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         $entry->setUrl($content['url']);
  79
  80         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
  81         if (false !== $domainName) {
  82             $entry->setDomainName($domainName);
  83         }
  84
  85         if (!empty($content['title'])) {
  86             $entry->setTitle($content['title']);
  87         } elseif (!empty($content['open_graph']['og_title'])) {
  88             $entry->setTitle($content['open_graph']['og_title']);
  89         }
  90
  91         $html = $content['html'];
  92         if (false === $html) {
  93             $html = $this->fetchingErrorMessage;
  94
  95             if (!empty($content['open_graph']['og_description'])) {
  96                 $html .= '<p><i>But we found a short description: </i></p>';
  97                 $html .= $content['open_graph']['og_description'];
  98             }
  99         }
 100
 101         $entry->setContent($html);
 102         $entry->setReadingTime(Utils::getReadingTime($html));
 103
 104         if (!empty($content['status'])) {
 105             $entry->setHttpStatus($content['status']);
 106         }
 107
 108         if (!empty($content['authors']) && is_array($content['authors'])) {
 109             $entry->setPublishedBy($content['authors']);
 110         }
 111
 112         if (!empty($content['all_headers'])) {
 113             $entry->setHeaders($content['all_headers']);
 114         }
 115
 116         if (!empty($content['date'])) {
 117             $this->updatePublishedAt($entry, $content['date']);
 118         }
 119
 120         if (!empty($content['language'])) {
 121             $this->updateLanguage($entry, $content['language']);
 122         }
 123
 124         if (!empty($content['open_graph']['og_image'])) {
 125             $this->updatePreviewPicture($entry, $content['open_graph']['og_image']);
 126         }
 127
 128         // if content is an image, define it as a preview too
 129         if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
 130             $this->updatePreviewPicture($entry, $content['url']);
 131         }
 132
 133         if (!empty($content['content_type'])) {
 134             $entry->setMimetype($content['content_type']);
 135         }
 136
 137         try {
 138             $this->tagger->tag($entry);
 139         } catch (\Exception $e) {
 140             $this->logger->error('Error while trying to automatically tag an entry.', [
 141                 'entry_url' => $content['url'],
 142                 'error_msg' => $e->getMessage(),
 143             ]);
 144         }
 145     }
 146
 147     /**
 148      * Validate that the given content has at least a title, an html and a url.
 149      *
 150      * @param array $content
 151      *
 152      * @return bool true if valid otherwise false
 153      */
 154     private function validateContent(array $content)
 155     {
 156         return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);
 157     }
 158
 159     /**
 160      * Use a Symfony validator to ensure the language is well formatted.
 161      *
 162      * @param Entry  $entry
 163      * @param string $value Language to validate and save
 164      */
 165     public function updateLanguage(Entry $entry, $value)
 166     {
 167         // some lang are defined as fr-FR, es-ES.
 168         // replacing - by _ might increase language support
 169         $value = str_replace('-', '_', $value);
 170
 171         $errors = $this->validator->validate(
 172             $value,
 173             (new LocaleConstraint())
 174         );
 175
 176         if (0 === count($errors)) {
 177             $entry->setLanguage($value);
 178
 179             return;
 180         }
 181
 182         $this->logger->warning('Language validation failed. ' . (string) $errors);
 183     }
 184
 185     /**
 186      * Use a Symfony validator to ensure the preview picture is a real url.
 187      *
 188      * @param Entry  $entry
 189      * @param string $value URL to validate and save
 190      */
 191     public function updatePreviewPicture(Entry $entry, $value)
 192     {
 193         $errors = $this->validator->validate(
 194             $value,
 195             (new UrlConstraint())
 196         );
 197
 198         if (0 === count($errors)) {
 199             $entry->setPreviewPicture($value);
 200
 201             return;
 202         }
 203
 204         $this->logger->warning('PreviewPicture validation failed. ' . (string) $errors);
 205     }
 206
 207     /**
 208      * Update date.
 209      *
 210      * @param Entry  $entry
 211      * @param string $value Date to validate and save
 212      */
 213     public function updatePublishedAt(Entry $entry, $value)
 214     {
 215         $date = $value;
 216
 217         // is it a timestamp?
 218         if (filter_var($date, FILTER_VALIDATE_INT) !== false) {
 219             $date = '@'.$value;
 220         }
 221
 222         try {
 223             $entry->setPublishedAt(new \DateTime($date));
 224         } catch (\Exception $e) {
 225             $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $entry->getUrl(), 'date' => $value]);
 226         }
 227     }
 228 }