contrib/git-svn/git-svn.perl

   1 #!/usr/bin/env perl
   2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3 # License: GPL v2 or later
   4 use warnings;
   5 use strict;
   6 use vars qw/    $AUTHOR $VERSION
   7                 $SVN_URL $SVN_INFO $SVN_WC $SVN_UUID
   8                 $GIT_SVN_INDEX $GIT_SVN
   9                 $GIT_DIR $REV_DIR $GIT_SVN_DIR/;
  10 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  11 $VERSION = '1.1.0-pre';
  12
  13 use Cwd qw/abs_path/;
  14 $GIT_DIR = abs_path($ENV{GIT_DIR} || '.git');
  15 $ENV{GIT_DIR} = $GIT_DIR;
  16
  17 my $LC_ALL = $ENV{LC_ALL};
  18 # make sure the svn binary gives consistent output between locales and TZs:
  19 $ENV{TZ} = 'UTC';
  20 $ENV{LC_ALL} = 'C';
  21
  22 # If SVN:: library support is added, please make the dependencies
  23 # optional and preserve the capability to use the command-line client.
  24 # use eval { require SVN::... } to make it lazy load
  25 # We don't use any modules not in the standard Perl distribution:
  26 use Carp qw/croak/;
  27 use IO::File qw//;
  28 use File::Basename qw/dirname basename/;
  29 use File::Path qw/mkpath/;
  30 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  31 use File::Spec qw//;
  32 use POSIX qw/strftime/;
  33 my $sha1 = qr/[a-f\d]{40}/;
  34 my $sha1_short = qr/[a-f\d]{4,40}/;
  35 my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
  36         $_find_copies_harder, $_l, $_cp_similarity,
  37         $_repack, $_repack_nr, $_repack_flags,
  38         $_template, $_shared, $_no_default_regex, $_no_graft_copy,
  39         $_version, $_upgrade, $_authors, $_branch_all_refs, @_opt_m);
  40 my (@_branch_from, %tree_map, %users);
  41 my ($_svn_co_url_revs, $_svn_pg_peg_revs);
  42 my @repo_path_split_cache;
  43
  44 my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
  45                 'branch|b=s' => \@_branch_from,
  46                 'branch-all-refs|B' => \$_branch_all_refs,
  47                 'authors-file|A=s' => \$_authors,
  48                 'repack:i' => \$_repack,
  49                 'repack-flags|repack-args|repack-opts=s' => \$_repack_flags);
  50
  51 my ($_trunk, $_tags, $_branches);
  52 my %multi_opts = ( 'trunk|T=s' => \$_trunk,
  53                 'tags|t=s' => \$_tags,
  54                 'branches|b=s' => \$_branches );
  55 my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
  56
  57 # yes, 'native' sets "\n".  Patches to fix this for non-*nix systems welcome:
  58 my %EOL = ( CR => "\015", LF => "\012", CRLF => "\015\012", native => "\012" );
  59
  60 my %cmd = (
  61         fetch => [ \&fetch, "Download new revisions from SVN",
  62                         { 'revision|r=s' => \$_revision, %fc_opts } ],
  63         init => [ \&init, "Initialize a repo for tracking" .
  64                           " (requires URL argument)",
  65                           \%init_opts ],
  66         commit => [ \&commit, "Commit git revisions to SVN",
  67                         {       'stdin|' => \$_stdin,
  68                                 'edit|e' => \$_edit,
  69                                 'rmdir' => \$_rmdir,
  70                                 'find-copies-harder' => \$_find_copies_harder,
  71                                 'l=i' => \$_l,
  72                                 'copy-similarity|C=i'=> \$_cp_similarity,
  73                                 %fc_opts,
  74                         } ],
  75         'show-ignore' => [ \&show_ignore, "Show svn:ignore listings", { } ],
  76         rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
  77                         { 'no-ignore-externals' => \$_no_ignore_ext,
  78                           'upgrade' => \$_upgrade } ],
  79         'graft-branches' => [ \&graft_branches,
  80                         'Detect merges/branches from already imported history',
  81                         { 'merge-rx|m' => \@_opt_m,
  82                           'no-default-regex' => \$_no_default_regex,
  83                           'no-graft-copy' => \$_no_graft_copy } ],
  84         'multi-init' => [ \&multi_init,
  85                         'Initialize multiple trees (like git-svnimport)',
  86                         { %multi_opts, %fc_opts } ],
  87         'multi-fetch' => [ \&multi_fetch,
  88                         'Fetch multiple trees (like git-svnimport)',
  89                         \%fc_opts ],
  90 );
  91
  92 my $cmd;
  93 for (my $i = 0; $i < @ARGV; $i++) {
  94         if (defined $cmd{$ARGV[$i]}) {
  95                 $cmd = $ARGV[$i];
  96                 splice @ARGV, $i, 1;
  97                 last;
  98         }
  99 };
 100
 101 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 102
 103 read_repo_config(\%opts);
 104 GetOptions(%opts, 'help|H|h' => \$_help,
 105                 'version|V' => \$_version,
 106                 'id|i=s' => \$GIT_SVN) or exit 1;
 107
 108 set_default_vals();
 109 usage(0) if $_help;
 110 version() if $_version;
 111 usage(1) unless defined $cmd;
 112 init_vars();
 113 load_authors() if $_authors;
 114 load_all_refs() if $_branch_all_refs;
 115 svn_compat_check();
 116 migration_check() unless $cmd =~ /^(?:init|multi-init)$/;
 117 $cmd{$cmd}->[0]->(@ARGV);
 118 exit 0;
 119
 120 ####################### primary functions ######################
 121 sub usage {
 122         my $exit = shift || 0;
 123         my $fd = $exit ? \*STDERR : \*STDOUT;
 124         print $fd <<"";
 125 git-svn - bidirectional operations between a single Subversion tree and git
 126 Usage: $0 <command> [options] [arguments]\n
 127
 128         print $fd "Available commands:\n" unless $cmd;
 129
 130         foreach (sort keys %cmd) {
 131                 next if $cmd && $cmd ne $_;
 132                 print $fd '  ',pack('A13',$_),$cmd{$_}->[1],"\n";
 133                 foreach (keys %{$cmd{$_}->[2]}) {
 134                         # prints out arguments as they should be passed:
 135                         my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 136                         print $fd ' ' x 17, join(', ', map { length $_ > 1 ?
 137                                                         "--$_" : "-$_" }
 138                                                 split /\|/,$_)," $x\n";
 139                 }
 140         }
 141         print $fd <<"";
 142 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 143 arbitrary identifier if you're tracking multiple SVN branches/repositories in
 144 one git repository and want to keep them separate.  See git-svn(1) for more
 145 information.
 146
 147         exit $exit;
 148 }
 149
 150 sub version {
 151         print "git-svn version $VERSION\n";
 152         exit 0;
 153 }
 154
 155 sub rebuild {
 156         $SVN_URL = shift or undef;
 157         my $newest_rev = 0;
 158         if ($_upgrade) {
 159                 sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD");
 160         } else {
 161                 check_upgrade_needed();
 162         }
 163
 164         my $pid = open(my $rev_list,'-|');
 165         defined $pid or croak $!;
 166         if ($pid == 0) {
 167                 exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!;
 168         }
 169         my $latest;
 170         while (<$rev_list>) {
 171                 chomp;
 172                 my $c = $_;
 173                 croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 174                 my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
 175                 next if (!@commit); # skip merges
 176                 my $id = $commit[$#commit];
 177                 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 178                                                 \s([a-f\d\-]+)$/x);
 179                 if (!$rev || !$uuid || !$url) {
 180                         # some of the original repositories I made had
 181                         # indentifiers like this:
 182                         ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)
 183                                                         \@([a-f\d\-]+)/x);
 184                         if (!$rev || !$uuid) {
 185                                 croak "Unable to extract revision or UUID from ",
 186                                         "$c, $id\n";
 187                         }
 188                 }
 189
 190                 # if we merged or otherwise started elsewhere, this is
 191                 # how we break out of it
 192                 next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
 193                 next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
 194
 195                 print "r$rev = $c\n";
 196                 unless (defined $latest) {
 197                         if (!$SVN_URL && !$url) {
 198                                 croak "SVN repository location required: $url\n";
 199                         }
 200                         $SVN_URL ||= $url;
 201                         $SVN_UUID ||= $uuid;
 202                         setup_git_svn();
 203                         $latest = $rev;
 204                 }
 205                 assert_revision_eq_or_unknown($rev, $c);
 206                 sys('git-update-ref',"svn/$GIT_SVN/revs/$rev",$c);
 207                 $newest_rev = $rev if ($rev > $newest_rev);
 208         }
 209         close $rev_list or croak $?;
 210         if (!chdir $SVN_WC) {
 211                 svn_cmd_checkout($SVN_URL, $latest, $SVN_WC);
 212                 chdir $SVN_WC or croak $!;
 213         }
 214
 215         $pid = fork;
 216         defined $pid or croak $!;
 217         if ($pid == 0) {
 218                 my @svn_up = qw(svn up);
 219                 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 220                 sys(@svn_up,"-r$newest_rev");
 221                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 222                 index_changes();
 223                 exec('git-write-tree') or croak $!;
 224         }
 225         waitpid $pid, 0;
 226         croak $? if $?;
 227
 228         if ($_upgrade) {
 229                 print STDERR <<"";
 230 Keeping deprecated refs/head/$GIT_SVN-HEAD for now.  Please remove it
 231 when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN
 232
 233         }
 234 }
 235
 236 sub init {
 237         $SVN_URL = shift or die "SVN repository location required " .
