git.rot13.org Git - webpac2/blob - run.pl

   1 #!/usr/bin/perl -w
   2
   3 use strict;
   4
   5 use Cwd qw/abs_path/;
   6 use File::Temp qw/tempdir/;
   7 use lib './lib';
   8
   9 use WebPAC::Common 0.02;
  10 use WebPAC::Parser 0.08;
  11 use WebPAC::Input 0.16;
  12 use WebPAC::Store 0.15;
  13 use WebPAC::Normalize 0.22;
  14 #use WebPAC::Output::TT;
  15 use WebPAC::Validate 0.11;
  16 use WebPAC::Output::MARC;
  17 use WebPAC::Config;
  18 use Getopt::Long;
  19 use File::Path;
  20 use Time::HiRes qw/time/;
  21 use File::Slurp;
  22 use Data::Dump qw/dump/;
  23 use Storable qw/dclone/;
  24 use Pod::Usage qw/pod2usage/;
  25 use LWP::Simple qw//;
  26
  27 use POSIX ":sys_wait_h"; # imports WNOHANG
  28
  29 =head1 NAME
  30
  31 run.pl - start WebPAC indexing
  32
  33 B<this command will probably go away. Don't get used to it!>
  34
  35 =head1 OPTIONS
  36
  37 =over 4
  38
  39 =item --offset 42
  40
  41 start loading (all) databases at offset 42
  42
  43 =item --limit 100
  44
  45 limit loading to 100 records
  46
  47 =item --clean
  48
  49 remove database and Hyper Estraier index before indexing
  50
  51 =item --only=database_name/input_filter
  52
  53 reindex just single database (legacy name is --one)
  54
  55 C</input_filter> is optional part which can be C<name>
  56 or C<type> from input
  57
  58 =item --config conf/config.yml
  59
  60 path to YAML configuration file
  61
  62 =item --stats
  63
  64 disable indexing, modify_* in configuration and dump statistics about field
  65 and subfield usage for each input
  66
  67 =item --validate path/to/validation_file
  68
  69 turn on extra validation of imput records, see L<WebPAC::Validation>
  70
  71 You can use special variables C<$database> and $C<$input> in this parametar
  72 like C<--validate 'conf/validate/$database-$input'> to construct filename
  73
  74 =item --validate-delimiters path/to/validate_delimiters_file
  75
  76 this option is used with C<--validate> to turn on extra validation of
  77 delimiters. If file is non existant, it will be created on first run.
  78
  79 =item --marc-generate
  80
  81 Generate MARC file. This will automatically be on if file contains C<marc*> directives.
  82 You can use this option as C<--no-marc-generate> to disable MARC generation.
  83
  84 =item --marc-lint
  85
  86 By default turned on if normalisation file has C<marc*> directives. You can disable lint
  87 messages with C<--no-marc-lint>.
