src/Wallabag/CoreBundle/Helper/EntriesExport.php

   1 <?php
   2
   3 namespace Wallabag\CoreBundle\Helper;
   4
   5 use Html2Text\Html2Text;
   6 use JMS\Serializer\SerializationContext;
   7 use JMS\Serializer\SerializerBuilder;
   8 use PHPePub\Core\EPub;
   9 use PHPePub\Core\Structure\OPF\DublinCore;
  10 use Symfony\Component\HttpFoundation\Response;
  11 use Symfony\Component\Translation\TranslatorInterface;
  12 use Wallabag\CoreBundle\Entity\Entry;
  13
  14 /**
  15  * This class doesn't have unit test BUT it's fully covered by a functional test with ExportControllerTest.
  16  */
  17 class EntriesExport
  18 {
  19     private $wallabagUrl;
  20     private $logoPath;
  21     private $translator;
  22     private $title = '';
  23     private $entries = [];
  24     private $author = 'wallabag';
  25     private $language = '';
  26
  27     /**
  28      * @param TranslatorInterface $translator  Translator service
  29      * @param string              $wallabagUrl Wallabag instance url
  30      * @param string              $logoPath    Path to the logo FROM THE BUNDLE SCOPE
  31      */
  32     public function __construct(TranslatorInterface $translator, $wallabagUrl, $logoPath)
  33     {
  34         $this->translator = $translator;
  35         $this->wallabagUrl = $wallabagUrl;
  36         $this->logoPath = $logoPath;
  37     }
  38
  39     /**
  40      * Define entries.
  41      *
  42      * @param array|Entry $entries An array of entries or one entry
  43      *
  44      * @return EntriesExport
  45      */
  46     public function setEntries($entries)
  47     {
  48         if (!\is_array($entries)) {
  49             $this->language = $entries->getLanguage();
  50             $entries = [$entries];
  51         }
  52
  53         $this->entries = $entries;
  54
  55         return $this;
  56     }
  57
  58     /**
  59      * Sets the category of which we want to get articles, or just one entry.
  60      *
  61      * @param string $method Method to get articles
  62      *
  63      * @return EntriesExport
  64      */
  65     public function updateTitle($method)
  66     {
  67         $this->title = $method . ' articles';
  68
  69         if ('entry' === $method) {
  70             $this->title = $this->entries[0]->getTitle();
  71         }
  72
  73         return $this;
  74     }
  75
  76     /**
  77      * Sets the author for one entry or category.
  78      *
  79      * The publishers are used, or the domain name if empty.
  80      *
  81      * @param string $method Method to get articles
  82      *
  83      * @return EntriesExport
  84      */
  85     public function updateAuthor($method)
  86     {
  87         if ('entry' !== $method) {
  88             $this->author = $method . ' authors';
  89
  90             return $this;
  91         }
  92
  93         $this->author = $this->entries[0]->getDomainName();
  94
  95         $publishedBy = $this->entries[0]->getPublishedBy();
  96         if (!empty($publishedBy)) {
  97             $this->author = implode(', ', $publishedBy);
  98         }
  99
 100         return $this;
 101     }
 102
 103     /**
 104      * Sets the output format.
 105      *
 106      * @param string $format
 107      *
 108      * @return Response
 109      */
 110     public function exportAs($format)
 111     {
 112         $functionName = 'produce' . ucfirst($format);
 113         if (method_exists($this, $functionName)) {
 114             return $this->$functionName();
 115         }
 116
 117         throw new \InvalidArgumentException(sprintf('The format "%s" is not yet supported.', $format));
 118     }
 119
 120     public function exportJsonData()
 121     {
 122         return $this->prepareSerializingContent('json');
 123     }
 124
 125     /**
 126      * Use PHPePub to dump a .epub file.
 127      *
 128      * @return Response
 129      */
 130     private function produceEpub()
 131     {
 132         /*
 133          * Start and End of the book
 134          */
 135         $content_start =
 136             "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
 137             . "<html xmlns=\"http://www.w3.org/1999/xhtml\" xmlns:epub=\"http://www.idpf.org/2007/ops\">\n"
 138             . '<head>'
 139             . "<meta http-equiv=\"Default-Style\" content=\"text/html; charset=utf-8\" />\n"
 140             . "<title>wallabag articles book</title>\n"
 141             . "</head>\n"
 142             . "<body>\n";
 143
 144         $bookEnd = "</body>\n</html>\n";
 145
 146         $book = new EPub(EPub::BOOK_VERSION_EPUB3);
 147
 148         /*
 149          * Book metadata
 150          */
 151
 152         $book->setTitle($this->title);
 153         // Could also be the ISBN number, prefered for published books, or a UUID.
 154         $book->setIdentifier($this->title, EPub::IDENTIFIER_URI);
 155         // Not needed, but included for the example, Language is mandatory, but EPub defaults to "en". Use RFC3066 Language codes, such as "en", "da", "fr" etc.
