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