Integrate HTMLPurifier into liberty.

1 files changed, 142 insertions, 1 deletions

diff --git a/LibertySystem.php b/LibertySystem.php
index 13ea58c..0faade0 100755
--- a/LibertySystem.php
+++ b/LibertySystem.php
@@ -3,7 +3,7 @@
 * System class for handling the liberty package
 *
 * @package  liberty
-* @version  $Header: /cvsroot/bitweaver/_bit_liberty/LibertySystem.php,v 1.68 2007/04/07 18:42:22 wjames5 Exp $
+* @version  $Header: /cvsroot/bitweaver/_bit_liberty/LibertySystem.php,v 1.69 2007/05/17 14:14:29 nickpalmer Exp $
 * @author   spider <spider@steelsun.com>
 */
 
@@ -106,6 +106,147 @@ class LibertySystem extends LibertyBase {
 		}
 	}
 
+	/**
+	 * Return the types of purification supported by purifyHtml
+	 * @returns an array of strings with the types
+	 */
+	function purifyHtmlMethods() {
+		return array('htmlpurifier' => "HTML Purifier", 
+					 'simple' => "Simple Purifier");
+	}
+	
+	/**                                                                     
+	 * Purify HTML from a string.                                           
+	 *                                                                      
+	 * @param string The string to be cleaned.                              
+	 * @returns string The sanitized string                                 
+	 */
+	function purifyHtml($pString) {
+		global $gBitSystem;
+		switch($gBitSystem->getConfig('liberty_html_purifier', 'simple')) {
+		case 'htmlpurifier':
+		    $pString = $this->advancedPurifyHtml($pString);
+		    break;
+			
+		case 'simple':
+		default:
+		    $pString = $this->simplePurifyHtml($pString);
+		    break;
+		}
+		
+		return $pString;
+	}
+	
+	function advancedPurifyHtml($pString) {
+		global $gHtmlPurifier, $gBitSystem;
+		if (!isset($gHtmlPurifier)) {
+		    $blacklistedTags = $gBitSystem->
+				getConfig('blacklisted_html_tags', '');
+			require_once(UTIL_PKG_PATH . 'htmlpurifier/HTMLPurifier.auto.php');
+			$config = HTMLPurifier_Config::createDefault();
+
+			if ($gBitSystem->getConfig('liberty_html_pure_escape_bad', 'y') == 'y') {
+				$config->set('Core', 'EscapeInvalidTags', true);
+				$config->set('Core', 'EscapeInvalidChildren', true);
+			}
+			if ($gBitSystem->getConfig('liberty_html_pure_disable_extern') == 'y') {
+				$config->set('URI', 'DisableExternal', true);
+			}
+			if ($gBitSystem->getConfig('liberty_html_pure_disable_extern_res', 'y') == 'y') {
+				$config->set('URI', 'DisableExternalResources', true);
+			}
+			if ($gBitSystem->getConfig('liberty_html_pure_disable_res') == 'y') {
+				$config->set('URI', 'DisableResources', true);
+			}
+			if ($gBitSystem->getConfig('liberty_html_pure_disable_uri') == 'y') {
+				$config->set('URI', 'Disable', true);
+			}
+			if ($gBitSystem->getConfig('liberty_html_pure_use_redirect') == 'y') {
+				$config->set('URI', 'Munge', LIBERTY_PKG_URL.'redirect.php?q=%s');
+			}
+			if ($gBitSystem->getConfig('liberty_html_pure_strict_html', 'y') == 'y') {
+				$config->set('HTML', 'Strict', true);
+			}
+			if ($gBitSystem->getConfig('liberty_html_pure_xhtml', 'n') == 'n') {
+			    $config->set('Core', 'XHTML', true);
+			}
+			
+			$def =& $config->getHTMLDefinition();
+			// HTMLPurifier doesn't have a blacklist feature. Duh guys!
+			// Note that this has to come last since the other configs
+			// may tweak the def.
+			foreach (explode(',',$blacklistedTags) as $tag) {
+				unset($def->info[$tag]);
+			}
+
+			$gHtmlPurifier = new HTMLPurifier($config);
+		}
+		$pString = $gHtmlPurifier->purify($pString);
+
+		/* There isn't an easy way to disable an attribute in HTMLPurifier */
+		$pString = $this->purifyStyle($pString);
+
+		return $pString;
+	}
+
+	/**
+	 * Removes all style both inline and attributes unless the user
+     * has permission to edit styles.
+	 */
+    function purifyStyle( $pText ) {
+		global $gBitUser;
+
+		$text = $pText;
+		// Yank style - both tag and inline attributes
+		// strip_tags has doesn't recognize that css within the style tags are not document text. To fix this do something similar to the following:
+		if( !$gBitUser->hasPermission( 'p_liberty_edit_html_style' ) ) {
+			$text = preg_replace( "/<style[^>]*>.*<\/style>/siU", '', $text );
+		}
+		$text = stripslashes($text);
+		if( !$gBitUser->hasPermission( 'p_liberty_edit_html_style' ) ) {
+			$text = preg_replace( "/ (style|class)=[\"]?([^\"]*)[\"]?/i", '', $text);
+		}
+
+		return $text;
+	}
+    
+	// This function is a menagerie of the techniques of the comments listed at
+	// http://www.php.net/manual/en/function.strip-tags.php - spiderr
+	function simplePurifyHtml( $pText ) {
+		global $gBitSystem, $gBitUser;
+		
+		// convert all HTML entites to catch people trying to sneak stuff by with things like &#123; etc..
+		if( function_exists( 'html_entity_decode' ) ) {
+			// quieten this down since it causes an error in PHP4
+			// http://bugs.php.net/bug.php?id=25670
+			$text = @html_entity_decode( $pText, ENT_COMPAT, 'UTF-8' );
+		} else {
+			$trans_tbl = get_html_translation_table(HTML_ENTITIES);
+			$trans_tbl = array_flip($trans_tbl);
+			$text = strtr($pText, $trans_tbl);
+		}
+		
+		// strip_tags() appears to become nauseated at the site of a <!DOCTYPE> declaration
+		$text = str_replace( '<!DOCTYPE', '<DOCTYPE', $text );
+		
+		$text = $this->purifyStyle($text);
+
+		// Strip all evil tags that remain
+		// this comes out of gBitSystem->getConfig() set in Liberty Admin
+		$acceptableTags = $gBitSystem->getConfig( 'approved_html_tags', DEFAULT_ACCEPTABLE_TAGS );
+		
+		// Destroy all script code "manually" - strip_tags will leave code inline as plain text
+		if( !preg_match( '/\<script\>/', $acceptableTags ) ) {
+			$text = preg_replace( "/(\<script)(.*?)(script\>)/si", '', $text );
+		}
+		
+		$text = strip_tags( $text, $acceptableTags );
+		$text = str_replace("<!--", "&lt;!--", $text);
+		$text = preg_replace("/(\<)(.*?)(--\>)/mi", "".nl2br("\\2")."", $text);
+		
+		return( $text );
+	}
+
 	// ****************************** Plugin Functions
 	/**
 	 * Load only active plugins from disk