contrib/git-svn/git-svn.perl

   1 #!/usr/bin/env perl
   2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3 # License: GPL v2 or later
   4 use warnings;
   5 use strict;
   6 use vars qw/    $AUTHOR $VERSION
   7                 $SVN_URL $SVN_INFO $SVN_WC $SVN_UUID
   8                 $GIT_SVN_INDEX $GIT_SVN
   9                 $GIT_DIR $REV_DIR/;
  10 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  11 $VERSION = '0.10.0';
  12 $GIT_DIR = $ENV{GIT_DIR} || "$ENV{PWD}/.git";
  13 # make sure the svn binary gives consistent output between locales and TZs:
  14 $ENV{TZ} = 'UTC';
  15 $ENV{LC_ALL} = 'C';
  16
  17 # If SVN:: library support is added, please make the dependencies
  18 # optional and preserve the capability to use the command-line client.
  19 # use eval { require SVN::... } to make it lazy load
  20 # We don't use any modules not in the standard Perl distribution:
  21 use Carp qw/croak/;
  22 use IO::File qw//;
  23 use File::Basename qw/dirname basename/;
  24 use File::Path qw/mkpath/;
  25 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  26 use File::Spec qw//;
  27 use POSIX qw/strftime/;
  28 my $sha1 = qr/[a-f\d]{40}/;
  29 my $sha1_short = qr/[a-f\d]{4,40}/;
  30 my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
  31         $_find_copies_harder, $_l, $_version, $_upgrade, $_authors);
  32 my (@_branch_from, %tree_map, %users);
  33
  34 my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
  35                 'branch|b=s' => \@_branch_from,
  36                 'authors-file|A=s' => \$_authors );
  37 my %cmd = (
  38         fetch => [ \&fetch, "Download new revisions from SVN",
  39                         { 'revision|r=s' => \$_revision, %fc_opts } ],
  40         init => [ \&init, "Initialize and fetch (import)", { } ],
  41         commit => [ \&commit, "Commit git revisions to SVN",
  42                         {       'stdin|' => \$_stdin,
  43                                 'edit|e' => \$_edit,
  44                                 'rmdir' => \$_rmdir,
  45                                 'find-copies-harder' => \$_find_copies_harder,
  46                                 'l=i' => \$_l,
  47                                 %fc_opts,
  48                         } ],
  49         'show-ignore' => [ \&show_ignore, "Show svn:ignore listings", { } ],
  50         rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
  51                         { 'no-ignore-externals' => \$_no_ignore_ext,
  52                           'upgrade' => \$_upgrade } ],
  53 );
  54 my $cmd;
  55 for (my $i = 0; $i < @ARGV; $i++) {
  56         if (defined $cmd{$ARGV[$i]}) {
  57                 $cmd = $ARGV[$i];
  58                 splice @ARGV, $i, 1;
  59                 last;
  60         }
  61 };
  62
  63 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
  64
  65 GetOptions(%opts, 'help|H|h' => \$_help,
  66                 'version|V' => \$_version,
  67                 'id|i=s' => \$GIT_SVN) or exit 1;
  68
  69 $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
  70 $GIT_SVN_INDEX = "$GIT_DIR/$GIT_SVN/index";
  71 $ENV{GIT_DIR} ||= $GIT_DIR;
  72 $SVN_URL = undef;
  73 $REV_DIR = "$GIT_DIR/$GIT_SVN/revs";
  74 $SVN_WC = "$GIT_DIR/$GIT_SVN/tree";
  75
  76 usage(0) if $_help;
  77 version() if $_version;
  78 usage(1) unless defined $cmd;
  79 load_authors() if $_authors;
  80 svn_check_ignore_externals();
  81 $cmd{$cmd}->[0]->(@ARGV);
  82 exit 0;
  83
  84 ####################### primary functions ######################
  85 sub usage {
  86         my $exit = shift || 0;
  87         my $fd = $exit ? \*STDERR : \*STDOUT;
  88         print $fd <<"";
  89 git-svn - bidirectional operations between a single Subversion tree and git
  90 Usage: $0 <command> [options] [arguments]\n
  91
  92         print $fd "Available commands:\n" unless $cmd;
  93
  94         foreach (sort keys %cmd) {
  95                 next if $cmd && $cmd ne $_;
  96                 print $fd '  ',pack('A13',$_),$cmd{$_}->[1],"\n";
  97                 foreach (keys %{$cmd{$_}->[2]}) {
  98                         # prints out arguments as they should be passed:
  99                         my $x = s#=s$## ? '<arg>' : s#=i$## ? '<num>' : '';
 100                         print $fd ' ' x 17, join(', ', map { length $_ > 1 ?
 101                                                         "--$_" : "-$_" }
 102                                                 split /\|/,$_)," $x\n";
 103                 }
 104         }
 105         print $fd <<"";
 106 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 107 arbitrary identifier if you're tracking multiple SVN branches/repositories in
 108 one git repository and want to keep them separate.  See git-svn(1) for more
 109 information.