 238                                 "as a command-line argument\n";
 239         $SVN_URL =~ s!/+$!!; # strip trailing slash
 240         unless (-d $GIT_DIR) {
 241                 my @init_db = ('git-init-db');
 242                 push @init_db, "--template=$_template" if defined $_template;
 243                 push @init_db, "--shared" if defined $_shared;
 244                 sys(@init_db);
 245         }
 246         setup_git_svn();
 247 }
 248
 249 sub fetch {
 250         my (@parents) = @_;
 251         check_upgrade_needed();
 252         $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 253         my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 254         unless ($_revision) {
 255                 $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 256         }
 257         push @log_args, "-r$_revision";
 258         push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 259
 260         my $svn_log = svn_log_raw(@log_args);
 261
 262         my $base = next_log_entry($svn_log) or croak "No base revision!\n";
 263         my $last_commit = undef;
 264         unless (-d $SVN_WC) {
 265                 svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC);
 266                 chdir $SVN_WC or croak $!;
 267                 read_uuid();
 268                 $last_commit = git_commit($base, @parents);
 269                 assert_tree($last_commit);
 270         } else {
 271                 chdir $SVN_WC or croak $!;
 272                 read_uuid();
 273                 eval { $last_commit = file_to_s("$REV_DIR/$base->{revision}") };
 274                 # looks like a user manually cp'd and svn switch'ed
 275                 unless ($last_commit) {
 276                         sys(qw/svn revert -R ./);
 277                         assert_svn_wc_clean($base->{revision});
 278                         $last_commit = git_commit($base, @parents);
 279                         assert_tree($last_commit);
 280                 }
 281         }
 282         my @svn_up = qw(svn up);
 283         push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 284         my $last = $base;
 285         while (my $log_msg = next_log_entry($svn_log)) {
 286                 assert_tree($last_commit);
 287                 if ($last->{revision} >= $log_msg->{revision}) {
 288                         croak "Out of order: last >= current: ",
 289                                 "$last->{revision} >= $log_msg->{revision}\n";
 290                 }
 291                 # Revert is needed for cases like:
 292                 # https://svn.musicpd.org/Jamming/trunk (r166:167), but
 293                 # I can't seem to reproduce something like that on a test...
 294                 sys(qw/svn revert -R ./);
 295                 assert_svn_wc_clean($last->{revision});
 296                 sys(@svn_up,"-r$log_msg->{revision}");
 297                 $last_commit = git_commit($log_msg, $last_commit, @parents);
 298                 $last = $log_msg;
 299         }
 300         unless (-e "$GIT_DIR/refs/heads/master") {
 301                 sys(qw(git-update-ref refs/heads/master),$last_commit);
 302         }
 303         close $svn_log->{fh};
 304         return $last;
 305 }
 306
 307 sub commit {
 308         my (@commits) = @_;
 309         check_upgrade_needed();
 310         if ($_stdin || !@commits) {
 311                 print "Reading from stdin...\n";
 312                 @commits = ();
 313                 while (<STDIN>) {
 314                         if (/\b($sha1_short)\b/o) {
 315                                 unshift @commits, $1;
 316                         }
 317                 }
 318         }
 319         my @revs;
 320         foreach my $c (@commits) {
 321                 chomp(my @tmp = safe_qx('git-rev-parse',$c));
 322                 if (scalar @tmp == 1) {
 323                         push @revs, $tmp[0];
 324                 } elsif (scalar @tmp > 1) {
 325                         push @revs, reverse (safe_qx('git-rev-list',@tmp));
 326                 } else {
 327                         die "Failed to rev-parse $c\n";
 328                 }
 329         }
 330         chomp @revs;
 331
 332         chdir $SVN_WC or croak "Unable to chdir $SVN_WC: $!\n";
 333         my $info = svn_info('.');
 334         my $fetched = fetch();
 335         if ($info->{Revision} != $fetched->{revision}) {
 336                 print STDERR "There are new revisions that were fetched ",
 337                                 "and need to be merged (or acknowledged) ",
 338                                 "before committing.\n";
 339                 exit 1;
 340         }
 341         $info = svn_info('.');
 342         read_uuid($info);
 343         my $svn_current_rev =  $info->{'Last Changed Rev'};
 344         foreach my $c (@revs) {
 345                 my $mods = svn_checkout_tree($svn_current_rev, $c);
 346                 if (scalar @$mods == 0) {
 347                         print "Skipping, no changes detected\n";
 348                         next;
 349                 }
 350                 $svn_current_rev = svn_commit_tree($svn_current_rev, $c);
 351         }
 352         print "Done committing ",scalar @revs," revisions to SVN\n";
 353 }
 354
 355 sub show_ignore {
 356         require File::Find or die $!;
 357         my $exclude_file = "$GIT_DIR/info/exclude";
 358         open my $fh, '<', $exclude_file or croak $!;
 359         chomp(my @excludes = (<$fh>));
 360         close $fh or croak $!;
 361
 362         $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
 363         chdir $SVN_WC or croak $!;
 364         my %ign;
 365         File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 366                 s#^\./##;
 367                 @{$ign{$_}} = svn_propget_base('svn:ignore', $_);
 368                 }}, no_chdir=>1},'.');
 369
 370         print "\n# /\n";
 371         foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 372         delete $ign{'.'};
 373         foreach my $i (sort keys %ign) {
 374                 print "\n# ",$i,"\n";
 375                 foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 376         }
 377 }
 378
 379 sub graft_branches {
 380         my $gr_file = "$GIT_DIR/info/grafts";
 381         my ($grafts, $comments) = read_grafts($gr_file);
 382         my $gr_sha1;
 383
 384         if (%$grafts) {
 385                 # temporarily disable our grafts file to make this idempotent
 386                 chomp($gr_sha1 = safe_qx(qw/git-hash-object -w/,$gr_file));
 387                 rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
 388         }
 389
 390         my $l_map = read_url_paths();
 391         my @re = map { qr/$_/is } @_opt_m if @_opt_m;
 392         unless ($_no_default_regex) {
 393                 push @re, (     qr/\b(?:merge|merging|merged)\s+(\S.+)/is,
 394                                 qr/\b(?:from|of)\s+(\S.+)/is );
 395         }
 396         foreach my $u (keys %$l_map) {
 397                 if (@re) {
 398                         foreach my $p (keys %{$l_map->{$u}}) {
 399                                 graft_merge_msg($grafts,$l_map,$u,$p);
 400                         }
 401                 }
 402                 graft_file_copy($grafts,$l_map,$u) unless $_no_graft_copy;
 403         }
 404
 405         write_grafts($grafts, $comments, $gr_file);
 406         unlink "$gr_file~$gr_sha1" if $gr_sha1;
 407 }
 408
 409 sub multi_init {
 410         my $url = shift;
 411         $_trunk ||= 'trunk';
 412         $_trunk =~ s#/+$##;
 413         $url =~ s#/+$## if $url;
 414         if ($_trunk !~ m#^[a-z\+]+://#) {
 415                 $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#);
 416                 unless ($url) {
 417                         print STDERR "E: '$_trunk' is not a complete URL ",
 418                                 "and a separate URL is not specified\n";
 419                         exit 1;
 420                 }
 421                 $_trunk = $url . $_trunk;
 422         }
 423         if ($GIT_SVN eq 'git-svn') {
 424                 print "GIT_SVN_ID set to 'trunk' for $_trunk\n";
 425                 $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
 426         }
 427         init_vars();
 428         init($_trunk);
 429         complete_url_ls_init($url, $_branches, '--branches/-b', '');
 430         complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/');
 431 }
 432
 433 sub multi_fetch {
 434         # try to do trunk first, since branches/tags
 435         # may be descended from it.
