contrib/git-svn/git-svn.perl

   1 #!/usr/bin/env perl
   2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3 # License: GPL v2 or later
   4 use warnings;
   5 use strict;
   6 use vars qw/    $AUTHOR $VERSION
   7                 $SVN_URL $SVN_INFO $SVN_WC $SVN_UUID
   8                 $GIT_SVN_INDEX $GIT_SVN
   9                 $GIT_DIR $REV_DIR/;
  10 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  11 $VERSION = '0.11.0';
  12 $GIT_DIR = $ENV{GIT_DIR} || "$ENV{PWD}/.git";
  13 # make sure the svn binary gives consistent output between locales and TZs:
  14 $ENV{TZ} = 'UTC';
  15 $ENV{LC_ALL} = 'C';
  16
  17 # If SVN:: library support is added, please make the dependencies
  18 # optional and preserve the capability to use the command-line client.
  19 # use eval { require SVN::... } to make it lazy load
  20 # We don't use any modules not in the standard Perl distribution:
  21 use Carp qw/croak/;
  22 use IO::File qw//;
  23 use File::Basename qw/dirname basename/;
  24 use File::Path qw/mkpath/;
  25 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  26 use File::Spec qw//;
  27 use POSIX qw/strftime/;
  28 my $sha1 = qr/[a-f\d]{40}/;
  29 my $sha1_short = qr/[a-f\d]{4,40}/;
  30 my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
  31         $_find_copies_harder, $_l, $_version, $_upgrade, $_authors);
  32 my (@_branch_from, %tree_map, %users);
  33 my $_svn_co_url_revs;
  34
  35 my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
  36                 'branch|b=s' => \@_branch_from,
  37                 'authors-file|A=s' => \$_authors );
  38 my %cmd = (
  39         fetch => [ \&fetch, "Download new revisions from SVN",
  40                         { 'revision|r=s' => \$_revision, %fc_opts } ],
  41         init => [ \&init, "Initialize and fetch (import)", { } ],
  42         commit => [ \&commit, "Commit git revisions to SVN",
  43                         {       'stdin|' => \$_stdin,
  44                                 'edit|e' => \$_edit,
  45                                 'rmdir' => \$_rmdir,
  46                                 'find-copies-harder' => \$_find_copies_harder,
  47                                 'l=i' => \$_l,
  48                                 %fc_opts,
  49                         } ],
  50         'show-ignore' => [ \&show_ignore, "Show svn:ignore listings", { } ],
  51         rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
  52                         { 'no-ignore-externals' => \$_no_ignore_ext,
  53                           'upgrade' => \$_upgrade } ],
  54 );
  55 my $cmd;
  56 for (my $i = 0; $i < @ARGV; $i++) {
  57         if (defined $cmd{$ARGV[$i]}) {
  58                 $cmd = $ARGV[$i];
  59                 splice @ARGV, $i, 1;
  60                 last;
  61         }
  62 };
  63
  64 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
  65
  66 GetOptions(%opts, 'help|H|h' => \$_help,
  67                 'version|V' => \$_version,
  68                 'id|i=s' => \$GIT_SVN) or exit 1;
  69
  70 $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
  71 $GIT_SVN_INDEX = "$GIT_DIR/$GIT_SVN/index";
  72 $ENV{GIT_DIR} ||= $GIT_DIR;
  73 $SVN_URL = undef;
  74 $REV_DIR = "$GIT_DIR/$GIT_SVN/revs";
  75 $SVN_WC = "$GIT_DIR/$GIT_SVN/tree";
  76
  77 usage(0) if $_help;
  78 version() if $_version;
  79 usage(1) unless defined $cmd;
  80 load_authors() if $_authors;
  81 svn_compat_check();
  82 $cmd{$cmd}->[0]->(@ARGV);
  83 exit 0;
  84
  85 ####################### primary functions ######################
  86 sub usage {
  87         my $exit = shift || 0;
  88         my $fd = $exit ? \*STDERR : \*STDOUT;
  89         print $fd <<"";
  90 git-svn - bidirectional operations between a single Subversion tree and git
  91 Usage: $0 <command> [options] [arguments]\n
  92
  93         print $fd "Available commands:\n" unless $cmd;
  94
  95         foreach (sort keys %cmd) {
  96                 next if $cmd && $cmd ne $_;
  97                 print $fd '  ',pack('A13',$_),$cmd{$_}->[1],"\n";
  98                 foreach (keys %{$cmd{$_}->[2]}) {
  99                         # prints out arguments as they should be passed:
 100                         my $x = s#=s$## ? '<arg>' : s#=i$## ? '<num>' : '';
 101                         print $fd ' ' x 17, join(', ', map { length $_ > 1 ?
 102                                                         "--$_" : "-$_" }
 103                                                 split /\|/,$_)," $x\n";
 104                 }
 105         }
 106         print $fd <<"";
 107 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 108 arbitrary identifier if you're tracking multiple SVN branches/repositories in
 109 one git repository and want to keep them separate.  See git-svn(1) for more
 110 information.
