parse-tags-only.php

<?php

/* this is a simple script to generate a tree hierarchy based on element tags
  by Stéfan Sinclair, no license (use and abuse as you wish, no reponsibility here) */
  
  
ini_set("memory_limit", "512M");

$xml = simplexml_load_file(dirname(__FILE__) . '/tagsOnly.xml');

$paths = array();
getChildren($xml, array('ORLANDO'));
echo json_encode(outputChildren('ORLANDO/ENTRY'));

function outputChildren($path) {
  global $paths, $path_keys;
	$children = array();
	if (isset($path_keys[$path])) {
		foreach (array_keys($path_keys[$path]) as $key) {
			if ($paths["$path/$key"]>50) {
				$children[] = outputChildren("$path/$key");
			}
		}
	}
	$result = array(
		"name" => substr($path, strrpos($path, '/')+1),
		"size" => $paths[$path]
	);
	if ($children) {$result['children']=$children;}
	return $result;
}

function getChildren($xml, $stack) {
	global $paths, $path_keys;
	foreach ($xml->children() as $element => $child) {
		$path_keys[implode($stack, "/")][$element]++;
		array_push($stack, $element);
		$paths[implode($stack, "/")]++;
		getChildren($child, $stack);
		array_pop($stack);
	}
}