 436         if (-d "$GIT_DIR/svn/trunk") {
 437                 print "Fetching trunk\n";
 438                 defined(my $pid = fork) or croak $!;
 439                 if (!$pid) {
 440                         $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
 441                         init_vars();
 442                         fetch(@_);
 443                         exit 0;
 444                 }
 445                 waitpid $pid, 0;
 446                 croak $? if $?;
 447         }
 448         rec_fetch('', "$GIT_DIR/svn", @_);
 449 }
 450
 451 ########################### utility functions #########################
 452
 453 sub rec_fetch {
 454         my ($pfx, $p, @args) = @_;
 455         my @dir;
 456         foreach (sort <$p/*>) {
 457                 if (-r "$_/info/url") {
 458                         $pfx .= '/' if $pfx && $pfx !~ m!/$!;
 459                         my $id = $pfx . basename $_;
 460                         next if $id eq 'trunk';
 461                         print "Fetching $id\n";
 462                         defined(my $pid = fork) or croak $!;
 463                         if (!$pid) {
 464                                 $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 465                                 init_vars();
 466                                 fetch(@args);
 467                                 exit 0;
 468                         }
 469                         waitpid $pid, 0;
 470                         croak $? if $?;
 471                 } elsif (-d $_) {
 472                         push @dir, $_;
 473                 }
 474         }
 475         foreach (@dir) {
 476                 my $x = $_;
 477                 $x =~ s!^\Q$GIT_DIR\E/svn/!!;
 478                 rec_fetch($x, $_);
 479         }
 480 }
 481
 482 sub complete_url_ls_init {
 483         my ($url, $var, $switch, $pfx) = @_;
 484         unless ($var) {
 485                 print STDERR "W: $switch not specified\n";
 486                 return;
 487         }
 488         $var =~ s#/+$##;
 489         if ($var !~ m#^[a-z\+]+://#) {
 490                 $var = '/' . $var if ($var !~ m#^/#);
 491                 unless ($url) {
 492                         print STDERR "E: '$var' is not a complete URL ",
 493                                 "and a separate URL is not specified\n";
 494                         exit 1;
 495                 }
 496                 $var = $url . $var;
 497         }
 498         chomp(my @ls = safe_qx(qw/svn ls --non-interactive/, $var));
 499         my $old = $GIT_SVN;
 500         defined(my $pid = fork) or croak $!;
 501         if (!$pid) {
 502                 foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
 503                         $u =~ s#/+$##;
 504                         if ($u !~ m!\Q$var\E/(.+)$!) {
 505                                 print STDERR "W: Unrecognized URL: $u\n";
 506                                 die "This should never happen\n";
 507                         }
 508                         my $id = $pfx.$1;
 509                         print "init $u => $id\n";
 510                         $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
 511                         init_vars();
 512                         init($u);
 513                 }
 514                 exit 0;
 515         }
 516         waitpid $pid, 0;
 517         croak $? if $?;
 518 }
 519
 520 sub common_prefix {
 521         my $paths = shift;
 522         my %common;
 523         foreach (@$paths) {
 524                 my @tmp = split m#/#, $_;
 525                 my $p = '';
 526                 while (my $x = shift @tmp) {
 527                         $p .= "/$x";
 528                         $common{$p} ||= 0;
 529                         $common{$p}++;
 530                 }
 531         }
 532         foreach (sort {length $b <=> length $a} keys %common) {
 533                 if ($common{$_} == @$paths) {
 534                         return $_;
 535                 }
 536         }
 537         return '';
 538 }
 539
 540 # this isn't funky-filename safe, but good enough for now...
 541 sub graft_file_copy {
 542         my ($grafts, $l_map, $u) = @_;
 543         my $paths = $l_map->{$u};
 544         my $pfx = common_prefix([keys %$paths]);
 545
 546         my $pid = open my $fh, '-|';
 547         defined $pid or croak $!;
 548         unless ($pid) {
 549                 exec(qw/svn log -v/, $u.$pfx) or croak $!;
 550         }
 551         my ($r, $mp) = (undef, undef);
 552         while (<$fh>) {
 553                 chomp;
 554                 if (/^\-{72}$/) {
 555                         $mp = $r = undef;
 556                 } elsif (/^r(\d+) \| /) {
 557                         $r = $1 unless defined $r;
 558                 } elsif (/^Changed paths:/) {
 559                         $mp = 1;
 560                 } elsif ($mp && m#^   [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) {
 561                         my $dbg = "r$r | $_";
 562                         my ($p1, $p0, $r0) = ($1, $2, $3);
 563                         my $c;
 564                         foreach my $x (keys %$paths) {
 565                                 next unless ($p1 =~ /^\Q$x\E/);
 566                                 my $i = $paths->{$x};
 567                                 my $f = "$GIT_DIR/svn/$i/revs/$r";
 568                                 unless (-r $f) {
 569                                         print STDERR "r$r of $i not imported,",
 570                                                                 " $dbg\n";
 571                                         next;
 572                                 }
 573                                 $c = file_to_s($f);
 574                         }
 575                         next unless $c;
 576                         foreach my $x (keys %$paths) {
 577                                 next unless ($p0 =~ /^\Q$x\E/);
 578                                 my $i = $paths->{$x};
 579                                 my $f = "$GIT_DIR/svn/$i/revs/$r0";
 580                                 while ($r0 && !-r $f) {
 581                                         # could be an older revision, too...
 582                                         $r0--;
 583                                         $f = "$GIT_DIR/svn/$i/revs/$r0";
 584                                 }
 585                                 unless (-r $f) {
 586                                         print STDERR "r$r0 of $i not imported,",
 587                                                                 " $dbg\n";
 588                                         next;
 589                                 }
 590                                 my $r1 = file_to_s($f);
 591                                 $grafts->{$c}->{$r1} = 1;
 592                         }
 593                 }
 594         }
 595 }
 596
 597 sub process_merge_msg_matches {
 598         my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
 599         my (@strong, @weak);
 600         foreach (@matches) {
 601                 # merging with ourselves is not interesting
 602                 next if $_ eq $p;
 603                 if ($l_map->{$u}->{$_}) {
 604                         push @strong, $_;
 605                 } else {
 606                         push @weak, $_;
 607                 }
 608         }
 609         foreach my $w (@weak) {
 610                 last if @strong;
 611                 # no exact match, use branch name as regexp.
