File Coverage

blib/lib/Lingua/EN/Titlecase/HTML.pm

Criterion	Covered	Total	%
statement	29	29	100.0
branch	13	14	92.8
condition	3	3	100.0
subroutine	6	6	100.0
pod	1	1	100.0
total	52	53	98.1

package Lingua::EN::Titlecase::HTML;
use strict;
use warnings;
use parent "Lingua::EN::Titlecase";
use HTML::TokeParser;

sub lexer : method {
    my $self = shift;
    return $self->{_lexer} if $self->{_lexer};

    my $wp = $self->word_punctuation;
    my $wordish = $self->wordish_rx;

    $self->{_lexer} = sub {
        unless ( $self->{_raw_html} ) {
            my $tmp = $self->{_raw_html} = shift;
            $self->{_parser} = HTML::TokeParser->new(\$tmp);
        }

        if ( defined $self->{__text} and length $self->{__text} )
        {
            $self->{__text} =~ s/\A($wordish)// and return [ "word", "$1" ];
            $self->{__text} =~ s/\A(.)//s and return [ undef, "$1" ];
        }
        elsif ( my $token = $self->{_parser}->get_token() )
        {
            return [ undef, $token->[-1] ] unless $token->[0] eq "T";
            $self->{__text} = $token->[1];
            return $self->{_lexer}->();
        }
        else
        {
            $self->{_raw_html} = undef; # reset for next possible pass
            $self->{_parser} = undef;
            return ();
        }
    };
}

1;

__END__