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 = 'Various 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         // 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.
 154         $book->setLanguage($this->language);
 155         $book->setDescription('Some articles saved on my wallabag');
 156
 157         $book->setAuthor($this->author, $this->author);
 158
 159         // I hope this is a non existant address :)
 160         $book->setPublisher('wallabag', 'wallabag');
 161         // Strictly not needed as the book date defaults to time().
 162         $book->setDate(time());
 163         $book->setSourceURL($this->wallabagUrl);
 164
 165         $book->addDublinCoreMetadata(DublinCore::CONTRIBUTOR, 'PHP');
 166         $book->addDublinCoreMetadata(DublinCore::CONTRIBUTOR, 'wallabag');
 167
 168         /*
 169          * Front page
 170          */
 171         if (file_exists($this->logoPath)) {
 172             $book->setCoverImage('Cover.png', file_get_contents($this->logoPath), 'image/png');
 173         }
 174
 175         $entryIds = [];
 176         $entryCount = \count($this->entries);
 177         $i = 0;
 178
 179         /*
 180          * Adding actual entries
 181          */
 182
 183         // set tags as subjects
 184         foreach ($this->entries as $entry) {
 185             ++$i;
 186             foreach ($entry->getTags() as $tag) {
 187                 $book->setSubject($tag->getLabel());
 188             }
 189             $filename = sha1($entry->getTitle());
 190
 191             $publishedBy = $entry->getPublishedBy();
 192             if (!empty($publishedBy)) {
 193                 $authors = implode(',', $publishedBy);
 194             } else {
 195                 $authors = $this->translator->trans('export.unknown');
 196             }
 197
 198             $titlepage = $content_start .
 199                 '<h1>' . $entry->getTitle() . '</h1>' .
 200                 '<dl>' .
 201                 '<dt>' . $this->translator->trans('entry.view.published_by') . '</dt><dd>' . $authors . '</dd>' .
 202                 '<dt>' . $this->translator->trans('entry.metadata.reading_time') . '</dt><dd>' . $this->translator->trans('entry.metadata.reading_time_minutes_short', ['%readingTime%' => $entry->getReadingTime()]) . '</dd>' .
 203                 '<dt>' . $this->translator->trans('entry.metadata.added_on') . '</dt><dd>' . $entry->getCreatedAt()->format('Y-m-d') . '</dd>' .
 204                 '<dt>' . $this->translator->trans('entry.metadata.address') . '</dt><dd><a href="' . $entry->getUrl() . '">' . $entry->getUrl() . '</a></dd>' .
 205                 '</dl>' .
 206                 $bookEnd;
 207             $book->addChapter("Entry {$i} of {$entryCount}", "{$filename}_cover.html", $titlepage, true, EPub::EXTERNAL_REF_ADD);
 208             $chapter = $content_start . $entry->getContent() . $bookEnd;
 209
 210             $entryIds[] = $entry->getId();
 211             $book->addChapter($entry->getTitle(), "{$filename}.html", $chapter, true, EPub::EXTERNAL_REF_ADD);
 212         }
 213
 214         $book->addChapter('Notices', 'Cover2.html', $content_start . $this->getExportInformation('PHPePub') . $bookEnd);
 215
 216         // Could also be the ISBN number, prefered for published books, or a UUID.
 217         $hash = sha1(sprintf('%s:%s', $this->wallabagUrl, implode(',', $entryIds)));
 218         $book->setIdentifier(sprintf('urn:wallabag:%s', $hash), EPub::IDENTIFIER_URI);
 219
 220         return Response::create(
 221             $book->getBook(),
 222             200,
 223             [
 224                 'Content-Description' => 'File Transfer',
 225                 'Content-type' => 'application/epub+zip',
 226                 'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.epub"',
 227                 'Content-Transfer-Encoding' => 'binary',
 228             ]
 229         );
 230     }
 231
 232     /**
 233      * Use PHPMobi to dump a .mobi file.
