src/Wallabag/CoreBundle/Helper/ContentProxy.php

   1 <?php
   2
   3 namespace Wallabag\CoreBundle\Helper;
   4
   5 use Graby\Graby;
   6 use Psr\Log\LoggerInterface;
   7 use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
   8 use Symfony\Component\Validator\Constraints\Locale as LocaleConstraint;
   9 use Symfony\Component\Validator\Constraints\Url as UrlConstraint;
  10 use Symfony\Component\Validator\Validator\ValidatorInterface;
  11 use Wallabag\CoreBundle\Entity\Entry;
  12 use Wallabag\CoreBundle\Tools\Utils;
  13
  14 /**
  15  * This kind of proxy class takes care of getting the content from an url
  16  * and updates the entry with what it found.
  17  */
  18 class ContentProxy
  19 {
  20     protected $graby;
  21     protected $tagger;
  22     protected $validator;
  23     protected $logger;
  24     protected $mimeGuesser;
  25     protected $fetchingErrorMessage;
  26     protected $eventDispatcher;
  27     protected $storeArticleHeaders;
  28
  29     public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage, $storeArticleHeaders = false)
  30     {
  31         $this->graby = $graby;
  32         $this->tagger = $tagger;
  33         $this->validator = $validator;
  34         $this->logger = $logger;
  35         $this->mimeGuesser = new MimeTypeExtensionGuesser();
  36         $this->fetchingErrorMessage = $fetchingErrorMessage;
  37         $this->storeArticleHeaders = $storeArticleHeaders;
  38     }
  39
  40     /**
  41      * Update entry using either fetched or provided content.
  42      *
  43      * @param Entry  $entry                Entry to update
  44      * @param string $url                  Url of the content
  45      * @param array  $content              Array with content provided for import with AT LEAST keys title, html, url to skip the fetchContent from the url
  46      * @param bool   $disableContentUpdate Whether to skip trying to fetch content using Graby
  47      */
  48     public function updateEntry(Entry $entry, $url, array $content = [], $disableContentUpdate = false)
  49     {
  50         $this->graby->toggleImgNoReferrer(true);
  51         if (!empty($content['html'])) {
  52             $content['html'] = $this->graby->cleanupHtml($content['html'], $url);
  53         }
  54
  55         if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) {
  56             $fetchedContent = $this->graby->fetchContent($url);
  57
  58             $fetchedContent['title'] = $this->sanitizeContentTitle(
  59                 $fetchedContent['title'],
  60                 isset($fetchedContent['headers']['content-type']) ? $fetchedContent['headers']['content-type'] : ''
  61             );
  62
  63             // when content is imported, we have information in $content
  64             // in case fetching content goes bad, we'll keep the imported information instead of overriding them
  65             if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage) {
  66                 $content = $fetchedContent;
  67             }
  68         }
  69
  70         // be sure to keep the url in case of error
  71         // so we'll be able to refetch it in the future
  72         $content['url'] = !empty($content['url']) ? $content['url'] : $url;
  73
  74         // In one case (at least in tests), url is empty here
  75         // so we set it using $url provided in the updateEntry call.
  76         // Not sure what are the other possible cases where this property is empty
  77         if (empty($entry->getUrl()) && !empty($url)) {
  78             $entry->setUrl($url);
  79         }
  80
  81         $entry->setGivenUrl($url);
  82
  83         $this->stockEntry($entry, $content);
  84     }
  85
  86     /**
  87      * Use a Symfony validator to ensure the language is well formatted.
  88      *
  89      * @param Entry  $entry
  90      * @param string $value Language to validate and save
  91      */
  92     public function updateLanguage(Entry $entry, $value)
  93     {
  94         // some lang are defined as fr-FR, es-ES.
  95         // replacing - by _ might increase language support
  96         $value = str_replace('-', '_', $value);
  97
  98         $errors = $this->validator->validate(
  99             $value,
 100             (new LocaleConstraint())
 101         );
 102
 103         if (0 === \count($errors)) {
 104             $entry->setLanguage($value);
 105
 106             return;
 107         }
 108
 109         $this->logger->warning('Language validation failed. ' . (string) $errors);
 110     }
 111
 112     /**
 113      * Use a Symfony validator to ensure the preview picture is a real url.
 114      *
 115      * @param Entry  $entry
 116      * @param string $value URL to validate and save
 117      */
 118     public function updatePreviewPicture(Entry $entry, $value)
 119     {
 120         $errors = $this->validator->validate(
 121             $value,
 122             (new UrlConstraint())
 123         );
 124
 125         if (0 === \count($errors)) {
 126             $entry->setPreviewPicture($value);
 127
 128             return;
 129         }
 130
 131         $this->logger->warning('PreviewPicture validation failed. ' . (string) $errors);
 132     }
 133
 134     /**
 135      * Update date.
 136      *
 137      * @param Entry  $entry
 138      * @param string $value Date to validate and save
 139      */
 140     public function updatePublishedAt(Entry $entry, $value)
 141     {
 142         $date = $value;
 143
 144         // is it a timestamp?
