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 $sha1_short = qr/[a-f\d]{6,40}/;
  36 my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
  37         $_find_copies_harder, $_l, $_version);
  38
  39 GetOptions(     'revision|r=s' => \$_revision,
  40                 'no-ignore-externals' => \$_no_ignore_ext,
  41                 'stdin|' => \$_stdin,
  42                 'edit|e' => \$_edit,
  43                 'rmdir' => \$_rmdir,
  44                 'help|H|h' => \$_help,
  45                 'find-copies-harder' => \$_find_copies_harder,
  46                 'l=i' => \$_l,
  47                 'version|V' => \$_version,
  48                 'no-stop-on-copy' => \$_no_stop_copy );
  49 my %cmd = (
  50         fetch => [ \&fetch, "Download new revisions from SVN" ],
  51         init => [ \&init, "Initialize and fetch (import)"],
  52         commit => [ \&commit, "Commit git revisions to SVN" ],
  53         'show-ignore' => [ \&show_ignore, "Show svn:ignore listings" ],
  54         rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)" ],
  55         help => [ \&usage, "Show help" ],
  56 );
  57 my $cmd;
  58 for (my $i = 0; $i < @ARGV; $i++) {
  59         if (defined $cmd{$ARGV[$i]}) {
  60                 $cmd = $ARGV[$i];
  61                 splice @ARGV, $i, 1;
  62                 last;
  63         }
  64 };
  65
  66 # we may be called as git-svn-(command), or git-svn(command).
  67 foreach (keys %cmd) {
  68         if (/git\-svn\-?($_)(?:\.\w+)?$/) {
  69                 $cmd = $1;
  70                 last;
  71         }
  72 }
  73 usage(0) if $_help;
  74 version() if $_version;
  75 usage(1) unless (defined $cmd);
  76 svn_check_ignore_externals();
  77 $cmd{$cmd}->[0]->(@ARGV);
  78 exit 0;
  79
  80 ####################### primary functions ######################
  81 sub usage {
  82         my $exit = shift || 0;
  83         my $fd = $exit ? \*STDERR : \*STDOUT;
  84         print $fd <<"";
  85 git-svn - bidirectional operations between a single Subversion tree and git
  86 Usage: $0 <command> [options] [arguments]\n
  87 Available commands:
  88
  89         foreach (sort keys %cmd) {
  90                 print $fd '  ',pack('A10',$_),$cmd{$_}->[1],"\n";
  91         }
  92         print $fd <<"";
  93 \nGIT_SVN_ID may be set in the environment to an arbitrary identifier if
  94 you're tracking multiple SVN branches/repositories in one git repository
  95 and want to keep them separate.
  96
  97         exit $exit;
  98 }
  99
 100 sub version {
 101         print "git-svn version $VERSION\n";
 102         exit 0;
 103 }
 104
 105 sub rebuild {
 106         $SVN_URL = shift or undef;
 107         my $repo_uuid;
 108         my $newest_rev = 0;
 109
 110         my $pid = open(my $rev_list,'-|');
 111         defined $pid or croak $!;
 112         if ($pid == 0) {
 113                 exec("git-rev-list","$GIT_SVN-HEAD") or croak $!;
 114         }
 115         my $first;
 116         while (<$rev_list>) {
 117                 chomp;
 118                 my $c = $_;
 119                 croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 120                 my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
 121                 next if (!@commit); # skip merges
 122                 my $id = $commit[$#commit];
 123                 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 124                                                 \s([a-f\d\-]+)$/x);
 125                 if (!$rev || !$uuid || !$url) {
 126                         # some of the original repositories I made had
 127                         # indentifiers like this:
 128                         ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)
 129                                                         \@([a-f\d\-]+)/x);
 130                         if (!$rev || !$uuid) {
 131                                 croak "Unable to extract revision or UUID from ",
 132                                         "$c, $id\n";
 133                         }
 134                 }
 135                 print "r$rev = $c\n";
 136                 unless (defined $first) {
 137                         if (!$SVN_URL && !$url) {
 138                                 croak "SVN repository location required: $url\n";
 139                         }
 140                         $SVN_URL ||= $url;
 141                         $repo_uuid = setup_git_svn();
 142                         $first = $rev;
 143                 }
 144                 if ($uuid ne $repo_uuid) {
 145                         croak "Repository UUIDs do not match!\ngot: $uuid\n",
 146                                                 "expected: $repo_uuid\n";
 147                 }
 148                 assert_revision_eq_or_unknown($rev, $c);
 149                 sys('git-update-ref',"$GIT_SVN/revs/$rev",$c);
 150                 $newest_rev = $rev if ($rev > $newest_rev);
 151         }
 152         close $rev_list or croak $?;
 153         if (!chdir $SVN_WC) {
 154                 my @svn_co = ('svn','co',"-r$first");
 155                 push @svn_co, '--ignore-externals' unless $_no_ignore_ext;
 156                 sys(@svn_co, $SVN_URL, $SVN_WC);
