XPath query on all HTML files in a (nested) directory

recursive-xpath.php

<?php

// path = directory containing HTML files
// query = XPath query to run on each file, e.g. "//object/@data"

list(, $path, $query) = $argv;
if(!$path || !$query) exit("Usage: $argv[0] path query\n");
if(!is_dir($path)) exit("$path is not a directory\n");

$files = new RecursiveIteratorIterator(new RecursiveDirectoryIterator($path));
$matches = new RegexIterator($files, '/\.html$/i', RegexIterator::GET_MATCH);

libxml_use_internal_errors(true);
$dom = new DOMDocument;
$items = array();

foreach ($matches as $file => $match) {
	$dom->loadHTMLFile($file);

	$xpath = new DOMXPath($dom);
	$nodes = $xpath->query($query);
	if (!$nodes->length) continue;

	foreach ($nodes as $node) {
		$items[] = $node->textContent;
	}
}

asort($items);

print_r($items);