#!/usr/bin/perl
# all_resources.pl
# Use this script to index the complete list of resources including all metadata
# as well as term association. The output of this file is HTML and the indexer is
# intended to be something like Swish-E.
# 2006-06-29 - escaped the output to remove swish-e errors - ELM
# 2006-01-12 - added pod - ELM
# first draft 8/09/2004 - RJF
=head1 NAME
all_resources.pl - dump all MyLibrary resources in a form indexable by swish-e
=head1 DESCRIPTION
This program loops through each resource in your MyLibrary database, formats each record as an HTML stream with very rich meta-data elements, adds a swish-e specific header to each stream, and sends the whole thing to STDOUT.
This program is intended to be run via another script (./index-resources.pl). The HTML's meta-data elements are expected to be echoed in a swish-e configuration file (./index-resources.cfg). These meta-data elements then become fields searchable in the resulting index.
=head1 SEE ALSO
index-resources.cfg and index-resources.pl
=head1 AUTHOR
Rob Fox
=cut
use MyLibrary::Core;
use strict;
# configure
MyLibrary::Config->instance( 'oss_and_xml' );
# first, get all of the resources
my @resource_ids = MyLibrary::Resource->get_ids();
foreach my $resource_id (@resource_ids) {
my $resource = MyLibrary::Resource->new(id => $resource_id);
my @related_terms = $resource->related_terms();
my %facet_ids = ();
my $output;
foreach my $related_term_id (@related_terms) {
my $term = MyLibrary::Term->new(id => $related_term_id);
my $facet_id = $term->facet_id();
if ($facet_ids{$facet_id}) {
my @active_terms = @{$facet_ids{$facet_id}};
push (@active_terms, $related_term_id);
$facet_ids{$facet_id} = \@active_terms;
} else {
my @active_terms = ();
push (@active_terms, $related_term_id);
$facet_ids{$facet_id} = \@active_terms;
}
}
$output = "\n\n\t" . &escape($resource->name()) . "\n";
$output .= "\tid()) . "\" />\n";
$output .= "\tfull_text()) . "\" />\n";
$output .= "\treference_linking()) . "\" />\n";
$output .= "\tcoverage_info()) . "\" />\n";
$output .= "\tnote()) . "\" />\n";
$output .= "\tlcd()) . "\" />\n";
# output locations
my @resource_locations = $resource->resource_locations();
foreach my $location (@resource_locations) {
$output .= "\tlocation()) . "\" />\n";
}
# output facets and terms
foreach my $facet_id (keys %facet_ids) {
my $facet = MyLibrary::Facet->new(id => $facet_id);
my $facet_name = $facet->facet_name();
$facet_name =~ s/ /_/g;
$facet_name =~ s/&/and/g;
my @term_ids = @{$facet_ids{$facet_id}};
foreach my $term_id (@term_ids) {
my $term = MyLibrary::Term->new(id => $term_id);
$output .= "\tterm_name()) . "\" />\n";
}
}
$output .= "\n\n\n";
$output .= "\t" . &escape($resource->name()) . "
\n";
$output .= "\t" . &escape($resource->note()) . "
\n";
$output .= "
\n\n\n";
# output the record
print "Path-Name: ${resource_id}\n";
print "Content-length: " . scalar(length $output) . "\n";
print "Document-Type: HTML*\n";
print "\n";
print "$output";
}
sub escape {
my $s = shift;
$s =~ s/&/&/g;
$s =~ s//>/g;
$s =~ s/'/'/g;
$s =~ s/"/"/g;
return $s;
}