Reverted shift change to old migration

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

diff --git a/app/Entities/Tools/PageContent.php b/app/Entities/Tools/PageContent.php
index 2c95862657b3c5741a8fbdc75c8dc363836e6c36..9f4ac2893f7fe0857acbf3ea476035f39b887cec 100644 (file)
--- a/app/Entities/Tools/PageContent.php
+++ b/app/Entities/Tools/PageContent.php
@@ -1,17 +1,20 @@
-<?php namespace BookStack\Entities\Tools;
+<?php
+
+namespace BookStack\Entities\Tools;

 
 

-use BookStack\Auth\Permissions\PermissionService;

 use BookStack\Entities\Models\Page;
 use BookStack\Entities\Models\Page;

+use BookStack\Entities\Tools\Markdown\CustomListItemRenderer;

 use BookStack\Entities\Tools\Markdown\CustomStrikeThroughExtension;
 use BookStack\Entities\Tools\Markdown\CustomStrikeThroughExtension;

+use BookStack\Exceptions\ImageUploadException;

 use BookStack\Facades\Theme;
 use BookStack\Theming\ThemeEvents;
 use BookStack\Facades\Theme;
 use BookStack\Theming\ThemeEvents;

-use BookStack\Util\HtmlContentFilter;
-use BookStack\Uploads\Image;

 use BookStack\Uploads\ImageRepo;
 use BookStack\Uploads\ImageRepo;

-use BookStack\Uploads\ImageService;
+use BookStack\Util\HtmlContentFilter;

 use DOMDocument;
 use DOMNodeList;
 use DOMXPath;
 use DOMDocument;
 use DOMNodeList;
 use DOMXPath;

+use Illuminate\Support\Str;
+use League\CommonMark\Block\Element\ListItem;

 use League\CommonMark\CommonMarkConverter;
 use League\CommonMark\Environment;
 use League\CommonMark\Extension\Table\TableExtension;
 use League\CommonMark\CommonMarkConverter;
 use League\CommonMark\Environment;
 use League\CommonMark\Extension\Table\TableExtension;


@@ -19,7 +22,6 @@ use League\CommonMark\Extension\TaskList\TaskListExtension;
 
 class PageContent
 {
 
 class PageContent
 {

-

     protected $page;
 
     /**
     protected $page;
 
     /**


@@ -35,7 +37,7 @@ class PageContent
      */
     public function setNewHTML(string $html)
     {
      */
     public function setNewHTML(string $html)
     {

-        $html = $this->saveBase64Images($this->page, $html);
+        $html = $this->extractBase64ImagesFromHtml($html);

         $this->page->html = $this->formatHtml($html);
         $this->page->text = $this->toPlainText();
         $this->page->markdown = '';
         $this->page->html = $this->formatHtml($html);
         $this->page->text = $this->toPlainText();
         $this->page->markdown = '';


@@ -46,6 +48,7 @@ class PageContent
      */
     public function setNewMarkdown(string $markdown)
     {
      */
     public function setNewMarkdown(string $markdown)
     {

+        $markdown = $this->extractBase64ImagesFromMarkdown($markdown);

         $this->page->markdown = $markdown;
         $html = $this->markdownToHtml($markdown);
         $this->page->html = $this->formatHtml($html);
         $this->page->markdown = $markdown;
         $html = $this->markdownToHtml($markdown);
         $this->page->html = $this->formatHtml($html);


@@ -63,51 +66,49 @@ class PageContent
         $environment->addExtension(new CustomStrikeThroughExtension());
         $environment = Theme::dispatch(ThemeEvents::COMMONMARK_ENVIRONMENT_CONFIGURE, $environment) ?? $environment;
         $converter = new CommonMarkConverter([], $environment);
         $environment->addExtension(new CustomStrikeThroughExtension());
         $environment = Theme::dispatch(ThemeEvents::COMMONMARK_ENVIRONMENT_CONFIGURE, $environment) ?? $environment;
         $converter = new CommonMarkConverter([], $environment);

+
+        $environment->addBlockRenderer(ListItem::class, new CustomListItemRenderer(), 10);
+

         return $converter->convertToHtml($markdown);
     }
 
     /**
         return $converter->convertToHtml($markdown);
     }
 
     /**

-     * Convert all base64 image data to saved images
+     * Convert all base64 image data to saved images.