 145         if (false !== filter_var($date, FILTER_VALIDATE_INT)) {
 146             $date = '@' . $date;
 147         }
 148
 149         try {
 150             // is it already a DateTime?
 151             // (it's inside the try/catch in case of fail to be parse time string)
 152             if (!$date instanceof \DateTime) {
 153                 $date = new \DateTime($date);
 154             }
 155
 156             $entry->setPublishedAt($date);
 157         } catch (\Exception $e) {
 158             $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $entry->getUrl(), 'date' => $value]);
 159         }
 160     }
 161
 162     /**
 163      * Helper to extract and save host from entry url.
 164      *
 165      * @param Entry $entry
 166      */
 167     public function setEntryDomainName(Entry $entry)
 168     {
 169         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
 170         if (false !== $domainName) {
 171             $entry->setDomainName($domainName);
 172         }
 173     }
 174
 175     /**
 176      * Helper to set a default title using:
 177      * - url basename, if applicable
 178      * - hostname.
 179      *
 180      * @param Entry $entry
 181      */
 182     public function setDefaultEntryTitle(Entry $entry)
 183     {
 184         $url = parse_url($entry->getUrl());
 185         $path = pathinfo($url['path'], PATHINFO_BASENAME);
 186
 187         if (empty($path)) {
 188             $path = $url['host'];
 189         }
 190
 191         $entry->setTitle($path);
 192     }
 193
 194     /**
 195      * Try to sanitize the title of the fetched content from wrong character encodings and invalid UTF-8 character.
 196      *
 197      * @param string $title
 198      * @param string $contentType
 199      *
 200      * @return string
 201      */
 202     private function sanitizeContentTitle($title, $contentType)
 203     {
 204         if ('application/pdf' === $contentType) {
 205             $title = $this->convertPdfEncodingToUTF8($title);
 206         }
 207
 208         return $this->sanitizeUTF8Text($title);
 209     }
 210
 211     /**
 212      * If the title from the fetched content comes from a PDF, then its very possible that the character encoding is not
 213      * UTF-8. This methods tries to identify the character encoding and translate the title to UTF-8.
 214      *
 215      * @param $title
 216      *
 217      * @return string (maybe contains invalid UTF-8 character)
 218      */
 219     private function convertPdfEncodingToUTF8($title)
 220     {
 221         // first try UTF-8 because its easier to detect its present/absence
 222         foreach (['UTF-8', 'UTF-16BE', 'WINDOWS-1252'] as $encoding) {
 223             if (mb_check_encoding($title, $encoding)) {
 224                 return mb_convert_encoding($title, 'UTF-8', $encoding);
 225             }
 226         }
 227
 228         return $title;
 229     }
 230
 231     /**
 232      * Remove invalid UTF-8 characters from the given string.
 233      *
 234      * @param string $rawText
 235      *
 236      * @return string
 237      */
 238     private function sanitizeUTF8Text($rawText)
 239     {
 240         if (mb_check_encoding($rawText, 'UTF-8')) {
 241             return $rawText;
 242         }
 243
 244         return iconv('UTF-8', 'UTF-8//IGNORE', $rawText);
 245     }
 246
 247     /**
 248      * Stock entry with fetched or imported content.
 249      * Will fall back to OpenGraph data if available.
 250      *
 251      * @param Entry $entry   Entry to stock
 252      * @param array $content Array with at least title, url & html
 253      */
 254     private function stockEntry(Entry $entry, array $content)
 255     {
 256         $this->updateOriginUrl($entry, $content['url']);
 257
 258         $this->setEntryDomainName($entry);
 259
 260         if (!empty($content['title'])) {
 261             $entry->setTitle($content['title']);
 262         }
 263
 264         if (empty($content['html'])) {
 265             $content['html'] = $this->fetchingErrorMessage;
 266
 267             if (!empty($content['description'])) {
 268                 $content['html'] .= '<p><i>But we found a short description: </i></p>';
 269                 $content['html'] .= $content['description'];
 270             }
 271         }
 272
 273         $entry->setContent($content['html']);
 274         $entry->setReadingTime(Utils::getReadingTime($content['html']));
 275
 276         if (!empty($content['status'])) {
 277             $entry->setHttpStatus($content['status']);
 278         }
 279
 280         if (!empty($content['authors']) && \is_array($content['authors'])) {
 281             $entry->setPublishedBy($content['authors']);
 282         }
 283
 284         if (!empty($content['headers'])) {
 285             $entry->setHeaders($content['headers']);
 286         }
 287
 288         if (!empty($content['date'])) {
 289             $this->updatePublishedAt($entry, $content['date']);
 290         }
 291
 292         if (!empty($content['language'])) {
 293             $this->updateLanguage($entry, $content['language']);
 294         }
 295
 296         $previewPictureUrl = '';
 297         if (!empty($content['image'])) {
 298             $previewPictureUrl = $content['image'];
 299         }
 300
 301         // if content is an image, define it as a preview too
 302         if (!empty($content['headers']['content-type']) && \in_array($this->mimeGuesser->guess($content['headers']['content-type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
 303             $previewPictureUrl = $content['url'];
 304         } elseif (empty($previewPictureUrl)) {
 305             $this->logger->debug('Extracting images from content to provide a default preview picture');
 306             $imagesUrls = DownloadImages::extractImagesUrlsFromHtml($content['html']);
 307             $this->logger->debug(\count($imagesUrls) . ' pictures found');
 308
 309             if (!empty($imagesUrls)) {
 310                 $previewPictureUrl = $imagesUrls[0];
 311             }
 312         }
 313
 314         if (!empty($content['headers']['content-type'])) {
 315             $entry->setMimetype($content['headers']['content-type']);
 316         }
 317
 318         if (!empty($previewPictureUrl)) {
 319             $this->updatePreviewPicture($entry, $previewPictureUrl);
 320         }
 321
 322         try {
 323             $this->tagger->tag($entry);
 324         } catch (\Exception $e) {
 325             $this->logger->error('Error while trying to automatically tag an entry.', [
 326                 'entry_url' => $content['url'],
 327                 'error_msg' => $e->getMessage(),
 328             ]);
 329         }
 330     }
 331
 332     /**
 333      * Update the origin_url field when a redirection occurs
 334      * This field is set if it is empty and new url does not match ignore list.
