public function indexDoc($doc, $url) { $fields = array(); $fields['url'] = urlencode($url); $fields['docId'] = md5($fields['url']); if ($oldDoc = $this->_documentExists($fields['docId'])) { if (md5($doc->body) == $oldDoc->checkSum) { Bbx_Log::write('No change in document', null, self::LOG); return; } else { $this->_deleteDocument($oldDoc->id); } } $contentSample = preg_replace('/([\\s\\r\\n\\t]*<[^>]+>[\\s\\r\\n\\t]*)+/', ' ', $doc->getHtmlBody()); $contentSample = utf8_excerpt(trim($contentSample), 128, false, true); $fields['contentSample'] = htmlentities($contentSample, ENT_COMPAT, 'UTF-8'); $this->_indexDocument($doc, $fields); }
function utf8_excerpt_for_content($text) { return utf8_excerpt($text, 'content'); }