 157                 chdir $SVN_WC or croak $!;
 158         }
 159
 160         $pid = fork;
 161         defined $pid or croak $!;
 162         if ($pid == 0) {
 163                 my @svn_up = qw(svn up);
 164                 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 165                 sys(@svn_up,"-r$newest_rev");
 166                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 167                 git_addremove();
 168                 exec('git-write-tree');
 169         }
 170         waitpid $pid, 0;
 171 }
 172
 173 sub init {
 174         $SVN_URL = shift or croak "SVN repository location required\n";
 175         unless (-d $GIT_DIR) {
 176                 sys('git-init-db');
 177         }
 178         setup_git_svn();
 179 }
 180
 181 sub fetch {
 182         my (@parents) = @_;
 183         $SVN_URL ||= file_to_s("$GIT_DIR/$GIT_SVN/info/url");
 184         my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 185         unless ($_revision) {
 186                 $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 187         }
 188         push @log_args, "-r$_revision";
 189         push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 190
 191         my $svn_log = svn_log_raw(@log_args);
 192         @$svn_log = sort { $a->{revision} <=> $b->{revision} } @$svn_log;
 193
 194         my $base = shift @$svn_log or croak "No base revision!\n";
 195         my $last_commit = undef;
 196         unless (-d $SVN_WC) {
 197                 my @svn_co = ('svn','co',"-r$base->{revision}");
 198                 push @svn_co,'--ignore-externals' unless $_no_ignore_ext;
 199                 sys(@svn_co, $SVN_URL, $SVN_WC);
 200                 chdir $SVN_WC or croak $!;
 201                 $last_commit = git_commit($base, @parents);
 202                 unless (-f "$GIT_DIR/refs/heads/master") {
 203                         sys(qw(git-update-ref refs/heads/master),$last_commit);
 204                 }
 205                 assert_svn_wc_clean($base->{revision}, $last_commit);
 206         } else {
 207                 chdir $SVN_WC or croak $!;
 208                 $last_commit = file_to_s("$REV_DIR/$base->{revision}");
 209         }
 210         my @svn_up = qw(svn up);
 211         push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 212         my $last_rev = $base->{revision};
 213         foreach my $log_msg (@$svn_log) {
 214                 assert_svn_wc_clean($last_rev, $last_commit);
 215                 $last_rev = $log_msg->{revision};
 216                 sys(@svn_up,"-r$last_rev");
 217                 $last_commit = git_commit($log_msg, $last_commit, @parents);
 218         }
 219         assert_svn_wc_clean($last_rev, $last_commit);
 220         return pop @$svn_log;
 221 }
 222
 223 sub commit {
 224         my (@commits) = @_;
 225         if ($_stdin || !@commits) {
 226                 print "Reading from stdin...\n";
 227                 @commits = ();
 228                 while (<STDIN>) {
 229                         if (/\b([a-f\d]{6,40})\b/) {
 230                                 unshift @commits, $1;
 231                         }
 232                 }
 233         }
 234         my @revs;
 235         foreach my $c (@commits) {
 236                 chomp(my @tmp = safe_qx('git-rev-parse',$c));
 237                 if (scalar @tmp == 1) {
 238                         push @revs, $tmp[0];
 239                 } elsif (scalar @tmp > 1) {
 240                         push @revs, reverse (safe_qx('git-rev-list',@tmp));
 241                 } else {
 242                         die "Failed to rev-parse $c\n";
 243                 }
 244         }
 245         chomp @revs;
 246
 247         fetch();
 248         chdir $SVN_WC or croak $!;
 249         my $svn_current_rev =  svn_info('.')->{'Last Changed Rev'};
 250         foreach my $c (@revs) {
 251                 print "Committing $c\n";
 252                 my $mods = svn_checkout_tree($svn_current_rev, $c);
 253                 if (scalar @$mods == 0) {
 254                         print "Skipping, no changes detected\n";
 255                         next;
 256                 }
 257                 $svn_current_rev = svn_commit_tree($svn_current_rev, $c);
 258         }
 259         print "Done committing ",scalar @revs," revisions to SVN\n";
 260
 261 }
 262
 263 sub show_ignore {
 264         require File::Find or die $!;
 265         my $exclude_file = "$GIT_DIR/info/exclude";
 266         open my $fh, '<', $exclude_file or croak $!;
 267         chomp(my @excludes = (<$fh>));
 268         close $fh or croak $!;
 269
 270         $SVN_URL ||= file_to_s("$GIT_DIR/$GIT_SVN/info/url");
 271         chdir $SVN_WC or croak $!;
 272         my %ign;
 273         File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 274                 s#^\./##;
 275                 @{$ign{$_}} = safe_qx(qw(svn propget svn:ignore),$_);
 276                 }}, no_chdir=>1},'.');
 277
 278         print "\n# /\n";
 279         foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 280         delete $ign{'.'};
 281         foreach my $i (sort keys %ign) {
 282                 print "\n# ",$i,"\n";
 283                 foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 284         }