 335      *
 336      * @param Entry  $entry
 337      * @param string $url
 338      */
 339     private function updateOriginUrl(Entry $entry, $url)
 340     {
 341         if (empty($url) || $entry->getUrl() === $url) {
 342             return false;
 343         }
 344
 345         $parsed_entry_url = parse_url($entry->getUrl());
 346         $parsed_content_url = parse_url($url);
 347
 348         /**
 349          * The following part computes the list of part changes between two
 350          * parse_url arrays.
 351          *
 352          * As array_diff_assoc only computes changes to go from the left array
 353          * to the right one, we make two differents arrays to have both
 354          * directions. We merge these two arrays and sort keys before passing
 355          * the result to the switch.
 356          *
 357          * The resulting array gives us all changing parts between the two
 358          * urls: scheme, host, path, query and/or fragment.
 359          */
 360         $diff_ec = array_diff_assoc($parsed_entry_url, $parsed_content_url);
 361         $diff_ce = array_diff_assoc($parsed_content_url, $parsed_entry_url);
 362
 363         $diff = array_merge($diff_ec, $diff_ce);
 364         $diff_keys = array_keys($diff);
 365         sort($diff_keys);
 366
 367         if ($this->ignoreUrl($entry->getUrl())) {
 368             $entry->setUrl($url);
 369
 370             return false;
 371         }
 372
 373         /**
 374          * This switch case lets us apply different behaviors according to
 375          * changing parts of urls.
 376          *
 377          * As $diff_keys is an array, we provide arrays as cases. ['path'] means
 378          * 'only the path is different between the two urls' whereas
 379          * ['fragment', 'query'] means 'only fragment and query string parts are
 380          * different between the two urls'.
 381          *
 382          * Note that values in $diff_keys are sorted.
 383          */
 384         switch ($diff_keys) {
 385             case ['path']:
 386                 if (($parsed_entry_url['path'] . '/' === $parsed_content_url['path']) // diff is trailing slash, we only replace the url of the entry
 387                     || ($url === urldecode($entry->getUrl()))) { // we update entry url if new url is a decoded version of it, see EntryRepository#findByUrlAndUserId
 388                     $entry->setUrl($url);
 389                 }
 390                 break;
 391             case ['scheme']:
 392                 $entry->setUrl($url);
 393                 break;
 394             case ['fragment']:
 395                 // noop
 396                 break;
 397             default:
 398                 if (empty($entry->getOriginUrl())) {
 399                     $entry->setOriginUrl($entry->getUrl());
 400                 }
 401                 $entry->setUrl($url);
 402                 break;
 403         }
 404     }
 405
 406     /**
 407      * Check entry url against an ignore list to replace with content url.
 408      *
 409      * XXX: move the ignore list in the database to let users handle it
 410      *
 411      * @param string $url url to test
 412      *
 413      * @return bool true if url matches ignore list otherwise false
 414      */
 415     private function ignoreUrl($url)
 416     {
 417         $ignored_hosts = ['feedproxy.google.com', 'feeds.reuters.com'];
 418         $ignored_patterns = ['https?://www\.lemonde\.fr/tiny.*'];
 419
 420         $parsed_url = parse_url($url);
 421
 422         $filtered = array_filter($ignored_hosts, function ($var) use ($parsed_url) {
 423             return $var === $parsed_url['host'];
 424         });
 425
 426         if ([] !== $filtered) {
 427             return true;
 428         }
 429
 430         $filtered = array_filter($ignored_patterns, function ($var) use ($url) {
 431             return preg_match("`$var`i", $url);
 432         });
 433
 434         if ([] !== $filtered) {
 435             return true;
 436         }
 437
 438         return false;
 439     }
 440
 441     /**
 442      * Validate that the given content has at least a title, an html and a url.
 443      *
 444      * @param array $content
 445      *
 446      * @return bool true if valid otherwise false
 447      */
 448     private function validateContent(array $content)
 449     {
 450         return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);
 451     }
 452 }