update the parser lib

This commit is contained in:
friendica 2012-11-18 23:44:58 -08:00
parent 55907cf599
commit fbf58ea0c4
2 changed files with 14 additions and 3 deletions

View File

@ -158,6 +158,8 @@ class HTML5_TreeBuilder {
if ($this->ignore_lf_token) $this->ignore_lf_token--;
$this->ignored = false;
$token['name'] = str_replace(':', '-', $token['name']);
// indenting is a little wonky, this can be changed later on
switch ($mode) {
@ -1429,7 +1431,7 @@ class HTML5_TreeBuilder {
case 'tbody': case 'td': case 'tfoot': case 'th': case 'thead': case 'tr':
// parse error
break;
/* A start tag token not covered by the previous entries */
default:
/* Reconstruct the active formatting elements, if any. */
@ -3038,10 +3040,16 @@ class HTML5_TreeBuilder {
private function insertElement($token, $append = true) {
$el = $this->dom->createElementNS(self::NS_HTML, $token['name']);
if (!empty($token['attr'])) {
foreach($token['attr'] as $attr) {
if(!$el->hasAttribute($attr['name'])) {
// mike@macgirvin.com 2011-11-17, check attribute name for
// validity (ignoring extenders and combiners) as illegal chars in names
// causes everything to abort
$valid = preg_match('/^[a-zA-Z\_\:]([\-a-zA-Z0-9\_\:\.]+$)/',$attr['name'],$matches);
if($attr['name'] && (!$el->hasAttribute($attr['name'])) && ($valid)) {
$el->setAttribute($attr['name'], $attr['value']);
}
}

View File

@ -9,6 +9,8 @@ function submit_content(&$a) {
if(! strlen($url))
exit;
logger('Updating: ' . $url);
$nurl = str_replace(array('https:','//www.'), array('http:','//'), $url);
@ -30,6 +32,7 @@ function submit_content(&$a) {
$parms = scrape_dfrn($url);
// logger('dir_submit: ' . print_r($parms,true));
if((! count($parms)) || (validate_dfrn($parms))) {
exit;