fix various html parse errors

diff --git a/include/Scrape.php b/include/Scrape.php
index 642b8e624c5601d1a7ab20c6b4f5c4e76fe5b8b3..f5b15c82fbeb7eafef5c0dc58a2e1502f59db89e 100644 (file)
--- 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 746422a137b5fe4b2bf2f8e2117a1b5b47c36b0b..c7669b381c7305441d427cf9dba291f778c2247a 100644 (file)
--- 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 ec99d1e0dc12a0c0ec1f206187ae958782d60f91..c6f5fd909fa2004157a64ba08f4add5579145d7d 100644 (file)
--- 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,'<?xml')) {
                require_once('library/HTML5/Parser.php');
-               $dom = @HTML5_Parser::parse($html);
+
+               try {
+                       $dom = HTML5_Parser::parse($html);
+               } catch (DOMException $e) {
+                       logger('lrdd: parse error: ' . $e);
+               }
 
                if($dom) {
                        $items = $dom->getElementsByTagName('link');

diff --git a/library/HTML5/TreeBuilder.php b/library/HTML5/TreeBuilder.php
index 03e2ee77fde7d33e8434254aebde167d81e5f0af..a64816f5678bcf205ac6a932b28de6f906dbf80e 100644 (file)
--- a/library/HTML5/TreeBuilder.php
+++ b/library/HTML5/TreeBuilder.php
@@ -3041,6 +3041,8 @@ class HTML5_TreeBuilder {
 
         if (!empty($token['attr'])) {
             foreach($token['attr'] as $attr) {
+                               // mike@macgirvin.com 2011-10-21, stray double quotes cause everything to abort
+                               $attr['name'] = str_replace('"','',$attr['name']);
                 if(!$el->hasAttribute($attr['name'])) {
                     $el->setAttribute($attr['name'], $attr['value']);
                 }

diff --git a/mod/parse_url.php b/mod/parse_url.php
index a238ecb2f915cabb071787f3e0156867611c7a27..7a520ba92f6e5dde83dc8dcf2978abbad803e714 100644 (file)
--- a/mod/parse_url.php
+++ b/mod/parse_url.php
@@ -88,9 +88,11 @@ function parse_url_content(&$a) {
        $purifier = new HTMLPurifier($config);
        $s = $purifier->purify($s);
 
-//     logger('parse_url: purified: ' . $s, LOGGER_DATA);
-
-       $dom = @HTML5_Parser::parse($s);
+       try {
+               $dom = HTML5_Parser::parse($s);
+       } catch (DOMException $e) {
+               logger('scrape_dfrn: parse error: ' . $e);
+       }
 
        if(! $dom) {
                echo sprintf($template,$url,$url,'') . $str_tags;

diff --git a/view/diaspora_vcard.tpl b/view/diaspora_vcard.tpl
index 684b1d18def18bc53b095d33ac0c5c57bd242ee4..e71fc33ef2456ddffbc3b3f91a93ea4ce8a5639f 100644 (file)
--- a/view/diaspora_vcard.tpl
+++ b/view/diaspora_vcard.tpl
@@ -33,19 +33,19 @@
        <dl class="entity_photo">
                <dt>Photo</dt>
                <dd>
-                       <img class="photo avatar" height="300px" width="300px" src="$diaspora.photo300">
+                       <img class="photo avatar" height="300" width="300" src="$diaspora.photo300">
                </dd>
        </dl>
        <dl class="entity_photo_medium">
                <dt>Photo</dt>
                <dd> 
-                       <img class="photo avatar" height="100px" width="100px" src="$diaspora.photo100">
+                       <img class="photo avatar" height="100" width="100" src="$diaspora.photo100">
                </dd>
        </dl>
        <dl class="entity_photo_small">
                <dt>Photo</dt>
                <dd>
-                       <img class="photo avatar" height="50px" width="50px" src="$diaspora.photo50">
+                       <img class="photo avatar" height="50" width="50" src="$diaspora.photo50">
                </dd>
        </dl>
        <dl class="entity_searchable">