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/$_no_metadata
   7             $_repack $_repack_flags $_use_svm_props $_head
   8             $_use_svnsync_props $no_reuse_existing
   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 our $_follow_parent  = 1;
  34 our $_minimize_url   = 'unset';
  35 our $default_repo_id = 'svn';
  36 our $default_ref_id  = $ENV{GIT_SVN_ID} || 'git-svn';
  37
  38 my ($_gc_nr, $_gc_period);
  39
  40 # properties that we do not log:
  41 my %SKIP_PROP;
  42 BEGIN {
  43         %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
  44                                         svn:special svn:executable
  45                                         svn:entry:committed-rev
  46                                         svn:entry:last-author
  47                                         svn:entry:uuid
  48                                         svn:entry:committed-date/;
  49
  50         # some options are read globally, but can be overridden locally
  51         # per [svn-remote "..."] section.  Command-line options will *NOT*
  52         # override options set in an [svn-remote "..."] section
  53         no strict 'refs';
  54         for my $option (qw/follow_parent no_metadata use_svm_props
  55                            use_svnsync_props/) {
  56                 my $key = $option;
  57                 $key =~ tr/_//d;
  58                 my $prop = "-$option";
  59                 *$option = sub {
  60                         my ($self) = @_;
  61                         return $self->{$prop} if exists $self->{$prop};
  62                         my $k = "svn-remote.$self->{repo_id}.$key";
  63                         eval { command_oneline(qw/config --get/, $k) };
  64                         if ($@) {
  65                                 $self->{$prop} = ${"Git::SVN::_$option"};
  66                         } else {
  67                                 my $v = command_oneline(qw/config --bool/,$k);
  68                                 $self->{$prop} = $v eq 'false' ? 0 : 1;
  69                         }
  70                         return $self->{$prop};
  71                 }
  72         }
  73 }
  74
  75
  76 my (%LOCKFILES, %INDEX_FILES);
  77 END {
  78         unlink keys %LOCKFILES if %LOCKFILES;
  79         unlink keys %INDEX_FILES if %INDEX_FILES;
  80 }
  81
  82 sub resolve_local_globs {
  83         my ($url, $fetch, $glob_spec) = @_;
  84         return unless defined $glob_spec;
  85         my $ref = $glob_spec->{ref};
  86         my $path = $glob_spec->{path};
  87         foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
  88                 next unless m#^$ref->{regex}$#;
  89                 my $p = $1;
  90                 my $pathname = desanitize_refname($path->full_path($p));
  91                 my $refname = desanitize_refname($ref->full_path($p));
  92                 if (my $existing = $fetch->{$pathname}) {
  93                         if ($existing ne $refname) {
  94                                 die "Refspec conflict:\n",
  95                                     "existing: $existing\n",
  96                                     " globbed: $refname\n";
  97                         }
  98                         my $u = (::cmt_metadata("$refname"))[0];
  99                         $u =~ s!^\Q$url\E(/|$)!! or die
 100                           "$refname: '$url' not found in '$u'\n";
 101                         if ($pathname ne $u) {
 102                                 warn "W: Refspec glob conflict ",
 103                                      "(ref: $refname):\n",
 104                                      "expected path: $pathname\n",
 105                                      "    real path: $u\n",
 106                                      "Continuing ahead with $u\n";
 107                                 next;
 108                         }
 109                 } else {
 110                         $fetch->{$pathname} = $refname;
 111                 }
 112         }
 113 }
 114
 115 sub parse_revision_argument {
 116         my ($base, $head) = @_;
 117         if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
 118                 return ($base, $head);
 119         }
 120         return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
 121         return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
 122         return ($head, $head) if ($::_revision eq 'HEAD');
 123         return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
 124         return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
 125         die "revision argument: $::_revision not understood by git-svn\n";
 126 }
 127
 128 sub fetch_all {
 129         my ($repo_id, $remotes) = @_;
 130         if (ref $repo_id) {
 131                 my $gs = $repo_id;
 132                 $repo_id = undef;
 133                 $repo_id = $gs->{repo_id};
 134         }
 135         $remotes ||= read_all_remotes();
 136         my $remote = $remotes->{$repo_id} or
 137                      die "[svn-remote \"$repo_id\"] unknown\n";
 138         my $fetch = $remote->{fetch};
 139         my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
 140         my (@gs, @globs);
 141         my $ra = Git::SVN::Ra->new($url);
 142         my $uuid = $ra->get_uuid;
 143         my $head = $ra->get_latest_revnum;
 144
 145         # ignore errors, $head revision may not even exist anymore
 146         eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
 147         warn "W: $@\n" if $@;
 148
 149         my $base = defined $fetch ? $head : 0;
 150
 151         # read the max revs for wildcard expansion (branches/*, tags/*)
 152         foreach my $t (qw/branches tags/) {
 153                 defined $remote->{$t} or next;
 154                 push @globs, @{$remote->{$t}};
 155
 156                 my $max_rev = eval { tmp_config(qw/--int --get/,
 157                                          "svn-remote.$repo_id.${t}-maxRev") };
 158                 if (defined $max_rev && ($max_rev < $base)) {
 159                         $base = $max_rev;
 160                 } elsif (!defined $max_rev) {
 161                         $base = 0;
 162                 }
 163         }
 164
 165         if ($fetch) {
 166                 foreach my $p (sort keys %$fetch) {
 167                         my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
 168                         my $lr = $gs->rev_map_max;
 169                         if (defined $lr) {
 170                                 $base = $lr if ($lr < $base);
 171                         }
 172                         push @gs, $gs;
 173                 }
 174         }
 175
 176         ($base, $head) = parse_revision_argument($base, $head);
 177         $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
 178 }
 179
 180 sub read_all_remotes {
 181         my $r = {};
 182         my $use_svm_props = eval { command_oneline(qw/config --bool
 183             svn.useSvmProps/) };
 184         $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
 185         my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
 186         foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
 187                 if (m!^(.+)\.fetch=$svn_refspec$!) {
 188                         my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
 189                         die("svn-remote.$remote: remote ref '$remote_ref' "
 190                             . "must start with 'refs/'\n")
 191                                 unless $remote_ref =~ m{^refs/};
 192                         $local_ref = uri_decode($local_ref);
 193                         $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
 194                         $r->{$remote}->{svm} = {} if $use_svm_props;
 195                 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
 196                         $r->{$1}->{svm} = {};
 197                 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
 198                         $r->{$1}->{url} = $2;
 199                 } elsif (m!^(.+)\.pushurl=\s*(.*)\s*$!) {
 200                         $r->{$1}->{pushurl} = $2;
 201                 } elsif (m!^(.+)\.ignore-refs=\s*(.*)\s*$!) {
 202                         $r->{$1}->{ignore_refs_regex} = $2;
 203                 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
 204                         my ($remote, $t, $local_ref, $remote_ref) =
 205                                                              ($1, $2, $3, $4);
 206                         die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
 207                             . "must start with 'refs/'\n")
 208                                 unless $remote_ref =~ m{^refs/};
 209                         $local_ref = uri_decode($local_ref);
 210
 211                         require Git::SVN::GlobSpec;
 212                         my $rs = {
 213                             t => $t,
 214                             remote => $remote,
 215                             path => Git::SVN::GlobSpec->new($local_ref, 1),
 216                             ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
 217                         if (length($rs->{ref}->{right}) != 0) {
 218                                 die "The '*' glob character must be the last ",
 219                                     "character of '$remote_ref'\n";
 220                         }
 221                         push @{ $r->{$remote}->{$t} }, $rs;
 222                 }
 223         }
 224
 225         map {
 226                 if (defined $r->{$_}->{svm}) {
 227                         my $svm;
 228                         eval {
 229                                 my $section = "svn-remote.$_";
 230                                 $svm = {
 231                                         source => tmp_config('--get',
 232                                             "$section.svm-source"),
 233                                         replace => tmp_config('--get',
 234                                             "$section.svm-replace"),
 235                                 }
 236                         };
 237                         $r->{$_}->{svm} = $svm;
 238                 }
 239         } keys %$r;
 240
 241         foreach my $remote (keys %$r) {
 242                 foreach ( grep { defined $_ }
 243                           map { $r->{$remote}->{$_} } qw(branches tags) ) {
 244                         foreach my $rs ( @$_ ) {
 245                                 $rs->{ignore_refs_regex} =
 246                                     $r->{$remote}->{ignore_refs_regex};
 247                         }
 248                 }
 249         }
 250
 251         $r;
 252 }
 253
 254 sub init_vars {
 255         $_gc_nr = $_gc_period = 1000;
 256         if (defined $_repack || defined $_repack_flags) {
 257                warn "Repack options are obsolete; they have no effect.\n";
 258         }
 259 }
 260
 261 sub verify_remotes_sanity {
 262         return unless -d $ENV{GIT_DIR};
 263         my %seen;
 264         foreach (command(qw/config -l/)) {
 265                 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
 266                         if ($seen{$1}) {
 267                                 die "Remote ref refs/remote/$1 is tracked by",
 268                                     "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
 269                                     "Please resolve this ambiguity in ",
 270                                     "your git configuration file before ",
 271                                     "continuing\n";
 272                         }
 273                         $seen{$1} = $_;
 274                 }
 275         }
 276 }
 277
 278 sub find_existing_remote {
 279         my ($url, $remotes) = @_;
 280         return undef if $no_reuse_existing;
 281         my $existing;
 282         foreach my $repo_id (keys %$remotes) {
 283                 my $u = $remotes->{$repo_id}->{url} or next;
 284                 next if $u ne $url;
 285                 $existing = $repo_id;
 286                 last;
 287         }
 288         $existing;
 289 }
 290
 291 sub init_remote_config {
 292         my ($self, $url, $no_write) = @_;
 293         $url =~ s!/+$!!; # strip trailing slash
 294         my $r = read_all_remotes();
 295         my $existing = find_existing_remote($url, $r);
 296         if ($existing) {
 297                 unless ($no_write) {
 298                         print STDERR "Using existing ",
 299                                      "[svn-remote \"$existing\"]\n";
 300                 }
 301                 $self->{repo_id} = $existing;
 302         } elsif ($_minimize_url) {
 303                 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
 304                 $existing = find_existing_remote($min_url, $r);
 305                 if ($existing) {
 306                         unless ($no_write) {
 307                                 print STDERR "Using existing ",
 308                                              "[svn-remote \"$existing\"]\n";
 309                         }
 310                         $self->{repo_id} = $existing;
 311                 }
 312                 if ($min_url ne $url) {
 313                         unless ($no_write) {
 314                                 print STDERR "Using higher level of URL: ",
 315                                              "$url => $min_url\n";
 316                         }
 317                         my $old_path = $self->path;
 318                         $url =~ s!^\Q$min_url\E(/|$)!!;
 319                         if (length $old_path) {
 320                                 $url .= "/$old_path";
 321                         }
 322                         $self->path($url);
 323                         $url = $min_url;
 324                 }
 325         }
 326         my $orig_url;
