feat(PageContent): set unique ids on nested headers

[bookstack] / app / Entities / Tools / PageContent.php

diff --git a/app/Entities/Tools/PageContent.php b/app/Entities/Tools/PageContent.php
index 724230a3d6a2de8bf29b3ca62a5056f880ae665f..c8204a18186893c874f03dcdb65e820e03c3d638 100644 (file)
--- a/app/Entities/Tools/PageContent.php
+++ b/app/Entities/Tools/PageContent.php
@@ -9,6 +9,7 @@ use BookStack\Exceptions\ImageUploadException;
 use BookStack\Facades\Theme;
 use BookStack\Theming\ThemeEvents;
 use BookStack\Uploads\ImageRepo;
+use BookStack\Uploads\ImageService;
 use BookStack\Util\HtmlContentFilter;
 use DOMDocument;
 use DOMNodeList;
@@ -130,7 +131,13 @@ class PageContent
         $imageInfo = $this->parseBase64ImageUri($uri);
 
         // Validate extension and content
-        if (empty($imageInfo['data']) || !$imageRepo->imageExtensionSupported($imageInfo['extension'])) {
+        if (empty($imageInfo['data']) || !ImageService::isExtensionSupported($imageInfo['extension'])) {
+            return '';
+        }
+
+        // Validate that the content is not over our upload limit
+        $uploadLimitBytes = (config('app.upload_limit') * 1000000);
+        if (strlen($imageInfo['data']) > $uploadLimitBytes) {
             return '';
         }
 
@@ -148,15 +155,17 @@ class PageContent
 
     /**
      * Parse a base64 image URI into the data and extension.
+     *
      * @return array{extension: array, data: string}
      */
     protected function parseBase64ImageUri(string $uri): array
     {
         [$dataDefinition, $base64ImageData] = explode(',', $uri, 2);
         $extension = strtolower(preg_split('/[\/;]/', $dataDefinition)[1] ?? '');
+
         return [
             'extension' => $extension,
-            'data' => base64_decode($base64ImageData) ?: '',
+            'data'      => base64_decode($base64ImageData) ?: '',
         ];
     }
 
@@ -184,6 +193,15 @@ class PageContent
             }
         }
 
+        // Set ids on nested header nodes
+        $nestedHeaders = $xPath->query('//body//*//h1|//body//*//h2|//body//*//h3|//body//*//h4|//body//*//h5|//body//*//h6');
+        foreach ($nestedHeaders as $nestedHeader) {
+            [$oldId, $newId] = $this->setUniqueId($nestedHeader, $idMap);
+            if ($newId && $newId !== $oldId) {
+                $this->updateLinks($xPath, '#' . $oldId, '#' . $newId);
+            }
+        }
+
         // Ensure no duplicate ids within child items
         $idElems = $xPath->query('//body//*//*[@id]');
         foreach ($idElems as $domElem) {
@@ -381,7 +399,7 @@ class PageContent
      */
     protected function fetchSectionOfPage(Page $page, string $sectionId): string
     {
-        $topLevelTags = ['table', 'ul', 'ol'];
+        $topLevelTags = ['table', 'ul', 'ol', 'pre'];
         $doc = $this->loadDocumentFromHtml($page->html);
 
         // Search included content for the id given and blank out if not exists.