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      * Stock entry with fetched or imported content.
  71      * Will fall back to OpenGraph data if available.
  72      *
  73      * @param Entry $entry   Entry to stock
  74      * @param array $content Array with at least title, url & html
  75      */
  76     private function stockEntry(Entry $entry, array $content)
  77     {
  78         $title = $content['title'];
  79         if (!$title && !empty($content['open_graph']['og_title'])) {
  80             $title = $content['open_graph']['og_title'];
  81         }
  82
  83         $html = $content['html'];
  84         if (false === $html) {
  85             $html = $this->fetchingErrorMessage;
  86
  87             if (!empty($content['open_graph']['og_description'])) {
  88                 $html .= '<p><i>But we found a short description: </i></p>';
  89                 $html .= $content['open_graph']['og_description'];
  90             }
  91         }
  92
  93         $entry->setUrl($content['url']);
  94         $entry->setTitle($title);
  95         $entry->setContent($html);
  96         $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
  97
  98         if (!empty($content['date'])) {
  99             $date = $content['date'];
 100
 101             // is it a timestamp?
 102             if (filter_var($date, FILTER_VALIDATE_INT) !== false) {
 103                 $date = '@' . $content['date'];
 104             }
 105
 106             try {
 107                 $entry->setPublishedAt(new \DateTime($date));
 108             } catch (\Exception $e) {
 109                 $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $content['url'], 'date' => $content['date']]);
 110             }
 111         }
 112
 113         if (!empty($content['authors']) && is_array($content['authors'])) {
 114             $entry->setPublishedBy($content['authors']);
 115         }
 116
 117         if (!empty($content['all_headers'])) {
 118             $entry->setHeaders($content['all_headers']);
 119         }
 120
 121         $this->validateAndSetLanguage(
 122             $entry,
 123             isset($content['language']) ? $content['language'] : null
 124         );
 125
 126         $this->validateAndSetPreviewPicture(
 127             $entry,
 128             isset($content['open_graph']['og_image']) ? $content['open_graph']['og_image'] : null
 129         );
 130
 131         // if content is an image, define it as a preview too
 132         if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
 133             $this->validateAndSetPreviewPicture(
 134                 $entry,
 135                 $content['url']
 136             );
 137         }
 138
 139         $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : '');
 140         $entry->setReadingTime(Utils::getReadingTime($html));
 141
 142         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
 143         if (false !== $domainName) {
 144             $entry->setDomainName($domainName);
 145         }
 146
 147         try {
 148             $this->tagger->tag($entry);
 149         } catch (\Exception $e) {
 150             $this->logger->error('Error while trying to automatically tag an entry.', [
 151                 'entry_url' => $content['url'],
 152                 'error_msg' => $e->getMessage(),
 153             ]);
 154         }
 155     }
 156
 157     /**
 158      * Validate that the given content has at least a title, an html and a url.
 159      *
 160      * @param array $content
 161      *
 162      * @return bool true if valid otherwise false
 163      */
 164     private function validateContent(array $content)
 165     {
 166         return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);
 167     }
 168
 169     /**
 170      * Use a Symfony validator to ensure the language is well formatted.
 171      *
 172      * @param Entry  $entry
 173      * @param string $value Language to validate
 174      */
 175     private function validateAndSetLanguage($entry, $value)
 176     {
 177         // some lang are defined as fr-FR, es-ES.
 178         // replacing - by _ might increase language support
 179         $value = str_replace('-', '_', $value);
 180
 181         $errors = $this->validator->validate(
 182             $value,
 183             (new LocaleConstraint())
 184         );
 185
 186         if (0 === count($errors)) {
 187             $entry->setLanguage($value);
 188
 189             return;
 190         }
 191
 192         $this->logger->warning('Language validation failed. ' . (string) $errors);
 193     }
 194
 195     /**
 196      * Use a Symfony validator to ensure the preview picture is a real url.
 197      *
 198      * @param Entry  $entry
 199      * @param string $value URL to validate
 200      */
 201     private function validateAndSetPreviewPicture($entry, $value)
 202     {
 203         $errors = $this->validator->validate(
 204             $value,
 205             (new UrlConstraint())
 206         );
 207
 208         if (0 === count($errors)) {
 209             $entry->setPreviewPicture($value);
 210
 211             return;
 212         }
 213
 214         $this->logger->warning('PreviewPicture validation failed. ' . (string) $errors);
 215     }
 216 }