 612                 my $re = qr/\Q$w\E/i;
 613                 foreach (keys %{$l_map->{$u}}) {
 614                         if (/$re/) {
 615                                 push @strong, $_;
 616                                 last;
 617                         }
 618                 }
 619                 last if @strong;
 620                 $w = basename($w);
 621                 $re = qr/\Q$w\E/i;
 622                 foreach (keys %{$l_map->{$u}}) {
 623                         if (/$re/) {
 624                                 push @strong, $_;
 625                                 last;
 626                         }
 627                 }
 628         }
 629         my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
 630                                         \s(?:[a-f\d\-]+)$/xsm);
 631         unless (defined $rev) {
 632                 ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
 633                                         \@(?:[a-f\d\-]+)/xsm);
 634                 return unless defined $rev;
 635         }
 636         foreach my $m (@strong) {
 637                 my ($r0, $s0) = find_rev_before($rev, $m);
 638                 $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
 639         }
 640 }
 641
 642 sub graft_merge_msg {
 643         my ($grafts, $l_map, $u, $p, @re) = @_;
 644
 645         my $x = $l_map->{$u}->{$p};
 646         my $rl = rev_list_raw($x);
 647         while (my $c = next_rev_list_entry($rl)) {
 648                 foreach my $re (@re) {
 649                         my (@br) = ($c->{m} =~ /$re/g);
 650                         next unless @br;
 651                         process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
 652                 }
 653         }
 654 }
 655
 656 sub read_uuid {
 657         return if $SVN_UUID;
 658         my $info = shift || svn_info('.');
 659         $SVN_UUID = $info->{'Repository UUID'} or
 660                                         croak "Repository UUID unreadable\n";
 661         s_to_file($SVN_UUID,"$GIT_SVN_DIR/info/uuid");
 662 }
 663
 664 sub quiet_run {
 665         my $pid = fork;
 666         defined $pid or croak $!;
 667         if (!$pid) {
 668                 open my $null, '>', '/dev/null' or croak $!;
 669                 open STDERR, '>&', $null or croak $!;
 670                 open STDOUT, '>&', $null or croak $!;
 671                 exec @_ or croak $!;
 672         }
 673         waitpid $pid, 0;
 674         return $?;
 675 }
 676
 677 sub repo_path_split {
 678         my $full_url = shift;
 679         $full_url =~ s#/+$##;
 680
 681         foreach (@repo_path_split_cache) {
 682                 if ($full_url =~ s#$_##) {
 683                         my $u = $1;
 684                         $full_url =~ s#^/+##;
 685                         return ($u, $full_url);
 686                 }
 687         }
 688
 689         my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i);
 690         $path =~ s#^/+##;
 691         my @paths = split(m#/+#, $path);
 692
 693         while (quiet_run(qw/svn ls --non-interactive/, $url)) {
 694                 my $n = shift @paths || last;
 695                 $url .= "/$n";
 696         }
 697         push @repo_path_split_cache, qr/^(\Q$url\E)/;
 698         $path = join('/',@paths);
 699         return ($url, $path);
 700 }
 701
 702 sub setup_git_svn {
 703         defined $SVN_URL or croak "SVN repository location required\n";
 704         unless (-d $GIT_DIR) {
 705                 croak "GIT_DIR=$GIT_DIR does not exist!\n";
 706         }
 707         mkpath([$GIT_SVN_DIR]);
 708         mkpath(["$GIT_SVN_DIR/info"]);
 709         mkpath([$REV_DIR]);
 710         s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
 711
 712         open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!;
 713         print $fd '.svn',"\n";
 714         close $fd or croak $!;
 715         my ($url, $path) = repo_path_split($SVN_URL);
 716         s_to_file($url, "$GIT_SVN_DIR/info/repo_url");
 717         s_to_file($path, "$GIT_SVN_DIR/info/repo_path");
 718 }
 719
 720 sub assert_svn_wc_clean {
 721         my ($svn_rev) = @_;
 722         croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
 723         my $lcr = svn_info('.')->{'Last Changed Rev'};
 724         if ($svn_rev != $lcr) {
 725                 print STDERR "Checking for copy-tree ... ";
 726                 my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
 727                                                 "-r$lcr:$svn_rev")));
 728                 if (@diff) {
 729                         croak "Nope!  Expected r$svn_rev, got r$lcr\n";
 730                 } else {
 731                         print STDERR "OK!\n";
 732                 }
 733         }
 734         my @status = grep(!/^Performing status on external/,(`svn status`));
 735         @status = grep(!/^\s*$/,@status);
 736         if (scalar @status) {
 737                 print STDERR "Tree ($SVN_WC) is not clean:\n";
 738                 print STDERR $_ foreach @status;
 739                 croak;
 740         }
 741 }
 742
 743 sub assert_tree {
 744         my ($treeish) = @_;
 745         croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
 746         chomp(my $type = `git-cat-file -t $treeish`);
 747         my $expected;
 748         while ($type eq 'tag') {
 749                 chomp(($treeish, $type) = `git-cat-file tag $treeish`);
 750         }
 751         if ($type eq 'commit') {
 752                 $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
 753                 ($expected) = ($expected =~ /^tree ($sha1)$/);
 754                 die "Unable to get tree from $treeish\n" unless $expected;
 755         } elsif ($type eq 'tree') {
 756                 $expected = $treeish;
 757         } else {
 758                 die "$treeish is a $type, expected tree, tag or commit\n";
 759         }
 760
 761         my $old_index = $ENV{GIT_INDEX_FILE};
 762         my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
 763         if (-e $tmpindex) {
 764                 unlink $tmpindex or croak $!;
 765         }
 766         $ENV{GIT_INDEX_FILE} = $tmpindex;
 767         index_changes(1);
 768         chomp(my $tree = `git-write-tree`);
 769         if ($old_index) {
 770                 $ENV{GIT_INDEX_FILE} = $old_index;
 771         } else {
 772                 delete $ENV{GIT_INDEX_FILE};
 773         }
 774         if ($tree ne $expected) {
 775                 croak "Tree mismatch, Got: $tree, Expected: $expected\n";
 776         }
 777         unlink $tmpindex;
 778 }
 779
 780 sub parse_diff_tree {
 781         my $diff_fh = shift;
 782         local $/ = "\0";
 783         my $state = 'meta';
 784         my @mods;
 785         while (<$diff_fh>) {
 786                 chomp $_; # this gets rid of the trailing "\0"
 787                 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
 788                                         $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
 789                         push @mods, {   mode_a => $1, mode_b => $2,
 790                                         sha1_b => $3, chg => $4 };
 791                         if ($4 =~ /^(?:C|R)$/) {
 792                                 $state = 'file_a';
 793                         } else {
 794                                 $state = 'file_b';
 795                         }
 796                 } elsif ($state eq 'file_a') {
 797                         my $x = $mods[$#mods] or croak "Empty array\n";
 798                         if ($x->{chg} !~ /^(?:C|R)$/) {
 799                                 croak "Error parsing $_, $x->{chg}\n";
 800                         }
 801                         $x->{file_a} = $_;
 802                         $state = 'file_b';
 803                 } elsif ($state eq 'file_b') {
 804                         my $x = $mods[$#mods] or croak "Empty array\n";
 805                         if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
 806                                 croak "Error parsing $_, $x->{chg}\n";
 807                         }
 808                         if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
 809                                 croak "Error parsing $_, $x->{chg}\n";
 810                         }
 811                         $x->{file_b} = $_;
 812                         $state = 'meta';
 813                 } else {
 814                         croak "Error parsing $_\n";
 815                 }
 816         }
 817         close $diff_fh or croak $!;
 818
 819         return \@mods;
 820 }
 821
 822 sub svn_check_prop_executable {
 823         my $m = shift;
 824         return if -l $m->{file_b};
 825         if ($m->{mode_b} =~ /755$/) {
 826                 chmod((0755 &~ umask),$m->{file_b}) or croak $!;
 827                 if ($m->{mode_a} !~ /755$/) {
 828                         sys(qw(svn propset svn:executable 1), $m->{file_b});
 829                 }
 830                 -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
 831         } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
 832                 sys(qw(svn propdel svn:executable), $m->{file_b});
 833                 chmod((0644 &~ umask),$m->{file_b}) or croak $!;
 834                 -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
 835         }
 836 }
 837
 838 sub svn_ensure_parent_path {
 839         my $dir_b = dirname(shift);
 840         svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
 841         mkpath([$dir_b]) unless (-d $dir_b);
 842         sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
 843 }
 844
 845 sub precommit_check {
 846         my $mods = shift;
 847         my (%rm_file, %rmdir_check, %added_check);
 848
 849         my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
 850         foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
 851                 if ($m->{chg} eq 'R') {
 852                         if (-d $m->{file_b}) {
 853                                 err_dir_to_file("$m->{file_a} => $m->{file_b}");
 854                         }
 855                         # dir/$file => dir/file/$file
 856                         my $dirname = dirname($m->{file_b});
 857                         while ($dirname ne File::Spec->curdir) {
 858                                 if ($dirname ne $m->{file_a}) {
 859                                         $dirname = dirname($dirname);
 860                                         next;
 861                                 }
 862                                 err_file_to_dir("$m->{file_a} => $m->{file_b}");
 863                         }
 864                         # baz/zzz => baz (baz is a file)
 865                         $dirname = dirname($m->{file_a});
 866                         while ($dirname ne File::Spec->curdir) {
 867                                 if ($dirname ne $m->{file_b}) {
 868                                         $dirname = dirname($dirname);
 869                                         next;
 870                                 }
 871                                 err_dir_to_file("$m->{file_a} => $m->{file_b}");
 872                         }
 873                 }
 874                 if ($m->{chg} =~ /^(D|R)$/) {
 875                         my $t = $1 eq 'D' ? 'file_b' : 'file_a';
 876                         $rm_file{ $m->{$t} } = 1;
 877                         my $dirname = dirname( $m->{$t} );
 878                         my $basename = basename( $m->{$t} );
 879                         $rmdir_check{$dirname}->{$basename} = 1;
 880                 } elsif ($m->{chg} =~ /^(?:A|C)$/) {
 881                         if (-d $m->{file_b}) {
 882                                 err_dir_to_file($m->{file_b});
 883                         }
 884                         my $dirname = dirname( $m->{file_b} );
 885                         my $basename = basename( $m->{file_b} );
 886                         $added_check{$dirname}->{$basename} = 1;
 887                         while ($dirname ne File::Spec->curdir) {
 888                                 if ($rm_file{$dirname}) {
 889                                         err_file_to_dir($m->{file_b});
 890                                 }
 891                                 $dirname = dirname $dirname;
 892                         }
 893                 }
 894         }
 895         return (\%rmdir_check, \%added_check);
 896
 897         sub err_dir_to_file {
 898                 my $file = shift;
 899                 print STDERR "Node change from directory to file ",
 900                                 "is not supported by Subversion: ",$file,"\n";
 901                 exit 1;
 902         }
 903         sub err_file_to_dir {
 904                 my $file = shift;
 905                 print STDERR "Node change from file to directory ",
 906                                 "is not supported by Subversion: ",$file,"\n";
 907                 exit 1;
 908         }
 909 }
 910
 911 sub svn_checkout_tree {
 912         my ($svn_rev, $treeish) = @_;
 913         my $from = file_to_s("$REV_DIR/$svn_rev");
 914         assert_tree($from);
 915         print "diff-tree $from $treeish\n";
 916         my $pid = open my $diff_fh, '-|';
 917         defined $pid or croak $!;
 918         if ($pid == 0) {
 919                 my @diff_tree = qw(git-diff-tree -z -r);
 920                 if ($_cp_similarity) {
 921                         push @diff_tree, "-C$_cp_similarity";
 922                 } else {
 923                         push @diff_tree, '-C';
 924                 }
 925                 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
 926                 push @diff_tree, "-l$_l" if defined $_l;
 927                 exec(@diff_tree, $from, $treeish) or croak $!;
 928         }
 929         my $mods = parse_diff_tree($diff_fh);
 930         unless (@$mods) {
 931                 # git can do empty commits, but SVN doesn't allow it...