 111
 112         exit $exit;
 113 }
 114
 115 sub version {
 116         print "git-svn version $VERSION\n";
 117         exit 0;
 118 }
 119
 120 sub rebuild {
 121         $SVN_URL = shift or undef;
 122         my $newest_rev = 0;
 123         if ($_upgrade) {
 124                 sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD");
 125         } else {
 126                 check_upgrade_needed();
 127         }
 128
 129         my $pid = open(my $rev_list,'-|');
 130         defined $pid or croak $!;
 131         if ($pid == 0) {
 132                 exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!;
 133         }
 134         my $latest;
 135         while (<$rev_list>) {
 136                 chomp;
 137                 my $c = $_;
 138                 croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 139                 my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
 140                 next if (!@commit); # skip merges
 141                 my $id = $commit[$#commit];
 142                 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 143                                                 \s([a-f\d\-]+)$/x);
 144                 if (!$rev || !$uuid || !$url) {
 145                         # some of the original repositories I made had
 146                         # indentifiers like this:
 147                         ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)
 148                                                         \@([a-f\d\-]+)/x);
 149                         if (!$rev || !$uuid) {
 150                                 croak "Unable to extract revision or UUID from ",
 151                                         "$c, $id\n";
 152                         }
 153                 }
 154
 155                 # if we merged or otherwise started elsewhere, this is
 156                 # how we break out of it
 157                 next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
 158                 next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
 159
 160                 print "r$rev = $c\n";
 161                 unless (defined $latest) {
 162                         if (!$SVN_URL && !$url) {
 163                                 croak "SVN repository location required: $url\n";
 164                         }
 165                         $SVN_URL ||= $url;
 166                         $SVN_UUID ||= $uuid;
 167                         setup_git_svn();
 168                         $latest = $rev;
 169                 }
 170                 assert_revision_eq_or_unknown($rev, $c);
 171                 sys('git-update-ref',"$GIT_SVN/revs/$rev",$c);
 172                 $newest_rev = $rev if ($rev > $newest_rev);
 173         }
 174         close $rev_list or croak $?;
 175         if (!chdir $SVN_WC) {
 176                 svn_cmd_checkout($SVN_URL, $latest, $SVN_WC);
 177                 chdir $SVN_WC or croak $!;
 178         }
 179
 180         $pid = fork;
 181         defined $pid or croak $!;
 182         if ($pid == 0) {
 183                 my @svn_up = qw(svn up);
 184                 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 185                 sys(@svn_up,"-r$newest_rev");
 186                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 187                 git_addremove();
 188                 exec('git-write-tree');
 189         }
 190         waitpid $pid, 0;
 191
 192         if ($_upgrade) {
 193                 print STDERR <<"";
 194 Keeping deprecated refs/head/$GIT_SVN-HEAD for now.  Please remove it
 195 when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN
 196
 197         }
 198 }
 199
 200 sub init {
 201         $SVN_URL = shift or croak "SVN repository location required\n";
 202         unless (-d $GIT_DIR) {
 203                 sys('git-init-db');
 204         }
 205         setup_git_svn();
 206 }
 207
 208 sub fetch {
 209         my (@parents) = @_;
 210         check_upgrade_needed();
 211         $SVN_URL ||= file_to_s("$GIT_DIR/$GIT_SVN/info/url");
 212         my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 213         unless ($_revision) {
 214                 $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 215         }
 216         push @log_args, "-r$_revision";
 217         push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 218
 219         my $svn_log = svn_log_raw(@log_args);
 220
 221         my $base = next_log_entry($svn_log) or croak "No base revision!\n";
 222         my $last_commit = undef;
 223         unless (-d $SVN_WC) {
 224                 svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC);
 225                 chdir $SVN_WC or croak $!;
 226                 read_uuid();
 227                 $last_commit = git_commit($base, @parents);
 228                 assert_svn_wc_clean($base->{revision}, $last_commit);
 229         } else {
 230                 chdir $SVN_WC or croak $!;
 231                 read_uuid();
 232                 $last_commit = file_to_s("$REV_DIR/$base->{revision}");
 233         }
 234         my @svn_up = qw(svn up);
 235         push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 236         my $last = $base;
 237         while (my $log_msg = next_log_entry($svn_log)) {
 238                 assert_svn_wc_clean($last->{revision}, $last_commit);
 239                 if ($last->{revision} >= $log_msg->{revision}) {
 240                         croak "Out of order: last >= current: ",
 241                                 "$last->{revision} >= $log_msg->{revision}\n";
 242                 }
 243                 sys(@svn_up,"-r$log_msg->{revision}");
 244                 $last_commit = git_commit($log_msg, $last_commit, @parents);
 245                 $last = $log_msg;
 246         }
 247         assert_svn_wc_clean($last->{revision}, $last_commit);
 248         unless (-e "$GIT_DIR/refs/heads/master") {
 249                 sys(qw(git-update-ref refs/heads/master),$last_commit);
 250         }
 251         return $last;
 252 }
 253
 254 sub commit {
 255         my (@commits) = @_;
 256         check_upgrade_needed();
 257         if ($_stdin || !@commits) {
 258                 print "Reading from stdin...\n";
 259                 @commits = ();
 260                 while (<STDIN>) {
 261                         if (/\b($sha1_short)\b/o) {
 262                                 unshift @commits, $1;
 263                         }
 264                 }
 265         }
