From e923bc4ab388d00ec757987003e39918756a7c59 Mon Sep 17 00:00:00 2001
From: Christian Schneider
Date: Mon, 12 Feb 2007 15:13:41 +0000
Subject: Added it_html::sanitize to make user-contributed html safe to use

---
 html.class | 50 ++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 48 insertions(+), 2 deletions(-)

diff --git a/html.class b/html.class
index 098e4b3..ccacfea 100644
--- a/html.class
+++ b/html.class
@@ -33,10 +33,11 @@ class it_html
 		'show_content_type' => true,
 		'show_favicon' => true,
 		'show_boot_dom' => true,
+		'staticallycallable' => "q,u,select",	# Those methods are statically callable (have same arguments as global stubs) but are a bit slower
+		'notexported' => "sanitize",	# Those methods are not exported
 	);
 	var $tags_seen = array('body' => true); # body always counts as seen
 	var $_hasnonewline = array();
-	var $_staticallycallable = "q,u,select";	# Those methods are statically callable (have same arguments as global stubs) but are a bit slower
 
 /**
  * Create a HTML object and global functions for all methods (exlcluding
@@ -64,6 +65,7 @@ function it_html($config = array())
 		$this->_htmltype = $this->_oldhtml ? "html" : "xhtml";
 	$this->_oldhtml = $this->_htmltype == "html";
 	$this->_hasnonewline = array_flip(explode(',', "dummy," . $this->_nonewlinetags)); # dummy keeps values >0
+	$notexported = array_flip(explode(',', "dummy," . $this->_notexported)); # dummy keeps values >0
 
 	# Create global functions for _tags
 	foreach (array_merge(explode(',', $this->_tags), explode(',', $this->_moretags)) as $func)
@@ -75,7 +77,7 @@ function it_html($config = array())
 	# Create global functions for it_html methods
 	foreach (get_class_methods(get_class($this)) as $func)
 	{
-		if (!preg_match('/^_/', $func) && !is_a($this, $func) && $func && !function_exists($func))
+		if (!preg_match('/^_/', $func) && !is_a($this, $func) && $func && !function_exists($func) && !$notexported[$func])
 			$code[$func] = "function $func() { \$args = func_get_args(); return \$GLOBALS['$this->_name']->$func(\$args); }";
 	}
 
@@ -223,6 +225,50 @@ function span($args)
 }
 
 
+/**
+ * Return HTML with all evil things stripped. Allowed are a coupld of simple
+ * tags like div, p, i, b, br without attributes, a with absolute href,
+ * img with absolute src url. Also ensures that tags are balanced.
+ * @param $html HTML string to be sanitized
+ * @return Sanitized HTML
+ */
+function sanitize($html)
+{
+	$result = "";
+	$html = it::replace('[\0\n\r\s]+' => " ", $html);
+	$urlpattern = 'https?://[^">]+';
+
+	if ($tag = it::match("(.*)<(div|p|i|b)[^>]*>(.*?)</\\2>(.*)", $html))
+	{
+		# Simple tags with content, no attributes kept
+		list($head, $tagname, $content, $tail) = $tag;
+		$result .= it_html::sanitize($head) . "<$tagname>" . it_html::sanitize($content) . "</$tagname>" . it_html::sanitize($tail);
+	}
+	else if ($tag = it::match('(.*)<a[^>]+?href="(' . $urlpattern . ')"[^>]*?>(.*?)</a>(.*)', $html))
+	{
+		# Link tags, keeps only href attribute
+		list($head, $href, $content, $tail) = $tag;
+		$result .= it_html::sanitize($head) . "<a href=\"$href\">" . it_html::sanitize($content) . "</a>" . it_html::sanitize($tail);
+	}
+	else if ($tag = it::match('(.*)<img[^>]+?src="(' . $urlpattern . ')"[^>]*?>(.*)', $html))
+	{
+		# Image tags, keeps only src attribute
+		list($head, $src, $tail) = $tag;
+		$result .= it_html::sanitize($head) . "<img src=\"$src\" />" . it_html::sanitize($tail);
+	}
+	else if ($tag = it::match("(.*)<(br)[^>]*>(.*)", $html))
+	{
+		# Simple tags without content, no attributes kept
+		list($head, $tagname, $tail) = $tag;
+		$result .= it_html::sanitize($head) . "<$tagname />" . it_html::sanitize($tail);
+	}
+	else
+		$result = Q(html_entity_decode(strip_tags($html)));
+
+	return $result;
+}
+
+
 /**
  * Shortcut: return htmlspecialchars($string);
  * @param $string String to encode with htmlspecialchars()
-- 
cgit v1.2.3