BookStack Code Mirror - bookstack/blob - app/Entities/Tools/ExportFormatter.php

   1 <?php namespace BookStack\Entities\Tools;
   2
   3 use BookStack\Entities\Models\Book;
   4 use BookStack\Entities\Models\Chapter;
   5 use BookStack\Entities\Models\Page;
   6 use BookStack\Uploads\ImageService;
   7 use DomPDF;
   8 use Exception;
   9 use SnappyPDF;
  10 use Throwable;
  11
  12 class ExportFormatter
  13 {
  14
  15     protected $imageService;
  16
  17     /**
  18      * ExportService constructor.
  19      */
  20     public function __construct(ImageService $imageService)
  21     {
  22         $this->imageService = $imageService;
  23     }
  24
  25     /**
  26      * Convert a page to a self-contained HTML file.
  27      * Includes required CSS & image content. Images are base64 encoded into the HTML.
  28      * @throws Throwable
  29      */
  30     public function pageToContainedHtml(Page $page)
  31     {
  32         $page->html = (new PageContent($page))->render();
  33         $pageHtml = view('pages.export', [
  34             'page' => $page,
  35             'format' => 'html',
  36         ])->render();
  37         return $this->containHtml($pageHtml);
  38     }
  39
  40     /**
  41      * Convert a chapter to a self-contained HTML file.
  42      * @throws Throwable
  43      */
  44     public function chapterToContainedHtml(Chapter $chapter)
  45     {
  46         $pages = $chapter->getVisiblePages();
  47         $pages->each(function ($page) {
  48             $page->html = (new PageContent($page))->render();
  49         });
  50         $html = view('chapters.export', [
  51             'chapter' => $chapter,
  52             'pages' => $pages,
  53             'format' => 'html',
  54         ])->render();
  55         return $this->containHtml($html);
  56     }
  57
  58     /**
  59      * Convert a book to a self-contained HTML file.
  60      * @throws Throwable
  61      */
  62     public function bookToContainedHtml(Book $book)
  63     {
  64         $bookTree = (new BookContents($book))->getTree(false, true);
  65         $html = view('books.export', [
  66             'book' => $book,
  67             'bookChildren' => $bookTree,
  68             'format' => 'html',
  69         ])->render();
  70         return $this->containHtml($html);
  71     }
  72
  73     /**
  74      * Convert a page to a PDF file.
  75      * @throws Throwable
  76      */
  77     public function pageToPdf(Page $page)
  78     {
  79         $page->html = (new PageContent($page))->render();
  80         $html = view('pages.export', [
  81             'page' => $page,
  82             'format' => 'pdf',
  83         ])->render();
  84         return $this->htmlToPdf($html);
  85     }
  86
  87     /**
  88      * Convert a chapter to a PDF file.
  89      * @throws Throwable
  90      */
  91     public function chapterToPdf(Chapter $chapter)
  92     {
  93         $pages = $chapter->getVisiblePages();
  94         $pages->each(function ($page) {
  95             $page->html = (new PageContent($page))->render();
  96         });
  97
  98         $html = view('chapters.export', [
  99             'chapter' => $chapter,
 100             'pages' => $pages,
 101             'format' => 'pdf',
 102         ])->render();
 103
 104         return $this->htmlToPdf($html);
 105     }
 106
 107     /**
 108      * Convert a book to a PDF file.
 109      * @throws Throwable
 110      */
 111     public function bookToPdf(Book $book)
 112     {
 113         $bookTree = (new BookContents($book))->getTree(false, true);
 114         $html = view('books.export', [
 115             'book' => $book,
 116             'bookChildren' => $bookTree,
 117             'format' => 'pdf',
 118         ])->render();
 119         return $this->htmlToPdf($html);
 120     }
 121
 122     /**
 123      * Convert normal web-page HTML to a PDF.
