perl/Git/SVN.pm

   1 package Git::SVN;
   2 use strict;
   3 use warnings;
   4 use Fcntl qw/:DEFAULT :seek/;
   5 use constant rev_map_fmt => 'NH40';
   6 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
   7             $_repack $_repack_flags $_use_svm_props $_head
   8             $_use_svnsync_props $no_reuse_existing $_minimize_url
   9             $_use_log_author $_add_author_from $_localtime/;
  10 use Carp qw/croak/;
  11 use File::Path qw/mkpath/;
  12 use File::Copy qw/copy/;
  13 use IPC::Open3;
  14 use Time::Local;
  15 use Memoize;  # core since 5.8.0, Jul 2002
  16 use Memoize::Storable;
  17 use POSIX qw(:signal_h);
  18
  19 use Git qw(
  20     command
  21     command_oneline
  22     command_noisy
  23     command_output_pipe
  24     command_close_pipe
  25 );
  26 use Git::SVN::Utils qw(fatal can_compress);
  27
  28 my $can_use_yaml;
  29 BEGIN {
  30         $can_use_yaml = eval { require Git::SVN::Memoize::YAML; 1};
  31 }
  32
  33 my ($_gc_nr, $_gc_period);
  34
  35 # properties that we do not log:
  36 my %SKIP_PROP;
  37 BEGIN {
  38         %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
  39                                         svn:special svn:executable
  40                                         svn:entry:committed-rev
  41                                         svn:entry:last-author
  42                                         svn:entry:uuid
  43                                         svn:entry:committed-date/;
  44
  45         # some options are read globally, but can be overridden locally
  46         # per [svn-remote "..."] section.  Command-line options will *NOT*
  47         # override options set in an [svn-remote "..."] section
  48         no strict 'refs';
  49         for my $option (qw/follow_parent no_metadata use_svm_props
  50                            use_svnsync_props/) {
  51                 my $key = $option;
  52                 $key =~ tr/_//d;
  53                 my $prop = "-$option";
  54                 *$option = sub {
  55                         my ($self) = @_;
  56                         return $self->{$prop} if exists $self->{$prop};
  57                         my $k = "svn-remote.$self->{repo_id}.$key";
  58                         eval { command_oneline(qw/config --get/, $k) };
  59                         if ($@) {
  60                                 $self->{$prop} = ${"Git::SVN::_$option"};
  61                         } else {
  62                                 my $v = command_oneline(qw/config --bool/,$k);
  63                                 $self->{$prop} = $v eq 'false' ? 0 : 1;
  64                         }
  65                         return $self->{$prop};
  66                 }
  67         }
  68 }
  69
  70
  71 my (%LOCKFILES, %INDEX_FILES);
  72 END {
  73         unlink keys %LOCKFILES if %LOCKFILES;
  74         unlink keys %INDEX_FILES if %INDEX_FILES;
  75 }
  76
  77 sub resolve_local_globs {
  78         my ($url, $fetch, $glob_spec) = @_;
  79         return unless defined $glob_spec;
  80         my $ref = $glob_spec->{ref};
  81         my $path = $glob_spec->{path};
  82         foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
  83                 next unless m#^$ref->{regex}$#;
  84                 my $p = $1;
  85                 my $pathname = desanitize_refname($path->full_path($p));
  86                 my $refname = desanitize_refname($ref->full_path($p));
  87                 if (my $existing = $fetch->{$pathname}) {
  88                         if ($existing ne $refname) {
  89                                 die "Refspec conflict:\n",
  90                                     "existing: $existing\n",
  91                                     " globbed: $refname\n";
  92                         }
  93                         my $u = (::cmt_metadata("$refname"))[0];
  94                         $u =~ s!^\Q$url\E(/|$)!! or die
  95                           "$refname: '$url' not found in '$u'\n";
  96                         if ($pathname ne $u) {
  97                                 warn "W: Refspec glob conflict ",
  98                                      "(ref: $refname):\n",
  99                                      "expected path: $pathname\n",
 100                                      "    real path: $u\n",
 101                                      "Continuing ahead with $u\n";
 102                                 next;
 103                         }
 104                 } else {
 105                         $fetch->{$pathname} = $refname;
 106                 }
 107         }
 108 }
 109
 110 sub parse_revision_argument {
 111         my ($base, $head) = @_;
 112         if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
 113                 return ($base, $head);
 114         }
 115         return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
 116         return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
 117         return ($head, $head) if ($::_revision eq 'HEAD');
 118         return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
 119         return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
 120         die "revision argument: $::_revision not understood by git-svn\n";
 121 }
 122
 123 sub fetch_all {
 124         my ($repo_id, $remotes) = @_;
 125         if (ref $repo_id) {
 126                 my $gs = $repo_id;
 127                 $repo_id = undef;
 128                 $repo_id = $gs->{repo_id};
 129         }
 130         $remotes ||= read_all_remotes();
 131         my $remote = $remotes->{$repo_id} or
 132                      die "[svn-remote \"$repo_id\"] unknown\n";
 133         my $fetch = $remote->{fetch};
 134         my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
 135         my (@gs, @globs);
 136         my $ra = Git::SVN::Ra->new($url);
 137         my $uuid = $ra->get_uuid;
 138         my $head = $ra->get_latest_revnum;
 139
 140         # ignore errors, $head revision may not even exist anymore
 141         eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
 142         warn "W: $@\n" if $@;
 143
 144         my $base = defined $fetch ? $head : 0;
 145
 146         # read the max revs for wildcard expansion (branches/*, tags/*)
 147         foreach my $t (qw/branches tags/) {
 148                 defined $remote->{$t} or next;
 149                 push @globs, @{$remote->{$t}};
 150
 151                 my $max_rev = eval { tmp_config(qw/--int --get/,
 152                                          "svn-remote.$repo_id.${t}-maxRev") };
 153                 if (defined $max_rev && ($max_rev < $base)) {
 154                         $base = $max_rev;
 155                 } elsif (!defined $max_rev) {
 156                         $base = 0;
 157                 }
 158         }
 159
 160         if ($fetch) {
 161                 foreach my $p (sort keys %$fetch) {
 162                         my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
 163                         my $lr = $gs->rev_map_max;
 164                         if (defined $lr) {
 165                                 $base = $lr if ($lr < $base);
 166                         }
 167                         push @gs, $gs;
 168                 }
 169         }
 170
 171         ($base, $head) = parse_revision_argument($base, $head);
 172         $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
 173 }
 174
 175 sub read_all_remotes {
 176         my $r = {};
 177         my $use_svm_props = eval { command_oneline(qw/config --bool
 178             svn.useSvmProps/) };
 179         $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
 180         my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
 181         foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
 182                 if (m!^(.+)\.fetch=$svn_refspec$!) {
 183                         my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
 184                         die("svn-remote.$remote: remote ref '$remote_ref' "
 185                             . "must start with 'refs/'\n")
 186                                 unless $remote_ref =~ m{^refs/};
 187                         $local_ref = uri_decode($local_ref);
 188                         $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
 189                         $r->{$remote}->{svm} = {} if $use_svm_props;
 190                 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
 191                         $r->{$1}->{svm} = {};
 192                 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
 193                         $r->{$1}->{url} = $2;
 194                 } elsif (m!^(.+)\.pushurl=\s*(.*)\s*$!) {
 195                         $r->{$1}->{pushurl} = $2;
 196                 } elsif (m!^(.+)\.ignore-refs=\s*(.*)\s*$!) {
 197                         $r->{$1}->{ignore_refs_regex} = $2;
 198                 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
 199                         my ($remote, $t, $local_ref, $remote_ref) =
 200                                                              ($1, $2, $3, $4);
 201                         die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
 202                             . "must start with 'refs/'\n")
 203                                 unless $remote_ref =~ m{^refs/};
 204                         $local_ref = uri_decode($local_ref);
 205                         my $rs = {
 206                             t => $t,
 207                             remote => $remote,
 208                             path => Git::SVN::GlobSpec->new($local_ref, 1),
 209                             ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
 210                         if (length($rs->{ref}->{right}) != 0) {
 211                                 die "The '*' glob character must be the last ",
 212                                     "character of '$remote_ref'\n";
 213                         }
 214                         push @{ $r->{$remote}->{$t} }, $rs;
 215                 }
 216         }
 217
 218         map {
 219                 if (defined $r->{$_}->{svm}) {
 220                         my $svm;
 221                         eval {
 222                                 my $section = "svn-remote.$_";
 223                                 $svm = {
 224                                         source => tmp_config('--get',
 225                                             "$section.svm-source"),
 226                                         replace => tmp_config('--get',
 227                                             "$section.svm-replace"),
 228                                 }
 229                         };
 230                         $r->{$_}->{svm} = $svm;
 231                 }
 232         } keys %$r;
 233
 234         foreach my $remote (keys %$r) {
 235                 foreach ( grep { defined $_ }
 236                           map { $r->{$remote}->{$_} } qw(branches tags) ) {
 237                         foreach my $rs ( @$_ ) {
 238                                 $rs->{ignore_refs_regex} =
 239                                     $r->{$remote}->{ignore_refs_regex};
 240                         }
 241                 }
 242         }
 243
 244         $r;
 245 }
 246
 247 sub init_vars {
 248         $_gc_nr = $_gc_period = 1000;
 249         if (defined $_repack || defined $_repack_flags) {
 250                warn "Repack options are obsolete; they have no effect.\n";
 251         }
 252 }
 253
 254 sub verify_remotes_sanity {
 255         return unless -d $ENV{GIT_DIR};
 256         my %seen;
 257         foreach (command(qw/config -l/)) {
 258                 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
 259                         if ($seen{$1}) {
 260                                 die "Remote ref refs/remote/$1 is tracked by",
 261                                     "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
 262                                     "Please resolve this ambiguity in ",
 263                                     "your git configuration file before ",
 264                                     "continuing\n";
 265                         }
 266                         $seen{$1} = $_;
 267                 }
 268         }
 269 }
 270
 271 sub find_existing_remote {
 272         my ($url, $remotes) = @_;
 273         return undef if $no_reuse_existing;
 274         my $existing;
 275         foreach my $repo_id (keys %$remotes) {
 276                 my $u = $remotes->{$repo_id}->{url} or next;
 277                 next if $u ne $url;
 278                 $existing = $repo_id;
 279                 last;
 280         }
 281         $existing;
 282 }
 283
 284 sub init_remote_config {
 285         my ($self, $url, $no_write) = @_;
 286         $url =~ s!/+$!!; # strip trailing slash
 287         my $r = read_all_remotes();
 288         my $existing = find_existing_remote($url, $r);
 289         if ($existing) {
 290                 unless ($no_write) {
 291                         print STDERR "Using existing ",
 292                                      "[svn-remote \"$existing\"]\n";
 293                 }
 294                 $self->{repo_id} = $existing;
 295         } elsif ($_minimize_url) {
 296                 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
 297                 $existing = find_existing_remote($min_url, $r);
 298                 if ($existing) {
 299                         unless ($no_write) {
 300                                 print STDERR "Using existing ",
 301                                              "[svn-remote \"$existing\"]\n";
 302                         }
 303                         $self->{repo_id} = $existing;
 304                 }
 305                 if ($min_url ne $url) {
 306                         unless ($no_write) {
 307                                 print STDERR "Using higher level of URL: ",
 308                                              "$url => $min_url\n";
 309                         }
 310                         my $old_path = $self->{path};
 311                         $self->{path} = $url;
