top-hosts-2015.php

<?php

$hosts = [];

$doc = new DOMDocument();
$doc->loadHTMLFile("report-2015.html");

$historyTable = $doc->getElementById('history');
$historyTableRows = $historyTable->getElementsByTagName('tr');

foreach ($historyTableRows as $historyTableRow) {
	$urlCell = $historyTableRow->childNodes[0];

	$url = trim($urlCell->firstChild->textContent);
    $urlHost = parse_url($url, PHP_URL_HOST);

    if (empty($urlHost)) {
        continue;
    }

    if (! isset($hosts[$urlHost])) {
        $hosts[$urlHost] = 1;
    } else {
        $hosts[$urlHost]++;
    }
}

arsort($hosts, SORT_NUMERIC);

file_put_contents('top-hosts-2015.json', json_encode($hosts));