File Coverage

blib/lib/WWW/Pastebin/PastebinCa/Retrieve.pm

Criterion	Covered	Total	%
statement	55	58	94.8
branch	29	34	85.2
condition	39	48	81.2
subroutine	7	7	100.0
pod			n/a
total	130	147	88.4

package WWW::Pastebin::PastebinCa::Retrieve;

use warnings;
use strict;

our $VERSION = '0.003';

use base 'WWW::Pastebin::Base::Retrieve';
use HTML::TokeParser::Simple;
use HTML::Entities;

sub _make_uri_and_id {
    my ( $self, $id ) = @_;

    my ( $private ) = $id =~ m{(?:http://)? (?:www\.)? (.+?) pastebin\.ca};

    $private = ''
        unless defined $private;

    $id =~ s{ ^ \s+ | (?:http://)? (?:www\.)?.*? pastebin\.ca/ | \s+ $}{}gxi;
    return ( URI->new("http://${private}pastebin.ca/$id"), $id );
}

sub _parse {
    my ( $self, $content ) = @_;
    return $self->_set_error( 'Nothing to parse (empty document retrieved)' )
        unless defined $content and length $content;

    my $parser = HTML::TokeParser::Simple->new( \$content );

    my %data;
    my %nav = (
        level       => 0,
        get_lang    => 0,
        get_name    => 0,
        get_date    => 0,
        get_desc    => 0,
    );
    while ( my $t = $parser->get_token ) {
        if ( $t->is_start_tag('h2')
            #and defined $t->get_attr('class')
            #and $t->get_attr('class') eq 'first'
        ) {

            $nav{level} = 1;
        }
        elsif ( $nav{level} == 1 and $t->is_start_tag('dt') ) {
            @nav{ qw(level  get_name) } = (2, 1);
        }
        elsif ( $nav{get_name} == 1 and $t->is_text ) {
            $data{name} = $t->as_is;
            $nav{get_name} = 0;
        }
        elsif ( $t->is_start_tag('p') and defined $t->get_attr('id')
            and $t->get_attr('id') eq 'des'
        ) {
            $nav{get_desc} = 1;
        }
        elsif ( $nav{get_desc} and $t->is_text ) {
            $data{desc} = $t->as_is;
            $nav{get_desc} = 0;
        }
        elsif ( $nav{level} == 2 and $t->is_start_tag('dd') ) {
            $nav{get_date} = 1;
            $nav{level}++;
        }
        elsif ( $nav{get_date} and $t->is_text ) {
            $data{post_date} = $t->as_is;
            $data{post_date} =~ s/\s+/ /g;
            $data{post_date} =~ s/ //g;
            $nav{get_date}   = 0;
            $nav{level} = 7;
        }
        elsif ( $nav{level} == 7 and $t->is_start_tag('span') ) {
            $nav{level}++;
        }
        elsif ( $t->is_start_tag('textarea')
            and defined $t->get_attr('name')
            and $t->get_attr('name') eq 'content' ) {
            $nav{get_paste} = 1;
        }
        elsif ( $nav{get_paste} and $t->is_text ) {
            $data{content} = $t->as_is;
            $nav{get_paste} = 0;
            $nav{get_lang} = 1;
        }
        elsif ( $nav{get_lang} == 1 and  $t->is_start_tag('select') ) {
            $nav{get_lang} = 2;
        }
        elsif ( $nav{get_lang} == 2 and $t->is_start_tag('option')
            and $t->get_attr('selected')
        ) {
            $nav{get_lang} = 3;
        }
        elsif ( $nav{get_lang} == 3 and $t->is_text ) {
            $data{language} = $t->as_is;
            $nav{success} = 1;
            last;
        }

    }
    unless ( $nav{success} ) {
        my $message = "Failed to parse paste.. ";
        $message .= $nav{level}
                  ? "\$nav{level} == $nav{level}"
                  : "that paste ID doesn't seem to exist";
        return $self->_set_error( $message );
    }

    decode_entities( $_ ) for values %data;

    $self->content( $data{content} );
    return \%data;
}

1;
__END__