/*
The PHP5 function below generates a semicolon-separated file like the following:
"Pork and Beans" acoustic with Tay Zonday;http://www.youtube.com/watch?v=QQnT3psRi7Q
12 Angry Men;http://www.youtube.com/watch?v=52y55e3LpDQ
12:01 PM, time-travel short film from 1990;http://www.youtube.com/watch?v=PhRXTX0HFgw
...
*/


function collectData() {
    $lines = array();
    for ($year = 2005; $year <= 2008; $year++) {
        for ($month = 1; $month <= 12; $month++) {
            $overviewUrl = 'http://waxy.org/links/archive/' . $year . '/' . padVar($month, 2) . '/index.shtml';
            echo '<strong>' . $overviewUrl . '</strong><br />'; doFlush();

            $dom = new domdocument;
            @$dom->loadHtmlFile($overviewUrl);
            $oXpath = new domXpath($dom);
            $nodes = $oXpath->query('//div[@class="item"]/a[@href]');
            foreach ($nodes as $node) {
                $url = $node->getAttribute('href');
                if ( striPos($url, 'http://www.youtube.com/watch') !== false ) {
                    $title = $node->firstChild->nodeValue;
                    $title = str_iReplace('Video: ', '', $title);
                    $line = str_iReplace(';', ' ', $title) . ';' . str_iReplace(';', ' ', $url);
                    $lines[] = $line; 
                    echo misc_toXml($line) . '<br />';
                }
            }
        }
    }
    sort($lines);
    setFileText( 'waxy.txt', implode("\r\n", $lines) );
}