 266         my @revs;
 267         foreach my $c (@commits) {
 268                 chomp(my @tmp = safe_qx('git-rev-parse',$c));
 269                 if (scalar @tmp == 1) {
 270                         push @revs, $tmp[0];
 271                 } elsif (scalar @tmp > 1) {
 272                         push @revs, reverse (safe_qx('git-rev-list',@tmp));
 273                 } else {
 274                         die "Failed to rev-parse $c\n";
 275                 }
 276         }
 277         chomp @revs;
 278
 279         fetch();
 280         chdir $SVN_WC or croak $!;
 281         my $info = svn_info('.');
 282         read_uuid($info);
 283         my $svn_current_rev =  $info->{'Last Changed Rev'};
 284         foreach my $c (@revs) {
 285                 my $mods = svn_checkout_tree($svn_current_rev, $c);
 286                 if (scalar @$mods == 0) {
 287                         print "Skipping, no changes detected\n";
 288                         next;
 289                 }
 290                 $svn_current_rev = svn_commit_tree($svn_current_rev, $c);
 291         }
 292         print "Done committing ",scalar @revs," revisions to SVN\n";
 293
 294 }
 295
 296 sub show_ignore {
 297         require File::Find or die $!;
 298         my $exclude_file = "$GIT_DIR/info/exclude";
 299         open my $fh, '<', $exclude_file or croak $!;
 300         chomp(my @excludes = (<$fh>));
 301         close $fh or croak $!;
 302
 303         $SVN_URL ||= file_to_s("$GIT_DIR/$GIT_SVN/info/url");
 304         chdir $SVN_WC or croak $!;
 305         my %ign;
 306         File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 307                 s#^\./##;
 308                 @{$ign{$_}} = safe_qx(qw(svn propget svn:ignore),$_);
 309                 }}, no_chdir=>1},'.');
 310
 311         print "\n# /\n";
 312         foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 313         delete $ign{'.'};
 314         foreach my $i (sort keys %ign) {
 315                 print "\n# ",$i,"\n";
 316                 foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 317         }
 318 }
 319
 320 ########################### utility functions #########################
 321
 322 sub read_uuid {
 323         return if $SVN_UUID;
 324         my $info = shift || svn_info('.');
 325         $SVN_UUID = $info->{'Repository UUID'} or
 326                                         croak "Repository UUID unreadable\n";
 327         s_to_file($SVN_UUID,"$GIT_DIR/$GIT_SVN/info/uuid");
 328 }
 329
 330 sub setup_git_svn {
 331         defined $SVN_URL or croak "SVN repository location required\n";
 332         unless (-d $GIT_DIR) {
 333                 croak "GIT_DIR=$GIT_DIR does not exist!\n";
 334         }
 335         mkpath(["$GIT_DIR/$GIT_SVN"]);
 336         mkpath(["$GIT_DIR/$GIT_SVN/info"]);
 337         mkpath([$REV_DIR]);
 338         s_to_file($SVN_URL,"$GIT_DIR/$GIT_SVN/info/url");
 339
 340         open my $fd, '>>', "$GIT_DIR/$GIT_SVN/info/exclude" or croak $!;
 341         print $fd '.svn',"\n";
 342         close $fd or croak $!;
 343 }
 344
 345 sub assert_svn_wc_clean {
 346         my ($svn_rev, $treeish) = @_;
 347         croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
 348         croak "$treeish is not a sha1!\n" unless ($treeish =~ /^$sha1$/o);
 349         my $lcr = svn_info('.')->{'Last Changed Rev'};
 350         if ($svn_rev != $lcr) {
 351                 print STDERR "Checking for copy-tree ... ";
 352                 # use
 353                 my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
 354                                                 "-r$lcr:$svn_rev")));
 355                 if (@diff) {
 356                         croak "Nope!  Expected r$svn_rev, got r$lcr\n";
 357                 } else {
 358                         print STDERR "OK!\n";
 359                 }
 360         }
 361         my @status = grep(!/^Performing status on external/,(`svn status`));
 362         @status = grep(!/^\s*$/,@status);
 363         if (scalar @status) {
 364                 print STDERR "Tree ($SVN_WC) is not clean:\n";
 365                 print STDERR $_ foreach @status;
 366                 croak;
 367         }
 368         assert_tree($treeish);
 369 }
 370
 371 sub assert_tree {
 372         my ($treeish) = @_;
 373         croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
 374         chomp(my $type = `git-cat-file -t $treeish`);
 375         my $expected;
 376         while ($type eq 'tag') {
 377                 chomp(($treeish, $type) = `git-cat-file tag $treeish`);
 378         }
 379         if ($type eq 'commit') {
 380                 $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
 381                 ($expected) = ($expected =~ /^tree ($sha1)$/);
 382                 die "Unable to get tree from $treeish\n" unless $expected;
 383         } elsif ($type eq 'tree') {
 384                 $expected = $treeish;
 385         } else {
 386                 die "$treeish is a $type, expected tree, tag or commit\n";
 387         }
 388
 389         my $old_index = $ENV{GIT_INDEX_FILE};
 390         my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
 391         if (-e $tmpindex) {
 392                 unlink $tmpindex or croak $!;
 393         }
 394         $ENV{GIT_INDEX_FILE} = $tmpindex;
 395         git_addremove();
 396         chomp(my $tree = `git-write-tree`);
 397         if ($old_index) {
 398                 $ENV{GIT_INDEX_FILE} = $old_index;
 399         } else {
 400                 delete $ENV{GIT_INDEX_FILE};
 401         }
 402         if ($tree ne $expected) {
 403                 croak "Tree mismatch, Got: $tree, Expected: $expected\n";
 404         }
 405 }
 406
 407 sub parse_diff_tree {
 408         my $diff_fh = shift;
 409         local $/ = "\0";
 410         my $state = 'meta';
 411         my @mods;
 412         while (<$diff_fh>) {
 413                 chomp $_; # this gets rid of the trailing "\0"
 414                 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
 415                                         $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
 416                         push @mods, {   mode_a => $1, mode_b => $2,
 417                                         sha1_b => $3, chg => $4 };
 418                         if ($4 =~ /^(?:C|R)$/) {
 419                                 $state = 'file_a';
 420                         } else {