  88
  89 =item --marc-xml
  90
  91 Creeate MARCXML file (this can be quite large)
  92
  93 =item --marc-dump
  94
  95 Force dump or input and marc record for debugging.
  96
  97 =item --parallel 4
  98
  99 Run databases in parallel (aproximatly same as number of processors in
 100 machine if you want to use full load)
 101
 102 =item --only-links
 103
 104 Create just links
 105
 106 =item --merge
 107
 108 Create merged index of databases which have links
 109
 110 =item --mirror http://www.example.com
 111
 112 Tries to download input path files from mirror URI
 113
 114 =back
 115
 116 =cut
 117
 118 my $offset;
 119 my $limit;
 120
 121 my $clean = 0;
 122 my $config_path;
 123 my $debug = 0;
 124 my $only_filter;
 125 my $stats = 0;
 126 my $validate_path;
 127 my $validate_delimiters_path;
 128 my $marc_generate = 1;
 129 my $marc_lint = 0;
 130 my $marc_dump = 0;
 131 my $marc_xml = 0;
 132 my $parallel = 0;
 133 my $only_links = 0;
 134 my $merge = 0;
 135 my $mirror;
 136 my $help;
 137
 138 my $log = _new WebPAC::Common()->_get_logger();
 139
 140 GetOptions(
 141         "limit=i" => \$limit,
 142         "offset=i" => \$offset,
 143         "clean" => \$clean,
 144         "one=s" => \$only_filter,
 145         "only=s" => \$only_filter,
 146         "config=s" => \$config_path,
 147         "debug+" => \$debug,
 148         "stats" => \$stats,
 149         "validate=s" => \$validate_path,
 150         "validate-delimiters=s" => \$validate_delimiters_path,
 151         "marc-generate!" => \$marc_generate,
 152         "marc-lint!" => \$marc_lint,
 153         "marc-dump!" => \$marc_dump,
 154         "marcxml!" => \$marc_xml,
 155         "parallel=i" => \$parallel,
 156         "only-links!" => \$only_links,
 157         "merge" => \$merge,
 158         "mirror=s" => \$mirror,
 159         "help" => \$help,
 160 );
 161
 162 $marc_generate = 0 if ( $validate_delimiters_path );
 163
 164 pod2usage(-verbose => 2) if ($help);
 165
 166 my $config = new WebPAC::Config( path => $config_path );
 167
 168 WebPAC::Normalize::_debug( $debug - 1 ) if $debug > 1;
 169
 170 #print "config = ",dump($config) if ($debug);
 171
 172 die "no databases in config file!\n" unless ($config->databases);
 173
 174 $log->info( "-" x 79 );
 175
 176 my $log_file = 'log';
 177
 178 if (-e $log_file ) {    # && -s $log_file > 5 * 1024 * 1024) {
 179         $log->info("moved old log with ", -s $log_file, " bytes to '${log_file}.old'");
 180         rename $log_file, "${log_file}.old" || $log->logwarn("can't rename $log_file to ${log_file}.old: $!");
 181 }
 182
 183 my $estcmd_fh;
 184 my $estcmd_path = './estcmd-merge.sh';
 185 if ($merge) {
 186         open($estcmd_fh, '>', $estcmd_path) || $log->logdie("can't open $estcmd_path: $!");
 187         print $estcmd_fh 'cd /data/estraier/_node/ || exit 1',$/;
 188         print $estcmd_fh 'sudo /etc/init.d/hyperestraier stop',$/;
 189         $log->info("created merge batch file $estcmd_path");
 190 }
 191
 192 my $validate;
 193 $validate = new WebPAC::Validate(
 194         delimiters => $config->webpac('delimiters'),
 195 ) if ($validate_path || $validate_delimiters_path);
 196
 197 my $use_indexer = $config->use_indexer;
 198 $stats ||= $validate;
 199 if ($stats) {
 200         $log->debug("disabled indexing for stats collection");
 201         $use_indexer = undef;
 202 } elsif ( $use_indexer ) {
 203         $log->info("using $use_indexer indexing engine...");
 204 }
 205
 206 # parse normalize files and create source files for lookup and normalization
 207
 208 my ($only_database,$only_input) = split(m#/#, $only_filter) if $only_filter;
 209
 210 my $parser = new WebPAC::Parser(
 211         config => $config,
 212         only_database => $only_database,
 213         only_input => $only_input,
 214 );
 215
 216 my $total_rows = 0;
 217 my $start_t = time();
 218
 219 my @links;
 220
 221 if ($parallel) {
 222         eval 'use Proc::Queue size => 1;';
 223         die $@ if $@;
 224         $log->info("Using $parallel processes for speedup");
 225         Proc::Queue::size($parallel);
 226 }
 227
 228 sub create_ds_config {
 229         my ($db_config, $database, $input, $mfn) = @_;
 230         my $c = dclone( $db_config );
 231         $c->{_} = $database || $log->logconfess("need database");
 232         $c->{_mfn} = $mfn || $log->logconfess("need mfn");
 233         $c->{input} = $input || $log->logconfess("need input");
 234         return $c;
 235 }
 236
 237 foreach my $database ( sort keys %{ $config->databases } ) {
 238         my $db_config = $config->databases->{$database};
 239
 240         next if ($only_database && $database !~ m/$only_database/i);
 241
 242         if ($parallel) {
 243                 my $f=fork;
 244                 if(defined ($f) and $f==0) {
 245                         $log->info("Created processes $$ for speedup");
 246                 } else {
 247                         next;
 248                 }
 249         }
 250