 156         $book->setLanguage($this->language);
 157         $book->setDescription('Some articles saved on my wallabag');
 158
 159         $book->setAuthor($this->author, $this->author);
 160
 161         // I hope this is a non existant address :)
 162         $book->setPublisher('wallabag', 'wallabag');
 163         // Strictly not needed as the book date defaults to time().
 164         $book->setDate(time());
 165         $book->setSourceURL($this->wallabagUrl);
 166
 167         $book->addDublinCoreMetadata(DublinCore::CONTRIBUTOR, 'PHP');
 168         $book->addDublinCoreMetadata(DublinCore::CONTRIBUTOR, 'wallabag');
 169
 170         /*
 171          * Front page
 172          */
 173         if (file_exists($this->logoPath)) {
 174             $book->setCoverImage('Cover.png', file_get_contents($this->logoPath), 'image/png');
 175         }
 176
 177         /*
 178          * Adding actual entries
 179          */
 180
 181         // set tags as subjects
 182         foreach ($this->entries as $entry) {
 183             foreach ($entry->getTags() as $tag) {
 184                 $book->setSubject($tag->getLabel());
 185             }
 186
 187             // the reader in Kobo Devices doesn't likes special caracters
 188             // in filenames, we limit to A-z/0-9
 189             $filename = preg_replace('/[^A-Za-z0-9\-]/', '', $entry->getTitle());
 190
 191             $titlepage = $content_start . '<h1>' . $entry->getTitle() . '</h1>' . $this->getExportInformation('PHPePub') . $bookEnd;
 192             $book->addChapter('Title', 'Title.html', $titlepage, true, EPub::EXTERNAL_REF_ADD);
 193             $chapter = $content_start . $entry->getContent() . $bookEnd;
 194             $book->addChapter($entry->getTitle(), htmlspecialchars($filename) . '.html', $chapter, true, EPub::EXTERNAL_REF_ADD);
 195         }
 196
 197         $book->buildTOC();
 198
 199         return Response::create(
 200             $book->getBook(),
 201             200,
 202             [
 203                 'Content-Description' => 'File Transfer',
 204                 'Content-type' => 'application/epub+zip',
 205                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.epub"',
 206                 'Content-Transfer-Encoding' => 'binary',
 207             ]
 208         );
 209     }
 210
 211     /**
 212      * Use PHPMobi to dump a .mobi file.
 213      *
 214      * @return Response
 215      */
 216     private function produceMobi()
 217     {
 218         $mobi = new \MOBI();
 219         $content = new \MOBIFile();
 220
 221         /*
 222          * Book metadata
 223          */
 224         $content->set('title', $this->title);
 225         $content->set('author', $this->author);
 226         $content->set('subject', $this->title);
 227
 228         /*
 229          * Front page
 230          */
 231         $content->appendParagraph($this->getExportInformation('PHPMobi'));
 232         if (file_exists($this->logoPath)) {
 233             $content->appendImage(imagecreatefrompng($this->logoPath));
 234         }
 235         $content->appendPageBreak();
 236
 237         /*
 238          * Adding actual entries
 239          */
 240         foreach ($this->entries as $entry) {
 241             $content->appendChapterTitle($entry->getTitle());
 242             $content->appendParagraph($entry->getContent());
 243             $content->appendPageBreak();
 244         }
 245         $mobi->setContentProvider($content);
 246
 247         // the browser inside Kindle Devices doesn't likes special caracters either, we limit to A-z/0-9
 248         $this->title = preg_replace('/[^A-Za-z0-9\-]/', '', $this->title);
 249
 250         return Response::create(
 251             $mobi->toString(),
 252             200,
 253             [
 254                 'Accept-Ranges' => 'bytes',
 255                 'Content-Description' => 'File Transfer',
 256                 'Content-type' => 'application/x-mobipocket-ebook',
 257                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.mobi"',
 258                 'Content-Transfer-Encoding' => 'binary',
 259             ]
 260         );
 261     }
 262
 263     /**
 264      * Use TCPDF to dump a .pdf file.
 265      *
 266      * @return Response
 267      */
 268     private function producePdf()
 269     {
 270         $pdf = new \TCPDF(PDF_PAGE_ORIENTATION, PDF_UNIT, PDF_PAGE_FORMAT, true, 'UTF-8', false);
 271
 272         /*
 273          * Book metadata
 274          */
 275         $pdf->SetCreator(PDF_CREATOR);
 276         $pdf->SetAuthor($this->author);
 277         $pdf->SetTitle($this->title);
 278         $pdf->SetSubject('Articles via wallabag');
 279         $pdf->SetKeywords('wallabag');
 280
 281         /*
 282          * Front page
 283          */
 284         $pdf->AddPage();
 285         $intro = '<h1>' . $this->title . '</h1>' . $this->getExportInformation('tcpdf');
 286
 287         $pdf->writeHTMLCell(0, 0, '', '', $intro, 0, 1, 0, true, '', true);
 288
 289         /*
 290          * Adding actual entries
 291          */
 292         foreach ($this->entries as $entry) {
 293             foreach ($entry->getTags() as $tag) {
 294                 $pdf->SetKeywords($tag->getLabel());
 295             }
 296
 297             $pdf->AddPage();
 298             $html = '<h1>' . $entry->getTitle() . '</h1>';
 299             $html .= $entry->getContent();
 300
 301             $pdf->writeHTMLCell(0, 0, '', '', $html, 0, 1, 0, true, '', true);
 302         }
 303
 304         // set image scale factor
 305         $pdf->setImageScale(PDF_IMAGE_SCALE_RATIO);
 306
 307         return Response::create(
 308             $pdf->Output('', 'S'),
 309             200,
 310             [
 311                 'Content-Description' => 'File Transfer',
 312                 'Content-type' => 'application/pdf',
 313                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.pdf"',
 314                 'Content-Transfer-Encoding' => 'binary',
 315             ]
 316         );
 317     }
 318
 319     /**
 320      * Inspired from CsvFileDumper.