 285 }
 286
 287 ########################### utility functions #########################
 288
 289 sub setup_git_svn {
 290         defined $SVN_URL or croak "SVN repository location required\n";
 291         unless (-d $GIT_DIR) {
 292                 croak "GIT_DIR=$GIT_DIR does not exist!\n";
 293         }
 294         mkpath(["$GIT_DIR/$GIT_SVN"]);
 295         mkpath(["$GIT_DIR/$GIT_SVN/info"]);
 296         mkpath([$REV_DIR]);
 297         s_to_file($SVN_URL,"$GIT_DIR/$GIT_SVN/info/url");
 298         my $uuid = svn_info($SVN_URL)->{'Repository UUID'} or
 299                                         croak "Repository UUID unreadable\n";
 300         s_to_file($uuid,"$GIT_DIR/$GIT_SVN/info/uuid");
 301
 302         open my $fd, '>>', "$GIT_DIR/$GIT_SVN/info/exclude" or croak $!;
 303         print $fd '.svn',"\n";
 304         close $fd or croak $!;
 305         return $uuid;
 306 }
 307
 308 sub assert_svn_wc_clean {
 309         my ($svn_rev, $treeish) = @_;
 310         croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
 311         croak "$treeish is not a sha1!\n" unless ($treeish =~ /^$sha1$/o);
 312         my $svn_info = svn_info('.');
 313         if ($svn_rev != $svn_info->{'Last Changed Rev'}) {
 314                 croak "Expected r$svn_rev, got r",
 315                                 $svn_info->{'Last Changed Rev'},"\n";
 316         }
 317         my @status = grep(!/^Performing status on external/,(`svn status`));
 318         @status = grep(!/^\s*$/,@status);
 319         if (scalar @status) {
 320                 print STDERR "Tree ($SVN_WC) is not clean:\n";
 321                 print STDERR $_ foreach @status;
 322                 croak;
 323         }
 324         assert_tree($treeish);
 325 }
 326
 327 sub assert_tree {
 328         my ($treeish) = @_;
 329         croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
 330         chomp(my $type = `git-cat-file -t $treeish`);
 331         my $expected;
 332         while ($type eq 'tag') {
 333                 chomp(($treeish, $type) = `git-cat-file tag $treeish`);
 334         }
 335         if ($type eq 'commit') {
 336                 $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
 337                 ($expected) = ($expected =~ /^tree ($sha1)$/);
 338                 die "Unable to get tree from $treeish\n" unless $expected;
 339         } elsif ($type eq 'tree') {
 340                 $expected = $treeish;
 341         } else {
 342                 die "$treeish is a $type, expected tree, tag or commit\n";
 343         }
 344
 345         my $old_index = $ENV{GIT_INDEX_FILE};
 346         my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
 347         if (-e $tmpindex) {
 348                 unlink $tmpindex or croak $!;
 349         }
 350         $ENV{GIT_INDEX_FILE} = $tmpindex;
 351         git_addremove();
 352         chomp(my $tree = `git-write-tree`);
 353         if ($old_index) {
 354                 $ENV{GIT_INDEX_FILE} = $old_index;
 355         } else {
 356                 delete $ENV{GIT_INDEX_FILE};
 357         }
 358         if ($tree ne $expected) {
 359                 croak "Tree mismatch, Got: $tree, Expected: $expected\n";
 360         }
 361 }
 362
 363 sub parse_diff_tree {
 364         my $diff_fh = shift;
 365         local $/ = "\0";
 366         my $state = 'meta';
 367         my @mods;
 368         while (<$diff_fh>) {
 369                 chomp $_; # this gets rid of the trailing "\0"
 370                 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
 371                                         $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
 372                         push @mods, {   mode_a => $1, mode_b => $2,
 373                                         sha1_b => $3, chg => $4 };
 374                         if ($4 =~ /^(?:C|R)$/) {
 375                                 $state = 'file_a';
 376                         } else {
 377                                 $state = 'file_b';
 378                         }
 379                 } elsif ($state eq 'file_a') {
 380                         my $x = $mods[$#mods] or croak "Empty array\n";
 381                         if ($x->{chg} !~ /^(?:C|R)$/) {
 382                                 croak "Error parsing $_, $x->{chg}\n";
 383                         }
 384                         $x->{file_a} = $_;
 385                         $state = 'file_b';
 386                 } elsif ($state eq 'file_b') {
 387                         my $x = $mods[$#mods] or croak "Empty array\n";
 388                         if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
 389                                 croak "Error parsing $_, $x->{chg}\n";
 390                         }
 391                         if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
 392                                 croak "Error parsing $_, $x->{chg}\n";
 393                         }
 394                         $x->{file_b} = $_;
 395                         $state = 'meta';
 396                 } else {
 397                         croak "Error parsing $_\n";
 398                 }
 399         }
 400         close $diff_fh or croak $!;
 401
