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/;
  10 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  11 $VERSION = '0.10.0';
  12 $GIT_DIR = $ENV{GIT_DIR} || "$ENV{PWD}/.git";
  13 # make sure the svn binary gives consistent output between locales and TZs:
  14 $ENV{TZ} = 'UTC';
  15 $ENV{LC_ALL} = 'C';
  16
  17 # If SVN:: library support is added, please make the dependencies
  18 # optional and preserve the capability to use the command-line client.
  19 # use eval { require SVN::... } to make it lazy load
  20 # We don't use any modules not in the standard Perl distribution:
  21 use Carp qw/croak/;
  22 use IO::File qw//;
  23 use File::Basename qw/dirname basename/;
  24 use File::Path qw/mkpath/;
  25 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  26 use File::Spec qw//;
  27 use POSIX qw/strftime/;
  28 my $sha1 = qr/[a-f\d]{40}/;
  29 my $sha1_short = qr/[a-f\d]{4,40}/;
  30 my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
  31         $_find_copies_harder, $_l, $_version, $_upgrade, $_authors);
  32 my (@_branch_from, %tree_map, %users);
  33
  34 my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
  35                 'branch|b=s' => \@_branch_from,
  36                 'authors-file|A=s' => \$_authors );
  37 my %cmd = (
  38         fetch => [ \&fetch, "Download new revisions from SVN",
  39                         { 'revision|r=s' => \$_revision, %fc_opts } ],
  40         init => [ \&init, "Initialize and fetch (import)", { } ],
  41         commit => [ \&commit, "Commit git revisions to SVN",
  42                         {       'stdin|' => \$_stdin,
  43                                 'edit|e' => \$_edit,
  44                                 'rmdir' => \$_rmdir,
  45                                 'find-copies-harder' => \$_find_copies_harder,
  46                                 'l=i' => \$_l,
  47                                 %fc_opts,
  48                         } ],
  49         'show-ignore' => [ \&show_ignore, "Show svn:ignore listings", { } ],
  50         rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
  51                         { 'no-ignore-externals' => \$_no_ignore_ext,
  52                           'upgrade' => \$_upgrade } ],
  53 );
  54 my $cmd;
  55 for (my $i = 0; $i < @ARGV; $i++) {
  56         if (defined $cmd{$ARGV[$i]}) {
  57                 $cmd = $ARGV[$i];
  58                 splice @ARGV, $i, 1;
  59                 last;
  60         }
  61 };
  62
  63 # we may be called as git-svn-(command), or git-svn(command).
  64 foreach (keys %cmd) {
  65         if (/git\-svn\-?($_)(?:\.\w+)?$/) {
  66                 $cmd = $1;
  67                 last;
  68         }
  69 }
  70
  71 my %opts;
  72 %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
  73
  74 GetOptions(%opts, 'help|H|h' => \$_help,
  75                 'version|V' => \$_version,
  76                 'id|i=s' => \$GIT_SVN) or exit 1;
  77
  78 $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
  79 $GIT_SVN_INDEX = "$GIT_DIR/$GIT_SVN/index";
  80 $ENV{GIT_DIR} ||= $GIT_DIR;
  81 $SVN_URL = undef;
  82 $REV_DIR = "$GIT_DIR/$GIT_SVN/revs";
  83 $SVN_WC = "$GIT_DIR/$GIT_SVN/tree";
  84
  85 usage(0) if $_help;
  86 version() if $_version;
  87 usage(1) unless defined $cmd;
  88 load_authors() if $_authors;
  89 svn_check_ignore_externals();
  90 $cmd{$cmd}->[0]->(@ARGV);
  91 exit 0;
  92
  93 ####################### primary functions ######################
  94 sub usage {
  95         my $exit = shift || 0;
  96         my $fd = $exit ? \*STDERR : \*STDOUT;
  97         print $fd <<"";
  98 git-svn - bidirectional operations between a single Subversion tree and git
  99 Usage: $0 <command> [options] [arguments]\n
 100 Available commands:
 101
 102         foreach (sort keys %cmd) {
 103                 print $fd '  ',pack('A13',$_),$cmd{$_}->[1],"\n";
 104         }
 105         print $fd <<"";
 106 \nGIT_SVN_ID may be set in the environment to an arbitrary identifier if
 107 you're tracking multiple SVN branches/repositories in one git repository
 108 and want to keep them separate.  See git-svn(1) for more information.
