contrib/git-svn/git-svn

   1 #!/usr/bin/env perl
   2 use warnings;
   3 use strict;
   4 use vars qw/    $AUTHOR $VERSION
   5                 $SVN_URL $SVN_INFO $SVN_WC
   6                 $GIT_SVN_INDEX $GIT_SVN
   7                 $GIT_DIR $REV_DIR/;
   8 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
   9 $VERSION = '0.9.0';
  10 $GIT_DIR = $ENV{GIT_DIR} || "$ENV{PWD}/.git";
  11 $GIT_SVN = $ENV{GIT_SVN_ID} || 'git-svn';
  12 $GIT_SVN_INDEX = "$GIT_DIR/$GIT_SVN/index";
  13 $ENV{GIT_DIR} ||= $GIT_DIR;
  14 $SVN_URL = undef;
  15 $REV_DIR = "$GIT_DIR/$GIT_SVN/revs";
  16 $SVN_WC = "$GIT_DIR/$GIT_SVN/tree";
  17
  18 # make sure the svn binary gives consistent output between locales and TZs:
  19 $ENV{TZ} = 'UTC';
  20 $ENV{LC_ALL} = 'C';
  21
  22 # If SVN:: library support is added, please make the dependencies
  23 # optional and preserve the capability to use the command-line client.
  24 # See what I do with XML::Simple to make the dependency optional.
  25 use Carp qw/croak/;
  26 use IO::File qw//;
  27 use File::Basename qw/dirname basename/;
  28 use File::Path qw/mkpath/;
  29 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  30 use File::Spec qw//;
  31 my $sha1 = qr/[a-f\d]{40}/;
  32 my $sha1_short = qr/[a-f\d]{6,40}/;
  33 my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit);
  34
  35 GetOptions(     'revision|r=s' => \$_revision,
  36                 'no-ignore-externals' => \$_no_ignore_ext,
  37                 'stdin|' => \$_stdin,
  38                 'edit|e' => \$_edit,
  39                 'rmdir' => \$_rmdir,
  40                 'help|H|h' => \$_help,
  41                 'no-stop-copy' => \$_no_stop_copy );
  42 my %cmd = (
  43         fetch => [ \&fetch, "Download new revisions from SVN" ],
  44         init => [ \&init, "Initialize and fetch (import)"],
  45         commit => [ \&commit, "Commit git revisions to SVN" ],
  46         rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)" ],
  47         help => [ \&usage, "Show help" ],
  48 );
  49 my $cmd;
  50 for (my $i = 0; $i < @ARGV; $i++) {
  51         if (defined $cmd{$ARGV[$i]}) {
  52                 $cmd = $ARGV[$i];
  53                 splice @ARGV, $i, 1;
  54                 last;
  55         }
  56 };
  57
  58 # we may be called as git-svn-(command), or git-svn(command).
  59 foreach (keys %cmd) {
  60         if (/git\-svn\-?($_)(?:\.\w+)?$/) {
  61                 $cmd = $1;
  62                 last;
  63         }
  64 }
  65 usage(0) if $_help;
  66 usage(1) unless (defined $cmd);
  67 svn_check_ignore_externals();
  68 $cmd{$cmd}->[0]->(@ARGV);
  69 exit 0;
  70
  71 ####################### primary functions ######################
  72 sub usage {
  73         my $exit = shift || 0;
  74         my $fd = $exit ? \*STDERR : \*STDOUT;
  75         print $fd <<"";
  76 git-svn - bidirectional operations between a single Subversion tree and git
  77 Usage: $0 <command> [options] [arguments]\n
  78 Available commands:
  79
  80         foreach (sort keys %cmd) {
  81                 print $fd '  ',pack('A10',$_),$cmd{$_}->[1],"\n";
  82         }
  83         print $fd <<"";
  84 \nGIT_SVN_ID may be set in the environment to an arbitrary identifier if
  85 you're tracking multiple SVN branches/repositories in one git repository
  86 and want to keep them separate.