 421                                 $state = 'file_b';
 422                         }
 423                 } elsif ($state eq 'file_a') {
 424                         my $x = $mods[$#mods] or croak "Empty array\n";
 425                         if ($x->{chg} !~ /^(?:C|R)$/) {
 426                                 croak "Error parsing $_, $x->{chg}\n";
 427                         }
 428                         $x->{file_a} = $_;
 429                         $state = 'file_b';
 430                 } elsif ($state eq 'file_b') {
 431                         my $x = $mods[$#mods] or croak "Empty array\n";
 432                         if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
 433                                 croak "Error parsing $_, $x->{chg}\n";
 434                         }
 435                         if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
 436                                 croak "Error parsing $_, $x->{chg}\n";
 437                         }
 438                         $x->{file_b} = $_;
 439                         $state = 'meta';
 440                 } else {
 441                         croak "Error parsing $_\n";
 442                 }
 443         }
 444         close $diff_fh or croak $!;
 445
 446         return \@mods;
 447 }
 448
 449 sub svn_check_prop_executable {
 450         my $m = shift;
 451         return if -l $m->{file_b};
 452         if ($m->{mode_b} =~ /755$/) {
 453                 chmod((0755 &~ umask),$m->{file_b}) or croak $!;
 454                 if ($m->{mode_a} !~ /755$/) {
 455                         sys(qw(svn propset svn:executable 1), $m->{file_b});
 456                 }
 457                 -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
 458         } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
 459                 sys(qw(svn propdel svn:executable), $m->{file_b});
 460                 chmod((0644 &~ umask),$m->{file_b}) or croak $!;
 461                 -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
 462         }
 463 }
 464
 465 sub svn_ensure_parent_path {
 466         my $dir_b = dirname(shift);
 467         svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
 468         mkpath([$dir_b]) unless (-d $dir_b);
 469         sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
 470 }
 471
 472 sub precommit_check {
 473         my $mods = shift;
 474         my (%rm_file, %rmdir_check, %added_check);
 475
 476         my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
 477         foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
 478                 if ($m->{chg} eq 'R') {
 479                         if (-d $m->{file_b}) {
 480                                 err_dir_to_file("$m->{file_a} => $m->{file_b}");
 481                         }
 482                         # dir/$file => dir/file/$file
 483                         my $dirname = dirname($m->{file_b});
 484                         while ($dirname ne File::Spec->curdir) {
 485                                 if ($dirname ne $m->{file_a}) {
 486                                         $dirname = dirname($dirname);
 487                                         next;
 488                                 }
 489                                 err_file_to_dir("$m->{file_a} => $m->{file_b}");
 490                         }
 491                         # baz/zzz => baz (baz is a file)
 492                         $dirname = dirname($m->{file_a});
 493                         while ($dirname ne File::Spec->curdir) {
 494                                 if ($dirname ne $m->{file_b}) {
 495                                         $dirname = dirname($dirname);
 496                                         next;
 497                                 }
 498                                 err_dir_to_file("$m->{file_a} => $m->{file_b}");
 499                         }
 500                 }
 501                 if ($m->{chg} =~ /^(D|R)$/) {
 502                         my $t = $1 eq 'D' ? 'file_b' : 'file_a';
 503                         $rm_file{ $m->{$t} } = 1;
 504                         my $dirname = dirname( $m->{$t} );
 505                         my $basename = basename( $m->{$t} );
 506                         $rmdir_check{$dirname}->{$basename} = 1;
 507                 } elsif ($m->{chg} =~ /^(?:A|C)$/) {
 508                         if (-d $m->{file_b}) {
 509                                 err_dir_to_file($m->{file_b});
 510                         }
 511                         my $dirname = dirname( $m->{file_b} );
 512                         my $basename = basename( $m->{file_b} );
 513                         $added_check{$dirname}->{$basename} = 1;
 514                         while ($dirname ne File::Spec->curdir) {
 515                                 if ($rm_file{$dirname}) {
 516                                         err_file_to_dir($m->{file_b});
 517                                 }
 518                                 $dirname = dirname $dirname;
 519                         }
 520                 }
 521         }
 522         return (\%rmdir_check, \%added_check);
 523
 524         sub err_dir_to_file {
 525                 my $file = shift;
 526                 print STDERR "Node change from directory to file ",
 527                                 "is not supported by Subversion: ",$file,"\n";
 528                 exit 1;
 529         }
 530         sub err_file_to_dir {
 531                 my $file = shift;
 532                 print STDERR "Node change from file to directory ",
 533                                 "is not supported by Subversion: ",$file,"\n";
 534                 exit 1;
 535         }
 536 }
 537
 538 sub svn_checkout_tree {
 539         my ($svn_rev, $treeish) = @_;
 540         my $from = file_to_s("$REV_DIR/$svn_rev");
 541         assert_svn_wc_clean($svn_rev,$from);
 542         print "diff-tree $from $treeish\n";
 543         my $pid = open my $diff_fh, '-|';
 544         defined $pid or croak $!;
 545         if ($pid == 0) {
 546                 my @diff_tree = qw(git-diff-tree -z -r -C);
 547                 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
 548                 push @diff_tree, "-l$_l" if defined $_l;
 549                 exec(@diff_tree, $from, $treeish) or croak $!;
 550         }
 551         my $mods = parse_diff_tree($diff_fh);
 552         unless (@$mods) {
 553                 # git can do empty commits, but SVN doesn't allow it...
