From 01164c8c2f802f0b4c2f0f7c4a83ed826768ae1a Mon Sep 17 00:00:00 2001 From: Friendika Date: Thu, 20 Oct 2011 16:48:07 -0700 Subject: fix various html parse errors --- include/Scrape.php | 27 ++++++++++++++++++++------- include/event.php | 7 ++++++- include/network.php | 7 ++++++- 3 files changed, 32 insertions(+), 9 deletions(-) (limited to 'include') diff --git a/include/Scrape.php b/include/Scrape.php index 642b8e624..f5b15c82f 100644 --- a/include/Scrape.php +++ b/include/Scrape.php @@ -30,8 +30,11 @@ function scrape_dfrn($url) { } } - - $dom = HTML5_Parser::parse($s); + try { + $dom = HTML5_Parser::parse($s); + } catch (DOMException $e) { + logger('scrape_dfrn: parse error: ' . $e); + } if(! $dom) return $ret; @@ -132,9 +135,11 @@ function scrape_meta($url) { } } - - - $dom = HTML5_Parser::parse($s); + try { + $dom = HTML5_Parser::parse($s); + } catch (DOMException $e) { + logger('scrape_meta: parse error: ' . $e); + } if(! $dom) return $ret; @@ -177,7 +182,11 @@ function scrape_vcard($url) { } } - $dom = HTML5_Parser::parse($s); + try { + $dom = HTML5_Parser::parse($s); + } catch (DOMException $e) { + logger('scrape_vcard: parse error: ' . $e); + } if(! $dom) return $ret; @@ -243,7 +252,11 @@ function scrape_feed($url) { } } - $dom = HTML5_Parser::parse($s); + try { + $dom = HTML5_Parser::parse($s); + } catch (DOMException $e) { + logger('scrape_feed: parse error: ' . $e); + } if(! $dom) return $ret; diff --git a/include/event.php b/include/event.php index 746422a13..c7669b381 100644 --- a/include/event.php +++ b/include/event.php @@ -53,7 +53,12 @@ function parse_event($h) { $ret = array(); - $dom = HTML5_Parser::parse($h); + + try { + $dom = HTML5_Parser::parse($h); + } catch (DOMException $e) { + logger('parse_event: parse error: ' . $e); + } if(! $dom) return $ret; diff --git a/include/network.php b/include/network.php index ec99d1e0d..c6f5fd909 100644 --- a/include/network.php +++ b/include/network.php @@ -429,7 +429,12 @@ function lrdd($uri) { // don't try and parse raw xml as html if(! strstr($html,'getElementsByTagName('link'); -- cgit v1.2.3