 110
 111         exit $exit;
 112 }
 113
 114 sub version {
 115         print "git-svn version $VERSION\n";
 116         exit 0;
 117 }
 118
 119 sub rebuild {
 120         $SVN_URL = shift or undef;
 121         my $newest_rev = 0;
 122         if ($_upgrade) {
 123                 sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD");
 124         } else {
 125                 check_upgrade_needed();
 126         }
 127
 128         my $pid = open(my $rev_list,'-|');
 129         defined $pid or croak $!;
 130         if ($pid == 0) {
 131                 exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!;
 132         }
 133         my $latest;
 134         while (<$rev_list>) {
 135                 chomp;
 136                 my $c = $_;
 137                 croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 138                 my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
 139                 next if (!@commit); # skip merges
 140                 my $id = $commit[$#commit];
 141                 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 142                                                 \s([a-f\d\-]+)$/x);
 143                 if (!$rev || !$uuid || !$url) {
 144                         # some of the original repositories I made had
 145                         # indentifiers like this:
 146                         ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)
 147                                                         \@([a-f\d\-]+)/x);
 148                         if (!$rev || !$uuid) {
 149                                 croak "Unable to extract revision or UUID from ",
 150                                         "$c, $id\n";
 151                         }
 152                 }
 153
 154                 # if we merged or otherwise started elsewhere, this is
 155                 # how we break out of it
 156                 next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
 157                 next if (defined $SVN_URL && ($url ne $SVN_URL));
 158
 159                 print "r$rev = $c\n";
 160                 unless (defined $latest) {
 161                         if (!$SVN_URL && !$url) {
 162                                 croak "SVN repository location required: $url\n";
 163                         }
 164                         $SVN_URL ||= $url;
 165                         $SVN_UUID ||= setup_git_svn();
 166                         $latest = $rev;
 167                 }
 168                 assert_revision_eq_or_unknown($rev, $c);
 169                 sys('git-update-ref',"$GIT_SVN/revs/$rev",$c);
 170                 $newest_rev = $rev if ($rev > $newest_rev);
 171         }
 172         close $rev_list or croak $?;
 173         if (!chdir $SVN_WC) {
 174                 my @svn_co = ('svn','co',"-r$latest");
 175                 push @svn_co, '--ignore-externals' unless $_no_ignore_ext;
 176                 sys(@svn_co, $SVN_URL, $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 $svn_log = svn_log_raw(@log_args);
 220         @$svn_log = sort { $a->{revision} <=> $b->{revision} } @$svn_log;
 221
 222         my $base = shift @$svn_log or croak "No base revision!\n";
 223         my $last_commit = undef;
 224         unless (-d $SVN_WC) {
 225                 my @svn_co = ('svn','co',"-r$base->{revision}");
 226                 push @svn_co,'--ignore-externals' unless $_no_ignore_ext;
 227                 sys(@svn_co, $SVN_URL, $SVN_WC);
 228                 chdir $SVN_WC or croak $!;
 229                 $last_commit = git_commit($base, @parents);
 230                 assert_svn_wc_clean($base->{revision}, $last_commit);
 231         } else {
 232                 chdir $SVN_WC or croak $!;
 233                 $last_commit = file_to_s("$REV_DIR/$base->{revision}");
 234         }
 235         my @svn_up = qw(svn up);
 236         push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 237         my $last_rev = $base->{revision};
 238         foreach my $log_msg (@$svn_log) {
 239                 assert_svn_wc_clean($last_rev, $last_commit);
 240                 $last_rev = $log_msg->{revision};
 241                 sys(@svn_up,"-r$last_rev");
 242                 $last_commit = git_commit($log_msg, $last_commit, @parents);
 243         }
 244         assert_svn_wc_clean($last_rev, $last_commit);
 245         unless (-e "$GIT_DIR/refs/heads/master") {
 246                 sys(qw(git-update-ref refs/heads/master),$last_commit);
 247         }
 248         return pop @$svn_log;
 249 }
 250
 251 sub commit {
 252         my (@commits) = @_;
 253         check_upgrade_needed();
 254         if ($_stdin || !@commits) {
 255                 print "Reading from stdin...\n";
 256                 @commits = ();
 257                 while (<STDIN>) {
 258                         if (/\b($sha1_short)\b/o) {
 259                                 unshift @commits, $1;
 260                         }
 261                 }
 262         }
 263         my @revs;
 264         foreach my $c (@commits) {
 265                 chomp(my @tmp = safe_qx('git-rev-parse',$c));
 266                 if (scalar @tmp == 1) {
 267                         push @revs, $tmp[0];
 268                 } elsif (scalar @tmp > 1) {
 269                         push @revs, reverse (safe_qx('git-rev-list',@tmp));
 270                 } else {
 271                         die "Failed to rev-parse $c\n";
 272                 }
 273         }
 274         chomp @revs;
 275
 276         fetch();
 277         chdir $SVN_WC or croak $!;
 278         my $svn_current_rev =  svn_info('.')->{'Last Changed Rev'};
 279         foreach my $c (@revs) {
 280                 my $mods = svn_checkout_tree($svn_current_rev, $c);
 281                 if (scalar @$mods == 0) {
 282                         print "Skipping, no changes detected\n";
 283                         next;
 284                 }
