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         $_find_copies_harder, $_l);
  35
  36 GetOptions(     'revision|r=s' => \$_revision,
  37                 'no-ignore-externals' => \$_no_ignore_ext,
  38                 'stdin|' => \$_stdin,
  39                 'edit|e' => \$_edit,
  40                 'rmdir' => \$_rmdir,
  41                 'help|H|h' => \$_help,
  42                 'find-copies-harder' => \$_find_copies_harder,
  43                 'l=i' => \$_l,
  44                 'no-stop-on-copy' => \$_no_stop_copy );
  45 my %cmd = (
  46         fetch => [ \&fetch, "Download new revisions from SVN" ],
  47         init => [ \&init, "Initialize and fetch (import)"],
  48         commit => [ \&commit, "Commit git revisions to SVN" ],
  49         rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)" ],
  50         help => [ \&usage, "Show help" ],
  51 );
  52 my $cmd;
  53 for (my $i = 0; $i < @ARGV; $i++) {
  54         if (defined $cmd{$ARGV[$i]}) {
  55                 $cmd = $ARGV[$i];
  56                 splice @ARGV, $i, 1;
  57                 last;
  58         }
  59 };
  60
  61 # we may be called as git-svn-(command), or git-svn(command).
  62 foreach (keys %cmd) {
  63         if (/git\-svn\-?($_)(?:\.\w+)?$/) {
  64                 $cmd = $1;
  65                 last;
  66         }
  67 }
  68 usage(0) if $_help;
  69 usage(1) unless (defined $cmd);
  70 svn_check_ignore_externals();
  71 $cmd{$cmd}->[0]->(@ARGV);
  72 exit 0;
  73
  74 ####################### primary functions ######################
  75 sub usage {
  76         my $exit = shift || 0;
  77         my $fd = $exit ? \*STDERR : \*STDOUT;
  78         print $fd <<"";
  79 git-svn - bidirectional operations between a single Subversion tree and git
  80 Usage: $0 <command> [options] [arguments]\n
  81 Available commands:
  82
  83         foreach (sort keys %cmd) {
  84                 print $fd '  ',pack('A10',$_),$cmd{$_}->[1],"\n";
  85         }
  86         print $fd <<"";
  87 \nGIT_SVN_ID may be set in the environment to an arbitrary identifier if
  88 you're tracking multiple SVN branches/repositories in one git repository
  89 and want to keep them separate.
  90
  91         exit $exit;
  92 }
  93
  94 sub rebuild {
  95         $SVN_URL = shift or undef;
  96         my $repo_uuid;
  97         my $newest_rev = 0;
  98
  99         my $pid = open(my $rev_list,'-|');
 100         defined $pid or croak $!;
 101         if ($pid == 0) {
 102                 exec("git-rev-list","$GIT_SVN-HEAD") or croak $!;
 103         }
 104         my $first;
 105         while (<$rev_list>) {
 106                 chomp;
 107                 my $c = $_;
 108                 croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
 109                 my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
 110                 next if (!@commit); # skip merges
 111                 my $id = $commit[$#commit];
 112                 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 113                                                 \s([a-f\d\-]+)$/x);
 114                 if (!$rev || !$uuid || !$url) {
 115                         # some of the original repositories I made had
 116                         # indentifiers like this:
 117                         ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)
 118                                                         \@([a-f\d\-]+)/x);
 119                         if (!$rev || !$uuid) {
 120                                 croak "Unable to extract revision or UUID from ",
 121                                         "$c, $id\n";
 122                         }
 123                 }
 124                 print "r$rev = $c\n";
 125                 unless (defined $first) {
 126                         if (!$SVN_URL && !$url) {
 127                                 croak "SVN repository location required: $url\n";
 128                         }
 129                         $SVN_URL ||= $url;
 130                         $repo_uuid = setup_git_svn();
 131                         $first = $rev;
 132                 }
 133                 if ($uuid ne $repo_uuid) {
 134                         croak "Repository UUIDs do not match!\ngot: $uuid\n",
 135                                                 "expected: $repo_uuid\n";
 136                 }
 137                 assert_revision_eq_or_unknown($rev, $c);
 138                 sys('git-update-ref',"$GIT_SVN/revs/$rev",$c);
 139                 $newest_rev = $rev if ($rev > $newest_rev);
 140         }
 141         close $rev_list or croak $?;
 142         if (!chdir $SVN_WC) {
 143                 my @svn_co = ('svn','co',"-r$first");
 144                 push @svn_co, '--ignore-externals' unless $_no_ignore_ext;
 145                 sys(@svn_co, $SVN_URL, $SVN_WC);
 146                 chdir $SVN_WC or croak $!;
