Integrate HTMLPurifier into liberty.

8 files changed, 250 insertions, 77 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
diff --git a/admin/admin_liberty_inc.php b/admin/admin_liberty_inc.php
index 2fe7f56..efcc085 100644
--- a/admin/admin_liberty_inc.php
+++ b/admin/admin_liberty_inc.php
@@ -18,6 +18,50 @@ $formLibertyFeatures = array(
 	),
 );
 
+$formLibertyHtmlPurifierFeatures = array(
+	'liberty_html_pure_escape_bad' => array(
+		'label' => 'Escape invalid HTML',
+		'note' => ' Escapes invlid HTML as text. Otherwise invalid HTML is silently dropped. See <a href="http://htmlpurifier.org/live/configdoc/plain.html#Core.EscapeInvalidTags">this</a> and <a href="http://htmlpurifier.org/live/configdoc/plain.html#Core.EscapeInvalidChildren">this</a> for more information.',
+		'default' => 'y'
+	),
+	'liberty_html_pure_disable_extern' => array(
+		'label' => 'Disable External Links',
+		'note' => 'Disables links to external websites which is effective against spam. See <a href="http://htmlpurifier.org/live/configdoc/plain.html#URI.DisableExternal">this</a> for more information.',
+		'default' => 'n'
+	),
+	'liberty_html_pure_disable_extern_res' => array(
+		'label' => 'Disable External Resounces',
+		'note' => 'Disables the embedding of external resource like images from other hosts. See <a href="http://htmlpurifier.org/live/configdoc/plain.html#URI.DisableExternalResources">this</a> for more information.',
+		'default' => 'y'
+	),
+	'liberty_html_pure_disable_res' => array(
+		'label' => 'Disable All Resources',
+		'note' => 'Disables the embedding of all resources preventing users from including pictures at all. See <a href="http://htmlpurifier.org/live/configdoc/plain.html#URI.DisableResources">this</a> for more information.',
+		'default' => 'n'
+	),
+	'liberty_html_pure_disable_uri' => array(
+		'label' => 'Disable all URIs',
+		'note' => 'Disables all URIs in all forms within submitted content. See <a href="http://htmlpurifier.org/live/configdoc/plain.html#URI.Disable">this</a> for more information.',
+		'default' => 'n'
+	),
+	'liberty_html_pure_use_redirect' => array(
+		'label' => 'Use Redirect',
+		'note' => 'Uses the redirect service in the Redirect URI. This can be handy to track clicks out and prevent leacks of PageRank. See <a href="http://htmlpurifier.org/live/configdoc/plain.html#URI.Munge">this</a> for more information.',
+		'default' => 'n'
+	),
+	'liberty_html_pure_strict_html' => array(
+		'label' => 'Force Strict',
+		'note' => 'Determines if the purification matches the Transitional or Strict rule sets. See <a href="http://htmlpurifier.org/live/configdoc/plain.html#HTML.Strict">this</a> for more information.',
+		'default' => 'y'
+	),
+	'liberty_html_pure_xhtml' => array(
+		'label'  => 'Force XHTML',
+		'note' => 'Determine if purification forces only XHTML tags or if it allows standard HTML.',
+		'default' => 'n'
+	)
+);
+$gBitSmarty->assign( 'formLibertyHtmlPurifierFeatures', $formLibertyHtmlPurifierFeatures );
+
 $formLibertyTextareaFeatures = array(
 	"liberty_textarea_height" => array(
 		'label' => 'Default Textarea Height',
@@ -99,17 +143,26 @@ $formValues = array( 'image_processor', 'liberty_attachment_link_format', 'comme
 
 if( !empty( $_REQUEST['change_prefs'] ) ) {
 	$errors = array();
-	$formFeatures = array_merge( $formLibertyFeatures, $formImageFeatures, $formCaptcha );
+	$formFeatures = array_merge( $formLibertyFeatures, $formImageFeatures, $formCaptcha, $formLibertyHtmlPurifierFeatures );
 	foreach( $formFeatures as $item => $data ) {
 		simple_set_toggle( $item, LIBERTY_PKG_NAME );
 	}
 	foreach( $formLibertyTextareaFeatures as $item => $data ) {
 		simple_set_value( $item, LIBERTY_PKG_NAME );
 	}
+	simple_set_value('liberty_html_purifier', LIBERTY_PKG_NAME );
 	simple_set_value( 'liberty_attachment_style', LIBERTY_PKG_NAME );
 	$gBitSystem->storeConfig('liberty_cache', $_REQUEST['liberty_cache'], LIBERTY_PKG_NAME );
 	$gBitSystem->storeConfig('liberty_auto_display_attachment_thumbs', $_REQUEST['liberty_auto_display_attachment_thumbs'], LIBERTY_PKG_NAME );
 
