Parse RSS Yandex Search

gistfile1.txt

#!/usr/bin/perl
use utf8;
use strict;

use lib '../../lib';
use common;

use XML::Feed;
use Storable qw(nstore retrieve);
use Data::Dumper;

my $F = 'blog.store';
my $E = @ARGV ? {} : eval { retrieve $F };

my $feed = XML::Feed->parse(URI->new(
	'http://blogs.yandex.ru/search.rss?ft=all&full=1&server=twitter.com%2C%20station20.ru%2C%20twitterlinks.ru&text=%22%D0%A1%D0%A2%D0%90%D0%9D%D0%A6%D0%98%D0%AF%202.0%22%20|%20%22station20%22&x_server=on'
)) or die XML::Feed->errstr;

my $count;
for ($feed->entries) {
	$DB->query(
		'insert into chat set type="blog", message=?, etc=?, created=now() - interval ? minute',
		_clean($_->content->body),
		join('::', $_->author, $_->title, $_->link),
		$count++ || 0 # fuck yandex - no time2iso(str2time($_->issued)) :)
	) unless $E->{ $_->link }++;
}
print $count;

nstore $E, $F unless @ARGV;

sub _clean($) {
	for (my $str = shift) {
		s{</?ns0:hlword>}{}sg;
		return $_;
	}
}