Add unit test for ip addess searching

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

diff --git a/app/Entities/Tools/PageContent.php b/app/Entities/Tools/PageContent.php
index b1323bc68ac0f02b6f3f799a59a965c0f7953fb9..b95131fce2e25561c22131d3d81e3dc4d30a8aac 100644 (file)
--- a/app/Entities/Tools/PageContent.php
+++ b/app/Entities/Tools/PageContent.php
@@ -12,6 +12,8 @@ use BookStack\Uploads\ImageRepo;
 use BookStack\Uploads\ImageService;
 use BookStack\Util\HtmlContentFilter;
 use DOMDocument;
+use DOMElement;
+use DOMNode;
 use DOMNodeList;
 use DOMXPath;
 use Illuminate\Support\Str;
@@ -135,6 +137,12 @@ class PageContent
             return '';
         }
 
+        // Validate that the content is not over our upload limit
+        $uploadLimitBytes = (config('app.upload_limit') * 1000000);
+        if (strlen($imageInfo['data']) > $uploadLimitBytes) {
+            return '';
+        }
+
         // Save image from data with a random name
         $imageName = 'embedded-image-' . Str::random(8) . '.' . $imageInfo['extension'];
 
@@ -150,7 +158,7 @@ class PageContent
     /**
      * Parse a base64 image URI into the data and extension.
      *
-     * @return array{extension: array, data: string}
+     * @return array{extension: string, data: string}
      */
     protected function parseBase64ImageUri(string $uri): array
     {
@@ -187,6 +195,15 @@ 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) {
@@ -222,9 +239,9 @@ class PageContent
      * A map for existing ID's should be passed in to check for current existence.
      * Returns a pair of strings in the format [old_id, new_id].
      */
-    protected function setUniqueId(\DOMNode $element, array &$idMap): array
+    protected function setUniqueId(DOMNode $element, array &$idMap): array
     {
-        if (get_class($element) !== 'DOMElement') {
+        if (!$element instanceof DOMElement) {
             return ['', ''];
         }
 
@@ -236,7 +253,7 @@ class PageContent
             return [$existingId, $existingId];
         }
 
-        // Create an unique id for the element
+        // Create a unique id for the element
         // Uses the content as a basis to ensure output is the same every time
         // the same content is passed through.
         $contentId = 'bkmrk-' . mb_substr(strtolower(preg_replace('/\s+/', '-', trim($element->nodeValue))), 0, 20);
@@ -306,7 +323,7 @@ class PageContent
      */
     protected function headerNodesToLevelList(DOMNodeList $nodeList): array
     {
-        $tree = collect($nodeList)->map(function ($header) {
+        $tree = collect($nodeList)->map(function (DOMElement $header) {
             $text = trim(str_replace("\xc2\xa0", '', $header->nodeValue));
             $text = mb_substr($text, 0, 100);
 
@@ -384,7 +401,7 @@ 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.