 932                 return $mods;
 933         }
 934         my ($rm, $add) = precommit_check($mods);
 935
 936         my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
 937         foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
 938                 if ($m->{chg} eq 'C') {
 939                         svn_ensure_parent_path( $m->{file_b} );
 940                         sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
 941                         apply_mod_line_blob($m);
 942                         svn_check_prop_executable($m);
 943                 } elsif ($m->{chg} eq 'D') {
 944                         sys(qw(svn rm --force), $m->{file_b});
 945                 } elsif ($m->{chg} eq 'R') {
 946                         svn_ensure_parent_path( $m->{file_b} );
 947                         sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
 948                         apply_mod_line_blob($m);
 949                         svn_check_prop_executable($m);
 950                 } elsif ($m->{chg} eq 'M') {
 951                         apply_mod_line_blob($m);
 952                         svn_check_prop_executable($m);
 953                 } elsif ($m->{chg} eq 'T') {
 954                         sys(qw(svn rm --force),$m->{file_b});
 955                         apply_mod_line_blob($m);
 956                         sys(qw(svn add --force), $m->{file_b});
 957                         svn_check_prop_executable($m);
 958                 } elsif ($m->{chg} eq 'A') {
 959                         svn_ensure_parent_path( $m->{file_b} );
 960                         apply_mod_line_blob($m);
 961                         sys(qw(svn add --force), $m->{file_b});
 962                         svn_check_prop_executable($m);
 963                 } else {
 964                         croak "Invalid chg: $m->{chg}\n";
 965                 }
 966         }
 967
 968         assert_tree($treeish);
 969         if ($_rmdir) { # remove empty directories
 970                 handle_rmdir($rm, $add);
 971         }
 972         assert_tree($treeish);
 973         return $mods;
 974 }
 975
 976 # svn ls doesn't work with respect to the current working tree, but what's
 977 # in the repository.  There's not even an option for it... *sigh*
 978 # (added files don't show up and removed files remain in the ls listing)
 979 sub svn_ls_current {
 980         my ($dir, $rm, $add) = @_;
 981         chomp(my @ls = safe_qx('svn','ls',$dir));
 982         my @ret = ();
 983         foreach (@ls) {
 984                 s#/$##; # trailing slashes are evil
 985                 push @ret, $_ unless $rm->{$dir}->{$_};
 986         }
 987         if (exists $add->{$dir}) {
 988                 push @ret, keys %{$add->{$dir}};
 989         }
 990         return \@ret;
 991 }
 992
 993 sub handle_rmdir {
 994         my ($rm, $add) = @_;
 995
 996         foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
 997                 my $ls = svn_ls_current($dir, $rm, $add);
 998                 next if (scalar @$ls);
 999                 sys(qw(svn rm --force),$dir);
1000
1001                 my $dn = dirname $dir;
1002                 $rm->{ $dn }->{ basename $dir } = 1;
1003                 $ls = svn_ls_current($dn, $rm, $add);
1004                 while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
1005                         sys(qw(svn rm --force),$dn);
1006                         $dir = basename $dn;
1007                         $dn = dirname $dn;
1008                         $rm->{ $dn }->{ $dir } = 1;
1009                         $ls = svn_ls_current($dn, $rm, $add);
1010                 }
1011         }
1012 }
1013
1014 sub svn_commit_tree {
1015         my ($svn_rev, $commit) = @_;
1016         my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
1017         my %log_msg = ( msg => '' );
1018         open my $msg, '>', $commit_msg or croak $!;
1019
1020         chomp(my $type = `git-cat-file -t $commit`);
1021         if ($type eq 'commit') {
1022                 my $pid = open my $msg_fh, '-|';
1023                 defined $pid or croak $!;
1024
1025                 if ($pid == 0) {
1026                         exec(qw(git-cat-file commit), $commit) or croak $!;
1027                 }
1028                 my $in_msg = 0;
1029                 while (<$msg_fh>) {
1030                         if (!$in_msg) {
1031                                 $in_msg = 1 if (/^\s*$/);
1032                         } elsif (/^git-svn-id: /) {
1033                                 # skip this, we regenerate the correct one
1034                                 # on re-fetch anyways
1035                         } else {
1036                                 print $msg $_ or croak $!;
1037                         }
1038                 }
1039                 close $msg_fh or croak $!;
1040         }
1041         close $msg or croak $!;
1042
1043         if ($_edit || ($type eq 'tree')) {
1044                 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1045                 system($editor, $commit_msg);
1046         }
1047
1048         # file_to_s removes all trailing newlines, so just use chomp() here:
1049         open $msg, '<', $commit_msg or croak $!;
1050         { local $/; chomp($log_msg{msg} = <$msg>); }
1051         close $msg or croak $!;
1052
1053         my ($oneline) = ($log_msg{msg} =~ /([^\n\r]+)/);
1054         print "Committing $commit: $oneline\n";
1055
1056         if (defined $LC_ALL) {
1057                 $ENV{LC_ALL} = $LC_ALL;
1058         } else {
1059                 delete $ENV{LC_ALL};
1060         }
1061         my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
1062         $ENV{LC_ALL} = 'C';
1063         unlink $commit_msg;
1064         my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/);
1065         if (!defined $committed) {
1066                 my $out = join("\n",@ci_output);
1067                 print STDERR "W: Trouble parsing \`svn commit' output:\n\n",
1068                                 $out, "\n\nAssuming English locale...";
1069                 ($committed) = ($out =~ /^Committed revision \d+\./sm);
1070                 defined $committed or die " FAILED!\n",
1071                         "Commit output failed to parse committed revision!\n",
1072                 print STDERR " OK\n";
1073         }
1074
1075         my @svn_up = qw(svn up);
1076         push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
1077         if ($committed == ($svn_rev + 1)) {
1078                 push @svn_up, "-r$committed";
1079                 sys(@svn_up);
1080                 my $info = svn_info('.');
1081                 my $date = $info->{'Last Changed Date'} or die "Missing date\n";
1082                 if ($info->{'Last Changed Rev'} != $committed) {
1083                         croak "$info->{'Last Changed Rev'} != $committed\n"
1084                 }
1085                 my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1086                                         /(\d{4})\-(\d\d)\-(\d\d)\s
1087                                          (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1088                                          or croak "Failed to parse date: $date\n";
1089                 $log_msg{date} = "$tz $Y-$m-$d $H:$M:$S";
1090                 $log_msg{author} = $info->{'Last Changed Author'};
1091                 $log_msg{revision} = $committed;
1092                 $log_msg{msg} .= "\n";
1093                 my $parent = file_to_s("$REV_DIR/$svn_rev");
1094                 git_commit(\%log_msg, $parent, $commit);
1095                 return $committed;
1096         }
1097         # resync immediately
1098         push @svn_up, "-r$svn_rev";
1099         sys(@svn_up);
1100         return fetch("$committed=$commit")->{revision};
1101 }
1102
1103 sub rev_list_raw {
1104         my (@args) = @_;
1105         my $pid = open my $fh, '-|';
1106         defined $pid or croak $!;
1107         if (!$pid) {
1108                 exec(qw/git-rev-list --pretty=raw/, @args) or croak $!;
1109         }
1110         return { fh => $fh, t => { } };
1111 }
1112
1113 sub next_rev_list_entry {
1114         my $rl = shift;
1115         my $fh = $rl->{fh};
1116         my $x = $rl->{t};
1117         while (<$fh>) {
1118                 if (/^commit ($sha1)$/o) {
1119                         if ($x->{c}) {
1120                                 $rl->{t} = { c => $1 };
1121                                 return $x;
1122                         } else {
1123                                 $x->{c} = $1;
1124                         }
1125                 } elsif (/^parent ($sha1)$/o) {
1126                         $x->{p}->{$1} = 1;
1127                 } elsif (s/^    //) {
1128                         $x->{m} ||= '';
1129                         $x->{m} .= $_;
1130                 }
1131         }
1132         return ($x != $rl->{t}) ? $x : undef;
1133 }
1134
1135 # read the entire log into a temporary file (which is removed ASAP)
1136 # and store the file handle + parser state
1137 sub svn_log_raw {
1138         my (@log_args) = @_;
1139         my $log_fh = IO::File->new_tmpfile or croak $!;
1140         my $pid = fork;
1141         defined $pid or croak $!;
1142         if (!$pid) {
1143                 open STDOUT, '>&', $log_fh or croak $!;
1144                 exec (qw(svn log), @log_args) or croak $!