      */
      */

-    public function saveBase64Images(Page $page, string $htmlText): string
+    protected function extractBase64ImagesFromHtml(string $htmlText): string

     {
     {

-        if ($htmlText == '') {
+        if (empty($htmlText) || strpos($htmlText, 'data:image') === false) {

             return $htmlText;
         }
 
             return $htmlText;
         }
 

-        libxml_use_internal_errors(true);
-        $doc = new DOMDocument();
-        $doc->loadHTML(mb_convert_encoding($htmlText, 'HTML-ENTITIES', 'UTF-8'));
+        $doc = $this->loadDocumentFromHtml($htmlText);

         $container = $doc->documentElement;
         $body = $container->childNodes->item(0);
         $childNodes = $body->childNodes;
         $xPath = new DOMXPath($doc);
         $container = $doc->documentElement;
         $body = $container->childNodes->item(0);
         $childNodes = $body->childNodes;
         $xPath = new DOMXPath($doc);

+        $imageRepo = app()->make(ImageRepo::class);

 
         // Get all img elements with image data blobs
         $imageNodes = $xPath->query('//img[contains(@src, \'data:image\')]');
 
         // Get all img elements with image data blobs
         $imageNodes = $xPath->query('//img[contains(@src, \'data:image\')]');

-        foreach($imageNodes as $imageNode) {
+        foreach ($imageNodes as $imageNode) {

             $imageSrc = $imageNode->getAttribute('src');
             $imageSrc = $imageNode->getAttribute('src');

+            [$dataDefinition, $base64ImageData] = explode(',', $imageSrc, 2);
+            $extension = strtolower(preg_split('/[\/;]/', $dataDefinition)[1] ?? 'png');

 
 

-            # Parse base64 data
-            $result = preg_match('"data:image/[a-zA-Z]*(;base64,[a-zA-Z0-9+/\\= ]*)"', $imageSrc, $matches);
-
-            if($result === 1) {
-                $base64ImageData = $matches[1];
-
-                $image = new Image();
-                $imageService = app()->make(ImageService::class);
-                $permissionService = app(PermissionService::class);
-                $imageRepo = new ImageRepo(new Image(), $imageService, $permissionService, $page);
+            // Validate extension
+            if (!$imageRepo->imageExtensionSupported($extension)) {
+                $imageNode->setAttribute('src', '');
+                continue;
+            }

 
 

-                # Use existing saveDrawing method used for Drawio diagrams
-                $image = $imageRepo->saveDrawing($base64ImageData, $page->id);
-                
-                // Create a new img element with the saved image URI
-                $newNode = $doc->createElement('img');
-                $newNode->setAttribute('src', $image->path);
+            // Save image from data with a random name
+            $imageName = 'embedded-image-' . Str::random(8) . '.' . $extension;

 
 

-                // Replace the old img element
-                $imageNode->parentNode->replaceChild($newNode, $imageNode);
+            try {
+                $image = $imageRepo->saveNewFromData($imageName, base64_decode($base64ImageData), 'gallery', $this->page->id);
+                $imageNode->setAttribute('src', $image->url);
+            } catch (ImageUploadException $exception) {
+                $imageNode->setAttribute('src', '');

             }
         }
 
             }
         }
 


@@ -120,19 +121,49 @@ class PageContent
         return $html;
     }
 
         return $html;
     }
 

+    /**
+     * Convert all inline base64 content to uploaded image files.
+     */
+    protected function extractBase64ImagesFromMarkdown(string $markdown)
+    {
+        $imageRepo = app()->make(ImageRepo::class);
+        $matches = [];
+        preg_match_all('/!\[.*?]\(.*?(data:image\/.*?)[)"\s]/', $markdown, $matches);
+
+        foreach ($matches[1] as $base64Match) {
+            [$dataDefinition, $base64ImageData] = explode(',', $base64Match, 2);
+            $extension = strtolower(preg_split('/[\/;]/', $dataDefinition)[1] ?? 'png');
+
+            // Validate extension
+            if (!$imageRepo->imageExtensionSupported($extension)) {
+                $markdown = str_replace($base64Match, '', $markdown);
+                continue;
+            }
+
+            // Save image from data with a random name
+            $imageName = 'embedded-image-' . Str::random(8) . '.' . $extension;
+
+            try {
+                $image = $imageRepo->saveNewFromData($imageName, base64_decode($base64ImageData), 'gallery', $this->page->id);
+                $markdown = str_replace($base64Match, $image->url, $markdown);
+            } catch (ImageUploadException $exception) {
+                $markdown = str_replace($base64Match, '', $markdown);
+            }
+        }
+
+        return $markdown;
+    }
+