 554                 return $mods;
 555         }
 556         my ($rm, $add) = precommit_check($mods);
 557
 558         my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
 559         foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
 560                 if ($m->{chg} eq 'C') {
 561                         svn_ensure_parent_path( $m->{file_b} );
 562                         sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
 563                         apply_mod_line_blob($m);
 564                         svn_check_prop_executable($m);
 565                 } elsif ($m->{chg} eq 'D') {
 566                         sys(qw(svn rm --force), $m->{file_b});
 567                 } elsif ($m->{chg} eq 'R') {
 568                         svn_ensure_parent_path( $m->{file_b} );
 569                         sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
 570                         apply_mod_line_blob($m);
 571                         svn_check_prop_executable($m);
 572                 } elsif ($m->{chg} eq 'M') {
 573                         apply_mod_line_blob($m);
 574                         svn_check_prop_executable($m);
 575                 } elsif ($m->{chg} eq 'T') {
 576                         sys(qw(svn rm --force),$m->{file_b});
 577                         apply_mod_line_blob($m);
 578                         sys(qw(svn add --force), $m->{file_b});
 579                         svn_check_prop_executable($m);
 580                 } elsif ($m->{chg} eq 'A') {
 581                         svn_ensure_parent_path( $m->{file_b} );
 582                         apply_mod_line_blob($m);
 583                         sys(qw(svn add --force), $m->{file_b});
 584                         svn_check_prop_executable($m);
 585                 } else {
 586                         croak "Invalid chg: $m->{chg}\n";
 587                 }
 588         }
 589
 590         assert_tree($treeish);
 591         if ($_rmdir) { # remove empty directories
 592                 handle_rmdir($rm, $add);
 593         }
 594         assert_tree($treeish);
 595         return $mods;
 596 }
 597
 598 # svn ls doesn't work with respect to the current working tree, but what's
 599 # in the repository.  There's not even an option for it... *sigh*
 600 # (added files don't show up and removed files remain in the ls listing)
 601 sub svn_ls_current {
 602         my ($dir, $rm, $add) = @_;
 603         chomp(my @ls = safe_qx('svn','ls',$dir));
 604         my @ret = ();
 605         foreach (@ls) {
 606                 s#/$##; # trailing slashes are evil
 607                 push @ret, $_ unless $rm->{$dir}->{$_};
 608         }
 609         if (exists $add->{$dir}) {
 610                 push @ret, keys %{$add->{$dir}};
 611         }
 612         return \@ret;
 613 }
 614
 615 sub handle_rmdir {
 616         my ($rm, $add) = @_;
 617
 618         foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
 619                 my $ls = svn_ls_current($dir, $rm, $add);
 620                 next if (scalar @$ls);
 621                 sys(qw(svn rm --force),$dir);
 622
 623                 my $dn = dirname $dir;
 624                 $rm->{ $dn }->{ basename $dir } = 1;
 625                 $ls = svn_ls_current($dn, $rm, $add);
 626                 while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
 627                         sys(qw(svn rm --force),$dn);
 628                         $dir = basename $dn;
 629                         $dn = dirname $dn;
 630                         $rm->{ $dn }->{ $dir } = 1;
 631                         $ls = svn_ls_current($dn, $rm, $add);
 632                 }
 633         }
 634 }
 635
 636 sub svn_commit_tree {
 637         my ($svn_rev, $commit) = @_;
 638         my $commit_msg = "$GIT_DIR/$GIT_SVN/.svn-commit.tmp.$$";
 639         my %log_msg = ( msg => '' );
 640         open my $msg, '>', $commit_msg or croak $!;
 641
 642         chomp(my $type = `git-cat-file -t $commit`);
 643         if ($type eq 'commit') {
 644                 my $pid = open my $msg_fh, '-|';
 645                 defined $pid or croak $!;
 646
 647                 if ($pid == 0) {
 648                         exec(qw(git-cat-file commit), $commit) or croak $!;
 649                 }
 650                 my $in_msg = 0;
 651                 while (<$msg_fh>) {
 652                         if (!$in_msg) {
 653                                 $in_msg = 1 if (/^\s*$/);
 654                         } elsif (/^git-svn-id: /) {
 655                                 # skip this, we regenerate the correct one
 656                                 # on re-fetch anyways
 657                         } else {
 658                                 print $msg $_ or croak $!;
 659                         }
 660                 }
 661                 close $msg_fh or croak $!;
 662         }
 663         close $msg or croak $!;
 664
 665         if ($_edit || ($type eq 'tree')) {
 666                 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 667                 system($editor, $commit_msg);
 668         }
 669
 670         # file_to_s removes all trailing newlines, so just use chomp() here:
 671         open $msg, '<', $commit_msg or croak $!;
 672         { local $/; chomp($log_msg{msg} = <$msg>); }
 673         close $msg or croak $!;
 674
 675         my ($oneline) = ($log_msg{msg} =~ /([^\n\r]+)/);
 676         print "Committing $commit: $oneline\n";
 677
 678         my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
 679         my ($committed) = grep(/^Committed revision \d+\./,@ci_output);
 680         unlink $commit_msg;
 681         defined $committed or croak
 682                         "Commit output failed to parse committed revision!\n",
 683                         join("\n",@ci_output),"\n";
 684         my ($rev_committed) = ($committed =~ /^Committed revision (\d+)\./);
 685
 686         my @svn_up = qw(svn up);
 687         push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 688         if ($rev_committed == ($svn_rev + 1)) {
 689                 push @svn_up, "-r$rev_committed";
 690                 sys(@svn_up);
 691                 my $info = svn_info('.');
 692                 my $date = $info->{'Last Changed Date'} or die "Missing date\n";
 693                 if ($info->{'Last Changed Rev'} != $rev_committed) {
 694                         croak "$info->{'Last Changed Rev'} != $rev_committed\n"
 695                 }
 696                 my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
 697                                         /(\d{4})\-(\d\d)\-(\d\d)\s
 698                                          (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
 699                                          or croak "Failed to parse date: $date\n";
 700                 $log_msg{date} = "$tz $Y-$m-$d $H:$M:$S";
 701                 $log_msg{author} = $info->{'Last Changed Author'};
 702                 $log_msg{revision} = $rev_committed;
 703                 $log_msg{msg} .= "\n";
 704                 my $parent = file_to_s("$REV_DIR/$svn_rev");
 705                 git_commit(\%log_msg, $parent, $commit);
 706                 return $rev_committed;
 707         }
 708         # resync immediately
 709         push @svn_up, "-r$svn_rev";
 710         sys(@svn_up);
 711         return fetch("$rev_committed=$commit")->{revision};
 712 }
 713
 714 # read the entire log into a temporary file (which is removed ASAP)
 715 # and store the file handle + parser state
 716 sub svn_log_raw {
 717         my (@log_args) = @_;
 718         my $log_fh = IO::File->new_tmpfile or croak $!;
 719         my $pid = fork;
 720         defined $pid or croak $!;
 721         if (!$pid) {
 722                 open STDOUT, '>&', $log_fh or croak $!;
 723                 exec (qw(svn log), @log_args) or croak $!