 312                         $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
 313                         if (length $old_path) {
 314                                 $self->{path} .= "/$old_path";
 315                         }
 316                         $url = $min_url;
 317                 }
 318         }
 319         my $orig_url;
 320         if (!$existing) {
 321                 # verify that we aren't overwriting anything:
 322                 $orig_url = eval {
 323                         command_oneline('config', '--get',
 324                                         "svn-remote.$self->{repo_id}.url")
 325                 };
 326                 if ($orig_url && ($orig_url ne $url)) {
 327                         die "svn-remote.$self->{repo_id}.url already set: ",
 328                             "$orig_url\nwanted to set to: $url\n";
 329                 }
 330         }
 331         my ($xrepo_id, $xpath) = find_ref($self->refname);
 332         if (!$no_write && defined $xpath) {
 333                 die "svn-remote.$xrepo_id.fetch already set to track ",
 334                     "$xpath:", $self->refname, "\n";
 335         }
 336         unless ($no_write) {
 337                 command_noisy('config',
 338                               "svn-remote.$self->{repo_id}.url", $url);
 339                 $self->{path} =~ s{^/}{};
 340                 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
 341                 command_noisy('config', '--add',
 342                               "svn-remote.$self->{repo_id}.fetch",
 343                               "$self->{path}:".$self->refname);
 344         }
 345         $self->{url} = $url;
 346 }
 347
 348 sub find_by_url { # repos_root and, path are optional
 349         my ($class, $full_url, $repos_root, $path) = @_;
 350
 351         return undef unless defined $full_url;
 352         remove_username($full_url);
 353         remove_username($repos_root) if defined $repos_root;
 354         my $remotes = read_all_remotes();
 355         if (defined $full_url && defined $repos_root && !defined $path) {
 356                 $path = $full_url;
 357                 $path =~ s#^\Q$repos_root\E(?:/|$)##;
 358         }
 359         foreach my $repo_id (keys %$remotes) {
 360                 my $u = $remotes->{$repo_id}->{url} or next;
 361                 remove_username($u);
 362                 next if defined $repos_root && $repos_root ne $u;
 363
 364                 my $fetch = $remotes->{$repo_id}->{fetch} || {};
 365                 foreach my $t (qw/branches tags/) {
 366                         foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
 367                                 resolve_local_globs($u, $fetch, $globspec);
 368                         }
 369                 }
 370                 my $p = $path;
 371                 my $rwr = rewrite_root({repo_id => $repo_id});
 372                 my $svm = $remotes->{$repo_id}->{svm}
 373                         if defined $remotes->{$repo_id}->{svm};
 374                 unless (defined $p) {
 375                         $p = $full_url;
 376                         my $z = $u;
 377                         my $prefix = '';
 378                         if ($rwr) {
 379                                 $z = $rwr;
 380                                 remove_username($z);
 381                         } elsif (defined $svm) {
 382                                 $z = $svm->{source};
 383                                 $prefix = $svm->{replace};
 384                                 $prefix =~ s#^\Q$u\E(?:/|$)##;
 385                                 $prefix =~ s#/$##;
 386                         }
 387                         $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
 388                 }
 389                 foreach my $f (keys %$fetch) {
 390                         next if $f ne $p;
 391                         return Git::SVN->new($fetch->{$f}, $repo_id, $f);
 392                 }
 393         }
 394         undef;
 395 }
 396
 397 sub init {
 398         my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
 399         my $self = _new($class, $repo_id, $ref_id, $path);
 400         if (defined $url) {
 401                 $self->init_remote_config($url, $no_write);
 402         }
 403         $self;
 404 }
 405
 406 sub find_ref {
 407         my ($ref_id) = @_;
 408         foreach (command(qw/config -l/)) {
 409                 next unless m!^svn-remote\.(.+)\.fetch=
 410                               \s*(.*?)\s*:\s*(.+?)\s*$!x;
 411                 my ($repo_id, $path, $ref) = ($1, $2, $3);
 412                 if ($ref eq $ref_id) {
 413                         $path = '' if ($path =~ m#^\./?#);
 414                         return ($repo_id, $path);
 415                 }
 416         }
 417         (undef, undef, undef);
 418 }
 419
 420 sub new {
 421         my ($class, $ref_id, $repo_id, $path) = @_;
 422         if (defined $ref_id && !defined $repo_id && !defined $path) {
 423                 ($repo_id, $path) = find_ref($ref_id);
 424                 if (!defined $repo_id) {
 425                         die "Could not find a \"svn-remote.*.fetch\" key ",
 426                             "in the repository configuration matching: ",
 427                             "$ref_id\n";
 428                 }
 429         }
 430         my $self = _new($class, $repo_id, $ref_id, $path);
 431         if (!defined $self->{path} || !length $self->{path}) {
 432                 my $fetch = command_oneline('config', '--get',
 433                                             "svn-remote.$repo_id.fetch",
 434                                             ":$ref_id\$") or
 435                      die "Failed to read \"svn-remote.$repo_id.fetch\" ",
 436                          "\":$ref_id\$\" in config\n";
 437                 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
 438         }
 439         $self->{path} =~ s{/+}{/}g;
 440         $self->{path} =~ s{\A/}{};
 441         $self->{path} =~ s{/\z}{};
 442         $self->{url} = command_oneline('config', '--get',
 443                                        "svn-remote.$repo_id.url") or
 444                   die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
 445         $self->{pushurl} = eval { command_oneline('config', '--get',
 446                                   "svn-remote.$repo_id.pushurl") };
 447         $self->rebuild;
 448         $self;
 449 }
 450
 451 sub refname {
 452         my ($refname) = $_[0]->{ref_id} ;
 453
 454         # It cannot end with a slash /, we'll throw up on this because
 455         # SVN can't have directories with a slash in their name, either:
 456         if ($refname =~ m{/$}) {
 457                 die "ref: '$refname' ends with a trailing slash, this is ",
 458                     "not permitted by git nor Subversion\n";
 459         }
 460
 461         # It cannot have ASCII control character space, tilde ~, caret ^,
 462         # colon :, question-mark ?, asterisk *, space, or open bracket [
 463         # anywhere.
 464         #
 465         # Additionally, % must be escaped because it is used for escaping
 466         # and we want our escaped refname to be reversible
 467         $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
 468
 469         # no slash-separated component can begin with a dot .
 470         # /.* becomes /%2E*
 471         $refname =~ s{/\.}{/%2E}g;
 472
 473         # It cannot have two consecutive dots .. anywhere
 474         # .. becomes %2E%2E
 475         $refname =~ s{\.\.}{%2E%2E}g;
 476
 477         # trailing dots and .lock are not allowed
 478         # .$ becomes %2E and .lock becomes %2Elock
 479         $refname =~ s{\.(?=$|lock$)}{%2E};
 480
 481         # the sequence @{ is used to access the reflog
 482         # @{ becomes %40{
 483         $refname =~ s{\@\{}{%40\{}g;
 484
 485         return $refname;
 486 }
 487
 488 sub desanitize_refname {
 489         my ($refname) = @_;
 490         $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
 491         return $refname;
 492 }
 493
 494 sub svm_uuid {
 495         my ($self) = @_;
 496         return $self->{svm}->{uuid} if $self->svm;
 497         $self->ra;
 498         unless ($self->{svm}) {
 499                 die "SVM UUID not cached, and reading remotely failed\n";
 500         }
 501         $self->{svm}->{uuid};
 502 }
 503
 504 sub svm {
 505         my ($self) = @_;
 506         return $self->{svm} if $self->{svm};
 507         my $svm;
 508         # see if we have it in our config, first:
 509         eval {
 510                 my $section = "svn-remote.$self->{repo_id}";
 511                 $svm = {
 512                   source => tmp_config('--get', "$section.svm-source"),
 513                   uuid => tmp_config('--get', "$section.svm-uuid"),
 514                   replace => tmp_config('--get', "$section.svm-replace"),
 515                 }
 516         };
 517         if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
 518                 $self->{svm} = $svm;
 519         }
 520         $self->{svm};
 521 }
 522
 523 sub _set_svm_vars {
 524         my ($self, $ra) = @_;
 525         return $ra if $self->svm;
 526
 527         my @err = ( "useSvmProps set, but failed to read SVM properties\n",
 528                     "(svm:source, svm:uuid) ",
 529                     "from the following URLs:\n" );
 530         sub read_svm_props {
 531                 my ($self, $ra, $path, $r) = @_;
 532                 my $props = ($ra->get_dir($path, $r))[2];
 533                 my $src = $props->{'svm:source'};
 534                 my $uuid = $props->{'svm:uuid'};
 535                 return undef if (!$src || !$uuid);
 536
 537                 chomp($src, $uuid);
 538
 539                 $uuid =~ m{^[0-9a-f\-]{30,}$}i
 540                     or die "doesn't look right - svm:uuid is '$uuid'\n";
 541
 542                 # the '!' is used to mark the repos_root!/relative/path
 543                 $src =~ s{/?!/?}{/};
 544                 $src =~ s{/+$}{}; # no trailing slashes please
 545                 # username is of no interest
 546                 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
 547
 548                 my $replace = $ra->{url};
 549                 $replace .= "/$path" if length $path;
 550
 551                 my $section = "svn-remote.$self->{repo_id}";
 552                 tmp_config("$section.svm-source", $src);
 553                 tmp_config("$section.svm-replace", $replace);
 554                 tmp_config("$section.svm-uuid", $uuid);
 555                 $self->{svm} = {
 556                         source => $src,
 557                         uuid => $uuid,
 558                         replace => $replace
 559                 };
 560         }
 561
 562         my $r = $ra->get_latest_revnum;
 563         my $path = $self->{path};
 564         my %tried;
 565         while (length $path) {
 566                 unless ($tried{"$self->{url}/$path"}) {
 567                         return $ra if $self->read_svm_props($ra, $path, $r);
 568                         $tried{"$self->{url}/$path"} = 1;
 569                 }
 570                 $path =~ s#/?[^/]+$##;
 571         }
 572         die "Path: '$path' should be ''\n" if $path ne '';
 573         return $ra if $self->read_svm_props($ra, $path, $r);
 574         $tried{"$self->{url}/$path"} = 1;
 575
 576         if ($ra->{repos_root} eq $self->{url}) {
 577                 die @err, (map { "  $_\n" } keys %tried), "\n";
 578         }
 579
 580         # nope, make sure we're connected to the repository root:
 581         my $ok;
 582         my @tried_b;
 583         $path = $ra->{svn_path};
 584         $ra = Git::SVN::Ra->new($ra->{repos_root});
 585         while (length $path) {
 586                 unless ($tried{"$ra->{url}/$path"}) {
 587                         $ok = $self->read_svm_props($ra, $path, $r);
 588                         last if $ok;
 589                         $tried{"$ra->{url}/$path"} = 1;
 590                 }
 591                 $path =~ s#/?[^/]+$##;
 592         }
 593         die "Path: '$path' should be ''\n" if $path ne '';
 594         $ok ||= $self->read_svm_props($ra, $path, $r);
 595         $tried{"$ra->{url}/$path"} = 1;
 596         if (!$ok) {
 597                 die @err, (map { "  $_\n" } keys %tried), "\n";
 598         }
 599         Git::SVN::Ra->new($self->{url});
 600 }
 601
 602 sub svnsync {
 603         my ($self) = @_;
 604         return $self->{svnsync} if $self->{svnsync};
 605
 606         if ($self->no_metadata) {
 607                 die "Can't have both 'noMetadata' and ",
 608                     "'useSvnsyncProps' options set!\n";
 609         }
 610         if ($self->rewrite_root) {
 611                 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
 612                     "options set!\n";
 613         }
 614         if ($self->rewrite_uuid) {
 615                 die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
 616                     "options set!\n";
 617         }
 618
 619         my $svnsync;
 620         # see if we have it in our config, first:
 621         eval {
 622                 my $section = "svn-remote.$self->{repo_id}";
 623
 624                 my $url = tmp_config('--get', "$section.svnsync-url");
 625                 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
 626                    die "doesn't look right - svn:sync-from-url is '$url'\n";
 627
 628                 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
 629                 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
 630                    die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
 631
 632                 $svnsync = { url => $url, uuid => $uuid }
 633         };
 634         if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
 635                 return $self->{svnsync} = $svnsync;
 636         }
 637
 638         my $err = "useSvnsyncProps set, but failed to read " .