 147         }
 148
 149         $pid = fork;
 150         defined $pid or croak $!;
 151         if ($pid == 0) {
 152                 my @svn_up = qw(svn up);
 153                 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 154                 sys(@svn_up,"-r$newest_rev");
 155                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 156                 git_addremove();
 157                 exec('git-write-tree');
 158         }
 159         waitpid $pid, 0;
 160 }
 161
 162 sub init {
 163         $SVN_URL = shift or croak "SVN repository location required\n";
 164         unless (-d $GIT_DIR) {
 165                 sys('git-init-db');
 166         }
 167         setup_git_svn();
 168 }
 169
 170 sub fetch {
 171         my (@parents) = @_;
 172         $SVN_URL ||= file_to_s("$GIT_DIR/$GIT_SVN/info/url");
 173         my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
 174         unless ($_revision) {
 175                 $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
 176         }
 177         push @log_args, "-r$_revision";
 178         push @log_args, '--stop-on-copy' unless $_no_stop_copy;
 179
 180         eval { require XML::Simple or croak $! };
 181         my $svn_log = $@ ? svn_log_raw(@log_args) : svn_log_xml(@log_args);
 182         @$svn_log = sort { $a->{revision} <=> $b->{revision} } @$svn_log;
 183
 184         my $base = shift @$svn_log or croak "No base revision!\n";
 185         my $last_commit = undef;
 186         unless (-d $SVN_WC) {
 187                 my @svn_co = ('svn','co',"-r$base->{revision}");
 188                 push @svn_co,'--ignore-externals' unless $_no_ignore_ext;
 189                 sys(@svn_co, $SVN_URL, $SVN_WC);
 190                 chdir $SVN_WC or croak $!;
 191                 $last_commit = git_commit($base, @parents);
 192                 unless (-f "$GIT_DIR/refs/heads/master") {
 193                         sys(qw(git-update-ref refs/heads/master),$last_commit);
 194                 }
 195                 assert_svn_wc_clean($base->{revision}, $last_commit);
 196         } else {
 197                 chdir $SVN_WC or croak $!;
 198                 $last_commit = file_to_s("$REV_DIR/$base->{revision}");
 199         }
 200         my @svn_up = qw(svn up);
 201         push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 202         my $last_rev = $base->{revision};
 203         foreach my $log_msg (@$svn_log) {
 204                 assert_svn_wc_clean($last_rev, $last_commit);
 205                 $last_rev = $log_msg->{revision};
 206                 sys(@svn_up,"-r$last_rev");
 207                 $last_commit = git_commit($log_msg, $last_commit, @parents);
 208         }
 209         assert_svn_wc_clean($last_rev, $last_commit);
 210         return pop @$svn_log;
 211 }
 212
 213 sub commit {
 214         my (@commits) = @_;
 215         if ($_stdin || !@commits) {
 216                 print "Reading from stdin...\n";
 217                 @commits = ();
 218                 while (<STDIN>) {
 219                         if (/\b([a-f\d]{6,40})\b/) {
 220                                 unshift @commits, $1;
 221                         }
 222                 }
 223         }
 224         my @revs;
 225         foreach my $c (@commits) {
 226                 chomp(my @tmp = safe_qx('git-rev-parse',$c));
 227                 if (scalar @tmp == 1) {
 228                         push @revs, $tmp[0];
 229                 } elsif (scalar @tmp > 1) {
 230                         push @revs, reverse (safe_qx('git-rev-list',@tmp));
 231                 } else {
 232                         die "Failed to rev-parse $c\n";
 233                 }
 234         }
 235         chomp @revs;
 236
 237         fetch();
 238         chdir $SVN_WC or croak $!;
 239         my $svn_current_rev =  svn_info('.')->{'Last Changed Rev'};
 240         foreach my $c (@revs) {
 241                 print "Committing $c\n";
 242                 svn_checkout_tree($svn_current_rev, $c);
 243                 $svn_current_rev = svn_commit_tree($svn_current_rev, $c);
 244         }
 245         print "Done committing ",scalar @revs," revisions to SVN\n";
 246
 247 }
 248
 249 ########################### utility functions #########################
 250
 251 sub setup_git_svn {
 252         defined $SVN_URL or croak "SVN repository location required\n";
 253         unless (-d $GIT_DIR) {
 254                 croak "GIT_DIR=$GIT_DIR does not exist!\n";
 255         }
 256         mkpath(["$GIT_DIR/$GIT_SVN"]);
 257         mkpath(["$GIT_DIR/$GIT_SVN/info"]);
 258         mkpath([$REV_DIR]);
 259         s_to_file($SVN_URL,"$GIT_DIR/$GIT_SVN/info/url");
 260         my $uuid = svn_info($SVN_URL)->{'Repository UUID'} or
 261                                         croak "Repository UUID unreadable\n";
 262         s_to_file($uuid,"$GIT_DIR/$GIT_SVN/info/uuid");