 327         if (!$existing) {
 328                 # verify that we aren't overwriting anything:
 329                 $orig_url = eval {
 330                         command_oneline('config', '--get',
 331                                         "svn-remote.$self->{repo_id}.url")
 332                 };
 333                 if ($orig_url && ($orig_url ne $url)) {
 334                         die "svn-remote.$self->{repo_id}.url already set: ",
 335                             "$orig_url\nwanted to set to: $url\n";
 336                 }
 337         }
 338         my ($xrepo_id, $xpath) = find_ref($self->refname);
 339         if (!$no_write && defined $xpath) {
 340                 die "svn-remote.$xrepo_id.fetch already set to track ",
 341                     "$xpath:", $self->refname, "\n";
 342         }
 343         unless ($no_write) {
 344                 command_noisy('config',
 345                               "svn-remote.$self->{repo_id}.url", $url);
 346                 my $path = $self->path;
 347                 $path =~ s{^/}{};
 348                 $path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
 349                 $self->path($path);
 350                 command_noisy('config', '--add',
 351                               "svn-remote.$self->{repo_id}.fetch",
 352                               $self->path.":".$self->refname);
 353         }
 354         $self->{url} = $url;
 355 }
 356
 357 sub find_by_url { # repos_root and, path are optional
 358         my ($class, $full_url, $repos_root, $path) = @_;
 359
 360         return undef unless defined $full_url;
 361         remove_username($full_url);
 362         remove_username($repos_root) if defined $repos_root;
 363         my $remotes = read_all_remotes();
 364         if (defined $full_url && defined $repos_root && !defined $path) {
 365                 $path = $full_url;
 366                 $path =~ s#^\Q$repos_root\E(?:/|$)##;
 367         }
 368         foreach my $repo_id (keys %$remotes) {
 369                 my $u = $remotes->{$repo_id}->{url} or next;
 370                 remove_username($u);
 371                 next if defined $repos_root && $repos_root ne $u;
 372
 373                 my $fetch = $remotes->{$repo_id}->{fetch} || {};
 374                 foreach my $t (qw/branches tags/) {
 375                         foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
 376                                 resolve_local_globs($u, $fetch, $globspec);
 377                         }
 378                 }
 379                 my $p = $path;
 380                 my $rwr = rewrite_root({repo_id => $repo_id});
 381                 my $svm = $remotes->{$repo_id}->{svm}
 382                         if defined $remotes->{$repo_id}->{svm};
 383                 unless (defined $p) {
 384                         $p = $full_url;
 385                         my $z = $u;
 386                         my $prefix = '';
 387                         if ($rwr) {
 388                                 $z = $rwr;
 389                                 remove_username($z);
 390                         } elsif (defined $svm) {
 391                                 $z = $svm->{source};
 392                                 $prefix = $svm->{replace};
 393                                 $prefix =~ s#^\Q$u\E(?:/|$)##;
 394                                 $prefix =~ s#/$##;
 395                         }
 396                         $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
 397                 }
 398                 foreach my $f (keys %$fetch) {
 399                         next if $f ne $p;
 400                         return Git::SVN->new($fetch->{$f}, $repo_id, $f);
 401                 }
 402         }
 403         undef;
 404 }
 405
 406 sub init {
 407         my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
 408         my $self = _new($class, $repo_id, $ref_id, $path);
 409         if (defined $url) {
 410                 $self->init_remote_config($url, $no_write);
 411         }
 412         $self;
 413 }
 414
 415 sub find_ref {
 416         my ($ref_id) = @_;
 417         foreach (command(qw/config -l/)) {
 418                 next unless m!^svn-remote\.(.+)\.fetch=
 419                               \s*(.*?)\s*:\s*(.+?)\s*$!x;
 420                 my ($repo_id, $path, $ref) = ($1, $2, $3);
 421                 if ($ref eq $ref_id) {
 422                         $path = '' if ($path =~ m#^\./?#);
 423                         return ($repo_id, $path);
 424                 }
 425         }
 426         (undef, undef, undef);
 427 }
 428
 429 sub new {
 430         my ($class, $ref_id, $repo_id, $path) = @_;
 431         if (defined $ref_id && !defined $repo_id && !defined $path) {
 432                 ($repo_id, $path) = find_ref($ref_id);
 433                 if (!defined $repo_id) {
 434                         die "Could not find a \"svn-remote.*.fetch\" key ",
 435                             "in the repository configuration matching: ",
 436                             "$ref_id\n";
 437                 }
 438         }
 439         my $self = _new($class, $repo_id, $ref_id, $path);
 440         if (!defined $self->path || !length $self->path) {
 441                 my $fetch = command_oneline('config', '--get',
 442                                             "svn-remote.$repo_id.fetch",
 443                                             ":$ref_id\$") or
 444                      die "Failed to read \"svn-remote.$repo_id.fetch\" ",
 445                          "\":$ref_id\$\" in config\n";
 446                 my($path) = split(/\s*:\s*/, $fetch);
 447                 $self->path($path);
 448         }
 449         {
 450                 my $path = $self->path;
 451                 $path =~ s{/+}{/}g;
 452                 $path =~ s{\A/}{};
 453                 $path =~ s{/\z}{};
 454                 $self->path($path);
 455         }
 456         $self->{url} = command_oneline('config', '--get',
 457                                        "svn-remote.$repo_id.url") or
 458                   die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
 459         $self->{pushurl} = eval { command_oneline('config', '--get',
 460                                   "svn-remote.$repo_id.pushurl") };
 461         $self->rebuild;
 462         $self;
 463 }
 464
 465 sub refname {
 466         my ($refname) = $_[0]->{ref_id} ;
 467
 468         # It cannot end with a slash /, we'll throw up on this because
 469         # SVN can't have directories with a slash in their name, either:
 470         if ($refname =~ m{/$}) {
 471                 die "ref: '$refname' ends with a trailing slash, this is ",
 472                     "not permitted by git nor Subversion\n";
 473         }
 474
 475         # It cannot have ASCII control character space, tilde ~, caret ^,
 476         # colon :, question-mark ?, asterisk *, space, or open bracket [
 477         # anywhere.
 478         #
 479         # Additionally, % must be escaped because it is used for escaping
 480         # and we want our escaped refname to be reversible
 481         $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
 482
 483         # no slash-separated component can begin with a dot .
 484         # /.* becomes /%2E*
 485         $refname =~ s{/\.}{/%2E}g;
 486
 487         # It cannot have two consecutive dots .. anywhere
 488         # .. becomes %2E%2E
 489         $refname =~ s{\.\.}{%2E%2E}g;
 490
 491         # trailing dots and .lock are not allowed
 492         # .$ becomes %2E and .lock becomes %2Elock
 493         $refname =~ s{\.(?=$|lock$)}{%2E};
 494
 495         # the sequence @{ is used to access the reflog
 496         # @{ becomes %40{
 497         $refname =~ s{\@\{}{%40\{}g;
 498
 499         return $refname;
 500 }
 501
 502 sub desanitize_refname {
 503         my ($refname) = @_;
 504         $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
 505         return $refname;
 506 }
 507
 508 sub svm_uuid {
 509         my ($self) = @_;
 510         return $self->{svm}->{uuid} if $self->svm;
 511         $self->ra;
 512         unless ($self->{svm}) {
 513                 die "SVM UUID not cached, and reading remotely failed\n";
 514         }
 515         $self->{svm}->{uuid};
 516 }
 517
 518 sub svm {
 519         my ($self) = @_;
 520         return $self->{svm} if $self->{svm};
 521         my $svm;
 522         # see if we have it in our config, first:
 523         eval {
 524                 my $section = "svn-remote.$self->{repo_id}";
 525                 $svm = {
 526                   source => tmp_config('--get', "$section.svm-source"),
 527                   uuid => tmp_config('--get', "$section.svm-uuid"),
 528                   replace => tmp_config('--get', "$section.svm-replace"),
 529                 }
 530         };
 531         if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
 532                 $self->{svm} = $svm;
 533         }
 534         $self->{svm};
 535 }
 536
 537 sub _set_svm_vars {
 538         my ($self, $ra) = @_;
 539         return $ra if $self->svm;
 540
 541         my @err = ( "useSvmProps set, but failed to read SVM properties\n",
 542                     "(svm:source, svm:uuid) ",
 543                     "from the following URLs:\n" );
 544         sub read_svm_props {
 545                 my ($self, $ra, $path, $r) = @_;
 546                 my $props = ($ra->get_dir($path, $r))[2];
 547                 my $src = $props->{'svm:source'};
 548                 my $uuid = $props->{'svm:uuid'};
 549                 return undef if (!$src || !$uuid);
 550
 551                 chomp($src, $uuid);
 552
 553                 $uuid =~ m{^[0-9a-f\-]{30,}$}i
 554                     or die "doesn't look right - svm:uuid is '$uuid'\n";
 555
 556                 # the '!' is used to mark the repos_root!/relative/path
 557                 $src =~ s{/?!/?}{/};
 558                 $src =~ s{/+$}{}; # no trailing slashes please
 559                 # username is of no interest
 560                 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
 561
 562                 my $replace = $ra->{url};
 563                 $replace .= "/$path" if length $path;
 564
 565                 my $section = "svn-remote.$self->{repo_id}";
 566                 tmp_config("$section.svm-source", $src);
 567                 tmp_config("$section.svm-replace", $replace);
 568                 tmp_config("$section.svm-uuid", $uuid);
 569                 $self->{svm} = {
 570                         source => $src,
 571                         uuid => $uuid,
 572                         replace => $replace
 573                 };
 574         }
 575
 576         my $r = $ra->get_latest_revnum;
 577         my $path = $self->path;
 578         my %tried;
 579         while (length $path) {
 580                 unless ($tried{"$self->{url}/$path"}) {
 581                         return $ra if $self->read_svm_props($ra, $path, $r);
 582                         $tried{"$self->{url}/$path"} = 1;
 583                 }
 584                 $path =~ s#/?[^/]+$##;
 585         }
 586         die "Path: '$path' should be ''\n" if $path ne '';
 587         return $ra if $self->read_svm_props($ra, $path, $r);
 588         $tried{"$self->{url}/$path"} = 1;
 589
 590         if ($ra->{repos_root} eq $self->{url}) {
 591                 die @err, (map { "  $_\n" } keys %tried), "\n";
 592         }
 593
 594         # nope, make sure we're connected to the repository root:
 595         my $ok;
 596         my @tried_b;
 597         $path = $ra->{svn_path};
 598         $ra = Git::SVN::Ra->new($ra->{repos_root});
 599         while (length $path) {
 600                 unless ($tried{"$ra->{url}/$path"}) {
 601                         $ok = $self->read_svm_props($ra, $path, $r);
 602                         last if $ok;
 603                         $tried{"$ra->{url}/$path"} = 1;
 604                 }
 605                 $path =~ s#/?[^/]+$##;
 606         }
 607         die "Path: '$path' should be ''\n" if $path ne '';
 608         $ok ||= $self->read_svm_props($ra, $path, $r);
 609         $tried{"$ra->{url}/$path"} = 1;
 610         if (!$ok) {
 611                 die @err, (map { "  $_\n" } keys %tried), "\n";
 612         }
 613         Git::SVN::Ra->new($self->{url});
 614 }
 615
 616 sub svnsync {
 617         my ($self) = @_;
 618         return $self->{svnsync} if $self->{svnsync};
 619
 620         if ($self->no_metadata) {
 621                 die "Can't have both 'noMetadata' and ",
 622                     "'useSvnsyncProps' options set!\n";
 623         }
 624         if ($self->rewrite_root) {
 625                 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
 626                     "options set!\n";
 627         }
 628         if ($self->rewrite_uuid) {
 629                 die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
 630                     "options set!\n";
 631         }
 632
 633         my $svnsync;
 634         # see if we have it in our config, first:
 635         eval {
 636                 my $section = "svn-remote.$self->{repo_id}";
 637
 638                 my $url = tmp_config('--get', "$section.svnsync-url");
 639                 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
 640                    die "doesn't look right - svn:sync-from-url is '$url'\n";
 641
 642                 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
 643                 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
 644                    die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
 645
 646                 $svnsync = { url => $url, uuid => $uuid }
 647         };
 648         if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
 649                 return $self->{svnsync} = $svnsync;
 650         }
 651
 652         my $err = "useSvnsyncProps set, but failed to read " .
