| 1 |
dpavlin |
57 |
#!/usr/bin/perl -w |
| 2 |
|
|
|
| 3 |
|
|
package MWS::Clucene; |
| 4 |
|
|
use strict; |
| 5 |
|
|
use warnings; |
| 6 |
|
|
|
| 7 |
|
|
use MWS::Indexer; |
| 8 |
|
|
our @ISA=qw(MWS::Indexer); |
| 9 |
|
|
|
| 10 |
|
|
our $VERSION = '0.01'; |
| 11 |
|
|
|
| 12 |
|
|
BEGIN { |
| 13 |
|
|
# this is so utterly wrong that I won't even start to talk |
| 14 |
|
|
# about it. But, FulltextSearch::Clucene still miss real |
| 15 |
|
|
# Makefile.PL |
| 16 |
|
|
my $dir = $ENV{'HOME'}.'/clucene.output'; |
| 17 |
|
|
$ENV{'LD_LIBRARY_PATH'} = "$dir/perl"; |
| 18 |
|
|
unshift @INC, "$dir/perl"; |
| 19 |
|
|
} |
| 20 |
|
|
|
| 21 |
|
|
use FulltextSearch::CLucene; |
| 22 |
|
|
use Carp; |
| 23 |
|
|
use Data::Dumper; |
| 24 |
|
|
|
| 25 |
|
|
=head1 NAME |
| 26 |
|
|
|
| 27 |
|
|
MWS::Clucene - index your data using CLucene |
| 28 |
|
|
|
| 29 |
|
|
=head1 DESCRIPTION |
| 30 |
|
|
|
| 31 |
|
|
Simple implementation to use Lucene port to C++ |
| 32 |
|
|
|
| 33 |
|
|
=head1 METHODS |
| 34 |
|
|
|
| 35 |
|
|
=head2 open_index |
| 36 |
|
|
|
| 37 |
|
|
$self->open_index; |
| 38 |
|
|
|
| 39 |
|
|
=cut |
| 40 |
|
|
|
| 41 |
|
|
sub open_index { |
| 42 |
|
|
my $self = shift; |
| 43 |
|
|
|
| 44 |
|
|
$self->{index} = FulltextSearch::CLucene->new( path => "./index" ) || croak "can't open index '",$self->{index_dir},"': $!"; |
| 45 |
|
|
|
| 46 |
|
|
return $self->{index}; |
| 47 |
|
|
|
| 48 |
|
|
} |
| 49 |
|
|
|
| 50 |
|
|
=head2 search_index |
| 51 |
|
|
|
| 52 |
|
|
my @results = $self->search_index('message:funny'); |
| 53 |
|
|
|
| 54 |
|
|
Date limits are, well, cludged and sort isn't supported! |
| 55 |
|
|
|
| 56 |
|
|
=cut |
| 57 |
|
|
|
| 58 |
|
|
sub search_index { |
| 59 |
|
|
my $self = shift; |
| 60 |
|
|
|
| 61 |
|
|
my $s = shift || croak "search_index needs query"; |
| 62 |
|
|
|
| 63 |
|
|
$self->open_index if (! $self->{index}); |
| 64 |
|
|
|
| 65 |
|
|
# kill sort:something [asc|desc] |
| 66 |
|
|
$s =~ s/sort:\w+\s+\w+//; |
| 67 |
|
|
|
| 68 |
|
|
my ($y,$m,$d); |
| 69 |
|
|
|
| 70 |
|
|
if ($s =~ s/date:"*(\d{4})(?:-(\d{2}))*(?:-(\d{2}))*"*//) { |
| 71 |
|
|
($y,$m,$d) = ($1,$2,$3); |
| 72 |
|
|
|
| 73 |
|
|
my ($df,$dt); # date from, to |
| 74 |
|
|
|
| 75 |
|
|
if ($y && $m && $d) { |
| 76 |
|
|
$df = $self->fmtdate($y,$m,$y); |
| 77 |
|
|
$dt = $df; |
| 78 |
|
|
} elsif ($y && $m) { |
| 79 |
|
|
$df = $self->fmtdate($y,$m)."-01"; |
| 80 |
|
|
$dt = $self->fmtdate($y,$m)."-31"; |
| 81 |
|
|
} elsif ($y) { |
| 82 |
|
|
$df = $self->fmtdate($y)."-01-01"; |
