X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=src%2FWallabag%2FCoreBundle%2FHelper%2FContentProxy.php;h=c7ccfe4a0e2f871c9daa1528b6394663e7ddfc55;hb=refs%2Fheads%2Furl-3529;hp=e62927443417b136223dfb9980bca8cff5465f29;hpb=843182c7cf428b5f6b8a1ff7057adc703c1e816e;p=github%2Fwallabag%2Fwallabag.git diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php index e6292744..c7ccfe4a 100644 --- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php +++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php @@ -4,10 +4,12 @@ namespace Wallabag\CoreBundle\Helper; use Graby\Graby; use Psr\Log\LoggerInterface; +use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser; +use Symfony\Component\Validator\Constraints\Locale as LocaleConstraint; +use Symfony\Component\Validator\Constraints\Url as UrlConstraint; +use Symfony\Component\Validator\Validator\ValidatorInterface; use Wallabag\CoreBundle\Entity\Entry; use Wallabag\CoreBundle\Tools\Utils; -use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser; -use Symfony\Component\Config\Definition\Exception\Exception; /** * This kind of proxy class take care of getting the content from an url @@ -17,73 +19,168 @@ class ContentProxy { protected $graby; protected $tagger; + protected $validator; protected $logger; protected $mimeGuesser; protected $fetchingErrorMessage; protected $eventDispatcher; + protected $storeArticleHeaders; - public function __construct(Graby $graby, RuleBasedTagger $tagger, LoggerInterface $logger, $fetchingErrorMessage) + public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage, $storeArticleHeaders = false) { $this->graby = $graby; $this->tagger = $tagger; + $this->validator = $validator; $this->logger = $logger; $this->mimeGuesser = new MimeTypeExtensionGuesser(); $this->fetchingErrorMessage = $fetchingErrorMessage; + $this->storeArticleHeaders = $storeArticleHeaders; } /** - * Update existing entry by fetching from URL using Graby. + * Update entry using either fetched or provided content. * - * @param Entry $entry Entry to update - * @param string $url Url to grab content for + * @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) + public function updateEntry(Entry $entry, $url, array $content = [], $disableContentUpdate = false) { - $content = $this->graby->fetchContent($url); + 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'] = $content['url'] ?: $url; + $content['url'] = !empty($content['url']) ? $content['url'] : $url; + + // In one case (at least in tests), url is empty here + // so we set it using $url provided in the updateEntry call. + // Not sure what are the other possible cases where this property is empty + if (empty($entry->getUrl()) && !empty($url)) { + $entry->setUrl($url); + } $this->stockEntry($entry, $content); } /** - * Import entry using either fetched or provided 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 Entry to update - * @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 + * @param Entry $entry + * @param string $value URL to validate and save */ - public function importEntry(Entry $entry, array $content, $disableContentUpdate = false) + 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 = $value; + + // is it a timestamp? + if (false !== filter_var($date, FILTER_VALIDATE_INT)) { + $date = '@' . $date; + } + try { - $this->validateContent($content); - } catch (\Exception $e) { - // validation failed but do we want to disable updating content? - if (true === $disableContentUpdate) { - throw $e; + // is it already a DateTime? + // (it's inside the try/catch in case of fail to be parse time string) + if (!$date instanceof \DateTime) { + $date = new \DateTime($date); } + + $entry->setPublishedAt($date); + } catch (\Exception $e) { + $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $entry->getUrl(), 'date' => $value]); } + } - if (false === $disableContentUpdate) { - try { - $fetchedContent = $this->graby->fetchContent($content['url']); - } catch (\Exception $e) { - $this->logger->error('Error while trying to fetch content from URL.', [ - 'entry_url' => $content['url'], - 'error_msg' => $e->getMessage(), - ]); - } + /** + * Helper to extract and save host from entry url. + * + * @param Entry $entry + */ + public function setEntryDomainName(Entry $entry) + { + $domainName = parse_url($entry->getUrl(), PHP_URL_HOST); + if (false !