 639                   "svnsync property: svn:sync-from-";
 640         my $rp = $self->ra->rev_proplist(0);
 641
 642         my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
 643         ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
 644                    die "doesn't look right - svn:sync-from-url is '$url'\n";
 645
 646         my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
 647         ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
 648                    die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
 649
 650         my $section = "svn-remote.$self->{repo_id}";
 651         tmp_config('--add', "$section.svnsync-uuid", $uuid);
 652         tmp_config('--add', "$section.svnsync-url", $url);
 653         return $self->{svnsync} = { url => $url, uuid => $uuid };
 654 }
 655
 656 # this allows us to memoize our SVN::Ra UUID locally and avoid a
 657 # remote lookup (useful for 'git svn log').
 658 sub ra_uuid {
 659         my ($self) = @_;
 660         unless ($self->{ra_uuid}) {
 661                 my $key = "svn-remote.$self->{repo_id}.uuid";
 662                 my $uuid = eval { tmp_config('--get', $key) };
 663                 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
 664                         $self->{ra_uuid} = $uuid;
 665                 } else {
 666                         die "ra_uuid called without URL\n" unless $self->{url};
 667                         $self->{ra_uuid} = $self->ra->get_uuid;
 668                         tmp_config('--add', $key, $self->{ra_uuid});
 669                 }
 670         }
 671         $self->{ra_uuid};
 672 }
 673
 674 sub _set_repos_root {
 675         my ($self, $repos_root) = @_;
 676         my $k = "svn-remote.$self->{repo_id}.reposRoot";
 677         $repos_root ||= $self->ra->{repos_root};
 678         tmp_config($k, $repos_root);
 679         $repos_root;
 680 }
 681
 682 sub repos_root {
 683         my ($self) = @_;
 684         my $k = "svn-remote.$self->{repo_id}.reposRoot";
 685         eval { tmp_config('--get', $k) } || $self->_set_repos_root;
 686 }
 687
 688 sub ra {
 689         my ($self) = shift;
 690         my $ra = Git::SVN::Ra->new($self->{url});
 691         $self->_set_repos_root($ra->{repos_root});
 692         if ($self->use_svm_props && !$self->{svm}) {
 693                 if ($self->no_metadata) {
 694                         die "Can't have both 'noMetadata' and ",
 695                             "'useSvmProps' options set!\n";
 696                 } elsif ($self->use_svnsync_props) {
 697                         die "Can't have both 'useSvnsyncProps' and ",
 698                             "'useSvmProps' options set!\n";
 699                 }
 700                 $ra = $self->_set_svm_vars($ra);
 701                 $self->{-want_revprops} = 1;
 702         }
 703         $ra;
 704 }
 705
 706 # prop_walk(PATH, REV, SUB)
 707 # -------------------------
 708 # Recursively traverse PATH at revision REV and invoke SUB for each
 709 # directory that contains a SVN property.  SUB will be invoked as
 710 # follows:  &SUB(gs, path, props);  where `gs' is this instance of
 711 # Git::SVN, `path' the path to the directory where the properties
 712 # `props' were found.  The `path' will be relative to point of checkout,
 713 # that is, if url://repo/trunk is the current Git branch, and that
 714 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
 715 # as `path' (note the trailing `/').
 716 sub prop_walk {
 717         my ($self, $path, $rev, $sub) = @_;
 718
 719         $path =~ s#^/##;
 720         my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
 721         $path =~ s#^/*#/#g;
 722         my $p = $path;
 723         # Strip the irrelevant part of the path.
 724         $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
 725         # Ensure the path is terminated by a `/'.
 726         $p =~ s#/*$#/#;
 727
 728         # The properties contain all the internal SVN stuff nobody
 729         # (usually) cares about.
 730         my $interesting_props = 0;
 731         foreach (keys %{$props}) {
 732                 # If it doesn't start with `svn:', it must be a
 733                 # user-defined property.
 734                 ++$interesting_props and next if $_ !~ /^svn:/;
 735                 # FIXME: Fragile, if SVN adds new public properties,
 736                 # this needs to be updated.
 737                 ++$interesting_props if /^svn:(?:ignore|keywords|executable
 738                                                  |eol-style|mime-type
 739                                                  |externals|needs-lock)$/x;
 740         }
 741         &$sub($self, $p, $props) if $interesting_props;
 742
 743         foreach (sort keys %$dirent) {
 744                 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
 745                 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
 746         }
 747 }
 748
 749 sub last_rev { ($_[0]->last_rev_commit)[0] }
 750 sub last_commit { ($_[0]->last_rev_commit)[1] }
 751
 752 # returns the newest SVN revision number and newest commit SHA1
 753 sub last_rev_commit {
 754         my ($self) = @_;
 755         if (defined $self->{last_rev} && defined $self->{last_commit}) {
 756                 return ($self->{last_rev}, $self->{last_commit});
 757         }
 758         my $c = ::verify_ref($self->refname.'^0');
 759         if ($c && !$self->use_svm_props && !$self->no_metadata) {
 760                 my $rev = (::cmt_metadata($c))[1];
 761                 if (defined $rev) {
 762                         ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
 763                         return ($rev, $c);
 764                 }
 765         }
 766         my $map_path = $self->map_path;
 767         unless (-e $map_path) {
 768                 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
 769                 return (undef, undef);
 770         }
 771         my ($rev, $commit) = $self->rev_map_max(1);
 772         ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
 773         return ($rev, $commit);
 774 }
 775
 776 sub get_fetch_range {
 777         my ($self, $min, $max) = @_;
 778         $max ||= $self->ra->get_latest_revnum;
 779         $min ||= $self->rev_map_max;
 780         (++$min, $max);
 781 }
 782
 783 sub tmp_config {
 784         my (@args) = @_;
 785         my $old_def_config = "$ENV{GIT_DIR}/svn/config";
 786         my $config = "$ENV{GIT_DIR}/svn/.metadata";
 787         if (! -f $config && -f $old_def_config) {
 788                 rename $old_def_config, $config or
 789                        die "Failed rename $old_def_config => $config: $!\n";
 790         }
 791         my $old_config = $ENV{GIT_CONFIG};
 792         $ENV{GIT_CONFIG} = $config;
 793         $@ = undef;
 794         my @ret = eval {
 795                 unless (-f $config) {
 796                         mkfile($config);
 797                         open my $fh, '>', $config or
 798                             die "Can't open $config: $!\n";
 799                         print $fh "; This file is used internally by ",
 800                                   "git-svn\n" or die
 801                                   "Couldn't write to $config: $!\n";
 802                         print $fh "; You should not have to edit it\n" or
 803                               die "Couldn't write to $config: $!\n";
 804                         close $fh or die "Couldn't close $config: $!\n";
 805                 }
 806                 command('config', @args);
 807         };
 808         my $err = $@;
 809         if (defined $old_config) {
 810                 $ENV{GIT_CONFIG} = $old_config;
 811         } else {
 812                 delete $ENV{GIT_CONFIG};
 813         }
 814         die $err if $err;
 815         wantarray ? @ret : $ret[0];
 816 }
 817
 818 sub tmp_index_do {
 819         my ($self, $sub) = @_;
 820         my $old_index = $ENV{GIT_INDEX_FILE};
 821         $ENV{GIT_INDEX_FILE} = $self->{index};
 822         $@ = undef;
 823         my @ret = eval {
 824                 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
 825                 mkpath([$dir]) unless -d $dir;
 826                 &$sub;
 827         };
 828         my $err = $@;
 829         if (defined $old_index) {
 830                 $ENV{GIT_INDEX_FILE} = $old_index;
 831         } else {
 832                 delete $ENV{GIT_INDEX_FILE};
 833         }
 834         die $err if $err;
 835         wantarray ? @ret : $ret[0];
 836 }
 837
 838 sub assert_index_clean {
 839         my ($self, $treeish) = @_;
 840
 841         $self->tmp_index_do(sub {
 842                 command_noisy('read-tree', $treeish) unless -e $self->{index};
 843                 my $x = command_oneline('write-tree');
 844                 my ($y) = (command(qw/cat-file commit/, $treeish) =~
 845                            /^tree ($::sha1)/mo);
 846                 return if $y eq $x;
 847
 848                 warn "Index mismatch: $y != $x\nrereading $treeish\n";
 849                 unlink $self->{index} or die "unlink $self->{index}: $!\n";
 850                 command_noisy('read-tree', $treeish);
 851                 $x = command_oneline('write-tree');
 852                 if ($y ne $x) {
 853                         fatal "trees ($treeish) $y != $x\n",
 854                               "Something is seriously wrong...";
 855                 }
 856         });
 857 }
 858
 859 sub get_commit_parents {
 860         my ($self, $log_entry) = @_;
 861         my (%seen, @ret, @tmp);
 862         # legacy support for 'set-tree'; this is only used by set_tree_cb:
 863         if (my $ip = $self->{inject_parents}) {
 864                 if (my $commit = delete $ip->{$log_entry->{revision}}) {
 865                         push @tmp, $commit;
 866                 }
 867         }
 868         if (my $cur = ::verify_ref($self->refname.'^0')) {
 869                 push @tmp, $cur;
 870         }
 871         if (my $ipd = $self->{inject_parents_dcommit}) {
 872                 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
 873                         push @tmp, @$commit;
 874                 }
 875         }
 876         push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
 877         while (my $p = shift @tmp) {
 878                 next if $seen{$p};
 879                 $seen{$p} = 1;
 880                 push @ret, $p;
 881         }
 882         @ret;
 883 }
 884
 885 sub rewrite_root {
 886         my ($self) = @_;
 887         return $self->{-rewrite_root} if exists $self->{-rewrite_root};
 888         my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
 889         my $rwr = eval { command_oneline(qw/config --get/, $k) };
 890         if ($rwr) {
 891                 $rwr =~ s#/+$##;
 892                 if ($rwr !~ m#^[a-z\+]+://#) {
 893                         die "$rwr is not a valid URL (key: $k)\n";
 894                 }
 895         }
 896         $self->{-rewrite_root} = $rwr;
 897 }
 898
 899 sub rewrite_uuid {
 900         my ($self) = @_;
 901         return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
 902         my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
 903         my $rwid = eval { command_oneline(qw/config --get/, $k) };
 904         if ($rwid) {
 905                 $rwid =~ s#/+$##;
 906                 if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
 907                         die "$rwid is not a valid UUID (key: $k)\n";
 908                 }
 909         }
 910         $self->{-rewrite_uuid} = $rwid;
 911 }
 912
 913 sub metadata_url {
 914         my ($self) = @_;
 915         ($self->rewrite_root || $self->{url}) .