 124      * @throws Exception
 125      */
 126     protected function htmlToPdf(string $html): string
 127     {
 128         $containedHtml = $this->containHtml($html);
 129         $useWKHTML = config('snappy.pdf.binary') !== false;
 130         if ($useWKHTML) {
 131             $pdf = SnappyPDF::loadHTML($containedHtml);
 132             $pdf->setOption('print-media-type', true);
 133         } else {
 134             $pdf = DomPDF::loadHTML($containedHtml);
 135         }
 136         return $pdf->output();
 137     }
 138
 139     /**
 140      * Bundle of the contents of a html file to be self-contained.
 141      * @throws Exception
 142      */
 143     protected function containHtml(string $htmlContent): string
 144     {
 145         $imageTagsOutput = [];
 146         preg_match_all("/\<img.*?src\=(\'|\")(.*?)(\'|\").*?\>/i", $htmlContent, $imageTagsOutput);
 147
 148         // Replace image src with base64 encoded image strings
 149         if (isset($imageTagsOutput[0]) && count($imageTagsOutput[0]) > 0) {
 150             foreach ($imageTagsOutput[0] as $index => $imgMatch) {
 151                 $oldImgTagString = $imgMatch;
 152                 $srcString = $imageTagsOutput[2][$index];
 153                 $imageEncoded = $this->imageService->imageUriToBase64($srcString);
 154                 if ($imageEncoded === null) {
 155                     $imageEncoded = $srcString;
 156                 }
 157                 $newImgTagString = str_replace($srcString, $imageEncoded, $oldImgTagString);
 158                 $htmlContent = str_replace($oldImgTagString, $newImgTagString, $htmlContent);
 159             }
 160         }
 161
 162         $linksOutput = [];
 163         preg_match_all("/\<a.*href\=(\'|\")(.*?)(\'|\").*?\>/i", $htmlContent, $linksOutput);
 164
 165         // Replace image src with base64 encoded image strings
 166         if (isset($linksOutput[0]) && count($linksOutput[0]) > 0) {
 167             foreach ($linksOutput[0] as $index => $linkMatch) {
 168                 $oldLinkString = $linkMatch;
 169                 $srcString = $linksOutput[2][$index];
 170                 if (strpos(trim($srcString), 'http') !== 0) {
 171                     $newSrcString = url($srcString);
 172                     $newLinkString = str_replace($srcString, $newSrcString, $oldLinkString);
 173                     $htmlContent = str_replace($oldLinkString, $newLinkString, $htmlContent);
 174                 }
 175             }
 176         }
 177
 178         // Replace any relative links with system domain
 179         return $htmlContent;
 180     }
 181
 182     /**
 183      * Converts the page contents into simple plain text.
 184      * This method filters any bad looking content to provide a nice final output.
 185      */
 186     public function pageToPlainText(Page $page): string
 187     {
 188         $html = (new PageContent($page))->render();
 189         $text = strip_tags($html);
 190         // Replace multiple spaces with single spaces
 191         $text = preg_replace('/\ {2,}/', ' ', $text);
 192         // Reduce multiple horrid whitespace characters.
 193         $text = preg_replace('/(\x0A|\xA0|\x0A|\r|\n){2,}/su', "\n\n", $text);
 194         $text = html_entity_decode($text);
 195         // Add title
 196         $text = $page->name . "\n\n" . $text;
 197         return $text;
 198     }
 199
 200     /**
 201      * Convert a chapter into a plain text string.
 202      */
 203     public function chapterToPlainText(Chapter $chapter): string
 204     {
 205         $text = $chapter->name . "\n\n";
 206         $text .= $chapter->description . "\n\n";
 207         foreach ($chapter->getVisiblePages() as $page) {
 208             $text .= $this->pageToPlainText($page);
 209         }
 210         return $text;
 211     }
 212
 213     /**
 214      * Convert a book into a plain text string.
 215      */
 216     public function bookToPlainText(Book $book): string
 217     {
 218         $bookTree = (new BookContents($book))->getTree(false, false);
 219         $text = $book->name . "\n\n";
 220         foreach ($bookTree as $bookChild) {
 221             if ($bookChild->isA('chapter')) {
 222                 $text .= $this->chapterToPlainText($bookChild);
 223             } else {
 224                 $text .= $this->pageToPlainText($bookChild);
 225             }
 226         }
 227         return $text;
 228     }
 229 }