MD Exports: Added HTML description conversion

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

diff --git a/app/Entities/Tools/ExportFormatter.php b/app/Entities/Tools/ExportFormatter.php
index 6779797d10b45fe336c43ec1d7521c181acb9ef6..0af68b8db3cfcd3f4f9957c0a949259a2144593b 100644 (file)
--- a/app/Entities/Tools/ExportFormatter.php
+++ b/app/Entities/Tools/ExportFormatter.php
@@ -8,9 +8,8 @@ use BookStack\Entities\Models\Page;
 use BookStack\Entities\Tools\Markdown\HtmlToMarkdown;
 use BookStack\Uploads\ImageService;
 use BookStack\Util\CspService;
-use DOMDocument;
+use BookStack\Util\HtmlDocument;
 use DOMElement;
-use DOMXPath;
 use Exception;
 use Throwable;
 
@@ -151,45 +150,36 @@ class ExportFormatter
     protected function htmlToPdf(string $html): string
     {
         $html = $this->containHtml($html);
-        $html = $this->replaceIframesWithLinks($html);
-        $html = $this->openDetailElements($html);
+        $doc = new HtmlDocument();
+        $doc->loadCompleteHtml($html);
 
-        return $this->pdfGenerator->fromHtml($html);
+        $this->replaceIframesWithLinks($doc);
+        $this->openDetailElements($doc);
+        $cleanedHtml = $doc->getHtml();
+
+        return $this->pdfGenerator->fromHtml($cleanedHtml);
     }
 
     /**
      * Within the given HTML content, Open any detail blocks.
      */
-    protected function openDetailElements(string $html): string
+    protected function openDetailElements(HtmlDocument $doc): void
     {
-        libxml_use_internal_errors(true);
-
-        $doc = new DOMDocument();
-        $doc->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'));
-        $xPath = new DOMXPath($doc);
-
-        $details = $xPath->query('//details');
+        $details = $doc->queryXPath('//details');
         /** @var DOMElement $detail */
         foreach ($details as $detail) {
             $detail->setAttribute('open', 'open');
         }
-
-        return $doc->saveHTML();
     }
 
     /**
-     * Within the given HTML content, replace any iframe elements
+     * Within the given HTML document, replace any iframe elements
      * with anchor links within paragraph blocks.
      */
-    protected function replaceIframesWithLinks(string $html): string
+    protected function replaceIframesWithLinks(HtmlDocument $doc): void
     {
-        libxml_use_internal_errors(true);
-
-        $doc = new DOMDocument();
-        $doc->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'));
-        $xPath = new DOMXPath($doc);
+        $iframes = $doc->queryXPath('//iframe');
 
-        $iframes = $xPath->query('//iframe');
         /** @var DOMElement $iframe */
         foreach ($iframes as $iframe) {
             $link = $iframe->getAttribute('src');
@@ -203,8 +193,6 @@ class ExportFormatter
             $paragraph->appendChild($anchor);
             $iframe->parentNode->replaceChild($paragraph, $iframe);
         }
-
-        return $doc->saveHTML();
     }
 
     /**
@@ -222,7 +210,7 @@ class ExportFormatter
             foreach ($imageTagsOutput[0] as $index => $imgMatch) {
                 $oldImgTagString = $imgMatch;
                 $srcString = $imageTagsOutput[2][$index];
-                $imageEncoded = $this->imageService->imageUriToBase64($srcString);
+                $imageEncoded = $this->imageService->imageUrlToBase64($srcString);
                 if ($imageEncoded === null) {
                     $imageEncoded = $srcString;
                 }
@@ -254,17 +242,20 @@ class ExportFormatter
      * Converts the page contents into simple plain text.
      * This method filters any bad looking content to provide a nice final output.
      */
-    public function pageToPlainText(Page $page): string
+    public function pageToPlainText(Page $page, bool $pageRendered = false, bool $fromParent = false): string
     {
-        $html = (new PageContent($page))->render();
-        $text = strip_tags($html);
+        $html = $pageRendered ? $page->html : (new PageContent($page))->render();
+        // Add proceeding spaces before tags so spaces remain between
+        // text within elements after stripping tags.
+        $html = str_replace('<', " <", $html);
+        $text = trim(strip_tags($html));
         // Replace multiple spaces with single spaces
-        $text = preg_replace('/\ {2,}/', ' ', $text);
+        $text = preg_replace('/ {2,}/', ' ', $text);
         // Reduce multiple horrid whitespace characters.
         $text = preg_replace('/(\x0A|\xA0|\x0A|\r|\n){2,}/su', "\n\n", $text);
         $text = html_entity_decode($text);
         // Add title
-        $text = $page->name . "\n\n" . $text;
+        $text = $page->name . ($fromParent ? "\n" : "\n\n") . $text;
 
         return $text;
     }
@@ -274,13 +265,15 @@ class ExportFormatter
      */
     public function chapterToPlainText(Chapter $chapter): string
     {
-        $text = $chapter->name . "\n\n";
-        $text .= $chapter->description . "\n\n";
+        $text = $chapter->name . "\n" . $chapter->description;
+        $text = trim($text) . "\n\n";
+
+        $parts = [];
         foreach ($chapter->getVisiblePages() as $page) {
-            $text .= $this->pageToPlainText($page);
+            $parts[] = $this->pageToPlainText($page, false, true);
         }
 
-        return $text;
+        return $text . implode("\n\n", $parts);
     }
 
     /**
@@ -288,17 +281,20 @@ class ExportFormatter
      */
     public function bookToPlainText(Book $book): string
     {
-        $bookTree = (new BookContents($book))->getTree(false, false);
-        $text = $book->name . "\n\n";
+        $bookTree = (new BookContents($book))->getTree(false, true);
+        $text = $book->name . "\n" . $book->description;
+        $text = rtrim($text) . "\n\n";
+
+        $parts = [];
         foreach ($bookTree as $bookChild) {
             if ($bookChild->isA('chapter')) {
-                $text .= $this->chapterToPlainText($bookChild);
+                $parts[] = $this->chapterToPlainText($bookChild);
             } else {
-                $text .= $this->pageToPlainText($bookChild);
+                $parts[] = $this->pageToPlainText($bookChild, true, true);
             }
         }
 
-        return $text;
+        return $text . implode("\n\n", $parts);
     }
 
     /**
@@ -319,7 +315,12 @@ class ExportFormatter
     public function chapterToMarkdown(Chapter $chapter): string
     {
         $text = '# ' . $chapter->name . "\n\n";
-        $text .= $chapter->description . "\n\n";
+
+        $description = (new HtmlToMarkdown($chapter->descriptionHtml()))->convert();
+        if ($description) {
+            $text .= $description . "\n\n";
+        }
+
         foreach ($chapter->pages as $page) {
             $text .= $this->pageToMarkdown($page) . "\n\n";
         }
@@ -334,6 +335,12 @@ class ExportFormatter
     {
         $bookTree = (new BookContents($book))->getTree(false, true);
         $text = '# ' . $book->name . "\n\n";
+
+        $description = (new HtmlToMarkdown($book->descriptionHtml()))->convert();
+        if ($description) {
+            $text .= $description . "\n\n";
+        }
+
         foreach ($bookTree as $bookChild) {
             if ($bookChild instanceof Chapter) {
                 $text .= $this->chapterToMarkdown($bookChild) . "\n\n";