 285                 $svn_current_rev = svn_commit_tree($svn_current_rev, $c);
 286         }
 287         print "Done committing ",scalar @revs," revisions to SVN\n";
 288
 289 }
 290
 291 sub show_ignore {
 292         require File::Find or die $!;
 293         my $exclude_file = "$GIT_DIR/info/exclude";
 294         open my $fh, '<', $exclude_file or croak $!;
 295         chomp(my @excludes = (<$fh>));
 296         close $fh or croak $!;
 297
 298         $SVN_URL ||= file_to_s("$GIT_DIR/$GIT_SVN/info/url");
 299         chdir $SVN_WC or croak $!;
 300         my %ign;
 301         File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
 302                 s#^\./##;
 303                 @{$ign{$_}} = safe_qx(qw(svn propget svn:ignore),$_);
 304                 }}, no_chdir=>1},'.');
 305
 306         print "\n# /\n";
 307         foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
 308         delete $ign{'.'};
 309         foreach my $i (sort keys %ign) {
 310                 print "\n# ",$i,"\n";
 311                 foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
 312         }
 313 }
 314
 315 ########################### utility functions #########################
 316
 317 sub setup_git_svn {
 318         defined $SVN_URL or croak "SVN repository location required\n";
 319         unless (-d $GIT_DIR) {
 320                 croak "GIT_DIR=$GIT_DIR does not exist!\n";
 321         }
 322         mkpath(["$GIT_DIR/$GIT_SVN"]);
 323         mkpath(["$GIT_DIR/$GIT_SVN/info"]);
 324         mkpath([$REV_DIR]);
 325         s_to_file($SVN_URL,"$GIT_DIR/$GIT_SVN/info/url");
 326         $SVN_UUID = svn_info($SVN_URL)->{'Repository UUID'} or
 327                                         croak "Repository UUID unreadable\n";
 328         s_to_file($SVN_UUID,"$GIT_DIR/$GIT_SVN/info/uuid");
 329
 330         open my $fd, '>>', "$GIT_DIR/$GIT_SVN/info/exclude" or croak $!;
 331         print $fd '.svn',"\n";
 332         close $fd or croak $!;
 333         return $SVN_UUID;
 334 }
 335
 336 sub assert_svn_wc_clean {
 337         my ($svn_rev, $treeish) = @_;
 338         croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
 339         croak "$treeish is not a sha1!\n" unless ($treeish =~ /^$sha1$/o);
 340         my $lcr = svn_info('.')->{'Last Changed Rev'};
 341         if ($svn_rev != $lcr) {
 342                 print STDERR "Checking for copy-tree ... ";
 343                 # use
 344                 my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
 345                                                 "-r$lcr:$svn_rev")));
 346                 if (@diff) {
 347                         croak "Nope!  Expected r$svn_rev, got r$lcr\n";
 348                 } else {
 349                         print STDERR "OK!\n";
 350                 }
 351         }
 352         my @status = grep(!/^Performing status on external/,(`svn status`));
 353         @status = grep(!/^\s*$/,@status);
 354         if (scalar @status) {
 355                 print STDERR "Tree ($SVN_WC) is not clean:\n";
 356                 print STDERR $_ foreach @status;
 357                 croak;
 358         }
 359         assert_tree($treeish);
 360 }
 361
 362 sub assert_tree {
 363         my ($treeish) = @_;
 364         croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
 365         chomp(my $type = `git-cat-file -t $treeish`);
 366         my $expected;
 367         while ($type eq 'tag') {
 368                 chomp(($treeish, $type) = `git-cat-file tag $treeish`);
 369         }
 370         if ($type eq 'commit') {
 371                 $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
 372                 ($expected) = ($expected =~ /^tree ($sha1)$/);
 373                 die "Unable to get tree from $treeish\n" unless $expected;
 374         } elsif ($type eq 'tree') {
 375                 $expected = $treeish;
 376         } else {
 377                 die "$treeish is a $type, expected tree, tag or commit\n";
 378         }
 379
 380         my $old_index = $ENV{GIT_INDEX_FILE};
 381         my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
 382         if (-e $tmpindex) {
 383                 unlink $tmpindex or croak $!;
 384         }
 385         $ENV{GIT_INDEX_FILE} = $tmpindex;
 386         git_addremove();
 387         chomp(my $tree = `git-write-tree`);
 388         if ($old_index) {
 389                 $ENV{GIT_INDEX_FILE} = $old_index;
 390         } else {
 391                 delete $ENV{GIT_INDEX_FILE};
 392         }
 393         if ($tree ne $expected) {
 394                 croak "Tree mismatch, Got: $tree, Expected: $expected\n";
 395         }
 396 }
 397
 398 sub parse_diff_tree {
 399         my $diff_fh = shift;
 400         local $/ = "\0";
 401         my $state = 'meta';
 402         my @mods;
 403         while (<$diff_fh>) {
 404                 chomp $_; # this gets rid of the trailing "\0"
 405                 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
 406                                         $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
 407                         push @mods, {   mode_a => $1, mode_b => $2,
 408                                         sha1_b => $3, chg => $4 };
 409                         if ($4 =~ /^(?:C|R)$/) {
 410                                 $state = 'file_a';
 411                         } else {
 412                                 $state = 'file_b';
 413                         }
 414                 } elsif ($state eq 'file_a') {
 415                         my $x = $mods[$#mods] or croak "Empty array\n";
 416                         if ($x->{chg} !~ /^(?:C|R)$/) {
 417                                 croak "Error parsing $_, $x->{chg}\n";
 418                         }
 419                         $x->{file_a} = $_;
 420                         $state = 'file_b';
 421                 } elsif ($state eq 'file_b') {