 109
 110         exit $exit;
 111 }
 112
 113 sub version {
 114         print "git-svn version $VERSION\n";
 115         exit 0;
 116 }
 117
 118 sub rebuild {
 119         $SVN_URL = shift or undef;
 120         my $newest_rev = 0;
 121         if ($_upgrade) {
 122                 sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD");
 123         } else {
 124                 check_upgrade_needed();
 125         }
 126
 127         my $pid = open(my $rev_list,'-|');
 128         defined $pid or croak $!;
 129         if ($pid == 0) {
 130                 exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!;
 131         }
 132         my $latest;
 133         while (<$rev_list>) {
 134                 chomp;
 135                 my $c = $_;
 136                 croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 137                 my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
 138                 next if (!@commit); # skip merges
 139                 my $id = $commit[$#commit];
 140                 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 141                                                 \s([a-f\d\-]+)$/x);
 142                 if (!$rev || !$uuid || !$url) {
 143                         # some of the original repositories I made had
 144                         # indentifiers like this:
 145                         ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)
 146                                                         \@([a-f\d\-]+)/x);
 147                         if (!$rev || !$uuid) {
 148                                 croak "Unable to extract revision or UUID from ",
 149                                         "$c, $id\n";
 150                         }
 151                 }
 152
 153                 # if we merged or otherwise started elsewhere, this is
 154                 # how we break out of it
 155                 next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
 156                 next if (defined $SVN_URL && ($url ne $SVN_URL));
 157
 158                 print "r$rev = $c\n";
 159                 unless (defined $latest) {
 160                         if (!$SVN_URL && !$url) {
 161                                 croak "SVN repository location required: $url\n";
 162                         }
 163                         $SVN_URL ||= $url;
 164                         $SVN_UUID ||= setup_git_svn();
 165                         $latest = $rev;
 166                 }
 167                 assert_revision_eq_or_unknown($rev, $c);
 168                 sys('git-update-ref',"$GIT_SVN/revs/$rev",$c);
 169                 $newest_rev = $rev if ($rev > $newest_rev);
 170         }
 171         close $rev_list or croak $?;
 172         if (!chdir $SVN_WC) {
 173                 my @svn_co = ('svn','co',"-r$latest");
 174                 push @svn_co, '--ignore-externals' unless $_no_ignore_ext;
 175                 sys(@svn_co, $SVN_URL, $SVN_WC);
 176                 chdir $SVN_WC or croak $!;
 177         }
 178
 179         $pid = fork;
 180         defined $pid or croak $!;
 181         if ($pid == 0) {
 182                 my @svn_up = qw(svn up);
 183                 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 184                 sys(@svn_up,"-r$newest_rev");
 185                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 186                 git_addremove();
 187                 exec('git-write-tree');
 188         }
 189         waitpid $pid, 0;
 190
 191         if ($_upgrade) {
 192                 print STDERR <<"";
 193 Keeping deprecated refs/head/$GIT_SVN-HEAD for now.  Please remove it
 194 when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN
 195
 196         }
 197 }
 198
 199 sub init {
 200         $SVN_URL = shift or croak "SVN repository location required\n";
 201         unless (-d $GIT_DIR) {
 202                 sys('git-init-db');
 203         }
 204         setup_git_svn();
 205 }
 206
 207 sub fetch {
 208         my (@parents) = @_;
 209         check_upgrade_needed();
 210         $SVN_URL ||= file_to_s("$GIT_DIR/$GIT_SVN/info/url");
 211         my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 212         unless ($_revision) {
 213                 $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 214         }
 215         push @log_args, "-r$_revision";
 216         push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 217
 218         my $svn_log = svn_log_raw(@log_args);
 219         @$svn_log = sort { $a->{revision} <=> $b->{revision} } @$svn_log;
 220
 221         my $base = shift @$svn_log or croak "No base revision!\n";
 222         my $last_commit = undef;
 223         unless (-d $SVN_WC) {
 224                 my @svn_co = ('svn','co',"-r$base->{revision}");
 225                 push @svn_co,'--ignore-externals' unless $_no_ignore_ext;
 226                 sys(@svn_co, $SVN_URL, $SVN_WC);
 227                 chdir $SVN_WC or croak $!;
 228                 $last_commit = git_commit($base, @parents);
 229                 assert_svn_wc_clean($base->{revision}, $last_commit);
 230         } else {
 231                 chdir $SVN_WC or croak $!;
 232                 $last_commit = file_to_s("$REV_DIR/$base->{revision}");
 233         }
 234         my @svn_up = qw(svn up);
 235         push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 236         my $last_rev = $base->{revision};
 237         foreach my $log_msg (@$svn_log) {
 238                 assert_svn_wc_clean($last_rev, $last_commit);
 239                 $last_rev = $log_msg->{revision};
 240                 sys(@svn_up,"-r$last_rev");
 241                 $last_commit = git_commit($log_msg, $last_commit, @parents);
 242         }
 243         assert_svn_wc_clean($last_rev, $last_commit);
 244         unless (-e "$GIT_DIR/refs/heads/master") {
 245                 sys(qw(git-update-ref refs/heads/master),$last_commit);
 246         }
 247         return pop @$svn_log;
 248 }
 249
 250 sub commit {
 251         my (@commits) = @_;
 252         check_upgrade_needed();
 253         if ($_stdin || !@commits) {
 254                 print "Reading from stdin...\n";
 255                 @commits = ();
 256                 while (<STDIN>) {
 257                         if (/\b($sha1_short)\b/o) {