 653                   "svnsync property: svn:sync-from-";
 654         my $rp = $self->ra->rev_proplist(0);
 655
 656         my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
 657         ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
 658                    die "doesn't look right - svn:sync-from-url is '$url'\n";
 659
 660         my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
 661         ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
 662                    die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
 663
 664         my $section = "svn-remote.$self->{repo_id}";
 665         tmp_config('--add', "$section.svnsync-uuid", $uuid);
 666         tmp_config('--add', "$section.svnsync-url", $url);
 667         return $self->{svnsync} = { url => $url, uuid => $uuid };
 668 }
 669
 670 # this allows us to memoize our SVN::Ra UUID locally and avoid a
 671 # remote lookup (useful for 'git svn log').
 672 sub ra_uuid {
 673         my ($self) = @_;
 674         unless ($self->{ra_uuid}) {
 675                 my $key = "svn-remote.$self->{repo_id}.uuid";
 676                 my $uuid = eval { tmp_config('--get', $key) };
 677                 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
 678                         $self->{ra_uuid} = $uuid;
 679                 } else {
 680                         die "ra_uuid called without URL\n" unless $self->{url};
 681                         $self->{ra_uuid} = $self->ra->get_uuid;
 682                         tmp_config('--add', $key, $self->{ra_uuid});
 683                 }
 684         }
 685         $self->{ra_uuid};
 686 }
 687
 688 sub _set_repos_root {
 689         my ($self, $repos_root) = @_;
 690         my $k = "svn-remote.$self->{repo_id}.reposRoot";
 691         $repos_root ||= $self->ra->{repos_root};
 692         tmp_config($k, $repos_root);
 693         $repos_root;
 694 }
 695
 696 sub repos_root {
 697         my ($self) = @_;
 698         my $k = "svn-remote.$self->{repo_id}.reposRoot";
 699         eval { tmp_config('--get', $k) } || $self->_set_repos_root;
 700 }
 701
 702 sub ra {
 703         my ($self) = shift;
 704         my $ra = Git::SVN::Ra->new($self->{url});
 705         $self->_set_repos_root($ra->{repos_root});
 706         if ($self->use_svm_props && !$self->{svm}) {
 707                 if ($self->no_metadata) {
 708                         die "Can't have both 'noMetadata' and ",
 709                             "'useSvmProps' options set!\n";
 710                 } elsif ($self->use_svnsync_props) {
 711                         die "Can't have both 'useSvnsyncProps' and ",
 712                             "'useSvmProps' options set!\n";
 713                 }
 714                 $ra = $self->_set_svm_vars($ra);
 715                 $self->{-want_revprops} = 1;
 716         }
 717         $ra;
 718 }
 719
 720 # prop_walk(PATH, REV, SUB)
 721 # -------------------------
 722 # Recursively traverse PATH at revision REV and invoke SUB for each
 723 # directory that contains a SVN property.  SUB will be invoked as
 724 # follows:  &SUB(gs, path, props);  where `gs' is this instance of
 725 # Git::SVN, `path' the path to the directory where the properties
 726 # `props' were found.  The `path' will be relative to point of checkout,
 727 # that is, if url://repo/trunk is the current Git branch, and that
 728 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
 729 # as `path' (note the trailing `/').
 730 sub prop_walk {
 731         my ($self, $path, $rev, $sub) = @_;
 732
 733         $path =~ s#^/##;
 734         my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
 735         $path =~ s#^/*#/#g;
 736         my $p = $path;
 737         # Strip the irrelevant part of the path.
 738         $p =~ s#^/+\Q@{[$self->path]}\E(/|$)#/#;
 739         # Ensure the path is terminated by a `/'.
 740         $p =~ s#/*$#/#;
 741
 742         # The properties contain all the internal SVN stuff nobody
 743         # (usually) cares about.
 744         my $interesting_props = 0;
 745         foreach (keys %{$props}) {
 746                 # If it doesn't start with `svn:', it must be a
 747                 # user-defined property.
 748                 ++$interesting_props and next if $_ !~ /^svn:/;
 749                 # FIXME: Fragile, if SVN adds new public properties,
 750                 # this needs to be updated.
 751                 ++$interesting_props if /^svn:(?:ignore|keywords|executable
 752                                                  |eol-style|mime-type
 753                                                  |externals|needs-lock)$/x;
 754         }
 755         &$sub($self, $p, $props) if $interesting_props;
 756
 757         foreach (sort keys %$dirent) {
 758                 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
 759                 $self->prop_walk($self->path . $p . $_, $rev, $sub);
 760         }
 761 }
 762
 763 sub last_rev { ($_[0]->last_rev_commit)[0] }
 764 sub last_commit { ($_[0]->last_rev_commit)[1] }
 765
 766 # returns the newest SVN revision number and newest commit SHA1
 767 sub last_rev_commit {
 768         my ($self) = @_;
 769         if (defined $self->{last_rev} && defined $self->{last_commit}) {
 770                 return ($self->{last_rev}, $self->{last_commit});
 771         }
 772         my $c = ::verify_ref($self->refname.'^0');
 773         if ($c && !$self->use_svm_props && !$self->no_metadata) {
 774                 my $rev = (::cmt_metadata($c))[1];
 775                 if (defined $rev) {
 776                         ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
 777                         return ($rev, $c);
 778                 }
 779         }
 780         my $map_path = $self->map_path;
 781         unless (-e $map_path) {
 782                 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
 783                 return (undef, undef);
 784         }
 785         my ($rev, $commit) = $self->rev_map_max(1);
 786         ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
 787         return ($rev, $commit);
 788 }
 789
 790 sub get_fetch_range {
 791         my ($self, $min, $max) = @_;
 792         $max ||= $self->ra->get_latest_revnum;
 793         $min ||= $self->rev_map_max;
 794         (++$min, $max);
 795 }
 796
 797 sub tmp_config {
 798         my (@args) = @_;
 799         my $old_def_config = "$ENV{GIT_DIR}/svn/config";
 800         my $config = "$ENV{GIT_DIR}/svn/.metadata";
 801         if (! -f $config && -f $old_def_config) {
 802                 rename $old_def_config, $config or
 803                        die "Failed rename $old_def_config => $config: $!\n";
 804         }
 805         my $old_config = $ENV{GIT_CONFIG};
 806         $ENV{GIT_CONFIG} = $config;
 807         $@ = undef;
 808         my @ret = eval {
 809                 unless (-f $config) {
 810                         mkfile($config);
 811                         open my $fh, '>', $config or
 812                             die "Can't open $config: $!\n";
 813                         print $fh "; This file is used internally by ",
 814                                   "git-svn\n" or die
 815                                   "Couldn't write to $config: $!\n";
 816                         print $fh "; You should not have to edit it\n" or
 817                               die "Couldn't write to $config: $!\n";
 818                         close $fh or die "Couldn't close $config: $!\n";
 819                 }
 820                 command('config', @args);
 821         };
 822         my $err = $@;
 823         if (defined $old_config) {
 824                 $ENV{GIT_CONFIG} = $old_config;
 825         } else {
 826                 delete $ENV{GIT_CONFIG};
 827         }
 828         die $err if $err;
 829         wantarray ? @ret : $ret[0];
 830 }
 831
 832 sub tmp_index_do {
 833         my ($self, $sub) = @_;
 834         my $old_index = $ENV{GIT_INDEX_FILE};
 835         $ENV{GIT_INDEX_FILE} = $self->{index};
 836         $@ = undef;
 837         my @ret = eval {
 838                 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
 839                 mkpath([$dir]) unless -d $dir;
 840                 &$sub;
 841         };
 842         my $err = $@;
 843         if (defined $old_index) {
 844                 $ENV{GIT_INDEX_FILE} = $old_index;
 845         } else {
 846                 delete $ENV{GIT_INDEX_FILE};
 847         }
 848         die $err if $err;
 849         wantarray ? @ret : $ret[0];
 850 }
 851
 852 sub assert_index_clean {
 853         my ($self, $treeish) = @_;
 854
 855         $self->tmp_index_do(sub {
 856                 command_noisy('read-tree', $treeish) unless -e $self->{index};
 857                 my $x = command_oneline('write-tree');
 858                 my ($y) = (command(qw/cat-file commit/, $treeish) =~
 859                            /^tree ($::sha1)/mo);
 860                 return if $y eq $x;
 861
 862                 warn "Index mismatch: $y != $x\nrereading $treeish\n";
 863                 unlink $self->{index} or die "unlink $self->{index}: $!\n";
 864                 command_noisy('read-tree', $treeish);
 865                 $x = command_oneline('write-tree');
 866                 if ($y ne $x) {
 867                         fatal "trees ($treeish) $y != $x\n",
 868                               "Something is seriously wrong...";
 869                 }
 870         });
 871 }
 872
 873 sub get_commit_parents {
 874         my ($self, $log_entry) = @_;
 875         my (%seen, @ret, @tmp);
 876         # legacy support for 'set-tree'; this is only used by set_tree_cb:
 877         if (my $ip = $self->{inject_parents}) {
 878                 if (my $commit = delete $ip->{$log_entry->{revision}}) {
 879                         push @tmp, $commit;
 880                 }
 881         }
 882         if (my $cur = ::verify_ref($self->refname.'^0')) {
 883                 push @tmp, $cur;
 884         }
 885         if (my $ipd = $self->{inject_parents_dcommit}) {
 886                 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
 887                         push @tmp, @$commit;
 888                 }
 889         }
 890         push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
 891         while (my $p = shift @tmp) {
 892                 next if $seen{$p};
 893                 $seen{$p} = 1;
 894                 push @ret, $p;
 895         }
 896         @ret;
 897 }
 898
 899 sub rewrite_root {
 900         my ($self) = @_;
 901         return $self->{-rewrite_root} if exists $self->{-rewrite_root};
 902         my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
 903         my $rwr = eval { command_oneline(qw/config --get/, $k) };
 904         if ($rwr) {
 905                 $rwr =~ s#/+$##;
 906                 if ($rwr !~ m#^[a-z\+]+://#) {
 907                         die "$rwr is not a valid URL (key: $k)\n";
 908                 }
 909         }
 910         $self->{-rewrite_root} = $rwr;
 911 }
 912
 913 sub rewrite_uuid {
 914         my ($self) = @_;
 915         return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
 916         my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
 917         my $rwid = eval { command_oneline(qw/config --get/, $k) };
 918         if ($rwid) {
 919                 $rwid =~ s#/+$##;
 920                 if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
 921                         die "$rwid is not a valid UUID (key: $k)\n";
 922                 }
 923         }
 924         $self->{-rewrite_uuid} = $rwid;
 925 }
 926
 927 sub metadata_url {
 928         my ($self) = @_;
 929         ($self->rewrite_root || $self->{url}) .