 263
 264         open my $fd, '>>', "$GIT_DIR/$GIT_SVN/info/exclude" or croak $!;
 265         print $fd '.svn',"\n";
 266         close $fd or croak $!;
 267         return $uuid;
 268 }
 269
 270 sub assert_svn_wc_clean {
 271         my ($svn_rev, $commit) = @_;
 272         croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
 273         croak "$commit is not a sha1!\n" unless ($commit =~ /^$sha1$/o);
 274         my $svn_info = svn_info('.');
 275         if ($svn_rev != $svn_info->{'Last Changed Rev'}) {
 276                 croak "Expected r$svn_rev, got r",
 277                                 $svn_info->{'Last Changed Rev'},"\n";
 278         }
 279         my @status = grep(!/^Performing status on external/,(`svn status`));
 280         @status = grep(!/^\s*$/,@status);
 281         if (scalar @status) {
 282                 print STDERR "Tree ($SVN_WC) is not clean:\n";
 283                 print STDERR $_ foreach @status;
 284                 croak;
 285         }
 286         my ($tree_a) = grep(/^tree $sha1$/o,`git-cat-file commit $commit`);
 287         $tree_a =~ s/^tree //;
 288         chomp $tree_a;
 289         chomp(my $tree_b = `GIT_INDEX_FILE=$GIT_SVN_INDEX git-write-tree`);
 290         if ($tree_a ne $tree_b) {
 291                 croak "$svn_rev != $commit, $tree_a != $tree_b\n";
 292         }
 293 }
 294
 295 sub parse_diff_tree {
 296         my $diff_fh = shift;
 297         local $/ = "\0";
 298         my $state = 'meta';
 299         my @mods;
 300         while (<$diff_fh>) {
 301                 chomp $_; # this gets rid of the trailing "\0"
 302                 print $_,"\n";
 303                 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
 304                                         $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
 305                         push @mods, {   mode_a => $1, mode_b => $2,
 306                                         sha1_b => $3, chg => $4 };
 307                         if ($4 =~ /^(?:C|R)$/) {
 308                                 $state = 'file_a';
 309                         } else {
 310                                 $state = 'file_b';
 311                         }
 312                 } elsif ($state eq 'file_a') {
 313                         my $x = $mods[$#mods] or croak __LINE__,": Empty array\n";
 314                         if ($x->{chg} !~ /^(?:C|R)$/) {
 315                                 croak __LINE__,": Error parsing $_, $x->{chg}\n";
 316                         }
 317                         $x->{file_a} = $_;
 318                         $state = 'file_b';
 319                 } elsif ($state eq 'file_b') {
 320                         my $x = $mods[$#mods] or croak __LINE__,": Empty array\n";
 321                         if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
 322                                 croak __LINE__,": Error parsing $_, $x->{chg}\n";
 323                         }
 324                         if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
 325                                 croak __LINE__,": Error parsing $_, $x->{chg}\n";
 326                         }
 327                         $x->{file_b} = $_;
 328                         $state = 'meta';
 329                 } else {
 330                         croak __LINE__,": Error parsing $_\n";
 331                 }
 332         }
 333         close $diff_fh or croak $!;
 334         return \@mods;
 335 }
 336
 337 sub svn_check_prop_executable {
 338         my $m = shift;
 339         if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
 340                 sys(qw(svn propset svn:executable 1), $m->{file_b});
 341         } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
 342                 sys(qw(svn propdel svn:executable), $m->{file_b});
 343         }
 344 }
 345
 346 sub svn_ensure_parent_path {
 347         my $dir_b = dirname(shift);
 348         svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
 349         mkpath([$dir_b]) unless (-d $dir_b);
 350         sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
 351 }
 352
 353 sub svn_checkout_tree {
 354         my ($svn_rev, $commit) = @_;
 355         my $from = file_to_s("$REV_DIR/$svn_rev");
 356         assert_svn_wc_clean($svn_rev,$from);
 357         print "diff-tree '$from' '$commit'\n";
 358         my $pid = open my $diff_fh, '-|';
 359         defined $pid or croak $!;
 360         if ($pid == 0) {
 361                 my @diff_tree = qw(git-diff-tree -z -r -C);
 362                 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
 363                 push @diff_tree, "-l$_l" if defined $_l;
 364                 exec(@diff_tree, $from, $commit) or croak $!;
 365         }
 366         my $mods = parse_diff_tree($diff_fh);
 367         unless (@$mods) {
 368                 # git can do empty commits, SVN doesn't allow it...
