trunk/scripts/dbi-indexer.pl

#!/usr/bin/perl -w

use strict;
use Search::Estraier 0.06;
use DBI;
use Data::Dumper;
use Encode qw/from_to/;

=head1 NAME

dbi-indexer.pl - example indexer of DBI sources for Search::Estraier

=cut

my $c = {
        node_url => 'http://localhost:1978/node/dbi',
        dbi => 'Pg:dbname=azop',
        sql => qq{
                select * from history_collection_view_cache
        },
        debug => 0,
};

# create and configure node
my $node = new Search::Estraier::Node(
        url => $c->{node_url},
        user => 'admin',
        passwd => 'admin',
        croak_on_error => 1,
        create => 1,
        debug => $c->{debug},
);

# create DBI connection
my $dbh = DBI->connect("DBI:$c->{dbi}","","") || die $DBI::errstr;

my $sth = $dbh->prepare($c->{sql}) || die $dbh->errstr();
$sth->execute() || die $sth->errstr();

warn "# columns: ",join(",",@{ $sth->{NAME} }),"\n" if ($c->{debug});

my $total = $sth->rows;
my $i = 1;

while (my $row = $sth->fetchrow_hashref() ) {

        warn "# row: ",Dumper($row) if ($c->{debug});

        # create document
        my $doc = new Search::Estraier::Document;

        $doc->add_attr('@uri', $row->{_id});

        printf "%4d ",$i;

        while (my ($col,$val) = each %{$row}) {

                if ($val) {
                        # change encoding?
                        from_to($val, 'ISO-8859-2', 'UTF-8');

                        # add attributes (make column usable from attribute search)
                        $doc->add_attr($col, $val);

                        # add body text to document (make it searchable using full-text index)
                        $doc->add_text($val);

                        print "R";
                } else {
                        print ".";
                }

        }

        print " ", int(( $i++ / $total) * 100), "%\n";

        warn "# doc draft: ",$doc->dump_draft, "\n" if ($c->{debug});

        die "error: ", $node->status,"\n" unless (eval { $node->put_doc($doc) });
}
1	dpavlin	131	#!/usr/bin/perl -w
2
3			use strict;
4	dpavlin	144	use Search::Estraier 0.06;
5	dpavlin	131	use DBI;
6			use Data::Dumper;
7			use Encode qw/from_to/;
8
9			=head1 NAME
10
11			dbi-indexer.pl - example indexer of DBI sources for Search::Estraier
12
13			=cut
14
15	dpavlin	144	my $c = {
16			node_url => 'http://localhost:1978/node/dbi',
17			dbi => 'Pg:dbname=azop',
18			sql => qq{
19			select * from history_collection_view_cache
20			},
21			debug => 0,
22	dpavlin	131	};
23
24			# create and configure node
25			my $node = new Search::Estraier::Node(
26	dpavlin	144	url => $c->{node_url},
27	dpavlin	131	user => 'admin',
28			passwd => 'admin',
29			croak_on_error => 1,
30	dpavlin	144	create => 1,
31			debug => $c->{debug},
32	dpavlin	131	);
33
34			# create DBI connection
35	dpavlin	144	my $dbh = DBI->connect("DBI:$c->{dbi}","","") \|\| die $DBI::errstr;
36	dpavlin	131
37	dpavlin	144	my $sth = $dbh->prepare($c->{sql}) \|\| die $dbh->errstr();
38	dpavlin	131	$sth->execute() \|\| die $sth->errstr();
39
40	dpavlin	144	warn "# columns: ",join(",",@{ $sth->{NAME} }),"\n" if ($c->{debug});
41	dpavlin	131
42			my $total = $sth->rows;
43			my $i = 1;
44
45			while (my $row = $sth->fetchrow_hashref() ) {
46
47	dpavlin	144	warn "# row: ",Dumper($row) if ($c->{debug});
48	dpavlin	131
49			# create document
50			my $doc = new Search::Estraier::Document;
51
52			$doc->add_attr('@uri', $row->{_id});
53
54			printf "%4d ",$i;
55
56			while (my ($col,$val) = each %{$row}) {
57
58			if ($val) {
59			# change encoding?
60			from_to($val, 'ISO-8859-2', 'UTF-8');
61
62			# add attributes (make column usable from attribute search)
63			$doc->add_attr($col, $val);
64
65			# add body text to document (make it searchable using full-text index)
66			$doc->add_text($val);
67
68			print "R";
69			} else {
70			print ".";
71			}
72
73			}
74
75			print " ", int(( $i++ / $total) * 100), "%\n";
76
77	dpavlin	144	warn "# doc draft: ",$doc->dump_draft, "\n" if ($c->{debug});
78	dpavlin	131
79			die "error: ", $node->status,"\n" unless (eval { $node->put_doc($doc) });
80			}