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         // 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         $entryIds = [];
 176
 177         /*
 178          * Adding actual entries
 179          */
 180
 181         // set tags as subjects
 182         foreach ($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             $entryIds[] = $entry->getId();
 197         }
 198
 199         // Could also be the ISBN number, prefered for published books, or a UUID.
 200         $hash = sha1(sprintf('%s:%s', $this->wallabagUrl, implode(',', $entryIds)));
 201         $book->setIdentifier(sprintf('urn:wallabag:%s', $hash), EPub::IDENTIFIER_URI);
 202
 203         return Response::create(
 204             $book->getBook(),
 205             200,
 206             [
 207                 'Content-Description' => 'File Transfer',
 208                 'Content-type' => 'application/epub+zip',
 209                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.epub"',
 210                 'Content-Transfer-Encoding' => 'binary',
 211             ]
 212         );
 213     }
 214
 215     /**
 216      * Use PHPMobi to dump a .mobi file.
 217      *
 218      * @return Response
 219      */
 220     private function produceMobi()
 221     {
 222         $mobi = new \MOBI();
 223         $content = new \MOBIFile();
 224
 225         /*
 226          * Book metadata
 227          */
 228         $content->set('title', $this->title);
 229         $content->set('author', $this->author);
 230         $content->set('subject', $this->title);
 231
 232         /*
 233          * Front page
 234          */
 235         $content->appendParagraph($this->getExportInformation('PHPMobi'));
 236         if (file_exists($this->logoPath)) {
 237             $content->appendImage(imagecreatefrompng($this->logoPath));
 238         }
 239         $content->appendPageBreak();
 240
 241         /*
 242          * Adding actual entries
 243          */
 244         foreach ($this->entries as $entry) {
 245             $content->appendChapterTitle($entry->getTitle());
 246             $content->appendParagraph($entry->getContent());
 247             $content->appendPageBreak();
 248         }
 249         $mobi->setContentProvider($content);
 250
 251         // the browser inside Kindle Devices doesn't likes special caracters either, we limit to A-z/0-9
 252         $this->title = preg_replace('/[^A-Za-z0-9\-]/', '', $this->title);
 253
 254         return Response::create(
 255             $mobi->toString(),
 256             200,
 257             [
 258                 'Accept-Ranges' => 'bytes',
 259                 'Content-Description' => 'File Transfer',
 260                 'Content-type' => 'application/x-mobipocket-ebook',
 261                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.mobi"',
 262                 'Content-Transfer-Encoding' => 'binary',
 263             ]
 264         );
 265     }
 266
 267     /**
 268      * Use TCPDF to dump a .pdf file.
 269      *
 270      * @return Response
 271      */
 272     private function producePdf()
 273     {
 274         $pdf = new \TCPDF(PDF_PAGE_ORIENTATION, PDF_UNIT, PDF_PAGE_FORMAT, true, 'UTF-8', false);
 275
 276         /*
 277          * Book metadata
 278          */
 279         $pdf->SetCreator(PDF_CREATOR);
 280         $pdf->SetAuthor($this->author);
 281         $pdf->SetTitle($this->title);
 282         $pdf->SetSubject('Articles via wallabag');
 283         $pdf->SetKeywords('wallabag');
 284
 285         /*
 286          * Front page
 287          */
 288         $pdf->AddPage();
 289         $intro = '<h1>' . $this->title . '</h1>' . $this->getExportInformation('tcpdf');
 290
 291         $pdf->writeHTMLCell(0, 0, '', '', $intro, 0, 1, 0, true, '', true);
 292
 293         /*
 294          * Adding actual entries
 295          */
 296         foreach ($this->entries as $entry) {
 297             foreach ($entry->getTags() as $tag) {
 298                 $pdf->SetKeywords($tag->getLabel());
 299             }
 300
 301             $pdf->AddPage();
 302             $html = '<h1>' . $entry->getTitle() . '</h1>';
 303             $html .= $entry->getContent();
 304
 305             $pdf->writeHTMLCell(0, 0, '', '', $html, 0, 1, 0, true, '', true);
 306         }
 307
 308         // set image scale factor
 309         $pdf->setImageScale(PDF_IMAGE_SCALE_RATIO);
 310
 311         return Response::create(
 312             $pdf->Output('', 'S'),
 313             200,
 314             [
 315                 'Content-Description' => 'File Transfer',
 316                 'Content-type' => 'application/pdf',
 317                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.pdf"',
 318                 'Content-Transfer-Encoding' => 'binary',
 319             ]
 320         );
 321     }
 322
 323     /**
 324      * Inspired from CsvFileDumper.