 369                 return $svn_rev;
 370         }
 371         my %rm;
 372         foreach my $m (@$mods) {
 373                 if ($m->{chg} eq 'C') {
 374                         svn_ensure_parent_path( $m->{file_b} );
 375                         sys(qw(svn cp),         $m->{file_a}, $m->{file_b});
 376                         blob_to_file(           $m->{sha1_b}, $m->{file_b});
 377                         svn_check_prop_executable($m);
 378                 } elsif ($m->{chg} eq 'D') {
 379                         $rm{dirname $m->{file_b}}->{basename $m->{file_b}} = 1;
 380                         sys(qw(svn rm --force), $m->{file_b});
 381                 } elsif ($m->{chg} eq 'R') {
 382                         svn_ensure_parent_path( $m->{file_b} );
 383                         sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
 384                         blob_to_file(           $m->{sha1_b}, $m->{file_b});
 385                         svn_check_prop_executable($m);
 386                         $rm{dirname $m->{file_a}}->{basename $m->{file_a}} = 1;
 387                 } elsif ($m->{chg} eq 'M') {
 388                         if ($m->{mode_b} =~ /^120/ && $m->{mode_a} =~ /^120/) {
 389                                 unlink $m->{file_b} or croak $!;
 390                                 blob_to_symlink($m->{sha1_b}, $m->{file_b});
 391                         } else {
 392                                 blob_to_file($m->{sha1_b}, $m->{file_b});
 393                         }
 394                         svn_check_prop_executable($m);
 395                 } elsif ($m->{chg} eq 'T') {
 396                         sys(qw(svn rm --force),$m->{file_b});
 397                         if ($m->{mode_b} =~ /^120/ && $m->{mode_a} =~ /^100/) {
 398                                 blob_to_symlink($m->{sha1_b}, $m->{file_b});
 399                         } else {
 400                                 blob_to_file($m->{sha1_b}, $m->{file_b});
 401                         }
 402                         svn_check_prop_executable($m);
 403                         sys(qw(svn add --force), $m->{file_b});
 404                 } elsif ($m->{chg} eq 'A') {
 405                         svn_ensure_parent_path( $m->{file_b} );
 406                         blob_to_file(           $m->{sha1_b}, $m->{file_b});
 407                         if ($m->{mode_b} =~ /755$/) {
 408                                 chmod 0755, $m->{file_b};
 409                         }
 410                         sys(qw(svn add --force), $m->{file_b});
 411                 } else {
 412                         croak "Invalid chg: $m->{chg}\n";
 413                 }
 414         }
 415         if ($_rmdir) {
 416                 my $old_index = $ENV{GIT_INDEX_FILE};
 417                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 418                 foreach my $dir (keys %rm) {
 419                         my $files = $rm{$dir};
 420                         my @files;
 421                         foreach (safe_qx('svn','ls',$dir)) {
 422                                 chomp;
 423                                 push @files, $_ unless $files->{$_};
 424                         }
 425                         sys(qw(svn rm),$dir) unless @files;
 426                 }
 427                 if ($old_index) {
 428                         $ENV{GIT_INDEX_FILE} = $old_index;
 429                 } else {
 430                         delete $ENV{GIT_INDEX_FILE};
 431                 }
 432         }
 433 }
 434
 435 sub svn_commit_tree {
 436         my ($svn_rev, $commit) = @_;
 437         my $commit_msg = "$GIT_DIR/$GIT_SVN/.svn-commit.tmp.$$";
 438         open my $msg, '>', $commit_msg  or croak $!;
 439