 930            (length $self->path ? '/' . $self->path : '');
 931 }
 932
 933 sub full_url {
 934         my ($self) = @_;
 935         $self->{url} . (length $self->path ? '/' . $self->path : '');
 936 }
 937
 938 sub full_pushurl {
 939         my ($self) = @_;
 940         if ($self->{pushurl}) {
 941                 return $self->{pushurl} . (length $self->path ? '/' .
 942                        $self->path : '');
 943         } else {
 944                 return $self->full_url;
 945         }
 946 }
 947
 948 sub set_commit_header_env {
 949         my ($log_entry) = @_;
 950         my %env;
 951         foreach my $ned (qw/NAME EMAIL DATE/) {
 952                 foreach my $ac (qw/AUTHOR COMMITTER/) {
 953                         $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
 954                 }
 955         }
 956
 957         $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
 958         $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
 959         $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
 960
 961         $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
 962                                                 ? $log_entry->{commit_name}
 963                                                 : $log_entry->{name};
 964         $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
 965                                                 ? $log_entry->{commit_email}
 966                                                 : $log_entry->{email};
 967         \%env;
 968 }
 969
 970 sub restore_commit_header_env {
 971         my ($env) = @_;
 972         foreach my $ned (qw/NAME EMAIL DATE/) {
 973                 foreach my $ac (qw/AUTHOR COMMITTER/) {
 974                         my $k = "GIT_${ac}_${ned}";
 975                         if (defined $env->{$k}) {
 976                                 $ENV{$k} = $env->{$k};
 977                         } else {
 978                                 delete $ENV{$k};
 979                         }
 980                 }
 981         }
 982 }
 983
 984 sub gc {
 985         command_noisy('gc', '--auto');
 986 };
 987
 988 sub do_git_commit {
 989         my ($self, $log_entry) = @_;
 990         my $lr = $self->last_rev;
 991         if (defined $lr && $lr >= $log_entry->{revision}) {
 992                 die "Last fetched revision of ", $self->refname,
 993                     " was r$lr, but we are about to fetch: ",
 994                     "r$log_entry->{revision}!\n";
 995         }
 996         if (my $c = $self->rev_map_get($log_entry->{revision})) {
 997                 croak "$log_entry->{revision} = $c already exists! ",
 998                       "Why are we refetching it?\n";
 999         }
1000         my $old_env = set_commit_header_env($log_entry);
1001         my $tree = $log_entry->{tree};
1002         if (!defined $tree) {
1003                 $tree = $self->tmp_index_do(sub {
1004                                             command_oneline('write-tree') });
1005         }
1006         die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1007
1008         my @exec = ('git', 'commit-tree', $tree);
1009         foreach ($self->get_commit_parents($log_entry)) {
1010                 push @exec, '-p', $_;
1011         }
1012         defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1013                                                                    or croak $!;
1014         binmode $msg_fh;
1015
1016         # we always get UTF-8 from SVN, but we may want our commits in
1017         # a different encoding.
1018         if (my $enc = Git::config('i18n.commitencoding')) {
1019                 require Encode;
1020                 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
1021         }
1022         print $msg_fh $log_entry->{log} or croak $!;
1023         restore_commit_header_env($old_env);
1024         unless ($self->no_metadata) {
1025                 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1026                               or croak $!;
1027         }
1028         $msg_fh->flush == 0 or croak $!;
1029         close $msg_fh or croak $!;
1030         chomp(my $commit = do { local $/; <$out_fh> });
1031         close $out_fh or croak $!;
1032         waitpid $pid, 0;
1033         croak $? if $?;
1034         if ($commit !~ /^$::sha1$/o) {
1035                 die "Failed to commit, invalid sha1: $commit\n";
1036         }
1037
1038         $self->rev_map_set($log_entry->{revision}, $commit, 1);
1039
1040         $self->{last_rev} = $log_entry->{revision};
1041         $self->{last_commit} = $commit;
1042         print "r$log_entry->{revision}" unless $::_q > 1;
1043         if (defined $log_entry->{svm_revision}) {
1044                  print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
1045                  $self->rev_map_set($log_entry->{svm_revision}, $commit,
1046                                    0, $self->svm_uuid);
1047         }
1048         print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
1049         if (--$_gc_nr == 0) {
1050                 $_gc_nr = $_gc_period;
1051                 gc();
1052         }
1053         return $commit;
1054 }
1055
1056 sub match_paths {
1057         my ($self, $paths, $r) = @_;
1058         return 1 if $self->path eq '';
1059         if (my $path = $paths->{"/".$self->path}) {
1060                 return ($path->{action} eq 'D') ? 0 : 1;
1061         }
1062         $self->{path_regex} ||= qr{^/\Q@{[$self->path]}\E/};
1063         if (grep /$self->{path_regex}/, keys %$paths) {
1064                 return 1;
1065         }
1066         my $c = '';
1067         foreach (split m#/#, $self->path) {
1068                 $c .= "/$_";
1069                 next unless ($paths->{$c} &&
1070                              ($paths->{$c}->{action} =~ /^[AR]$/));
1071                 if ($self->ra->check_path($self->path, $r) ==
1072                     $SVN::Node::dir) {
1073                         return 1;
1074                 }
1075         }
1076         return 0;
1077 }
1078
1079 sub find_parent_branch {
1080         my ($self, $paths, $rev) = @_;
1081         return undef unless $self->follow_parent;
1082         unless (defined $paths) {
1083                 my $err_handler = $SVN::Error::handler;
1084                 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1085                 $self->ra->get_log([$self->path], $rev, $rev, 0, 1, 1,
1086                                    sub { $paths = $_[0] });
1087                 $SVN::Error::handler = $err_handler;
1088         }
1089         return undef unless defined $paths;
1090
1091         # look for a parent from another branch:
1092         my @b_path_components = split m#/#, $self->path;
1093         my @a_path_components;
1094         my $i;
1095         while (@b_path_components) {
1096                 $i = $paths->{'/'.join('/', @b_path_components)};
1097                 last if $i && defined $i->{copyfrom_path};
1098                 unshift(@a_path_components, pop(@b_path_components));
1099         }
1100         return undef unless defined $i && defined $i->{copyfrom_path};
1101         my $branch_from = $i->{copyfrom_path};
1102         if (@a_path_components) {
1103                 print STDERR "branch_from: $branch_from => ";
1104                 $branch_from .= '/'.join('/', @a_path_components);
1105                 print STDERR $branch_from, "\n";
1106         }
1107         my $r = $i->{copyfrom_rev};
1108         my $repos_root = $self->ra->{repos_root};
1109         my $url = $self->ra->{url};
1110         my $new_url = $url . $branch_from;
1111         print STDERR  "Found possible branch point: ",
1112                       "$new_url => ", $self->full_url, ", $r\n"
1113                       unless $::_q > 1;
1114         $branch_from =~ s#^/##;
1115         my $gs = $self->other_gs($new_url, $url,
1116                                  $branch_from, $r, $self->{ref_id});
1117         my ($r0, $parent) = $gs->find_rev_before($r, 1);
1118         {
1119                 my ($base, $head);
1120                 if (!defined $r0 || !defined $parent) {
1121                         ($base, $head) = parse_revision_argument(0, $r);
1122                 } else {
1123                         if ($r0 < $r) {
1124                                 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
1125                                         0, 1, sub { $base = $_[1] - 1 });
1126                         }
1127                 }
1128                 if (defined $base && $base <= $r) {
1129                         $gs->fetch($base, $r);
1130                 }
1131                 ($r0, $parent) = $gs->find_rev_before($r, 1);
1132         }
1133         if (defined $r0 && defined $parent) {
1134                 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
1135                              unless $::_q > 1;
1136                 my $ed;
1137                 if ($self->ra->can_do_switch) {
1138                         $self->assert_index_clean($parent);
1139                         print STDERR "Following parent with do_switch\n"
1140                                      unless $::_q > 1;
1141                         # do_switch works with svn/trunk >= r22312, but that
1142                         # is not included with SVN 1.4.3 (the latest version
1143                         # at the moment), so we can't rely on it
1144                         $self->{last_rev} = $r0;
1145                         $self->{last_commit} = $parent;
1146                         $ed = Git::SVN::Fetcher->new($self, $gs->{path});
1147                         $gs->ra->gs_do_switch($r0, $rev, $gs,
1148                                               $self->full_url, $ed)
1149                           or die "SVN connection failed somewhere...\n";
1150                 } elsif ($self->ra->trees_match($new_url, $r0,
1151                                                 $self->full_url, $rev)) {
1152                         print STDERR "Trees match:\n",
1153                                      "  $new_url\@$r0\n",
1154                                      "  ${\$self->full_url}\@$rev\n",
1155                                      "Following parent with no changes\n"
1156                                      unless $::_q > 1;
1157                         $self->tmp_index_do(sub {
1158                             command_noisy('read-tree', $parent);
1159                         });
1160                         $self->{last_commit} = $parent;
1161                 } else {
1162                         print STDERR "Following parent with do_update\n"
1163                                      unless $::_q > 1;
1164                         $ed = Git::SVN::Fetcher->new($self);
1165                         $self->ra->gs_do_update($rev, $rev, $self, $ed)
1166                           or die "SVN connection failed somewhere...\n";
1167                 }
1168                 print STDERR "Successfully followed parent\n" unless $::_q > 1;
1169                 return $self->make_log_entry($rev, [$parent], $ed);
1170         }
1171         return undef;
1172 }
1173
1174 sub do_fetch {
1175         my ($self, $paths, $rev) = @_;
1176         my $ed;
1177         my ($last_rev, @parents);
1178         if (my $lc = $self->last_commit) {
1179                 # we can have a branch that was deleted, then re-added
1180                 # under the same name but copied from another path, in
1181                 # which case we'll have multiple parents (we don't
1182                 # want to break the original ref, nor lose copypath info):
1183                 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1184                         push @{$log_entry->{parents}}, $lc;
1185                         return $log_entry;
1186                 }
1187                 $ed = Git::SVN::Fetcher->new($self);
1188                 $last_rev = $self->{last_rev};