  87
  88         exit $exit;
  89 }
  90
  91 sub rebuild {
  92         $SVN_URL = shift or undef;
  93         my $repo_uuid;
  94         my $newest_rev = 0;
  95
  96         my $pid = open(my $rev_list,'-|');
  97         defined $pid or croak $!;
  98         if ($pid == 0) {
  99                 exec("git-rev-list","$GIT_SVN-HEAD") or croak $!;
 100         }
 101         my $first;
 102         while (<$rev_list>) {
 103                 chomp;
 104                 my $c = $_;
 105                 croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 106                 my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
 107                 next if (!@commit); # skip merges
 108                 my $id = $commit[$#commit];
 109                 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 110                                                 \s([a-f\d\-]+)$/x);
 111                 if (!$rev || !$uuid || !$url) {
 112                         # some of the original repositories I made had
 113                         # indentifiers like this:
 114                         ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)
 115                                                         \@([a-f\d\-]+)/x);
 116                         if (!$rev || !$uuid) {
 117                                 croak "Unable to extract revision or UUID from ",
 118                                         "$c, $id\n";
 119                         }
 120                 }
 121                 print "r$rev = $c\n";
 122                 unless (defined $first) {
 123                         if (!$SVN_URL && !$url) {
 124                                 croak "SVN repository location required: $url\n";
 125                         }
 126                         $SVN_URL ||= $url;
 127                         $repo_uuid = setup_git_svn();
 128                         $first = $rev;
 129                 }
 130                 if ($uuid ne $repo_uuid) {
 131                         croak "Repository UUIDs do not match!\ngot: $uuid\n",
 132                                                 "expected: $repo_uuid\n";
 133                 }
 134                 assert_revision_eq_or_unknown($rev, $c);
 135                 sys('git-update-ref',"$GIT_SVN/revs/$rev",$c);
 136                 $newest_rev = $rev if ($rev > $newest_rev);
 137         }
 138         close $rev_list or croak $?;
 139         if (!chdir $SVN_WC) {
 140                 my @svn_co = ('svn','co',"-r$first");
 141                 push @svn_co, '--ignore-externals' unless $_no_ignore_ext;
 142                 sys(@svn_co, $SVN_URL, $SVN_WC);
 143                 chdir $SVN_WC or croak $!;
 144         }
 145
 146         $pid = fork;
 147         defined $pid or croak $!;
 148         if ($pid == 0) {
 149                 my @svn_up = qw(svn up);
 150                 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 151                 sys(@svn_up,"-r$newest_rev");
 152                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 153                 git_addremove();
 154                 exec('git-write-tree');
 155         }
 156         waitpid $pid, 0;
 157 }
 158
 159 sub init {
 160         $SVN_URL = shift or croak "SVN repository location required\n";
 161         unless (-d $GIT_DIR) {
 162                 sys('git-init-db');
 163         }
 164         setup_git_svn();
 165 }
 166
 167 sub fetch {
 168         my (@parents) = @_;
 169         $SVN_URL ||= file_to_s("$GIT_DIR/$GIT_SVN/info/url");
 170         my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 171         unless ($_revision) {
 172                 $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 173         }
 174         push @log_args, "-r$_revision";
 175         push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 176
 177         eval { require XML::Simple or croak $! };
 178         my $svn_log = $@ ? svn_log_raw(@log_args) : svn_log_xml(@log_args);
