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