+	if( !empty($_REQUEST['blacklisted_html_tags']) ) {
+	    $tags = preg_replace( '/\s/', '', $_REQUEST['blacklisted_html_tags'] );
+		if (strlen( $tags ) > 250) {
+			$tags = substr( $tags, 0, 250 );
+			$errors['blacklist'] = 'The blacklisted tags list has been shortened. You can only have 250 characters for blacklisted tags.';
+		}
+		$gBitSystem->storeConfig('blacklisted_html_tags', $tags , LIBERTY_PKG_NAME );
+	}
 	if( $_REQUEST['approved_html_tags'] != DEFAULT_ACCEPTABLE_TAGS ) {
 		$tags = preg_replace( '/\s/', '', $_REQUEST['approved_html_tags'] );
 		$lastAngle = strrpos( $tags, '>' ) + 1;
diff --git a/admin/schema_inc.php b/admin/schema_inc.php
index 6af69d0..1c95086 100644
--- a/admin/schema_inc.php
+++ b/admin/schema_inc.php
@@ -249,6 +249,8 @@ $gBitInstaller->registerPreferences( LIBERTY_PKG_NAME, array(
 	array(LIBERTY_PKG_NAME, 'liberty_action_log', 'y'),
 //	array(LIBERTY_PKG_NAME, 'liberty_attachment_link_format', 'wiki') not needed anymore since we use js in the edit page now (depends on format of content)
 //	array(LIBERTY_PKG_NAME, 'liberty_attachment_style', 'standard'),
+	// The default for new installs is htmlpurifier old stays simple
+	array(LIBERTY_PKG_NAME, 'liberty_html_purifier', 'htmlpurifier'),
 ) );
 
 $gBitInstaller->registerSchemaDefault( LIBERTY_PKG_NAME, array(
diff --git a/icons/bitweaver/htmlpurifier.png b/icons/bitweaver/htmlpurifier.png
new file mode 100644
index 0000000..119c4b9
--- /dev/null
+++ b/icons/bitweaver/htmlpurifier.png
diff --git a/plugins/format.bbcode.php b/plugins/format.bbcode.php
index ba37155..17e58f9 100644
--- a/plugins/format.bbcode.php
+++ b/plugins/format.bbcode.php
@@ -1,6 +1,6 @@
 <?php
 /**
- * @version  $Revision: 1.8 $
+ * @version  $Revision: 1.9 $
  * @package  liberty
  * @subpackage plugins_format
  */
@@ -9,7 +9,6 @@ global $gLibertySystem;
 
 /**
  * run 'pear install Text_Wiki_BBCode-alpha' to install the library,
- * you also need to enable the HTML plugin for now to due to dependency on the purge_html function
  */ 
 require_once('PEAR.php');
 
@@ -35,12 +34,9 @@ $pluginParams = array (
 $gLibertySystem->registerPlugin( PLUGIN_GUID_BBCODE, $pluginParams );
 
 function bbcode_verify_data( &$pParamHash ) {
+    	global $gLibertySystem;
 	$errorMsg = NULL;
-	if( !function_exists( 'purge_html' ) && include_once( LIBERTY_PKG_PATH.'plugins/format.bithtml.php' ) ) {
-		$pParamHash['content_store']['data'] = purge_html( $pParamHash['edit'] );
-	} else {
-		$pParamHash['content_store']['data'] = $pParamHash['edit'];
-	}
+	$pParamHash['content_store']['data'] = $gLibertySystem->purifyHtml( $pParamHash['edit']);
 	return $errorMsg;
 }
 
diff --git a/plugins/format.bithtml.php b/plugins/format.bithtml.php
index 1cd856b..bde0529 100644
--- a/plugins/format.bithtml.php
+++ b/plugins/format.bithtml.php
@@ -1,6 +1,6 @@
 <?php
 /**
- * @version  $Revision: 1.13 $
+ * @version  $Revision: 1.14 $
  * @package  liberty
  * @subpackage plugins_format
  */
@@ -26,56 +26,12 @@ $pluginParams = array (
 $gLibertySystem->registerPlugin( PLUGIN_GUID_BITHTML, $pluginParams );
 
 function bithtml_verify_data( &$pParamHash ) {
+    	global $gLibertySystem;
 	$errorMsg = NULL;
-	$pParamHash['content_store']['data'] = purge_html( $pParamHash['edit'] );
+	$pParamHash['content_store']['data'] = $gLibertySystem->purifyHtml( $pParamHash['edit'] );
 	return $errorMsg;
 }
 
-// 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 purge_html( $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', $text );
-
-	// Yank style - both tag and inline attributes
-	// strip_tags has doesn't recognize that css within the style tags are not document text. To fix this do something similar to the following:
-	if( !$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);
-	}
-
-	// 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( $text );
-}
-
 function bithtml_save_data( &$pParamHash ) {
 	static $parser;
 	if( empty( $parser ) ) {
diff --git a/plugins/format.tikiwiki.php b/plugins/format.tikiwiki.php
index df591e7..e2fd12a 100644
--- a/plugins/format.tikiwiki.php
+++ b/plugins/format.tikiwiki.php
@@ -1,6 +1,6 @@
 <?php
 /**
- * @version  $Revision: 1.93 $
+ * @version  $Revision: 1.94 $
  * @package  liberty
  */
 global $gLibertySystem;
@@ -759,7 +759,7 @@ class TikiWikiParser extends BitBase {
 	}
 
 	function parse_data( $pParseHash, &$pCommonObject ) {
-		global $gBitSystem, $gBitUser, $page;
+		global $gBitSystem, $gLibertySystem, $gBitUser, $page;
 
