9 |
use KinoSearch::Analysis::PolyAnalyzer; |
use KinoSearch::Analysis::PolyAnalyzer; |
10 |
use Encode qw/from_to/; |
use Encode qw/from_to/; |
11 |
use Data::Dumper; |
use Data::Dumper; |
12 |
|
use Storable; |
13 |
|
|
14 |
=head1 NAME |
=head1 NAME |
15 |
|
|
17 |
|
|
18 |
=head1 VERSION |
=head1 VERSION |
19 |
|
|
20 |
Version 0.01 |
Version 0.02 |
21 |
|
|
22 |
=cut |
=cut |
23 |
|
|
24 |
our $VERSION = '0.01'; |
our $VERSION = '0.02'; |
25 |
|
|
26 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
27 |
|
|
100 |
analyzer => $analyzer, |
analyzer => $analyzer, |
101 |
); |
); |
102 |
|
|
103 |
|
my $fields_path = $self->{index_path} . '/fields.storable'; |
104 |
|
$fields_path =~ s#//#/#g; |
105 |
|
if (-e $fields_path) { |
106 |
|
$self->{fields} = retrieve($fields_path) || |
107 |
|
$log->warn("can't open $fields_path: $!"); |
108 |
|
} else { |
109 |
|
$log->error("This will be dummy run since no fields statistics are found!"); |
110 |
|
$log->error("You will have to re-run indexing to get search results!"); |
111 |
|
$self->{dummy_run} = 1; |
112 |
|
} |
113 |
|
$self->{fields_path} = $fields_path; |
114 |
|
|
115 |
foreach my $f (@{ $self->{fields} }) { |
foreach my $f (@{ $self->{fields} }) { |
116 |
$self->{invindex}->spec_field( |
$self->{invindex}->spec_field( |
117 |
name => $f, |
name => $f, |
171 |
sub add_value($$$$$) { |
sub add_value($$$$$) { |
172 |
my ($self,$log,$doc,$n,$v) = @_; |
my ($self,$log,$doc,$n,$v) = @_; |
173 |
return unless ($v); |
return unless ($v); |
174 |
|
|
175 |
|
$self->{value_usage}->{$n}++; |
176 |
|
return if ($self->{dummy_run}); |
177 |
|
|
178 |
eval { $doc->set_value($n, $self->convert($v) ) }; |
eval { $doc->set_value($n, $self->convert($v) ) }; |
179 |
$log->warn("can't insert: $n = $v") if ($@); |
$log->warn("can't insert: $n = $v") if ($@); |
180 |
} |
} |
225 |
sub finish { |
sub finish { |
226 |
my $self = shift; |
my $self = shift; |
227 |
|
|
228 |
$self->_get_logger()->info("finish index writing to disk"); |
my $log = $self->_get_logger(); |
229 |
|
|
230 |
|
$log->info("finish index writing to disk"); |
231 |
$self->{invindex}->finish; |
$self->{invindex}->finish; |
232 |
|
|
233 |
|
$log->info("writing value usage file"); |
234 |
|
|
235 |
|
# add fields from last run |
236 |
|
map { $self->{value_usage}->{$_}++ } @{ $self->{fields} }; |
237 |
|
|
238 |
|
my @fields = keys %{ $self->{value_usage} }; |
239 |
|
store \@fields, $self->{fields_path} || |
240 |
|
$log->warn("can't write $self->{fields_path}: $!"); |
241 |
|
|
242 |
} |
} |
243 |
|
|
244 |
=head2 convert |
=head2 convert |