1145         }
1146         waitpid $pid, 0;
1147         croak $? if $?;
1148         seek $log_fh, 0, 0 or croak $!;
1149         return { state => 'sep', fh => $log_fh };
1150 }
1151
1152 sub next_log_entry {
1153         my $log = shift; # retval of svn_log_raw()
1154         my $ret = undef;
1155         my $fh = $log->{fh};
1156
1157         while (<$fh>) {
1158                 chomp;
1159                 if (/^\-{72}$/) {
1160                         if ($log->{state} eq 'msg') {
1161                                 if ($ret->{lines}) {
1162                                         $ret->{msg} .= $_."\n";
1163                                         unless(--$ret->{lines}) {
1164                                                 $log->{state} = 'sep';
1165                                         }
1166                                 } else {
1167                                         croak "Log parse error at: $_\n",
1168                                                 $ret->{revision},
1169                                                 "\n";
1170                                 }
1171                                 next;
1172                         }
1173                         if ($log->{state} ne 'sep') {
1174                                 croak "Log parse error at: $_\n",
1175                                         "state: $log->{state}\n",
1176                                         $ret->{revision},
1177                                         "\n";
1178                         }
1179                         $log->{state} = 'rev';
1180
1181                         # if we have an empty log message, put something there:
1182                         if ($ret) {
1183                                 $ret->{msg} ||= "\n";
1184                                 delete $ret->{lines};
1185                                 return $ret;
1186                         }
1187                         next;
1188                 }
1189                 if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
1190                         my $rev = $1;
1191                         my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
1192                         ($lines) = ($lines =~ /(\d+)/);
1193                         my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1194                                         /(\d{4})\-(\d\d)\-(\d\d)\s
1195                                          (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1196                                          or croak "Failed to parse date: $date\n";
1197                         $ret = {        revision => $rev,
1198                                         date => "$tz $Y-$m-$d $H:$M:$S",
1199                                         author => $author,
1200                                         lines => $lines,
1201                                         msg => '' };
1202                         if (defined $_authors && ! defined $users{$author}) {
1203                                 die "Author: $author not defined in ",
1204                                                 "$_authors file\n";
1205                         }
1206                         $log->{state} = 'msg_start';
1207                         next;
1208                 }
1209                 # skip the first blank line of the message:
1210                 if ($log->{state} eq 'msg_start' && /^$/) {
1211                         $log->{state} = 'msg';
1212                 } elsif ($log->{state} eq 'msg') {
1213                         if ($ret->{lines}) {
1214                                 $ret->{msg} .= $_."\n";
1215                                 unless (--$ret->{lines}) {
1216                                         $log->{state} = 'sep';
1217                                 }
1218                         } else {
1219                                 croak "Log parse error at: $_\n",
1220                                         $ret->{revision},"\n";
1221                         }
1222                 }
1223         }
1224         return $ret;
1225 }
1226
1227 sub svn_info {
1228         my $url = shift || $SVN_URL;
1229
1230         my $pid = open my $info_fh, '-|';
1231         defined $pid or croak $!;
1232
1233         if ($pid == 0) {
1234                 exec(qw(svn info),$url) or croak $!;
1235         }
1236
1237         my $ret = {};
1238         # only single-lines seem to exist in svn info output
1239         while (<$info_fh>) {
1240                 chomp $_;
1241                 if (m#^([^:]+)\s*:\s*(\S.*)$#) {
1242                         $ret->{$1} = $2;
1243                         push @{$ret->{-order}}, $1;
1244                 }
1245         }
1246         close $info_fh or croak $!;
1247         return $ret;
1248 }
1249
1250 sub sys { system(@_) == 0 or croak $? }
1251
1252 sub eol_cp {
1253         my ($from, $to) = @_;
1254         my $es = svn_propget_base('svn:eol-style', $to);
1255         open my $rfd, '<', $from or croak $!;
1256         binmode $rfd or croak $!;
1257         open my $wfd, '>', $to or croak $!;
1258         binmode $wfd or croak $!;
1259
1260         my $eol = $EOL{$es} or undef;
1261         my $buf;
1262         use bytes;
1263         while (1) {
1264                 my ($r, $w, $t);
1265                 defined($r = sysread($rfd, $buf, 4096)) or croak $!;
1266                 return unless $r;
1267                 if ($eol) {
1268                         if ($buf =~ /\015$/) {
1269                                 my $c;
1270                                 defined($r = sysread($rfd,$c,1)) or croak $!;
1271                                 $buf .= $c if $r > 0;
1272                         }
1273                         $buf =~ s/(?:\015\012|\015|\012)/$eol/gs;
1274                         $r = length($buf);
1275                 }
1276                 for ($w = 0; $w < $r; $w += $t) {
1277                         $t = syswrite($wfd, $buf, $r - $w, $w) or croak $!;
1278                 }
1279         }
1280         no bytes;
1281 }
1282
1283 sub do_update_index {
1284         my ($z_cmd, $cmd, $no_text_base) = @_;
1285
1286         my $z = open my $p, '-|';
1287         defined $z or croak $!;
1288         unless ($z) { exec @$z_cmd or croak $! }
1289
1290         my $pid = open my $ui, '|-';
1291         defined $pid or croak $!;
1292         unless ($pid) {
1293                 exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!;
1294         }
1295         local $/ = "\0";
1296         while (my $x = <$p>) {
1297                 chomp $x;
1298                 if (!$no_text_base && lstat $x && ! -l _ &&
1299                                 svn_propget_base('svn:keywords', $x)) {
1300                         my $mode = -x _ ? 0755 : 0644;
1301                         my ($v,$d,$f) = File::Spec->splitpath($x);
1302                         my $tb = File::Spec->catfile($d, '.svn', 'tmp',
1303                                                 'text-base',"$f.svn-base");
1304                         $tb =~ s#^/##;
1305                         unless (-f $tb) {
1306                                 $tb = File::Spec->catfile($d, '.svn',
1307                                                 'text-base',"$f.svn-base");
1308                                 $tb =~ s#^/##;
1309                         }
1310                         unlink $x or croak $!;
1311                         eol_cp($tb, $x);
1312                         chmod(($mode &~ umask), $x) or croak $!;
1313                 }
1314                 print $ui $x,"\0";
1315         }
1316         close $ui or croak $!;
1317 }
1318
1319 sub index_changes {
1320         my $no_text_base = shift;
1321         do_update_index([qw/git-diff-files --name-only -z/],
1322                         'remove',
1323                         $no_text_base);
1324         do_update_index([qw/git-ls-files -z --others/,
1325                                 "--exclude-from=$GIT_SVN_DIR/info/exclude"],
1326                         'add',
1327                         $no_text_base);
1328 }
1329
1330 sub s_to_file {
1331         my ($str, $file, $mode) = @_;
1332         open my $fd,'>',$file or croak $!;
1333         print $fd $str,"\n" or croak $!;
1334         close $fd or croak $!;
1335         chmod ($mode &~ umask, $file) if (defined $mode);
1336 }
1337
1338 sub file_to_s {
1339         my $file = shift;
1340         open my $fd,'<',$file or croak "$!: file: $file\n";
1341         local $/;
1342         my $ret = <$fd>;
1343         close $fd or croak $!;
1344         $ret =~ s/\s*$//s;
1345         return $ret;
1346 }
1347
1348 sub assert_revision_unknown {
1349         my $revno = shift;
1350         if (-f "$REV_DIR/$revno") {
1351                 croak "$REV_DIR/$revno already exists! ",
1352                                 "Why are we refetching it?";
1353         }
1354 }
1355
1356 sub trees_eq {
1357         my ($x, $y) = @_;
1358         my @x = safe_qx('git-cat-file','commit',$x);
1359         my @y = safe_qx('git-cat-file','commit',$y);
1360         if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
1361                                 || $y[0] !~ /^tree $sha1\n$/) {
1362                 print STDERR "Trees not equal: $y[0] != $x[0]\n";