 402         return \@mods;
 403 }
 404
 405 sub svn_check_prop_executable {
 406         my $m = shift;
 407         return if -l $m->{file_b};
 408         if ($m->{mode_b} =~ /755$/) {
 409                 chmod((0755 &~ umask),$m->{file_b}) or croak $!;
 410                 if ($m->{mode_a} !~ /755$/) {
 411                         sys(qw(svn propset svn:executable 1), $m->{file_b});
 412                 }
 413                 -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
 414         } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
 415                 sys(qw(svn propdel svn:executable), $m->{file_b});
 416                 chmod((0644 &~ umask),$m->{file_b}) or croak $!;
 417                 -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
 418         }
 419 }
 420
 421 sub svn_ensure_parent_path {
 422         my $dir_b = dirname(shift);
 423         svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
 424         mkpath([$dir_b]) unless (-d $dir_b);
 425         sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
 426 }
 427
 428 sub precommit_check {
 429         my $mods = shift;
 430         my (%rm_file, %rmdir_check, %added_check);
 431
 432         my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
 433         foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
 434                 if ($m->{chg} eq 'R') {
 435                         if (-d $m->{file_b}) {
 436                                 err_dir_to_file("$m->{file_a} => $m->{file_b}");
 437                         }
 438                         # dir/$file => dir/file/$file
 439                         my $dirname = dirname($m->{file_b});
 440                         while ($dirname ne File::Spec->curdir) {
 441                                 if ($dirname ne $m->{file_a}) {
 442                                         $dirname = dirname($dirname);
 443                                         next;
 444                                 }
 445                                 err_file_to_dir("$m->{file_a} => $m->{file_b}");
 446                         }
 447                         # baz/zzz => baz (baz is a file)
 448                         $dirname = dirname($m->{file_a});
 449                         while ($dirname ne File::Spec->curdir) {
 450                                 if ($dirname ne $m->{file_b}) {
 451                                         $dirname = dirname($dirname);
 452                                         next;
 453                                 }
 454                                 err_dir_to_file("$m->{file_a} => $m->{file_b}");
 455                         }
 456                 }
 457                 if ($m->{chg} =~ /^(D|R)$/) {
 458                         my $t = $1 eq 'D' ? 'file_b' : 'file_a';
 459                         $rm_file{ $m->{$t} } = 1;
 460                         my $dirname = dirname( $m->{$t} );
 461                         my $basename = basename( $m->{$t} );
 462                         $rmdir_check{$dirname}->{$basename} = 1;
 463                 } elsif ($m->{chg} =~ /^(?:A|C)$/) {
 464                         if (-d $m->{file_b}) {
 465                                 err_dir_to_file($m->{file_b});
 466                         }
 467                         my $dirname = dirname( $m->{file_b} );
 468                         my $basename = basename( $m->{file_b} );
 469                         $added_check{$dirname}->{$basename} = 1;
 470                         while ($dirname ne File::Spec->curdir) {
 471                                 if ($rm_file{$dirname}) {
 472                                         err_file_to_dir($m->{file_b});
 473                                 }
 474                                 $dirname = dirname $dirname;
 475                         }
 476                 }
 477         }
 478         return (\%rmdir_check, \%added_check);
 479
 480         sub err_dir_to_file {
 481                 my $file = shift;
 482                 print STDERR "Node change from directory to file ",
 483                                 "is not supported by Subversion: ",$file,"\n";
 484                 exit 1;
 485         }
 486         sub err_file_to_dir {
 487                 my $file = shift;
 488                 print STDERR "Node change from file to directory ",
 489                                 "is not supported by Subversion: ",$file,"\n";
 490                 exit 1;
 491         }
 492 }
 493
 494 sub svn_checkout_tree {
 495         my ($svn_rev, $treeish) = @_;
 496         my $from = file_to_s("$REV_DIR/$svn_rev");
 497         assert_svn_wc_clean($svn_rev,$from);
 498         print "diff-tree '$from' '$treeish'\n";
 499         my $pid = open my $diff_fh, '-|';
 500         defined $pid or croak $!;
 501         if ($pid == 0) {
 502                 my @diff_tree = qw(git-diff-tree -z -r -C);
 503                 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
 504                 push @diff_tree, "-l$_l" if defined $_l;
 505                 exec(@diff_tree, $from, $treeish) or croak $!;
 506         }
 507         my $mods = parse_diff_tree($diff_fh);
 508         unless (@$mods) {
 509                 # git can do empty commits, SVN doesn't allow it...