 440         chomp(my $type = `git-cat-file -t $commit`);
 441         if ($type eq 'commit') {
 442                 my $pid = open my $msg_fh, '-|';
 443                 defined $pid or croak $!;
 444
 445                 if ($pid == 0) {
 446                         exec(qw(git-cat-file commit), $commit) or croak $!;
 447                 }
 448                 my $in_msg = 0;
 449                 while (<$msg_fh>) {
 450                         if (!$in_msg) {
 451                                 $in_msg = 1 if (/^\s*$/);
 452                         } else {
 453                                 print $msg $_ or croak $!;
 454                         }
 455                 }
 456                 close $msg_fh or croak $!;
 457         }
 458         close $msg or croak $!;
 459
 460         if ($_edit || ($type eq 'tree')) {
 461                 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 462                 system($editor, $commit_msg);
 463         }
 464         my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
 465         my ($committed) = grep(/^Committed revision \d+\./,@ci_output);
 466         unlink $commit_msg;
 467         defined $committed or croak
 468                         "Commit output failed to parse committed revision!\n",
 469                         join("\n",@ci_output),"\n";
 470         my ($rev_committed) = ($committed =~ /^Committed revision (\d+)\./);
 471
 472         # resync immediately
 473         my @svn_up = (qw(svn up), "-r$svn_rev");
 474         push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
 475         sys(@svn_up);
 476         return fetch("$rev_committed=$commit")->{revision};
 477 }
 478
 479 sub svn_log_xml {
 480         my (@log_args) = @_;
 481         my $log_fh = IO::File->new_tmpfile or croak $!;
 482
 483         my $pid = fork;
 484         defined $pid or croak $!;
 485
 486         if ($pid == 0) {
 487                 open STDOUT, '>&', $log_fh or croak $!;
 488                 exec (qw(svn log --xml), @log_args) or croak $!
 489         }
 490
 491         waitpid $pid, 0;
 492         croak $? if $?;
 493
 494         seek $log_fh, 0, 0;
 495         my @svn_log;
 496         my $log = XML::Simple::XMLin( $log_fh,
 497                                 ForceArray => ['path','revision','logentry'],
 498                                 KeepRoot => 0,
 499                                 KeyAttr => {    logentry => '+revision',
 500                                                 paths => '+path' },
 501                         )->{logentry};
 502         foreach my $r (sort {$a <=> $b} keys %$log) {
 503                 my $log_msg = $log->{$r};
 504                 my ($Y,$m,$d,$H,$M,$S) = ($log_msg->{date} =~
 505                                         /(\d{4})\-(\d\d)\-(\d\d)T
 506                                          (\d\d)\:(\d\d)\:(\d\d)\.\d+Z$/x)
 507                                          or croak "Failed to parse date: ",
 508                                                  $log->{$r}->{date};
 509                 $log_msg->{date} = "+0000 $Y-$m-$d $H:$M:$S";
 510
 511                 # XML::Simple can't handle <msg></msg> as a string:
 512                 if (ref $log_msg->{msg} eq 'HASH') {
 513                         $log_msg->{msg} = "\n";
 514                 } else {
 515                         $log_msg->{msg} .= "\n";
 516                 }
 517                 push @svn_log, $log->{$r};
 518         }
 519         return \@svn_log;
 520 }
 521
 522 sub svn_log_raw {
 523         my (@log_args) = @_;
 524         my $pid = open my $log_fh,'-|';
 525         defined $pid or croak $!;
 526
 527         if ($pid == 0) {
 528                 exec (qw(svn log), @log_args) or croak $!
