graby = $graby; $this->tagger = $tagger; $this->validator = $validator; $this->logger = $logger; $this->mimeGuesser = new MimeTypeExtensionGuesser(); $this->fetchingErrorMessage = $fetchingErrorMessage; } /** * Update entry using either fetched or provided content. * * @param Entry $entry Entry to update * @param string $url Url of the content * @param array $content Array with content provided for import with AT LEAST keys title, html, url to skip the fetchContent from the url * @param bool $disableContentUpdate Whether to skip trying to fetch content using Graby */ public function updateEntry(Entry $entry, $url, array $content = [], $disableContentUpdate = false) { if (!empty($content['html'])) { $content['html'] = $this->graby->cleanupHtml($content['html'], $url); } if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) { $fetchedContent = $this->graby->fetchContent($url); // when content is imported, we have information in $content // in case fetching content goes bad, we'll keep the imported information instead of overriding them if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage) { $content = $fetchedContent; } } // be sure to keep the url in case of error // so we'll be able to refetch it in the future $content['url'] = !empty($content['url']) ? $content['url'] : $url; $this->stockEntry($entry, $content); } /** * Use a Symfony validator to ensure the language is well formatted. * * @param Entry $entry * @param string $value Language to validate and save */ public function updateLanguage(Entry $entry, $value) { // some lang are defined as fr-FR, es-ES. // replacing - by _ might increase language support $value = str_replace('-', '_', $value); $errors = $this->validator->validate( $value, (new LocaleConstraint()) ); if (0 === count($errors)) { $entry->setLanguage($value); return; } $this->logger->warning('Language validation failed. ' . (string) $errors); } /** * Use a Symfony validator to ensure the preview picture is a real url. * * @param Entry $entry * @param string $value URL to validate and save */ public function updatePreviewPicture(Entry $entry, $value) { $errors = $this->validator->validate( $value, (new UrlConstraint()) ); if (0 === count($errors)) { $entry->setPreviewPicture($value); return; } $this->logger->warning('PreviewPicture validation failed. ' . (string) $errors); } /** * Update date. * * @param Entry $entry * @param string $value Date to validate and save */ public function updatePublishedAt(Entry $entry, $value) { $date = $value instanceof \DateTime ? $value->date : $value; // is it a timestamp? if (filter_var($date, FILTER_VALIDATE_INT) !== false) { $date = '@' . $value; } try { $entry->setPublishedAt(new \DateTime($date)); } catch (\Exception $e) { $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $entry->getUrl(), 'date' => $value]); } } /** * Stock entry with fetched or imported content. * Will fall back to OpenGraph data if available. * * @param Entry $entry Entry to stock * @param array $content Array with at least title, url & html */ private function stockEntry(Entry $entry, array $content) { $entry->setUrl($content['url']); $domainName = parse_url($entry->getUrl(), PHP_URL_HOST); if (false !== $domainName) { $entry->setDomainName($domainName); } if (!empty($content['title'])) { $entry->setTitle($content['title']); } elseif (!empty($content['open_graph']['og_title'])) { $entry->setTitle($content['open_graph']['og_title']); } $html = $content['html']; if (false === $html) { $html = $this->fetchingErrorMessage; if (!empty($content['open_graph']['og_description'])) { $html .= '

But we found a short description:

'; $html .= $content['open_graph']['og_description']; } } $entry->setContent($html); $entry->setReadingTime(Utils::getReadingTime($html)); if (!empty($content['status'])) { $entry->setHttpStatus($content['status']); } if (!empty($content['authors']) && is_array($content['authors'])) { $entry->setPublishedBy($content['authors']); } if (!empty($content['all_headers'])) { $entry->setHeaders($content['all_headers']); } if (!empty($content['date'])) { $this->updatePublishedAt($entry, $content['date']); } if (!empty($content['language'])) { $this->updateLanguage($entry, $content['language']); } if (!empty($content['open_graph']['og_image'])) { $this->updatePreviewPicture($entry, $content['open_graph']['og_image']); } // if content is an image, define it as a preview too if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) { $this->updatePreviewPicture($entry, $content['url']); } if (!empty($content['content_type'])) { $entry->setMimetype($content['content_type']); } try { $this->tagger->tag($entry); } catch (\Exception $e) { $this->logger->error('Error while trying to automatically tag an entry.', [ 'entry_url' => $content['url'], 'error_msg' => $e->getMessage(), ]); } } /** * Validate that the given content has at least a title, an html and a url. * * @param array $content * * @return bool true if valid otherwise false */ private function validateContent(array $content) { return !empty($content['title']) && !empty($content['html']) && !empty($content['url']); } }