 179         @$svn_log = sort { $a->{revision} <=> $b->{revision} } @$svn_log;
 180
 181         my $base = shift @$svn_log or croak "No base revision!\n";
 182         my $last_commit = undef;
 183         unless (-d $SVN_WC) {
 184                 my @svn_co = ('svn','co',"-r$base->{revision}");
 185                 push @svn_co,'--ignore-externals' unless $_no_ignore_ext;
 186                 sys(@svn_co, $SVN_URL, $SVN_WC);
 187                 chdir $SVN_WC or croak $!;
 188                 $last_commit = git_commit($base, @parents);
 189                 unless (-f "$GIT_DIR/refs/heads/master") {
 190                         sys(qw(git-update-ref refs/heads/master),$last_commit);
 191                 }
 192                 assert_svn_wc_clean($base->{revision}, $last_commit);
 193         } else {
 194                 chdir $SVN_WC or croak $!;
 195                 $last_commit = file_to_s("$REV_DIR/$base->{revision}");
 196         }
 197         my @svn_up = qw(svn up);
 198         push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 199         my $last_rev = $base->{revision};
 200         foreach my $log_msg (@$svn_log) {
 201                 assert_svn_wc_clean($last_rev, $last_commit);
 202                 $last_rev = $log_msg->{revision};
 203                 sys(@svn_up,"-r$last_rev");
 204                 $last_commit = git_commit($log_msg, $last_commit, @parents);
 205         }
 206         assert_svn_wc_clean($last_rev, $last_commit);
 207         return pop @$svn_log;
 208 }
 209
 210 sub commit {
 211         my (@commits) = @_;
 212         if ($_stdin || !@commits) {
 213                 print "Reading from stdin...\n";
 214                 @commits = ();
 215                 while (<STDIN>) {
 216                         if (/^([a-f\d]{6,40})\b/) {
 217                                 unshift @commits, $1;
 218                         }
 219                 }
 220         }
 221         my @revs;
 222         foreach (@commits) {
 223                 push @revs, (safe_qx('git-rev-parse',$_));
 224         }
 225         chomp @revs;
 226
 227         fetch();
 228         chdir $SVN_WC or croak $!;
 229         my $svn_current_rev =  svn_info('.')->{'Last Changed Rev'};
 230         foreach my $c (@revs) {
 231                 print "Committing $c\n";
 232                 svn_checkout_tree($svn_current_rev, $c);
 233                 $svn_current_rev = svn_commit_tree($svn_current_rev, $c);
 234         }
 235         print "Done committing ",scalar @revs," revisions to SVN\n";
 236
 237 }
 238
 239 ########################### utility functions #########################
 240
 241 sub setup_git_svn {
 242         defined $SVN_URL or croak "SVN repository location required\n";
 243         unless (-d $GIT_DIR) {
 244                 croak "GIT_DIR=$GIT_DIR does not exist!\n";
 245         }
 246         mkpath(["$GIT_DIR/$GIT_SVN"]);
 247         mkpath(["$GIT_DIR/$GIT_SVN/info"]);
 248         mkpath([$REV_DIR]);
 249         s_to_file($SVN_URL,"$GIT_DIR/$GIT_SVN/info/url");
 250         my $uuid = svn_info($SVN_URL)->{'Repository UUID'} or
 251                                         croak "Repository UUID unreadable\n";
 252         s_to_file($uuid,"$GIT_DIR/$GIT_SVN/info/uuid");
 253
 254         open my $fd, '>>', "$GIT_DIR/$GIT_SVN/info/exclude" or croak $!;
 255         print $fd '.svn',"\n";
 256         close $fd or croak $!;
 257         return $uuid;
 258 }
 259
 260 sub assert_svn_wc_clean {
 261         my ($svn_rev, $commit) = @_;
 262         croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
 263         croak "$commit is not a sha1!\n" unless ($commit =~ /^$sha1$/o);
 264         my $svn_info = svn_info('.');
 265         if ($svn_rev != $svn_info->{'Last Changed Rev'}) {
 266                 croak "Expected r$svn_rev, got r",
 267                                 $svn_info->{'Last Changed Rev'},"\n";
 268         }
 269         my @status = grep(!/^Performing status on external/,(`svn status`));
