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";
        my $results = $index->Query($s);

        my @res_ids;

        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;
        }

        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			my $results = $index->Query($s);
50
51	dpavlin	14	my @res_ids;
52	dpavlin	12
53	dpavlin	14	while ( my $r = $results->NextResult ) {
54	dpavlin	13
55	dpavlin	14	sub p($$) {
56			my ($r,$prop) = @_;
57			$prop = $r->Property($prop);
58			$prop =~ s/##lf##/\n/gs;
59			return $utf2iso->convert($prop);
60			}
61	dpavlin	13
62	dpavlin	14	my $id = p($r,"swishdocpath");
63			push @res_ids, $id;
64
65	dpavlin	13	foreach my $p (qw(from to cc bcc)) {
66	dpavlin	14	@{$self->{cache}->{$id}->{$p}} = split(/##/, p($r,$p.'_phrase'));
67	dpavlin	13	}
68
69			foreach my $p (qw(subject body date)) {
70	dpavlin	14	$self->{cache}->{$id}->{$p} = p($r,$p);
71	dpavlin	13	}
72
73			# this is redundant, but needed for templates later...
74			$self->{cache}->{$id}->{'id'} = $id;
75	dpavlin	12	}
76
77	dpavlin	14	return @res_ids;
78	dpavlin	12	}
79
80			sub add_index {
81			my $self = shift;
82	dpavlin	14
83			my $mbox_id = shift \|\| croak "add_index needs mbox_id";
84			my $document = shift \|\| croak "add_index needs document";
85
86			my ($mbox,$id) = split(/\s/,$mbox_id,2);
87
88			my $xml = qq{<message>};
89			foreach my $tag (keys %$document) {
90			my $data = $document->{$tag};
91	dpavlin	16	next if (! $data \|\| $data eq '');
92	dpavlin	14	# save [cr/]lf before conversion to XML
93			$data =~ s/\n\r/##lf##/gs;
94			$data =~ s/\n/##lf##/gs;
95	dpavlin	16	$xml .= "<$tag><![CDATA[".$data."]]></$tag>\n";
96	dpavlin	14	}
97			$xml .= qq{</message>};
98
99			$xml = $iso2utf->convert($xml);
100			use bytes; # as opposed to chars
101			print "Path-Name: $mbox $id\n";
102			print "Content-Length: ".(length($xml)+1)."\n";
103			print "Document-Type: XML\n\n$xml\n";
104
105	dpavlin	12	}
106
107			sub close_index {
108			my $self = shift;
109
110			}
111
112			1;