 251         my $indexer;
 252         if ($use_indexer && $parser->have_rules( 'search', $database )) {
 253
 254                 my $cfg_name = $use_indexer;
 255                 $cfg_name =~ s/\-.*$//;
 256
 257                 my $indexer_config = $config->get( $cfg_name ) || $log->logdie("can't find '$cfg_name' part in confguration");
 258                 $indexer_config->{database} = $database;
 259                 $indexer_config->{clean} = $clean;
 260                 $indexer_config->{label} = $db_config->{name};
 261
 262                 # force clean if database has links
 263                 $indexer_config->{clean} = 1 if ($db_config->{links});
 264
 265                 if ($use_indexer eq 'hyperestraier') {
 266
 267                         # open Hyper Estraier database
 268                         require WebPAC::Output::Estraier;
 269                         $indexer = new WebPAC::Output::Estraier( %{ $indexer_config } );
 270
 271                 } elsif ($use_indexer eq 'hyperestraier-native') {
 272
 273                         # open Hyper Estraier database
 274                         require WebPAC::Output::EstraierNative;
 275                         $indexer = new WebPAC::Output::EstraierNative( %{ $indexer_config } );
 276
 277                 } elsif ($use_indexer eq 'kinosearch') {
 278
 279                         die "no longer supported";
 280
 281                 } else {
 282                         $log->logdie("unknown use_indexer: $use_indexer");
 283                 }
 284
 285                 $log->logdie("can't continue without valid indexer") unless ($indexer);
 286         }
 287
 288
 289         #
 290         # store Hyper Estraier links to other databases
 291         #
 292         if (ref($db_config->{links}) eq 'ARRAY' && $use_indexer) {
 293                 foreach my $link (@{ $db_config->{links} }) {
 294                         if ($use_indexer eq 'hyperestraier') {
 295                                 if ($merge) {
 296                                         print $estcmd_fh 'sudo -u www-data estcmd merge ' . $database . ' ' . $link->{to},$/;
 297                                 } else {
 298                                         $log->info("saving link $database -> $link->{to} [$link->{credit}]");
 299                                         push @links, sub {
 300                                                 $log->info("adding link $database -> $link->{to} [$link->{credit}]");
 301                                                 $indexer->add_link(
 302                                                         from => $database,
 303                                                         to => $link->{to},
 304                                                         credit => $link->{credit},
 305                                                 );
 306                                         };
 307                                 }
 308                         } else {
 309                                 $log->warn("NOT IMPLEMENTED WITH $use_indexer: adding link $database -> $link->{to} [$link->{credit}]");
 310                         }
 311                 }
 312         }
 313         next if ($only_links);
 314
 315
 316         #
 317         # now WebPAC::Store
 318         #
 319         my $store = new WebPAC::Store({
 320                 debug => $debug,
 321         });
 322
 323
 324         #
 325         # prepare output
 326         #
 327         my @outputs = force_array( $db_config->{output}, sub {
 328                 $log->error("Database $database doesn't have any outputs defined. Do you want to remove it from configuration?" );
 329         } );
 330
 331         my @output_modules;
 332
 333         foreach my $output ( @outputs ) {
 334
 335 #warn '## output = ',dump( $output );
 336
 337                 my $module = $output->{module} || $log->logdie("need module in output section of $database");
 338                 $module = 'WebPAC::Output::' . $module unless $module =~ m/::/;
 339
 340                 $log->debug("loading output module $module");
 341                 eval "require $module";
 342
 343                 # add database to arugemnts for output filter
 344                 $output->{database} = $database;
 345                 $output->{clean} = $clean;
 346
 347                 $log->debug("calling $module->new(",dump( $output ),")");
 348                 my $out = new $module->new( $output );
 349                 if ( $out->init ) {
 350                         push @output_modules, $out;
 351                 } else {
 352                         $log->warn("SKIPPED $module");
 353                 }
 354         }
 355
 356
 357         #
 358         # now, iterate through input formats
 359         #
 360
 361
 362         my @inputs = force_array( $db_config->{input}, sub {
 363                 $log->info("database $database doesn't have inputs defined");
 364         } );
 365
 366         if ( -e 'out/debug' ) { # FIXME flag?
