package ExtractText;

use strict;
use Exporter ();
use HTML::Parser;
use LWP::UserAgent;
use Carp qw( croak );

our ( @ISA ) = qw( Exporter );
our ( @EXPORT_OK ) = qw( extract_text );

sub extract_text {
    shift( @_ ) if ( $_[0] eq 'ExtractText' );
    my $uri = shift;

    die( "Single parameter to extract_links() must be a URI to process" )
        unless ( defined( $uri ) );

    my $ua = LWP::UserAgent->new();
    my $res = $ua->get( $uri );

    croak( "Fetch of '$uri' failed: ", $res->status_line() )
        unless( $res->code() == 200 );

    my $parser = HTML::Parser->new(
        text_h => [ \&_parser_text, 'self,dtext,is_cdata' ]
    );
    $parser->parse( $res->content() );

    return( @{ $parser->{_extracted} } );
}

sub _parser_text {
    my ($self, $dtext, $is_cdata) = @_;

    push( @{ $self->{_extracted} }, $dtext )
        unless ( $is_cdata );
}

1;