 510                 return $mods;
 511         }
 512         my ($rm, $add) = precommit_check($mods);
 513
 514         my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
 515         foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
 516                 if ($m->{chg} eq 'C') {
 517                         svn_ensure_parent_path( $m->{file_b} );
 518                         sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
 519                         apply_mod_line_blob($m);
 520                         svn_check_prop_executable($m);
 521                 } elsif ($m->{chg} eq 'D') {
 522                         sys(qw(svn rm --force), $m->{file_b});
 523                 } elsif ($m->{chg} eq 'R') {
 524                         svn_ensure_parent_path( $m->{file_b} );
 525                         sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
 526                         apply_mod_line_blob($m);
 527                         svn_check_prop_executable($m);
 528                 } elsif ($m->{chg} eq 'M') {
 529                         apply_mod_line_blob($m);
 530                         svn_check_prop_executable($m);
 531                 } elsif ($m->{chg} eq 'T') {
 532                         sys(qw(svn rm --force),$m->{file_b});
 533                         apply_mod_line_blob($m);
 534                         sys(qw(svn add --force), $m->{file_b});
 535                         svn_check_prop_executable($m);
 536                 } elsif ($m->{chg} eq 'A') {
 537                         svn_ensure_parent_path( $m->{file_b} );
 538                         apply_mod_line_blob($m);
 539                         sys(qw(svn add --force), $m->{file_b});
 540                         svn_check_prop_executable($m);
 541                 } else {
 542                         croak "Invalid chg: $m->{chg}\n";
 543                 }
 544         }
 545
 546         assert_tree($treeish);
 547         if ($_rmdir) { # remove empty directories
 548                 handle_rmdir($rm, $add);
 549         }
 550         assert_tree($treeish);
 551         return $mods;
 552 }
 553
 554 # svn ls doesn't work with respect to the current working tree, but what's
 555 # in the repository.  There's not even an option for it... *sigh*
 556 # (added files don't show up and removed files remain in the ls listing)
 557 sub svn_ls_current {
 558         my ($dir, $rm, $add) = @_;
 559         chomp(my @ls = safe_qx('svn','ls',$dir));
 560         my @ret = ();
 561         foreach (@ls) {
 562                 s#/$##; # trailing slashes are evil
 563                 push @ret, $_ unless $rm->{$dir}->{$_};
 564         }
 565         if (exists $add->{$dir}) {
 566                 push @ret, keys %{$add->{$dir}};
 567         }
 568         return \@ret;
 569 }
 570
 571 sub handle_rmdir {
 572         my ($rm, $add) = @_;
 573
 574         foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
 575                 my $ls = svn_ls_current($dir, $rm, $add);
 576                 next if (scalar @$ls);
 577                 sys(qw(svn rm --force),$dir);
 578
 579                 my $dn = dirname $dir;
 580                 $rm->{ $dn }->{ basename $dir } = 1;
 581                 $ls = svn_ls_current($dn, $rm, $add);
 582                 while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
 583                         sys(qw(svn rm --force),$dn);
 584                         $dir = basename $dn;
 585                         $dn = dirname $dn;
 586                         $rm->{ $dn }->{ $dir } = 1;
 587                         $ls = svn_ls_current($dn, $rm, $add);
 588                 }
 589         }
 590 }
 591
 592 sub svn_commit_tree {
 593         my ($svn_rev, $commit) = @_;
 594         my $commit_msg = "$GIT_DIR/$GIT_SVN/.svn-commit.tmp.$$";
 595         my %log_msg = ( msg => '' );
 596         open my $msg, '>', $commit_msg  or croak $!;
 597
 598         chomp(my $type = `git-cat-file -t $commit`);
 599         if ($type eq 'commit') {
 600                 my $pid = open my $msg_fh, '-|';
 601                 defined $pid or croak $!;
 602
 603                 if ($pid == 0) {
 604                         exec(qw(git-cat-file commit), $commit) or croak $!;
 605                 }
 606                 my $in_msg = 0;
 607                 while (<$msg_fh>) {
 608                         if (!$in_msg) {
 609                                 $in_msg = 1 if (/^\s*$/);
 610                         } else {
 611                                 $log_msg{msg} .= $_;
 612                                 print $msg $_ or croak $!;
 613                         }
 614                 }
 615                 close $msg_fh or croak $!;
 616         }
 617         close $msg or croak $!;
 618
 619         if ($_edit || ($type eq 'tree')) {
 620                 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 621                 system($editor, $commit_msg);