 422                         my $x = $mods[$#mods] or croak "Empty array\n";
 423                         if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
 424                                 croak "Error parsing $_, $x->{chg}\n";
 425                         }
 426                         if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
 427                                 croak "Error parsing $_, $x->{chg}\n";
 428                         }
 429                         $x->{file_b} = $_;
 430                         $state = 'meta';
 431                 } else {
 432                         croak "Error parsing $_\n";
 433                 }
 434         }
 435         close $diff_fh or croak $!;
 436
 437         return \@mods;
 438 }
 439
 440 sub svn_check_prop_executable {
 441         my $m = shift;
 442         return if -l $m->{file_b};
 443         if ($m->{mode_b} =~ /755$/) {
 444                 chmod((0755 &~ umask),$m->{file_b}) or croak $!;
 445                 if ($m->{mode_a} !~ /755$/) {
 446                         sys(qw(svn propset svn:executable 1), $m->{file_b});
 447                 }
 448                 -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
 449         } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
 450                 sys(qw(svn propdel svn:executable), $m->{file_b});
 451                 chmod((0644 &~ umask),$m->{file_b}) or croak $!;
 452                 -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
 453         }
 454 }
 455
 456 sub svn_ensure_parent_path {
 457         my $dir_b = dirname(shift);
 458         svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
 459         mkpath([$dir_b]) unless (-d $dir_b);
 460         sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
 461 }
 462
 463 sub precommit_check {
 464         my $mods = shift;
 465         my (%rm_file, %rmdir_check, %added_check);
 466
 467         my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
 468         foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
 469                 if ($m->{chg} eq 'R') {
 470                         if (-d $m->{file_b}) {
 471                                 err_dir_to_file("$m->{file_a} => $m->{file_b}");
 472                         }
 473                         # dir/$file => dir/file/$file
 474                         my $dirname = dirname($m->{file_b});
 475                         while ($dirname ne File::Spec->curdir) {
 476                                 if ($dirname ne $m->{file_a}) {
 477                                         $dirname = dirname($dirname);
 478                                         next;
 479                                 }
 480                                 err_file_to_dir("$m->{file_a} => $m->{file_b}");
 481                         }
 482                         # baz/zzz => baz (baz is a file)
 483                         $dirname = dirname($m->{file_a});
 484                         while ($dirname ne File::Spec->curdir) {
 485                                 if ($dirname ne $m->{file_b}) {
 486                                         $dirname = dirname($dirname);
 487                                         next;
 488                                 }
 489                                 err_dir_to_file("$m->{file_a} => $m->{file_b}");
 490                         }
 491                 }
 492                 if ($m->{chg} =~ /^(D|R)$/) {
 493                         my $t = $1 eq 'D' ? 'file_b' : 'file_a';
 494                         $rm_file{ $m->{$t} } = 1;
 495                         my $dirname = dirname( $m->{$t} );
 496                         my $basename = basename( $m->{$t} );
 497                         $rmdir_check{$dirname}->{$basename} = 1;
 498                 } elsif ($m->{chg} =~ /^(?:A|C)$/) {
 499                         if (-d $m->{file_b}) {
 500                                 err_dir_to_file($m->{file_b});
 501                         }
 502                         my $dirname = dirname( $m->{file_b} );
 503                         my $basename = basename( $m->{file_b} );
 504                         $added_check{$dirname}->{$basename} = 1;
 505                         while ($dirname ne File::Spec->curdir) {
 506                                 if ($rm_file{$dirname}) {
 507                                         err_file_to_dir($m->{file_b});
 508                                 }
 509                                 $dirname = dirname $dirname;
 510                         }
 511                 }
 512         }
 513         return (\%rmdir_check, \%added_check);
 514
 515         sub err_dir_to_file {
 516                 my $file = shift;
 517                 print STDERR "Node change from directory to file ",
 518                                 "is not supported by Subversion: ",$file,"\n";
 519                 exit 1;
 520         }
 521         sub err_file_to_dir {
 522                 my $file = shift;
 523                 print STDERR "Node change from file to directory ",
 524                                 "is not supported by Subversion: ",$file,"\n";
 525                 exit 1;
 526         }
 527 }
 528
 529 sub svn_checkout_tree {
 530         my ($svn_rev, $treeish) = @_;
 531         my $from = file_to_s("$REV_DIR/$svn_rev");
 532         assert_svn_wc_clean($svn_rev,$from);
 533         print "diff-tree $from $treeish\n";
 534         my $pid = open my $diff_fh, '-|';
 535         defined $pid or croak $!;
 536         if ($pid == 0) {
 537                 my @diff_tree = qw(git-diff-tree -z -r -C);
 538                 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
 539                 push @diff_tree, "-l$_l" if defined $_l;
 540                 exec(@diff_tree, $from, $treeish) or croak $!;
 541         }
 542         my $mods = parse_diff_tree($diff_fh);
 543         unless (@$mods) {
 544                 # git can do empty commits, but SVN doesn't allow it...