 325      *
 326      * @return Response
 327      */
 328     private function produceCsv()
 329     {
 330         $delimiter = ';';
 331         $enclosure = '"';
 332         $handle = fopen('php://memory', 'b+r');
 333
 334         fputcsv($handle, ['Title', 'URL', 'Content', 'Tags', 'MIME Type', 'Language', 'Creation date'], $delimiter, $enclosure);
 335
 336         foreach ($this->entries as $entry) {
 337             fputcsv(
 338                 $handle,
 339                 [
 340                     $entry->getTitle(),
 341                     $entry->getURL(),
 342                     // remove new line to avoid crazy results
 343                     str_replace(["\r\n", "\r", "\n"], '', $entry->getContent()),
 344                     implode(', ', $entry->getTags()->toArray()),
 345                     $entry->getMimetype(),
 346                     $entry->getLanguage(),
 347                     $entry->getCreatedAt()->format('d/m/Y h:i:s'),
 348                 ],
 349                 $delimiter,
 350                 $enclosure
 351             );
 352         }
 353
 354         rewind($handle);
 355         $output = stream_get_contents($handle);
 356         fclose($handle);
 357
 358         return Response::create(
 359             $output,
 360             200,
 361             [
 362                 'Content-type' => 'application/csv',
 363                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.csv"',
 364                 'Content-Transfer-Encoding' => 'UTF-8',
 365             ]
 366         );
 367     }
 368
 369     /**
 370      * Dump a JSON file.
 371      *
 372      * @return Response
 373      */
 374     private function produceJson()
 375     {
 376         return Response::create(
 377             $this->prepareSerializingContent('json'),
 378             200,
 379             [
 380                 'Content-type' => 'application/json',
 381                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.json"',
 382                 'Content-Transfer-Encoding' => 'UTF-8',
 383             ]
 384         );
 385     }
 386
 387     /**
 388      * Dump a XML file.
 389      *
 390      * @return Response
 391      */
 392     private function produceXml()
 393     {
 394         return Response::create(
 395             $this->prepareSerializingContent('xml'),
 396             200,
 397             [
 398                 'Content-type' => 'application/xml',
 399                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.xml"',
 400                 'Content-Transfer-Encoding' => 'UTF-8',
 401             ]
 402         );
 403     }
 404
 405     /**
 406      * Dump a TXT file.
 407      *
 408      * @return Response
 409      */
 410     private function produceTxt()
 411     {
 412         $content = '';
 413         $bar = str_repeat('=', 100);
 414         foreach ($this->entries as $entry) {
 415             $content .= "\n\n" . $bar . "\n\n" . $entry->getTitle() . "\n\n" . $bar . "\n\n";
 416             $html = new Html2Text($entry->getContent(), ['do_links' => 'none', 'width' => 100]);
 417             $content .= $html->getText();
 418         }
 419
 420         return Response::create(
 421             $content,
 422             200,
 423             [
 424                 'Content-type' => 'text/plain',
 425                 'Content-Disposition' => 'attachment; filename="' . $this->title . '.txt"',
 426                 'Content-Transfer-Encoding' => 'UTF-8',
 427             ]
 428         );
 429     }
 430
 431     /**
 432      * Return a Serializer object for producing processes that need it (JSON & XML).
 433      *
 434      * @param string $format
 435      *
 436      * @return string
 437      */
 438     private function prepareSerializingContent($format)
 439     {
 440         $serializer = SerializerBuilder::create()->build();
 441
 442         return $serializer->serialize(
 443             $this->entries,
 444             $format,
 445             SerializationContext::create()->setGroups(['entries_for_user'])
 446         );
 447     }
 448
 449     /**
 450      * Return a kind of footer / information for the epub.
 451      *
 452      * @param string $type Generator of the export, can be: tdpdf, PHPePub, PHPMobi
 453      *
 454      * @return string
 455      */
 456     private function getExportInformation($type)
 457     {
 458         $info = $this->translator->trans('export.footer_template', [
 459             '%method%' => $type,
 460         ]);
 461
 462         if ('tcpdf' === $type) {
 463             return str_replace('%IMAGE%', '<img src="' . $this->logoPath . '" />', $info);
 464         }
 465
 466         return str_replace('%IMAGE%', '', $info);
 467     }
 468 }