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;

        croak "search_index needs query" if (! @_);

        my $index = $self->open_index;

        my $sw;

        foreach my $s (@_) {

                if ($s =~ /^\s*(\w+):(.+)\s*$/) {
                        $sw .= "$1=($2)";
                } else {
                        # and/or/not operators
                        $sw .= " $s ";
                }
        }

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

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

        # 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}} = ();
                        foreach my $v (split(/##/, p($r,$p.'_phrase'))) {
                                push @{$self->{cache}->{$id}->{$p}}, $v;
                                $self->add_counter($p,$v);
                        }
                }

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

                $self->add_counter_calendar(p($r,'date_utime'));

                # 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	dpavlin	21	croak "search_index needs query" if (! @_);
40	dpavlin	12
41			my $index = $self->open_index;
42
43	dpavlin	21	my $sw;
44
45			foreach my $s (@_) {
46
47			if ($s =~ /^\s(\w+):(.+)\s$/) {
48			$sw .= "$1=($2)";
49			} else {
50			# and/or/not operators
51			$sw .= " $s ";
52			}
53	dpavlin	12	}
54
55	dpavlin	21	print STDERR "swish search: $sw\n";
56	dpavlin	18
57			# convert to UTF-8
58	dpavlin	21	$sw = $iso2utf->convert($sw) \|\| $sw;
59			my $results = $index->Query($sw);
60	dpavlin	12
61	dpavlin	17	# store total number of hits
62			$self->{'total_hits'} = $results->Hits;
63
64	dpavlin	14	my @res_ids;
65	dpavlin	12
66	dpavlin	18	my $count = 1;
67	dpavlin	17
68	dpavlin	14	while ( my $r = $results->NextResult ) {
69	dpavlin	13
70	dpavlin	14	sub p($$) {
71			my ($r,$prop) = @_;
72			$prop = $r->Property($prop);
73			$prop =~ s/##lf##/\n/gs;
74			return $utf2iso->convert($prop);
75			}
76	dpavlin	13
77	dpavlin	14	my $id = p($r,"swishdocpath");
78			push @res_ids, $id;
79
80	dpavlin	13	foreach my $p (qw(from to cc bcc)) {
81	dpavlin	19	@{$self->{cache}->{$id}->{$p}} = ();
82			foreach my $v (split(/##/, p($r,$p.'_phrase'))) {
83			push @{$self->{cache}->{$id}->{$p}}, $v;
84			$self->add_counter($p,$v);
85			}
86	dpavlin	13	}
87
88			foreach my $p (qw(subject body date)) {
89	dpavlin	14	$self->{cache}->{$id}->{$p} = p($r,$p);
90	dpavlin	13	}
91
92	dpavlin	20	$self->add_counter_calendar(p($r,'date_utime'));
93
94	dpavlin	13	# this is redundant, but needed for templates later...
95			$self->{cache}->{$id}->{'id'} = $id;
96	dpavlin	17
97	dpavlin	18	last if (++$count > $self->{max_results});
98	dpavlin	12	}
99
100	dpavlin	14	return @res_ids;
101	dpavlin	12	}
102
103			sub add_index {
104			my $self = shift;
105	dpavlin	14
106			my $mbox_id = shift \|\| croak "add_index needs mbox_id";
107			my $document = shift \|\| croak "add_index needs document";
108
109			my ($mbox,$id) = split(/\s/,$mbox_id,2);
110
111			my $xml = qq{<message>};
112			foreach my $tag (keys %$document) {
113			my $data = $document->{$tag};
114	dpavlin	16	next if (! $data \|\| $data eq '');
115	dpavlin	14	# save [cr/]lf before conversion to XML
116			$data =~ s/\n\r/##lf##/gs;
117			$data =~ s/\n/##lf##/gs;
118	dpavlin	16	$xml .= "<$tag><![CDATA[".$data."]]></$tag>\n";
119	dpavlin	14	}
120			$xml .= qq{</message>};
121
122			$xml = $iso2utf->convert($xml);
123			use bytes; # as opposed to chars
124			print "Path-Name: $mbox $id\n";
125			print "Content-Length: ".(length($xml)+1)."\n";
126			print "Document-Type: XML\n\n$xml\n";
127
128	dpavlin	12	}
129
130			sub close_index {
131			my $self = shift;
132
133			}
134
135			1;