Browse Source

Moving the clean up code to a central place.

pull/1242/head
Michael Vogel 7 years ago
parent
commit
04106ff6f5
  1. 5
      include/Scrape.php
  2. 6
      library/HTML5/Parser.php

5
include/Scrape.php

@ -263,11 +263,6 @@ function scrape_feed($url) {
}
try {
// Cleanup invalid HTML
$doc = new DOMDocument();
@$doc->loadHTML($s);
$s = $doc->saveHTML();
$dom = HTML5_Parser::parse($s);
} catch (DOMException $e) {
logger('scrape_feed: parse error: ' . $e);

6
library/HTML5/Parser.php

@ -17,6 +17,12 @@ class HTML5_Parser
* @return Parsed HTML as DOMDocument
*/
static public function parse($text, $builder = null) {
// Cleanup invalid HTML
$doc = new DOMDocument();
@$doc->loadHTML($text);
$text = $doc->saveHTML();
$tokenizer = new HTML5_Tokenizer($text, $builder);
$tokenizer->parse();
return $tokenizer->save();

Loading…
Cancel
Save