Semantic-Engine/EPrints/index.pl

#!/usr/bin/perl -w

use strict;
use Semantic::API;
use Data::Dump qw/dump/;
use EPrints qw/_x/;
use lib '/home/dpavlin/stem-hr/';
use StemHR;

warn dump( StemHR->stem('kuæni') );

my $debug = shift @ARGV;

my $type = 'slogovi';

my $dbh = EPrints->dbh;
my $sth = $dbh->prepare(qq{
SELECT
        archive_title.eprintid as id,
        title
FROM archive_title 
}) || die $dbh->errstr();
$sth->execute() || die $sth->errstr();

my $indexer = Semantic::API::Index->new(
        storage => 'sqlite',
        database => 'eprints.db',
        collection => 'EPrints'
);


$indexer->add_word_filters( minimum_length => 3,
                            too_many_numbers => 10,
                            maximum_word_length => 15 );

# use this encoding for any incoming text
#$indexer->set_default_encoding( "utf8"); 

my $total = 0;

while (my $row = $sth->fetchrow_hashref ) {
        EPrints->id( $row->{id} );
        my ( $title, $keywords, $abstract ) = (
                _x( $row->{title} ),
                EPrints->lookup( 'keywords' ),
                EPrints->lookup( 'abstract' )
        );
        my @body = split( /\W*\s+\W*/, "$title $title $title $keywords $keywords $abstract" );
        my $body;
        foreach my $word ( @body ) {
#               $body .= StemHR->stem( $word ) . ' ';
                $body .= join(" ",EPrints::slogovi( $word )) . ' ';
        }

        warn "body: $body\n" if $debug;

    $indexer->index( $row->{id}, join(" ", @body, $body ) );
        $total++;
        print STDERR _x( $row->{id}, " ", $row->{title} ), "\n";
}


print STDERR "\nNow adding $total items to the database...";
$indexer->finish(); 
print STDERR "done!\n"; 

1	dpavlin	1	#!/usr/bin/perl -w
2
3			use strict;
4			use Semantic::API;
5			use Data::Dump qw/dump/;
6			use EPrints qw/_x/;
7			use lib '/home/dpavlin/stem-hr/';
8			use StemHR;
9
10			warn dump( StemHR->stem('kuæni') );
11
12			my $debug = shift @ARGV;
13
14	dpavlin	4	my $type = 'slogovi';
15
16	dpavlin	1	my $dbh = EPrints->dbh;
17			my $sth = $dbh->prepare(qq{
18			SELECT
19			archive_title.eprintid as id,
20			title
21			FROM archive_title
22			}) \|\| die $dbh->errstr();
23			$sth->execute() \|\| die $sth->errstr();
24
25			my $indexer = Semantic::API::Index->new(
26			storage => 'sqlite',
27			database => 'eprints.db',
28			collection => 'EPrints'
29			);
30
31
32			$indexer->add_word_filters( minimum_length => 3,
33			too_many_numbers => 10,
34			maximum_word_length => 15 );
35
36			# use this encoding for any incoming text
37	dpavlin	4	#$indexer->set_default_encoding( "utf8");
38	dpavlin	1
39			my $total = 0;
40
41			while (my $row = $sth->fetchrow_hashref ) {
42			EPrints->id( $row->{id} );
43			my ( $title, $keywords, $abstract ) = (
44			_x( $row->{title} ),
45			EPrints->lookup( 'keywords' ),
46			EPrints->lookup( 'abstract' )
47			);
48			my @body = split( /\W\s+\W/, "$title $title $title $keywords $keywords $abstract" );
49			my $body;
50			foreach my $word ( @body ) {
51	dpavlin	4	# $body .= StemHR->stem( $word ) . ' ';
52			$body .= join(" ",EPrints::slogovi( $word )) . ' ';
53	dpavlin	1	}
54
55			warn "body: $body\n" if $debug;
56
57			$indexer->index( $row->{id}, join(" ", @body, $body ) );
58			$total++;
59			print STDERR _x( $row->{id}, " ", $row->{title} ), "\n";
60			}
61
62
63			print STDERR "\nNow adding $total items to the database...";
64			$indexer->finish();
65			print STDERR "done!\n";
66