 916            (length $self->{path} ? '/' . $self->{path} : '');
 917 }
 918
 919 sub full_url {
 920         my ($self) = @_;
 921         $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
 922 }
 923
 924 sub full_pushurl {
 925         my ($self) = @_;
 926         if ($self->{pushurl}) {
 927                 return $self->{pushurl} . (length $self->{path} ? '/' .
 928                        $self->{path} : '');
 929         } else {
 930                 return $self->full_url;
 931         }
 932 }
 933
 934 sub set_commit_header_env {
 935         my ($log_entry) = @_;
 936         my %env;
 937         foreach my $ned (qw/NAME EMAIL DATE/) {
 938                 foreach my $ac (qw/AUTHOR COMMITTER/) {
 939                         $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
 940                 }
 941         }
 942
 943         $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
 944         $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
 945         $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
 946
 947         $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
 948                                                 ? $log_entry->{commit_name}
 949                                                 : $log_entry->{name};
 950         $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
 951                                                 ? $log_entry->{commit_email}
 952                                                 : $log_entry->{email};
 953         \%env;
 954 }
 955
 956 sub restore_commit_header_env {
 957         my ($env) = @_;
 958         foreach my $ned (qw/NAME EMAIL DATE/) {
 959                 foreach my $ac (qw/AUTHOR COMMITTER/) {
 960                         my $k = "GIT_${ac}_${ned}";
 961                         if (defined $env->{$k}) {
 962                                 $ENV{$k} = $env->{$k};
 963                         } else {
 964                                 delete $ENV{$k};
 965                         }
 966                 }
 967         }
 968 }
 969
 970 sub gc {
 971         command_noisy('gc', '--auto');
 972 };
 973
 974 sub do_git_commit {
 975         my ($self, $log_entry) = @_;
 976         my $lr = $self->last_rev;
 977         if (defined $lr && $lr >= $log_entry->{revision}) {
 978                 die "Last fetched revision of ", $self->refname,
 979                     " was r$lr, but we are about to fetch: ",
 980                     "r$log_entry->{revision}!\n";
 981         }
 982         if (my $c = $self->rev_map_get($log_entry->{revision})) {
 983                 croak "$log_entry->{revision} = $c already exists! ",
 984                       "Why are we refetching it?\n";
 985         }
 986         my $old_env = set_commit_header_env($log_entry);
 987         my $tree = $log_entry->{tree};
 988         if (!defined $tree) {
 989                 $tree = $self->tmp_index_do(sub {
 990                                             command_oneline('write-tree') });
 991         }
 992         die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
 993
 994         my @exec = ('git', 'commit-tree', $tree);
 995         foreach ($self->get_commit_parents($log_entry)) {
 996                 push @exec, '-p', $_;
 997         }
 998         defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
 999                                                                    or croak $!;
1000         binmode $msg_fh;
1001
1002         # we always get UTF-8 from SVN, but we may want our commits in
1003         # a different encoding.
1004         if (my $enc = Git::config('i18n.commitencoding')) {
1005                 require Encode;
1006                 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
1007         }
1008         print $msg_fh $log_entry->{log} or croak $!;
1009         restore_commit_header_env($old_env);
1010         unless ($self->no_metadata) {
1011                 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1012                               or croak $!;
1013         }
1014         $msg_fh->flush == 0 or croak $!;
1015         close $msg_fh or croak $!;
1016         chomp(my $commit = do { local $/; <$out_fh> });
1017         close $out_fh or croak $!;
1018         waitpid $pid, 0;
1019         croak $? if $?;
1020         if ($commit !~ /^$::sha1$/o) {
1021                 die "Failed to commit, invalid sha1: $commit\n";
1022         }
1023
1024         $self->rev_map_set($log_entry->{revision}, $commit, 1);
1025
1026         $self->{last_rev} = $log_entry->{revision};
1027         $self->{last_commit} = $commit;
1028         print "r$log_entry->{revision}" unless $::_q > 1;
1029         if (defined $log_entry->{svm_revision}) {
1030                  print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
1031                  $self->rev_map_set($log_entry->{svm_revision}, $commit,
1032                                    0, $self->svm_uuid);
1033         }
1034         print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
1035         if (--$_gc_nr == 0) {
1036                 $_gc_nr = $_gc_period;
1037                 gc();
1038         }
1039         return $commit;
1040 }
1041
1042 sub match_paths {
1043         my ($self, $paths, $r) = @_;
1044         return 1 if $self->{path} eq '';
1045         if (my $path = $paths->{"/$self->{path}"}) {
1046                 return ($path->{action} eq 'D') ? 0 : 1;
1047         }
1048         $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1049         if (grep /$self->{path_regex}/, keys %$paths) {
1050                 return 1;
1051         }
1052         my $c = '';
1053         foreach (split m#/#, $self->{path}) {
1054                 $c .= "/$_";
1055                 next unless ($paths->{$c} &&
1056                              ($paths->{$c}->{action} =~ /^[AR]$/));
1057                 if ($self->ra->check_path($self->{path}, $r) ==
1058                     $SVN::Node::dir) {
1059                         return 1;
1060                 }
1061         }
1062         return 0;
1063 }
1064
1065 sub find_parent_branch {
1066         my ($self, $paths, $rev) = @_;
1067         return undef unless $self->follow_parent;
1068         unless (defined $paths) {
1069                 my $err_handler = $SVN::Error::handler;
1070                 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1071                 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
1072                                    sub { $paths = $_[0] });
1073                 $SVN::Error::handler = $err_handler;
1074         }
1075         return undef unless defined $paths;
1076
1077         # look for a parent from another branch:
1078         my @b_path_components = split m#/#, $self->{path};
1079         my @a_path_components;
1080         my $i;
1081         while (@b_path_components) {
1082                 $i = $paths->{'/'.join('/', @b_path_components)};
1083                 last if $i && defined $i->{copyfrom_path};
1084                 unshift(@a_path_components, pop(@b_path_components));
1085         }
1086         return undef unless defined $i && defined $i->{copyfrom_path};
1087         my $branch_from = $i->{copyfrom_path};
1088         if (@a_path_components) {
1089                 print STDERR "branch_from: $branch_from => ";
1090                 $branch_from .= '/'.join('/', @a_path_components);
1091                 print STDERR $branch_from, "\n";
1092         }
1093         my $r = $i->{copyfrom_rev};
1094         my $repos_root = $self->ra->{repos_root};
1095         my $url = $self->ra->{url};
1096         my $new_url = $url . $branch_from;
1097         print STDERR  "Found possible branch point: ",
1098                       "$new_url => ", $self->full_url, ", $r\n"
1099                       unless $::_q > 1;
1100         $branch_from =~ s#^/##;
1101         my $gs = $self->other_gs($new_url, $url,
1102                                  $branch_from, $r, $self->{ref_id});
1103         my ($r0, $parent) = $gs->find_rev_before($r, 1);
1104         {
1105                 my ($base, $head);
1106                 if (!defined $r0 || !defined $parent) {
1107                         ($base, $head) = parse_revision_argument(0, $r);
1108                 } else {
1109                         if ($r0 < $r) {
1110                                 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
1111                                         0, 1, sub { $base = $_[1] - 1 });
1112                         }
1113                 }
1114                 if (defined $base && $base <= $r) {
1115                         $gs->fetch($base, $r);
1116                 }
1117                 ($r0, $parent) = $gs->find_rev_before($r, 1);
1118         }
1119         if (defined $r0 && defined $parent) {
1120                 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
1121                              unless $::_q > 1;
1122                 my $ed;
1123                 if ($self->ra->can_do_switch) {
1124                         $self->assert_index_clean($parent);
1125                         print STDERR "Following parent with do_switch\n"
1126                                      unless $::_q > 1;
1127                         # do_switch works with svn/trunk >= r22312, but that
1128                         # is not included with SVN 1.4.3 (the latest version
1129                         # at the moment), so we can't rely on it
1130                         $self->{last_rev} = $r0;
1131                         $self->{last_commit} = $parent;
1132                         $ed = Git::SVN::Fetcher->new($self, $gs->{path});
1133                         $gs->ra->gs_do_switch($r0, $rev, $gs,
1134                                               $self->full_url, $ed)
1135                           or die "SVN connection failed somewhere...\n";
1136                 } elsif ($self->ra->trees_match($new_url, $r0,
1137                                                 $self->full_url, $rev)) {
1138                         print STDERR "Trees match:\n",
1139                                      "  $new_url\@$r0\n",
1140                                      "  ${\$self->full_url}\@$rev\n",
1141                                      "Following parent with no changes\n"
1142                                      unless $::_q > 1;
1143                         $self->tmp_index_do(sub {
1144                             command_noisy('read-tree', $parent);
1145                         });
1146                         $self->{last_commit} = $parent;
1147                 } else {
1148                         print STDERR "Following parent with do_update\n"
1149                                      unless $::_q > 1;
1150                         $ed = Git::SVN::Fetcher->new($self);
1151                         $self->ra->gs_do_update($rev, $rev, $self, $ed)
1152                           or die "SVN connection failed somewhere...\n";
1153                 }
1154                 print STDERR "Successfully followed parent\n" unless $::_q > 1;
1155                 return $self->make_log_entry($rev, [$parent], $ed);
1156         }
1157         return undef;
1158 }
1159
1160 sub do_fetch {
1161         my ($self, $paths, $rev) = @_;
1162         my $ed;
1163         my ($last_rev, @parents);
1164         if (my $lc = $self->last_commit) {
1165                 # we can have a branch that was deleted, then re-added