1189                 $ed->{c} = $lc;
1190                 @parents = ($lc);
1191         } else {
1192                 $last_rev = $rev;
1193                 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1194                         return $log_entry;
1195                 }
1196                 $ed = Git::SVN::Fetcher->new($self);
1197         }
1198         unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1199                 die "SVN connection failed somewhere...\n";
1200         }
1201         $self->make_log_entry($rev, \@parents, $ed);
1202 }
1203
1204 sub mkemptydirs {
1205         my ($self, $r) = @_;
1206
1207         sub scan {
1208                 my ($r, $empty_dirs, $line) = @_;
1209                 if (defined $r && $line =~ /^r(\d+)$/) {
1210                         return 0 if $1 > $r;
1211                 } elsif ($line =~ /^  \+empty_dir: (.+)$/) {
1212                         $empty_dirs->{$1} = 1;
1213                 } elsif ($line =~ /^  \-empty_dir: (.+)$/) {
1214                         my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
1215                         delete @$empty_dirs{@d};
1216                 }
1217                 1; # continue
1218         };
1219
1220         my %empty_dirs = ();
1221         my $gz_file = "$self->{dir}/unhandled.log.gz";
1222         if (-f $gz_file) {
1223                 if (!can_compress()) {
1224                         warn "Compress::Zlib could not be found; ",
1225                              "empty directories in $gz_file will not be read\n";
1226                 } else {
1227                         my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
1228                                 die "Unable to open $gz_file: $!\n";
1229                         my $line;
1230                         while ($gz->gzreadline($line) > 0) {
1231                                 scan($r, \%empty_dirs, $line) or last;
1232                         }
1233                         $gz->gzclose;
1234                 }
1235         }
1236
1237         if (open my $fh, '<', "$self->{dir}/unhandled.log") {
1238                 binmode $fh or croak "binmode: $!";
1239                 while (<$fh>) {
1240                         scan($r, \%empty_dirs, $_) or last;
1241                 }
1242                 close $fh;
1243         }
1244
1245         my $strip = qr/\A\Q@{[$self->path]}\E(?:\/|$)/;
1246         foreach my $d (sort keys %empty_dirs) {
1247                 $d = uri_decode($d);
1248                 $d =~ s/$strip//;
1249                 next unless length($d);
1250                 next if -d $d;
1251                 if (-e $d) {
1252                         warn "$d exists but is not a directory\n";
1253                 } else {
1254                         print "creating empty directory: $d\n";
1255                         mkpath([$d]);
1256                 }
1257         }
1258 }
1259
1260 sub get_untracked {
1261         my ($self, $ed) = @_;
1262         my @out;
1263         my $h = $ed->{empty};
1264         foreach (sort keys %$h) {
1265                 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1266                 push @out, "  $act: " . uri_encode($_);
1267                 warn "W: $act: $_\n";
1268         }
1269         foreach my $t (qw/dir_prop file_prop/) {
1270                 $h = $ed->{$t} or next;
1271                 foreach my $path (sort keys %$h) {
1272                         my $ppath = $path eq '' ? '.' : $path;
1273                         foreach my $prop (sort keys %{$h->{$path}}) {
1274                                 next if $SKIP_PROP{$prop};
1275                                 my $v = $h->{$path}->{$prop};
1276                                 my $t_ppath_prop = "$t: " .
1277                                                     uri_encode($ppath) . ' ' .
1278                                                     uri_encode($prop);
1279                                 if (defined $v) {
1280                                         push @out, "  +$t_ppath_prop " .
1281                                                    uri_encode($v);
1282                                 } else {
1283                                         push @out, "  -$t_ppath_prop";
1284                                 }
1285                         }
1286                 }
1287         }
1288         foreach my $t (qw/absent_file absent_directory/) {
1289                 $h = $ed->{$t} or next;
1290                 foreach my $parent (sort keys %$h) {
1291                         foreach my $path (sort @{$h->{$parent}}) {
1292                                 push @out, "  $t: " .
1293                                            uri_encode("$parent/$path");
1294                                 warn "W: $t: $parent/$path ",
1295                                      "Insufficient permissions?\n";
1296                         }
1297                 }
1298         }
1299         \@out;
1300 }
1301
1302 sub get_tz {
1303         # some systmes don't handle or mishandle %z, so be creative.
1304         my $t = shift || time;
1305         my $gm = timelocal(gmtime($t));
1306         my $sign = qw( + + - )[ $t <=> $gm ];
1307         return sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
1308 }
1309
1310 # parse_svn_date(DATE)
1311 # --------------------
1312 # Given a date (in UTC) from Subversion, return a string in the format
1313 # "<TZ Offset> <local date/time>" that Git will use.
1314 #
1315 # By default the parsed date will be in UTC; if $Git::SVN::_localtime
1316 # is true we'll convert it to the local timezone instead.
1317 sub parse_svn_date {
1318         my $date = shift || return '+0000 1970-01-01 00:00:00';
1319         my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1320                                             (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
1321                                          croak "Unable to parse date: $date\n";
1322         my $parsed_date;    # Set next.
1323
1324         if ($Git::SVN::_localtime) {
1325                 # Translate the Subversion datetime to an epoch time.
1326                 # Begin by switching ourselves to $date's timezone, UTC.
1327                 my $old_env_TZ = $ENV{TZ};
1328                 $ENV{TZ} = 'UTC';
1329
1330                 my $epoch_in_UTC =
1331                     POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
1332
1333                 # Determine our local timezone (including DST) at the
1334                 # time of $epoch_in_UTC.  $Git::SVN::Log::TZ stored the
1335                 # value of TZ, if any, at the time we were run.
1336                 if (defined $Git::SVN::Log::TZ) {
1337                         $ENV{TZ} = $Git::SVN::Log::TZ;
1338                 } else {
1339                         delete $ENV{TZ};
1340                 }
1341
1342                 my $our_TZ = get_tz();
1343
1344                 # This converts $epoch_in_UTC into our local timezone.
1345                 my ($sec, $min, $hour, $mday, $mon, $year,
1346                     $wday, $yday, $isdst) = localtime($epoch_in_UTC);
1347
1348                 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
1349                                        $our_TZ, $year + 1900, $mon + 1,
1350                                        $mday, $hour, $min, $sec);
1351
1352                 # Reset us to the timezone in effect when we entered
1353                 # this routine.
1354                 if (defined $old_env_TZ) {
1355                         $ENV{TZ} = $old_env_TZ;
1356                 } else {
1357                         delete $ENV{TZ};
1358                 }
1359         } else {
1360                 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
1361         }
1362
1363         return $parsed_date;
1364 }
1365
1366 sub other_gs {
1367         my ($self, $new_url, $url,
1368             $branch_from, $r, $old_ref_id) = @_;
1369         my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
1370         unless ($gs) {
1371                 my $ref_id = $old_ref_id;
1372                 $ref_id =~ s/\@\d+-*$//;
1373                 $ref_id .= "\@$r";
1374                 # just grow a tail if we're not unique enough :x
1375                 $ref_id .= '-' while find_ref($ref_id);
1376                 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
1377                 if ($u =~ s#^\Q$url\E(/|$)##) {
1378                         $p = $u;
1379                         $u = $url;
1380                         $repo_id = $self->{repo_id};
1381                 }
1382                 while (1) {
1383                         # It is possible to tag two different subdirectories at
1384                         # the same revision.  If the url for an existing ref
1385                         # does not match, we must either find a ref with a
1386                         # matching url or create a new ref by growing a tail.
1387                         $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
1388                         my (undef, $max_commit) = $gs->rev_map_max(1);
1389                         last if (!$max_commit);
1390                         my ($url) = ::cmt_metadata($max_commit);
1391                         last if ($url eq $gs->metadata_url);
1392                         $ref_id .= '-';
1393                 }
1394                 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
1395         }
1396         $gs
1397 }
1398
1399 sub call_authors_prog {
1400         my ($orig_author) = @_;
1401         $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
1402         my $author = `$::_authors_prog $orig_author`;
1403         if ($? != 0) {
1404                 die "$::_authors_prog failed with exit code $?\n"
1405         }
1406         if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
1407                 my ($name, $email) = ($1, $2);
1408                 $email = undef if length $2 == 0;
1409                 return [$name, $email];
1410         } else {
1411                 die "Author: $orig_author: $::_authors_prog returned "
1412                         . "invalid author format: $author\n";
1413         }
1414 }
1415
1416 sub check_author {
1417         my ($author) = @_;
1418         if (!defined $author || length $author == 0) {
1419                 $author = '(no author)';
1420         }
1421         if (!defined $::users{$author}) {
1422                 if (defined $::_authors_prog) {
1423                         $::users{$author} = call_authors_prog($author);
1424                 } elsif (defined $::_authors) {
1425                         die "Author: $author not defined in $::_authors file\n";
1426                 }
1427         }
1428         $author;
1429 }
1430
1431 sub find_extra_svk_parents {
1432         my ($self, $ed, $tickets, $parents) = @_;
1433         # aha!  svk:merge property changed...
1434         my @tickets = split "\n", $tickets;
1435         my @known_parents;
1436         for my $ticket ( @tickets ) {
1437                 my ($uuid, $path, $rev) = split /:/, $ticket;
1438                 if ( $uuid eq $self->ra_uuid ) {
1439                         my $url = $self->{url};
1440                         my $repos_root = $url;
1441                         my $branch_from = $path;
1442                         $branch_from =~ s{^/}{};
1443                         my $gs = $self->other_gs($repos_root."/".$branch_from,
1444                                                  $url,
1445                                                  $branch_from,
1446                                                  $rev,
1447                                                  $self->{ref_id});
1448                         if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
1449                                 # wahey!  we found it, but it might be
1450                                 # an old one (!)
