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 use Symfony\Component\Config\Definition\Exception\Exception;
  11
  12 /**
  13  * This kind of proxy class take care of getting the content from an url
  14  * and update the entry with what it found.
  15  */
  16 class ContentProxy
  17 {
  18     protected $graby;
  19     protected $tagger;
  20     protected $logger;
  21     protected $mimeGuesser;
  22     protected $fetchingErrorMessage;
  23     protected $eventDispatcher;
  24
  25     public function __construct(Graby $graby, RuleBasedTagger $tagger, LoggerInterface $logger, $fetchingErrorMessage)
  26     {
  27         $this->graby = $graby;
  28         $this->tagger = $tagger;
  29         $this->logger = $logger;
  30         $this->mimeGuesser = new MimeTypeExtensionGuesser();
  31         $this->fetchingErrorMessage = $fetchingErrorMessage;
  32     }
  33
  34     /**
  35      * Update existing entry by fetching from URL using Graby.
  36      *
  37      * @param Entry  $entry   Entry to update
  38      * @param string $url     Url to grab content for
  39      */
  40     public function updateEntry(Entry $entry, $url)
  41     {
  42         $content = $this->graby->fetchContent($url);
  43
  44         $this->stockEntry($entry, $content);
  45     }
  46
  47     /**
  48      * Import entry using either fetched or provided content.
  49      *
  50      * @param Entry  $entry                Entry to update
  51      * @param array  $content              Array with content provided for import with AT LEAST keys title, html, url to skip the fetchContent from the url
  52      * @param bool   $disableContentUpdate Whether to skip trying to fetch content using Graby
  53      */
  54     public function importEntry(Entry $entry, array $content, $disableContentUpdate = false)
  55     {
  56         $this->validateContent($content);
  57
  58         if (false === $disableContentUpdate) {
  59             try {
  60                 $fetchedContent = $this->graby->fetchContent($content['url']);
  61             } catch (\Exception $e) {
  62                 $this->logger->error('Error while trying to fetch content from URL.', [
  63                     'entry_url' => $content['url'],
  64                     'error_msg' => $e->getMessage(),
  65                 ]);
  66             }
  67
  68             // when content is imported, we have information in $content
  69             // in case fetching content goes bad, we'll keep the imported information instead of overriding them
  70             if ($fetchedContent['html'] !== $this->fetchingErrorMessage) {
  71                 $content = $fetchedContent;
  72             }
  73         }
  74
  75         $this->stockEntry($entry, $content);
  76     }
  77
  78     /**
  79      * Stock entry with fetched or imported content.
  80      * Will fall back to OpenGraph data if available.
  81      *
  82      * @param Entry  $entry   Entry to stock
  83      * @param array  $content Array with at least title and URL
  84      */
  85     private function stockEntry(Entry $entry, array $content)
  86     {
  87         $title = $content['title'];
  88         if (!$title && !empty($content['open_graph']['og_title'])) {
  89             $title = $content['open_graph']['og_title'];
  90         }
  91
  92         $html = $content['html'];
  93         if (false === $html) {
  94             $html = $this->fetchingErrorMessage;
  95
  96             if (!empty($content['open_graph']['og_description'])) {
  97                 $html .= '<p><i>But we found a short description: </i></p>';
  98                 $html .= $content['open_graph']['og_description'];
  99             }
 100         }
 101
 102         $entry->setUrl($content['url']);
 103         $entry->setTitle($title);
 104         $entry->setContent($html);
 105         $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
 106
 107         if (!empty($content['date'])) {
 108             $date = $content['date'];
 109
 110             // is it a timestamp?
 111             if (filter_var($date, FILTER_VALIDATE_INT) !== false) {
 112                 $date = '@'.$content['date'];
 113             }
 114
 115             try {
 116                 $entry->setPublishedAt(new \DateTime($date));
 117             } catch (\Exception $e) {
 118                 $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $url, 'date' => $content['date']]);
 119             }
 120         }
 121
 122         if (!empty($content['authors'])) {
 123             $entry->setPublishedBy($content['authors']);
 124         }
 125
 126         if (!empty($content['all_headers'])) {
 127             $entry->setHeaders($content['all_headers']);
 128         }
 129
 130         $entry->setLanguage(isset($content['language']) ? $content['language'] : '');
 131         $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : '');
 132         $entry->setReadingTime(Utils::getReadingTime($html));
 133
 134         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
 135         if (false !== $domainName) {
 136             $entry->setDomainName($domainName);
 137         }
 138
 139         if (!empty($content['open_graph']['og_image'])) {
 140             $entry->setPreviewPicture($content['open_graph']['og_image']);
 141         }
 142
 143         // if content is an image define as a preview too
 144         if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
 145             $entry->setPreviewPicture($content['url']);
 146         }
 147
 148         try {
 149             $this->tagger->tag($entry);
 150         } catch (\Exception $e) {
 151             $this->logger->error('Error while trying to automatically tag an entry.', [
 152                 'entry_url' => $content['url'],
 153                 'error_msg' => $e->getMessage(),
 154             ]);
 155         }
 156     }
 157
 158     /**
 159      * Validate that the given content has at least a title, an html and a url.
 160      *
 161      * @param array $content
 162      */
 163     private function validateContent(array $content)
 164     {
 165         if (!empty($content['title']))) {
 166             throw new Exception('Missing title from imported entry!');
 167         }
 168
 169         if (!empty($content['url']))) {
 170             throw new Exception('Missing URL from imported entry!');
 171         }
 172
 173         if (!empty($content['html']))) {
 174             throw new Exception('Missing html from imported entry!');
 175         }
 176     }
 177 }