| 83 |
|
|
$dt = $self->fmtdate($y)."-12-31"; |
| 84 |
|
|
} |
| 85 |
|
|
print STDERR "Plucene query $s [from $df to $dt]\n" if ($self->{debug}); |
| 86 |
|
|
return $self->{index}->search_during($s, $df => $dt); |
| 87 |
|
|
} else { |
| 88 |
|
|
print STDERR "Plucene query $s\n" if ($self->{debug}); |
| 89 |
|
|
return $self->{index}->search($s); |
| 90 |
|
|
} |
| 91 |
|
|
} |
| 92 |
|
|
|
| 93 |
|
|
=head2 add_index |
| 94 |
|
|
|
| 95 |
|
|
$self->add_index($mbox_id, $document); |
| 96 |
|
|
|
| 97 |
|
|
This method will try to fake "message" field which is used to search whole |
| 98 |
|
|
message (including headers) with body and subject. Duplicating content is |
| 99 |
|
|
wrong, but using Plucene::Simple has it's limitations. |
| 100 |
|
|
|
| 101 |
|
|
It will also remove time from date to make Plucene happy. |
| 102 |
|
|
|
| 103 |
|
|
=cut |
| 104 |
|
|
|
| 105 |
|
|
sub add_index { |
| 106 |
|
|
my $self = shift; |
| 107 |
|
|
|
| 108 |
|
|
my $mbox_id = shift || croak "add_index needs mbox_id"; |
| 109 |
|
|
my $document = shift || croak "add_index needs document"; |
| 110 |
|
|
|
| 111 |
|
|
$self->open_index if (! $self->{index}); |
| 112 |
|
|
|
| 113 |
|
|
# does index withh document allready exist? |
| 114 |
|
|
if (-f $self->{index_dir}."/segmets") { |
| 115 |
|
|
return if ($self->{index}->search( field => "id:", query => $document->{id} )); |
| 116 |
|
|
} |
| 117 |
|
|
|
| 118 |
|
|
print STDERR "add_index($mbox_id)\n" if ($self->{debug}); |
| 119 |
|
|
|
| 120 |
|
|
$document->{message} = ($document->{body} || '') . ($document->{subject} || ''); |
| 121 |
|
|
|
| 122 |
|
|
$document->{date} =~ s/\d{4}-\d+-\d+.*//; |
| 123 |
|
|
|
| 124 |
|
|
$self->{index}->new_document; |
| 125 |
|
|
foreach my $f (keys %{$document}) { |
| 126 |
|
|
if (lc($f) eq 'date') { |
| 127 |
|
|
$self->{index}->add_date( field => $f, value => $document->{$f}); |
| 128 |
|
|
} else { |
| 129 |
|
|
$self->{index}->add_field( field => $f, value => $document->{$f}); |
| 130 |
|
|
} |
| 131 |
|
|
} |
| 132 |
|
|
|
| 133 |
|
|
$self->{index}->insert_document; |
| 134 |
|
|
} |
| 135 |
|
|
|
| 136 |
|
|
=head2 close_index |
| 137 |
|
|
|
| 138 |
|
|
$self->close_index; |
| 139 |
|
|
|
| 140 |
|
|
This will also optimize Plucene index file. |
| 141 |
|
|
|
| 142 |
|
|
=cut |
| 143 |
|
|
|
| 144 |
|
|
sub close_index { |
| 145 |
|
|
my $self = shift; |
| 146 |
|
|
|
| 147 |
|
|
$self->{index}->close; |
| 148 |
|
|
} |
| 149 |
|
|
|
| 150 |
|
|
1; |
Parent Directory
| 
Revision Log