1166                 # under the same name but copied from another path, in
1167                 # which case we'll have multiple parents (we don't
1168                 # want to break the original ref, nor lose copypath info):
1169                 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1170                         push @{$log_entry->{parents}}, $lc;
1171                         return $log_entry;
1172                 }
1173                 $ed = Git::SVN::Fetcher->new($self);
1174                 $last_rev = $self->{last_rev};
1175                 $ed->{c} = $lc;
1176                 @parents = ($lc);
1177         } else {
1178                 $last_rev = $rev;
1179                 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1180                         return $log_entry;
1181                 }
1182                 $ed = Git::SVN::Fetcher->new($self);
1183         }
1184         unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1185                 die "SVN connection failed somewhere...\n";
1186         }
1187         $self->make_log_entry($rev, \@parents, $ed);
1188 }
1189
1190 sub mkemptydirs {
1191         my ($self, $r) = @_;
1192
1193         sub scan {
1194                 my ($r, $empty_dirs, $line) = @_;
1195                 if (defined $r && $line =~ /^r(\d+)$/) {
1196                         return 0 if $1 > $r;
1197                 } elsif ($line =~ /^  \+empty_dir: (.+)$/) {
1198                         $empty_dirs->{$1} = 1;
1199                 } elsif ($line =~ /^  \-empty_dir: (.+)$/) {
1200                         my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
1201                         delete @$empty_dirs{@d};
1202                 }
1203                 1; # continue
1204         };
1205
1206         my %empty_dirs = ();
1207         my $gz_file = "$self->{dir}/unhandled.log.gz";
1208         if (-f $gz_file) {
1209                 if (!can_compress()) {
1210                         warn "Compress::Zlib could not be found; ",
1211                              "empty directories in $gz_file will not be read\n";
1212                 } else {
1213                         my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
1214                                 die "Unable to open $gz_file: $!\n";
1215                         my $line;
1216                         while ($gz->gzreadline($line) > 0) {
1217                                 scan($r, \%empty_dirs, $line) or last;
1218                         }
1219                         $gz->gzclose;
1220                 }
1221         }
1222
1223         if (open my $fh, '<', "$self->{dir}/unhandled.log") {
1224                 binmode $fh or croak "binmode: $!";
1225                 while (<$fh>) {
1226                         scan($r, \%empty_dirs, $_) or last;
1227                 }
1228                 close $fh;
1229         }
1230
1231         my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
1232         foreach my $d (sort keys %empty_dirs) {
1233                 $d = uri_decode($d);
1234                 $d =~ s/$strip//;
1235                 next unless length($d);
1236                 next if -d $d;
1237                 if (-e $d) {
1238                         warn "$d exists but is not a directory\n";
1239                 } else {
1240                         print "creating empty directory: $d\n";
1241                         mkpath([$d]);
1242                 }
1243         }
1244 }
1245
1246 sub get_untracked {
1247         my ($self, $ed) = @_;
1248         my @out;
1249         my $h = $ed->{empty};
1250         foreach (sort keys %$h) {
1251                 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1252                 push @out, "  $act: " . uri_encode($_);
1253                 warn "W: $act: $_\n";
1254         }
1255         foreach my $t (qw/dir_prop file_prop/) {
1256                 $h = $ed->{$t} or next;
1257                 foreach my $path (sort keys %$h) {
1258                         my $ppath = $path eq '' ? '.' : $path;
1259                         foreach my $prop (sort keys %{$h->{$path}}) {
1260                                 next if $SKIP_PROP{$prop};
1261                                 my $v = $h->{$path}->{$prop};
1262                                 my $t_ppath_prop = "$t: " .
1263                                                     uri_encode($ppath) . ' ' .
1264                                                     uri_encode($prop);
1265                                 if (defined $v) {
1266                                         push @out, "  +$t_ppath_prop " .
1267                                                    uri_encode($v);
1268                                 } else {
1269                                         push @out, "  -$t_ppath_prop";
1270                                 }
1271                         }
1272                 }
1273         }
1274         foreach my $t (qw/absent_file absent_directory/) {
1275                 $h = $ed->{$t} or next;
1276                 foreach my $parent (sort keys %$h) {
1277                         foreach my $path (sort @{$h->{$parent}}) {
1278                                 push @out, "  $t: " .
1279                                            uri_encode("$parent/$path");
1280                                 warn "W: $t: $parent/$path ",
1281                                      "Insufficient permissions?\n";
1282                         }
1283                 }
1284         }
1285         \@out;
1286 }
1287
1288 sub get_tz {
1289         # some systmes don't handle or mishandle %z, so be creative.
1290         my $t = shift || time;
1291         my $gm = timelocal(gmtime($t));
1292         my $sign = qw( + + - )[ $t <=> $gm ];
1293         return sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
1294 }
1295
1296 # parse_svn_date(DATE)
1297 # --------------------
1298 # Given a date (in UTC) from Subversion, return a string in the format
1299 # "<TZ Offset> <local date/time>" that Git will use.
1300 #
1301 # By default the parsed date will be in UTC; if $Git::SVN::_localtime
1302 # is true we'll convert it to the local timezone instead.
1303 sub parse_svn_date {
1304         my $date = shift || return '+0000 1970-01-01 00:00:00';
1305         my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1306                                             (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
1307                                          croak "Unable to parse date: $date\n";
1308         my $parsed_date;    # Set next.
1309
1310         if ($Git::SVN::_localtime) {
1311                 # Translate the Subversion datetime to an epoch time.
1312                 # Begin by switching ourselves to $date's timezone, UTC.
1313                 my $old_env_TZ = $ENV{TZ};
1314                 $ENV{TZ} = 'UTC';
1315
1316                 my $epoch_in_UTC =
1317                     POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
1318
1319                 # Determine our local timezone (including DST) at the
1320                 # time of $epoch_in_UTC.  $Git::SVN::Log::TZ stored the
1321                 # value of TZ, if any, at the time we were run.
1322                 if (defined $Git::SVN::Log::TZ) {
1323                         $ENV{TZ} = $Git::SVN::Log::TZ;
1324                 } else {
1325                         delete $ENV{TZ};
1326                 }
1327
1328                 my $our_TZ = get_tz();
1329
1330                 # This converts $epoch_in_UTC into our local timezone.
1331                 my ($sec, $min, $hour, $mday, $mon, $year,
1332                     $wday, $yday, $isdst) = localtime($epoch_in_UTC);
1333
1334                 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
1335                                        $our_TZ, $year + 1900, $mon + 1,
1336                                        $mday, $hour, $min, $sec);
1337
1338                 # Reset us to the timezone in effect when we entered
1339                 # this routine.
1340                 if (defined $old_env_TZ) {
1341                         $ENV{TZ} = $old_env_TZ;
1342                 } else {
1343                         delete $ENV{TZ};
1344                 }
1345         } else {
1346                 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
1347         }
1348
1349         return $parsed_date;
1350 }
1351
1352 sub other_gs {
1353         my ($self, $new_url, $url,
1354             $branch_from, $r, $old_ref_id) = @_;
1355         my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
1356         unless ($gs) {
1357                 my $ref_id = $old_ref_id;
1358                 $ref_id =~ s/\@\d+-*$//;
1359                 $ref_id .= "\@$r";
1360                 # just grow a tail if we're not unique enough :x
1361                 $ref_id .= '-' while find_ref($ref_id);
1362                 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
1363                 if ($u =~ s#^\Q$url\E(/|$)##) {
1364                         $p = $u;
1365                         $u = $url;
1366                         $repo_id = $self->{repo_id};
1367                 }
1368                 while (1) {
1369                         # It is possible to tag two different subdirectories at
1370                         # the same revision.  If the url for an existing ref
1371                         # does not match, we must either find a ref with a
1372                         # matching url or create a new ref by growing a tail.
1373                         $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
1374                         my (undef, $max_commit) = $gs->rev_map_max(1);
1375                         last if (!$max_commit);
1376                         my ($url) = ::cmt_metadata($max_commit);
1377                         last if ($url eq $gs->metadata_url);
1378                         $ref_id .= '-';
1379                 }
1380                 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
1381         }
1382         $gs
1383 }
1384
1385 sub call_authors_prog {
1386         my ($orig_author) = @_;
1387         $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
1388         my $author = `$::_authors_prog $orig_author`;
1389         if ($? != 0) {
1390                 die "$::_authors_prog failed with exit code $?\n"
1391         }
1392         if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
1393                 my ($name, $email) = ($1, $2);
1394                 $email = undef if length $2 == 0;
1395                 return [$name, $email];
1396         } else {
1397                 die "Author: $orig_author: $::_authors_prog returned "
1398                         . "invalid author format: $author\n";
1399         }
1400 }
1401
1402 sub check_author {
1403         my ($author) = @_;
1404         if (!defined $author || length $author == 0) {
1405                 $author = '(no author)';
1406         }
1407         if (!defined $::users{$author}) {
1408                 if (defined $::_authors_prog) {
1409                         $::users{$author} = call_authors_prog($author);
1410                 } elsif (defined $::_authors) {
1411                         die "Author: $author not defined in $::_authors file\n";
1412                 }
1413         }
1414         $author;
1415 }
1416
1417 sub find_extra_svk_parents {
1418         my ($self, $ed, $tickets, $parents) = @_;
1419         # aha!  svk:merge property changed...