 258                                 unshift @commits, $1;
 259                         }
 260                 }
 261         }
 262         my @revs;
 263         foreach my $c (@commits) {
 264                 chomp(my @tmp = safe_qx('git-rev-parse',$c));
 265                 if (scalar @tmp == 1) {
 266                         push @revs, $tmp[0];
 267                 } elsif (scalar @tmp > 1) {
 268                         push @revs, reverse (safe_qx('git-rev-list',@tmp));
 269                 } else {
 270                         die "Failed to rev-parse $c\n";
 271                 }
 272         }
 273         chomp @revs;
 274
 275         fetch();
 276         chdir $SVN_WC or croak $!;
 277         my $svn_current_rev =  svn_info('.')->{'Last Changed Rev'};
 278         foreach my $c (@revs) {
 279                 my $mods = svn_checkout_tree($svn_current_rev, $c);
 280                 if (scalar @$mods == 0) {
 281                         print "Skipping, no changes detected\n";
 282                         next;
 283                 }
 284                 $svn_current_rev = svn_commit_tree($svn_current_rev, $c);
 285         }
 286         print "Done committing ",scalar @revs," revisions to SVN\n";
 287
 288 }
 289
 290 sub show_ignore {
 291         require File::Find or die $!;
 292         my $exclude_file = "$GIT_DIR/info/exclude";
 293         open my $fh, '<', $exclude_file or croak $!;
 294         chomp(my @excludes = (<$fh>));
 295         close $fh or croak $!;
 296
 297         $SVN_URL ||= file_to_s("$GIT_DIR/$GIT_SVN/info/url");
 298         chdir $SVN_WC or croak $!;
 299         my %ign;
 300         File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 301                 s#^\./##;
 302                 @{$ign{$_}} = safe_qx(qw(svn propget svn:ignore),$_);
 303                 }}, no_chdir=>1},'.');
 304
 305         print "\n# /\n";
 306         foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 307         delete $ign{'.'};
 308         foreach my $i (sort keys %ign) {
 309                 print "\n# ",$i,"\n";
 310                 foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 311         }
 312 }
 313
 314 ########################### utility functions #########################
 315
 316 sub setup_git_svn {
 317         defined $SVN_URL or croak "SVN repository location required\n";
 318         unless (-d $GIT_DIR) {
 319                 croak "GIT_DIR=$GIT_DIR does not exist!\n";
 320         }
 321         mkpath(["$GIT_DIR/$GIT_SVN"]);
 322         mkpath(["$GIT_DIR/$GIT_SVN/info"]);
 323         mkpath([$REV_DIR]);
 324         s_to_file($SVN_URL,"$GIT_DIR/$GIT_SVN/info/url");
 325         $SVN_UUID = svn_info($SVN_URL)->{'Repository UUID'} or
 326                                         croak "Repository UUID unreadable\n";
 327         s_to_file($SVN_UUID,"$GIT_DIR/$GIT_SVN/info/uuid");
 328
 329         open my $fd, '>>', "$GIT_DIR/$GIT_SVN/info/exclude" or croak $!;
 330         print $fd '.svn',"\n";
 331         close $fd or croak $!;
 332         return $SVN_UUID;
 333 }
 334
 335 sub assert_svn_wc_clean {
 336         my ($svn_rev, $treeish) = @_;
 337         croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
 338         croak "$treeish is not a sha1!\n" unless ($treeish =~ /^$sha1$/o);
 339         my $svn_info = svn_info('.');
 340         if ($svn_rev != $svn_info->{'Last Changed Rev'}) {
 341                 croak "Expected r$svn_rev, got r",
 342                                 $svn_info->{'Last Changed Rev'},"\n";
 343         }
 344         my @status = grep(!/^Performing status on external/,(`svn status`));
 345         @status = grep(!/^\s*$/,@status);
 346         if (scalar @status) {
 347                 print STDERR "Tree ($SVN_WC) is not clean:\n";
 348                 print STDERR $_ foreach @status;
 349                 croak;
 350         }
 351         assert_tree($treeish);
 352 }
 353
 354 sub assert_tree {
 355         my ($treeish) = @_;
 356         croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
 357         chomp(my $type = `git-cat-file -t $treeish`);
 358         my $expected;
 359         while ($type eq 'tag') {
 360                 chomp(($treeish, $type) = `git-cat-file tag $treeish`);
 361         }
 362         if ($type eq 'commit') {
 363                 $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
 364                 ($expected) = ($expected =~ /^tree ($sha1)$/);
 365                 die "Unable to get tree from $treeish\n" unless $expected;
 366         } elsif ($type eq 'tree') {
 367                 $expected = $treeish;
 368         } else {
 369                 die "$treeish is a $type, expected tree, tag or commit\n";
 370         }
 371
 372         my $old_index = $ENV{GIT_INDEX_FILE};
 373         my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
 374         if (-e $tmpindex) {
 375                 unlink $tmpindex or croak $!;
 376         }
 377         $ENV{GIT_INDEX_FILE} = $tmpindex;
 378         git_addremove();
 379         chomp(my $tree = `git-write-tree`);
 380         if ($old_index) {
 381                 $ENV{GIT_INDEX_FILE} = $old_index;
 382         } else {
 383                 delete $ENV{GIT_INDEX_FILE};
 384         }
 385         if ($tree ne $expected) {
 386                 croak "Tree mismatch, Got: $tree, Expected: $expected\n";
 387         }
 388 }
 389
 390 sub parse_diff_tree {
 391         my $diff_fh = shift;
 392         local $/ = "\0";
 393         my $state = 'meta';
 394         my @mods;
 395         while (<$diff_fh>) {
 396                 chomp $_; # this gets rid of the trailing "\0"
 397                 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
 398                                         $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
 399                         push @mods, {   mode_a => $1, mode_b => $2,
 400                                         sha1_b => $3, chg => $4 };
 401                         if ($4 =~ /^(?:C|R)$/) {
 402                                 $state = 'file_a';
 403                         } else {
 404                                 $state = 'file_b';
 405                         }
 406                 } elsif ($state eq 'file_a') {
 407                         my $x = $mods[$#mods] or croak "Empty array\n";