 367                 my $out;
 368                 foreach my $i ( @inputs ) {
 369                         warn dump( $i );
 370                         next unless defined $i->{normalize};
 371                         warn dump( $i->{normalize} );
 372                         foreach my $normalize ( @{ $i->{normalize} } ) {
 373                                 my $path = $normalize->{path};
 374                                 $out .= qq/\n##\n## $path\n##\n\n/;
 375                                 $out .= read_file( $path );
 376                         }
 377                 }
 378                 my $all = "out/debug/all-normalize.pl";
 379                 write_file( $all, $out );
 380                 warn "### all normalize for this input saved to: $all";
 381         };
 382
 383         foreach my $input (@inputs) {
 384
 385                 my $input_name = $input->{name} || $log->logdie("input without a name isn't valid: ",dump($input));
 386
 387                 if ( $input->{skip} ) {
 388                         $log->info("skip $input_name");
 389                         next;
 390                 }
 391
 392                 next if defined $only_input && $input_name !~ m#$only_input#i;
 393
 394                 my $type = lc($input->{type});
 395
 396                 # FIXME check if input module exists
 397                 my $input_module = $input->{module};
 398
 399                 if ( ! $input_module ) {
 400                         if ( grep(/$type/, $config->webpac('inputs')) ) {
 401                                 $input_module = $config->webpac('inputs')->{$type};
 402                         } else {
 403                                 $log->logdie("I know only how to handle input types ", join(",", $config->webpac('inputs') ), " not '$type'!" );
 404                         }
 405                 }
 406
 407                 my @lookups = $parser->have_lookup_create($database, $input);
 408
 409                 $log->info("working on $database/$input_name with $input_module on $input->{path}",
 410                         @lookups ? " creating lookups: ".join(", ", @lookups) : ""
 411                 );
 412
 413                 if ($stats) {
 414                         # disable modification of records if --stats is in use
 415                         delete($input->{modify_records});
 416                         delete($input->{modify_file});
 417                 }
 418
 419                 if ( $mirror ) {
 420                         my $path = $input->{path} || die "no input path in ",dump( $input );
 421
 422                         my $base = $path;
 423                         $base =~ s{/[^/]+$}{};
 424                         mkpath $base unless -e $base;
 425
 426                         my $rc = LWP::Simple::mirror( "$mirror/$path", $path );
 427                         if (LWP::Simple::is_error( $rc )) {
 428                                 die "can't mirror $mirror/$path -> $path [$rc]";
 429                         } else {
 430                                 $log->info( "mirror ", $path, " [$rc] ", -s $path, " bytes" );
 431                         }
 432
 433                 }
 434
 435                 my $input_db = new WebPAC::Input(
 436                         module => $input_module,
 437                         limit => $limit || $input->{limit},
 438                         offset => $offset || $input->{offset},
 439                         recode => $input->{recode},
 440                         stats => $stats,
 441                         modify_records => $input->{modify_records},
 442                         modify_file => $input->{modify_file},
 443                         input_config => $input,
 444                 );
 445                 $log->logdie("can't create input using $input_module") unless ($input);
 446
 447                 if (defined( $input->{lookup} )) {
 448                         $log->warn("$database/$input_name has depriciated lookup definition, removing it...");
 449                         delete( $input->{lookup} );
 450                 }
 451
 452                 my $lookup_coderef;
 453
 454                 if (@lookups) {
 455
 456                         my $rules = $parser->lookup_create_rules($database, $input) || $log->logdie("no rules found for $database/$input");
 457
 458                         $lookup_coderef = sub {
 459                                 my $rec = shift || die "need rec!";
 460                                 my $mfn = $rec->{'000'}->[0] || die "need mfn in 000";
 461
 462                                 WebPAC::Normalize::data_structure(
 463                                         row => $rec,
 464                                         rules => $rules,
 465                                         config => create_ds_config( $db_config, $database, $input, $mfn ),
 466                                 );
