X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=src%2FWallabag%2FCoreBundle%2FHelper%2FContentProxy.php;h=c73b8eafb7aa6de27d45b55794b28ac9dae464ab;hb=7aba665e484c5c36ee029219a999a427d864ff22;hp=8019df42a6c51bd642c78074f6eceb6ec0e91657;hpb=114c55c0a6eade2ba6c53fe25f61cc58cca91620;p=github%2Fwallabag%2Fwallabag.git diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php index 8019df42..c73b8eaf 100644 --- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php +++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php @@ -3,11 +3,10 @@ namespace Wallabag\CoreBundle\Helper; use Graby\Graby; -use Psr\Log\LoggerInterface as Logger; +use Psr\Log\LoggerInterface; use Wallabag\CoreBundle\Entity\Entry; -use Wallabag\CoreBundle\Entity\Tag; use Wallabag\CoreBundle\Tools\Utils; -use Wallabag\CoreBundle\Repository\TagRepository; +use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser; /** * This kind of proxy class take care of getting the content from an url @@ -18,18 +17,21 @@ class ContentProxy protected $graby; protected $tagger; protected $logger; - protected $tagRepository; + protected $mimeGuesser; + protected $fetchingErrorMessage; + protected $eventDispatcher; - public function __construct(Graby $graby, RuleBasedTagger $tagger, TagRepository $tagRepository, Logger $logger) + public function __construct(Graby $graby, RuleBasedTagger $tagger, LoggerInterface $logger, $fetchingErrorMessage) { $this->graby = $graby; $this->tagger = $tagger; $this->logger = $logger; - $this->tagRepository = $tagRepository; + $this->mimeGuesser = new MimeTypeExtensionGuesser(); + $this->fetchingErrorMessage = $fetchingErrorMessage; } /** - * Fetch content using graby and hydrate given entry with results information. + * Fetch content using graby and hydrate given $entry with results information. * In case we couldn't find content, we'll try to use Open Graph data. * * We can also force the content, in case of an import from the v1 for example, so the function won't @@ -37,27 +39,36 @@ class ContentProxy * * @param Entry $entry Entry to update * @param string $url Url to grab content for - * @param array $content An array with AT LEAST keys title, html, url, language & content_type to skip the fetchContent from the url - * - * @return Entry + * @param array $content An array with AT LEAST keys title, html, url to skip the fetchContent from the url */ public function updateEntry(Entry $entry, $url, array $content = []) { + // ensure content is a bit cleaned up + if (!empty($content['html'])) { + $content['html'] = $this->graby->cleanupHtml($content['html'], $url); + } + // do we have to fetch the content or the provided one is ok? if (empty($content) || false === $this->validateContent($content)) { - $content = $this->graby->fetchContent($url); + $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; + } } $title = $content['title']; - if (!$title && isset($content['open_graph']['og_title'])) { + if (!$title && !empty($content['open_graph']['og_title'])) { $title = $content['open_graph']['og_title']; } $html = $content['html']; if (false === $html) { - $html = '

Unable to retrieve readable content.

'; + $html = $this->fetchingErrorMessage; - if (isset($content['open_graph']['og_description'])) { + if (!empty($content['open_graph']['og_description'])) { $html .= '

But we found a short description:

'; $html .= $content['open_graph']['og_description']; } @@ -66,8 +77,33 @@ class ContentProxy $entry->setUrl($content['url'] ?: $url); $entry->setTitle($title); $entry->setContent($html); - $entry->setLanguage($content['language']); - $entry->setMimetype($content['content_type']); + $entry->setHttpStatus(isset($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']; + } + + 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['authors'])) { + $entry->setPublishedBy($content['authors']); + } + + if (!empty($content['all_headers'])) { + $entry->setHeaders($content['all_headers']); + } + + $entry->setLanguage(isset($content['language']) ? $content['language'] : ''); + $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : ''); $entry->setReadingTime(Utils::getReadingTime($html)); $domainName = parse_url($entry->getUrl(), PHP_URL_HOST); @@ -75,10 +111,15 @@ class ContentProxy $entry->setDomainName($domainName); } - if (isset($content['open_graph']['og_image'])) { + if (!empty($content['open_graph']['og_image'])) { $entry->setPreviewPicture($content['open_graph']['og_image']); } + // if content is an image define 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']); + } + try { $this->tagger->tag($entry); } catch (\Exception $e) { @@ -87,56 +128,6 @@ class ContentProxy 'error_msg' => $e->getMessage(), ]); } - - return $entry; - } - - /** - * Assign some tags to an entry. - * - * @param Entry $entry - * @param array|string $tags An array of tag or a string coma separated of tag - * @param array $entitiesReady Entities from the EntityManager which are persisted but not yet flushed - * It is mostly to fix duplicate tag on import @see http://stackoverflow.com/a/7879164/569101 - */ - public function assignTagsToEntry(Entry $entry, $tags, array $entitiesReady = []) - { - if (!is_array($tags)) { - $tags = explode(',', $tags); - } - - // keeps only Tag entity from the "not yet flushed entities" - $tagsNotYetFlushed = []; - foreach ($entitiesReady as $entity) { - if ($entity instanceof Tag) { - $tagsNotYetFlushed[$entity->getLabel()] = $entity; - } - } - - foreach ($tags as $label) { - $label = trim($label); - - // avoid empty tag - if (0 === strlen($label)) { - continue; - } - - if (isset($tagsNotYetFlushed[$label])) { - $tagEntity = $tagsNotYetFlushed[$label]; - } else { - $tagEntity = $this->tagRepository->findOneByLabel($label); - - if (is_null($tagEntity)) { - $tagEntity = new Tag(); - $tagEntity->setLabel($label); - } - } - - // only add the tag on the entry if the relation doesn't exist - if (false === $entry->getTags()->contains($tagEntity)) { - $entry->addTag($tagEntity); - } - } } /** @@ -149,6 +140,6 @@ class ContentProxy */ private function validateContent(array $content) { - return isset($content['title']) && isset($content['html']) && isset($content['url']) && isset($content['language']) && isset($content['content_type']); + return !empty($content['title']) && !empty($content['html']) && !empty($content['url']); } }