 270         @status = grep(!/^\s*$/,@status);
 271         if (scalar @status) {
 272                 print STDERR "Tree ($SVN_WC) is not clean:\n";
 273                 print STDERR $_ foreach @status;
 274                 croak;
 275         }
 276         my ($tree_a) = grep(/^tree $sha1$/o,`git-cat-file commit $commit`);
 277         $tree_a =~ s/^tree //;
 278         chomp $tree_a;
 279         chomp(my $tree_b = `GIT_INDEX_FILE=$GIT_SVN_INDEX git-write-tree`);
 280         if ($tree_a ne $tree_b) {
 281                 croak "$svn_rev != $commit, $tree_a != $tree_b\n";
 282         }
 283 }
 284
 285 sub parse_diff_tree {
 286         my $diff_fh = shift;
 287         local $/ = "\0";
 288         my $state = 'meta';
 289         my @mods;
 290         while (<$diff_fh>) {
 291                 chomp $_; # this gets rid of the trailing "\0"
 292                 print $_,"\n";
 293                 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
 294                                         $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
 295                         push @mods, {   mode_a => $1, mode_b => $2,
 296                                         sha1_b => $3, chg => $4 };
 297                         if ($4 =~ /^(?:C|R)$/) {
 298                                 $state = 'file_a';
 299                         } else {
 300                                 $state = 'file_b';
 301                         }
 302                 } elsif ($state eq 'file_a') {
 303                         my $x = $mods[$#mods] or croak __LINE__,": Empty array\n";
 304                         if ($x->{chg} !~ /^(?:C|R)$/) {
 305                                 croak __LINE__,": Error parsing $_, $x->{chg}\n";
 306                         }
 307                         $x->{file_a} = $_;
 308                         $state = 'file_b';
 309                 } elsif ($state eq 'file_b') {
 310                         my $x = $mods[$#mods] or croak __LINE__,": Empty array\n";
 311                         if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
 312                                 croak __LINE__,": Error parsing $_, $x->{chg}\n";
 313                         }
 314                         if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
 315                                 croak __LINE__,": Error parsing $_, $x->{chg}\n";
 316                         }
 317                         $x->{file_b} = $_;
 318                         $state = 'meta';
 319                 } else {
 320                         croak __LINE__,": Error parsing $_\n";
 321                 }
 322         }
 323         close $diff_fh or croak $!;
 324         return \@mods;
 325 }
 326
 327 sub svn_check_prop_executable {
 328         my $m = shift;
 329         if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
 330                 sys(qw(svn propset svn:executable 1), $m->{file_b});
 331         } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
 332                 sys(qw(svn propdel svn:executable), $m->{file_b});
 333         }
 334 }
 335
 336 sub svn_ensure_parent_path {
 337         my $dir_b = dirname(shift);
 338         svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
 339         mkpath([$dir_b]) unless (-d $dir_b);
 340         sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
 341 }
 342
 343 sub svn_checkout_tree {
 344         my ($svn_rev, $commit) = @_;
 345         my $from = file_to_s("$REV_DIR/$svn_rev");
 346         assert_svn_wc_clean($svn_rev,$from);
 347         print "diff-tree '$from' '$commit'\n";
 348         my $pid = open my $diff_fh, '-|';
 349         defined $pid or croak $!;
 350         if ($pid == 0) {
 351                 exec(qw(git-diff-tree -z -r -C), $from, $commit) or croak $!;
 352         }
 353         my $mods = parse_diff_tree($diff_fh);
 354         unless (@$mods) {
 355                 # git can do empty commits, SVN doesn't allow it...
