mws/trunk/MWS_swish.pm

#!/usr/bin/perl -w

use strict;

#
# simple implementation to use SWISH-e with queryies like
# Lucene (subject:something)
#

use SWISH::API;
use Text::Iconv;
use Data::Dumper;

my $iso2utf = Text::Iconv->new('ISO-8859-2','UTF-8');
my $utf2iso = Text::Iconv->new('UTF-8','ISO-8859-2');

sub open_index {
        my $self = shift;

        my $swish = $self->{index};

        if (! $swish) {

                my $index_file = $self->{index_file} || croak "open_index needs index filename";
                $index_file .= "/swish-e";
                print STDERR "opening index '$index_file'\n";
                $swish = SWISH::API->new($index_file);
                $swish->AbortLastError if $swish->Error;

                $self->{index} = $swish;
        }

        return $swish;
}

sub search_index {
        my $self = shift;

        my $s = shift || croak "search_index needs query";

        my $index = $self->open_index;

        if ($s =~ /:/) {
                my ($fld,$val) = split(/:/,$s,2);
                $s = "$fld=($val)";
        }

        print STDERR "swish search: $s\n";

        # convert to UTF-8
        $s = $iso2utf->convert($s) || $s;
        my $results = $index->Query($s);

        # store total number of hits
        $self->{'total_hits'} = $results->Hits;

        my @res_ids;

        my $count = 1;

        while ( my $r = $results->NextResult ) {

                sub p($$) {
                        my ($r,$prop) = @_;
                        $prop = $r->Property($prop);
                        $prop =~ s/##lf##/\n/gs;
                        return $utf2iso->convert($prop);
                }

                my $id = p($r,"swishdocpath");
                push @res_ids, $id;

                foreach my $p (qw(from to cc bcc)) {
                        @{$self->{cache}->{$id}->{$p}} = split(/##/, p($r,$p.'_phrase'));
                }

                foreach my $p (qw(subject body date)) {
                        $self->{cache}->{$id}->{$p} = p($r,$p);
                }

                # this is redundant, but needed for templates later...
                $self->{cache}->{$id}->{'id'} = $id;

                last if (++$count > $self->{max_results});
        }

        return @res_ids;
}

sub add_index {
        my $self = shift;

        my $mbox_id = shift || croak "add_index needs mbox_id";
        my $document = shift || croak "add_index needs document";

        my ($mbox,$id) = split(/\s/,$mbox_id,2);

        my $xml = qq{<message>};
        foreach my $tag (keys %$document) {
                my $data = $document->{$tag};
                next if (! $data || $data eq '');
                # save [cr/]lf before conversion to XML
                $data =~ s/\n\r/##lf##/gs;
                $data =~ s/\n/##lf##/gs;
                $xml .= "<$tag><![CDATA[".$data."]]></$tag>\n";
        }
        $xml .= qq{</message>};

        $xml = $iso2utf->convert($xml);
        use bytes;      # as opposed to chars
        print "Path-Name: $mbox $id\n";
        print "Content-Length: ".(length($xml)+1)."\n";
        print "Document-Type: XML\n\n$xml\n";

}

sub close_index {
        my $self = shift;

}

1;
1	dpavlin	12	#!/usr/bin/perl -w
2
3			use strict;
4
5			#
6			# simple implementation to use SWISH-e with queryies like
7			# Lucene (subject:something)
8			#
9
10			use SWISH::API;
11	dpavlin	14	use Text::Iconv;
12			use Data::Dumper;
13	dpavlin	12
14	dpavlin	14	my $iso2utf = Text::Iconv->new('ISO-8859-2','UTF-8');
15			my $utf2iso = Text::Iconv->new('UTF-8','ISO-8859-2');
16
17	dpavlin	12	sub open_index {
18			my $self = shift;
19
20			my $swish = $self->{index};
21
22			if (! $swish) {
23
24			my $index_file = $self->{index_file} \|\| croak "open_index needs index filename";
25			$index_file .= "/swish-e";
26			print STDERR "opening index '$index_file'\n";
27			$swish = SWISH::API->new($index_file);
28			$swish->AbortLastError if $swish->Error;
29
30			$self->{index} = $swish;
31			}
32
33			return $swish;
34			}
35
36			sub search_index {
37			my $self = shift;
38
39			my $s = shift \|\| croak "search_index needs query";
40
41			my $index = $self->open_index;
42
43			if ($s =~ /:/) {
44			my ($fld,$val) = split(/:/,$s,2);
45			$s = "$fld=($val)";
46			}
47
48			print STDERR "swish search: $s\n";
49	dpavlin	18
50			# convert to UTF-8
51			$s = $iso2utf->convert($s) \|\| $s;
52	dpavlin	12	my $results = $index->Query($s);
53
54	dpavlin	17	# store total number of hits
55			$self->{'total_hits'} = $results->Hits;
56
57	dpavlin	14	my @res_ids;
58	dpavlin	12
59	dpavlin	18	my $count = 1;
60	dpavlin	17
61	dpavlin	14	while ( my $r = $results->NextResult ) {
62	dpavlin	13
63	dpavlin	14	sub p($$) {
64			my ($r,$prop) = @_;
65			$prop = $r->Property($prop);
66			$prop =~ s/##lf##/\n/gs;
67			return $utf2iso->convert($prop);
68			}
69	dpavlin	13
70	dpavlin	14	my $id = p($r,"swishdocpath");
71			push @res_ids, $id;
72
73	dpavlin	13	foreach my $p (qw(from to cc bcc)) {
74	dpavlin	14	@{$self->{cache}->{$id}->{$p}} = split(/##/, p($r,$p.'_phrase'));
75	dpavlin	13	}
76
77			foreach my $p (qw(subject body date)) {
78	dpavlin	14	$self->{cache}->{$id}->{$p} = p($r,$p);
79	dpavlin	13	}
80
81			# this is redundant, but needed for templates later...
82			$self->{cache}->{$id}->{'id'} = $id;
83	dpavlin	17
84	dpavlin	18	last if (++$count > $self->{max_results});
85	dpavlin	12	}
86
87	dpavlin	14	return @res_ids;
88	dpavlin	12	}
89
90			sub add_index {
91			my $self = shift;
92	dpavlin	14
93			my $mbox_id = shift \|\| croak "add_index needs mbox_id";
94			my $document = shift \|\| croak "add_index needs document";
95
96			my ($mbox,$id) = split(/\s/,$mbox_id,2);
97
98			my $xml = qq{<message>};
99			foreach my $tag (keys %$document) {
100			my $data = $document->{$tag};
101	dpavlin	16	next if (! $data \|\| $data eq '');
102	dpavlin	14	# save [cr/]lf before conversion to XML
103			$data =~ s/\n\r/##lf##/gs;
104			$data =~ s/\n/##lf##/gs;
105	dpavlin	16	$xml .= "<$tag><![CDATA[".$data."]]></$tag>\n";
106	dpavlin	14	}
107			$xml .= qq{</message>};
108
109			$xml = $iso2utf->convert($xml);
110			use bytes; # as opposed to chars
111			print "Path-Name: $mbox $id\n";
112			print "Content-Length: ".(length($xml)+1)."\n";
113			print "Document-Type: XML\n\n$xml\n";
114
115	dpavlin	12	}
116
117			sub close_index {
118			my $self = shift;
119
120			}
121
122			1;