1420         my @tickets = split "\n", $tickets;
1421         my @known_parents;
1422         for my $ticket ( @tickets ) {
1423                 my ($uuid, $path, $rev) = split /:/, $ticket;
1424                 if ( $uuid eq $self->ra_uuid ) {
1425                         my $url = $self->{url};
1426                         my $repos_root = $url;
1427                         my $branch_from = $path;
1428                         $branch_from =~ s{^/}{};
1429                         my $gs = $self->other_gs($repos_root."/".$branch_from,
1430                                                  $url,
1431                                                  $branch_from,
1432                                                  $rev,
1433                                                  $self->{ref_id});
1434                         if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
1435                                 # wahey!  we found it, but it might be
1436                                 # an old one (!)
1437                                 push @known_parents, [ $rev, $commit ];
1438                         }
1439                 }
1440         }
1441         # Ordering matters; highest-numbered commit merge tickets
1442         # first, as they may account for later merge ticket additions
1443         # or changes.
1444         @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
1445         for my $parent ( @known_parents ) {
1446                 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
1447                 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
1448                 my $new;
1449                 while ( <$msg_fh> ) {
1450                         $new=1;last;
1451                 }
1452                 command_close_pipe($msg_fh, $ctx);
1453                 if ( $new ) {
1454                         print STDERR
1455                             "Found merge parent (svk:merge ticket): $parent\n";
1456                         push @$parents, $parent;
1457                 }
1458         }
1459 }
1460
1461 sub lookup_svn_merge {
1462         my $uuid = shift;
1463         my $url = shift;
1464         my $merge = shift;
1465
1466         my ($source, $revs) = split ":", $merge;
1467         my $path = $source;
1468         $path =~ s{^/}{};
1469         my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
1470         if ( !$gs ) {
1471                 warn "Couldn't find revmap for $url$source\n";
1472                 return;
1473         }
1474         my @ranges = split ",", $revs;
1475         my ($tip, $tip_commit);
1476         my @merged_commit_ranges;
1477         # find the tip
1478         for my $range ( @ranges ) {
1479                 my ($bottom, $top) = split "-", $range;
1480                 $top ||= $bottom;
1481                 my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
1482                 my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
1483
1484                 unless ($top_commit and $bottom_commit) {
1485                         warn "W:unknown path/rev in svn:mergeinfo "
1486                                 ."dirprop: $source:$range\n";
1487                         next;
1488                 }
1489
1490                 if (scalar(command('rev-parse', "$bottom_commit^@"))) {
1491                         push @merged_commit_ranges,
1492                              "$bottom_commit^..$top_commit";
1493                 } else {
1494                         push @merged_commit_ranges, "$top_commit";
1495                 }
1496
1497                 if ( !defined $tip or $top > $tip ) {
1498                         $tip = $top;
1499                         $tip_commit = $top_commit;
1500                 }
1501         }
1502         return ($tip_commit, @merged_commit_ranges);
1503 }
1504
1505 sub _rev_list {
1506         my ($msg_fh, $ctx) = command_output_pipe(
1507                 "rev-list", @_,
1508                );
1509         my @rv;
1510         while ( <$msg_fh> ) {
1511                 chomp;
1512                 push @rv, $_;
1513         }
1514         command_close_pipe($msg_fh, $ctx);
1515         @rv;
1516 }
1517
1518 sub check_cherry_pick {
1519         my $base = shift;
1520         my $tip = shift;
1521         my $parents = shift;
1522         my @ranges = @_;
1523         my %commits = map { $_ => 1 }
1524                 _rev_list("--no-merges", $tip, "--not", $base, @$parents, "--");
1525         for my $range ( @ranges ) {
1526                 delete @commits{_rev_list($range, "--")};
1527         }
1528         for my $commit (keys %commits) {
1529                 if (has_no_changes($commit)) {
1530                         delete $commits{$commit};
1531                 }
1532         }
1533         return (keys %commits);
1534 }
1535
1536 sub has_no_changes {
1537         my $commit = shift;
1538
1539         my @revs = split / /, command_oneline(
1540                 qw(rev-list --parents -1 -m), $commit);
1541
1542         # Commits with no parents, e.g. the start of a partial branch,
1543         # have changes by definition.
1544         return 1 if (@revs < 2);
1545
1546         # Commits with multiple parents, e.g a merge, have no changes
1547         # by definition.
1548         return 0 if (@revs > 2);
1549
1550         return (command_oneline("rev-parse", "$commit^{tree}") eq
1551                 command_oneline("rev-parse", "$commit~1^{tree}"));
1552 }
1553
1554 sub tie_for_persistent_memoization {
1555         my $hash = shift;
1556         my $path = shift;
1557
1558         if ($can_use_yaml) {
1559                 tie %$hash => 'Git::SVN::Memoize::YAML', "$path.yaml";
1560         } else {
1561                 tie %$hash => 'Memoize::Storable', "$path.db", 'nstore';
1562         }
1563 }
1564
1565 # The GIT_DIR environment variable is not always set until after the command
1566 # line arguments are processed, so we can't memoize in a BEGIN block.
1567 {
1568         my $memoized = 0;
1569
1570         sub memoize_svn_mergeinfo_functions {
1571                 return if $memoized;
1572                 $memoized = 1;
1573
1574                 my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
1575                 mkpath([$cache_path]) unless -d $cache_path;
1576
1577                 my %lookup_svn_merge_cache;
1578                 my %check_cherry_pick_cache;
1579                 my %has_no_changes_cache;
1580
1581                 tie_for_persistent_memoization(\%lookup_svn_merge_cache,
1582                     "$cache_path/lookup_svn_merge");
1583                 memoize 'lookup_svn_merge',
1584                         SCALAR_CACHE => 'FAULT',
1585                         LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
1586                 ;
1587
1588                 tie_for_persistent_memoization(\%check_cherry_pick_cache,
1589                     "$cache_path/check_cherry_pick");
1590                 memoize 'check_cherry_pick',
1591                         SCALAR_CACHE => 'FAULT',
1592                         LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
1593                 ;
1594
1595                 tie_for_persistent_memoization(\%has_no_changes_cache,
1596                     "$cache_path/has_no_changes");
1597                 memoize 'has_no_changes',
1598                         SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
1599                         LIST_CACHE => 'FAULT',
1600                 ;
1601         }
1602
1603         sub unmemoize_svn_mergeinfo_functions {
1604                 return if not $memoized;
1605                 $memoized = 0;
1606
1607                 Memoize::unmemoize 'lookup_svn_merge';
1608                 Memoize::unmemoize 'check_cherry_pick';
1609                 Memoize::unmemoize 'has_no_changes';
1610         }
1611
1612         Memoize::memoize 'Git::SVN::repos_root';
1613 }
1614
1615 END {
1616         # Force cache writeout explicitly instead of waiting for
1617         # global destruction to avoid segfault in Storable:
1618         # http://rt.cpan.org/Public/Bug/Display.html?id=36087
1619         unmemoize_svn_mergeinfo_functions();
1620 }
1621
1622 sub parents_exclude {
1623         my $parents = shift;
1624         my @commits = @_;
1625         return unless @commits;
1626
1627         my @excluded;
1628         my $excluded;
1629         do {
1630                 my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
1631                 $excluded = command_oneline(@cmd);
1632                 if ( $excluded ) {
1633                         my @new;
1634                         my $found;
1635                         for my $commit ( @commits ) {
1636                                 if ( $commit eq $excluded ) {
1637                                         push @excluded, $commit;
1638                                         $found++;
1639                                         last;
1640                                 }
1641                                 else {
1642                                         push @new, $commit;
1643                                 }
1644                         }
1645                         die "saw commit '$excluded' in rev-list output, "
1646                                 ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
1647                                         unless $found;
1648                         @commits = @new;
1649                 }
1650         }
1651                 while ($excluded and @commits);
1652
1653         return @excluded;
1654 }
1655
1656
1657 # note: this function should only be called if the various dirprops
1658 # have actually changed
1659 sub find_extra_svn_parents {
1660         my ($self, $ed, $mergeinfo, $parents) = @_;
1661         # aha!  svk:merge property changed...
1662
1663         memoize_svn_mergeinfo_functions();
1664
1665         # We first search for merged tips which are not in our
1666         # history.  Then, we figure out which git revisions are in
1667         # that tip, but not this revision.  If all of those revisions
1668         # are now marked as merge, we can add the tip as a parent.
1669         my @merges = split "\n", $mergeinfo;
1670         my @merge_tips;
1671         my $url = $self->{url};
1672         my $uuid = $self->ra_uuid;
1673         my %ranges;
1674         for my $merge ( @merges ) {
1675                 my ($tip_commit, @ranges) =
1676                         lookup_svn_merge( $uuid, $url, $merge );
1677                 unless (!$tip_commit or
1678                                 grep { $_ eq $tip_commit } @$parents ) {
1679                         push @merge_tips, $tip_commit;
1680                         $ranges{$tip_commit} = \@ranges;
1681                 } else {
1682                         push @merge_tips, undef;
1683                 }
1684         }
1685
1686         my %excluded = map { $_ => 1 }
1687                 parents_exclude($parents, grep { defined } @merge_tips);
1688
1689         # check merge tips for new parents
1690         my @new_parents;
1691         for my $merge_tip ( @merge_tips ) {
1692                 my $spec = shift @merges;
1693                 next unless $merge_tip and $excluded{$merge_tip};
1694
1695                 my $ranges = $ranges{$merge_tip};
1696
1697                 # check out 'new' tips
1698                 my $merge_base;
1699                 eval {
1700                         $merge_base = command_oneline(
1701                                 "merge-base",
1702                                 @$parents, $merge_tip,
1703                         );
1704                 };
1705                 if ($@) {
1706                         die "An error occurred during merge-base"
1707                                 unless $@->isa("Git::Error::Command");
1708
1709                         warn "W: Cannot find common ancestor between ".
