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