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->title . '.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         // the browser inside Kindle Devices doesn't likes special caracters either, we limit to A-z/0-9
 269         $this->title = preg_replace('/[^A-Za-z0-9\-]/', '', $this->title);
 270
 271         return Response::create(
 272             $mobi->toString(),
 273             200,
 274             [
 275                 'Accept-Ranges' => 'bytes',
 276                 'Content-Description' => 'File Transfer',
 277                 'Content-type' => 'application/x-mobipocket-ebook',
 278                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.mobi"',
 279                 'Content-Transfer-Encoding' => 'binary',
 280             ]
 281         );
 282     }
 283
 284     /**
 285      * Use TCPDF to dump a .pdf file.
 286      *
 287      * @return Response
 288      */
 289     private function producePdf()
 290     {
 291         $pdf = new \TCPDF(PDF_PAGE_ORIENTATION, PDF_UNIT, PDF_PAGE_FORMAT, true, 'UTF-8', false);
 292
 293         /*
 294          * Book metadata
 295          */
 296         $pdf->SetCreator(PDF_CREATOR);
 297         $pdf->SetAuthor($this->author);
 298         $pdf->SetTitle($this->title);
 299         $pdf->SetSubject('Articles via wallabag');
 300         $pdf->SetKeywords('wallabag');
 301
 302         /*
 303          * Adding actual entries
 304          */
 305         foreach ($this->entries as $entry) {
 306             foreach ($entry->getTags() as $tag) {
 307                 $pdf->SetKeywords($tag->getLabel());
 308             }
 309
 310             $publishedBy = $entry->getPublishedBy();
 311             if (!empty($publishedBy)) {
 312                 $authors = implode(',', $publishedBy);
 313             } else {
 314                 $authors = $this->translator->trans('export.unknown');
 315             }
 316
 317             $pdf->addPage();
 318             $html = '<h1>' . $entry->getTitle() . '</h1>' .
 319                 '<dl>' .
 320                 '<dt>' . $this->translator->trans('entry.view.published_by') . '</dt><dd>' . $authors . '</dd>' .
 321                 '<dt>' . $this->translator->trans('entry.metadata.reading_time') . '</dt><dd>' . $this->translator->trans('entry.metadata.reading_time_minutes_short', ['%readingTime%' => $entry->getReadingTime()]) . '</dd>' .
 322                 '<dt>' . $this->translator->trans('entry.metadata.added_on') . '</dt><dd>' . $entry->getCreatedAt()->format('Y-m-d') . '</dd>' .
 323                 '<dt>' . $this->translator->trans('entry.metadata.address') . '</dt><dd><a href="' . $entry->getUrl() . '">' . $entry->getUrl() . '</a></dd>' .
 324                 '</dl>';
 325             $pdf->writeHTMLCell(0, 0, '', '', $html, 0, 1, 0, true, '', true);
 326
 327             $pdf->AddPage();
 328             $html = '<h1>' . $entry->getTitle() . '</h1>';
 329             $html .= $entry->getContent();
 330
 331             $pdf->writeHTMLCell(0, 0, '', '', $html, 0, 1, 0, true, '', true);
 332         }
 333
 334         /*
 335          * Last page
 336          */
 337         $pdf->AddPage();
 338         $html = $this->getExportInformation('tcpdf');
 339
 340         $pdf->writeHTMLCell(0, 0, '', '', $html, 0, 1, 0, true, '', true);
 341
 342         // set image scale factor
 343         $pdf->setImageScale(PDF_IMAGE_SCALE_RATIO);
 344
 345         return Response::create(
 346             $pdf->Output('', 'S'),
 347             200,
 348             [
 349                 'Content-Description' => 'File Transfer',
 350                 'Content-type' => 'application/pdf',
 351                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.pdf"',
 352                 'Content-Transfer-Encoding' => 'binary',
 353             ]
 354         );
 355     }
 356
 357     /**
 358      * Inspired from CsvFileDumper.
