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         $book->buildTOC();
 178
 179         /*
 180          * Adding actual entries
 181          */
 182
 183         // set tags as subjects
 184         foreach ($this->entries as $entry) {
 185             foreach ($entry->getTags() as $tag) {
 186                 $book->setSubject($tag->getLabel());
 187             }
 188
 189             // the reader in Kobo Devices doesn't likes special caracters
 190             // in filenames, we limit to A-z/0-9
 191             $filename = preg_replace('/[^A-Za-z0-9\-]/', '', $entry->getTitle());
 192
 193             $titlepage = "<h1>".$entry->getTitle()."</h1>";
 194             $chapter = $content_start . $titlepage . $entry->getContent() . $bookEnd;
 195             $book->addChapter($entry->getTitle(), htmlspecialchars($filename) . '.html', $chapter, true, EPub::EXTERNAL_REF_ADD);
 196         }
 197
 198         $book->addChapter('Notices', 'Cover2.html', $content_start . $this->getExportInformation('PHPePub') . $bookEnd);
 199
 200         return Response::create(
 201             $book->getBook(),
 202             200,
 203             [
 204                 'Content-Description' => 'File Transfer',
 205                 'Content-type' => 'application/epub+zip',
 206                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.epub"',
 207                 'Content-Transfer-Encoding' => 'binary',
 208             ]
 209         );
 210     }
 211
 212     /**
 213      * Use PHPMobi to dump a .mobi file.
 214      *
 215      * @return Response
 216      */
 217     private function produceMobi()
 218     {
 219         $mobi = new \MOBI();
 220         $content = new \MOBIFile();
 221
 222         /*
 223          * Book metadata
 224          */
 225         $content->set('title', $this->title);
 226         $content->set('author', $this->author);
 227         $content->set('subject', $this->title);
 228
 229         /*
 230          * Front page
 231          */
 232         $content->appendParagraph($this->getExportInformation('PHPMobi'));
 233         if (file_exists($this->logoPath)) {
 234             $content->appendImage(imagecreatefrompng($this->logoPath));
 235         }
 236         $content->appendPageBreak();
 237
 238         /*
 239          * Adding actual entries
 240          */
 241         foreach ($this->entries as $entry) {
 242             $content->appendChapterTitle($entry->getTitle());
 243             $content->appendParagraph($entry->getContent());
 244             $content->appendPageBreak();
 245         }
 246         $mobi->setContentProvider($content);
 247
 248         // the browser inside Kindle Devices doesn't likes special caracters either, we limit to A-z/0-9
 249         $this->title = preg_replace('/[^A-Za-z0-9\-]/', '', $this->title);
 250
 251         return Response::create(
 252             $mobi->toString(),
 253             200,
 254             [
 255                 'Accept-Ranges' => 'bytes',
 256                 'Content-Description' => 'File Transfer',
 257                 'Content-type' => 'application/x-mobipocket-ebook',
 258                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.mobi"',
 259                 'Content-Transfer-Encoding' => 'binary',
 260             ]
 261         );
 262     }
 263
 264     /**
 265      * Use TCPDF to dump a .pdf file.
 266      *
 267      * @return Response
 268      */
 269     private function producePdf()
 270     {
 271         $pdf = new \TCPDF(PDF_PAGE_ORIENTATION, PDF_UNIT, PDF_PAGE_FORMAT, true, 'UTF-8', false);
 272
 273         /*
 274          * Book metadata
 275          */
 276         $pdf->SetCreator(PDF_CREATOR);
 277         $pdf->SetAuthor($this->author);
 278         $pdf->SetTitle($this->title);
 279         $pdf->SetSubject('Articles via wallabag');
 280         $pdf->SetKeywords('wallabag');
 281
 282         /*
 283          * Front page
 284          */
 285         $pdf->AddPage();
 286         $intro = '<h1>' . $this->title . '</h1>' . $this->getExportInformation('tcpdf');
 287
 288         $pdf->writeHTMLCell(0, 0, '', '', $intro, 0, 1, 0, true, '', true);
 289
 290         /*
 291          * Adding actual entries
 292          */
 293         foreach ($this->entries as $entry) {
 294             foreach ($entry->getTags() as $tag) {
 295                 $pdf->SetKeywords($tag->getLabel());
 296             }
 297
 298             $pdf->AddPage();
 299             $html = '<h1>' . $entry->getTitle() . '</h1>';
 300             $html .= $entry->getContent();
 301
 302             $pdf->writeHTMLCell(0, 0, '', '', $html, 0, 1, 0, true, '', true);
 303         }
 304
 305         // set image scale factor
 306         $pdf->setImageScale(PDF_IMAGE_SCALE_RATIO);
 307
 308         return Response::create(
 309             $pdf->Output('', 'S'),
 310             200,
 311             [
 312                 'Content-Description' => 'File Transfer',
 313                 'Content-type' => 'application/pdf',
 314                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.pdf"',
 315                 'Content-Transfer-Encoding' => 'binary',
 316             ]
 317         );
 318     }
 319
 320     /**
 321      * Inspired from CsvFileDumper.
