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\Validator\Constraints\Language as LanguageConstraint;
  11 use Symfony\Component\Validator\Constraints\Url as UrlConstraint;
  12 use Symfony\Component\Validator\Validator\ValidatorInterface;
  13
  14 /**
  15  * This kind of proxy class take care of getting the content from an url
  16  * and update the entry with what it found.
  17  */
  18 class ContentProxy
  19 {
  20     protected $graby;
  21     protected $tagger;
  22     protected $logger;
  23     protected $mimeGuesser;
  24     protected $fetchingErrorMessage;
  25     protected $eventDispatcher;
  26
  27     public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage)
  28     {
  29         $this->graby = $graby;
  30         $this->tagger = $tagger;
  31         $this->validator = $validator;
  32         $this->logger = $logger;
  33         $this->mimeGuesser = new MimeTypeExtensionGuesser();
  34         $this->fetchingErrorMessage = $fetchingErrorMessage;
  35     }
  36
  37     /**
  38      * Update entry using either fetched or provided content.
  39      *
  40      * @param Entry  $entry                Entry to update
  41      * @param string $url                  Url of the content
  42      * @param array  $content              Array with content provided for import with AT LEAST keys title, html, url to skip the fetchContent from the url
  43      * @param bool   $disableContentUpdate Whether to skip trying to fetch content using Graby
  44      */
  45     public function updateEntry(Entry $entry, $url, array $content = [], $disableContentUpdate = false)
  46     {
  47         if (!empty($content['html'])) {
  48             $content['html'] = $this->graby->cleanupHtml($content['html'], $url);
  49         }
  50
  51         if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) {
  52             $fetchedContent = $this->graby->fetchContent($url);
  53
  54             // when content is imported, we have information in $content
  55             // in case fetching content goes bad, we'll keep the imported information instead of overriding them
  56             if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage) {
  57                 $content = $fetchedContent;
  58             }
  59         }
  60
  61         // be sure to keep the url in case of error
  62         // so we'll be able to refetch it in the future
  63         $content['url'] = !empty($content['url']) ? $content['url'] : $url;
  64
  65         $this->stockEntry($entry, $content);
  66     }
  67
  68     /**
  69      * Stock entry with fetched or imported content.
  70      * Will fall back to OpenGraph data if available.
  71      *
  72      * @param Entry $entry   Entry to stock
  73      * @param array $content Array with at least title, url & html
  74      */
  75     private function stockEntry(Entry $entry, array $content)
  76     {
  77         $title = $content['title'];
  78         if (!$title && !empty($content['open_graph']['og_title'])) {
  79             $title = $content['open_graph']['og_title'];
  80         }
  81
  82         $html = $content['html'];
  83         if (false === $html) {
  84             $html = $this->fetchingErrorMessage;
  85
  86             if (!empty($content['open_graph']['og_description'])) {
  87                 $html .= '<p><i>But we found a short description: </i></p>';
  88                 $html .= $content['open_graph']['og_description'];
  89             }
  90         }
  91
  92         $entry->setUrl($content['url']);
  93         $entry->setTitle($title);
  94         $entry->setContent($html);
  95         $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
  96
  97         if (!empty($content['date'])) {
  98             $date = $content['date'];
  99
 100             // is it a timestamp?
 101             if (filter_var($date, FILTER_VALIDATE_INT) !== false) {
 102                 $date = '@'.$content['date'];
 103             }
 104
 105             try {
 106                 $entry->setPublishedAt(new \DateTime($date));
 107             } catch (\Exception $e) {
 108                 $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $content['url'], 'date' => $content['date']]);
 109             }
 110         }
 111
 112         if (!empty($content['authors']) && is_array($content['authors'])) {
 113             $entry->setPublishedBy($content['authors']);
 114         }
 115
 116         if (!empty($content['all_headers'])) {
 117             $entry->setHeaders($content['all_headers']);
 118         }
 119
 120         $this->validateAndSetLanguage(
 121             $entry,
 122             isset($content['language']) ? $content['language'] : ''
 123         );
 124
 125         $this->validateAndSetPreviewPicture(
 126             $entry,
 127             isset($content['open_graph']['og_image']) ? $content['open_graph']['og_image'] : ''
 128         );
 129
 130         // if content is an image define as a preview too
 131         if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
 132             $this->validateAndSetPreviewPicture(
 133                 $entry,
 134                 $content['url']
 135             );
 136         }
 137
 138         $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : '');
 139         $entry->setReadingTime(Utils::getReadingTime($html));
 140
 141         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
 142         if (false !== $domainName) {
 143             $entry->setDomainName($domainName);
 144         }
 145
 146         try {
 147             $this->tagger->tag($entry);
 148         } catch (\Exception $e) {
 149             $this->logger->error('Error while trying to automatically tag an entry.', [
 150                 'entry_url' => $content['url'],
 151                 'error_msg' => $e->getMessage(),
 152             ]);
 153         }
 154     }
 155
 156     /**
 157      * Validate that the given content has at least a title, an html and a url.
 158      *
 159      * @param array $content
 160      *
 161      * @return bool true if valid otherwise false
 162      */
 163     private function validateContent(array $content)
 164     {
 165         return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);
 166     }
 167
 168     /**
 169      * Use a Symfony validator to ensure the language is well formatted.
 170      *
 171      * @param Entry  $entry
 172      * @param string $value Language to validate
 173      */
 174     private function validateAndSetLanguage($entry, $value)
 175     {
 176         $errors = $this->validator->validate(
 177             $value,
 178             (new LanguageConstraint())
 179         );
 180
 181         if (0 === count($errors)) {
 182             $entry->setLanguage($value);
 183
 184             return;
 185         }
 186
 187         $this->logger->warning('Language validation failed. '.(string) $errors);
 188     }
 189
 190     /**
 191      * Use a Symfony validator to ensure the preview picture is a real url.
 192      *
 193      * @param Entry  $entry
 194      * @param string $value URL to validate
 195      */
 196     private function validateAndSetPreviewPicture($entry, $value)
 197     {
 198         $errors = $this->validator->validate(
 199             $value,
 200             (new UrlConstraint())
 201         );
 202
 203         if (0 === count($errors)) {
 204             $entry->setPreviewPicture($value);
 205
 206             return;
 207         }
 208
 209         $this->logger->warning('PreviewPicture validation failed. '.(string) $errors);
 210     }
 211 }