Parsing RSS with RSSLite


 # simple script that parses an rss file using XML::RSSLite module

use strict;
use IO::File;
use LWP::Simple;
use HTML::Entities;

use XML::RSSLite;

my $lewrss =  "http://someurl.com/file.rss";

my $xml = get($lewrss);
die "Could not retrieve $lewrss" unless $xml;

# print $xml;
# exit;

if ( length($xml) < 1000 ) {
    exit 1;
}

$xml =~ s|<description>.*?</description>|<description></description>|gs;
$xml =~ s|<content:encoded>.*?</content:encoded>|<content:encoded></content:encoded>|gs;

my %result;

parseRSS(\%result, \$xml);

my $website = $result{'title'};

my $prevdate = "";

    foreach my $item (@{$result{'item'}}) {
        my $link  = $item->{'link'};
        my $domain;
        if ( $link =~ m|http://(.*?)/| ) {
            $domain = $1;
            $domain =~ s|^www\.||;
        }
        my $title = $item->{'title'};
        my $date;
        if ( exists($item->{'pubDate'}) ) {
            $date = $item->{'pubDate'};
        } elsif ( exists($item->{'dc:date'}) ) {
            $date = $item->{'dc:date'};
        }
        $date = substr($date, 0, 16);

        if ( $date ne $prevdate ) {
            print "<p><strong>$date</strong></p>\n";
            $prevdate = $date;
        }         
            print "<a href=\"$link\">$title</a> <small> - $domain</small><br />\n";
    }


exit 0;

code=yes