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     protected $storeArticleHeaders;
  28
  29     public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage, $storeArticleHeaders = false)
  30     {
  31         $this->graby = $graby;
  32         $this->tagger = $tagger;
  33         $this->validator = $validator;
  34         $this->logger = $logger;
  35         $this->mimeGuesser = new MimeTypeExtensionGuesser();
  36         $this->fetchingErrorMessage = $fetchingErrorMessage;
  37         $this->storeArticleHeaders = $storeArticleHeaders;
  38     }
  39
  40     /**
  41      * Update entry using either fetched or provided content.
  42      *
  43      * @param Entry  $entry                Entry to update
  44      * @param string $url                  Url of the content
  45      * @param array  $content              Array with content provided for import with AT LEAST keys title, html, url to skip the fetchContent from the url
  46      * @param bool   $disableContentUpdate Whether to skip trying to fetch content using Graby
  47      */
  48     public function updateEntry(Entry $entry, $url, array $content = [], $disableContentUpdate = false)
  49     {
  50         if (!empty($content['html'])) {
  51             $content['html'] = $this->graby->cleanupHtml($content['html'], $url);
  52         }
  53
  54         if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) {
  55             $fetchedContent = $this->graby->fetchContent($url);
  56
  57             // when content is imported, we have information in $content
  58             // in case fetching content goes bad, we'll keep the imported information instead of overriding them
  59             if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage) {
  60                 $content = $fetchedContent;
  61             }
  62         }
  63
  64         // be sure to keep the url in case of error
  65         // so we'll be able to refetch it in the future
  66         $content['url'] = !empty($content['url']) ? $content['url'] : $url;
  67
  68         $this->stockEntry($entry, $content);
  69     }
  70
  71     /**
  72      * Use a Symfony validator to ensure the language is well formatted.
  73      *
  74      * @param Entry  $entry
  75      * @param string $value Language to validate and save
  76      */
  77     public function updateLanguage(Entry $entry, $value)
  78     {
  79         // some lang are defined as fr-FR, es-ES.
  80         // replacing - by _ might increase language support
  81         $value = str_replace('-', '_', $value);
  82
  83         $errors = $this->validator->validate(
  84             $value,
  85             (new LocaleConstraint())
  86         );
  87
  88         if (0 === \count($errors)) {
  89             $entry->setLanguage($value);
  90
  91             return;
  92         }
  93
  94         $this->logger->warning('Language validation failed. ' . (string) $errors);
  95     }
  96
  97     /**
  98      * Use a Symfony validator to ensure the preview picture is a real url.
  99      *
 100      * @param Entry  $entry
 101      * @param string $value URL to validate and save
 102      */
 103     public function updatePreviewPicture(Entry $entry, $value)
 104     {
 105         $errors = $this->validator->validate(
 106             $value,
 107             (new UrlConstraint())
 108         );
 109
 110         if (0 === \count($errors)) {
 111             $entry->setPreviewPicture($value);
 112
 113             return;
 114         }
 115
 116         $this->logger->warning('PreviewPicture validation failed. ' . (string) $errors);
 117     }
 118
 119     /**
 120      * Update date.
 121      *
 122      * @param Entry  $entry
 123      * @param string $value Date to validate and save
 124      */
 125     public function updatePublishedAt(Entry $entry, $value)
 126     {
 127         $date = $value;
 128
 129         // is it a timestamp?
 130         if (false !== filter_var($date, FILTER_VALIDATE_INT)) {
 131             $date = '@' . $date;
 132         }
 133
 134         try {
 135             // is it already a DateTime?
 136             // (it's inside the try/catch in case of fail to be parse time string)
 137             if (!$date instanceof \DateTime) {
 138                 $date = new \DateTime($date);
 139             }
 140
 141             $entry->setPublishedAt($date);
 142         } catch (\Exception $e) {
 143             $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $entry->getUrl(), 'date' => $value]);
 144         }
 145     }
 146
 147     /**
 148      * Helper to extract and save host from entry url.
 149      *
 150      * @param Entry $entry
 151      */
 152     public function setEntryDomainName(Entry $entry)
 153     {
 154         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
 155         if (false !== $domainName) {
 156             $entry->setDomainName($domainName);
 157         }
 158     }
 159
 160     /**
 161      * Helper to set a default title using:
 162      * - url basename, if applicable
 163      * - hostname.
 164      *
 165      * @param Entry $entry
 166      */
 167     public function setDefaultEntryTitle(Entry $entry)
 168     {
 169         $url = parse_url($entry->getUrl());
 170         $path = pathinfo($url['path'], PATHINFO_BASENAME);
 171
 172         if (empty($path)) {
 173             $path = $url['host'];
 174         }
 175
 176         $entry->setTitle($path);
 177     }
 178
 179     /**
 180      * Stock entry with fetched or imported content.
 181      * Will fall back to OpenGraph data if available.
 182      *
 183      * @param Entry $entry   Entry to stock
 184      * @param array $content Array with at least title, url & html
 185      */
 186     private function stockEntry(Entry $entry, array $content)
 187     {
 188         $entry->setUrl($content['url']);
 189
 190         $this->setEntryDomainName($entry);
 191
 192         if (!empty($content['title'])) {
 193             $entry->setTitle($content['title']);
 194         } elseif (!empty($content['open_graph']['og_title'])) {
 195             $entry->setTitle($content['open_graph']['og_title']);
 196         }
 197
 198         $html = $content['html'];
 199         if (false === $html) {
 200             $html = $this->fetchingErrorMessage;
 201
 202             if (!empty($content['open_graph']['og_description'])) {
 203                 $html .= '<p><i>But we found a short description: </i></p>';
 204                 $html .= $content['open_graph']['og_description'];
 205             }
 206         }
 207
 208         $entry->setContent($html);
 209         $entry->setReadingTime(Utils::getReadingTime($html));
 210
 211         if (!empty($content['status'])) {
 212             $entry->setHttpStatus($content['status']);
 213         }
 214
 215         if (!empty($content['authors']) && \is_array($content['authors'])) {
 216             $entry->setPublishedBy($content['authors']);
 217         }
 218
 219         if (!empty($content['all_headers']) && $this->storeArticleHeaders) {
 220             $entry->setHeaders($content['all_headers']);
 221         }
 222
 223         if (!empty($content['date'])) {
 224             $this->updatePublishedAt($entry, $content['date']);
 225         }
 226
 227         if (!empty($content['language'])) {
 228             $this->updateLanguage($entry, $content['language']);
 229         }
 230
 231         if (!empty($content['open_graph']['og_image'])) {
 232             $this->updatePreviewPicture($entry, $content['open_graph']['og_image']);
 233         }
 234
 235         // if content is an image, define it as a preview too
 236         if (!empty($content['content_type']) && \in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
 237             $this->updatePreviewPicture($entry, $content['url']);
 238         }
 239
 240         if (!empty($content['content_type'])) {
 241             $entry->setMimetype($content['content_type']);
 242         }
 243
 244         try {
 245             $this->tagger->tag($entry);
 246         } catch (\Exception $e) {
 247             $this->logger->error('Error while trying to automatically tag an entry.', [
 248                 'entry_url' => $content['url'],
 249                 'error_msg' => $e->getMessage(),
 250             ]);
 251         }
 252     }
 253
 254     /**
 255      * Validate that the given content has at least a title, an html and a url.
 256      *
 257      * @param array $content
 258      *
 259      * @return bool true if valid otherwise false
 260      */
 261     private function validateContent(array $content)
 262     {
 263         return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);
 264     }
 265 }