 321      *
 322      * @return Response
 323      */
 324     private function produceCsv()
 325     {
 326         $delimiter = ';';
 327         $enclosure = '"';
 328         $handle = fopen('php://memory', 'b+r');
 329
 330         fputcsv($handle, ['Title', 'URL', 'Content', 'Tags', 'MIME Type', 'Language', 'Creation date'], $delimiter, $enclosure);
 331
 332         foreach ($this->entries as $entry) {
 333             fputcsv(
 334                 $handle,
 335                 [
 336                     $entry->getTitle(),
 337                     $entry->getURL(),
 338                     // remove new line to avoid crazy results
 339                     str_replace(["\r\n", "\r", "\n"], '', $entry->getContent()),
 340                     implode(', ', $entry->getTags()->toArray()),
 341                     $entry->getMimetype(),
 342                     $entry->getLanguage(),
 343                     $entry->getCreatedAt()->format('d/m/Y h:i:s'),
 344                 ],
 345                 $delimiter,
 346                 $enclosure
 347             );
 348         }
 349
 350         rewind($handle);
 351         $output = stream_get_contents($handle);
 352         fclose($handle);
 353
 354         return Response::create(
 355             $output,
 356             200,
 357             [
 358                 'Content-type' => 'application/csv',
 359                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.csv"',
 360                 'Content-Transfer-Encoding' => 'UTF-8',
 361             ]
 362         );
 363     }
 364
 365     /**
 366      * Dump a JSON file.
 367      *
 368      * @return Response
 369      */
 370     private function produceJson()
 371     {
 372         return Response::create(
 373             $this->prepareSerializingContent('json'),
 374             200,
 375             [
 376                 'Content-type' => 'application/json',
 377                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.json"',
 378                 'Content-Transfer-Encoding' => 'UTF-8',
 379             ]
 380         );
 381     }
 382
 383     /**
 384      * Dump a XML file.
 385      *
 386      * @return Response
 387      */
 388     private function produceXml()
 389     {
 390         return Response::create(
 391             $this->prepareSerializingContent('xml'),
 392             200,
 393             [
 394                 'Content-type' => 'application/xml',
 395                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.xml"',
 396                 'Content-Transfer-Encoding' => 'UTF-8',
 397             ]
 398         );
 399     }
 400
 401     /**
 402      * Dump a TXT file.
 403      *
 404      * @return Response
 405      */
 406     private function produceTxt()
 407     {
 408         $content = '';
 409         $bar = str_repeat('=', 100);
 410         foreach ($this->entries as $entry) {
 411             $content .= "\n\n" . $bar . "\n\n" . $entry->getTitle() . "\n\n" . $bar . "\n\n";
 412             $html = new Html2Text($entry->getContent(), ['do_links' => 'none', 'width' => 100]);
 413             $content .= $html->getText();
 414         }
 415
 416         return Response::create(
 417             $content,
 418             200,
 419             [
 420                 'Content-type' => 'text/plain',
 421                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.txt"',
 422                 'Content-Transfer-Encoding' => 'UTF-8',
 423             ]
 424         );
 425     }
 426
 427     /**
 428      * Return a Serializer object for producing processes that need it (JSON & XML).
 429      *
 430      * @param string $format
 431      *
 432      * @return string
 433      */
 434     private function prepareSerializingContent($format)
 435     {
 436         $serializer = SerializerBuilder::create()->build();
 437
 438         return $serializer->serialize(
 439             $this->entries,
 440             $format,
 441             SerializationContext::create()->setGroups(['entries_for_user'])
 442         );
 443     }
 444
 445     /**
 446      * Return a kind of footer / information for the epub.
 447      *
 448      * @param string $type Generator of the export, can be: tdpdf, PHPePub, PHPMobi
 449      *
 450      * @return string
 451      */
 452     private function getExportInformation($type)
 453     {
 454         $info = $this->translator->trans('export.footer_template', [
 455             '%method%' => $type,
 456         ]);
 457
 458         if ('tcpdf' === $type) {
 459             return str_replace('%IMAGE%', '<img src="' . $this->logoPath . '" />', $info);
 460         }
 461
 462         return str_replace('%IMAGE%', '', $info);
 463     }
 464 }