     /**
      * Formats a page's html to be tagged correctly within the system.
      */
     protected function formatHtml(string $htmlText): string
     {
     /**
      * Formats a page's html to be tagged correctly within the system.
      */
     protected function formatHtml(string $htmlText): string
     {

-        if ($htmlText == '') {
+        if (empty($htmlText)) {

             return $htmlText;
         }
 
             return $htmlText;
         }
 

-        libxml_use_internal_errors(true);
-        $doc = new DOMDocument();
-        $doc->loadHTML(mb_convert_encoding($htmlText, 'HTML-ENTITIES', 'UTF-8'));
-
+        $doc = $this->loadDocumentFromHtml($htmlText);

         $container = $doc->documentElement;
         $body = $container->childNodes->item(0);
         $childNodes = $body->childNodes;
         $container = $doc->documentElement;
         $body = $container->childNodes->item(0);
         $childNodes = $body->childNodes;


@@ -171,7 +202,7 @@ class PageContent
     protected function updateLinks(DOMXPath $xpath, string $old, string $new)
     {
         $old = str_replace('"', '', $old);
     protected function updateLinks(DOMXPath $xpath, string $old, string $new)
     {
         $old = str_replace('"', '', $old);

-        $matchingLinks = $xpath->query('//body//*//*[@href="'.$old.'"]');
+        $matchingLinks = $xpath->query('//body//*//*[@href="' . $old . '"]');

         foreach ($matchingLinks as $domElem) {
             $domElem->setAttribute('href', $new);
         }
         foreach ($matchingLinks as $domElem) {
             $domElem->setAttribute('href', $new);
         }


@@ -180,7 +211,7 @@ class PageContent
     /**
      * Set a unique id on the given DOMElement.
      * A map for existing ID's should be passed in to check for current existence.
     /**
      * Set a unique id on the given DOMElement.
      * 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]
+     * 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
     {


@@ -192,6 +223,7 @@ class PageContent
         $existingId = $element->getAttribute('id');
         if (strpos($existingId, 'bkmrk') === 0 && !isset($idMap[$existingId])) {
             $idMap[$existingId] = true;
         $existingId = $element->getAttribute('id');
         if (strpos($existingId, 'bkmrk') === 0 && !isset($idMap[$existingId])) {
             $idMap[$existingId] = true;

+

             return [$existingId, $existingId];
         }
 
             return [$existingId, $existingId];
         }
 


@@ -209,6 +241,7 @@ class PageContent
 
         $element->setAttribute('id', $newId);
         $idMap[$newId] = true;
 
         $element->setAttribute('id', $newId);
         $idMap[$newId] = true;

+

         return [$existingId, $newId];
     }
 
         return [$existingId, $newId];
     }
 


@@ -218,15 +251,16 @@ class PageContent
     protected function toPlainText(): string
     {
         $html = $this->render(true);
     protected function toPlainText(): string
     {
         $html = $this->render(true);

+

         return html_entity_decode(strip_tags($html));
     }
 
     /**
         return html_entity_decode(strip_tags($html));
     }
 
     /**

-     * Render the page for viewing
+     * Render the page for viewing.

      */
      */

-    public function render(bool $blankIncludes = false) : string
+    public function render(bool $blankIncludes = false): string

     {
     {

-        $content = $this->page->html;
+        $content = $this->page->html ?? '';

 
         if (!config('app.allow_content_scripts')) {
             $content = HtmlContentFilter::removeScripts($content);
 
         if (!config('app.allow_content_scripts')) {
             $content = HtmlContentFilter::removeScripts($content);


@@ -242,7 +276,7 @@ class PageContent
     }
 
     /**
     }
 
     /**

-     * Parse the headers on the page to get a navigation menu
+     * Parse the headers on the page to get a navigation menu.

      */
     public function getNavigation(string $htmlContent): array
     {
      */
     public function getNavigation(string $htmlContent): array
     {


@@ -250,11 +284,9 @@ class PageContent
             return [];
         }
 
             return [];
         }
 

-        libxml_use_internal_errors(true);
-        $doc = new DOMDocument();
-        $doc->loadHTML(mb_convert_encoding($htmlContent, 'HTML-ENTITIES', 'UTF-8'));
+        $doc = $this->loadDocumentFromHtml($htmlContent);

         $xPath = new DOMXPath($doc);
         $xPath = new DOMXPath($doc);