 408                         if ($x->{chg} !~ /^(?:C|R)$/) {
 409                                 croak "Error parsing $_, $x->{chg}\n";
 410                         }
 411                         $x->{file_a} = $_;
 412                         $state = 'file_b';
 413                 } elsif ($state eq 'file_b') {
 414                         my $x = $mods[$#mods] or croak "Empty array\n";
 415                         if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
 416                                 croak "Error parsing $_, $x->{chg}\n";
 417                         }
 418                         if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
 419                                 croak "Error parsing $_, $x->{chg}\n";
 420                         }
 421                         $x->{file_b} = $_;
 422                         $state = 'meta';
 423                 } else {
 424                         croak "Error parsing $_\n";
 425                 }
 426         }
 427         close $diff_fh or croak $!;
 428
 429         return \@mods;
 430 }
 431
 432 sub svn_check_prop_executable {
 433         my $m = shift;
 434         return if -l $m->{file_b};
 435         if ($m->{mode_b} =~ /755$/) {
 436                 chmod((0755 &~ umask),$m->{file_b}) or croak $!;
 437                 if ($m->{mode_a} !~ /755$/) {
 438                         sys(qw(svn propset svn:executable 1), $m->{file_b});
 439                 }
 440                 -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
 441         } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
 442                 sys(qw(svn propdel svn:executable), $m->{file_b});
 443                 chmod((0644 &~ umask),$m->{file_b}) or croak $!;
 444                 -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
 445         }
 446 }
 447
 448 sub svn_ensure_parent_path {
 449         my $dir_b = dirname(shift);
 450         svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
 451         mkpath([$dir_b]) unless (-d $dir_b);
 452         sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
 453 }
 454
 455 sub precommit_check {
 456         my $mods = shift;
 457         my (%rm_file, %rmdir_check, %added_check);
 458
 459         my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
 460         foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
 461                 if ($m->{chg} eq 'R') {
 462                         if (-d $m->{file_b}) {
 463                                 err_dir_to_file("$m->{file_a} => $m->{file_b}");
 464                         }
 465                         # dir/$file => dir/file/$file
 466                         my $dirname = dirname($m->{file_b});
 467                         while ($dirname ne File::Spec->curdir) {
 468                                 if ($dirname ne $m->{file_a}) {
 469                                         $dirname = dirname($dirname);
 470                                         next;
 471                                 }
 472                                 err_file_to_dir("$m->{file_a} => $m->{file_b}");
 473                         }
 474                         # baz/zzz => baz (baz is a file)
 475                         $dirname = dirname($m->{file_a});
 476                         while ($dirname ne File::Spec->curdir) {
 477                                 if ($dirname ne $m->{file_b}) {
 478                                         $dirname = dirname($dirname);
 479                                         next;
 480                                 }
 481                                 err_dir_to_file("$m->{file_a} => $m->{file_b}");
 482                         }
 483                 }
 484                 if ($m->{chg} =~ /^(D|R)$/) {
 485                         my $t = $1 eq 'D' ? 'file_b' : 'file_a';
 486                         $rm_file{ $m->{$t} } = 1;
 487                         my $dirname = dirname( $m->{$t} );
 488                         my $basename = basename( $m->{$t} );
 489                         $rmdir_check{$dirname}->{$basename} = 1;
 490                 } elsif ($m->{chg} =~ /^(?:A|C)$/) {
 491                         if (-d $m->{file_b}) {
 492                                 err_dir_to_file($m->{file_b});
 493                         }
 494                         my $dirname = dirname( $m->{file_b} );
 495                         my $basename = basename( $m->{file_b} );
 496                         $added_check{$dirname}->{$basename} = 1;
 497                         while ($dirname ne File::Spec->curdir) {
 498                                 if ($rm_file{$dirname}) {
 499                                         err_file_to_dir($m->{file_b});
 500                                 }
 501                                 $dirname = dirname $dirname;
 502                         }
 503                 }
 504         }
 505         return (\%rmdir_check, \%added_check);
 506
 507         sub err_dir_to_file {
 508                 my $file = shift;
 509                 print STDERR "Node change from directory to file ",
 510                                 "is not supported by Subversion: ",$file,"\n";
 511                 exit 1;
 512         }
 513         sub err_file_to_dir {
 514                 my $file = shift;
 515                 print STDERR "Node change from file to directory ",
 516                                 "is not supported by Subversion: ",$file,"\n";
 517                 exit 1;
 518         }
 519 }
 520
 521 sub svn_checkout_tree {
 522         my ($svn_rev, $treeish) = @_;
 523         my $from = file_to_s("$REV_DIR/$svn_rev");
 524         assert_svn_wc_clean($svn_rev,$from);
 525         print "diff-tree $from $treeish\n";
 526         my $pid = open my $diff_fh, '-|';
 527         defined $pid or croak $!;
 528         if ($pid == 0) {
 529                 my @diff_tree = qw(git-diff-tree -z -r -C);
 530                 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
 531                 push @diff_tree, "-l$_l" if defined $_l;
 532                 exec(@diff_tree, $from, $treeish) or croak $!;
 533         }
 534         my $mods = parse_diff_tree($diff_fh);
 535         unless (@$mods) {
 536                 # git can do empty commits, but SVN doesn't allow it...
