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 $content = [])
  47     {
  48         // ensure content is a bit cleaned up
  49         if (!empty($content['html'])) {
  50             $content['html'] = $this->graby->cleanupHtml($content['html'], $url);
  51         }
  52
  53         // do we have to fetch the content or the provided one is ok?
  54         if (empty($content) || false === $this->validateContent($content)) {
  55             $fetchedContent = $this->graby->fetchContent($url);
  56
  57             // when content is imported, we have information in $content
  58             // in case fetching content goes bad, we'll keep the imported information instead of overriding them
  59             if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage) {
  60                 $content = $fetchedContent;
  61             }
  62         }
  63
  64         $title = $content['title'];
  65         if (!$title && !empty($content['open_graph']['og_title'])) {
  66             $title = $content['open_graph']['og_title'];
  67         }
  68
  69         $html = $content['html'];
  70         if (false === $html) {
  71             $html = $this->fetchingErrorMessage;
  72
  73             if (!empty($content['open_graph']['og_description'])) {
  74                 $html .= '<p><i>But we found a short description: </i></p>';
  75                 $html .= $content['open_graph']['og_description'];
  76             }
  77         }
  78
  79         $entry->setUrl($content['url'] ?: $url);
  80         $entry->setTitle($title);
  81         $entry->setContent($html);
  82         $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
  83
  84         if (!empty($content['date'])) {
  85             try {
  86                 $entry->setPublishedAt(new \DateTime($content['date']));
  87             } catch (\Exception $e) {
  88                 $this->logger->warn('Error while defining date', ['e' => $e, 'url' => $url, 'date' => $content['date']]);
  89             }
  90         }
  91
  92         if (!empty($content['authors'])) {
  93             $entry->setPublishedBy($content['authors']);
  94         }
  95
  96         if (!empty($content['all_headers'])) {
  97             $entry->setHeaders($content['all_headers']);
  98         }
  99
 100         $entry->setLanguage(isset($content['language']) ? $content['language'] : '');
 101         $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : '');
 102         $entry->setReadingTime(Utils::getReadingTime($html));
 103
 104         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
 105         if (false !== $domainName) {
 106             $entry->setDomainName($domainName);
 107         }
 108
 109         if (!empty($content['open_graph']['og_image'])) {
 110             $entry->setPreviewPicture($content['open_graph']['og_image']);
 111         }
 112
 113         // if content is an image define as a preview too
 114         if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
 115             $entry->setPreviewPicture($content['url']);
 116         }
 117
 118         try {
 119             $this->tagger->tag($entry);
 120         } catch (\Exception $e) {
 121             $this->logger->error('Error while trying to automatically tag an entry.', [
 122                 'entry_url' => $url,
 123                 'error_msg' => $e->getMessage(),
 124             ]);
 125         }
 126
 127         return $entry;
 128     }
 129
 130     /**
 131      * Validate that the given content as enough value to be used
 132      * instead of fetch the content from the url.
 133      *
 134      * @param array $content
 135      *
 136      * @return bool true if valid otherwise false
 137      */
 138     private function validateContent(array $content)
 139     {
 140         return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);
 141     }
 142 }