 467
 468                                 #warn "current lookup: ", dump(WebPAC::Normalize::_get_lookup());
 469                         };
 470
 471                         WebPAC::Normalize::_set_lookup( undef );
 472
 473                         $log->debug("created lookup_coderef using:\n$rules");
 474
 475                 };
 476
 477                 my $lookup_jar;
 478
 479                 my $maxmfn = $input_db->open(
 480                         path => $input->{path},
 481                         input_encoding => $input->{encoding},   # database encoding
 482                         lookup_coderef => $lookup_coderef,
 483                         lookup => $lookup_jar,
 484                         %{ $input },
 485                         load_row => sub {
 486                                 my $a = shift;
 487                                 return $store->load_row(
 488                                         database => $database,
 489                                         input => $input_name,
 490                                         id => $a->{id},
 491                                 );
 492                         },
 493                         save_row => sub {
 494                                 my $a = shift;
 495                                 return $store->save_row(
 496                                         database => $database,
 497                                         input => $input_name,
 498                                         id => $a->{id},
 499                                         row => $a->{row},
 500                                 );
 501                         },
 502
 503                 );
 504
 505                 my $lookup_data = WebPAC::Normalize::_get_lookup();
 506
 507                 if (defined( $lookup_data->{$database}->{$input_name} )) {
 508                         $log->debug("created following lookups: ", sub { dump( $lookup_data ) } );
 509
 510                         foreach my $key (keys %{ $lookup_data->{$database}->{$input_name} }) {
 511                                 $store->save_lookup(
 512                                         database => $database,
 513                                         input => $input_name,
 514                                         key => $key,
 515                                         data => $lookup_data->{$database}->{$input_name}->{$key},
 516                                 );
 517                         }
 518                 }
 519
 520                 my $report_fh;
 521                 if ($stats || $validate) {
 522                         my $out_report = 'out/report'; # FIXME move to config
 523                         mkpath $out_report unless -e $out_report;
 524                         my $path = "$out_report/${database}-${input_name}.txt";
 525                         open($report_fh, '>', $path) || $log->logdie("can't open $path: $!");
 526
 527                         print $report_fh "Report for database '$database' input '$input_name' records ",
 528                                 $offset || 1, "-", $limit || $input->{limit} || $maxmfn, "\n\n";
 529                         $log->info("Generating report file $path");
 530
 531                         if ( $validate ) {
 532                                 $validate->read_validate_file( $validate->fill_in( $validate_path, database => $database, input => $input_name ) ) if ( $validate_path );
 533                                 $validate->read_validate_delimiters_file( $validate->fill_in( $validate_delimiters_path, database => $database, input => $input_name ) ) if ( $validate_delimiters_path );
 534                         }
 535                 }
 536
 537                 my $marc;
 538                 if ($marc_generate && $parser->have_rules( 'marc', $database, $input_name )) {
 539
 540                         my $out_marc = 'out/marc'; # FIXME move to config
 541                         mkpath $out_marc unless -e $out_marc;
 542
 543                         $marc = new WebPAC::Output::MARC(
 544                                 path => "$out_marc/${database}-${input_name}",
 545                                 lint => $marc_lint,
 546                                 dump => $marc_dump,
 547                                 marcxml => $marc_xml,
 548                         );
 549                 }
 550
 551                 my $rules = $parser->normalize_rules($database,$input_name);
 552                 if ( ! $rules ) {
 553                         $log->logwarn("no normalize rules for $database/$input_name", $input_db->input_module->can('normalize') ? " using normalize from input module" : '');
 554                         next;
 555                 }
 556
 557                 $log->debug("parsed normalize rules:\n$rules");
 558
 559                 # reset position in database
 560                 $input_db->seek(1);