 356                 return $svn_rev;
 357         }
 358         my %rm;
 359         foreach my $m (@$mods) {
 360                 if ($m->{chg} eq 'C') {
 361                         svn_ensure_parent_path( $m->{file_b} );
 362                         sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
 363                         blob_to_file(           $m->{sha1_b}, $m->{file_b});
 364                         svn_check_prop_executable($m);
 365                 } elsif ($m->{chg} eq 'D') {
 366                         $rm{dirname $m->{file_b}}->{basename $m->{file_b}} = 1;
 367                         sys(qw(svn rm --force), $m->{file_b});
 368                 } elsif ($m->{chg} eq 'R') {
 369                         svn_ensure_parent_path( $m->{file_b} );
 370                         sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
 371                         blob_to_file(           $m->{sha1_b}, $m->{file_b});
 372                         svn_check_prop_executable($m);
 373                         $rm{dirname $m->{file_a}}->{basename $m->{file_a}} = 1;
 374                 } elsif ($m->{chg} eq 'M') {
 375                         if ($m->{mode_b} =~ /^120/ && $m->{mode_a} =~ /^120/) {
 376                                 unlink $m->{file_b} or croak $!;
 377                                 blob_to_symlink($m->{sha1_b}, $m->{file_b});
 378                         } else {
 379                                 blob_to_file($m->{sha1_b}, $m->{file_b});
 380                         }
 381                         svn_check_prop_executable($m);
 382                 } elsif ($m->{chg} eq 'T') {
 383                         sys(qw(svn rm --force),$m->{file_b});
 384                         if ($m->{mode_b} =~ /^120/ && $m->{mode_a} =~ /^100/) {
 385                                 blob_to_symlink($m->{sha1_b}, $m->{file_b});
 386                         } else {
 387                                 blob_to_file($m->{sha1_b}, $m->{file_b});
 388                         }
 389                         svn_check_prop_executable($m);
 390                         sys(qw(svn add --force), $m->{file_b});
 391                 } elsif ($m->{chg} eq 'A') {
 392                         svn_ensure_parent_path( $m->{file_b} );
 393                         blob_to_file(           $m->{sha1_b}, $m->{file_b});
 394                         if ($m->{mode_b} =~ /755$/) {
 395                                 chmod 0755, $m->{file_b};
 396                         }
 397                         sys(qw(svn add --force), $m->{file_b});
 398                 } else {
 399                         croak "Invalid chg: $m->{chg}\n";
 400                 }
 401         }
 402         if ($_rmdir) {
 403                 my $old_index = $ENV{GIT_INDEX_FILE};
 404                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 405                 foreach my $dir (keys %rm) {
 406                         my $files = $rm{$dir};
 407                         my @files;
 408                         foreach (safe_qx('svn','ls',$dir)) {
 409                                 chomp;
 410                                 push @files, $_ unless $files->{$_};
 411                         }
 412                         sys(qw(svn rm),$dir) unless @files;
 413                 }
 414                 if ($old_index) {
 415                         $ENV{GIT_INDEX_FILE} = $old_index;
 416                 } else {
 417                         delete $ENV{GIT_INDEX_FILE};
 418                 }
 419         }
 420 }
 421
 422 sub svn_commit_tree {
 423         my ($svn_rev, $commit) = @_;
 424         my $commit_msg = "$GIT_DIR/$GIT_SVN/.svn-commit.tmp.$$";
 425         open my $msg, '>', $commit_msg  or croak $!;
 426
 427         chomp(my $type = `git-cat-file -t $commit`);
 428         if ($type eq 'commit') {
 429                 my $pid = open my $msg_fh, '-|';
 430                 defined $pid or croak $!;
 431
 432                 if ($pid == 0) {
 433                         exec(qw(git-cat-file commit), $commit) or croak $!;
 434                 }
 435                 my $in_msg = 0;
 436                 while (<$msg_fh>) {
 437                         if (!$in_msg) {
 438                                 $in_msg = 1 if (/^\s*$/);
 439                         } else {
 440                                 print $msg $_ or croak $!;
 441                         }
 442                 }
 443                 close $msg_fh or croak $!;
 444         }
 445         close $msg or croak $!;
 446
 447         if ($_edit || ($type eq 'tree')) {
 448                 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 449                 system($editor, $commit_msg);
 450         }
 451         my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
 452         my ($committed) = grep(/^Committed revision \d+\./,@ci_output);
 453         unlink $commit_msg;
 454         defined $committed or croak
 455                         "Commit output failed to parse committed revision!\n",
 456                         join("\n",@ci_output),"\n";
 457         my ($rev_committed) = ($committed =~ /^Committed revision (\d+)\./);
 458
 459         # resync immediately
 460         my @svn_up = (qw(svn up), "-r$svn_rev");
 461         push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 462         sys(@svn_up);
 463         return fetch("$rev_committed=$commit")->{revision};
 464 }
 465
 466 sub svn_log_xml {
 467         my (@log_args) = @_;
 468         my $log_fh = IO::File->new_tmpfile or croak $!;
 469
 470         my $pid = fork;
 471         defined $pid or croak $!;
 472
 473         if ($pid == 0) {
 474                 open STDOUT, '>&', $log_fh or croak $!;
 475                 exec (qw(svn log --xml), @log_args) or croak $!
