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;

        my $sort_map = {
                'date' => 'date_utime',
                'from' => 'from_phrase',
                'to' => 'to_phrase',
                'rank' => 'swishrank',
        };

        my $sort_by;

        foreach my $s (@_) {

                if ($s =~ /^\s*(\w+):(.+)\s*$/) {
                        my ($f,$v) = ($1,$2);
                        if (lc($f) eq "sort") {
                                my ($sf,$sv) = split(/ /,$v,2);
                                $sort_by = $sort_map->{$sf} || croak "unsupported sort by field $v - fix sort_map";
                                $sort_by .= " $sv";
                        } else {
                                $sw .= "$f=($v)";
                        }
                } else {
                        # and/or/not operators
                        $sw .= " $s ";
                }
        }

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

        # convert to UTF-8
        $sw = $iso2utf->convert($sw) || $sw;

        my $search = $index->New_Search_Object;
        $search->SetSort( $sort_by );
        my $results = $search->Execute($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 date_utime)) {
                        $self->{cache}->{$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	dpavlin	24	my $sort_map = {
46			'date' => 'date_utime',
47			'from' => 'from_phrase',
48			'to' => 'to_phrase',
49			'rank' => 'swishrank',
50			};
51
52			my $sort_by;
53
54	dpavlin	21	foreach my $s (@_) {
55
56			if ($s =~ /^\s(\w+):(.+)\s$/) {
57	dpavlin	24	my ($f,$v) = ($1,$2);
58			if (lc($f) eq "sort") {
59			my ($sf,$sv) = split(/ /,$v,2);
60			$sort_by = $sort_map->{$sf} \|\| croak "unsupported sort by field $v - fix sort_map";
61			$sort_by .= " $sv";
62			} else {
63			$sw .= "$f=($v)";
64			}
65	dpavlin	21	} else {
66			# and/or/not operators
67			$sw .= " $s ";
68			}
69	dpavlin	12	}
70
71	dpavlin	21	print STDERR "swish search: $sw\n";
72	dpavlin	18
73			# convert to UTF-8
74	dpavlin	21	$sw = $iso2utf->convert($sw) \|\| $sw;
75	dpavlin	12
76	dpavlin	24	my $search = $index->New_Search_Object;
77			$search->SetSort( $sort_by );
78			my $results = $search->Execute($sw);
79
80	dpavlin	17	# store total number of hits
81			$self->{'total_hits'} = $results->Hits;
82
83	dpavlin	14	my @res_ids;
84	dpavlin	12
85	dpavlin	18	my $count = 1;
86	dpavlin	17
87	dpavlin	14	while ( my $r = $results->NextResult ) {
88	dpavlin	13
89	dpavlin	14	sub p($$) {
90			my ($r,$prop) = @_;
91			$prop = $r->Property($prop);
92			$prop =~ s/##lf##/\n/gs;
93			return $utf2iso->convert($prop);
94			}
95	dpavlin	13
96	dpavlin	14	my $id = p($r,"swishdocpath");
97			push @res_ids, $id;
98
99	dpavlin	13	foreach my $p (qw(from to cc bcc)) {
100	dpavlin	19	@{$self->{cache}->{$id}->{$p}} = ();
101			foreach my $v (split(/##/, p($r,$p.'_phrase'))) {
102			push @{$self->{cache}->{$id}->{$p}}, $v;
103			$self->add_counter($p,$v);
104			}
105	dpavlin	13	}
106
107	dpavlin	23	foreach my $p (qw(subject body date date_utime)) {
108	dpavlin	14	$self->{cache}->{$id}->{$p} = p($r,$p);
109	dpavlin	13	}
110
111	dpavlin	20	$self->add_counter_calendar(p($r,'date_utime'));
112
113	dpavlin	13	# this is redundant, but needed for templates later...
114			$self->{cache}->{$id}->{'id'} = $id;
115	dpavlin	17
116	dpavlin	18	last if (++$count > $self->{max_results});
117	dpavlin	12	}
118
119	dpavlin	14	return @res_ids;
120	dpavlin	12	}
121
122			sub add_index {
123			my $self = shift;
124	dpavlin	14
125			my $mbox_id = shift \|\| croak "add_index needs mbox_id";
126			my $document = shift \|\| croak "add_index needs document";
127
128			my ($mbox,$id) = split(/\s/,$mbox_id,2);
129
130			my $xml = qq{<message>};
131			foreach my $tag (keys %$document) {
132			my $data = $document->{$tag};
133	dpavlin	16	next if (! $data \|\| $data eq '');
134	dpavlin	14	# save [cr/]lf before conversion to XML
135			$data =~ s/\n\r/##lf##/gs;
136			$data =~ s/\n/##lf##/gs;
137	dpavlin	16	$xml .= "<$tag><![CDATA[".$data."]]></$tag>\n";
138	dpavlin	14	}
139			$xml .= qq{</message>};
140
141			$xml = $iso2utf->convert($xml);
142			use bytes; # as opposed to chars
143			print "Path-Name: $mbox $id\n";
144			print "Content-Length: ".(length($xml)+1)."\n";
145			print "Document-Type: XML\n\n$xml\n";
146
147	dpavlin	12	}
148
149			sub close_index {
150			my $self = shift;
151
152			}
153
154			1;