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 Wallabag\CoreBundle\Entity\Entry;
   8 use Wallabag\CoreBundle\Tools\Utils;
   9 use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
  10
  11 /**
  12  * This kind of proxy class take care of getting the content from an url
  13  * and update the entry with what it found.
  14  */
  15 class ContentProxy
  16 {
  17     protected $graby;
  18     protected $tagger;
  19     protected $logger;
  20     protected $mimeGuesser;
  21     protected $fetchingErrorMessage;
  22     protected $eventDispatcher;
  23
  24     public function __construct(Graby $graby, RuleBasedTagger $tagger, LoggerInterface $logger, $fetchingErrorMessage)
  25     {
  26         $this->graby = $graby;
  27         $this->tagger = $tagger;
  28         $this->logger = $logger;
  29         $this->mimeGuesser = new MimeTypeExtensionGuesser();
  30         $this->fetchingErrorMessage = $fetchingErrorMessage;
  31     }
  32
  33     /**
  34      * Fetch content using graby and hydrate given entry with results information.
  35      * In case we couldn't find content, we'll try to use Open Graph data.
  36      *
  37      * We can also force the content, in case of an import from the v1 for example, so the function won't
  38      * fetch the content from the website but rather use information given with the $content parameter.
  39      *
  40      * @param Entry  $entry   Entry to update
  41      * @param string $url     Url to grab content for
  42      * @param array  $content An array with AT LEAST keys title, html, url, language & content_type to skip the fetchContent from the url
  43      *
  44      * @return Entry
  45      */
  46     public function updateEntry(Entry $entry, $url, array $content = [])
  47     {
  48         // do we have to fetch the content or the provided one is ok?
  49         if (empty($content) || false === $this->validateContent($content)) {
  50             $fetchedContent = $this->graby->fetchContent($url);
  51
  52             // when content is imported, we have information in $content
  53             // in case fetching content goes bad, we'll keep the imported information instead of overriding them
  54             if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage) {
  55                 $content = $fetchedContent;
  56             }
  57         }
  58
  59         $title = $content['title'];
  60         if (!$title && isset($content['open_graph']['og_title'])) {
  61             $title = $content['open_graph']['og_title'];
  62         }
  63
  64         $html = $content['html'];
  65         if (false === $html) {
  66             $html = $this->fetchingErrorMessage;
  67
  68             if (isset($content['open_graph']['og_description'])) {
  69                 $html .= '<p><i>But we found a short description: </i></p>';
  70                 $html .= $content['open_graph']['og_description'];
  71             }
  72         }
  73
  74         $entry->setUrl($content['url'] ?: $url);
  75         $entry->setTitle($title);
  76         $entry->setContent($html);
  77         $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
  78
  79         if (isset($content['date']) && null !== $content['date'] && '' !== $content['date']) {
  80             $entry->setPublishedAt(new \DateTime($content['date']));
  81         }
  82
  83         if (!empty($content['authors'])) {
  84             $entry->setPublishedBy($content['authors']);
  85         }
  86
  87         if (!empty($content['all_headers'])) {
  88             $entry->setHeaders($content['all_headers']);
  89         }
  90
  91         $entry->setLanguage(isset($content['language']) ? $content['language'] : '');
  92         $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : '');
  93         $entry->setReadingTime(Utils::getReadingTime($html));
  94
  95         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
  96         if (false !== $domainName) {
  97             $entry->setDomainName($domainName);
  98         }
  99
 100         if (isset($content['open_graph']['og_image']) && $content['open_graph']['og_image']) {
 101             $entry->setPreviewPicture($content['open_graph']['og_image']);
 102         }
 103
 104         // if content is an image define as a preview too
 105         if (isset($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
 106             $entry->setPreviewPicture($content['url']);
 107         }
 108
 109         try {
 110             $this->tagger->tag($entry);
 111         } catch (\Exception $e) {
 112             $this->logger->error('Error while trying to automatically tag an entry.', [
 113                 'entry_url' => $url,
 114                 'error_msg' => $e->getMessage(),
 115             ]);
 116         }
 117
 118         return $entry;
 119     }
 120
 121     /**
 122      * Validate that the given content as enough value to be used
 123      * instead of fetch the content from the url.
 124      *
 125      * @param array $content
 126      *
 127      * @return bool true if valid otherwise false
 128      */
 129     private function validateContent(array $content)
 130     {
 131         return isset($content['title']) && isset($content['html']) && isset($content['url']) && isset($content['language']) && isset($content['content_type']);
 132     }
 133 }