 537                 return $mods;
 538         }
 539         my ($rm, $add) = precommit_check($mods);
 540
 541         my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
 542         foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
 543                 if ($m->{chg} eq 'C') {
 544                         svn_ensure_parent_path( $m->{file_b} );
 545                         sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
 546                         apply_mod_line_blob($m);
 547                         svn_check_prop_executable($m);
 548                 } elsif ($m->{chg} eq 'D') {
 549                         sys(qw(svn rm --force), $m->{file_b});
 550                 } elsif ($m->{chg} eq 'R') {
 551                         svn_ensure_parent_path( $m->{file_b} );
 552                         sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
 553                         apply_mod_line_blob($m);
 554                         svn_check_prop_executable($m);
 555                 } elsif ($m->{chg} eq 'M') {
 556                         apply_mod_line_blob($m);
 557                         svn_check_prop_executable($m);
 558                 } elsif ($m->{chg} eq 'T') {
 559                         sys(qw(svn rm --force),$m->{file_b});
 560                         apply_mod_line_blob($m);
 561                         sys(qw(svn add --force), $m->{file_b});
 562                         svn_check_prop_executable($m);
 563                 } elsif ($m->{chg} eq 'A') {
 564                         svn_ensure_parent_path( $m->{file_b} );
 565                         apply_mod_line_blob($m);
 566                         sys(qw(svn add --force), $m->{file_b});
 567                         svn_check_prop_executable($m);
 568                 } else {
 569                         croak "Invalid chg: $m->{chg}\n";
 570                 }
 571         }
 572
 573         assert_tree($treeish);
 574         if ($_rmdir) { # remove empty directories
 575                 handle_rmdir($rm, $add);
 576         }
 577         assert_tree($treeish);
 578         return $mods;
 579 }
 580
 581 # svn ls doesn't work with respect to the current working tree, but what's
 582 # in the repository.  There's not even an option for it... *sigh*
 583 # (added files don't show up and removed files remain in the ls listing)
 584 sub svn_ls_current {
 585         my ($dir, $rm, $add) = @_;
 586         chomp(my @ls = safe_qx('svn','ls',$dir));
 587         my @ret = ();
 588         foreach (@ls) {
 589                 s#/$##; # trailing slashes are evil
 590                 push @ret, $_ unless $rm->{$dir}->{$_};
 591         }
 592         if (exists $add->{$dir}) {
 593                 push @ret, keys %{$add->{$dir}};
 594         }
 595         return \@ret;
 596 }
 597
 598 sub handle_rmdir {
 599         my ($rm, $add) = @_;
 600
 601         foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
 602                 my $ls = svn_ls_current($dir, $rm, $add);
 603                 next if (scalar @$ls);
 604                 sys(qw(svn rm --force),$dir);
 605
 606                 my $dn = dirname $dir;
 607                 $rm->{ $dn }->{ basename $dir } = 1;
 608                 $ls = svn_ls_current($dn, $rm, $add);
 609                 while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
 610                         sys(qw(svn rm --force),$dn);
 611                         $dir = basename $dn;
 612                         $dn = dirname $dn;
 613                         $rm->{ $dn }->{ $dir } = 1;
 614                         $ls = svn_ls_current($dn, $rm, $add);
 615                 }
 616         }
 617 }
 618
 619 sub svn_commit_tree {
 620         my ($svn_rev, $commit) = @_;
 621         my $commit_msg = "$GIT_DIR/$GIT_SVN/.svn-commit.tmp.$$";
 622         my %log_msg = ( msg => '' );
 623         open my $msg, '>', $commit_msg or croak $!;
 624
 625         chomp(my $type = `git-cat-file -t $commit`);
 626         if ($type eq 'commit') {
 627                 my $pid = open my $msg_fh, '-|';
 628                 defined $pid or croak $!;
 629
 630                 if ($pid == 0) {
 631                         exec(qw(git-cat-file commit), $commit) or croak $!;
 632                 }
 633                 my $in_msg = 0;
 634                 while (<$msg_fh>) {
 635                         if (!$in_msg) {
 636                                 $in_msg = 1 if (/^\s*$/);
 637                         } elsif (/^git-svn-id: /) {
 638                                 # skip this, we regenerate the correct one
 639                                 # on re-fetch anyways
 640                         } else {
 641                                 print $msg $_ or croak $!;
 642                         }
 643                 }
 644                 close $msg_fh or croak $!;
 645         }
 646         close $msg or croak $!;
 647
 648         if ($_edit || ($type eq 'tree')) {
 649                 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 650                 system($editor, $commit_msg);
 651         }
 652
 653         # file_to_s removes all trailing newlines, so just use chomp() here:
 654         open $msg, '<', $commit_msg or croak $!;
 655         { local $/; chomp($log_msg{msg} = <$msg>); }
 656         close $msg or croak $!;
 657
 658         my ($oneline) = ($log_msg{msg} =~ /([^\n\r]+)/);
 659         print "Committing $commit: $oneline\n";
 660
 661         my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
 662         my ($committed) = grep(/^Committed revision \d+\./,@ci_output);
 663         unlink $commit_msg;
 664         defined $committed or croak
 665                         "Commit output failed to parse committed revision!\n",
 666                         join("\n",@ci_output),"\n";
 667         my ($rev_committed) = ($committed =~ /^Committed revision (\d+)\./);
 668
 669         my @svn_up = qw(svn up);
 670         push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 671         if ($rev_committed == ($svn_rev + 1)) {
 672                 push @svn_up, "-r$rev_committed";
 673                 sys(@svn_up);
 674                 my $info = svn_info('.');