 		$data      = $pParseHash['data'];
 		$contentId = $pParseHash['content_id'];
@@ -783,17 +783,7 @@ class TikiWikiParser extends BitBase {
 
 		// disable HTML in wiki page for now - very disruptive. should be changed into a per page setting - xing
 		if( !empty( $contentPrefs['content_enter_html'] ) ) {
-			// this is copied and pasted from format.bithtml.php - xing
-			// 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 ) ) {
-				$data = preg_replace( "/(\<script)(.*?)(script\>)/si", '', $data );
-			}
-
-			$data = strip_tags( $data, $acceptableTags );
+		    $data = $gLibertySystem->purifyHtml($data);
 		} elseif( !$gBitSystem->isFeatureActive( 'content_allow_html' ) ) {
 			// convert HTML to chars
 			$data = htmlspecialchars( $data, ENT_NOQUOTES, 'UTF-8' );
diff --git a/templates/admin_liberty.tpl b/templates/admin_liberty.tpl
index 93ad3fc..46c6ec3 100644
--- a/templates/admin_liberty.tpl
+++ b/templates/admin_liberty.tpl
@@ -1,5 +1,5 @@
 {strip}
-{form}
+{form}	
 	{legend legend="General Settings"}
 		{foreach from=$formLibertyFeatures key=item item=output}
 			<div class="row">
@@ -38,21 +38,56 @@
 		</div>
 
 		<div class="row">
-			{formlabel label="Acceptable HTML tags" for="approved_html_tags"}
-			{formfeedback warning=$errors.warning}
+			{formlabel label="Liberty Cache" for="liberty_cache"}
 			{forminput}
-				<input type="text" id="approved_html_tags" name="approved_html_tags" size="50" maxlength="250" value="{$approved_html_tags|escape}" />
-				{formhelp note="List of allowed HTML tags. All other tags will be stripped when users save content. This will affect all format plugins."}
+				{html_options name=liberty_cache id=liberty_cache values=$cacheTimes options=$cacheTimes selected=$gBitSystem->getConfig('liberty_cache')}
+				{formhelp note='Cache all parsed content. This will dramatically reduce load on the server if pages are called frequently.' page=''}
 			{/forminput}
 		</div>
+	{/legend}
 
+	{legend legend="HTML Cleanup"}
 		<div class="row">
-			{formlabel label="Liberty Cache" for="liberty_cache"}
+			{formlabel label="Purification System"}
 			{forminput}
-				{html_options name=liberty_cache id=liberty_cache values=$cacheTimes options=$cacheTimes selected=$gBitSystem->getConfig('liberty_cache')}
-				{formhelp note='Cache all parsed content. This will dramatically reduce load on the server if pages are called frequently.' page=''}
+				{html_options name=liberty_html_purifier options=$gLibertySystem->purifyHtmlMethods() selected=$gBitSystem->getConfig('liberty_html_purifier', 'simple')}
+				{formhelp note="Which system should be used to purify incoming HTML. The simple algorithm is faster but <strong>far less</strong> robust and secure than <a href=http://htmlpurifier.org>HTML Purifier</a> which has a much richer feature set. HTMLPurifier is recommended to protect against the most XSS attacks. The Simple system is known to <strong>fail XSS smoke tests</strong> and is therefore not recommended."}
 			{/forminput}
 		</div>
+
+
+		{legend legend="Simple Purifier Features"}
+			<div class="row">
+				{formlabel label="Acceptable HTML tags" for="approved_html_tags"}
+				{formfeedback warning=$errors.warning}
+				{forminput}
+					<input type="text" id="approved_html_tags" name="approved_html_tags" size="50" maxlength="250" value="{$approved_html_tags|escape}" />
+					{formhelp note="List of allowed HTML tags. All other tags will be stripped when users save content. This will affect all format plugins and all purification systems."}
+				{/forminput}
+			</div>
+		{/legend}
+
+		{legend legend="HTMLPurifier Features"}
+			<div class="row">
+				{formlabel label="Blacklisted HTML tags" for="blacklisted_html_tags"}
+				{formfeedback warning=$errors.blacklist}
+				{forminput}
+					<input type="text" id="blacklisted_html_tags" name="blacklisted_html_tags" size="50" maxlength="250" value="{$gBitSystem->getConfig('blacklisted_html_tags')|escape}" />
+					{formhelp note="A comma seperated list of tags that should NOT be allowed in any content."}
+				{/forminput}
+			</div>
+
+			{foreach from=$formLibertyHtmlPurifierFeatures key=item item=output}
+				<div class="row">
+					{formlabel label=`$output.label` for=$item}
+					{forminput}
+						{html_checkboxes name="$item" values="y" checked=$gBitSystem->getConfig($item, $output.default) labels=false id=$item}
+						{formhelp note=`$output.note` page=`$output.page`}
+					{/forminput}
+				</div>
+			{/foreach}
+		{/legend}
+
 	{/legend}
 
 	{legend legend="Captcha Settings"}