scrape-fu

scrape-fu.php

<?php
require 'phpQuery/phpQuery.php';

$prefix = 'http://fyrrum.dk/portal.php';

$doc = phpQuery::newDocumentHTML(file_get_contents(
    $prefix
));

$captures = array();
foreach ($doc['#page-body table dl.icon'] as $node) {
    
    $capture = array();
    
    $href = pq($node)->find('a.topictitle:first');
    
    if ($href->count() == 0) {
        continue;
    }
    
    $capture['topic'] = $href->text();
    $url = $href->attr('href');
    if ($url[0] == '.') {
        $url = substr($url, 1);
    }
    $capture['url'] = $prefix . $url;
    
    $last = pq($node)->find('dd.lastpost');
    $capture['last_post_by'] = $last->find('a:first')->text();
        
    $at = $last->find('span')->text();
    $parts = explode("\n", $at);
    $capture['last_post_at'] = trim($parts[1]);
    
    $captures[] = $capture;
}

foreach ($captures as $c) {
    printf("%s / %s / %s\n", $c['last_post_at'], $c['topic'], $c['last_post_by']);
}
?>