<?php
set_time_limit(0);

$contents = file_get_contents("http://nl.wikipedia.org/wiki/Lijst_van_gemeenten_in_het_Vlaams_Gewest");
$lines = explode("\n", $contents);
$start = false;
$file = "E:\Projects\WebApps\gemeentes.xml";

$xml = "<?xml version=\"1.0\" encoding=\"UTF-8\" ?>\n<gemeentes>\n";
$aantal = 0;

foreach($lines as $line) {
    if(preg_match("~<a href=\"(/wiki/.+?)\"~", $line, $matches)) {
        if($matches[1] == "/wiki/Aalst_(Oost-Vlaanderen)") {
            $start = true;
        }

        if($matches[1] == "/wiki/Categorie:Alles") {
            $start = false;
        }

        if($start == true) {
            $c = file_get_contents("http://nl.wikipedia.org" . $matches[1]);
            if(preg_match("~Overige informatie~", $c)) {
                $start = strpos($c, "<a href=\"/wiki/Postcode\">Postcode</a>");
                $end = strpos($c, "<a href=\"/wiki/Netnummer\"");
                $c = substr($c, $start, $end-$start);

                preg_match_all("~(\d{4})~", $c, $results);
                $c2 = substr($c, strpos($c, "Deelgemeente</a></b><br>"));
                preg_match_all("~(<a.*>)?([a-zA-Z\-\s]+)(</a>)?(<br( /)?>|</td>)~", $c2, $results2);

                if(count($results[1]) == count($results2[2])) {
                    $count = count($results[1]);
                    $aantal += $count;
                    for($i = 0 ; $i < $count ; $i++) {
                        $xml .= "\t<gemeente>\n";
                        $xml .= "\t\t<postcode>" . $results[1][$i] . "</postcode>\n";
                        $xml .= "\t\t<naam>" . trim($results2[2][$i]) . "</naam>\n";
                        $xml .= "\t</gemeente>\n";
                    }
                }
            }
        }
    }
}

$xml .= "\t<totaal aantal=\"$aantal\" />\n";
$xml .= "</gemeentes>";
//lege lijn weghalen
$xml = preg_replace("~^\n$~", "", $xml);

file_put_contents($file, $xml);

header('content-type: text/xml');
echo $xml;
?>