1451                                 push @known_parents, [ $rev, $commit ];
1452                         }
1453                 }
1454         }
1455         # Ordering matters; highest-numbered commit merge tickets
1456         # first, as they may account for later merge ticket additions
1457         # or changes.
1458         @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
1459         for my $parent ( @known_parents ) {
1460                 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
1461                 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
1462                 my $new;
1463                 while ( <$msg_fh> ) {
1464                         $new=1;last;
1465                 }
1466                 command_close_pipe($msg_fh, $ctx);
1467                 if ( $new ) {
1468                         print STDERR
1469                             "Found merge parent (svk:merge ticket): $parent\n";
1470                         push @$parents, $parent;
1471                 }
1472         }
1473 }
1474
1475 sub lookup_svn_merge {
1476         my $uuid = shift;
1477         my $url = shift;
1478         my $merge = shift;
1479
1480         my ($source, $revs) = split ":", $merge;
1481         my $path = $source;
1482         $path =~ s{^/}{};
1483         my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
1484         if ( !$gs ) {
1485                 warn "Couldn't find revmap for $url$source\n";
1486                 return;
1487         }
1488         my @ranges = split ",", $revs;
1489         my ($tip, $tip_commit);
1490         my @merged_commit_ranges;
1491         # find the tip
1492         for my $range ( @ranges ) {
1493                 my ($bottom, $top) = split "-", $range;
1494                 $top ||= $bottom;
1495                 my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
1496                 my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
1497
1498                 unless ($top_commit and $bottom_commit) {
1499                         warn "W:unknown path/rev in svn:mergeinfo "
1500                                 ."dirprop: $source:$range\n";
1501                         next;
1502                 }
1503
1504                 if (scalar(command('rev-parse', "$bottom_commit^@"))) {
1505                         push @merged_commit_ranges,
1506                              "$bottom_commit^..$top_commit";
1507                 } else {
1508                         push @merged_commit_ranges, "$top_commit";
1509                 }
1510
1511                 if ( !defined $tip or $top > $tip ) {
1512                         $tip = $top;
1513                         $tip_commit = $top_commit;
1514                 }
1515         }
1516         return ($tip_commit, @merged_commit_ranges);
1517 }
1518
1519 sub _rev_list {
1520         my ($msg_fh, $ctx) = command_output_pipe(
1521                 "rev-list", @_,
1522                );
1523         my @rv;
1524         while ( <$msg_fh> ) {
1525                 chomp;
1526                 push @rv, $_;
1527         }
1528         command_close_pipe($msg_fh, $ctx);
1529         @rv;
1530 }
1531
1532 sub check_cherry_pick {
1533         my $base = shift;
1534         my $tip = shift;
1535         my $parents = shift;
1536         my @ranges = @_;
1537         my %commits = map { $_ => 1 }
1538                 _rev_list("--no-merges", $tip, "--not", $base, @$parents, "--");
1539         for my $range ( @ranges ) {
1540                 delete @commits{_rev_list($range, "--")};
1541         }
1542         for my $commit (keys %commits) {
1543                 if (has_no_changes($commit)) {
1544                         delete $commits{$commit};
1545                 }
1546         }
1547         return (keys %commits);
1548 }
1549
1550 sub has_no_changes {
1551         my $commit = shift;
1552
1553         my @revs = split / /, command_oneline(
1554                 qw(rev-list --parents -1 -m), $commit);
1555
1556         # Commits with no parents, e.g. the start of a partial branch,
1557         # have changes by definition.
1558         return 1 if (@revs < 2);
1559
1560         # Commits with multiple parents, e.g a merge, have no changes
1561         # by definition.
1562         return 0 if (@revs > 2);
1563
1564         return (command_oneline("rev-parse", "$commit^{tree}") eq
1565                 command_oneline("rev-parse", "$commit~1^{tree}"));
1566 }
1567
1568 sub tie_for_persistent_memoization {
1569         my $hash = shift;
1570         my $path = shift;
1571
1572         if ($can_use_yaml) {
1573                 tie %$hash => 'Git::SVN::Memoize::YAML', "$path.yaml";
1574         } else {
1575                 tie %$hash => 'Memoize::Storable', "$path.db", 'nstore';
1576         }
1577 }
1578
1579 # The GIT_DIR environment variable is not always set until after the command
1580 # line arguments are processed, so we can't memoize in a BEGIN block.
1581 {
1582         my $memoized = 0;
1583
1584         sub memoize_svn_mergeinfo_functions {
1585                 return if $memoized;
1586                 $memoized = 1;
1587
1588                 my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
1589                 mkpath([$cache_path]) unless -d $cache_path;
1590
1591                 my %lookup_svn_merge_cache;
1592                 my %check_cherry_pick_cache;
1593                 my %has_no_changes_cache;
1594
1595                 tie_for_persistent_memoization(\%lookup_svn_merge_cache,
1596                     "$cache_path/lookup_svn_merge");
1597                 memoize 'lookup_svn_merge',
1598                         SCALAR_CACHE => 'FAULT',
1599                         LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
1600                 ;
1601
1602                 tie_for_persistent_memoization(\%check_cherry_pick_cache,
1603                     "$cache_path/check_cherry_pick");
1604                 memoize 'check_cherry_pick',
1605                         SCALAR_CACHE => 'FAULT',
1606                         LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
1607                 ;
1608
1609                 tie_for_persistent_memoization(\%has_no_changes_cache,
1610                     "$cache_path/has_no_changes");
1611                 memoize 'has_no_changes',
1612                         SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
1613                         LIST_CACHE => 'FAULT',
1614                 ;
1615         }
1616
1617         sub unmemoize_svn_mergeinfo_functions {
1618                 return if not $memoized;
1619                 $memoized = 0;
1620
1621                 Memoize::unmemoize 'lookup_svn_merge';
1622                 Memoize::unmemoize 'check_cherry_pick';
1623                 Memoize::unmemoize 'has_no_changes';
1624         }
1625
1626         Memoize::memoize 'Git::SVN::repos_root';
1627 }
1628
1629 END {
1630         # Force cache writeout explicitly instead of waiting for
1631         # global destruction to avoid segfault in Storable:
1632         # http://rt.cpan.org/Public/Bug/Display.html?id=36087
1633         unmemoize_svn_mergeinfo_functions();
1634 }
1635
1636 sub parents_exclude {
1637         my $parents = shift;
1638         my @commits = @_;
1639         return unless @commits;
1640
1641         my @excluded;
1642         my $excluded;
1643         do {
1644                 my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
1645                 $excluded = command_oneline(@cmd);
1646                 if ( $excluded ) {
1647                         my @new;
1648                         my $found;
1649                         for my $commit ( @commits ) {
1650                                 if ( $commit eq $excluded ) {
1651                                         push @excluded, $commit;
1652                                         $found++;
1653                                         last;
1654                                 }
1655                                 else {
1656                                         push @new, $commit;
1657                                 }
1658                         }
1659                         die "saw commit '$excluded' in rev-list output, "
1660                                 ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
1661                                         unless $found;
1662                         @commits = @new;
1663                 }
1664         }
1665                 while ($excluded and @commits);
1666
1667         return @excluded;
1668 }
1669
1670
1671 # note: this function should only be called if the various dirprops
1672 # have actually changed
1673 sub find_extra_svn_parents {
1674         my ($self, $ed, $mergeinfo, $parents) = @_;
1675         # aha!  svk:merge property changed...
1676
1677         memoize_svn_mergeinfo_functions();
1678
1679         # We first search for merged tips which are not in our
1680         # history.  Then, we figure out which git revisions are in
1681         # that tip, but not this revision.  If all of those revisions
1682         # are now marked as merge, we can add the tip as a parent.
1683         my @merges = split "\n", $mergeinfo;
1684         my @merge_tips;
1685         my $url = $self->{url};
1686         my $uuid = $self->ra_uuid;
1687         my %ranges;
1688         for my $merge ( @merges ) {
1689                 my ($tip_commit, @ranges) =
1690                         lookup_svn_merge( $uuid, $url, $merge );
1691                 unless (!$tip_commit or
1692                                 grep { $_ eq $tip_commit } @$parents ) {
1693                         push @merge_tips, $tip_commit;
1694                         $ranges{$tip_commit} = \@ranges;
1695                 } else {
1696                         push @merge_tips, undef;
1697                 }
1698         }
1699
1700         my %excluded = map { $_ => 1 }
1701                 parents_exclude($parents, grep { defined } @merge_tips);
1702
1703         # check merge tips for new parents
1704         my @new_parents;
1705         for my $merge_tip ( @merge_tips ) {
1706                 my $spec = shift @merges;
1707                 next unless $merge_tip and $excluded{$merge_tip};
1708
1709                 my $ranges = $ranges{$merge_tip};
1710
1711                 # check out 'new' tips
1712                 my $merge_base;
1713                 eval {
1714                         $merge_base = command_oneline(
1715                                 "merge-base",
1716                                 @$parents, $merge_tip,
1717                         );
1718                 };
1719                 if ($@) {
1720                         die "An error occurred during merge-base"
1721                                 unless $@->isa("Git::Error::Command");
1722
1723                         warn "W: Cannot find common ancestor between ".
