure_path($dir);
3234        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3235                                $pbat,$self->{r},$self->{pool});
3236        print "\t$m->{chg}\t$m->{file_b}\n" unless $q;
3237        $self->chg_file($fbat, $m);
3238        $self->close_file($fbat,undef,$self->{pool});
3239}
3240
3241sub T { shift->M(@_) }
3242
3243sub change_file_prop {
3244        my ($self, $fbat, $pname, $pval) = @_;
3245        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3246}
3247
3248sub chg_file {
3249        my ($self, $fbat, $m) = @_;
3250        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3251                $self->change_file_prop($fbat,'svn:executable','*');
3252        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3253                $self->change_file_prop($fbat,'svn:executable',undef);
3254        }
3255        my $fh = IO::File->new_tmpfile or croak $!;
3256        if ($m->{mode_b} =~ /^120/) {
3257                print $fh 'link ' or croak $!;
3258                $self->change_file_prop($fbat,'svn:special','*');
3259        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3260                $self->change_file_prop($fbat,'svn:special',undef);
3261        }
3262        defined(my $pid = fork) or croak $!;
3263        if (!$pid) {
3264                open STDOUT, '>&', $fh or croak $!;
3265                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3266        }
3267        waitpid $pid, 0;
3268        croak $? if $?;
3269        $fh->flush == 0 or croak $!;
3270        seek $fh, 0, 0 or croak $!;
3271
3272        my $md5 = Digest::MD5->new;
3273        $md5->addfile($fh) or croak $!;
3274        seek $fh, 0, 0 or croak $!;
3275
3276        my $exp = $md5->hexdigest;
3277        my $atd = $self->apply_textdelta($fbat, undef, $self->{pool});
3278        my $got = SVN::TxDelta::send_stream($fh, @$atd, $self->{pool});
3279        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3280
3281        close $fh or croak $!;
3282}
3283
3284sub D {
3285        my ($self, $m, $q) = @_;
3286        my ($dir, $file) = split_path($m->{file_b});
3287        my $pbat = $self->ensure_path($dir);
3288        print "\tD\t$m->{file_b}\n" unless $q;
3289        $self->delete_entry($m->{file_b}, $pbat);
3290}
3291
3292sub close_edit {
3293        my ($self) = @_;
3294        my ($p,$bat) = ($self->{pool}, $self->{bat});
3295        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3296                $self->close_directory($bat->{$_}, $p);
3297        }
3298        $self->SUPER::close_edit($p);
3299        $p->clear;
3300}
3301
3302sub abort_edit {
3303        my ($self) = @_;
3304        $self->SUPER::abort_edit($self->{pool});
3305        $self->{pool}->clear;
3306}
3307
3308__END__
3309
3310Data structures:
3311
3312$svn_log hashref (as returned by svn_log_raw)
3313{
3314        fh => file handle of the log file,
3315        state => state of the log file parser (sep/msg/rev/msg_start...)
3316}
3317
3318$log_msg hashref as returned by next_log_entry($svn_log)
3319{
3320        msg => 'whitespace-formatted log entry
3321',                                              # trailing newline is preserved
3322        revision => '8',                        # integer
3323        date => '2004-02-24T17:01:44.108345Z',  # commit date
3324        author => 'committer name'
3325};
3326
3327
3328@mods = array of diff-index line hashes, each element represents one line
3329        of diff-index output
3330
3331diff-index line ($m hash)
3332{
3333        mode_a => first column of diff-index output, no leading ':',
3334        mode_b => second column of diff-index output,
3335        sha1_b => sha1sum of the final blob,
3336        chg => change type [MCRADT],
3337        file_a => original file name of a file (iff chg is 'C' or 'R')
3338        file_b => new/current file name of a file (any chg)
3339}
3340;
3341
3342# retval of read_url_paths{,_all}();
3343$l_map = {
3344        # repository root url
3345        'https://svn.musicpd.org' => {
3346                # repository path               # GIT_SVN_ID
3347                'mpd/trunk'             =>      'trunk',
3348                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
3349        },
3350}
3351
3352Notes:
3353        I don't trust the each() function on unless I created %hash myself
3354        because the internal iterator may not have started at base.