Content: Updated purifier and content caching

- Updated page content cache to use app version in cache key
- Moved purifier cache into framework to better work with existing
  expected folders.
- Added app version check to purifier so that it will reset its own
  cache on app version change.

app/Entities/Tools/PageContent.php

@@ -2,6 +2,7 @@
 
 namespace BookStack\Entities\Tools;
 
+use BookStack\App\AppVersion;
 use BookStack\Entities\Models\Page;
 use BookStack\Entities\Queries\PageQueries;
 use BookStack\Entities\Tools\Markdown\MarkdownToHtml;
@@ -321,13 +322,12 @@ class PageContent
         $cacheKey = $this->getContentCacheKey($doc->getBodyInnerHtml());
         $cached = cache()->get($cacheKey, null);
         if ($cached !== null) {
-//            return $cached;
+            return $cached;
         }
 
         $filterConfig = HtmlContentFilterConfig::fromConfigString(config('app.content_filtering'));
         $filter = new HtmlContentFilter($filterConfig);
         $filtered = $filter->filterDocument($doc);
-//        $filtered = $doc->getBodyInnerHtml();
 
         $cacheTime = 86400 * 7; // 1 week
         cache()->put($cacheKey, $filtered, $cacheTime);
@@ -340,7 +340,8 @@ class PageContent
         $contentHash = md5($html);
         $contentId = $this->page->id;
         $contentTime = $this->page->updated_at->timestamp;
-        return "page-content-cache::{$contentId}::{$contentTime}::{$contentHash}";
+        $appVersion = AppVersion::get();
+        return "page-content-cache::{$appVersion}::{$contentId}::{$contentTime}::{$contentHash}";
     }
 
     /**

app/Util/ConfiguredHtmlPurifier.php

@@ -2,19 +2,29 @@
 
 namespace BookStack\Util;
 
+use BookStack\App\AppVersion;
 use HTMLPurifier;
 use HTMLPurifier_Config;
+use HTMLPurifier_DefinitionCache_Serializer;
 use HTMLPurifier_HTML5Config;
 use HTMLPurifier_HTMLDefinition;
 
+/**
+ * Provides a configured HTML Purifier instance.
+ * https://github.com/ezyang/htmlpurifier
+ * Also uses this to extend support to HTML5 elements:
+ * https://github.com/xemlock/htmlpurifier-html5
+ */
 class ConfiguredHtmlPurifier
 {
     protected HTMLPurifier $purifier;
+    protected static bool $cachedChecked = false;
 
     public function __construct()
     {
         $config = HTMLPurifier_HTML5Config::createDefault();
         $this->setConfig($config);
+        $this->resetCacheIfNeeded($config);
 
         $htmlDef = $config->getDefinition('HTML', true, true);
         if ($htmlDef instanceof HTMLPurifier_HTMLDefinition) {
@@ -24,9 +34,29 @@ class ConfiguredHtmlPurifier
         $this->purifier = new HTMLPurifier($config);
     }
 
+    protected function resetCacheIfNeeded(HTMLPurifier_Config $config): void
+    {
+        if (self::$cachedChecked) {
+            return;
+        }
+
+        $cachedForVersion = cache('htmlpurifier::cache-version');
+        $appVersion = AppVersion::get();
+        if ($cachedForVersion !== $appVersion) {
+            foreach (['HTML', 'CSS', 'URI'] as $name) {
+                $cache = new HTMLPurifier_DefinitionCache_Serializer($name);
+                $cache->flush($config);
+            }
+            cache()->set('htmlpurifier::cache-version', $appVersion);
+        }
+
+        self::$cachedChecked = true;
+    }
+
     protected function setConfig(HTMLPurifier_Config $config): void
     {
-        $config->set('Cache.SerializerPath', storage_path('purifier'));
+        $config->set('Cache.SerializerPath', storage_path('framework/purifier'));
+        $config->set('Core.AllowHostnameUnderscore', true);
         $config->set('CSS.AllowTricky', true);
         $config->set('HTML.SafeIframe', true);
         $config->set('Attr.EnableID', true);
@@ -44,7 +74,7 @@ class ConfiguredHtmlPurifier
             'file' => true,
         ]);
 
-         $config->set('Cache.DefinitionImpl', null); // Disable cache during testing
+         // $config->set('Cache.DefinitionImpl', null); // Disable cache during testing
     }
 
     public function configureDefinition(HTMLPurifier_HTMLDefinition $definition): void