 561
 562                 # generate name of config key for indexer (strip everything after -)
 563                 my $indexer_config = $use_indexer;
 564                 $indexer_config =~ s/^(\w+)-?.*$/$1/g if ($indexer_config);
 565
 566                 my $lookup_hash;
 567                 my $depends = $parser->depends($database,$input_name);
 568
 569                 if ($depends) {
 570                         $log->debug("$database/$input_name depends on: ", dump($depends)) if ($depends);
 571                         $log->logdie("parser->depends didn't return HASH") unless (ref($depends) eq 'HASH');
 572
 573                         foreach my $db (keys %$depends) {
 574                                 foreach my $i (keys %{$depends->{$db}}) {
 575                                         foreach my $k (keys %{$depends->{$db}->{$i}}) {
 576                                                 my $t = time();
 577                                                 $log->debug("loading lookup $db/$i");
 578                                                 $lookup_hash->{$db}->{$i}->{$k} = $store->load_lookup(
 579                                                         database => $db,
 580                                                         input => $i,
 581                                                         key => $k,
 582                                                 );
 583                                                 $log->debug(sprintf("lookup $db/$i took %.2fs", time() - $t));
 584                                         }
 585                                 }
 586                         }
 587
 588                         $log->debug("lookup_hash = ", sub { dump( $lookup_hash ) });
 589                 }
 590
 591
 592                 # setup input name for all output filters
 593                 foreach my $out ( @output_modules ) {
 594                         if ( $out->can('input') ) {
 595                                 $out->input( $input_name );
 596                         } else {
 597                                 $log->warn("output filter ",ref($out)," doesn't support input name");
 598                         }
 599                 }
 600
 601
 602                 foreach my $pos ( 0 ... $input_db->size ) {
 603
 604                         my $row = $input_db->fetch || next;
 605
 606                         $total_rows++;
 607
 608                         my $mfn = $row->{'000'}->[0];
 609
 610                         if (! $mfn || $mfn !~ m{^\d+$}) {
 611                                 $log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos");
 612                                 $mfn = $pos;
 613                                 push @{ $row->{'000'} }, $pos;
 614                         }
 615
 616                         foreach my $out ( @output_modules ) {
 617                                 $out->add_row( $mfn, $row ) if $out->can('add_row');
 618                         }
 619
 620                         if ($validate) {
 621                                 if ( my $errors = $validate->validate_rec( $row, $input_db->dump_ascii ) ) {
 622                                         $log->error( "MFN $mfn validation error:\n",
 623                                                 $validate->report_error( $errors )
 624                                         );
 625                                 }
 626                                 next;   # validation doesn't create any output
 627                         }
 628
 629                         my $ds;
 630
 631                         if ($rules) {
 632
 633                                 $ds = WebPAC::Normalize::data_structure(
 634                                         row => $row,
 635                                         rules => $rules,
 636                                         lookup => $lookup_hash,
 637                                         config => create_ds_config( $db_config, $database, $input, $mfn ),
 638                                         marc_encoding => 'utf-8',
 639                                         load_row_coderef => sub {
 640                                                 my ($database,$input,$mfn) = @_;
 641 #warn "### load_row($database,$input,$mfn) from data_structure\n";
 642                                                 return $store->load_row(
 643                                                         database => $database,
 644                                                         input => $input,
 645                                                         id => $mfn,
 646                                                 );
 647                                         },
 648                                 );
 649
 650                         } elsif ( $input_db->input_module->can('normalize') ) {
 651                                 $ds = $input_db->input_module->normalize( $mfn );