 724         }
 725         waitpid $pid, 0;
 726         croak if $?;
 727         seek $log_fh, 0, 0 or croak $!;
 728         return { state => 'sep', fh => $log_fh };
 729 }
 730
 731 sub next_log_entry {
 732         my $log = shift; # retval of svn_log_raw()
 733         my $ret = undef;
 734         my $fh = $log->{fh};
 735
 736         while (<$fh>) {
 737                 chomp;
 738                 if (/^\-{72}$/) {
 739                         if ($log->{state} eq 'msg') {
 740                                 if ($ret->{lines}) {
 741                                         $ret->{msg} .= $_."\n";
 742                                         unless(--$ret->{lines}) {
 743                                                 $log->{state} = 'sep';
 744                                         }
 745                                 } else {
 746                                         croak "Log parse error at: $_\n",
 747                                                 $ret->{revision},
 748                                                 "\n";
 749                                 }
 750                                 next;
 751                         }
 752                         if ($log->{state} ne 'sep') {
 753                                 croak "Log parse error at: $_\n",
 754                                         "state: $log->{state}\n",
 755                                         $ret->{revision},
 756                                         "\n";
 757                         }
 758                         $log->{state} = 'rev';
 759
 760                         # if we have an empty log message, put something there:
 761                         if ($ret) {
 762                                 $ret->{msg} ||= "\n";
 763                                 delete $ret->{lines};
 764                                 return $ret;
 765                         }
 766                         next;
 767                 }
 768                 if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
 769                         my $rev = $1;
 770                         my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
 771                         ($lines) = ($lines =~ /(\d+)/);
 772                         my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
 773                                         /(\d{4})\-(\d\d)\-(\d\d)\s
 774                                          (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
 775                                          or croak "Failed to parse date: $date\n";
 776                         $ret = {        revision => $rev,
 777                                         date => "$tz $Y-$m-$d $H:$M:$S",
 778                                         author => $author,
 779                                         lines => $lines,
 780                                         msg => '' };
 781                         if (defined $_authors && ! defined $users{$author}) {
 782                                 die "Author: $author not defined in ",
 783                                                 "$_authors file\n";
 784                         }
 785                         $log->{state} = 'msg_start';
 786                         next;
 787                 }
 788                 # skip the first blank line of the message:
 789                 if ($log->{state} eq 'msg_start' && /^$/) {
 790                         $log->{state} = 'msg';
 791                 } elsif ($log->{state} eq 'msg') {
 792                         if ($ret->{lines}) {
 793                                 $ret->{msg} .= $_."\n";
 794                                 unless (--$ret->{lines}) {
 795                                         $log->{state} = 'sep';
 796                                 }
 797                         } else {
 798                                 croak "Log parse error at: $_\n",
 799                                         $ret->{revision},"\n";
 800                         }
 801                 }
 802         }
 803         return $ret;
 804 }
 805
 806 sub svn_info {
 807         my $url = shift || $SVN_URL;
 808
 809         my $pid = open my $info_fh, '-|';
 810         defined $pid or croak $!;
 811
 812         if ($pid == 0) {
 813                 exec(qw(svn info),$url) or croak $!;
 814         }
 815
 816         my $ret = {};
 817         # only single-lines seem to exist in svn info output
 818         while (<$info_fh>) {
 819                 chomp $_;
 820                 if (m#^([^:]+)\s*:\s*(\S.*)$#) {
 821                         $ret->{$1} = $2;
 822                         push @{$ret->{-order}}, $1;
 823                 }
 824         }
 825         close $info_fh or croak $!;
 826         return $ret;
 827 }
 828
 829 sub sys { system(@_) == 0 or croak $? }
 830
 831 sub git_addremove {
 832         system( "git-diff-files --name-only -z ".
 833                                 " | git-update-index --remove -z --stdin && ".
 834                 "git-ls-files -z --others ".
 835                         "'--exclude-from=$GIT_DIR/$GIT_SVN/info/exclude'".
 836                                 " | git-update-index --add -z --stdin"
 837                 ) == 0 or croak $?
