From 01164c8c2f802f0b4c2f0f7c4a83ed826768ae1a Mon Sep 17 00:00:00 2001
From: Friendika <info@friendika.com>
Date: Thu, 20 Oct 2011 16:48:07 -0700
Subject: [PATCH] fix various html parse errors

---
 include/Scrape.php            | 27 ++++++++++++++++++++-------
 include/event.php             |  7 ++++++-
 include/network.php           |  7 ++++++-
 library/HTML5/TreeBuilder.php |  2 ++
 mod/parse_url.php             |  8 +++++---
 view/diaspora_vcard.tpl       |  6 +++---
 6 files changed, 42 insertions(+), 15 deletions(-)

diff --git a/include/Scrape.php b/include/Scrape.php
index 642b8e624c..f5b15c82fb 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 $ret;
diff --git a/include/event.php b/include/event.php
index 746422a137..c7669b381c 100644
--- a/include/event.php
+++ b/include/event.php
@@ -53,7 +53,12 @@ function parse_event($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 $ret;
diff --git a/include/network.php b/include/network.php
index ec99d1e0dc..c6f5fd909f 100644
--- a/include/network.php
+++ b/include/network.php
@@ -429,7 +429,12 @@ function lrdd($uri) {
 	// don't try and parse raw xml as html
 	if(! strstr($html,'<?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 03e2ee77fd..a64816f567 100644
--- 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'], $attr['value']);
                 }
diff --git a/mod/parse_url.php b/mod/parse_url.php
index a238ecb2f9..7a520ba92f 100644
--- a/mod/parse_url.php
+++ b/mod/parse_url.php
@@ -88,9 +88,11 @@ function parse_url_content(&$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,'') . $str_tags;
diff --git a/view/diaspora_vcard.tpl b/view/diaspora_vcard.tpl
index 684b1d18de..e71fc33ef2 100644
--- a/view/diaspora_vcard.tpl
+++ b/view/diaspora_vcard.tpl
@@ -33,19 +33,19 @@
 	<dl class="entity_photo">
 		<dt>Photo</dt>
 		<dd>
-			<img class="photo avatar" height="300px" width="300px" src="$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">