 476         }
 477
 478         waitpid $pid, 0;
 479         croak $? if $?;
 480
 481         seek $log_fh, 0, 0;
 482         my @svn_log;
 483         my $log = XML::Simple::XMLin( $log_fh,
 484                                 ForceArray => ['path','revision','logentry'],
 485                                 KeepRoot => 0,
 486                                 KeyAttr => {    logentry => '+revision',
 487                                                 paths => '+path' },
 488                         )->{logentry};
 489         foreach my $r (sort {$a <=> $b} keys %$log) {
 490                 my $log_msg = $log->{$r};
 491                 my ($Y,$m,$d,$H,$M,$S) = ($log_msg->{date} =~
 492                                         /(\d{4})\-(\d\d)\-(\d\d)T
 493                                          (\d\d)\:(\d\d)\:(\d\d)\.\d+Z$/x)
 494                                          or croak "Failed to parse date: ",
 495                                                  $log->{$r}->{date};
 496                 $log_msg->{date} = "+0000 $Y-$m-$d $H:$M:$S";
 497
 498                 # XML::Simple can't handle <msg></msg> as a string:
 499                 if (ref $log_msg->{msg} eq 'HASH') {
 500                         $log_msg->{msg} = "\n";
 501                 } else {
 502                         $log_msg->{msg} .= "\n";
 503                 }
 504                 push @svn_log, $log->{$r};
 505         }
 506         return \@svn_log;
 507 }
 508
 509 sub svn_log_raw {
 510         my (@log_args) = @_;
 511         my $pid = open my $log_fh,'-|';
 512         defined $pid or croak $!;
 513
 514         if ($pid == 0) {
 515                 exec (qw(svn log), @log_args) or croak $!
 516         }
 517
 518         my @svn_log;
 519         my $state;
 520         while (<$log_fh>) {
 521                 chomp;
 522                 if (/^\-{72}$/) {
 523                         $state = 'rev';
 524
 525                         # if we have an empty log message, put something there:
 526                         if (@svn_log) {
 527                                 $svn_log[$#svn_log]->{msg} ||= "\n";
 528                         }
 529                         next;
 530                 }
 531                 if ($state eq 'rev' && s/^r(\d+)\s*\|\s*//) {
 532                         my $rev = $1;
 533                         my ($author, $date) = split(/\s*\|\s*/, $_, 2);
 534                         my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
 535                                         /(\d{4})\-(\d\d)\-(\d\d)\s
 536                                          (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
 537                                          or croak "Failed to parse date: $date\n";
 538                         my %log_msg = ( revision => $rev,
 539                                         date => "$tz $Y-$m-$d $H:$M:$S",
 540                                         author => $author,
 541                                         msg => '' );
 542                         push @svn_log, \%log_msg;
 543                         $state = 'msg_start';
 544                         next;
 545                 }
 546                 # skip the first blank line of the message:
 547                 if ($state eq 'msg_start' && /^$/) {
 548                         $state = 'msg';
 549                 } elsif ($state eq 'msg') {
 550                         $svn_log[$#svn_log]->{msg} .= $_."\n";
 551                 }
 552         }
 553         close $log_fh or croak $?;
 554         return \@svn_log;
 555 }
 556
 557 sub svn_info {
 558         my $url = shift || $SVN_URL;
 559
 560         my $pid = open my $info_fh, '-|';
 561         defined $pid or croak $!;
 562
 563         if ($pid == 0) {
 564                 exec(qw(svn info),$url) or croak $!;
 565         }
 566
 567         my $ret = {};
 568         # only single-lines seem to exist in svn info output
 569         while (<$info_fh>) {
 570                 chomp $_;
 571                 if (m#^([^:]+)\s*:\s*(\S*)$#) {
 572                         $ret->{$1} = $2;
 573                         push @{$ret->{-order}}, $1;
 574                 }
 575         }
 576         close $info_fh or croak $!;
 577         return $ret;
 578 }
 579
 580 sub sys { system(@_) == 0 or croak $? }
 581
 582 sub git_addremove {
 583         system( "git-diff-files --name-only -z ".