 622         }
 623         my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
 624         my ($committed) = grep(/^Committed revision \d+\./,@ci_output);
 625         unlink $commit_msg;
 626         defined $committed or croak
 627                         "Commit output failed to parse committed revision!\n",
 628                         join("\n",@ci_output),"\n";
 629         my ($rev_committed) = ($committed =~ /^Committed revision (\d+)\./);
 630
 631         my @svn_up = qw(svn up);
 632         push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 633         if ($rev_committed == ($svn_rev + 1)) {
 634                 push @svn_up, "-r$rev_committed";
 635                 sys(@svn_up);
 636                 my $info = svn_info('.');
 637                 my $date = $info->{'Last Changed Date'} or die "Missing date\n";
 638                 if ($info->{'Last Changed Rev'} != $rev_committed) {
 639                         croak "$info->{'Last Changed Rev'} != $rev_committed\n"
 640                 }
 641                 my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
 642                                         /(\d{4})\-(\d\d)\-(\d\d)\s
 643                                          (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
 644                                          or croak "Failed to parse date: $date\n";
 645                 $log_msg{date} = "$tz $Y-$m-$d $H:$M:$S";
 646                 $log_msg{author} = $info->{'Last Changed Author'};
 647                 $log_msg{revision} = $rev_committed;
 648                 $log_msg{msg} .= "\n";
 649                 my $parent = file_to_s("$REV_DIR/$svn_rev");
 650                 git_commit(\%log_msg, $parent, $commit);
 651                 return $rev_committed;
 652         }
 653         # resync immediately
 654         push @svn_up, "-r$svn_rev";
 655         sys(@svn_up);
 656         return fetch("$rev_committed=$commit")->{revision};
 657 }
 658
 659 sub svn_log_raw {
 660         my (@log_args) = @_;
 661         my $pid = open my $log_fh,'-|';
 662         defined $pid or croak $!;
 663
 664         if ($pid == 0) {
 665                 exec (qw(svn log), @log_args) or croak $!
 666         }
 667
 668         my @svn_log;
 669         my $state = 'sep';
 670         while (<$log_fh>) {
 671                 chomp;
 672                 if (/^\-{72}$/) {
 673                         if ($state eq 'msg') {
 674                                 if ($svn_log[$#svn_log]->{lines}) {
 675                                         $svn_log[$#svn_log]->{msg} .= $_."\n";
 676                                         unless(--$svn_log[$#svn_log]->{lines}) {
 677                                                 $state = 'sep';
 678                                         }
 679                                 } else {
 680                                         croak "Log parse error at: $_\n",
 681                                                 $svn_log[$#svn_log]->{revision},
 682                                                 "\n";
 683                                 }
 684                                 next;
 685                         }
 686                         if ($state ne 'sep') {
 687                                 croak "Log parse error at: $_\n",
 688                                         "state: $state\n",
 689                                         $svn_log[$#svn_log]->{revision},
 690                                         "\n";
 691                         }
 692                         $state = 'rev';
 693
 694                         # if we have an empty log message, put something there:
 695                         if (@svn_log) {
 696                                 $svn_log[$#svn_log]->{msg} ||= "\n";
 697                                 delete $svn_log[$#svn_log]->{lines};
 698                         }
 699                         next;
 700                 }
 701                 if ($state eq 'rev' && s/^r(\d+)\s*\|\s*//) {
 702                         my $rev = $1;
 703                         my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
 704                         ($lines) = ($lines =~ /(\d+)/);
 705                         my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
 706                                         /(\d{4})\-(\d\d)\-(\d\d)\s
 707                                          (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
 708                                          or croak "Failed to parse date: $date\n";
 709                         my %log_msg = ( revision => $rev,
 710                                         date => "$tz $Y-$m-$d $H:$M:$S",
 711                                         author => $author,
 712                                         lines => $lines,
 713                                         msg => '' );
