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             try {
  49                 $fetchedContent = $this->graby->fetchContent($url);
  50             } catch (\Exception $e) {
  51                 $this->logger->error('Error while trying to fetch content from URL.', [
  52                     'entry_url' => $url,
  53                     'error_msg' => $e->getMessage(),
  54                 ]);
  55             }
  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         // be sure to keep the url in case of error
  65         // so we'll be able to refetch it in the future
  66         $content['url'] = !empty($content['url']) ? $content['url'] : $url;
  67
  68         $this->stockEntry($entry, $content);
  69     }
  70
  71     /**
  72      * Stock entry with fetched or imported content.
  73      * Will fall back to OpenGraph data if available.
  74      *
  75      * @param Entry $entry   Entry to stock
  76      * @param array $content Array with at least title and URL
  77      */
  78     private function stockEntry(Entry $entry, array $content)
  79     {
  80         $title = $content['title'];
  81         if (!$title && !empty($content['open_graph']['og_title'])) {
  82             $title = $content['open_graph']['og_title'];
  83         }
  84
  85         $html = $content['html'];
  86         if (false === $html) {
  87             $html = $this->fetchingErrorMessage;
  88
  89             if (!empty($content['open_graph']['og_description'])) {
  90                 $html .= '<p><i>But we found a short description: </i></p>';
  91                 $html .= $content['open_graph']['og_description'];
  92             }
  93         }
  94
  95         $entry->setUrl($content['url']);
  96         $entry->setTitle($title);
  97         $entry->setContent($html);
  98         $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
  99
 100         if (!empty($content['date'])) {
 101             $date = $content['date'];
 102
 103             // is it a timestamp?
 104             if (filter_var($date, FILTER_VALIDATE_INT) !== false) {
 105                 $date = '@'.$content['date'];
 106             }
 107
 108             try {
 109                 $entry->setPublishedAt(new \DateTime($date));
 110             } catch (\Exception $e) {
 111                 $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $content['url'], 'date' => $content['date']]);
 112             }
 113         }
 114
 115         if (!empty($content['authors'])) {
 116             $entry->setPublishedBy($content['authors']);
 117         }
 118
 119         if (!empty($content['all_headers'])) {
 120             $entry->setHeaders($content['all_headers']);
 121         }
 122
 123         $entry->setLanguage(isset($content['language']) ? $content['language'] : '');
 124         $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : '');
 125         $entry->setReadingTime(Utils::getReadingTime($html));
 126
 127         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
 128         if (false !== $domainName) {
 129             $entry->setDomainName($domainName);
 130         }
 131
 132         if (!empty($content['open_graph']['og_image'])) {
 133             $entry->setPreviewPicture($content['open_graph']['og_image']);
 134         }
 135
 136         // if content is an image define as a preview too
 137         if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
 138             $entry->setPreviewPicture($content['url']);
 139         }
 140
 141         try {
 142             $this->tagger->tag($entry);
 143         } catch (\Exception $e) {
 144             $this->logger->error('Error while trying to automatically tag an entry.', [
 145                 'entry_url' => $content['url'],
 146                 'error_msg' => $e->getMessage(),
 147             ]);
 148         }
 149     }
 150
 151     /**
 152      * Validate that the given content has at least a title, an html and a url.
 153      *
 154      * @param array $content
 155      *
 156      * @return bool true if valid otherwise false
 157      */
 158     private function validateContent(array $content)
 159     {
 160         return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);
 161     }
 162 }