1363                 return 0
1364         }
1365         return 1;
1366 }
1367
1368 sub assert_revision_eq_or_unknown {
1369         my ($revno, $commit) = @_;
1370         if (-f "$REV_DIR/$revno") {
1371                 my $current = file_to_s("$REV_DIR/$revno");
1372                 if (($commit ne $current) && !trees_eq($commit, $current)) {
1373                         croak "$REV_DIR/$revno already exists!\n",
1374                                 "current: $current\nexpected: $commit\n";
1375                 }
1376                 return;
1377         }
1378 }
1379
1380 sub git_commit {
1381         my ($log_msg, @parents) = @_;
1382         assert_revision_unknown($log_msg->{revision});
1383         my $out_fh = IO::File->new_tmpfile or croak $!;
1384
1385         map_tree_joins() if (@_branch_from && !%tree_map);
1386
1387         # commit parents can be conditionally bound to a particular
1388         # svn revision via: "svn_revno=commit_sha1", filter them out here:
1389         my @exec_parents;
1390         foreach my $p (@parents) {
1391                 next unless defined $p;
1392                 if ($p =~ /^(\d+)=($sha1_short)$/o) {
1393                         if ($1 == $log_msg->{revision}) {
1394                                 push @exec_parents, $2;
1395                         }
1396                 } else {
1397                         push @exec_parents, $p if $p =~ /$sha1_short/o;
1398                 }
1399         }
1400
1401         my $pid = fork;
1402         defined $pid or croak $!;
1403         if ($pid == 0) {
1404                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
1405                 index_changes();
1406                 chomp(my $tree = `git-write-tree`);
1407                 croak $? if $?;
1408                 if (exists $tree_map{$tree}) {
1409                         my %seen_parent = map { $_ => 1 } @exec_parents;
1410                         foreach (@{$tree_map{$tree}}) {
1411                                 # MAXPARENT is defined to 16 in commit-tree.c:
1412                                 if ($seen_parent{$_} || @exec_parents > 16) {
1413                                         next;
1414                                 }
1415                                 push @exec_parents, $_;
1416                                 $seen_parent{$_} = 1;
1417                         }
1418                 }
1419                 my $msg_fh = IO::File->new_tmpfile or croak $!;
1420                 print $msg_fh $log_msg->{msg}, "\ngit-svn-id: ",
1421                                         "$SVN_URL\@$log_msg->{revision}",
1422                                         " $SVN_UUID\n" or croak $!;
1423                 $msg_fh->flush == 0 or croak $!;
1424                 seek $msg_fh, 0, 0 or croak $!;
1425
1426                 set_commit_env($log_msg);
1427
1428                 my @exec = ('git-commit-tree',$tree);
1429                 push @exec, '-p', $_  foreach @exec_parents;
1430                 open STDIN, '<&', $msg_fh or croak $!;
1431                 open STDOUT, '>&', $out_fh or croak $!;
1432                 exec @exec or croak $!;
1433         }
1434         waitpid($pid,0);
1435         croak $? if $?;
1436
1437         $out_fh->flush == 0 or croak $!;
1438         seek $out_fh, 0, 0 or croak $!;
1439         chomp(my $commit = do { local $/; <$out_fh> });
1440         if ($commit !~ /^$sha1$/o) {
1441                 croak "Failed to commit, invalid sha1: $commit\n";
1442         }
1443         my @update_ref = ('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
1444         if (my $primary_parent = shift @exec_parents) {
1445                 $pid = fork;
1446                 defined $pid or croak $!;
1447                 if (!$pid) {
1448                         close STDERR;
1449                         close STDOUT;
1450                         exec 'git-rev-parse','--verify',
1451                                         "refs/remotes/$GIT_SVN^0" or croak $!;
1452                 }
1453                 waitpid $pid, 0;
1454                 push @update_ref, $primary_parent unless $?;
1455         }
1456         sys(@update_ref);
1457         sys('git-update-ref',"svn/$GIT_SVN/revs/$log_msg->{revision}",$commit);
1458         print "r$log_msg->{revision} = $commit\n";
1459         if ($_repack && (--$_repack_nr == 0)) {
1460                 $_repack_nr = $_repack;
1461                 sys("git repack $_repack_flags");
1462         }
1463         return $commit;
1464 }
1465
1466 sub set_commit_env {
1467         my ($log_msg) = @_;
1468         my $author = $log_msg->{author};
1469         my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
1470                                 : ($author,"$author\@$SVN_UUID");
1471         $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1472         $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1473         $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
1474 }
1475
1476 sub apply_mod_line_blob {
1477         my $m = shift;
1478         if ($m->{mode_b} =~ /^120/) {
1479                 blob_to_symlink($m->{sha1_b}, $m->{file_b});
1480         } else {
1481                 blob_to_file($m->{sha1_b}, $m->{file_b});
1482         }
1483 }
1484
1485 sub blob_to_symlink {
1486         my ($blob, $link) = @_;
1487         defined $link or croak "\$link not defined!\n";
1488         croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
1489         if (-l $link || -f _) {
1490                 unlink $link or croak $!;
1491         }
1492
1493         my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
1494         symlink $dest, $link or croak $!;
1495 }
1496
1497 sub blob_to_file {
1498         my ($blob, $file) = @_;
1499         defined $file or croak "\$file not defined!\n";
1500         croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
1501         if (-l $file || -f _) {
1502                 unlink $file or croak $!;
1503         }
1504
1505         open my $blob_fh, '>', $file or croak "$!: $file\n";
1506         my $pid = fork;
1507         defined $pid or croak $!;
1508
1509         if ($pid == 0) {
1510                 open STDOUT, '>&', $blob_fh or croak $!;
1511                 exec('git-cat-file','blob',$blob) or croak $!;
1512         }
1513         waitpid $pid, 0;
1514         croak $? if $?;
1515
1516         close $blob_fh or croak $!;
1517 }
1518
1519 sub safe_qx {
1520         my $pid = open my $child, '-|';
1521         defined $pid or croak $!;
1522         if ($pid == 0) {
1523                 exec(@_) or croak $!;
1524         }
1525         my @ret = (<$child>);
1526         close $child or croak $?;
1527         die $? if $?; # just in case close didn't error out
1528         return wantarray ? @ret : join('',@ret);
1529 }
1530
1531 sub svn_compat_check {
1532         my @co_help = safe_qx(qw(svn co -h));
1533         unless (grep /ignore-externals/,@co_help) {
1534                 print STDERR "W: Installed svn version does not support ",
1535                                 "--ignore-externals\n";
1536                 $_no_ignore_ext = 1;
1537         }
1538         if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
1539                 $_svn_co_url_revs = 1;
1540         }
1541         if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) {
1542                 $_svn_pg_peg_revs = 1;
1543         }
1544
1545         # I really, really hope nobody hits this...
1546         unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
1547                 print STDERR <<'';
1548 W: The installed svn version does not support the --stop-on-copy flag in
1549    the log command.
1550    Lets hope the directory you're tracking is not a branch or tag
1551    and was never moved within the repository...
1552
1553                 $_no_stop_copy = 1;
1554         }
1555 }
1556
1557 # *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
1558 # (and they won't honor URL@<rev> without -r<rev>, too!)
1559 sub svn_cmd_checkout {
1560         my ($url, $rev, $dir) = @_;
1561         my @cmd = ('svn','co', "-r$rev");
1562         push @cmd, '--ignore-externals' unless $_no_ignore_ext;
1563         $url .= "\@$rev" if $_svn_co_url_revs;
1564         sys(@cmd, $url, $dir);
1565 }
1566
1567 sub check_upgrade_needed {
1568         my $old = eval {
1569                 my $pid = open my $child, '-|';
1570                 defined $pid or croak $!;
1571                 if ($pid == 0) {
1572                         close STDERR;
1573                         exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $!;
1574                 }
1575                 my @ret = (<$child>);
1576                 close $child or croak $?;
1577                 die $? if $?; # just in case close didn't error out
1578                 return wantarray ? @ret : join('',@ret);
1579         };
1580         return unless $old;
1581         my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
1582         if ($@ || !$head) {
1583                 print STDERR "Please run: $0 rebuild --upgrade\n";
1584                 exit 1;
1585         }
1586 }
1587
1588 # fills %tree_map with a reverse mapping of trees to commits.  Useful
1589 # for finding parents to commit on.