 359      *
 360      * @return Response
 361      */
 362     private function produceCsv()
 363     {
 364         $delimiter = ';';
 365         $enclosure = '"';
 366         $handle = fopen('php://memory', 'b+r');
 367
 368         fputcsv($handle, ['Title', 'URL', 'Content', 'Tags', 'MIME Type', 'Language', 'Creation date'], $delimiter, $enclosure);
 369
 370         foreach ($this->entries as $entry) {
 371             fputcsv(
 372                 $handle,
 373                 [
 374                     $entry->getTitle(),
 375                     $entry->getURL(),
 376                     // remove new line to avoid crazy results
 377                     str_replace(["\r\n", "\r", "\n"], '', $entry->getContent()),
 378                     implode(', ', $entry->getTags()->toArray()),
 379                     $entry->getMimetype(),
 380                     $entry->getLanguage(),
 381                     $entry->getCreatedAt()->format('d/m/Y h:i:s'),
 382                 ],
 383                 $delimiter,
 384                 $enclosure
 385             );
 386         }
 387
 388         rewind($handle);
 389         $output = stream_get_contents($handle);
 390         fclose($handle);
 391
 392         return Response::create(
 393             $output,
 394             200,
 395             [
 396                 'Content-type' => 'application/csv',
 397                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.csv"',
 398                 'Content-Transfer-Encoding' => 'UTF-8',
 399             ]
 400         );
 401     }
 402
 403     /**
 404      * Dump a JSON file.
 405      *
 406      * @return Response
 407      */
 408     private function produceJson()
 409     {
 410         return Response::create(
 411             $this->prepareSerializingContent('json'),
 412             200,
 413             [
 414                 'Content-type' => 'application/json',
 415                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.json"',
 416                 'Content-Transfer-Encoding' => 'UTF-8',
 417             ]
 418         );
 419     }
 420
 421     /**
 422      * Dump a XML file.
 423      *
 424      * @return Response
 425      */
 426     private function produceXml()
 427     {
 428         return Response::create(
 429             $this->prepareSerializingContent('xml'),
 430             200,
 431             [
 432                 'Content-type' => 'application/xml',
 433                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.xml"',
 434                 'Content-Transfer-Encoding' => 'UTF-8',
 435             ]
 436         );
 437     }
 438
 439     /**
 440      * Dump a TXT file.
 441      *
 442      * @return Response
 443      */
 444     private function produceTxt()
 445     {
 446         $content = '';
 447         $bar = str_repeat('=', 100);
 448         foreach ($this->entries as $entry) {
 449             $content .= "\n\n" . $bar . "\n\n" . $entry->getTitle() . "\n\n" . $bar . "\n\n";
 450             $html = new Html2Text($entry->getContent(), ['do_links' => 'none', 'width' => 100]);
 451             $content .= $html->getText();
 452         }
 453
 454         return Response::create(
 455             $content,
 456             200,
 457             [
 458                 'Content-type' => 'text/plain',
 459                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.txt"',
 460                 'Content-Transfer-Encoding' => 'UTF-8',
 461             ]
 462         );
 463     }
 464
 465     /**
 466      * Return a Serializer object for producing processes that need it (JSON & XML).
 467      *
 468      * @param string $format
 469      *
 470      * @return string
 471      */
 472     private function prepareSerializingContent($format)
 473     {
 474         $serializer = SerializerBuilder::create()->build();
 475
 476         return $serializer->serialize(
 477             $this->entries,
 478             $format,
 479             SerializationContext::create()->setGroups(['entries_for_user'])
 480         );
 481     }
 482
 483     /**
 484      * Return a kind of footer / information for the epub.
 485      *
 486      * @param string $type Generator of the export, can be: tdpdf, PHPePub, PHPMobi
 487      *
 488      * @return string
 489      */
 490     private function getExportInformation($type)
 491     {
 492         $info = $this->translator->trans('export.footer_template', [
 493             '%method%' => $type,
 494         ]);
 495
 496         if ('tcpdf' === $type) {
 497             return str_replace('%IMAGE%', '<img src="' . $this->logoPath . '" />', $info);
 498         }
 499
 500         return str_replace('%IMAGE%', '', $info);
 501     }
 502 }