 234      *
 235      * @return Response
 236      */
 237     private function produceMobi()
 238     {
 239         $mobi = new \MOBI();
 240         $content = new \MOBIFile();
 241
 242         /*
 243          * Book metadata
 244          */
 245         $content->set('title', $this->title);
 246         $content->set('author', $this->author);
 247         $content->set('subject', $this->title);
 248
 249         /*
 250          * Front page
 251          */
 252         $content->appendParagraph($this->getExportInformation('PHPMobi'));
 253         if (file_exists($this->logoPath)) {
 254             $content->appendImage(imagecreatefrompng($this->logoPath));
 255         }
 256         $content->appendPageBreak();
 257
 258         /*
 259          * Adding actual entries
 260          */
 261         foreach ($this->entries as $entry) {
 262             $content->appendChapterTitle($entry->getTitle());
 263             $content->appendParagraph($entry->getContent());
 264             $content->appendPageBreak();
 265         }
 266         $mobi->setContentProvider($content);
 267
 268         return Response::create(
 269             $mobi->toString(),
 270             200,
 271             [
 272                 'Accept-Ranges' => 'bytes',
 273                 'Content-Description' => 'File Transfer',
 274                 'Content-type' => 'application/x-mobipocket-ebook',
 275                 'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.mobi"',
 276                 'Content-Transfer-Encoding' => 'binary',
 277             ]
 278         );
 279     }
 280
 281     /**
 282      * Use TCPDF to dump a .pdf file.
 283      *
 284      * @return Response
 285      */
 286     private function producePdf()
 287     {
 288         $pdf = new \TCPDF(PDF_PAGE_ORIENTATION, PDF_UNIT, PDF_PAGE_FORMAT, true, 'UTF-8', false);
 289
 290         /*
 291          * Book metadata
 292          */
 293         $pdf->SetCreator(PDF_CREATOR);
 294         $pdf->SetAuthor($this->author);
 295         $pdf->SetTitle($this->title);
 296         $pdf->SetSubject('Articles via wallabag');
 297         $pdf->SetKeywords('wallabag');
 298
 299         /*
 300          * Adding actual entries
 301          */
 302         foreach ($this->entries as $entry) {
 303             foreach ($entry->getTags() as $tag) {
 304                 $pdf->SetKeywords($tag->getLabel());
 305             }
 306
 307             $publishedBy = $entry->getPublishedBy();
 308             if (!empty($publishedBy)) {
 309                 $authors = implode(',', $publishedBy);
 310             } else {
 311                 $authors = $this->translator->trans('export.unknown');
 312             }
 313
 314             $pdf->addPage();
 315             $html = '<h1>' . $entry->getTitle() . '</h1>' .
 316                 '<dl>' .
 317                 '<dt>' . $this->translator->trans('entry.view.published_by') . '</dt><dd>' . $authors . '</dd>' .
 318                 '<dt>' . $this->translator->trans('entry.metadata.reading_time') . '</dt><dd>' . $this->translator->trans('entry.metadata.reading_time_minutes_short', ['%readingTime%' => $entry->getReadingTime()]) . '</dd>' .
 319                 '<dt>' . $this->translator->trans('entry.metadata.added_on') . '</dt><dd>' . $entry->getCreatedAt()->format('Y-m-d') . '</dd>' .
 320                 '<dt>' . $this->translator->trans('entry.metadata.address') . '</dt><dd><a href="' . $entry->getUrl() . '">' . $entry->getUrl() . '</a></dd>' .
 321                 '</dl>';
 322             $pdf->writeHTMLCell(0, 0, '', '', $html, 0, 1, 0, true, '', true);
 323
 324             $pdf->AddPage();
 325             $html = '<h1>' . $entry->getTitle() . '</h1>';
 326             $html .= $entry->getContent();
 327
 328             $pdf->writeHTMLCell(0, 0, '', '', $html, 0, 1, 0, true, '', true);
 329         }
 330
 331         /*
 332          * Last page
 333          */
 334         $pdf->AddPage();
 335         $html = $this->getExportInformation('tcpdf');
 336
 337         $pdf->writeHTMLCell(0, 0, '', '', $html, 0, 1, 0, true, '', true);
 338
 339         // set image scale factor
 340         $pdf->setImageScale(PDF_IMAGE_SCALE_RATIO);
 341
 342         return Response::create(
 343             $pdf->Output('', 'S'),
 344             200,
 345             [
 346                 'Content-Description' => 'File Transfer',
 347                 'Content-type' => 'application/pdf',
 348                 'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.pdf"',
 349                 'Content-Transfer-Encoding' => 'binary',
 350             ]
 351         );
 352     }
 353
 354     /**
 355      * Inspired from CsvFileDumper.