 675                 my $date = $info->{'Last Changed Date'} or die "Missing date\n";
 676                 if ($info->{'Last Changed Rev'} != $rev_committed) {
 677                         croak "$info->{'Last Changed Rev'} != $rev_committed\n"
 678                 }
 679                 my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
 680                                         /(\d{4})\-(\d\d)\-(\d\d)\s
 681                                          (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
 682                                          or croak "Failed to parse date: $date\n";
 683                 $log_msg{date} = "$tz $Y-$m-$d $H:$M:$S";
 684                 $log_msg{author} = $info->{'Last Changed Author'};
 685                 $log_msg{revision} = $rev_committed;
 686                 $log_msg{msg} .= "\n";
 687                 my $parent = file_to_s("$REV_DIR/$svn_rev");
 688                 git_commit(\%log_msg, $parent, $commit);
 689                 return $rev_committed;
 690         }
 691         # resync immediately
 692         push @svn_up, "-r$svn_rev";
 693         sys(@svn_up);
 694         return fetch("$rev_committed=$commit")->{revision};
 695 }
 696
 697 sub svn_log_raw {
 698         my (@log_args) = @_;
 699         my $pid = open my $log_fh,'-|';
 700         defined $pid or croak $!;
 701
 702         if ($pid == 0) {
 703                 exec (qw(svn log), @log_args) or croak $!
 704         }
 705
 706         my @svn_log;
 707         my $state = 'sep';
 708         while (<$log_fh>) {
 709                 chomp;
 710                 if (/^\-{72}$/) {
 711                         if ($state eq 'msg') {
 712                                 if ($svn_log[$#svn_log]->{lines}) {
 713                                         $svn_log[$#svn_log]->{msg} .= $_."\n";
 714                                         unless(--$svn_log[$#svn_log]->{lines}) {
 715                                                 $state = 'sep';
 716                                         }
 717                                 } else {
 718                                         croak "Log parse error at: $_\n",
 719                                                 $svn_log[$#svn_log]->{revision},
 720                                                 "\n";
 721                                 }
 722                                 next;
 723                         }
 724                         if ($state ne 'sep') {
 725                                 croak "Log parse error at: $_\n",
 726                                         "state: $state\n",
 727                                         $svn_log[$#svn_log]->{revision},
 728                                         "\n";
 729                         }
 730                         $state = 'rev';
 731
 732                         # if we have an empty log message, put something there:
 733                         if (@svn_log) {
 734                                 $svn_log[$#svn_log]->{msg} ||= "\n";
 735                                 delete $svn_log[$#svn_log]->{lines};
 736                         }
 737                         next;
 738                 }
 739                 if ($state eq 'rev' && s/^r(\d+)\s*\|\s*//) {
 740                         my $rev = $1;
 741                         my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
 742                         ($lines) = ($lines =~ /(\d+)/);
 743                         my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
 744                                         /(\d{4})\-(\d\d)\-(\d\d)\s
 745                                          (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
 746                                          or croak "Failed to parse date: $date\n";
 747                         my %log_msg = ( revision => $rev,
 748                                         date => "$tz $Y-$m-$d $H:$M:$S",
 749                                         author => $author,
 750                                         lines => $lines,
 751                                         msg => '' );
 752                         if (defined $_authors && ! defined $users{$author}) {
 753                                 die "Author: $author not defined in ",
 754                                                 "$_authors file\n";
 755                         }
 756                         push @svn_log, \%log_msg;
 757                         $state = 'msg_start';
 758                         next;
 759                 }
 760                 # skip the first blank line of the message:
 761                 if ($state eq 'msg_start' && /^$/) {
 762                         $state = 'msg';
 763                 } elsif ($state eq 'msg') {
 764                         if ($svn_log[$#svn_log]->{lines}) {
 765                                 $svn_log[$#svn_log]->{msg} .= $_."\n";
 766                                 unless (--$svn_log[$#svn_log]->{lines}) {
 767                                         $state = 'sep';
 768                                 }
 769                         } else {
 770                                 croak "Log parse error at: $_\n",
 771                                         $svn_log[$#svn_log]->{revision},"\n";
 772                         }
 773                 }
 774         }
 775         close $log_fh or croak $?;
 776         return \@svn_log;
 777 }
 778
 779 sub svn_info {
 780         my $url = shift || $SVN_URL;
 781
 782         my $pid = open my $info_fh, '-|';
 783         defined $pid or croak $!;
 784
 785         if ($pid == 0) {
 786                 exec(qw(svn info),$url) or croak $!;
 787         }
 788
 789         my $ret = {};
 790         # only single-lines seem to exist in svn info output
 791         while (<$info_fh>) {
 792                 chomp $_;
 793                 if (m#^([^:]+)\s*:\s*(\S.*)$#) {
 794                         $ret->{$1} = $2;
 795                         push @{$ret->{-order}}, $1;
 796                 }
 797         }
 798         close $info_fh or croak $!;
 799         return $ret;
 800 }
 801
 802 sub sys { system(@_) == 0 or croak $? }
 803
 804 sub git_addremove {
 805         system( "git-diff-files --name-only -z ".