-        $headers = $xPath->query("//h1|//h2|//h3|//h4|//h5|//h6");
+        $headers = $xPath->query('//h1|//h2|//h3|//h4|//h5|//h6');

 
         return $headers ? $this->headerNodesToLevelList($headers) : [];
     }
 
         return $headers ? $this->headerNodesToLevelList($headers) : [];
     }


@@ -271,9 +303,9 @@ class PageContent
 
             return [
                 'nodeName' => strtolower($header->nodeName),
 
             return [
                 'nodeName' => strtolower($header->nodeName),

-                'level' => intval(str_replace('h', '', $header->nodeName)),
-                'link' => '#' . $header->getAttribute('id'),
-                'text' => $text,
+                'level'    => intval(str_replace('h', '', $header->nodeName)),
+                'link'     => '#' . $header->getAttribute('id'),
+                'text'     => $text,

             ];
         })->filter(function ($header) {
             return mb_strlen($header['text']) > 0;
             ];
         })->filter(function ($header) {
             return mb_strlen($header['text']) > 0;


@@ -283,6 +315,7 @@ class PageContent
         $levelChange = ($tree->pluck('level')->min() - 1);
         $tree = $tree->map(function ($header) use ($levelChange) {
             $header['level'] -= ($levelChange);
         $levelChange = ($tree->pluck('level')->min() - 1);
         $tree = $tree->map(function ($header) use ($levelChange) {
             $header['level'] -= ($levelChange);

+

             return $header;
         });
 
             return $header;
         });
 


@@ -292,7 +325,7 @@ class PageContent
     /**
      * Remove any page include tags within the given HTML.
      */
     /**
      * Remove any page include tags within the given HTML.
      */

-    protected function blankPageIncludes(string $html) : string
+    protected function blankPageIncludes(string $html): string

     {
         return preg_replace("/{{@\s?([0-9].*?)}}/", '', $html);
     }
     {
         return preg_replace("/{{@\s?([0-9].*?)}}/", '', $html);
     }


@@ -300,7 +333,7 @@ class PageContent
     /**
      * Parse any include tags "{{@<page_id>#section}}" to be part of the page.
      */
     /**
      * Parse any include tags "{{@<page_id>#section}}" to be part of the page.
      */

-    protected function parsePageIncludes(string $html) : string
+    protected function parsePageIncludes(string $html): string

     {
         $matches = [];
         preg_match_all("/{{@\s?([0-9].*?)}}/", $html, $matches);
     {
         $matches = [];
         preg_match_all("/{{@\s?([0-9].*?)}}/", $html, $matches);


@@ -316,6 +349,7 @@ class PageContent
             }
 
             // Find page and skip this if page not found
             }
 
             // Find page and skip this if page not found

+            /** @var ?Page $matchedPage */

             $matchedPage = Page::visible()->find($pageId);
             if ($matchedPage === null) {
                 $html = str_replace($fullMatch, '', $html);
             $matchedPage = Page::visible()->find($pageId);
             if ($matchedPage === null) {
                 $html = str_replace($fullMatch, '', $html);


@@ -336,16 +370,13 @@ class PageContent
         return $html;
     }
 
         return $html;
     }
 

-

     /**
      * Fetch the content from a specific section of the given page.
      */
     protected function fetchSectionOfPage(Page $page, string $sectionId): string
     {
         $topLevelTags = ['table', 'ul', 'ol'];
     /**
      * Fetch the content from a specific section of the given page.
      */
     protected function fetchSectionOfPage(Page $page, string $sectionId): string
     {
         $topLevelTags = ['table', 'ul', 'ol'];

-        $doc = new DOMDocument();
-        libxml_use_internal_errors(true);
-        $doc->loadHTML(mb_convert_encoding('<body>'.$page->html.'</body>', 'HTML-ENTITIES', 'UTF-8'));
+        $doc = $this->loadDocumentFromHtml($page->html);

 
         // Search included content for the id given and blank out if not exists.
         $matchingElem = $doc->getElementById($sectionId);
 
         // Search included content for the id given and blank out if not exists.
         $matchingElem = $doc->getElementById($sectionId);


@@ -368,4 +399,17 @@ class PageContent
 
         return $innerContent;
     }
 
         return $innerContent;
     }

+
+    /**
+     * Create and load a DOMDocument from the given html content.
+     */
+    protected function loadDocumentFromHtml(string $html): DOMDocument
+    {
+        libxml_use_internal_errors(true);
+        $doc = new DOMDocument();
+        $html = '<body>' . $html . '</body>';
+        $doc->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'));
+
+        return $doc;
+    }

 }
 }