 714                         push @svn_log, \%log_msg;
 715                         $state = 'msg_start';
 716                         next;
 717                 }
 718                 # skip the first blank line of the message:
 719                 if ($state eq 'msg_start' && /^$/) {
 720                         $state = 'msg';
 721                 } elsif ($state eq 'msg') {
 722                         if ($svn_log[$#svn_log]->{lines}) {
 723                                 $svn_log[$#svn_log]->{msg} .= $_."\n";
 724                                 unless (--$svn_log[$#svn_log]->{lines}) {
 725                                         $state = 'sep';
 726                                 }
 727                         } else {
 728                                 croak "Log parse error at: $_\n",
 729                                         $svn_log[$#svn_log]->{revision},"\n";
 730                         }
 731                 }
 732         }
 733         close $log_fh or croak $?;
 734         return \@svn_log;
 735 }
 736
 737 sub svn_info {
 738         my $url = shift || $SVN_URL;
 739
 740         my $pid = open my $info_fh, '-|';
 741         defined $pid or croak $!;
 742
 743         if ($pid == 0) {
 744                 exec(qw(svn info),$url) or croak $!;
 745         }
 746
 747         my $ret = {};
 748         # only single-lines seem to exist in svn info output
 749         while (<$info_fh>) {
 750                 chomp $_;
 751                 if (m#^([^:]+)\s*:\s*(\S.*)$#) {
 752                         $ret->{$1} = $2;
 753                         push @{$ret->{-order}}, $1;
 754                 }
 755         }
 756         close $info_fh or croak $!;
 757         return $ret;
 758 }
 759
 760 sub sys { system(@_) == 0 or croak $? }
 761
 762 sub git_addremove {
 763         system( "git-diff-files --name-only -z ".
 764                                 " | git-update-index --remove -z --stdin && ".
 765                 "git-ls-files -z --others ".
 766                         "'--exclude-from=$GIT_DIR/$GIT_SVN/info/exclude'".
 767                                 " | git-update-index --add -z --stdin"
 768                 ) == 0 or croak $?
 769 }
 770
 771 sub s_to_file {
 772         my ($str, $file, $mode) = @_;
 773         open my $fd,'>',$file or croak $!;
 774         print $fd $str,"\n" or croak $!;
 775         close $fd or croak $!;
 776         chmod ($mode &~ umask, $file) if (defined $mode);
 777 }
 778
 779 sub file_to_s {
 780         my $file = shift;
 781         open my $fd,'<',$file or croak "$!: file: $file\n";
 782         local $/;
 783         my $ret = <$fd>;
 784         close $fd or croak $!;
 785         $ret =~ s/\s*$//s;
 786         return $ret;
 787 }
 788
 789 sub assert_revision_unknown {
 790         my $revno = shift;
 791         if (-f "$REV_DIR/$revno") {
 792                 croak "$REV_DIR/$revno already exists! ",
 793                                 "Why are we refetching it?";
 794         }
 795 }
 796
 797 sub assert_revision_eq_or_unknown {
 798         my ($revno, $commit) = @_;
 799         if (-f "$REV_DIR/$revno") {
 800                 my $current = file_to_s("$REV_DIR/$revno");
 801                 if ($commit ne $current) {
 802                         croak "$REV_DIR/$revno already exists!\n",
 803                                 "current: $current\nexpected: $commit\n";
 804                 }
 805                 return;
 806         }
 807 }
 808
 809 sub git_commit {
 810         my ($log_msg, @parents) = @_;
 811         assert_revision_unknown($log_msg->{revision});
 812         my $out_fh = IO::File->new_tmpfile or croak $!;
 813         my $info = svn_info('.');
 814         my $uuid = $info->{'Repository UUID'};
 815         defined $uuid or croak "Unable to get Repository UUID\n";
 816
 817         # commit parents can be conditionally bound to a particular
 818         # svn revision via: "svn_revno=commit_sha1", filter them out here:
 819         my @exec_parents;
 820         foreach my $p (@parents) {
 821                 next unless defined $p;
 822                 if ($p =~ /^(\d+)=($sha1_short)$/o) {
 823                         if ($1 == $log_msg->{revision}) {
 824                                 push @exec_parents, $2;
 825                         }
 826                 } else {
 827                         push @exec_parents, $p if $p =~ /$sha1_short/o;
 828                 }
 829         }
 830
 831         my $pid = fork;
 832         defined $pid or croak $!;
 833         if ($pid == 0) {
 834                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 835                 git_addremove();
 836                 chomp(my $tree = `git-write-tree`);
 837                 croak if $?;
 838                 my $msg_fh = IO::File->new_tmpfile or croak $!;
 839                 print $msg_fh $log_msg->{msg}, "\ngit-svn-id: ",
