| line | stmt | bran | cond | sub | pod | time | code | 
| 1 |  |  |  |  |  |  | package KinoSearch1::Analysis::Stopalizer; | 
| 2 | 4 |  |  | 4 |  | 35228 | use strict; | 
|  | 4 |  |  |  |  | 13 |  | 
|  | 4 |  |  |  |  | 143 |  | 
| 3 | 4 |  |  | 4 |  | 22 | use warnings; | 
|  | 4 |  |  |  |  | 8 |  | 
|  | 4 |  |  |  |  | 101 |  | 
| 4 | 4 |  |  | 4 |  | 468 | use KinoSearch1::Util::ToolSet; | 
|  | 4 |  |  |  |  | 10 |  | 
|  | 4 |  |  |  |  | 608 |  | 
| 5 | 4 |  |  | 4 |  | 23 | use base qw( KinoSearch1::Analysis::Analyzer ); | 
|  | 4 |  |  |  |  | 8 |  | 
|  | 4 |  |  |  |  | 916 |  | 
| 6 |  |  |  |  |  |  |  | 
| 7 |  |  |  |  |  |  | BEGIN { | 
| 8 | 4 |  |  | 4 |  | 56 | __PACKAGE__->init_instance_vars( | 
| 9 |  |  |  |  |  |  | # constructor params / members | 
| 10 |  |  |  |  |  |  | stoplist => undef, | 
| 11 |  |  |  |  |  |  | ); | 
| 12 |  |  |  |  |  |  | } | 
| 13 |  |  |  |  |  |  |  | 
| 14 | 4 |  |  | 4 |  | 3308 | use Lingua::StopWords; | 
|  | 4 |  |  |  |  | 1303 |  | 
|  | 4 |  |  |  |  | 1432 |  | 
| 15 |  |  |  |  |  |  |  | 
| 16 |  |  |  |  |  |  | sub init_instance { | 
| 17 | 3 |  |  | 3 | 1 | 8 | my $self = shift; | 
| 18 | 3 |  |  |  |  | 37 | my $language = $self->{language} = lc( $self->{language} ); | 
| 19 |  |  |  |  |  |  |  | 
| 20 |  |  |  |  |  |  | # verify a supplied stoplist | 
| 21 | 3 | 100 |  |  |  | 18 | if ( defined $self->{stoplist} ) { | 
| 22 | 1 | 50 |  |  |  | 11 | croak("stoplist must be a hashref") | 
| 23 |  |  |  |  |  |  | unless reftype( $self->{stoplist} ) eq 'HASH'; | 
| 24 |  |  |  |  |  |  | } | 
| 25 |  |  |  |  |  |  | else { | 
| 26 |  |  |  |  |  |  | # create a stoplist if language was supplied | 
| 27 | 2 | 50 |  |  |  | 21 | if ( $language =~ /^(?:da|de|en|es|fr|it|nl|no|pt|ru|sv)$/ ) { | 
|  |  | 0 |  |  |  |  |  | 
| 28 | 2 |  |  |  |  | 14 | $self->{stoplist} = Lingua::StopWords::getStopWords($language); | 
| 29 |  |  |  |  |  |  | } | 
| 30 |  |  |  |  |  |  | # No Finnish stoplist, though we have a stemmmer. | 
| 31 |  |  |  |  |  |  | elsif ( $language eq 'fi' ) { | 
| 32 | 0 |  |  |  |  |  | $self->{stoplist} = {}; | 
| 33 |  |  |  |  |  |  | } | 
| 34 |  |  |  |  |  |  | else { | 
| 35 | 0 |  |  |  |  |  | confess "Invalid language: '$language'"; | 
| 36 |  |  |  |  |  |  | } | 
| 37 |  |  |  |  |  |  | } | 
| 38 |  |  |  |  |  |  | } | 
| 39 |  |  |  |  |  |  |  | 
| 40 |  |  |  |  |  |  | 1; | 
| 41 |  |  |  |  |  |  |  | 
| 42 |  |  |  |  |  |  | __END__ |