 838 }
 839
 840 sub s_to_file {
 841         my ($str, $file, $mode) = @_;
 842         open my $fd,'>',$file or croak $!;
 843         print $fd $str,"\n" or croak $!;
 844         close $fd or croak $!;
 845         chmod ($mode &~ umask, $file) if (defined $mode);
 846 }
 847
 848 sub file_to_s {
 849         my $file = shift;
 850         open my $fd,'<',$file or croak "$!: file: $file\n";
 851         local $/;
 852         my $ret = <$fd>;
 853         close $fd or croak $!;
 854         $ret =~ s/\s*$//s;
 855         return $ret;
 856 }
 857
 858 sub assert_revision_unknown {
 859         my $revno = shift;
 860         if (-f "$REV_DIR/$revno") {
 861                 croak "$REV_DIR/$revno already exists! ",
 862                                 "Why are we refetching it?";
 863         }
 864 }
 865
 866 sub trees_eq {
 867         my ($x, $y) = @_;
 868         my @x = safe_qx('git-cat-file','commit',$x);
 869         my @y = safe_qx('git-cat-file','commit',$y);
 870         if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
 871                                 || $y[0] !~ /^tree $sha1\n$/) {
 872                 print STDERR "Trees not equal: $y[0] != $x[0]\n";
 873                 return 0
 874         }
 875         return 1;
 876 }
 877
 878 sub assert_revision_eq_or_unknown {
 879         my ($revno, $commit) = @_;
 880         if (-f "$REV_DIR/$revno") {
 881                 my $current = file_to_s("$REV_DIR/$revno");
 882                 if (($commit ne $current) && !trees_eq($commit, $current)) {
 883                         croak "$REV_DIR/$revno already exists!\n",
 884                                 "current: $current\nexpected: $commit\n";
 885                 }
 886                 return;
 887         }
 888 }
 889
 890 sub git_commit {
 891         my ($log_msg, @parents) = @_;
 892         assert_revision_unknown($log_msg->{revision});
 893         my $out_fh = IO::File->new_tmpfile or croak $!;
 894
 895         map_tree_joins() if (@_branch_from && !%tree_map);
 896
 897         # commit parents can be conditionally bound to a particular
 898         # svn revision via: "svn_revno=commit_sha1", filter them out here:
 899         my @exec_parents;
 900         foreach my $p (@parents) {
 901                 next unless defined $p;
 902                 if ($p =~ /^(\d+)=($sha1_short)$/o) {
 903                         if ($1 == $log_msg->{revision}) {
 904                                 push @exec_parents, $2;
 905                         }
 906                 } else {
 907                         push @exec_parents, $p if $p =~ /$sha1_short/o;
 908                 }
 909         }
 910
 911         my $pid = fork;
 912         defined $pid or croak $!;
 913         if ($pid == 0) {
 914                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 915                 git_addremove();
 916                 chomp(my $tree = `git-write-tree`);
 917                 croak if $?;
 918                 if (exists $tree_map{$tree}) {
 919                         my %seen_parent = map { $_ => 1 } @exec_parents;
 920                         foreach (@{$tree_map{$tree}}) {
 921                                 # MAXPARENT is defined to 16 in commit-tree.c:
 922                                 if ($seen_parent{$_} || @exec_parents > 16) {
 923                                         next;
 924                                 }
 925                                 push @exec_parents, $_;
 926                                 $seen_parent{$_} = 1;
 927                         }
 928                 }
 929                 my $msg_fh = IO::File->new_tmpfile or croak $!;
 930                 print $msg_fh $log_msg->{msg}, "\ngit-svn-id: ",
 931                                         "$SVN_URL\@$log_msg->{revision}",
 932                                         " $SVN_UUID\n" or croak $!;
 933                 $msg_fh->flush == 0 or croak $!;
 934                 seek $msg_fh, 0, 0 or croak $!;
 935
 936                 set_commit_env($log_msg);
 937
 938                 my @exec = ('git-commit-tree',$tree);
 939                 push @exec, '-p', $_  foreach @exec_parents;
 940                 open STDIN, '<&', $msg_fh or croak $!;
 941                 open STDOUT, '>&', $out_fh or croak $!;
 942                 exec @exec or croak $!;
 943         }
 944         waitpid($pid,0);
 945         croak if $?;
 946
 947         $out_fh->flush == 0 or croak $!;
 948         seek $out_fh, 0, 0 or croak $!;
 949         chomp(my $commit = do { local $/; <$out_fh> });
 950         if ($commit !~ /^$sha1$/o) {
 951                 croak "Failed to commit, invalid sha1: $commit\n";
 952         }
 953         my @update_ref = ('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
 954         if (my $primary_parent = shift @exec_parents) {
 955                 $pid = fork;
 956                 defined $pid or croak $!;
 957                 if (!$pid) {
 958                         close STDERR;
 959                         close STDOUT;
 960                         exec 'git-rev-parse','--verify',
 961                                                 "refs/remotes/$GIT_SVN^0";
 962                 }
 963                 waitpid $pid, 0;
 964                 push @update_ref, $primary_parent unless $?;
 965         }
 966         sys(@update_ref);
 967         sys('git-update-ref',"$GIT_SVN/revs/$log_msg->{revision}",$commit);
 968         print "r$log_msg->{revision} = $commit\n";
 969         return $commit;
 970 }
 971
 972 sub set_commit_env {
 973         my ($log_msg) = @_;
 974         my $author = $log_msg->{author};
 975         my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
 976                                 : ($author,"$author\@$SVN_UUID");
 977         $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
 978         $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
 979         $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
 980 }
 981
 982 sub apply_mod_line_blob {
 983         my $m = shift;
 984         if ($m->{mode_b} =~ /^120/) {
 985                 blob_to_symlink($m->{sha1_b}, $m->{file_b});
 986         } else {
 987                 blob_to_file($m->{sha1_b}, $m->{file_b});
 988         }
 989 }
 990
 991 sub blob_to_symlink {
 992         my ($blob, $link) = @_;
