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         // In one case (at least in tests), url is empty here
  69         // so we set it using $url provided in the updateEntry call.
  70         // Not sure what are the other possible cases where this property is empty
  71         if (empty($entry->getUrl()) && !empty($url))
  72         {
  73             $entry->setUrl($url);
  74         }
  75
  76         $this->stockEntry($entry, $content);
  77     }
  78
  79     /**
  80      * Use a Symfony validator to ensure the language is well formatted.
  81      *
  82      * @param Entry  $entry
  83      * @param string $value Language to validate and save
  84      */
  85     public function updateLanguage(Entry $entry, $value)
  86     {
  87         // some lang are defined as fr-FR, es-ES.
  88         // replacing - by _ might increase language support
  89         $value = str_replace('-', '_', $value);
  90
  91         $errors = $this->validator->validate(
  92             $value,
  93             (new LocaleConstraint())
  94         );
  95
  96         if (0 === count($errors)) {
  97             $entry->setLanguage($value);
  98
  99             return;
 100         }
 101
 102         $this->logger->warning('Language validation failed. ' . (string) $errors);
 103     }
 104
 105     /**
 106      * Use a Symfony validator to ensure the preview picture is a real url.
 107      *
 108      * @param Entry  $entry
 109      * @param string $value URL to validate and save
 110      */
 111     public function updatePreviewPicture(Entry $entry, $value)
 112     {
 113         $errors = $this->validator->validate(
 114             $value,
 115             (new UrlConstraint())
 116         );
 117
 118         if (0 === count($errors)) {
 119             $entry->setPreviewPicture($value);
 120
 121             return;
 122         }
 123
 124         $this->logger->warning('PreviewPicture validation failed. ' . (string) $errors);
 125     }
 126
 127     /**
 128      * Update date.
 129      *
 130      * @param Entry  $entry
 131      * @param string $value Date to validate and save
 132      */
 133     public function updatePublishedAt(Entry $entry, $value)
 134     {
 135         $date = $value;
 136
 137         // is it a timestamp?
 138         if (false !== filter_var($date, FILTER_VALIDATE_INT)) {
 139             $date = '@' . $date;
 140         }
 141
 142         try {
 143             // is it already a DateTime?
 144             // (it's inside the try/catch in case of fail to be parse time string)
 145             if (!$date instanceof \DateTime) {
 146                 $date = new \DateTime($date);
 147             }
 148
 149             $entry->setPublishedAt($date);
 150         } catch (\Exception $e) {
 151             $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $entry->getUrl(), 'date' => $value]);
 152         }
 153     }
 154
 155     /**
 156      * Helper to extract and save host from entry url.
 157      *
 158      * @param Entry $entry
 159      */
 160     public function setEntryDomainName(Entry $entry)
 161     {
 162         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
 163         if (false !== $domainName) {
 164             $entry->setDomainName($domainName);
 165         }
 166     }
 167
 168     /**
 169      * Helper to set a default title using:
 170      * - url basename, if applicable
 171      * - hostname.
 172      *
 173      * @param Entry $entry
 174      */
 175     public function setDefaultEntryTitle(Entry $entry)
 176     {
 177         $url = parse_url($entry->getUrl());
 178         $path = pathinfo($url['path'], PATHINFO_BASENAME);
 179
 180         if (empty($path)) {
 181             $path = $url['host'];
 182         }
 183
 184         $entry->setTitle($path);
 185     }
 186
 187     /**
 188      * Stock entry with fetched or imported content.
 189      * Will fall back to OpenGraph data if available.
 190      *
 191      * @param Entry $entry   Entry to stock
 192      * @param array $content Array with at least title, url & html
 193      */
 194     private function stockEntry(Entry $entry, array $content)
 195     {
 196         $entry->setUrl($content['url']);
 197
 198         $this->setEntryDomainName($entry);
 199
 200         if (!empty($content['title'])) {
 201             $entry->setTitle($content['title']);
 202         } elseif (!empty($content['open_graph']['og_title'])) {
 203             $entry->setTitle($content['open_graph']['og_title']);
 204         }
 205
 206         $html = $content['html'];
 207         if (false === $html) {
 208             $html = $this->fetchingErrorMessage;
 209
 210             if (!empty($content['open_graph']['og_description'])) {
 211                 $html .= '<p><i>But we found a short description: </i></p>';
 212                 $html .= $content['open_graph']['og_description'];
 213             }
 214         }
 215
 216         $entry->setContent($html);
 217         $entry->setReadingTime(Utils::getReadingTime($html));
 218
 219         if (!empty($content['status'])) {
 220             $entry->setHttpStatus($content['status']);
 221         }
 222
 223         if (!empty($content['authors']) && is_array($content['authors'])) {
 224             $entry->setPublishedBy($content['authors']);
 225         }
 226
 227         if (!empty($content['all_headers']) && $this->storeArticleHeaders) {
 228             $entry->setHeaders($content['all_headers']);
 229         }
 230
 231         if (!empty($content['date'])) {
 232             $this->updatePublishedAt($entry, $content['date']);
 233         }
 234
 235         if (!empty($content['language'])) {
 236             $this->updateLanguage($entry, $content['language']);
 237         }
 238
 239         if (!empty($content['open_graph']['og_image'])) {
 240             $this->updatePreviewPicture($entry, $content['open_graph']['og_image']);
 241         }
 242
 243         // if content is an image, define it as a preview too
 244         if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
 245             $this->updatePreviewPicture($entry, $content['url']);
 246         }
 247
 248         if (!empty($content['content_type'])) {
 249             $entry->setMimetype($content['content_type']);
 250         }
 251
 252         try {
 253             $this->tagger->tag($entry);
 254         } catch (\Exception $e) {
 255             $this->logger->error('Error while trying to automatically tag an entry.', [
 256                 'entry_url' => $content['url'],
 257                 'error_msg' => $e->getMessage(),
 258             ]);
 259         }
 260     }
 261
 262     /**
 263      * Validate that the given content has at least a title, an html and a url.
 264      *
 265      * @param array $content
 266      *
 267      * @return bool true if valid otherwise false
 268      */
 269     private function validateContent(array $content)
 270     {
 271         return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);
 272     }
 273 }