1724                              "@$parents and $merge_tip. Ignoring merge info.\n";
1725                         next;
1726                 }
1727
1728                 # double check that there are no missing non-merge commits
1729                 my (@incomplete) = check_cherry_pick(
1730                         $merge_base, $merge_tip,
1731                         $parents,
1732                         @$ranges,
1733                        );
1734
1735                 if ( @incomplete ) {
1736                         warn "W:svn cherry-pick ignored ($spec) - missing "
1737                                 .@incomplete." commit(s) (eg $incomplete[0])\n";
1738                 } else {
1739                         warn
1740                                 "Found merge parent (svn:mergeinfo prop): ",
1741                                         $merge_tip, "\n";
1742                         push @new_parents, $merge_tip;
1743                 }
1744         }
1745
1746         # cater for merges which merge commits from multiple branches
1747         if ( @new_parents > 1 ) {
1748                 for ( my $i = 0; $i <= $#new_parents; $i++ ) {
1749                         for ( my $j = 0; $j <= $#new_parents; $j++ ) {
1750                                 next if $i == $j;
1751                                 next unless $new_parents[$i];
1752                                 next unless $new_parents[$j];
1753                                 my $revs = command_oneline(
1754                                         "rev-list", "-1",
1755                                         "$new_parents[$i]..$new_parents[$j]",
1756                                        );
1757                                 if ( !$revs ) {
1758                                         undef($new_parents[$j]);
1759                                 }
1760                         }
1761                 }
1762         }
1763         push @$parents, grep { defined } @new_parents;
1764 }
1765
1766 sub make_log_entry {
1767         my ($self, $rev, $parents, $ed) = @_;
1768         my $untracked = $self->get_untracked($ed);
1769
1770         my @parents = @$parents;
1771         my $ps = $ed->{path_strip} || "";
1772         for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
1773                 my $props = $ed->{dir_prop}{$path};
1774                 if ( $props->{"svk:merge"} ) {
1775                         $self->find_extra_svk_parents
1776                                 ($ed, $props->{"svk:merge"}, \@parents);
1777                 }
1778                 if ( $props->{"svn:mergeinfo"} ) {
1779                         $self->find_extra_svn_parents
1780                                 ($ed,
1781                                  $props->{"svn:mergeinfo"},
1782                                  \@parents);
1783                 }
1784         }
1785
1786         open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1787         print $un "r$rev\n" or croak $!;
1788         print $un $_, "\n" foreach @$untracked;
1789         my %log_entry = ( parents => \@parents, revision => $rev,
1790                           log => '');
1791
1792         my $headrev;
1793         my $logged = delete $self->{logged_rev_props};
1794         if (!$logged || $self->{-want_revprops}) {
1795                 my $rp = $self->ra->rev_proplist($rev);
1796                 foreach (sort keys %$rp) {
1797                         my $v = $rp->{$_};
1798                         if (/^svn:(author|date|log)$/) {
1799                                 $log_entry{$1} = $v;
1800                         } elsif ($_ eq 'svm:headrev') {
1801                                 $headrev = $v;
1802                         } else {
1803                                 print $un "  rev_prop: ", uri_encode($_), ' ',
1804                                           uri_encode($v), "\n";
1805                         }
1806                 }
1807         } else {
1808                 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1809         }
1810         close $un or croak $!;
1811
1812         $log_entry{date} = parse_svn_date($log_entry{date});
1813         $log_entry{log} .= "\n";
1814         my $author = $log_entry{author} = check_author($log_entry{author});
1815         my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1816                                                        : ($author, undef);
1817
1818         my ($commit_name, $commit_email) = ($name, $email);
1819         if ($_use_log_author) {
1820                 my $name_field;
1821                 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
1822                         $name_field = $1;
1823                 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
1824                         $name_field = $1;
1825                 }
1826                 if (!defined $name_field) {
1827                         if (!defined $email) {
1828                                 $email = $name;
1829                         }
1830                 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
1831                         ($name, $email) = ($1, $2);
1832                 } elsif ($name_field =~ /(.*)@/) {
1833                         ($name, $email) = ($1, $name_field);
1834                 } else {
1835                         ($name, $email) = ($name_field, $name_field);
1836                 }
1837         }
1838         if (defined $headrev && $self->use_svm_props) {
1839                 if ($self->rewrite_root) {
1840                         die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
1841                             "options set!\n";
1842                 }
1843                 if ($self->rewrite_uuid) {
1844                         die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
1845                             "options set!\n";
1846                 }
1847                 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
1848                 # we don't want "SVM: initializing mirror for junk" ...
1849                 return undef if $r == 0;
1850                 my $svm = $self->svm;
1851                 if ($uuid ne $svm->{uuid}) {
1852                         die "UUID mismatch on SVM path:\n",
1853                             "expected: $svm->{uuid}\n",
1854                             "     got: $uuid\n";
1855                 }
1856                 my $full_url = $self->full_url;
1857                 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
1858                              die "Failed to replace '$svm->{replace}' with ",
1859                                  "'$svm->{source}' in $full_url\n";
1860                 # throw away username for storing in records
1861                 remove_username($full_url);
1862                 $log_entry{metadata} = "$full_url\@$r $uuid";
1863                 $log_entry{svm_revision} = $r;
1864                 $email ||= "$author\@$uuid";
1865                 $commit_email ||= "$author\@$uuid";
1866         } elsif ($self->use_svnsync_props) {
1867                 my $full_url = $self->svnsync->{url};
1868                 $full_url .= "/".$self->path if length $self->path;
1869                 remove_username($full_url);
1870                 my $uuid = $self->svnsync->{uuid};
1871                 $log_entry{metadata} = "$full_url\@$rev $uuid";
1872                 $email ||= "$author\@$uuid";
1873                 $commit_email ||= "$author\@$uuid";
1874         } else {
1875                 my $url = $self->metadata_url;
1876                 remove_username($url);
1877                 my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
1878                 $log_entry{metadata} = "$url\@$rev " . $uuid;
1879                 $email ||= "$author\@" . $uuid;
1880                 $commit_email ||= "$author\@" . $uuid;
1881         }
1882         $log_entry{name} = $name;
1883         $log_entry{email} = $email;
1884         $log_entry{commit_name} = $commit_name;
1885         $log_entry{commit_email} = $commit_email;
1886         \%log_entry;
1887 }
1888
1889 sub fetch {
1890         my ($self, $min_rev, $max_rev, @parents) = @_;
1891         my ($last_rev, $last_commit) = $self->last_rev_commit;
1892         my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1893         $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1894 }
1895
1896 sub set_tree_cb {
1897         my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1898         $self->{inject_parents} = { $rev => $tree };
1899         $self->fetch(undef, undef);
1900 }
1901
1902 sub set_tree {
1903         my ($self, $tree) = (shift, shift);
1904         my $log_entry = ::get_commit_entry($tree);
1905         unless ($self->{last_rev}) {
1906                 fatal("Must have an existing revision to commit");
1907         }
1908         my %ed_opts = ( r => $self->{last_rev},
1909                         log => $log_entry->{log},
1910                         ra => $self->ra,
1911                         tree_a => $self->{last_commit},
1912                         tree_b => $tree,
1913                         editor_cb => sub {
1914                                $self->set_tree_cb($log_entry, $tree, @_) },
1915                         svn_path => $self->path );
1916         if (!Git::SVN::Editor->new(\%ed_opts)->apply_diff) {
1917                 print "No changes\nr$self->{last_rev} = $tree\n";
1918         }
1919 }
1920
1921 sub rebuild_from_rev_db {
1922         my ($self, $path) = @_;
1923         my $r = -1;
1924         open my $fh, '<', $path or croak "open: $!";
1925         binmode $fh or croak "binmode: $!";
1926         while (<$fh>) {
1927                 length($_) == 41 or croak "inconsistent size in ($_) != 41";
1928                 chomp($_);
1929                 ++$r;
1930                 next if $_ eq ('0' x 40);
1931                 $self->rev_map_set($r, $_);
1932                 print "r$r = $_\n";
1933         }
1934         close $fh or croak "close: $!";
1935         unlink $path or croak "unlink: $!";
1936 }
1937
1938 sub rebuild {
1939         my ($self) = @_;
1940         my $map_path = $self->map_path;
1941         my $partial = (-e $map_path && ! -z $map_path);
1942         return unless ::verify_ref($self->refname.'^0');
1943         if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
1944                 my $rev_db = $self->rev_db_path;
1945                 $self->rebuild_from_rev_db($rev_db);
1946                 if ($self->use_svm_props) {
1947                         my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
1948                         $self->rebuild_from_rev_db($svm_rev_db);
1949                 }
1950                 $self->unlink_rev_db_symlink;
1951                 return;
1952         }
1953         print "Rebuilding $map_path ...\n" if (!$partial);
1954         my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
1955                 (undef, undef));
1956         my ($log, $ctx) =
1957             command_output_pipe(qw/rev-list --pretty=raw --reverse/,
1958                                 ($head ? "$head.." : "") . $self->refname,
1959                                 '--');
1960         my $metadata_url = $self->metadata_url;
1961         remove_username($metadata_url);
1962         my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
1963         my $c;
1964         while (<$log>) {
1965                 if ( m{^commit ($::sha1)$} ) {
1966                         $c = $1;
1967                         next;
1968                 }
1969                 next unless s{^\s*(git-svn-id:)}{$1};
1970                 my ($url, $rev, $uuid) = ::extract_metadata($_);
1971                 remove_username($url);
1972
1973                 # ignore merges (from set-tree)
1974                 next if (!defined $rev || !$uuid);
1975
1976                 # if we merged or otherwise started elsewhere, this is
1977                 # how we break out of it
1978                 if (($uuid ne $svn_uuid) ||
1979                     ($metadata_url && $url && ($url ne $metadata_url))) {
1980                         next;
1981                 }
1982                 if ($partial && $head) {
1983                         print "Partial-rebuilding $map_path ...\n";
1984                         print "Currently at $base_rev = $head\n";
1985                         $head = undef;
1986                 }
1987
1988                 $self->rev_map_set($rev, $c);
1989                 print "r$rev = $c\n";
1990         }
1991         command_close_pipe($log, $ctx);
1992         print "Done rebuilding $map_path\n" if (!$partial || !$head);
1993         my $rev_db_path = $self->rev_db_path;
1994         if (-f $self->rev_db_path) {
1995                 unlink $self->rev_db_path or croak "unlink: $!";
1996         }
1997         $self->unlink_rev_db_symlink;
1998 }
1999
2000 # rev_map:
2001 # Tie::File seems to be prone to offset errors if revisions get sparse,
2002 # it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
2003 # one of my favorite modules is out :<  Next up would be one of the DBM
2004 # modules, but I'm not sure which is most portable...
2005 #
2006 # This is the replacement for the rev_db format, which was too big
2007 # and inefficient for large repositories with a lot of sparse history
2008 # (mainly tags)
2009 #
2010 # The format is this:
2011 #   - 24 bytes for every record,
2012 #     * 4 bytes for the integer representing an SVN revision number
2013 #     * 20 bytes representing the sha1 of a git commit
2014 #   - No empty padding records like the old format
2015 #     (except the last record, which can be overwritten)
2016 #   - new records are written append-only since SVN revision numbers
2017 #     increase monotonically
2018 #   - lookups on SVN revision number are done via a binary search
2019 #   - Piping the file to xxd -c24 is a good way of dumping it for
2020 #     viewing or editing (piped back through xxd -r), should the need
2021 #     ever arise.