 840                                         "$SVN_URL\@$log_msg->{revision}",
 841                                         " $uuid\n" or croak $!;
 842                 $msg_fh->flush == 0 or croak $!;
 843                 seek $msg_fh, 0, 0 or croak $!;
 844
 845                 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} =
 846                                                 $log_msg->{author};
 847                 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
 848                                                 $log_msg->{author}."\@$uuid";
 849                 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} =
 850                                                 $log_msg->{date};
 851                 my @exec = ('git-commit-tree',$tree);
 852                 push @exec, '-p', $_  foreach @exec_parents;
 853                 open STDIN, '<&', $msg_fh or croak $!;
 854                 open STDOUT, '>&', $out_fh or croak $!;
 855                 exec @exec or croak $!;
 856         }
 857         waitpid($pid,0);
 858         croak if $?;
 859
 860         $out_fh->flush == 0 or croak $!;
 861         seek $out_fh, 0, 0 or croak $!;
 862         chomp(my $commit = do { local $/; <$out_fh> });
 863         if ($commit !~ /^$sha1$/o) {
 864                 croak "Failed to commit, invalid sha1: $commit\n";
 865         }
 866         my @update_ref = ('git-update-ref',"refs/heads/$GIT_SVN-HEAD",$commit);
 867         if (my $primary_parent = shift @exec_parents) {
 868                 push @update_ref, $primary_parent;
 869         }
 870         sys(@update_ref);
 871         sys('git-update-ref',"$GIT_SVN/revs/$log_msg->{revision}",$commit);
 872         print "r$log_msg->{revision} = $commit\n";
 873         return $commit;
 874 }
 875
 876 sub apply_mod_line_blob {
 877         my $m = shift;
 878         if ($m->{mode_b} =~ /^120/) {
 879                 blob_to_symlink($m->{sha1_b}, $m->{file_b});
 880         } else {
 881                 blob_to_file($m->{sha1_b}, $m->{file_b});
 882         }
 883 }
 884
 885 sub blob_to_symlink {
 886         my ($blob, $link) = @_;
 887         defined $link or croak "\$link not defined!\n";
 888         croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
 889         if (-l $link || -f _) {
 890                 unlink $link or croak $!;
 891         }
 892
 893         my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
 894         symlink $dest, $link or croak $!;
 895 }
 896
 897 sub blob_to_file {
 898         my ($blob, $file) = @_;
 899         defined $file or croak "\$file not defined!\n";
 900         croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
 901         if (-l $file || -f _) {
 902                 unlink $file or croak $!;
 903         }
 904
 905         open my $blob_fh, '>', $file or croak "$!: $file\n";
 906         my $pid = fork;
 907         defined $pid or croak $!;
 908
 909         if ($pid == 0) {
 910                 open STDOUT, '>&', $blob_fh or croak $!;
 911                 exec('git-cat-file','blob',$blob);
 912         }
 913         waitpid $pid, 0;
 914         croak $? if $?;
 915
 916         close $blob_fh or croak $!;
 917 }
 918
 919 sub safe_qx {
 920         my $pid = open my $child, '-|';
 921         defined $pid or croak $!;
 922         if ($pid == 0) {
 923                 exec(@_) or croak $?;
 924         }
 925         my @ret = (<$child>);
 926         close $child or croak $?;
 927         die $? if $?; # just in case close didn't error out
 928         return wantarray ? @ret : join('',@ret);
 929 }
 930
 931 sub svn_check_ignore_externals {
 932         return if $_no_ignore_ext;
 933         unless (grep /ignore-externals/,(safe_qx(qw(svn co -h)))) {
 934                 print STDERR "W: Installed svn version does not support ",
 935                                 "--ignore-externals\n";
 936                 $_no_ignore_ext = 1;
 937         }
 938 }
 939 __END__
 940
 941 Data structures:
 942
 943 @svn_log = array of log_msg hashes
 944
 945 $log_msg hash
 946 {
 947         msg => 'whitespace-formatted log entry
 948 ',                                              # trailing newline is preserved
 949         revision => '8',                        # integer
 950         date => '2004-02-24T17:01:44.108345Z',  # commit date
 951         author => 'committer name'
 952 };
 953
 954
 955 @mods = array of diff-index line hashes, each element represents one line
 956         of diff-index output
 957
 958 diff-index line ($m hash)
 959 {
 960         mode_a => first column of diff-index output, no leading ':',
 961         mode_b => second column of diff-index output,
 962         sha1_b => sha1sum of the final blob,
 963         chg => change type [MCRAD],
 964         file_a => original file name of a file (iff chg is 'C' or 'R')
 965         file_b => new/current file name of a file (any chg)
 966 }
 967 ;