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 Data::Dumper;
   8 use lib './lib';
   9
  10 use WebPAC::Common 0.02;
  11 use WebPAC::Lookup;
  12 use WebPAC::Input::ISIS;
  13 use WebPAC::Store 0.03;
  14 use WebPAC::Normalize::XML;
  15 use WebPAC::Output::TT;
  16 use WebPAC::Output::Estraier 0.05;
  17 use YAML qw/LoadFile/;
  18 use LWP::Simple;
  19
  20 my $limit = shift @ARGV;
  21
  22 my $config = LoadFile('conf/config.yml');
  23
  24 print "config = ",Dumper($config);
  25
  26 die "no databases in config file!\n" unless ($config->{databases});
  27
  28 my $total_rows = 0;
  29
  30 while (my ($database, $db_config) = each %{ $config->{databases} }) {
  31
  32         my $log = _new WebPAC::Common()->_get_logger();
  33
  34         #
  35         # open Hyper Estraier database
  36         #
  37
  38         my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration");
  39         $est_config->{database} = $database;
  40
  41         $log->info("using HyperEstraier URL $est_config->{masterurl}");
  42
  43         my $est = new WebPAC::Output::Estraier(
  44                 %{ $est_config },
  45         );
  46
  47         #
  48         # now WebPAC::Store
  49         #
  50         my $abs_path = abs_path($0);
  51         $abs_path =~ s#/[^/]*$#/#;
  52
  53         my $db_path = $config->{webpac}->{db_path} . '/' . $database;
  54
  55         $log->info("working on $database in $db_path");
  56
  57         my $db = new WebPAC::Store(
  58                 path => $db_path,
  59                 database => $database,
  60                 debug => 1,
  61         );
  62
  63
  64         #
  65         # now, iterate through input formats
  66         #
  67
  68         my @inputs;
  69         if (ref($db_config->{input}) eq 'ARRAY') {
  70                 @inputs = @{ $db_config->{input} };
  71         } elsif ($db_config->{input}) {
  72                 push @inputs, $db_config->{input};
  73         } else {
  74                 $log->info("database $database doesn't have inputs defined");
  75         }
  76
  77         foreach my $input (@inputs) {
  78
  79                 my $type = lc($input->{type});
  80
  81                 die "I know only how to handle input type isis, not '$type'!\n" unless ($type eq 'isis');
  82
  83                 my $lookup = new WebPAC::Lookup(
  84                         lookup_file => $input->{lookup},
  85                 );
  86
  87                 $log->info("working on input $input->{path} [$input->{type}]");
  88
  89                 my $isis = new WebPAC::Input::ISIS(
  90                         code_page => $config->{webpac}->{webpac_encoding},
  91                         limit_mfn => $input->{limit},
  92                         lookup => $lookup,
  93                 );
  94
  95                 my $maxmfn = $isis->open(
  96                         filename => $input->{path},
  97                         code_page => $input->{encoding},        # database encoding
  98                 );
  99
 100                 my $n = new WebPAC::Normalize::XML(
 101                 #       filter => { 'foo' => sub { shift } },
 102                         db => $db,
 103                         lookup_regex => $lookup->regex,
 104                         lookup => $lookup,
 105                         prefix => $input->{name},
 106                 );
 107
 108                 my $normalize_path = $input->{normalize}->{path};
 109
 110                 if ($normalize_path =~ m/\.xml$/i) {
 111                         $n->open(
 112                                 tag => $input->{normalize}->{tag},
 113                                 xml_file => $input->{normalize}->{path},
 114                         );
 115                 } elsif ($normalize_path =~ m/\.(?:yml|yaml)$/i) {
 116                         $n->open_yaml(
 117                                 path => $normalize_path,
 118                                 tag => $input->{normalize}->{tag},
 119                         );
 120                 }
 121
 122                 for ( 0 ... $isis->size ) {
 123
 124                         my $row = $isis->fetch || next;
 125
 126                         my $mfn = $row->{'000'}->[0] || die "can't find MFN";
 127
 128                         my $ds = $n->data_structure($row);
 129
 130                         $est->add(
 131                                 id => $input->{name} . "#" . $mfn,
 132                                 ds => $ds,
 133                                 type => $config->{hyperestraier}->{type},
 134                         );
 135
 136                         $total_rows++;
 137                 }
 138
 139         };
 140
 141         $log->info("$total_rows records indexed");
 142
 143         #
 144         # add Hyper Estraier links to other databases
 145         #
 146         if (ref($db_config->{links}) eq 'ARRAY') {
 147                 foreach my $link (@{ $db_config->{links} }) {
 148                         $log->info("adding link $database -> $link->{to} [$link->{credit}]");
 149                         $est->add_link(
 150                                 from => $database,
 151                                 to => $link->{to},
 152                                 credit => $link->{credit},
 153                         );
 154                 }
 155         }
 156
 157 }
 158