 806                                 " | git-update-index --remove -z --stdin && ".
 807                 "git-ls-files -z --others ".
 808                         "'--exclude-from=$GIT_DIR/$GIT_SVN/info/exclude'".
 809                                 " | git-update-index --add -z --stdin"
 810                 ) == 0 or croak $?
 811 }
 812
 813 sub s_to_file {
 814         my ($str, $file, $mode) = @_;
 815         open my $fd,'>',$file or croak $!;
 816         print $fd $str,"\n" or croak $!;
 817         close $fd or croak $!;
 818         chmod ($mode &~ umask, $file) if (defined $mode);
 819 }
 820
 821 sub file_to_s {
 822         my $file = shift;
 823         open my $fd,'<',$file or croak "$!: file: $file\n";
 824         local $/;
 825         my $ret = <$fd>;
 826         close $fd or croak $!;
 827         $ret =~ s/\s*$//s;
 828         return $ret;
 829 }
 830
 831 sub assert_revision_unknown {
 832         my $revno = shift;
 833         if (-f "$REV_DIR/$revno") {
 834                 croak "$REV_DIR/$revno already exists! ",
 835                                 "Why are we refetching it?";
 836         }
 837 }
 838
 839 sub assert_revision_eq_or_unknown {
 840         my ($revno, $commit) = @_;
 841         if (-f "$REV_DIR/$revno") {
 842                 my $current = file_to_s("$REV_DIR/$revno");
 843                 if ($commit ne $current) {
 844                         croak "$REV_DIR/$revno already exists!\n",
 845                                 "current: $current\nexpected: $commit\n";
 846                 }
 847                 return;
 848         }
 849 }
 850
 851 sub git_commit {
 852         my ($log_msg, @parents) = @_;
 853         assert_revision_unknown($log_msg->{revision});
 854         my $out_fh = IO::File->new_tmpfile or croak $!;
 855         $SVN_UUID ||= svn_info('.')->{'Repository UUID'};
 856
 857         map_tree_joins() if (@_branch_from && !%tree_map);
 858
 859         # commit parents can be conditionally bound to a particular
 860         # svn revision via: "svn_revno=commit_sha1", filter them out here:
 861         my @exec_parents;
 862         foreach my $p (@parents) {
 863                 next unless defined $p;
 864                 if ($p =~ /^(\d+)=($sha1_short)$/o) {
 865                         if ($1 == $log_msg->{revision}) {
 866                                 push @exec_parents, $2;
 867                         }
 868                 } else {
 869                         push @exec_parents, $p if $p =~ /$sha1_short/o;
 870                 }
 871         }
 872
 873         my $pid = fork;
 874         defined $pid or croak $!;
 875         if ($pid == 0) {
 876                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 877                 git_addremove();
 878                 chomp(my $tree = `git-write-tree`);
 879                 croak if $?;
 880                 if (exists $tree_map{$tree}) {
 881                         my %seen_parent = map { $_ => 1 } @exec_parents;
 882                         foreach (@{$tree_map{$tree}}) {
 883                                 # MAXPARENT is defined to 16 in commit-tree.c:
 884                                 if ($seen_parent{$_} || @exec_parents > 16) {
 885                                         next;
 886                                 }
 887                                 push @exec_parents, $_;
 888                                 $seen_parent{$_} = 1;
 889                         }
 890                 }
 891                 my $msg_fh = IO::File->new_tmpfile or croak $!;
 892                 print $msg_fh $log_msg->{msg}, "\ngit-svn-id: ",
 893                                         "$SVN_URL\@$log_msg->{revision}",
 894                                         " $SVN_UUID\n" or croak $!;
 895                 $msg_fh->flush == 0 or croak $!;
 896                 seek $msg_fh, 0, 0 or croak $!;
 897
 898                 set_commit_env($log_msg);
 899
 900                 my @exec = ('git-commit-tree',$tree);
 901                 push @exec, '-p', $_  foreach @exec_parents;
 902                 open STDIN, '<&', $msg_fh or croak $!;
 903                 open STDOUT, '>&', $out_fh or croak $!;
 904                 exec @exec or croak $!;
 905         }
 906         waitpid($pid,0);
 907         croak if $?;
 908
 909         $out_fh->flush == 0 or croak $!;
 910         seek $out_fh, 0, 0 or croak $!;
 911         chomp(my $commit = do { local $/; <$out_fh> });
 912         if ($commit !~ /^$sha1$/o) {
 913                 croak "Failed to commit, invalid sha1: $commit\n";
 914         }
 915         my @update_ref = ('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
 916         if (my $primary_parent = shift @exec_parents) {
 917                 push @update_ref, $primary_parent;
 918         }
 919         sys(@update_ref);
 920         sys('git-update-ref',"$GIT_SVN/revs/$log_msg->{revision}",$commit);
 921         print "r$log_msg->{revision} = $commit\n";
 922         return $commit;
 923 }
 924
 925 sub set_commit_env {
 926         my ($log_msg) = @_;
 927         my $author = $log_msg->{author};
 928         my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
 929                                 : ($author,"$author\@$SVN_UUID");
 930         $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
 931         $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
 932         $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
 933 }
 934
 935 sub apply_mod_line_blob {