1710                              "@$parents and $merge_tip. Ignoring merge info.\n";
1711                         next;
1712                 }
1713
1714                 # double check that there are no missing non-merge commits
1715                 my (@incomplete) = check_cherry_pick(
1716                         $merge_base, $merge_tip,
1717                         $parents,
1718                         @$ranges,
1719                        );
1720
1721                 if ( @incomplete ) {
1722                         warn "W:svn cherry-pick ignored ($spec) - missing "
1723                                 .@incomplete." commit(s) (eg $incomplete[0])\n";
1724                 } else {
1725                         warn
1726                                 "Found merge parent (svn:mergeinfo prop): ",
1727                                         $merge_tip, "\n";
1728                         push @new_parents, $merge_tip;
1729                 }
1730         }
1731
1732         # cater for merges which merge commits from multiple branches
1733         if ( @new_parents > 1 ) {
1734                 for ( my $i = 0; $i <= $#new_parents; $i++ ) {
1735                         for ( my $j = 0; $j <= $#new_parents; $j++ ) {
1736                                 next if $i == $j;
1737                                 next unless $new_parents[$i];
1738                                 next unless $new_parents[$j];
1739                                 my $revs = command_oneline(
1740                                         "rev-list", "-1",
1741                                         "$new_parents[$i]..$new_parents[$j]",
1742                                        );
1743                                 if ( !$revs ) {
1744                                         undef($new_parents[$j]);
1745                                 }
1746                         }
1747                 }
1748         }
1749         push @$parents, grep { defined } @new_parents;
1750 }
1751
1752 sub make_log_entry {
1753         my ($self, $rev, $parents, $ed) = @_;
1754         my $untracked = $self->get_untracked($ed);
1755
1756         my @parents = @$parents;
1757         my $ps = $ed->{path_strip} || "";
1758         for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
1759                 my $props = $ed->{dir_prop}{$path};
1760                 if ( $props->{"svk:merge"} ) {
1761                         $self->find_extra_svk_parents
1762                                 ($ed, $props->{"svk:merge"}, \@parents);
1763                 }
1764                 if ( $props->{"svn:mergeinfo"} ) {
1765                         $self->find_extra_svn_parents
1766                                 ($ed,
1767                                  $props->{"svn:mergeinfo"},
1768                                  \@parents);
1769                 }
1770         }
1771
1772         open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1773         print $un "r$rev\n" or croak $!;
1774         print $un $_, "\n" foreach @$untracked;
1775         my %log_entry = ( parents => \@parents, revision => $rev,
1776                           log => '');
1777
1778         my $headrev;
1779         my $logged = delete $self->{logged_rev_props};
1780         if (!$logged || $self->{-want_revprops}) {
1781                 my $rp = $self->ra->rev_proplist($rev);
1782                 foreach (sort keys %$rp) {
1783                         my $v = $rp->{$_};
1784                         if (/^svn:(author|date|log)$/) {
1785                                 $log_entry{$1} = $v;
1786                         } elsif ($_ eq 'svm:headrev') {
1787                                 $headrev = $v;
1788                         } else {
1789                                 print $un "  rev_prop: ", uri_encode($_), ' ',
1790                                           uri_encode($v), "\n";
1791                         }
1792                 }
1793         } else {
1794                 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1795         }
1796         close $un or croak $!;
1797
1798         $log_entry{date} = parse_svn_date($log_entry{date});
1799         $log_entry{log} .= "\n";
1800         my $author = $log_entry{author} = check_author($log_entry{author});
1801         my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1802                                                        : ($author, undef);
1803
1804         my ($commit_name, $commit_email) = ($name, $email);
1805         if ($_use_log_author) {
1806                 my $name_field;
1807                 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
1808                         $name_field = $1;
1809                 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
1810                         $name_field = $1;
1811                 }
1812                 if (!defined $name_field) {
1813                         if (!defined $email) {
1814                                 $email = $name;
1815                         }
1816                 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
1817                         ($name, $email) = ($1, $2);
1818                 } elsif ($name_field =~ /(.*)@/) {
1819                         ($name, $email) = ($1, $name_field);
1820                 } else {
1821                         ($name, $email) = ($name_field, $name_field);
1822                 }
1823         }
1824         if (defined $headrev && $self->use_svm_props) {
1825                 if ($self->rewrite_root) {
1826                         die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
1827                             "options set!\n";
1828                 }
1829                 if ($self->rewrite_uuid) {
1830                         die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
1831                             "options set!\n";
1832                 }
1833                 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
1834                 # we don't want "SVM: initializing mirror for junk" ...
1835                 return undef if $r == 0;
1836                 my $svm = $self->svm;
1837                 if ($uuid ne $svm->{uuid}) {
1838                         die "UUID mismatch on SVM path:\n",
1839                             "expected: $svm->{uuid}\n",
1840                             "     got: $uuid\n";
1841                 }
1842                 my $full_url = $self->full_url;
1843                 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
1844                              die "Failed to replace '$svm->{replace}' with ",
1845                                  "'$svm->{source}' in $full_url\n";
1846                 # throw away username for storing in records
1847                 remove_username($full_url);
1848                 $log_entry{metadata} = "$full_url\@$r $uuid";
1849                 $log_entry{svm_revision} = $r;
1850                 $email ||= "$author\@$uuid";
1851                 $commit_email ||= "$author\@$uuid";
1852         } elsif ($self->use_svnsync_props) {
1853                 my $full_url = $self->svnsync->{url};
1854                 $full_url .= "/$self->{path}" if length $self->{path};
1855                 remove_username($full_url);
1856                 my $uuid = $self->svnsync->{uuid};
1857                 $log_entry{metadata} = "$full_url\@$rev $uuid";
1858                 $email ||= "$author\@$uuid";
1859                 $commit_email ||= "$author\@$uuid";
1860         } else {
1861                 my $url = $self->metadata_url;
1862                 remove_username($url);
1863                 my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
1864                 $log_entry{metadata} = "$url\@$rev " . $uuid;
1865                 $email ||= "$author\@" . $uuid;
1866                 $commit_email ||= "$author\@" . $uuid;
1867         }
1868         $log_entry{name} = $name;
1869         $log_entry{email} = $email;
1870         $log_entry{commit_name} = $commit_name;
1871         $log_entry{commit_email} = $commit_email;
1872         \%log_entry;
1873 }
1874
1875 sub fetch {
1876         my ($self, $min_rev, $max_rev, @parents) = @_;
1877         my ($last_rev, $last_commit) = $self->last_rev_commit;
1878         my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1879         $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1880 }
1881
1882 sub set_tree_cb {
1883         my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1884         $self->{inject_parents} = { $rev => $tree };
1885         $self->fetch(undef, undef);
1886 }
1887
1888 sub set_tree {
1889         my ($self, $tree) = (shift, shift);
1890         my $log_entry = ::get_commit_entry($tree);
1891         unless ($self->{last_rev}) {
1892                 fatal("Must have an existing revision to commit");
1893         }
1894         my %ed_opts = ( r => $self->{last_rev},
1895                         log => $log_entry->{log},
1896                         ra => $self->ra,
1897                         tree_a => $self->{last_commit},
1898                         tree_b => $tree,
1899                         editor_cb => sub {
1900                                $self->set_tree_cb($log_entry, $tree, @_) },
1901                         svn_path => $self->{path} );
1902         if (!Git::SVN::Editor->new(\%ed_opts)->apply_diff) {
1903                 print "No changes\nr$self->{last_rev} = $tree\n";
1904         }
1905 }
1906
1907 sub rebuild_from_rev_db {
1908         my ($self, $path) = @_;
1909         my $r = -1;
1910         open my $fh, '<', $path or croak "open: $!";
1911         binmode $fh or croak "binmode: $!";
1912         while (<$fh>) {
1913                 length($_) == 41 or croak "inconsistent size in ($_) != 41";
1914                 chomp($_);
1915                 ++$r;
1916                 next if $_ eq ('0' x 40);
1917                 $self->rev_map_set($r, $_);
1918                 print "r$r = $_\n";
1919         }
1920         close $fh or croak "close: $!";
1921         unlink $path or croak "unlink: $!";
1922 }
1923
1924 sub rebuild {
1925         my ($self) = @_;
1926         my $map_path = $self->map_path;
1927         my $partial = (-e $map_path && ! -z $map_path);
1928         return unless ::verify_ref($self->refname.'^0');
1929         if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
1930                 my $rev_db = $self->rev_db_path;
1931                 $self->rebuild_from_rev_db($rev_db);
1932                 if ($self->use_svm_props) {
1933                         my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
1934                         $self->rebuild_from_rev_db($svm_rev_db);
1935                 }
1936                 $self->unlink_rev_db_symlink;
1937                 return;
1938         }
1939         print "Rebuilding $map_path ...\n" if (!$partial);
1940         my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
1941                 (undef, undef));
1942         my ($log, $ctx) =
1943             command_output_pipe(qw/rev-list --pretty=raw --reverse/,
1944                                 ($head ? "$head.." : "") . $self->refname,
1945                                 '--');
1946         my $metadata_url = $self->metadata_url;
1947         remove_username($metadata_url);
1948         my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
1949         my $c;
1950         while (<$log>) {
1951                 if ( m{^commit ($::sha1)$} ) {
1952                         $c = $1;
1953                         next;
1954                 }
1955                 next unless s{^\s*(git-svn-id:)}{$1};
1956                 my ($url, $rev, $uuid) = ::extract_metadata($_);
1957                 remove_username($url);
1958
1959                 # ignore merges (from set-tree)
1960                 next if (!defined $rev || !$uuid);
1961
1962                 # if we merged or otherwise started elsewhere, this is
1963                 # how we break out of it
1964                 if (($uuid ne $svn_uuid) ||
1965                     ($metadata_url && $url && ($url ne $metadata_url))) {
1966                         next;
1967                 }
1968                 if ($partial && $head) {
1969                         print "Partial-rebuilding $map_path ...\n";
1970                         print "Currently at $base_rev = $head\n";
1971                         $head = undef;
1972                 }
1973
1974                 $self->rev_map_set($rev, $c);
1975                 print "r$rev = $c\n";
1976         }
1977         command_close_pipe($log, $ctx);
1978         print "Done rebuilding $map_path\n" if (!$partial || !$head);
1979         my $rev_db_path = $self->rev_db_path;
1980         if (-f $self->rev_db_path) {
1981                 unlink $self->rev_db_path or croak "unlink: $!";
1982         }
1983         $self->unlink_rev_db_symlink;
1984 }
1985
1986 # rev_map:
1987 # Tie::File seems to be prone to offset errors if revisions get sparse,
1988 # it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
1989 # one of my favorite modules is out :<  Next up would be one of the DBM
1990 # modules, but I'm not sure which is most portable...
1991 #
1992 # This is the replacement for the rev_db format, which was too big
1993 # and inefficient for large repositories with a lot of sparse history
1994 # (mainly tags)
1995 #
1996 # The format is this:
1997 #   - 24 bytes for every record,
1998 #     * 4 bytes for the integer representing an SVN revision number
1999 #     * 20 bytes representing the sha1 of a git commit
2000 #   - No empty padding records like the old format
2001 #     (except the last record, which can be overwritten)
2002 #   - new records are written append-only since SVN revision numbers
2003 #     increase monotonically
2004 #   - lookups on SVN revision number are done via a binary search
2005 #   - Piping the file to xxd -c24 is a good way of dumping it for
2006 #     viewing or editing (piped back through xxd -r), should the need
2007 #     ever arise.
