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      * Update entry using either fetched or provided content.
  35      *
  36      * @param Entry  $entry                Entry to update
  37      * @param string $url                  Url of the content
  38      * @param array  $content              Array with content provided for import with AT LEAST keys title, html, url to skip the fetchContent from the url
  39      * @param bool   $disableContentUpdate Whether to skip trying to fetch content using Graby
  40      */
  41     public function updateEntry(Entry $entry, $url, array $content = [], $disableContentUpdate = false)
  42     {
  43         if (!empty($content['html'])) {
  44             $content['html'] = $this->graby->cleanupHtml($content['html'], $url);
  45         }
  46
  47         if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) {
  48             $fetchedContent = $this->graby->fetchContent($url);
  49
  50             // when content is imported, we have information in $content
  51             // in case fetching content goes bad, we'll keep the imported information instead of overriding them
  52             if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage) {
  53                 $content = $fetchedContent;
  54             }
  55         }
  56
  57         // be sure to keep the url in case of error
  58         // so we'll be able to refetch it in the future
  59         $content['url'] = !empty($content['url']) ? $content['url'] : $url;
  60
  61         $this->stockEntry($entry, $content);
  62     }
  63
  64     /**
  65      * Stock entry with fetched or imported content.
  66      * Will fall back to OpenGraph data if available.
  67      *
  68      * @param Entry $entry   Entry to stock
  69      * @param array $content Array with at least title, url & html
  70      */
  71     private function stockEntry(Entry $entry, array $content)
  72     {
  73         $title = $content['title'];
  74         if (!$title && !empty($content['open_graph']['og_title'])) {
  75             $title = $content['open_graph']['og_title'];
  76         }
  77
  78         $html = $content['html'];
  79         if (false === $html) {
  80             $html = $this->fetchingErrorMessage;
  81
  82             if (!empty($content['open_graph']['og_description'])) {
  83                 $html .= '<p><i>But we found a short description: </i></p>';
  84                 $html .= $content['open_graph']['og_description'];
  85             }
  86         }
  87
  88         $entry->setUrl($content['url']);
  89         $entry->setTitle($title);
  90         $entry->setContent($html);
  91         $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
  92
  93         if (!empty($content['date'])) {
  94             $date = $content['date'];
  95
  96             // is it a timestamp?
  97             if (filter_var($date, FILTER_VALIDATE_INT) !== false) {
  98                 $date = '@'.$content['date'];
  99             }
 100
 101             try {
 102                 $entry->setPublishedAt(new \DateTime($date));
 103             } catch (\Exception $e) {
 104                 $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $content['url'], 'date' => $content['date']]);
 105             }
 106         }
 107
 108         if (!empty($content['authors']) && is_array($content['authors'])) {
 109             $entry->setPublishedBy($content['authors']);
 110         }
 111
 112         if (!empty($content['all_headers'])) {
 113             $entry->setHeaders($content['all_headers']);
 114         }
 115
 116         $entry->setLanguage(isset($content['language']) ? $content['language'] : '');
 117         $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : '');
 118         $entry->setReadingTime(Utils::getReadingTime($html));
 119
 120         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
 121         if (false !== $domainName) {
 122             $entry->setDomainName($domainName);
 123         }
 124
 125         if (!empty($content['open_graph']['og_image'])) {
 126             $entry->setPreviewPicture($content['open_graph']['og_image']);
 127         }
 128
 129         // if content is an image define as a preview too
 130         if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
 131             $entry->setPreviewPicture($content['url']);
 132         }
 133
 134         try {
 135             $this->tagger->tag($entry);
 136         } catch (\Exception $e) {
 137             $this->logger->error('Error while trying to automatically tag an entry.', [
 138                 'entry_url' => $content['url'],
 139                 'error_msg' => $e->getMessage(),
 140             ]);
 141         }
 142     }
 143
 144     /**
 145      * Validate that the given content has at least a title, an html and a url.
 146      *
 147      * @param array $content
 148      *
 149      * @return bool true if valid otherwise false
 150      */
 151     private function validateContent(array $content)
 152     {
 153         return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);
 154     }
 155 }