 529         }
 530
 531         my @svn_log;
 532         my $state;
 533         while (<$log_fh>) {
 534                 chomp;
 535                 if (/^\-{72}$/) {
 536                         $state = 'rev';
 537
 538                         # if we have an empty log message, put something there:
 539                         if (@svn_log) {
 540                                 $svn_log[$#svn_log]->{msg} ||= "\n";
 541                         }
 542                         next;
 543                 }
 544                 if ($state eq 'rev' && s/^r(\d+)\s*\|\s*//) {
 545                         my $rev = $1;
 546                         my ($author, $date) = split(/\s*\|\s*/, $_, 2);
 547                         my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
 548                                         /(\d{4})\-(\d\d)\-(\d\d)\s
 549                                          (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
 550                                          or croak "Failed to parse date: $date\n";
 551                         my %log_msg = ( revision => $rev,
 552                                         date => "$tz $Y-$m-$d $H:$M:$S",
 553                                         author => $author,
 554                                         msg => '' );
 555                         push @svn_log, \%log_msg;
 556                         $state = 'msg_start';
 557                         next;
 558                 }
 559                 # skip the first blank line of the message:
 560                 if ($state eq 'msg_start' && /^$/) {
 561                         $state = 'msg';
 562                 } elsif ($state eq 'msg') {
 563                         $svn_log[$#svn_log]->{msg} .= $_."\n";
 564                 }
 565         }
 566         close $log_fh or croak $?;
 567         return \@svn_log;
 568 }
 569
 570 sub svn_info {
 571         my $url = shift || $SVN_URL;
 572
 573         my $pid = open my $info_fh, '-|';
 574         defined $pid or croak $!;
 575
 576         if ($pid == 0) {
 577                 exec(qw(svn info),$url) or croak $!;
 578         }
 579
 580         my $ret = {};
 581         # only single-lines seem to exist in svn info output
 582         while (<$info_fh>) {
 583                 chomp $_;
 584                 if (m#^([^:]+)\s*:\s*(\S*)$#) {
 585                         $ret->{$1} = $2;
 586                         push @{$ret->{-order}}, $1;
 587                 }
 588         }
 589         close $info_fh or croak $!;
 590         return $ret;
 591 }
 592
 593 sub sys { system(@_) == 0 or croak $? }
 594
 595 sub git_addremove {
 596         system( "git-diff-files --name-only -z ".
 597                                 " | git-update-index --remove -z --stdin; ".
 598                 "git-ls-files -z --others ".
 599                         "'--exclude-from=$GIT_DIR/$GIT_SVN/info/exclude'".
 600                                 " | git-update-index --add -z --stdin; "
 601                 ) == 0 or croak $?
 602 }
 603
 604 sub s_to_file {
 605         my ($str, $file, $mode) = @_;
 606         open my $fd,'>',$file or croak $!;
 607         print $fd $str,"\n" or croak $!;
 608         close $fd or croak $!;
 609         chmod ($mode &~ umask, $file) if (defined $mode);
 610 }
 611
 612 sub file_to_s {
 613         my $file = shift;
 614         open my $fd,'<',$file or croak "$!: file: $file\n";
 615         local $/;
 616         my $ret = <$fd>;
 617         close $fd or croak $!;
 618         $ret =~ s/\s*$//s;
 619         return $ret;
 620 }
 621
 622 sub assert_revision_unknown {
 623         my $revno = shift;
 624         if (-f "$REV_DIR/$revno") {
 625                 croak "$REV_DIR/$revno already exists! ",
 626                                 "Why are we refetching it?";
 627         }
 628 }
 629
 630 sub assert_revision_eq_or_unknown {
 631         my ($revno, $commit) = @_;
 632         if (-f "$REV_DIR/$revno") {
 633                 my $current = file_to_s("$REV_DIR/$revno");
 634                 if ($commit ne $current) {
 635                         croak "$REV_DIR/$revno already exists!\n",
 636                                 "current: $current\nexpected: $commit\n";
 637                 }
 638                 return;
 639         }
 640 }
 641
 642 sub git_commit {
 643         my ($log_msg, @parents) = @_;
 644         assert_revision_unknown($log_msg->{revision});
 645         my $out_fh = IO::File->new_tmpfile or croak $!;
 646         my $info = svn_info('.');
 647         my $uuid = $info->{'Repository UUID'};
 648         defined $uuid or croak "Unable to get Repository UUID\n";
 649
 650         # commit parents can be conditionally bound to a particular
 651         # svn revision via: "svn_revno=commit_sha1", filter them out here:
 652         my @exec_parents;
 653         foreach my $p (@parents) {
 654                 next unless defined $p;
 655                 if ($p =~ /^(\d+)=($sha1_short)$/o) {
 656                         if ($1 == $log_msg->{revision}) {
 657                                 push @exec_parents, $2;