 545                 return $mods;
 546         }
 547         my ($rm, $add) = precommit_check($mods);
 548
 549         my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
 550         foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
 551                 if ($m->{chg} eq 'C') {
 552                         svn_ensure_parent_path( $m->{file_b} );
 553                         sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
 554                         apply_mod_line_blob($m);
 555                         svn_check_prop_executable($m);
 556                 } elsif ($m->{chg} eq 'D') {
 557                         sys(qw(svn rm --force), $m->{file_b});
 558                 } elsif ($m->{chg} eq 'R') {
 559                         svn_ensure_parent_path( $m->{file_b} );
 560                         sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
 561                         apply_mod_line_blob($m);
 562                         svn_check_prop_executable($m);
 563                 } elsif ($m->{chg} eq 'M') {
 564                         apply_mod_line_blob($m);
 565                         svn_check_prop_executable($m);
 566                 } elsif ($m->{chg} eq 'T') {
 567                         sys(qw(svn rm --force),$m->{file_b});
 568                         apply_mod_line_blob($m);
 569                         sys(qw(svn add --force), $m->{file_b});
 570                         svn_check_prop_executable($m);
 571                 } elsif ($m->{chg} eq 'A') {
 572                         svn_ensure_parent_path( $m->{file_b} );
 573                         apply_mod_line_blob($m);
 574                         sys(qw(svn add --force), $m->{file_b});
 575                         svn_check_prop_executable($m);
 576                 } else {
 577                         croak "Invalid chg: $m->{chg}\n";
 578                 }
 579         }
 580
 581         assert_tree($treeish);
 582         if ($_rmdir) { # remove empty directories
 583                 handle_rmdir($rm, $add);
 584         }
 585         assert_tree($treeish);
 586         return $mods;
 587 }
 588
 589 # svn ls doesn't work with respect to the current working tree, but what's
 590 # in the repository.  There's not even an option for it... *sigh*
 591 # (added files don't show up and removed files remain in the ls listing)
 592 sub svn_ls_current {
 593         my ($dir, $rm, $add) = @_;
 594         chomp(my @ls = safe_qx('svn','ls',$dir));
 595         my @ret = ();
 596         foreach (@ls) {
 597                 s#/$##; # trailing slashes are evil
 598                 push @ret, $_ unless $rm->{$dir}->{$_};
 599         }
 600         if (exists $add->{$dir}) {
 601                 push @ret, keys %{$add->{$dir}};
 602         }
 603         return \@ret;
 604 }
 605
 606 sub handle_rmdir {
 607         my ($rm, $add) = @_;
 608
 609         foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
 610                 my $ls = svn_ls_current($dir, $rm, $add);
 611                 next if (scalar @$ls);
 612                 sys(qw(svn rm --force),$dir);
 613
 614                 my $dn = dirname $dir;
 615                 $rm->{ $dn }->{ basename $dir } = 1;
 616                 $ls = svn_ls_current($dn, $rm, $add);
 617                 while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
 618                         sys(qw(svn rm --force),$dn);
 619                         $dir = basename $dn;
 620                         $dn = dirname $dn;
 621                         $rm->{ $dn }->{ $dir } = 1;
 622                         $ls = svn_ls_current($dn, $rm, $add);
 623                 }
 624         }
 625 }
 626
 627 sub svn_commit_tree {
 628         my ($svn_rev, $commit) = @_;
 629         my $commit_msg = "$GIT_DIR/$GIT_SVN/.svn-commit.tmp.$$";
 630         my %log_msg = ( msg => '' );
 631         open my $msg, '>', $commit_msg or croak $!;
 632
 633         chomp(my $type = `git-cat-file -t $commit`);
 634         if ($type eq 'commit') {
 635                 my $pid = open my $msg_fh, '-|';
 636                 defined $pid or croak $!;
 637
 638                 if ($pid == 0) {
 639                         exec(qw(git-cat-file commit), $commit) or croak $!;
 640                 }
 641                 my $in_msg = 0;
 642                 while (<$msg_fh>) {
 643                         if (!$in_msg) {
 644                                 $in_msg = 1 if (/^\s*$/);
 645                         } elsif (/^git-svn-id: /) {
 646                                 # skip this, we regenerate the correct one
 647                                 # on re-fetch anyways
 648                         } else {
 649                                 print $msg $_ or croak $!;
 650                         }
 651                 }
 652                 close $msg_fh or croak $!;
 653         }
 654         close $msg or croak $!;
 655
 656         if ($_edit || ($type eq 'tree')) {
 657                 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 658                 system($editor, $commit_msg);
 659         }
 660
 661         # file_to_s removes all trailing newlines, so just use chomp() here:
 662         open $msg, '<', $commit_msg or croak $!;
 663         { local $/; chomp($log_msg{msg} = <$msg>); }
 664         close $msg or croak $!;
 665
 666         my ($oneline) = ($log_msg{msg} =~ /([^\n\r]+)/);
 667         print "Committing $commit: $oneline\n";
 668
 669         my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
 670         my ($committed) = grep(/^Committed revision \d+\./,@ci_output);
 671         unlink $commit_msg;
 672         defined $committed or croak
 673                         "Commit output failed to parse committed revision!\n",
 674                         join("\n",@ci_output),"\n";
 675         my ($rev_committed) = ($committed =~ /^Committed revision (\d+)\./);
 676
 677         my @svn_up = qw(svn up);
 678         push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 679         if ($rev_committed == ($svn_rev + 1)) {
 680                 push @svn_up, "-r$rev_committed";
 681                 sys(@svn_up);
 682                 my $info = svn_info('.');