 322      *
 323      * @return Response
 324      */
 325     private function produceCsv()
 326     {
 327         $delimiter = ';';
 328         $enclosure = '"';
 329         $handle = fopen('php://memory', 'rb+');
 330
 331         fputcsv($handle, ['Title', 'URL', 'Content', 'Tags', 'MIME Type', 'Language', 'Creation date'], $delimiter, $enclosure);
 332
 333         foreach ($this->entries as $entry) {
 334             fputcsv(
 335                 $handle,
 336                 [
 337                     $entry->getTitle(),
 338                     $entry->getURL(),
 339                     // remove new line to avoid crazy results
 340                     str_replace(["\r\n", "\r", "\n"], '', $entry->getContent()),
 341                     implode(', ', $entry->getTags()->toArray()),
 342                     $entry->getMimetype(),
 343                     $entry->getLanguage(),
 344                     $entry->getCreatedAt()->format('d/m/Y h:i:s'),
 345                 ],
 346                 $delimiter,
 347                 $enclosure
 348             );
 349         }
 350
 351         rewind($handle);
 352         $output = stream_get_contents($handle);
 353         fclose($handle);
 354
 355         return Response::create(
 356             $output,
 357             200,
 358             [
 359                 'Content-type' => 'application/csv',
 360                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.csv"',
 361                 'Content-Transfer-Encoding' => 'UTF-8',
 362             ]
 363         );
 364     }
 365
 366     /**
 367      * Dump a JSON file.
 368      *
 369      * @return Response
 370      */
 371     private function produceJson()
 372     {
 373         return Response::create(
 374             $this->prepareSerializingContent('json'),
 375             200,
 376             [
 377                 'Content-type' => 'application/json',
 378                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.json"',
 379                 'Content-Transfer-Encoding' => 'UTF-8',
 380             ]
 381         );
 382     }
 383
 384     /**
 385      * Dump a XML file.
 386      *
 387      * @return Response
 388      */
 389     private function produceXml()
 390     {
 391         return Response::create(
 392             $this->prepareSerializingContent('xml'),
 393             200,
 394             [
 395                 'Content-type' => 'application/xml',
 396                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.xml"',
 397                 'Content-Transfer-Encoding' => 'UTF-8',
 398             ]
 399         );
 400     }
 401
 402     /**
 403      * Dump a TXT file.
 404      *
 405      * @return Response
 406      */
 407     private function produceTxt()
 408     {
 409         $content = '';
 410         $bar = str_repeat('=', 100);
 411         foreach ($this->entries as $entry) {
 412             $content .= "\n\n" . $bar . "\n\n" . $entry->getTitle() . "\n\n" . $bar . "\n\n";
 413             $html = new Html2Text($entry->getContent(), ['do_links' => 'none', 'width' => 100]);
 414             $content .= $html->getText();
 415         }
 416
 417         return Response::create(
 418             $content,
 419             200,
 420             [
 421                 'Content-type' => 'text/plain',
 422                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.txt"',
 423                 'Content-Transfer-Encoding' => 'UTF-8',
 424             ]
 425         );
 426     }
 427
 428     /**
 429      * Return a Serializer object for producing processes that need it (JSON & XML).
 430      *
 431      * @param string $format
 432      *
 433      * @return string
 434      */
 435     private function prepareSerializingContent($format)
 436     {
 437         $serializer = SerializerBuilder::create()->build();
 438
 439         return $serializer->serialize(
 440             $this->entries,
 441             $format,
 442             SerializationContext::create()->setGroups(['entries_for_user'])
 443         );
 444     }
 445
 446     /**
 447      * Return a kind of footer / information for the epub.
 448      *
 449      * @param string $type Generator of the export, can be: tdpdf, PHPePub, PHPMobi
 450      *
 451      * @return string
 452      */
 453     private function getExportInformation($type)
 454     {
 455         $info = $this->translator->trans('export.footer_template', [
 456             '%method%' => $type,
 457         ]);
 458
 459         if ('tcpdf' === $type) {
 460             return str_replace('%IMAGE%', '<img src="' . $this->logoPath . '" />', $info);
 461         }
 462
 463         return str_replace('%IMAGE%', '', $info);
 464     }
 465 }