BookStack Code Mirror - bookstack/blob - app/Entities/ExportService.php

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