1590 sub map_tree_joins {
1591         my %seen;
1592         foreach my $br (@_branch_from) {
1593                 my $pid = open my $pipe, '-|';
1594                 defined $pid or croak $!;
1595                 if ($pid == 0) {
1596                         exec(qw(git-rev-list --topo-order --pretty=raw), $br)
1597                                                                 or croak $!;
1598                 }
1599                 while (<$pipe>) {
1600                         if (/^commit ($sha1)$/o) {
1601                                 my $commit = $1;
1602
1603                                 # if we've seen a commit,
1604                                 # we've seen its parents
1605                                 last if $seen{$commit};
1606                                 my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
1607                                 unless (defined $tree) {
1608                                         die "Failed to parse commit $commit\n";
1609                                 }
1610                                 push @{$tree_map{$tree}}, $commit;
1611                                 $seen{$commit} = 1;
1612                         }
1613                 }
1614                 close $pipe; # we could be breaking the pipe early
1615         }
1616 }
1617
1618 sub load_all_refs {
1619         if (@_branch_from) {
1620                 print STDERR '--branch|-b parameters are ignored when ',
1621                         "--branch-all-refs|-B is passed\n";
1622         }
1623
1624         # don't worry about rev-list on non-commit objects/tags,
1625         # it shouldn't blow up if a ref is a blob or tree...
1626         chomp(@_branch_from = `git-rev-parse --symbolic --all`);
1627 }
1628
1629 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1630 sub load_authors {
1631         open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1632         while (<$authors>) {
1633                 chomp;
1634                 next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
1635                 my ($user, $name, $email) = ($1, $2, $3);
1636                 $users{$user} = [$name, $email];
1637         }
1638         close $authors or croak $!;
1639 }
1640
1641 sub svn_propget_base {
1642         my ($p, $f) = @_;
1643         $f .= '@BASE' if $_svn_pg_peg_revs;
1644         return safe_qx(qw/svn propget/, $p, $f);
1645 }
1646
1647 sub git_svn_each {
1648         my $sub = shift;
1649         foreach (`git-rev-parse --symbolic --all`) {
1650                 next unless s#^refs/remotes/##;
1651                 chomp $_;
1652                 next unless -f "$GIT_DIR/svn/$_/info/url";
1653                 &$sub($_);
1654         }
1655 }
1656
1657 sub migration_check {
1658         return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
1659         print "Upgrading repository...\n";
1660         unless (-d "$GIT_DIR/svn") {
1661                 mkdir "$GIT_DIR/svn" or croak $!;
1662         }
1663         print "Data from a previous version of git-svn exists, but\n\t",
1664                                 "$GIT_SVN_DIR\n\t(required for this version ",
1665                                 "($VERSION) of git-svn) does not.\n";
1666
1667         foreach my $x (`git-rev-parse --symbolic --all`) {
1668                 next unless $x =~ s#^refs/remotes/##;
1669                 chomp $x;
1670                 next unless -f "$GIT_DIR/$x/info/url";
1671                 my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
1672                 next unless $u;
1673                 my $dn = dirname("$GIT_DIR/svn/$x");
1674                 mkpath([$dn]) unless -d $dn;
1675                 rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
1676                 my ($url, $path) = repo_path_split($u);
1677                 s_to_file($url, "$GIT_DIR/svn/$x/info/repo_url");
1678                 s_to_file($path, "$GIT_DIR/svn/$x/info/repo_path");
1679         }
1680         print "Done upgrading.\n";
1681 }
1682
1683 sub find_rev_before {
1684         my ($r, $git_svn_id) = @_;
1685         my @revs = map { basename $_ } <$GIT_DIR/svn/$git_svn_id/revs/*>;
1686         foreach my $r0 (sort { $b <=> $a } @revs) {
1687                 next if $r0 >= $r;
1688                 return ($r0, file_to_s("$GIT_DIR/svn/$git_svn_id/revs/$r0"));
1689         }
1690         return (undef, undef);
1691 }
1692
1693 sub init_vars {
1694         $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
1695         $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
1696         $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
1697         $SVN_URL = undef;
1698         $REV_DIR = "$GIT_SVN_DIR/revs";
1699         $SVN_WC = "$GIT_SVN_DIR/tree";
1700 }
1701
1702 # convert GetOpt::Long specs for use by git-repo-config
1703 sub read_repo_config {
1704         return unless -d $GIT_DIR;
1705         my $opts = shift;
1706         foreach my $o (keys %$opts) {
1707                 my $v = $opts->{$o};
1708                 my ($key) = ($o =~ /^([a-z\-]+)/);
1709                 $key =~ s/-//g;
1710                 my $arg = 'git-repo-config';
1711                 $arg .= ' --int' if ($o =~ /[:=]i$/);
1712                 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1713                 if (ref $v eq 'ARRAY') {
1714                         chomp(my @tmp = `$arg --get-all svn.$key`);
1715                         @$v = @tmp if @tmp;
1716                 } else {
1717                         chomp(my $tmp = `$arg --get svn.$key`);
1718                         if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
1719                                 $$v = $tmp;
1720                         }
1721                 }
1722         }
1723 }
1724
1725 sub set_default_vals {
1726         if (defined $_repack) {
1727                 $_repack = 1000 if ($_repack <= 0);
1728                 $_repack_nr = $_repack;
1729                 $_repack_flags ||= '';
1730         }
1731 }
1732
1733 sub read_grafts {
1734         my $gr_file = shift;
1735         my ($grafts, $comments) = ({}, {});
1736         if (open my $fh, '<', $gr_file) {
1737                 my @tmp;
1738                 while (<$fh>) {
1739                         if (/^($sha1)\s+/) {
1740                                 my $c = $1;
1741                                 if (@tmp) {
1742                                         @{$comments->{$c}} = @tmp;
1743                                         @tmp = ();
1744                                 }
1745                                 foreach my $p (split /\s+/, $_) {
1746                                         $grafts->{$c}->{$p} = 1;
1747                                 }
1748                         } else {
1749                                 push @tmp, $_;
1750                         }
1751                 }
1752                 close $fh or croak $!;
1753                 @{$comments->{'END'}} = @tmp if @tmp;
1754         }
1755         return ($grafts, $comments);
1756 }
1757
1758 sub write_grafts {
1759         my ($grafts, $comments, $gr_file) = @_;
1760
1761         open my $fh, '>', $gr_file or croak $!;
1762         foreach my $c (sort keys %$grafts) {
1763                 if ($comments->{$c}) {
1764                         print $fh $_ foreach @{$comments->{$c}};
1765                 }
1766                 my $p = $grafts->{$c};
1767                 delete $p->{$c}; # commits are not self-reproducing...
1768                 my $pid = open my $ch, '-|';
1769                 defined $pid or croak $!;
1770                 if (!$pid) {
1771                         exec(qw/git-cat-file commit/, $c) or croak $!;
1772                 }
1773                 while (<$ch>) {
1774                         if (/^parent ([a-f\d]{40})/) {
1775                                 $p->{$1} = 1;
1776                         } else {
1777                                 last unless /^\S/i;
1778                         }
1779                 }
1780                 close $ch; # breaking the pipe
1781                 print $fh $c, ' ', join(' ', sort keys %$p),"\n";
1782         }
1783         if ($comments->{'END'}) {
1784                 print $fh $_ foreach @{$comments->{'END'}};
1785         }
1786         close $fh or croak $!;
1787 }
1788
1789 sub read_url_paths {
1790         my $l_map = {};
1791         git_svn_each(sub { my $x = shift;
1792                         my $u = file_to_s("$GIT_DIR/svn/$x/info/repo_url");
1793                         my $p = file_to_s("$GIT_DIR/svn/$x/info/repo_path");
1794                         # we hate trailing slashes
1795                         if ($u =~ s#(?:^\/+|\/+$)##g) {
1796                                 s_to_file($u,"$GIT_DIR/svn/$x/info/repo_url");
1797                         }
1798                         if ($p =~ s#(?:^\/+|\/+$)##g) {
1799                                 s_to_file($p,"$GIT_DIR/svn/$x/info/repo_path");
1800                         }
1801                         $l_map->{$u}->{$p} = $x;
1802                         });
1803         return $l_map;
1804 }
1805
1806 __END__
1807
1808 Data structures:
1809
1810 $svn_log hashref (as returned by svn_log_raw)
1811 {
1812         fh => file handle of the log file,
1813         state => state of the log file parser (sep/msg/rev/msg_start...)
1814 }
1815
1816 $log_msg hashref as returned by next_log_entry($svn_log)
1817 {
1818         msg => 'whitespace-formatted log entry
1819 ',                                              # trailing newline is preserved
1820         revision => '8',                        # integer
1821         date => '2004-02-24T17:01:44.108345Z',  # commit date
1822         author => 'committer name'
1823 };
1824
1825
1826 @mods = array of diff-index line hashes, each element represents one line
1827         of diff-index output
1828
1829 diff-index line ($m hash)
1830 {
1831         mode_a => first column of diff-index output, no leading ':',
1832         mode_b => second column of diff-index output,
1833         sha1_b => sha1sum of the final blob,
1834         chg => change type [MCRADT],
1835         file_a => original file name of a file (iff chg is 'C' or 'R')
1836         file_b => new/current file name of a file (any chg)
1837 }
1838 ;