 683                 my $date = $info->{'Last Changed Date'} or die "Missing date\n";
 684                 if ($info->{'Last Changed Rev'} != $rev_committed) {
 685                         croak "$info->{'Last Changed Rev'} != $rev_committed\n"
 686                 }
 687                 my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
 688                                         /(\d{4})\-(\d\d)\-(\d\d)\s
 689                                          (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
 690                                          or croak "Failed to parse date: $date\n";
 691                 $log_msg{date} = "$tz $Y-$m-$d $H:$M:$S";
 692                 $log_msg{author} = $info->{'Last Changed Author'};
 693                 $log_msg{revision} = $rev_committed;
 694                 $log_msg{msg} .= "\n";
 695                 my $parent = file_to_s("$REV_DIR/$svn_rev");
 696                 git_commit(\%log_msg, $parent, $commit);
 697                 return $rev_committed;
 698         }
 699         # resync immediately
 700         push @svn_up, "-r$svn_rev";
 701         sys(@svn_up);
 702         return fetch("$rev_committed=$commit")->{revision};
 703 }
 704
 705 sub svn_log_raw {
 706         my (@log_args) = @_;
 707         my $pid = open my $log_fh,'-|';
 708         defined $pid or croak $!;
 709
 710         if ($pid == 0) {
 711                 exec (qw(svn log), @log_args) or croak $!
 712         }
 713
 714         my @svn_log;
 715         my $state = 'sep';
 716         while (<$log_fh>) {
 717                 chomp;
 718                 if (/^\-{72}$/) {
 719                         if ($state eq 'msg') {
 720                                 if ($svn_log[$#svn_log]->{lines}) {
 721                                         $svn_log[$#svn_log]->{msg} .= $_."\n";
 722                                         unless(--$svn_log[$#svn_log]->{lines}) {
 723                                                 $state = 'sep';
 724                                         }
 725                                 } else {
 726                                         croak "Log parse error at: $_\n",
 727                                                 $svn_log[$#svn_log]->{revision},
 728                                                 "\n";
 729                                 }
 730                                 next;
 731                         }
 732                         if ($state ne 'sep') {
 733                                 croak "Log parse error at: $_\n",
 734                                         "state: $state\n",
 735                                         $svn_log[$#svn_log]->{revision},
 736                                         "\n";
 737                         }
 738                         $state = 'rev';
 739
 740                         # if we have an empty log message, put something there:
 741                         if (@svn_log) {
 742                                 $svn_log[$#svn_log]->{msg} ||= "\n";
 743                                 delete $svn_log[$#svn_log]->{lines};
 744                         }
 745                         next;
 746                 }
 747                 if ($state eq 'rev' && s/^r(\d+)\s*\|\s*//) {
 748                         my $rev = $1;
 749                         my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
 750                         ($lines) = ($lines =~ /(\d+)/);
 751                         my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
 752                                         /(\d{4})\-(\d\d)\-(\d\d)\s
 753                                          (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
 754                                          or croak "Failed to parse date: $date\n";
 755                         my %log_msg = ( revision => $rev,
 756                                         date => "$tz $Y-$m-$d $H:$M:$S",
 757                                         author => $author,
 758                                         lines => $lines,
 759                                         msg => '' );
 760                         if (defined $_authors && ! defined $users{$author}) {
 761                                 die "Author: $author not defined in ",
 762                                                 "$_authors file\n";
 763                         }
 764                         push @svn_log, \%log_msg;
 765                         $state = 'msg_start';
 766                         next;
 767                 }
 768                 # skip the first blank line of the message:
 769                 if ($state eq 'msg_start' && /^$/) {
 770                         $state = 'msg';
 771                 } elsif ($state eq 'msg') {
 772                         if ($svn_log[$#svn_log]->{lines}) {
 773                                 $svn_log[$#svn_log]->{msg} .= $_."\n";
 774                                 unless (--$svn_log[$#svn_log]->{lines}) {
 775                                         $state = 'sep';
 776                                 }
 777                         } else {
 778                                 croak "Log parse error at: $_\n",
 779                                         $svn_log[$#svn_log]->{revision},"\n";
 780                         }
 781                 }
 782         }
 783         close $log_fh or croak $?;
 784         return \@svn_log;
 785 }
 786
 787 sub svn_info {
 788         my $url = shift || $SVN_URL;
 789
 790         my $pid = open my $info_fh, '-|';
 791         defined $pid or croak $!;
 792
 793         if ($pid == 0) {
 794                 exec(qw(svn info),$url) or croak $!;
 795         }
 796
 797         my $ret = {};
 798         # only single-lines seem to exist in svn info output
 799         while (<$info_fh>) {
 800                 chomp $_;
 801                 if (m#^([^:]+)\s*:\s*(\S.*)$#) {
 802                         $ret->{$1} = $2;
 803                         push @{$ret->{-order}}, $1;
 804                 }
 805         }
 806         close $info_fh or croak $!;
 807         return $ret;
 808 }
 809
 810 sub sys { system(@_) == 0 or croak $? }
 811
 812 sub git_addremove {
 813         system( "git-diff-files --name-only -z ".