2008 #   - The last record can be padding revision with an all-zero sha1
2009 #     This is used to optimize fetch performance when using multiple
2010 #     "fetch" directives in .git/config
2011 #
2012 # These files are disposable unless noMetadata or useSvmProps is set
2013
2014 sub _rev_map_set {
2015         my ($fh, $rev, $commit) = @_;
2016
2017         binmode $fh or croak "binmode: $!";
2018         my $size = (stat($fh))[7];
2019         ($size % 24) == 0 or croak "inconsistent size: $size";
2020
2021         my $wr_offset = 0;
2022         if ($size > 0) {
2023                 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2024                 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2025                 $read == 24 or croak "read only $read bytes (!= 24)";
2026                 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2027                 if ($last_commit eq ('0' x40)) {
2028                         if ($size >= 48) {
2029                                 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2030                                 $read = sysread($fh, $buf, 24) or
2031                                     croak "read: $!";
2032                                 $read == 24 or
2033                                     croak "read only $read bytes (!= 24)";
2034                                 ($last_rev, $last_commit) =
2035                                     unpack(rev_map_fmt, $buf);
2036                                 if ($last_commit eq ('0' x40)) {
2037                                         croak "inconsistent .rev_map\n";
2038                                 }
2039                         }
2040                         if ($last_rev >= $rev) {
2041                                 croak "last_rev is higher!: $last_rev >= $rev";
2042                         }
2043                         $wr_offset = -24;
2044                 }
2045         }
2046         sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2047         syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2048           croak "write: $!";
2049 }
2050
2051 sub _rev_map_reset {
2052         my ($fh, $rev, $commit) = @_;
2053         my $c = _rev_map_get($fh, $rev);
2054         $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
2055         my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
2056         truncate $fh, $offset or croak "truncate: $!";
2057 }
2058
2059 sub mkfile {
2060         my ($path) = @_;
2061         unless (-e $path) {
2062                 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2063                 mkpath([$dir]) unless -d $dir;
2064                 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2065                 close $fh or die "Couldn't close (create) $path: $!\n";
2066         }
2067 }
2068
2069 sub rev_map_set {
2070         my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2071         defined $commit or die "missing arg3\n";
2072         length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2073         my $db = $self->map_path($uuid);
2074         my $db_lock = "$db.lock";
2075         my $sigmask;
2076         $update_ref ||= 0;
2077         if ($update_ref) {
2078                 $sigmask = POSIX::SigSet->new();
2079                 my $signew = POSIX::SigSet->new(SIGINT, SIGHUP, SIGTERM,
2080                         SIGALRM, SIGUSR1, SIGUSR2);
2081                 sigprocmask(SIG_BLOCK, $signew, $sigmask) or
2082                         croak "Can't block signals: $!";
2083         }
2084         mkfile($db);
2085
2086         $LOCKFILES{$db_lock} = 1;
2087         my $sync;
2088         # both of these options make our .rev_db file very, very important
2089         # and we can't afford to lose it because rebuild() won't work
2090         if ($self->use_svm_props || $self->no_metadata) {
2091                 $sync = 1;
2092                 copy($db, $db_lock) or die "rev_map_set(@_): ",
2093                                            "Failed to copy: ",
2094                                            "$db => $db_lock ($!)\n";
2095         } else {
2096                 rename $db, $db_lock or die "rev_map_set(@_): ",
2097                                             "Failed to rename: ",
2098                                             "$db => $db_lock ($!)\n";
2099         }
2100
2101         sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2102              or croak "Couldn't open $db_lock: $!\n";
2103         $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
2104                                  _rev_map_set($fh, $rev, $commit);
2105         if ($sync) {
2106                 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2107                 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2108         }
2109         close $fh or croak $!;
2110         if ($update_ref) {
2111                 $_head = $self;
2112                 my $note = "";
2113                 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
2114                 command_noisy('update-ref', '-m', "r$rev$note",
2115                               $self->refname, $commit);
2116         }
2117         rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2118                                     "$db_lock => $db ($!)\n";
2119         delete $LOCKFILES{$db_lock};
2120         if ($update_ref) {
2121                 sigprocmask(SIG_SETMASK, $sigmask) or
2122                         croak "Can't restore signal mask: $!";
2123         }
2124 }
2125
2126 # If want_commit, this will return an array of (rev, commit) where
2127 # commit _must_ be a valid commit in the archive.
2128 # Otherwise, it'll return the max revision (whether or not the
2129 # commit is valid or just a 0x40 placeholder).
2130 sub rev_map_max {
2131         my ($self, $want_commit) = @_;
2132         $self->rebuild;
2133         my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
2134         $want_commit ? ($r, $c) : $r;
2135 }
2136
2137 sub rev_map_max_norebuild {
2138         my ($self, $want_commit) = @_;
2139         my $map_path = $self->map_path;
2140         stat $map_path or return $want_commit ? (0, undef) : 0;
2141         sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2142         binmode $fh or croak "binmode: $!";
2143         my $size = (stat($fh))[7];
2144         ($size % 24) == 0 or croak "inconsistent size: $size";
2145
2146         if ($size == 0) {
2147                 close $fh or croak "close: $!";
2148                 return $want_commit ? (0, undef) : 0;
2149         }
2150
2151         sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2152         sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2153         my ($r, $c) = unpack(rev_map_fmt, $buf);
2154         if ($want_commit && $c eq ('0' x40)) {
2155                 if ($size < 48) {
2156                         return $want_commit ? (0, undef) : 0;
2157                 }
2158                 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2159                 sysread($fh, $buf, 24) == 24 or croak "read: $!";
2160                 ($r, $c) = unpack(rev_map_fmt, $buf);
2161                 if ($c eq ('0'x40)) {
2162                         croak "Penultimate record is all-zeroes in $map_path";
2163                 }
2164         }
2165         close $fh or croak "close: $!";
2166         $want_commit ? ($r, $c) : $r;
2167 }
2168
2169 sub rev_map_get {
2170         my ($self, $rev, $uuid) = @_;
2171         my $map_path = $self->map_path($uuid);
2172         return undef unless -e $map_path;
2173
2174         sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2175         my $c = _rev_map_get($fh, $rev);
2176         close($fh) or croak "close: $!";
2177         $c
2178 }
2179
2180 sub _rev_map_get {
2181         my ($fh, $rev) = @_;
2182
2183         binmode $fh or croak "binmode: $!";
2184         my $size = (stat($fh))[7];
2185         ($size % 24) == 0 or croak "inconsistent size: $size";
2186
2187         if ($size == 0) {
2188                 return undef;
2189         }
2190
2191         my ($l, $u) = (0, $size - 24);
2192         my ($r, $c, $buf);
2193
2194         while ($l <= $u) {
2195                 my $i = int(($l/24 + $u/24) / 2) * 24;
2196                 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2197                 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2198                 my ($r, $c) = unpack(rev_map_fmt, $buf);
2199
2200                 if ($r < $rev) {
2201                         $l = $i + 24;
2202                 } elsif ($r > $rev) {
2203                         $u = $i - 24;
2204                 } else { # $r == $rev
2205                         return $c eq ('0' x 40) ? undef : $c;
2206                 }
2207         }
2208         undef;
2209 }
2210
2211 # Finds the first svn revision that exists on (if $eq_ok is true) or
2212 # before $rev for the current branch.  It will not search any lower
2213 # than $min_rev.  Returns the git commit hash and svn revision number
2214 # if found, else (undef, undef).
2215 sub find_rev_before {
2216         my ($self, $rev, $eq_ok, $min_rev) = @_;
2217         --$rev unless $eq_ok;
2218         $min_rev ||= 1;
2219         my $max_rev = $self->rev_map_max;
2220         $rev = $max_rev if ($rev > $max_rev);
2221         while ($rev >= $min_rev) {
2222                 if (my $c = $self->rev_map_get($rev)) {
2223                         return ($rev, $c);
2224                 }
2225                 --$rev;
2226         }
2227         return (undef, undef);
2228 }
2229
2230 # Finds the first svn revision that exists on (if $eq_ok is true) or
2231 # after $rev for the current branch.  It will not search any higher
2232 # than $max_rev.  Returns the git commit hash and svn revision number
2233 # if found, else (undef, undef).
2234 sub find_rev_after {
2235         my ($self, $rev, $eq_ok, $max_rev) = @_;
2236         ++$rev unless $eq_ok;
2237         $max_rev ||= $self->rev_map_max;
2238         while ($rev <= $max_rev) {
2239                 if (my $c = $self->rev_map_get($rev)) {
2240                         return ($rev, $c);
2241                 }
2242                 ++$rev;
2243         }
2244         return (undef, undef);
2245 }
2246
2247 sub _new {
2248         my ($class, $repo_id, $ref_id, $path) = @_;
2249         unless (defined $repo_id && length $repo_id) {
2250                 $repo_id = $default_repo_id;
2251         }
2252         unless (defined $ref_id && length $ref_id) {
2253                 # Access the prefix option from the git-svn main program if it's loaded.
2254                 my $prefix = defined &::opt_prefix ? ::opt_prefix() : "";
2255                 $_[2] = $ref_id =
2256                              "refs/remotes/$prefix$default_ref_id";
2257         }
2258         $_[1] = $repo_id;
2259         my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2260
2261         # Older repos imported by us used $GIT_DIR/svn/foo instead of
2262         # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
2263         if ($ref_id =~ m{^refs/remotes/(.*)}) {
2264                 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
2265                 if (-d $old_dir && ! -d $dir) {
2266                         $dir = $old_dir;
2267                 }
2268         }
2269
2270         $_[3] = $path = '' unless (defined $path);
2271         mkpath([$dir]);
2272         bless {
2273                 ref_id => $ref_id, dir => $dir, index => "$dir/index",
2274                 path => $path, config => "$ENV{GIT_DIR}/svn/config",
2275                 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
2276 }
2277
2278 # for read-only access of old .rev_db formats
2279 sub unlink_rev_db_symlink {
2280         my ($self) = @_;
2281         my $link = $self->rev_db_path;
2282         $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
2283         if (-l $link) {
2284                 unlink $link or croak "unlink: $link failed!";
2285         }
2286 }
2287
2288 sub rev_db_path {
2289         my ($self, $uuid) = @_;
2290         my $db_path = $self->map_path($uuid);
2291         $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
2292             or croak "map_path: $db_path does not contain '/.rev_map.' !";
2293         $db_path;
2294 }
2295
2296 # the new replacement for .rev_db
2297 sub map_path {
2298         my ($self, $uuid) = @_;
2299         $uuid ||= $self->ra_uuid;
2300         "$self->{map_root}.$uuid";
2301 }
2302
2303 sub uri_encode {
2304         my ($f) = @_;
2305         $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2306         $f
2307 }
2308
2309 sub uri_decode {
2310         my ($f) = @_;
2311         $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
2312         $f
2313 }
2314
2315 sub remove_username {
2316         $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2317 }
2318
2319 1;