/**
* Extracts category tags from wikitext and returns a hash with an array
* of categories data and a modified version of the wikitext with the category
* tags removed.
*
* @param String $wikitext
* @param Boolean $force default === false to skip cache set it to true
*
* @return Array
*/
public static function extractCategoriesFromWikitext($wikitext, $force = false, $lang = null)
{
wfProfileIn(__METHOD__);
if (!$force && is_array(self::$data)) {
wfProfileOut(__METHOD__);
return self::$data;
}
$app = F::app();
// enable changes in Preprocessor and Parser
$app->wg->CategorySelectEnabled = true;
// prepare Parser
$app->wg->Parser->startExternalParse($app->wg->Title, new ParserOptions(), OT_WIKI);
// get DOM tree [PPNode_DOM class] as an XML string
$xml = $app->wg->Parser->preprocessToDom($wikitext)->__toString();
// disable changes in Preprocessor and Parser
$app->wg->CategorySelectEnabled = false;
// add encoding information
$xml = '<?xml version="1.0" encoding="UTF-8"?>' . $xml;
//init variables
self::$nodeLevel = 0;
self::getDefaultNamespaces($lang);
// we will ignore categories added inside following list of tags (BugId:8208)
self::$tagsWhiteList = array_keys($app->wg->Parser->mTagHooks);
// and includeonly tags (BugId:99450)
self::$tagsWhiteList[] = 'includeonly';
//create XML DOM document from provided XML
$dom = new DOMDocument();
$dom->loadXML($xml);
//get everything under main node
$root = $dom->getElementsByTagName('root')->item(0);
self::$frame = $app->wg->Parser->getPreprocessor()->newFrame();
$categories = self::parseNode($root);
// make wikitext from DOM tree
$modifiedWikitext = self::$frame->expand($root, PPFrame::NO_TEMPLATES | PPFrame::RECOVER_COMMENTS);
// replace markers back to wikitext
$modifiedWikitext = $app->wg->Parser->mStripState->unstripBoth($modifiedWikitext);
self::$data = array('categories' => $categories, 'wikitext' => rtrim($modifiedWikitext));
wfProfileOut(__METHOD__);
return self::$data;
}