Extract attribute values from JATS wiki tables

jats-wiki-table-extract.php

<?php

$doc = new DOMDocument;
$doc->loadHTMLFile('http://jatswiki.org/wiki/Attribute_values');

$data = [];

$xpath = new DOMXPath($doc);
foreach ($xpath->query('//table') as $table) {
	$type = $xpath->evaluate('string(preceding-sibling::h2[1])', $table);
	$type = preg_replace('/ values$/', '', $type);

	$keys = [];
	foreach ($xpath->query('tr[1]/th', $table) as $heading) {
		$keys[] = strtolower(trim($heading->textContent));
	}

	$items = [];
	foreach ($xpath->query('tr[position()>1]', $table) as $row) {
		$item = [];
		foreach ($xpath->query('td', $row) as $index => $cell) {
			$item[$keys[$index]] = trim($cell->textContent);
		}

		$items[] = $item;
	}

	$data[$type] = $items;
}

print json_encode($data, JSON_PRETTY_PRINT);