 652                         }
 653
 654                         if ( $ds ) {
 655                                 $log->debug("ds = ", sub { dump($ds) });
 656
 657                                 $store->save_ds(
 658                                         database => $database,
 659                                         input => $input_name,
 660                                         id => $mfn,
 661                                         ds => $ds,
 662                                 ) if !$stats;
 663
 664                                 $indexer->add(
 665                                         id => "${input_name}/${mfn}",
 666                                         ds => $ds,
 667                                         type => $config->get($indexer_config)->{type},
 668                                 ) if $indexer;
 669
 670                                 foreach my $out ( @output_modules ) {
 671                                         $out->add( $mfn, $ds ) if $out->can('add');
 672                                 }
 673
 674                         } else {
 675                                 $log->warn("record $pos didn't produce any output after normalization rules!") unless $marc;
 676                         }
 677                         if ($marc) {
 678                                 my $i = 0;
 679
 680                                 while (my $fields = WebPAC::Normalize::MARC::_get_marc_fields( fetch_next => 1 ) ) {
 681                                         $marc->add(
 682                                                 id => $mfn . ( $i ? "/$i" : '' ),
 683                                                 fields => $fields,
 684                                                 leader => WebPAC::Normalize::MARC::_get_marc_leader(),
 685                                                 row => $row,
 686                                         );
 687                                         $i++;
 688                                 }
 689
 690                                 $log->info("Created $i instances of MFN $mfn\n") if ($i > 1);
 691                         }
 692
 693                 }
 694
 695                 if ($validate) {
 696                         my $errors = $validate->report;
 697                         if ($errors) {
 698                                 $log->info("validation errors:\n$errors\n" );
 699                                 print $report_fh "$errors\n" if ($report_fh);
 700                         }
 701
 702                         print $report_fh "\nAll possible subfields/delimiter templates:\n", $validate->delimiters_templates( report => 1, current_input => 1 ), "\n\n";
 703
 704                         # must be last thing that touches $validate for this input
 705                         $validate->reset;
 706                 }
 707
 708                 if ($stats) {
 709                         my $s = $input_db->stats;
 710                         $log->info("statistics of fields usage:\n$s");
 711                         print $report_fh "Statistics of fields usage:\n$s" if ($report_fh);
 712                 }
 713
 714                 # close MARC file
 715                 $marc->finish if ($marc);
 716
 717                 # close report
 718                 close($report_fh) if ($report_fh);
 719         }
 720
 721         $indexer->finish if $indexer && $indexer->can('finish');
 722
 723         foreach my $out ( @output_modules ) {
 724                 $out->finish if $out->can('finish');
 725         }
 726
 727         my $dt = time() - $start_t;
 728         $log->info("$total_rows records ", $indexer ? "indexed " : "",
 729                 sprintf("in %.2f sec [%.2f rec/sec]",
 730                         $dt, ($total_rows / $dt)
 731                 )
 732         );
 733
 734
 735         # end forked process
 736         if ($parallel) {
 737                 $log->info("parallel process $$ finished");
 738                 exit(0);
 739         }
 740
 741 }
 742
 743 if ($parallel) {
 744         # wait all children to finish
 745         sleep(1) while wait != -1;
 746         $log->info("all parallel processes finished");
 747 }
 748
 749 # save new delimiters if needed
 750 $validate->save_delimiters_templates if ( $validate_delimiters_path );
 751
 752 #
 753 # handle links or merge after indexing
 754 #
 755
 756 if ($merge) {
 757         print $estcmd_fh 'sudo /etc/init.d/hyperestraier start',$/;
 758         close($estcmd_fh);
 759         chmod 0700, $estcmd_path || $log->warn("can't chmod 0700 $estcmd_path: $!");
 760         system $estcmd_path;
 761 } else {
 762         foreach my $link (@links) {
 763                 $log->logdie("coderef in link ", Dumper($link), " is ", ref($link), " and not CODE") unless (ref($link) eq 'CODE');
 764                 $link->();
 765         }
 766 }