 993         defined $link or croak "\$link not defined!\n";
 994         croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
 995         if (-l $link || -f _) {
 996                 unlink $link or croak $!;
 997         }
 998
 999         my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
1000         symlink $dest, $link or croak $!;
1001 }
1002
1003 sub blob_to_file {
1004         my ($blob, $file) = @_;
1005         defined $file or croak "\$file not defined!\n";
1006         croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
1007         if (-l $file || -f _) {
1008                 unlink $file or croak $!;
1009         }
1010
1011         open my $blob_fh, '>', $file or croak "$!: $file\n";
1012         my $pid = fork;
1013         defined $pid or croak $!;
1014
1015         if ($pid == 0) {
1016                 open STDOUT, '>&', $blob_fh or croak $!;
1017                 exec('git-cat-file','blob',$blob);
1018         }
1019         waitpid $pid, 0;
1020         croak $? if $?;
1021
1022         close $blob_fh or croak $!;
1023 }
1024
1025 sub safe_qx {
1026         my $pid = open my $child, '-|';
1027         defined $pid or croak $!;
1028         if ($pid == 0) {
1029                 exec(@_) or croak $?;
1030         }
1031         my @ret = (<$child>);
1032         close $child or croak $?;
1033         die $? if $?; # just in case close didn't error out
1034         return wantarray ? @ret : join('',@ret);
1035 }
1036
1037 sub svn_compat_check {
1038         my @co_help = safe_qx(qw(svn co -h));
1039         unless (grep /ignore-externals/,@co_help) {
1040                 print STDERR "W: Installed svn version does not support ",
1041                                 "--ignore-externals\n";
1042                 $_no_ignore_ext = 1;
1043         }
1044         if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
1045                 $_svn_co_url_revs = 1;
1046         }
1047
1048         # I really, really hope nobody hits this...
1049         unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
1050                 print STDERR <<'';
1051 W: The installed svn version does not support the --stop-on-copy flag in
1052    the log command.
1053    Lets hope the directory you're tracking is not a branch or tag
1054    and was never moved within the repository...
1055
1056                 $_no_stop_copy = 1;
1057         }
1058 }
1059
1060 # *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
1061 # (and they won't honor URL@<rev> without -r<rev>, too!)
1062 sub svn_cmd_checkout {
1063         my ($url, $rev, $dir) = @_;
1064         my @cmd = ('svn','co', "-r$rev");
1065         push @cmd, '--ignore-externals' unless $_no_ignore_ext;
1066         $url .= "\@$rev" if $_svn_co_url_revs;
1067         sys(@cmd, $url, $dir);
1068 }
1069
1070 sub check_upgrade_needed {
1071         my $old = eval {
1072                 my $pid = open my $child, '-|';
1073                 defined $pid or croak $!;
1074                 if ($pid == 0) {
1075                         close STDERR;
1076                         exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $?;
1077                 }
1078                 my @ret = (<$child>);
1079                 close $child or croak $?;
1080                 die $? if $?; # just in case close didn't error out
1081                 return wantarray ? @ret : join('',@ret);
1082         };
1083         return unless $old;
1084         my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
1085         if ($@ || !$head) {
1086                 print STDERR "Please run: $0 rebuild --upgrade\n";
1087                 exit 1;
1088         }
1089 }
1090
1091 # fills %tree_map with a reverse mapping of trees to commits.  Useful
1092 # for finding parents to commit on.
1093 sub map_tree_joins {
1094         foreach my $br (@_branch_from) {
1095                 my $pid = open my $pipe, '-|';
1096                 defined $pid or croak $!;
1097                 if ($pid == 0) {
1098                         exec(qw(git-rev-list --pretty=raw), $br) or croak $?;
1099                 }
1100                 while (<$pipe>) {
1101                         if (/^commit ($sha1)$/o) {
1102                                 my $commit = $1;
1103                                 my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
1104                                 unless (defined $tree) {
1105                                         die "Failed to parse commit $commit\n";
1106                                 }
1107                                 push @{$tree_map{$tree}}, $commit;
1108                         }
1109                 }
1110                 close $pipe or croak $?;
1111         }
1112 }
1113
1114 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1115 sub load_authors {
1116         open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1117         while (<$authors>) {
1118                 chomp;
1119                 next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
1120                 my ($user, $name, $email) = ($1, $2, $3);
1121                 $users{$user} = [$name, $email];
1122         }
1123         close $authors or croak $!;
1124 }
1125
1126 __END__
1127
1128 Data structures:
1129
1130 $svn_log hashref (as returned by svn_log_raw)
1131 {
1132         fh => file handle of the log file,
1133         state => state of the log file parser (sep/msg/rev/msg_start...)
1134 }
1135
1136 $log_msg hashref as returned by next_log_entry($svn_log)
1137 {
1138         msg => 'whitespace-formatted log entry
1139 ',                                              # trailing newline is preserved
1140         revision => '8',                        # integer
1141         date => '2004-02-24T17:01:44.108345Z',  # commit date
1142         author => 'committer name'
1143 };
1144
1145
1146 @mods = array of diff-index line hashes, each element represents one line
1147         of diff-index output
1148
1149 diff-index line ($m hash)
1150 {
1151         mode_a => first column of diff-index output, no leading ':',
1152         mode_b => second column of diff-index output,
1153         sha1_b => sha1sum of the final blob,
1154         chg => change type [MCRADT],
1155         file_a => original file name of a file (iff chg is 'C' or 'R')
1156         file_b => new/current file name of a file (any chg)
1157 }
1158 ;