MDL-25826 integrate HTMLPurifier 4.3.0 and improve performance

The new HTMLPurifier finally caches the schema properly eliminating both extra CPU cycles and disk writes. The repeated dir exists tests might cause problems on NFS shares.

diff --git a/lib/simpletest/testpurifier.php b/lib/simpletest/testpurifier.php
new file mode 100644 (file)
index 0000000..e69de29

diff --git a/lib/thirdpartylibs.xml b/lib/thirdpartylibs.xml
index eb8ec85..94a349e 100644 (file)
--- a/lib/thirdpartylibs.xml
+++ b/lib/thirdpartylibs.xml
@@ -67,7 +67,7 @@
     <location>htmlpurifier</location>
     <name>HTML Purifier</name>
     <license>LGPL</license>
-    <version>4.2.0</version>
+    <version>4.3.0</version>
     <licenseversion>2.1+</licenseversion>
   </library>
   <library>

diff --git a/lib/weblib.php b/lib/weblib.php
index cb65a20..9e1b939 100644 (file)
--- a/lib/weblib.php
+++ b/lib/weblib.php
@@ -1509,20 +1509,21 @@ function clean_text($text, $format = FORMAT_HTML, $options = array()) {
 function purify_html($text, $options = array()) {
     global $CFG;
 
-    // this can not be done only once because we sometimes need to reset the cache
-    $cachedir = $CFG->dataroot.'/cache/htmlpurifier';
-    check_dir_exists($cachedir);
-
     $type = !empty($options['allowid']) ? 'allowid' : 'normal';
     static $purifiers = array();
     if (empty($purifiers[$type])) {
+
+        // make sure the serializer dir exists, it should be fine if it disappears later during cache reset
+        $cachedir = $CFG->dataroot.'/cache/htmlpurifier';
+        check_dir_exists($cachedir);
+
         require_once $CFG->libdir.'/htmlpurifier/HTMLPurifier.safe-includes.php';
         $config = HTMLPurifier_Config::createDefault();
 
         $config->set('HTML.DefinitionID', 'moodlehtml');
-        $config->set('HTML.DefinitionRev', 1);
+        $config->set('HTML.DefinitionRev', 2);
         $config->set('Cache.SerializerPath', $cachedir);
-        //$config->set('Cache.SerializerPermission', $CFG->directorypermissions); // it would be nice to get this upstream
+        $config->set('Cache.SerializerPermissions', $CFG->directorypermissions);
         $config->set('Core.NormalizeNewlines', false);
         $config->set('Core.ConvertDocumentToFragment', true);
         $config->set('Core.Encoding', 'UTF-8');
@@ -1540,12 +1541,13 @@ function purify_html($text, $options = array()) {
             $config->set('Attr.EnableID', true);
         }
 
-        $def = $config->getHTMLDefinition(true);
-        $def->addElement('nolink', 'Block', 'Flow', array());                       // skip our filters inside
-        $def->addElement('tex', 'Inline', 'Inline', array());                       // tex syntax, equivalent to $$xx$$
-        $def->addElement('algebra', 'Inline', 'Inline', array());                   // algebra syntax, equivalent to @@xx@@
-        $def->addElement('lang', 'Block', 'Flow', array(), array('lang'=>'CDATA')); // old anf future style multilang - only our hacked lang attribute
-        $def->addAttribute('span', 'xxxlang', 'CDATA');                             // current problematic multilang
+        if ($def = $config->maybeGetRawHTMLDefinition()) {
+            $def->addElement('nolink', 'Block', 'Flow', array());                       // skip our filters inside
+            $def->addElement('tex', 'Inline', 'Inline', array());                       // tex syntax, equivalent to $$xx$$
+            $def->addElement('algebra', 'Inline', 'Inline', array());                   // algebra syntax, equivalent to @@xx@@
+            $def->addElement('lang', 'Block', 'Flow', array(), array('lang'=>'CDATA')); // old and future style multilang - only our hacked lang attribute
+            $def->addAttribute('span', 'xxxlang', 'CDATA');                             // current problematic multilang
+        }
 
         $purifier = new HTMLPurifier($config);
         $purifiers[$type] = $purifier;