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         // be sure to keep the url in case of error
  45         // so we'll be able to refetch it in the future
  46         $content['url'] = $content['url'] ?: $url;
  47
  48         $this->stockEntry($entry, $content);
  49     }
  50
  51     /**
  52      * Import entry using either fetched or provided content.
  53      *
  54      * @param Entry  $entry                Entry to update
  55      * @param array  $content              Array with content provided for import with AT LEAST keys title, html, url to skip the fetchContent from the url
  56      * @param bool   $disableContentUpdate Whether to skip trying to fetch content using Graby
  57      */
  58     public function importEntry(Entry $entry, array $content, $disableContentUpdate = false)
  59     {
  60         try {
  61             $this->validateContent($content);
  62         } catch (\Exception $e) {
  63             // validation failed but do we want to disable updating content?
  64             if (true === $disableContentUpdate) {
  65                 throw $e;
  66             }
  67         }
  68
  69         if (false === $disableContentUpdate) {
  70             try {
  71                 $fetchedContent = $this->graby->fetchContent($content['url']);
  72             } catch (\Exception $e) {
  73                 $this->logger->error('Error while trying to fetch content from URL.', [
  74                     'entry_url' => $content['url'],
  75                     'error_msg' => $e->getMessage(),
  76                 ]);
  77             }
  78
  79             // when content is imported, we have information in $content
  80             // in case fetching content goes bad, we'll keep the imported information instead of overriding them
  81             if ($fetchedContent['html'] !== $this->fetchingErrorMessage) {
  82                 $content = $fetchedContent;
  83             }
  84         }
  85
  86         $this->stockEntry($entry, $content);
  87     }
  88
  89     /**
  90      * Stock entry with fetched or imported content.
  91      * Will fall back to OpenGraph data if available.
  92      *
  93      * @param Entry $entry   Entry to stock
  94      * @param array $content Array with at least title and URL
  95      */
  96     private function stockEntry(Entry $entry, array $content)
  97     {
  98         $title = $content['title'];
  99         if (!$title && !empty($content['open_graph']['og_title'])) {
 100             $title = $content['open_graph']['og_title'];
 101         }
 102
 103         $html = $content['html'];
 104         if (false === $html) {
 105             $html = $this->fetchingErrorMessage;
 106
 107             if (!empty($content['open_graph']['og_description'])) {
 108                 $html .= '<p><i>But we found a short description: </i></p>';
 109                 $html .= $content['open_graph']['og_description'];
 110             }
 111         }
 112
 113         $entry->setUrl($content['url']);
 114         $entry->setTitle($title);
 115         $entry->setContent($html);
 116         $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
 117
 118         if (!empty($content['date'])) {
 119             $date = $content['date'];
 120
 121             // is it a timestamp?
 122             if (filter_var($date, FILTER_VALIDATE_INT) !== false) {
 123                 $date = '@'.$content['date'];
 124             }
 125
 126             try {
 127                 $entry->setPublishedAt(new \DateTime($date));
 128             } catch (\Exception $e) {
 129                 $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $url, 'date' => $content['date']]);
 130             }
 131         }
 132
 133         if (!empty($content['authors'])) {
 134             $entry->setPublishedBy($content['authors']);
 135         }
 136
 137         if (!empty($content['all_headers'])) {
 138             $entry->setHeaders($content['all_headers']);
 139         }
 140
 141         $entry->setLanguage(isset($content['language']) ? $content['language'] : '');
 142         $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : '');
 143         $entry->setReadingTime(Utils::getReadingTime($html));
 144
 145         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
 146         if (false !== $domainName) {
 147             $entry->setDomainName($domainName);
 148         }
 149
 150         if (!empty($content['open_graph']['og_image'])) {
 151             $entry->setPreviewPicture($content['open_graph']['og_image']);
 152         }
 153
 154         // if content is an image define as a preview too
 155         if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
 156             $entry->setPreviewPicture($content['url']);
 157         }
 158
 159         try {
 160             $this->tagger->tag($entry);
 161         } catch (\Exception $e) {
 162             $this->logger->error('Error while trying to automatically tag an entry.', [
 163                 'entry_url' => $content['url'],
 164                 'error_msg' => $e->getMessage(),
 165             ]);
 166         }
 167     }
 168
 169     /**
 170      * Validate that the given content has at least a title, an html and a url.
 171      *
 172      * @param array $content
 173      */
 174     private function validateContent(array $content)
 175     {
 176         if (empty($content['title'])) {
 177             throw new Exception('Missing title from imported entry!');
 178         }
 179
 180         if (empty($content['url'])) {
 181             throw new Exception('Missing URL from imported entry!');
 182         }
 183
 184         if (empty($content['html'])) {
 185             throw new Exception('Missing html from imported entry!');
 186         }
 187     }
 188 }