Merge pull request #1242 from annando/1412-stale-gprobe

Bugfix for stale gprobe.php
This commit is contained in:
Tobias Diekershoff 2014-12-10 06:16:35 +01:00
commit 15d5e6cd45
2 changed files with 11 additions and 2 deletions

View file

@ -10,7 +10,7 @@ function gprobe_run(&$argv, &$argc){
if(is_null($a)) { if(is_null($a)) {
$a = new App; $a = new App;
} }
if(is_null($db)) { if(is_null($db)) {
@include(".htconfig.php"); @include(".htconfig.php");
require_once("include/dba.php"); require_once("include/dba.php");
@ -37,6 +37,8 @@ function gprobe_run(&$argv, &$argc){
dbesc(normalise_link($url)) dbesc(normalise_link($url))
); );
logger("gprobe start for ".normalise_link($url), LOGGER_DEBUG);
if(! count($r)) { if(! count($r)) {
$arr = probe_url($url); $arr = probe_url($url);
@ -55,7 +57,8 @@ function gprobe_run(&$argv, &$argc){
} }
if(count($r)) if(count($r))
poco_load(0,0,$r[0]['id'], str_replace('/profile/','/poco/',$r[0]['url'])); poco_load(0,0,$r[0]['id'], str_replace('/profile/','/poco/',$r[0]['url']));
logger("gprobe end for ".normalise_link($url), LOGGER_DEBUG);
return; return;
} }

View file

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