 814                                 " | git-update-index --remove -z --stdin && ".
 815                 "git-ls-files -z --others ".
 816                         "'--exclude-from=$GIT_DIR/$GIT_SVN/info/exclude'".
 817                                 " | git-update-index --add -z --stdin"
 818                 ) == 0 or croak $?
 819 }
 820
 821 sub s_to_file {
 822         my ($str, $file, $mode) = @_;
 823         open my $fd,'>',$file or croak $!;
 824         print $fd $str,"\n" or croak $!;
 825         close $fd or croak $!;
 826         chmod ($mode &~ umask, $file) if (defined $mode);
 827 }
 828
 829 sub file_to_s {
 830         my $file = shift;
 831         open my $fd,'<',$file or croak "$!: file: $file\n";
 832         local $/;
 833         my $ret = <$fd>;
 834         close $fd or croak $!;
 835         $ret =~ s/\s*$//s;
 836         return $ret;
 837 }
 838
 839 sub assert_revision_unknown {
 840         my $revno = shift;
 841         if (-f "$REV_DIR/$revno") {
 842                 croak "$REV_DIR/$revno already exists! ",
 843                                 "Why are we refetching it?";
 844         }
 845 }
 846
 847 sub assert_revision_eq_or_unknown {
 848         my ($revno, $commit) = @_;
 849         if (-f "$REV_DIR/$revno") {
 850                 my $current = file_to_s("$REV_DIR/$revno");
 851                 if ($commit ne $current) {
 852                         croak "$REV_DIR/$revno already exists!\n",
 853                                 "current: $current\nexpected: $commit\n";
 854                 }
 855                 return;
 856         }
 857 }
 858
 859 sub git_commit {
 860         my ($log_msg, @parents) = @_;
 861         assert_revision_unknown($log_msg->{revision});
 862         my $out_fh = IO::File->new_tmpfile or croak $!;
 863         $SVN_UUID ||= svn_info('.')->{'Repository UUID'};
 864
 865         map_tree_joins() if (@_branch_from && !%tree_map);
 866
 867         # commit parents can be conditionally bound to a particular
 868         # svn revision via: "svn_revno=commit_sha1", filter them out here:
 869         my @exec_parents;
 870         foreach my $p (@parents) {
 871                 next unless defined $p;
 872                 if ($p =~ /^(\d+)=($sha1_short)$/o) {
 873                         if ($1 == $log_msg->{revision}) {
 874                                 push @exec_parents, $2;
 875                         }
 876                 } else {
 877                         push @exec_parents, $p if $p =~ /$sha1_short/o;
 878                 }
 879         }
 880
 881         my $pid = fork;
 882         defined $pid or croak $!;
 883         if ($pid == 0) {
 884                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 885                 git_addremove();
 886                 chomp(my $tree = `git-write-tree`);
 887                 croak if $?;
 888                 if (exists $tree_map{$tree}) {
 889                         my %seen_parent = map { $_ => 1 } @exec_parents;
 890                         foreach (@{$tree_map{$tree}}) {
 891                                 # MAXPARENT is defined to 16 in commit-tree.c:
 892                                 if ($seen_parent{$_} || @exec_parents > 16) {
 893                                         next;
 894                                 }
 895                                 push @exec_parents, $_;
 896                                 $seen_parent{$_} = 1;
 897                         }
 898                 }
 899                 my $msg_fh = IO::File->new_tmpfile or croak $!;
 900                 print $msg_fh $log_msg->{msg}, "\ngit-svn-id: ",
 901                                         "$SVN_URL\@$log_msg->{revision}",
 902                                         " $SVN_UUID\n" or croak $!;
 903                 $msg_fh->flush == 0 or croak $!;
 904                 seek $msg_fh, 0, 0 or croak $!;
 905
 906                 set_commit_env($log_msg);
 907
 908                 my @exec = ('git-commit-tree',$tree);
 909                 push @exec, '-p', $_  foreach @exec_parents;
 910                 open STDIN, '<&', $msg_fh or croak $!;
 911                 open STDOUT, '>&', $out_fh or croak $!;
 912                 exec @exec or croak $!;
 913         }
 914         waitpid($pid,0);
 915         croak if $?;
 916
 917         $out_fh->flush == 0 or croak $!;
 918         seek $out_fh, 0, 0 or croak $!;
 919         chomp(my $commit = do { local $/; <$out_fh> });
 920         if ($commit !~ /^$sha1$/o) {
 921                 croak "Failed to commit, invalid sha1: $commit\n";
 922         }
 923         my @update_ref = ('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
 924         if (my $primary_parent = shift @exec_parents) {
 925                 push @update_ref, $primary_parent;
 926         }
 927         sys(@update_ref);
 928         sys('git-update-ref',"$GIT_SVN/revs/$log_msg->{revision}",$commit);
 929         print "r$log_msg->{revision} = $commit\n";
 930         return $commit;
 931 }
 932
 933 sub set_commit_env {
 934         my ($log_msg) = @_;
 935         my $author = $log_msg->{author};
 936         my ($name,$email) = defined $users{$author} ?  @{$users{$author}}
 937                                 : ($author,"$author\@$SVN_UUID");
 938         $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
 939         $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
 940         $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
 941 }
 942
 943 sub apply_mod_line_blob {
