MDL-36297 caching: added for HTML purified text

lang/en/cache.php

@@ -36,6 +36,7 @@ $string['cacheadmin'] = 'Cache administration';
 $string['cacheconfig'] = 'Configuration';
 $string['cachedef_databasemeta'] = 'Database meta information';
 $string['cachedef_eventinvalidation'] = 'Event invalidation';
+$string['cachedef_htmlpurifier'] = 'HTML Purifier - cleaned content';
 $string['cachedef_locking'] = 'Locking';
 $string['cachedef_questiondata'] = 'Question definitions';
 $string['cachedef_string'] = 'Language string cache';

lib/db/caches.php

@@ -75,4 +75,12 @@ $definitions = array(
         'datasource' => 'question_finder',
         'datasourcefile' => 'question/engine/bank.php',
     ),
+
+    // HTML Purifier cache
+    // This caches the html purifier cleaned text. This is done because the text is usually cleaned once for every user
+    // and context combo. Text caching handles caching for the combonation, this cache is responsible for caching the
+    // cleaned text which is shareable.
+    'htmlpurifier' => array(
+        'mode' => cache_store::MODE_APPLICATION,
+    )
 );

lib/weblib.php

@@ -1578,8 +1578,21 @@ function is_purify_html_necessary($text) {
 function purify_html($text, $options = array()) {
     global $CFG;
 
-    $type = !empty($options['allowid']) ? 'allowid' : 'normal';
     static $purifiers = array();
+    static $caches = array();
+
+    $type = !empty($options['allowid']) ? 'allowid' : 'normal';
+
+    if (!array_key_exists($type, $caches)) {
+        $caches[$type] = cache::make('core', 'htmlpurifier', array('type' => $type));
+    }
+    $cache = $caches[$type];
+
+    $filteredtext = $cache->get($text);
+    if ($filteredtext !== false) {
+        return $filteredtext;
+    }
+
     if (empty($purifiers[$type])) {
 
         // make sure the serializer dir exists, it should be fine if it disappears later during cache reset
@@ -1627,15 +1640,17 @@ function purify_html($text, $options = array()) {
 
     $multilang = (strpos($text, 'class="multilang"') !== false);
 
+    $filteredtext = $text;
     if ($multilang) {
-        $text = preg_replace('/<span(\s+lang="([a-zA-Z0-9_-]+)"|\s+class="multilang"){2}\s*>/', '<span xxxlang="${2}">', $text);
+        $filteredtext = preg_replace('/<span(\s+lang="([a-zA-Z0-9_-]+)"|\s+class="multilang"){2}\s*>/', '<span xxxlang="${2}">', $filteredtext);
     }
-    $text = $purifier->purify($text);
+    $filteredtext = $purifier->purify($filteredtext);
     if ($multilang) {
-        $text = preg_replace('/<span xxxlang="([a-zA-Z0-9_-]+)">/', '<span lang="${1}" class="multilang">', $text);
+        $filteredtext = preg_replace('/<span xxxlang="([a-zA-Z0-9_-]+)">/', '<span lang="${1}" class="multilang">', $filteredtext);
     }
+    $cache->set($text, $filteredtext);
 
-    return $text;
+    return $filteredtext;
 }
 
 /**