 584                                 " | git-update-index --remove -z --stdin; ".
 585                 "git-ls-files -z --others ".
 586                         "'--exclude-from=$GIT_DIR/$GIT_SVN/info/exclude'".
 587                                 " | git-update-index --add -z --stdin; "
 588                 ) == 0 or croak $?
 589 }
 590
 591 sub s_to_file {
 592         my ($str, $file, $mode) = @_;
 593         open my $fd,'>',$file or croak $!;
 594         print $fd $str,"\n" or croak $!;
 595         close $fd or croak $!;
 596         chmod ($mode &~ umask, $file) if (defined $mode);
 597 }
 598
 599 sub file_to_s {
 600         my $file = shift;
 601         open my $fd,'<',$file or croak "$!: file: $file\n";
 602         local $/;
 603         my $ret = <$fd>;
 604         close $fd or croak $!;
 605         $ret =~ s/\s*$//s;
 606         return $ret;
 607 }
 608
 609 sub assert_revision_unknown {
 610         my $revno = shift;
 611         if (-f "$REV_DIR/$revno") {
 612                 croak "$REV_DIR/$revno already exists! ",
 613                                 "Why are we refetching it?";
 614         }
 615 }
 616
 617 sub assert_revision_eq_or_unknown {
 618         my ($revno, $commit) = @_;
 619         if (-f "$REV_DIR/$revno") {
 620                 my $current = file_to_s("$REV_DIR/$revno");
 621                 if ($commit ne $current) {
 622                         croak "$REV_DIR/$revno already exists!\n",
 623                                 "current: $current\nexpected: $commit\n";
 624                 }
 625                 return;
 626         }
 627 }
 628
 629 sub git_commit {
 630         my ($log_msg, @parents) = @_;
 631         assert_revision_unknown($log_msg->{revision});
 632         my $out_fh = IO::File->new_tmpfile or croak $!;
 633         my $info = svn_info('.');
 634         my $uuid = $info->{'Repository UUID'};
 635         defined $uuid or croak "Unable to get Repository UUID\n";
 636
 637         # commit parents can be conditionally bound to a particular
 638         # svn revision via: "svn_revno=commit_sha1", filter them out here:
 639         my @exec_parents;
 640         foreach my $p (@parents) {
 641                 next unless defined $p;
 642                 if ($p =~ /^(\d+)=($sha1_short)$/o) {
 643                         if ($1 == $log_msg->{revision}) {
 644                                 push @exec_parents, $2;
 645                         }
 646                 } else {
 647                         push @exec_parents, $p if $p =~ /$sha1_short/o;
 648                 }
 649         }
 650
 651         my $pid = fork;
 652         defined $pid or croak $!;
 653         if ($pid == 0) {
 654                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 655                 git_addremove();
 656                 chomp(my $tree = `git-write-tree`);
 657                 croak if $?;
 658                 my $msg_fh = IO::File->new_tmpfile or croak $!;
 659                 print $msg_fh $log_msg->{msg}, "\ngit-svn-id: ",
 660                                         "$SVN_URL\@$log_msg->{revision}",
 661                                         " $uuid\n" or croak $!;
 662                 $msg_fh->flush == 0 or croak $!;
 663                 seek $msg_fh, 0, 0 or croak $!;
 664
 665                 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} =
 666                                                 $log_msg->{author};
 667                 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
 668                                                 $log_msg->{author}."\@$uuid";
 669                 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} =
 670                                                 $log_msg->{date};