 658                         }
 659                 } else {
 660                         push @exec_parents, $p if $p =~ /$sha1_short/o;
 661                 }
 662         }
 663
 664         my $pid = fork;
 665         defined $pid or croak $!;
 666         if ($pid == 0) {
 667                 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
 668                 git_addremove();
 669                 chomp(my $tree = `git-write-tree`);
 670                 croak if $?;
 671                 my $msg_fh = IO::File->new_tmpfile or croak $!;
 672                 print $msg_fh $log_msg->{msg}, "\ngit-svn-id: ",
 673                                         "$SVN_URL\@$log_msg->{revision}",
 674                                         " $uuid\n" or croak $!;
 675                 $msg_fh->flush == 0 or croak $!;
 676                 seek $msg_fh, 0, 0 or croak $!;
 677
 678                 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} =
 679                                                 $log_msg->{author};
 680                 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
 681                                                 $log_msg->{author}."\@$uuid";
 682                 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} =
 683                                                 $log_msg->{date};
 684                 my @exec = ('git-commit-tree',$tree);
 685                 push @exec, '-p', $_  foreach @exec_parents;
 686                 open STDIN, '<&', $msg_fh or croak $!;
 687                 open STDOUT, '>&', $out_fh or croak $!;
 688                 exec @exec or croak $!;
 689         }
 690         waitpid($pid,0);
 691         croak if $?;
 692
 693         $out_fh->flush == 0 or croak $!;
 694         seek $out_fh, 0, 0 or croak $!;
 695         chomp(my $commit = do { local $/; <$out_fh> });
 696         if ($commit !~ /^$sha1$/o) {
 697                 croak "Failed to commit, invalid sha1: $commit\n";
 698         }
 699         my @update_ref = ('git-update-ref',"refs/heads/$GIT_SVN-HEAD",$commit);
 700         if (my $primary_parent = shift @exec_parents) {
 701                 push @update_ref, $primary_parent;
 702         }
 703         sys(@update_ref);
 704         sys('git-update-ref',"$GIT_SVN/revs/$log_msg->{revision}",$commit);
 705         print "r$log_msg->{revision} = $commit\n";
 706         return $commit;
 707 }
 708
 709 sub blob_to_symlink {
 710         my ($blob, $link) = @_;
 711         defined $link or croak "\$link not defined!\n";
 712         croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
 713         my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
 714         symlink $dest, $link or croak $!;
 715 }
 716
 717 sub blob_to_file {
 718         my ($blob, $file) = @_;
 719         defined $file or croak "\$file not defined!\n";
 720         croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
 721         open my $blob_fh, '>', $file or croak "$!: $file\n";
 722         my $pid = fork;
 723         defined $pid or croak $!;
 724
 725         if ($pid == 0) {
 726                 open STDOUT, '>&', $blob_fh or croak $!;
 727                 exec('git-cat-file','blob',$blob);
 728         }
 729         waitpid $pid, 0;
 730         croak $? if $?;
 731
 732         close $blob_fh or croak $!;
 733 }
 734
 735 sub safe_qx {
 736         my $pid = open my $child, '-|';
 737         defined $pid or croak $!;
 738         if ($pid == 0) {
 739                 exec(@_) or croak $?;
 740         }
 741         my @ret = (<$child>);
 742         close $child or croak $?;
 743         die $? if $?; # just in case close didn't error out
 744         return wantarray ? @ret : join('',@ret);
 745 }
 746
 747 sub svn_check_ignore_externals {
 748         return if $_no_ignore_ext;
 749         unless (grep /ignore-externals/,(safe_qx(qw(svn co -h)))) {
 750                 print STDERR "W: Installed svn version does not support ",
 751                                 "--ignore-externals\n";
 752                 $_no_ignore_ext = 1;
 753         }
 754 }
 755 __END__
 756
 757 Data structures:
 758
 759 @svn_log = array of log_msg hashes
 760
 761 $log_msg hash
 762 {
 763         msg => 'whitespace-formatted log entry
 764 ',                                              # trailing newline is preserved
 765         revision => '8',                        # integer
 766         date => '2004-02-24T17:01:44.108345Z',  # commit date
 767         author => 'committer name'
 768 };
 769
 770
 771 @mods = array of diff-index line hashes, each element represents one line
 772         of diff-index output
 773
 774 diff-index line ($m hash)
 775 {
 776         mode_a => first column of diff-index output, no leading ':',
 777         mode_b => second column of diff-index output,
 778         sha1_b => sha1sum of the final blob,
 779         chg => change type [MCRAD],
 780         file_a => original file name of a file (iff chg is 'C' or 'R')
 781         file_b => new/current file name of a file (any chg)
 782 }
 783 ;