X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=src%2FWallabag%2FCoreBundle%2FHelper%2FContentProxy.php;h=bfaa197695941af57dde87e751f95d1a1d0cc30c;hb=9fe87bc2e20fa95573287a61ef9798cc15648187;hp=d45aef8894d6d0fe154efc7fb6344e2840059a30;hpb=54c2d164a362e64a320438b439bf9dd6d2c02424;p=github%2Fwallabag%2Fwallabag.git diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php index d45aef88..bfaa1976 100644 --- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php +++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php @@ -5,9 +5,7 @@ namespace Wallabag\CoreBundle\Helper; use Graby\Graby; 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; /** @@ -19,37 +17,34 @@ 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, LoggerInterface $logger, $fetchingErrorMessage) + 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. - * In case we couldn't find content, we'll try to use Open Graph data. + * Update entry using either fetched or provided content. * - * We can also force the content, in case of an import from the v1 for example, so the function won't - * fetch the content from the website but rather use information given with the $content parameter. - * - * @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 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 = []) + public function updateEntry(Entry $entry, $url, array $content = [], $disableContentUpdate = false) { - // do we have to fetch the content or the provided one is ok? - if (empty($content) || false === $this->validateContent($content)) { + 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 @@ -59,8 +54,24 @@ class ContentProxy } } + // 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); + } + + /** + * 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) + { $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']; } @@ -68,25 +79,40 @@ class ContentProxy if (false === $html) { $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']; } } - $entry->setUrl($content['url'] ?: $url); + $entry->setUrl($content['url']); $entry->setTitle($title); $entry->setContent($html); $entry->setHttpStatus(isset($content['status']) ? $content['status'] : ''); - if (isset($content['date']) && null !== $content['date'] && '' !== $content['date']) { - $entry->setPublishedAt(new \DateTime($content['date'])); + 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' => $content['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)); @@ -96,12 +122,12 @@ class ContentProxy $entry->setDomainName($domainName); } - if (isset($content['open_graph']['og_image']) && $content['open_graph']['og_image']) { + if (!empty($content['open_graph']['og_image'])) { $entry->setPreviewPicture($content['open_graph']['og_image']); } // if content is an image define as a preview too - if (isset($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) { + if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) { $entry->setPreviewPicture($content['url']); } @@ -109,65 +135,14 @@ class ContentProxy $this->tagger->tag($entry); } catch (\Exception $e) { $this->logger->error('Error while trying to automatically tag an entry.', [ - 'entry_url' => $url, + 'entry_url' => $content['url'], '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); - } - } } /** - * Validate that the given content as enough value to be used - * instead of fetch the content from the url. + * Validate that the given content has at least a title, an html and a url. * * @param array $content * @@ -175,6 +150,6 @@ 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']); } }