File Coverage

blib/lib/WWW/Dict/Zdic.pm

Criterion	Covered	Total	%
statement	30	56	53.5
branch	0	4	0.0
condition	0	2	0.0
subroutine	10	12	83.3
pod	2	2	100.0
total	42	76	55.2

package WWW::Dict::Zdic;

use warnings;
use strict;
use v5.8.0;
our $VERSION = '0.0.4';

use base 'WWW::Dict';

use WWW::Mechanize;
use Encode;
use URI::Escape;
use HTML::TagParser;
use HTML::Entities;
use Class::Field qw'field const';

# Module implementation here

field ua => -init => 'WWW::Mechanize->new()';
const dict_url => 'http://www.zdic.net';
field word => '';

sub define {
    my $self = shift;
    my $word = shift;
    $self->word($word);
    $word = URI::Escape::uri_escape_utf8($word);
    $word =~ s/%/Zdic/g;
    my $url = $self->dict_url . "/zd/zi/${word}.htm";
    $self->ua->get($url);
    $self->parse_content( $self->ua->content() );
}

sub parse_content {
    my $self = shift;
    my $html = HTML::TagParser->new;
    $html->parse(shift);
    my @def;
    for my $elem ( $html->getElementsByTagName("div") ) {
        my $attr = $elem->attributes;
        next unless ( 'tab-page' eq ( $attr->{class} || '' ) );
        my $innerText = $elem->innerText;
        if (! Encode::is_utf8($innerText)) {
            $innerText = Encode::decode('utf8', $innerText);
            $innerText = decode_entities( $innerText );
        }
        my @d = grep { !/tabPane|^\s*$/ }
          map {
            s/--.*?Zdic.net.*?--//g;
            s/\t//g;
            $_;
          }
          split( /\r?\n/, $innerText );
        push @def,
          {
            category   => $d[0],
            definition => join( "\n", @d[ 1 .. $#d ] )
          };
    }
    return \@def;
}

1;    # Magic true value required at end of module
__END__