 356      *
 357      * @return Response
 358      */
 359     private function produceCsv()
 360     {
 361         $delimiter = ';';
 362         $enclosure = '"';
 363         $handle = fopen('php://memory', 'b+r');
 364
 365         fputcsv($handle, ['Title', 'URL', 'Content', 'Tags', 'MIME Type', 'Language', 'Creation date'], $delimiter, $enclosure);
 366
 367         foreach ($this->entries as $entry) {
 368             fputcsv(
 369                 $handle,
 370                 [
 371                     $entry->getTitle(),
 372                     $entry->getURL(),
 373                     // remove new line to avoid crazy results
 374                     str_replace(["\r\n", "\r", "\n"], '', $entry->getContent()),
 375                     implode(', ', $entry->getTags()->toArray()),
 376                     $entry->getMimetype(),
 377                     $entry->getLanguage(),
 378                     $entry->getCreatedAt()->format('d/m/Y h:i:s'),
 379                 ],
 380                 $delimiter,
 381                 $enclosure
 382             );
 383         }
 384
 385         rewind($handle);
 386         $output = stream_get_contents($handle);
 387         fclose($handle);
 388
 389         return Response::create(
 390             $output,
 391             200,
 392             [
 393                 'Content-type' => 'application/csv',
 394                 'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.csv"',
 395                 'Content-Transfer-Encoding' => 'UTF-8',
 396             ]
 397         );
 398     }
 399
 400     /**
 401      * Dump a JSON file.
 402      *
 403      * @return Response
 404      */
 405     private function produceJson()
 406     {
 407         return Response::create(
 408             $this->prepareSerializingContent('json'),
 409             200,
 410             [
 411                 'Content-type' => 'application/json',
 412                 'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.json"',
 413                 'Content-Transfer-Encoding' => 'UTF-8',
 414             ]
 415         );
 416     }
 417
 418     /**
 419      * Dump a XML file.
 420      *
 421      * @return Response
 422      */
 423     private function produceXml()
 424     {
 425         return Response::create(
 426             $this->prepareSerializingContent('xml'),
 427             200,
 428             [
 429                 'Content-type' => 'application/xml',
 430                 'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.xml"',
 431                 'Content-Transfer-Encoding' => 'UTF-8',
 432             ]
 433         );
 434     }
 435
 436     /**
 437      * Dump a TXT file.
 438      *
 439      * @return Response
 440      */
 441     private function produceTxt()
 442     {
 443         $content = '';
 444         $bar = str_repeat('=', 100);
 445         foreach ($this->entries as $entry) {
 446             $content .= "\n\n" . $bar . "\n\n" . $entry->getTitle() . "\n\n" . $bar . "\n\n";
 447             $html = new Html2Text($entry->getContent(), ['do_links' => 'none', 'width' => 100]);
 448             $content .= $html->getText();
 449         }
 450
 451         return Response::create(
 452             $content,
 453             200,
 454             [
 455                 'Content-type' => 'text/plain',
 456                 'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.txt"',
 457                 'Content-Transfer-Encoding' => 'UTF-8',
 458             ]
 459         );
 460     }
 461
 462     /**
 463      * Return a Serializer object for producing processes that need it (JSON & XML).
 464      *
 465      * @param string $format
 466      *
 467      * @return string
 468      */
 469     private function prepareSerializingContent($format)
 470     {
 471         $serializer = SerializerBuilder::create()->build();
 472
 473         return $serializer->serialize(
 474             $this->entries,
 475             $format,
 476             SerializationContext::create()->setGroups(['entries_for_user'])
 477         );
 478     }
 479
 480     /**
 481      * Return a kind of footer / information for the epub.
 482      *
 483      * @param string $type Generator of the export, can be: tdpdf, PHPePub, PHPMobi
 484      *
 485      * @return string
 486      */
 487     private function getExportInformation($type)
 488     {
 489         $info = $this->translator->trans('export.footer_template', [
 490             '%method%' => $type,
 491         ]);
 492
 493         if ('tcpdf' === $type) {
 494             return str_replace('%IMAGE%', '<img src="' . $this->logoPath . '" />', $info);
 495         }
 496
 497         return str_replace('%IMAGE%', '', $info);
 498     }
 499
 500     /**
 501      * Return a sanitized version of the title by applying translit iconv
 502      * and removing non alphanumeric characters, - and space.
 503      *
 504      * @return string Sanitized filename
 505      */
 506     private function getSanitizedFilename()
 507     {
 508         return preg_replace('/[^A-Za-z0-9\- \']/', '', iconv('utf-8', 'us-ascii//TRANSLIT', $this->title));
 509     }
 510 }