 671                 my @exec = ('git-commit-tree',$tree);
 672                 push @exec, '-p', $_  foreach @exec_parents;
 673                 open STDIN, '<&', $msg_fh or croak $!;
 674                 open STDOUT, '>&', $out_fh or croak $!;
 675                 exec @exec or croak $!;
 676         }
 677         waitpid($pid,0);
 678         croak if $?;
 679
 680         $out_fh->flush == 0 or croak $!;
 681         seek $out_fh, 0, 0 or croak $!;
 682         chomp(my $commit = do { local $/; <$out_fh> });
 683         if ($commit !~ /^$sha1$/o) {
 684                 croak "Failed to commit, invalid sha1: $commit\n";
 685         }
 686         my @update_ref = ('git-update-ref',"refs/heads/$GIT_SVN-HEAD",$commit);
 687         if (my $primary_parent = shift @exec_parents) {
 688                 push @update_ref, $primary_parent;
 689         }
 690         sys(@update_ref);
 691         sys('git-update-ref',"$GIT_SVN/revs/$log_msg->{revision}",$commit);
 692         print "r$log_msg->{revision} = $commit\n";
 693         return $commit;
 694 }
 695
 696 sub blob_to_symlink {
 697         my ($blob, $link) = @_;
 698         defined $link or croak "\$link not defined!\n";
 699         croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
 700         my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
 701         symlink $dest, $link or croak $!;
 702 }
 703
 704 sub blob_to_file {
 705         my ($blob, $file) = @_;
 706         defined $file or croak "\$file not defined!\n";
 707         croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
 708         open my $blob_fh, '>', $file or croak "$!: $file\n";
 709         my $pid = fork;
 710         defined $pid or croak $!;
 711
 712         if ($pid == 0) {
 713                 open STDOUT, '>&', $blob_fh or croak $!;
 714                 exec('git-cat-file','blob',$blob);
 715         }
 716         waitpid $pid, 0;
 717         croak $? if $?;
 718
 719         close $blob_fh or croak $!;
 720 }
 721
 722 sub safe_qx {
 723         my $pid = open my $child, '-|';
 724         defined $pid or croak $!;
 725         if ($pid == 0) {
 726                 exec(@_) or croak $?;
 727         }
 728         my @ret = (<$child>);
 729         close $child or croak $?;
 730         die $? if $?; # just in case close didn't error out
 731         return wantarray ? @ret : join('',@ret);
 732 }
 733
 734 sub svn_check_ignore_externals {
 735         return if $_no_ignore_ext;
 736         unless (grep /ignore-externals/,(safe_qx(qw(svn co -h)))) {
 737                 print STDERR "W: Installed svn version does not support ",
 738                                 "--ignore-externals\n";
 739                 $_no_ignore_ext = 1;
 740         }
 741 }
 742 __END__
 743
 744 Data structures:
 745
 746 @svn_log = array of log_msg hashes
 747
 748 $log_msg hash
 749 {
 750         msg => 'whitespace-formatted log entry
 751 ',                                              # trailing newline is preserved
 752         revision => '8',                        # integer
 753         date => '2004-02-24T17:01:44.108345Z',  # commit date
 754         author => 'committer name'
 755 };
 756
 757
 758 @mods = array of diff-index line hashes, each element represents one line
 759         of diff-index output
 760
 761 diff-index line ($m hash)
 762 {
 763         mode_a => first column of diff-index output, no leading ':',
 764         mode_b => second column of diff-index output,
 765         sha1_b => sha1sum of the final blob,
 766         chg => change type [MCRAD],
 767         file_a => original file name of a file (iff chg is 'C' or 'R')
 768         file_b => new/current file name of a file (any chg)
 769 }
 770 ;