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