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 Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
   8 use Symfony\Component\Validator\Constraints\Locale as LocaleConstraint;
   9 use Symfony\Component\Validator\Constraints\Url as UrlConstraint;
  10 use Symfony\Component\Validator\Validator\ValidatorInterface;
  11 use Wallabag\CoreBundle\Entity\Entry;
  12 use Wallabag\CoreBundle\Tools\Utils;
  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 $validator;
  23     protected $logger;
  24     protected $mimeGuesser;
  25     protected $fetchingErrorMessage;
  26     protected $eventDispatcher;
  27
  28     public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage)
  29     {
  30         $this->graby = $graby;
  31         $this->tagger = $tagger;
  32         $this->validator = $validator;
  33         $this->logger = $logger;
  34         $this->mimeGuesser = new MimeTypeExtensionGuesser();
  35         $this->fetchingErrorMessage = $fetchingErrorMessage;
  36     }
  37
  38     /**
  39      * Update entry using either fetched or provided content.
  40      *
  41      * @param Entry  $entry                Entry to update
  42      * @param string $url                  Url of the content
  43      * @param array  $content              Array with content provided for import with AT LEAST keys title, html, url to skip the fetchContent from the url
  44      * @param bool   $disableContentUpdate Whether to skip trying to fetch content using Graby
  45      */
  46     public function updateEntry(Entry $entry, $url, array $content = [], $disableContentUpdate = false)
  47     {
  48         if (!empty($content['html'])) {
  49             $content['html'] = $this->graby->cleanupHtml($content['html'], $url);
  50         }
  51
  52         if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) {
  53             $fetchedContent = $this->graby->fetchContent($url);
  54
  55             // when content is imported, we have information in $content
  56             // in case fetching content goes bad, we'll keep the imported information instead of overriding them
  57             if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage) {
  58                 $content = $fetchedContent;
  59             }
  60         }
  61
  62         // be sure to keep the url in case of error
  63         // so we'll be able to refetch it in the future
  64         $content['url'] = !empty($content['url']) ? $content['url'] : $url;
  65
  66         $this->stockEntry($entry, $content);
  67     }
  68
  69     /**
  70      * Use a Symfony validator to ensure the language is well formatted.
  71      *
  72      * @param Entry  $entry
  73      * @param string $value Language to validate and save
  74      */
  75     public function updateLanguage(Entry $entry, $value)
  76     {
  77         // some lang are defined as fr-FR, es-ES.
  78         // replacing - by _ might increase language support
  79         $value = str_replace('-', '_', $value);
  80
  81         $errors = $this->validator->validate(
  82             $value,
  83             (new LocaleConstraint())
  84         );
  85
  86         if (0 === count($errors)) {
  87             $entry->setLanguage($value);
  88
  89             return;
  90         }
  91
  92         $this->logger->warning('Language validation failed. ' . (string) $errors);
  93     }
  94
  95     /**
  96      * Use a Symfony validator to ensure the preview picture is a real url.
  97      *
  98      * @param Entry  $entry
  99      * @param string $value URL to validate and save
 100      */
 101     public function updatePreviewPicture(Entry $entry, $value)
 102     {
 103         $errors = $this->validator->validate(
 104             $value,
 105             (new UrlConstraint())
 106         );
 107
 108         if (0 === count($errors)) {
 109             $entry->setPreviewPicture($value);
 110
 111             return;
 112         }
 113
 114         $this->logger->warning('PreviewPicture validation failed. ' . (string) $errors);
 115     }
 116
 117     /**
 118      * Update date.
 119      *
 120      * @param Entry  $entry
 121      * @param string $value Date to validate and save
 122      */
 123     public function updatePublishedAt(Entry $entry, $value)
 124     {
 125         $date = $value;
 126
 127         // is it a timestamp?
 128         if (filter_var($date, FILTER_VALIDATE_INT) !== false) {
 129             $date = '@' . $value;
 130         }
 131
 132         try {
 133             $entry->setPublishedAt(new \DateTime($date));
 134         } catch (\Exception $e) {
 135             $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $entry->getUrl(), 'date' => $value]);
 136         }
 137     }
 138
 139     /**
 140      * Stock entry with fetched or imported content.
 141      * Will fall back to OpenGraph data if available.
 142      *
 143      * @param Entry $entry   Entry to stock
 144      * @param array $content Array with at least title, url & html
 145      */
 146     private function stockEntry(Entry $entry, array $content)
 147     {
 148         $entry->setUrl($content['url']);
 149
 150         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
 151         if (false !== $domainName) {
 152             $entry->setDomainName($domainName);
 153         }
 154
 155         if (!empty($content['title'])) {
 156             $entry->setTitle($content['title']);
 157         } elseif (!empty($content['open_graph']['og_title'])) {
 158             $entry->setTitle($content['open_graph']['og_title']);
 159         }
 160
 161         $html = $content['html'];
 162         if (false === $html) {
 163             $html = $this->fetchingErrorMessage;
 164
 165             if (!empty($content['open_graph']['og_description'])) {
 166                 $html .= '<p><i>But we found a short description: </i></p>';
 167                 $html .= $content['open_graph']['og_description'];
 168             }
 169         }
 170
 171         $entry->setContent($html);
 172         $entry->setReadingTime(Utils::getReadingTime($html));
 173
 174         if (!empty($content['status'])) {
 175             $entry->setHttpStatus($content['status']);
 176         }
 177
 178         if (!empty($content['authors']) && is_array($content['authors'])) {
 179             $entry->setPublishedBy($content['authors']);
 180         }
 181
 182         if (!empty($content['all_headers'])) {
 183             $entry->setHeaders($content['all_headers']);
 184         }
 185
 186         if (!empty($content['date'])) {
 187             $this->updatePublishedAt($entry, $content['date']);
 188         }
 189
 190         if (!empty($content['language'])) {
 191             $this->updateLanguage($entry, $content['language']);
 192         }
 193
 194         if (!empty($content['open_graph']['og_image'])) {
 195             $this->updatePreviewPicture($entry, $content['open_graph']['og_image']);
 196         }
 197
 198         // if content is an image, define it as a preview too
 199         if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
 200             $this->updatePreviewPicture($entry, $content['url']);
 201         }
 202
 203         if (!empty($content['content_type'])) {
 204             $entry->setMimetype($content['content_type']);
 205         }
 206
 207         try {
 208             $this->tagger->tag($entry);
 209         } catch (\Exception $e) {
 210             $this->logger->error('Error while trying to automatically tag an entry.', [
 211                 'entry_url' => $content['url'],
 212                 'error_msg' => $e->getMessage(),
 213             ]);
 214         }
 215     }
 216
 217     /**
 218      * Validate that the given content has at least a title, an html and a url.
 219      *
 220      * @param array $content
 221      *
 222      * @return bool true if valid otherwise false
 223      */
 224     private function validateContent(array $content)
 225     {
 226         return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);
 227     }
 228 }