 944         my $m = shift;
 945         if ($m->{mode_b} =~ /^120/) {
 946                 blob_to_symlink($m->{sha1_b}, $m->{file_b});
 947         } else {
 948                 blob_to_file($m->{sha1_b}, $m->{file_b});
 949         }
 950 }
 951
 952 sub blob_to_symlink {
 953         my ($blob, $link) = @_;
 954         defined $link or croak "\$link not defined!\n";
 955         croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
 956         if (-l $link || -f _) {
 957                 unlink $link or croak $!;
 958         }
 959
 960         my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
 961         symlink $dest, $link or croak $!;
 962 }
 963
 964 sub blob_to_file {
 965         my ($blob, $file) = @_;
 966         defined $file or croak "\$file not defined!\n";
 967         croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
 968         if (-l $file || -f _) {
 969                 unlink $file or croak $!;
 970         }
 971
 972         open my $blob_fh, '>', $file or croak "$!: $file\n";
 973         my $pid = fork;
 974         defined $pid or croak $!;
 975
 976         if ($pid == 0) {
 977                 open STDOUT, '>&', $blob_fh or croak $!;
 978                 exec('git-cat-file','blob',$blob);
 979         }
 980         waitpid $pid, 0;
 981         croak $? if $?;
 982
 983         close $blob_fh or croak $!;
 984 }
 985
 986 sub safe_qx {
 987         my $pid = open my $child, '-|';
 988         defined $pid or croak $!;
 989         if ($pid == 0) {
 990                 exec(@_) or croak $?;
 991         }
 992         my @ret = (<$child>);
 993         close $child or croak $?;
 994         die $? if $?; # just in case close didn't error out
 995         return wantarray ? @ret : join('',@ret);
 996 }
 997
 998 sub svn_check_ignore_externals {
 999         return if $_no_ignore_ext;
1000         unless (grep /ignore-externals/,(safe_qx(qw(svn co -h)))) {
1001                 print STDERR "W: Installed svn version does not support ",
1002                                 "--ignore-externals\n";
1003                 $_no_ignore_ext = 1;
1004         }
1005 }
1006
1007 sub check_upgrade_needed {
1008         my $old = eval {
1009                 my $pid = open my $child, '-|';
1010                 defined $pid or croak $!;
1011                 if ($pid == 0) {
1012                         close STDERR;
1013                         exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $?;
1014                 }
1015                 my @ret = (<$child>);
1016                 close $child or croak $?;
1017                 die $? if $?; # just in case close didn't error out
1018                 return wantarray ? @ret : join('',@ret);
1019         };
1020         return unless $old;
1021         my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
1022         if ($@ || !$head) {
1023                 print STDERR "Please run: $0 rebuild --upgrade\n";
1024                 exit 1;
1025         }
1026 }
1027
1028 # fills %tree_map with a reverse mapping of trees to commits.  Useful
1029 # for finding parents to commit on.
1030 sub map_tree_joins {
1031         foreach my $br (@_branch_from) {
1032                 my $pid = open my $pipe, '-|';
1033                 defined $pid or croak $!;
1034                 if ($pid == 0) {
1035                         exec(qw(git-rev-list --pretty=raw), $br) or croak $?;
1036                 }
1037                 while (<$pipe>) {
1038                         if (/^commit ($sha1)$/o) {
1039                                 my $commit = $1;
1040                                 my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
1041                                 unless (defined $tree) {
1042                                         die "Failed to parse commit $commit\n";
1043                                 }
1044                                 push @{$tree_map{$tree}}, $commit;
1045                         }
1046                 }
1047                 close $pipe or croak $?;
1048         }
1049 }
1050
1051 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1052 sub load_authors {
1053         open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1054         while (<$authors>) {
1055                 chomp;
1056                 next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
1057                 my ($user, $name, $email) = ($1, $2, $3);
1058                 $users{$user} = [$name, $email];
1059         }
1060         close $authors or croak $!;
1061 }
1062
1063 __END__
1064
1065 Data structures:
1066
1067 @svn_log = array of log_msg hashes
1068
1069 $log_msg hash
1070 {
1071         msg => 'whitespace-formatted log entry
1072 ',                                              # trailing newline is preserved
1073         revision => '8',                        # integer
1074         date => '2004-02-24T17:01:44.108345Z',  # commit date
1075         author => 'committer name'
1076 };
1077
1078
1079 @mods = array of diff-index line hashes, each element represents one line
1080         of diff-index output
1081
1082 diff-index line ($m hash)
1083 {
1084         mode_a => first column of diff-index output, no leading ':',
1085         mode_b => second column of diff-index output,
1086         sha1_b => sha1sum of the final blob,
1087         chg => change type [MCRADT],
1088         file_a => original file name of a file (iff chg is 'C' or 'R')
1089         file_b => new/current file name of a file (any chg)
1090 }
1091 ;