<?php
$karp = file_get_contents('http://spraakbanken.gu.se/ws/karp-sok?op=stat&format=json');
$json = json_decode($karp, true);

$karp_stats = array();
foreach($json['entries'] as $entry){
	if($entry['resource']){
		$karp_stats[$entry['resource']] = $entry['#text'];	
	}
}

$files = glob('metadata/lexicon/*.{xml}', GLOB_BRACE);
foreach($files as $file) {
	$xml = simplexml_load_file($file);
	
	$identifier = "";
	
	$identifiers = $xml->identificationInfo->identifier;

    foreach ($identifiers as $id) {
		$id = (string)$id;
		if(array_key_exists($id, $karp_stats)){
			$identifier = $id;
			break;
		}
	}
	
	echo $file."\n";

	if(array_key_exists($identifier, $karp_stats)){
		if(is_numeric($karp_stats[$identifier])){
			$updated = false;
			$dom = new DOMDocument;
			$dom->load($file);
			
			echo "\tidentifier: ".$identifier."\n";
			echo "\tkarp entries: ".$karp_stats[$identifier]."\n";
			
			$nodes = $dom->getElementsByTagName('sizeInfo');
			foreach ($nodes as $node) {
				$sizeUnit = $node->getElementsByTagName('sizeUnit')->item(0);
				$size = $node->getElementsByTagName('size')->item(0);
				
				if($sizeUnit->nodeValue == 'entries'){
					echo "\told entries: {$size->nodeValue} \n";
					if($size->nodeValue != $karp_stats[$identifier]){
						$size->nodeValue = $karp_stats[$identifier];
						echo "\tupdated {$sizeUnit->nodeValue} to {$size->nodeValue}\n";
						$updated = true;
					}else{
						echo "\tno update for number of entires\n";
					}
				}
			}
			
			if($updated){
				$dom->save($file);
				echo "\tsaved '{$file}' \n";
			}else{
				echo "\tno changes to '{$file}' \n";
			}
		}else{
			echo "\tentries : missing?\n";
		}
	}else{
		echo "\terror : resource not in karp\n";
	}
}
?>