== $domainName) { + $entry->setDomainName($domainName); + } + } - // when content is imported, we have information in $content - // in case fetching content goes bad, we'll keep the imported information instead of overriding them - if ($fetchedContent['html'] !== $this->fetchingErrorMessage) { - $content = $fetchedContent; - } + /** + * Helper to set a default title using: + * - url basename, if applicable + * - hostname. + * + * @param Entry $entry + */ + public function setDefaultEntryTitle(Entry $entry) + { + $url = parse_url($entry->getUrl()); + $path = pathinfo($url['path'], PATHINFO_BASENAME); + + if (empty($path)) { + $path = $url['host']; } - $this->stockEntry($entry, $content); + $entry->setTitle($path); } /** @@ -91,13 +188,26 @@ class ContentProxy * Will fall back to OpenGraph data if available. * * @param Entry $entry Entry to stock - * @param array $content Array with at least title and URL + * @param array $content Array with at least title, url & html */ private function stockEntry(Entry $entry, array $content) { - $title = $content['title']; - if (!$title && !empty($content['open_graph']['og_title'])) { - $title = $content['open_graph']['og_title']; + // When a redirection occurs while fetching an entry + // we move the original url in origin_url property if empty + // and set the entry url with the final value + if (!empty($content['url']) && $entry->getUrl() !== $content['url']) { + if (empty($entry->getOriginUrl())) { + $entry->setOriginUrl($entry->getUrl()); + } + $entry->setUrl($content['url']); + } + + $this->setEntryDomainName($entry); + + 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']; @@ -110,50 +220,40 @@ class ContentProxy } } - $entry->setUrl($content['url']); - $entry->setTitle($title); $entry->setContent($html); - $entry->setHttpStatus(isset($content['status']) ? $content['status'] : ''); - - if (!empty($content['date'])) { - $date = $content['date']; - - // is it a timestamp? - if (filter_var($date, FILTER_VALIDATE_INT) !== false) { - $date = '@'.$content['date']; - } + $entry->setReadingTime(Utils::getReadingTime($html)); - try { - $entry->setPublishedAt(new \DateTime($date)); - } catch (\Exception $e) { - $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $url, 'date' => $content['date']]); - } + if (!empty($content['status'])) { + $entry->setHttpStatus($content['status']); } - if (!empty($content['authors'])) { + if (!empty($content['authors']) && is_array($content['authors'])) { $entry->setPublishedBy($content['authors']); } - if (!empty($content['all_headers'])) { + if (!empty($content['all_headers']) && $this->storeArticleHeaders) { $entry->setHeaders($content['all_headers']); } - $entry->setLanguage(isset($content['language']) ? $content['language'] : ''); - $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : ''); - $entry->setReadingTime(Utils::getReadingTime($html)); + if (!empty($content['date'])) { + $this->updatePublishedAt($entry, $content['date']); + } - $domainName = parse_url($entry->getUrl(), PHP_URL_HOST); - if (false !== $domainName) { - $entry->setDomainName($domainName); + if (!empty($content['language'])) { + $this->updateLanguage($entry, $content['language']); } if (!empty($content['open_graph']['og_image'])) { - $entry->setPreviewPicture($content['open_graph']['og_image']); + $this->updatePreviewPicture($entry, $content['open_graph']['og_image']); } - // if content is an image define as a preview too + // if content is an image, define it as a preview too if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) { - $entry->setPreviewPicture($content['url']); + $this->updatePreviewPicture($entry, $content['url']); + } + + if (!empty($content['content_type'])) { + $entry->setMimetype($content['content_type']); } try { @@ -170,19 +270,11 @@ class ContentProxy * 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 $content) { - if (empty($content['title'])) { - throw new Exception('Missing title from imported entry!'); - } - - if (empty($content['url'])) { - throw new Exception('Missing URL from imported entry!'); - } - - if (empty($content['html'])) { - throw new Exception('Missing html from imported entry!'); - } + return !empty($content['title']) && !empty($content['html']) && !empty($content['url']); } }