2022 #   - The last record can be padding revision with an all-zero sha1
2023 #     This is used to optimize fetch performance when using multiple
2024 #     "fetch" directives in .git/config
2025 #
2026 # These files are disposable unless noMetadata or useSvmProps is set
2027
2028 sub _rev_map_set {
2029         my ($fh, $rev, $commit) = @_;
2030
2031         binmode $fh or croak "binmode: $!";
2032         my $size = (stat($fh))[7];
2033         ($size % 24) == 0 or croak "inconsistent size: $size";
2034
2035         my $wr_offset = 0;
2036         if ($size > 0) {
2037                 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2038                 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2039                 $read == 24 or croak "read only $read bytes (!= 24)";
2040                 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2041                 if ($last_commit eq ('0' x40)) {
2042                         if ($size >= 48) {
2043                                 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2044                                 $read = sysread($fh, $buf, 24) or
2045                                     croak "read: $!";
2046                                 $read == 24 or
2047                                     croak "read only $read bytes (!= 24)";
2048                                 ($last_rev, $last_commit) =
2049                                     unpack(rev_map_fmt, $buf);
2050                                 if ($last_commit eq ('0' x40)) {
2051                                         croak "inconsistent .rev_map\n";
2052                                 }
2053                         }
2054                         if ($last_rev >= $rev) {
2055                                 croak "last_rev is higher!: $last_rev >= $rev";
2056                         }
2057                         $wr_offset = -24;
2058                 }
2059         }
2060         sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2061         syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2062           croak "write: $!";
2063 }
2064
2065 sub _rev_map_reset {
2066         my ($fh, $rev, $commit) = @_;
2067         my $c = _rev_map_get($fh, $rev);
2068         $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
2069         my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
2070         truncate $fh, $offset or croak "truncate: $!";
2071 }
2072
2073 sub mkfile {
2074         my ($path) = @_;
2075         unless (-e $path) {
2076                 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2077                 mkpath([$dir]) unless -d $dir;
2078                 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2079                 close $fh or die "Couldn't close (create) $path: $!\n";
2080         }
2081 }
2082
2083 sub rev_map_set {
2084         my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2085         defined $commit or die "missing arg3\n";
2086         length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2087         my $db = $self->map_path($uuid);
2088         my $db_lock = "$db.lock";
2089         my $sigmask;
2090         $update_ref ||= 0;
2091         if ($update_ref) {
2092                 $sigmask = POSIX::SigSet->new();
2093                 my $signew = POSIX::SigSet->new(SIGINT, SIGHUP, SIGTERM,
2094                         SIGALRM, SIGUSR1, SIGUSR2);
2095                 sigprocmask(SIG_BLOCK, $signew, $sigmask) or
2096                         croak "Can't block signals: $!";
2097         }
2098         mkfile($db);
2099
2100         $LOCKFILES{$db_lock} = 1;
2101         my $sync;
2102         # both of these options make our .rev_db file very, very important
2103         # and we can't afford to lose it because rebuild() won't work
2104         if ($self->use_svm_props || $self->no_metadata) {
2105                 $sync = 1;
2106                 copy($db, $db_lock) or die "rev_map_set(@_): ",
2107                                            "Failed to copy: ",
2108                                            "$db => $db_lock ($!)\n";
2109         } else {
2110                 rename $db, $db_lock or die "rev_map_set(@_): ",
2111                                             "Failed to rename: ",
2112                                             "$db => $db_lock ($!)\n";
2113         }
2114
2115         sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2116              or croak "Couldn't open $db_lock: $!\n";
2117         $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
2118                                  _rev_map_set($fh, $rev, $commit);
2119         if ($sync) {
2120                 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2121                 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2122         }
2123         close $fh or croak $!;
2124         if ($update_ref) {
2125                 $_head = $self;
2126                 my $note = "";
2127                 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
2128                 command_noisy('update-ref', '-m', "r$rev$note",
2129                               $self->refname, $commit);
2130         }
2131         rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2132                                     "$db_lock => $db ($!)\n";
2133         delete $LOCKFILES{$db_lock};
2134         if ($update_ref) {
2135                 sigprocmask(SIG_SETMASK, $sigmask) or
2136                         croak "Can't restore signal mask: $!";
2137         }
2138 }
2139
2140 # If want_commit, this will return an array of (rev, commit) where
2141 # commit _must_ be a valid commit in the archive.
2142 # Otherwise, it'll return the max revision (whether or not the
2143 # commit is valid or just a 0x40 placeholder).
2144 sub rev_map_max {
2145         my ($self, $want_commit) = @_;
2146         $self->rebuild;
2147         my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
2148         $want_commit ? ($r, $c) : $r;
2149 }
2150
2151 sub rev_map_max_norebuild {
2152         my ($self, $want_commit) = @_;
2153         my $map_path = $self->map_path;
2154         stat $map_path or return $want_commit ? (0, undef) : 0;
2155         sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2156         binmode $fh or croak "binmode: $!";
2157         my $size = (stat($fh))[7];
2158         ($size % 24) == 0 or croak "inconsistent size: $size";
2159
2160         if ($size == 0) {
2161                 close $fh or croak "close: $!";
2162                 return $want_commit ? (0, undef) : 0;
2163         }
2164
2165         sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2166         sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2167         my ($r, $c) = unpack(rev_map_fmt, $buf);
2168         if ($want_commit && $c eq ('0' x40)) {
2169                 if ($size < 48) {
2170                         return $want_commit ? (0, undef) : 0;
2171                 }
2172                 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2173                 sysread($fh, $buf, 24) == 24 or croak "read: $!";
2174                 ($r, $c) = unpack(rev_map_fmt, $buf);
2175                 if ($c eq ('0'x40)) {
2176                         croak "Penultimate record is all-zeroes in $map_path";
2177                 }
2178         }
2179         close $fh or croak "close: $!";
2180         $want_commit ? ($r, $c) : $r;
2181 }
2182
2183 sub rev_map_get {
2184         my ($self, $rev, $uuid) = @_;
2185         my $map_path = $self->map_path($uuid);
2186         return undef unless -e $map_path;
2187
2188         sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2189         my $c = _rev_map_get($fh, $rev);
2190         close($fh) or croak "close: $!";
2191         $c
2192 }
2193
2194 sub _rev_map_get {
2195         my ($fh, $rev) = @_;
2196
2197         binmode $fh or croak "binmode: $!";
2198         my $size = (stat($fh))[7];
2199         ($size % 24) == 0 or croak "inconsistent size: $size";
2200
2201         if ($size == 0) {
2202                 return undef;
2203         }
2204
2205         my ($l, $u) = (0, $size - 24);
2206         my ($r, $c, $buf);
2207
2208         while ($l <= $u) {
2209                 my $i = int(($l/24 + $u/24) / 2) * 24;
2210                 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2211                 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2212                 my ($r, $c) = unpack(rev_map_fmt, $buf);
2213
2214                 if ($r < $rev) {
2215                         $l = $i + 24;
2216                 } elsif ($r > $rev) {
2217                         $u = $i - 24;
2218                 } else { # $r == $rev
2219                         return $c eq ('0' x 40) ? undef : $c;
2220                 }
2221         }
2222         undef;
2223 }
2224
2225 # Finds the first svn revision that exists on (if $eq_ok is true) or
2226 # before $rev for the current branch.  It will not search any lower
2227 # than $min_rev.  Returns the git commit hash and svn revision number
2228 # if found, else (undef, undef).
2229 sub find_rev_before {
2230         my ($self, $rev, $eq_ok, $min_rev) = @_;
2231         --$rev unless $eq_ok;
2232         $min_rev ||= 1;
2233         my $max_rev = $self->rev_map_max;
2234         $rev = $max_rev if ($rev > $max_rev);
2235         while ($rev >= $min_rev) {
2236                 if (my $c = $self->rev_map_get($rev)) {
2237                         return ($rev, $c);
2238                 }
2239                 --$rev;
2240         }
2241         return (undef, undef);
2242 }
2243
2244 # Finds the first svn revision that exists on (if $eq_ok is true) or
2245 # after $rev for the current branch.  It will not search any higher
2246 # than $max_rev.  Returns the git commit hash and svn revision number
2247 # if found, else (undef, undef).
2248 sub find_rev_after {
2249         my ($self, $rev, $eq_ok, $max_rev) = @_;
2250         ++$rev unless $eq_ok;
2251         $max_rev ||= $self->rev_map_max;
2252         while ($rev <= $max_rev) {
2253                 if (my $c = $self->rev_map_get($rev)) {
2254                         return ($rev, $c);
2255                 }
2256                 ++$rev;
2257         }
2258         return (undef, undef);
2259 }
2260
2261 sub _new {
2262         my ($class, $repo_id, $ref_id, $path) = @_;
2263         unless (defined $repo_id && length $repo_id) {
2264                 $repo_id = $default_repo_id;
2265         }
2266         unless (defined $ref_id && length $ref_id) {
2267                 # Access the prefix option from the git-svn main program if it's loaded.
2268                 my $prefix = defined &::opt_prefix ? ::opt_prefix() : "";
2269                 $_[2] = $ref_id =
2270                              "refs/remotes/$prefix$default_ref_id";
2271         }
2272         $_[1] = $repo_id;
2273         my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2274
2275         # Older repos imported by us used $GIT_DIR/svn/foo instead of
2276         # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
2277         if ($ref_id =~ m{^refs/remotes/(.*)}) {
2278                 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
2279                 if (-d $old_dir && ! -d $dir) {
2280                         $dir = $old_dir;
2281                 }
2282         }
2283
2284         $_[3] = $path = '' unless (defined $path);
2285         mkpath([$dir]);
2286         my $obj = bless {
2287                 ref_id => $ref_id, dir => $dir, index => "$dir/index",
2288                 config => "$ENV{GIT_DIR}/svn/config",
2289                 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
2290
2291         # Ensure it gets canonicalized
2292         $obj->path($path);
2293
2294         return $obj;
2295 }
2296
2297 sub path {
2298         my $self = shift;
2299
2300         if (@_) {
2301                 my $path = shift;
2302                 $self->{path} = $path;
2303                 return;
2304         }
2305
2306         return $self->{path};
2307 }
2308
2309 # for read-only access of old .rev_db formats
2310 sub unlink_rev_db_symlink {
2311         my ($self) = @_;
2312         my $link = $self->rev_db_path;
2313         $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
2314         if (-l $link) {
2315                 unlink $link or croak "unlink: $link failed!";
2316         }
2317 }
2318
2319 sub rev_db_path {
2320         my ($self, $uuid) = @_;
2321         my $db_path = $self->map_path($uuid);
2322         $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
2323             or croak "map_path: $db_path does not contain '/.rev_map.' !";
2324         $db_path;
2325 }
2326
2327 # the new replacement for .rev_db
2328 sub map_path {
2329         my ($self, $uuid) = @_;
2330         $uuid ||= $self->ra_uuid;
2331         "$self->{map_root}.$uuid";
2332 }
2333
2334 sub uri_encode {
2335         my ($f) = @_;
2336         $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2337         $f
2338 }
2339
2340 sub uri_decode {
2341         my ($f) = @_;
2342         $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
2343         $f
2344 }
2345
2346 sub remove_username {
2347         $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2348 }
2349
2350 1;