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