 936         my $m = shift;
 937         if ($m->{mode_b} =~ /^120/) {
 938                 blob_to_symlink($m->{sha1_b}, $m->{file_b});
 939         } else {
 940                 blob_to_file($m->{sha1_b}, $m->{file_b});
 941         }
 942 }
 943
 944 sub blob_to_symlink {
 945         my ($blob, $link) = @_;
 946         defined $link or croak "\$link not defined!\n";
 947         croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
 948         if (-l $link || -f _) {
 949                 unlink $link or croak $!;
 950         }
 951
 952         my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
 953         symlink $dest, $link or croak $!;
 954 }
 955
 956 sub blob_to_file {
 957         my ($blob, $file) = @_;
 958         defined $file or croak "\$file not defined!\n";
 959         croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
 960         if (-l $file || -f _) {
 961                 unlink $file or croak $!;
 962         }
 963
 964         open my $blob_fh, '>', $file or croak "$!: $file\n";
 965         my $pid = fork;
 966         defined $pid or croak $!;
 967
 968         if ($pid == 0) {
 969                 open STDOUT, '>&', $blob_fh or croak $!;
 970                 exec('git-cat-file','blob',$blob);
 971         }
 972         waitpid $pid, 0;
 973         croak $? if $?;
 974
 975         close $blob_fh or croak $!;
 976 }
 977
 978 sub safe_qx {
 979         my $pid = open my $child, '-|';
 980         defined $pid or croak $!;
 981         if ($pid == 0) {
 982                 exec(@_) or croak $?;
 983         }
 984         my @ret = (<$child>);
 985         close $child or croak $?;
 986         die $? if $?; # just in case close didn't error out
 987         return wantarray ? @ret : join('',@ret);
 988 }
 989
 990 sub svn_check_ignore_externals {
 991         return if $_no_ignore_ext;
 992         unless (grep /ignore-externals/,(safe_qx(qw(svn co -h)))) {
 993                 print STDERR "W: Installed svn version does not support ",
 994                                 "--ignore-externals\n";
 995                 $_no_ignore_ext = 1;
 996         }
 997 }
 998
 999 sub check_upgrade_needed {
1000         my $old = eval {
1001                 my $pid = open my $child, '-|';
1002                 defined $pid or croak $!;
1003                 if ($pid == 0) {
1004                         close STDERR;
1005                         exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $?;
1006                 }
1007                 my @ret = (<$child>);
1008                 close $child or croak $?;
1009                 die $? if $?; # just in case close didn't error out
1010                 return wantarray ? @ret : join('',@ret);
1011         };
1012         return unless $old;
1013         my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
1014         if ($@ || !$head) {
1015                 print STDERR "Please run: $0 rebuild --upgrade\n";
1016                 exit 1;
1017         }
1018 }
1019
1020 # fills %tree_map with a reverse mapping of trees to commits.  Useful
1021 # for finding parents to commit on.
1022 sub map_tree_joins {
1023         foreach my $br (@_branch_from) {
1024                 my $pid = open my $pipe, '-|';
1025                 defined $pid or croak $!;
1026                 if ($pid == 0) {
1027                         exec(qw(git-rev-list --pretty=raw), $br) or croak $?;
1028                 }
1029                 while (<$pipe>) {
1030                         if (/^commit ($sha1)$/o) {
1031                                 my $commit = $1;
1032                                 my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
1033                                 unless (defined $tree) {
1034                                         die "Failed to parse commit $commit\n";
1035                                 }
1036                                 push @{$tree_map{$tree}}, $commit;
1037                         }
1038                 }
1039                 close $pipe or croak $?;
1040         }
1041 }
1042
1043 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1044 sub load_authors {
1045         open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1046         while (<$authors>) {
1047                 chomp;
1048                 next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
1049                 my ($user, $name, $email) = ($1, $2, $3);
1050                 $users{$user} = [$name, $email];
1051         }
1052         close $authors or croak $!;
1053 }
1054
1055 __END__
1056
1057 Data structures:
1058
1059 @svn_log = array of log_msg hashes
1060
1061 $log_msg hash
1062 {
1063         msg => 'whitespace-formatted log entry
1064 ',                                              # trailing newline is preserved
1065         revision => '8',                        # integer
1066         date => '2004-02-24T17:01:44.108345Z',  # commit date
1067         author => 'committer name'
1068 };
1069
1070
1071 @mods = array of diff-index line hashes, each element represents one line
1072         of diff-index output
1073
1074 diff-index line ($m hash)
1075 {
1076         mode_a => first column of diff-index output, no leading ':',
1077         mode_b => second column of diff-index output,
1078         sha1_b => sha1sum of the final blob,
1079         chg => change type [MCRADT],
1080         file_a => original file name of a file (iff chg is 'C' or 'R')
1081         file_b => new/current file name of a file (any chg)
1082 }
1083 ;