#!/usr/bin/perl

# search-great-ideas.pl - query the great ideas against the great books

# Eric Lease Morgan <eric_morgan@infomotions.com>
# August 8, 2010 - initial investigations; based on search.pl


# define
use constant SOLR   => 'http://localhost:8983/solr/great-books';
use constant IDEAS => '/var/www/html/main/sandbox/great-books/etc/ideas.txt';

# require
use strict;
use WebService::Solr;
use Lingua::Stem::Snowball;

# initalize
my $solr    = WebService::Solr->new( SOLR );
my @ideas   = &slurp_words( IDEAS );
my $stemmer = Lingua::Stem::Snowball->new( lang => 'en' );

# process each idea
foreach my $idea ( sort @ideas ) {

	# search for the idea and its stem
	my $response_idea = $solr->search( $idea );
	my $stem = $stemmer->stem( $idea );	
	my $response_stem = $solr->search( $stem );

	# get the number of hits and display
	my $hit_idea = $response_idea->pager->total_entries;
	my $hit_stem = $response_stem->pager->total_entries;
	print "Your search for $idea and $stem found $hit_idea and $hit_stem document(s), respectively.\n";

}

# done
print "\n";
exit;


sub slurp_words {

	my $file = shift;
	
	my @words = ();
	
	open ( S, " < $file" ) or die "Can't open $file ($!)\n";
	while ( <S> ) {
	
		chop;
		next if ( ! $_ );  # blank line
		next if ( /^#/ );  # comments
		push @words, $_;
		
	}
	
	close S;
	return @words;

}