X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=src%2FWallabag%2FCoreBundle%2FHelper%2FContentProxy.php;h=1a2a330ff085f5e922da295fbd5c908611936545;hb=fc040c749dec0275e562182562c1c1cb89e6cfa1;hp=f222dd88c26fced386f0328d03105e9194e9812d;hpb=20da238413d1c6cc360d58a13df33eb199fa5f05;p=github%2Fwallabag%2Fwallabag.git diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php index f222dd88..1a2a330f 100644 --- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php +++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php @@ -4,11 +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\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 @@ -18,39 +19,41 @@ class ContentProxy { protected $graby; protected $tagger; + protected $validator; protected $logger; - protected $tagRepository; protected $mimeGuesser; protected $fetchingErrorMessage; + protected $eventDispatcher; + protected $storeArticleHeaders; - public function __construct(Graby $graby, RuleBasedTagger $tagger, TagRepository $tagRepository, 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->tagRepository = $tagRepository; $this->mimeGuesser = new MimeTypeExtensionGuesser(); $this->fetchingErrorMessage = $fetchingErrorMessage; + $this->storeArticleHeaders = $storeArticleHeaders; } /** - * 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 - * 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 + * Update entry using either fetched or provided content. * - * @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); + $fetchedContent['title'] = $this->sanitizeContentTitle($fetchedContent['title'], $fetchedContent['content_type']); // 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 @@ -59,107 +62,303 @@ class ContentProxy } } - $title = $content['title']; - if (!$title && isset($content['open_graph']['og_title'])) { - $title = $content['open_graph']['og_title']; + // 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; + + // 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); + } + + /** + * 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 = $value; + + // is it a timestamp? + if (false !== filter_var($date, FILTER_VALIDATE_INT)) { + $date = '@' . $date; + } + + try { + // is it already a DateTime? + // (it's inside the try/catch in case of fail to be parse time string) + if (!$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]); + } + } + + /** + * 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); + } + } + + /** + * 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']; + } + + $entry->setTitle($path); + } + + /** + * Try to sanitize the title of the fetched content from wrong character encodings and invalid UTF-8 character. + * + * @param $title + * @param $contentType + * + * @return string + */ + private function sanitizeContentTitle($title, $contentType) + { + if ('application/pdf' === $contentType) { + $title = $this->convertPdfEncodingToUTF8($title); + } + + return $this->sanitizeUTF8Text($title); + } + + /** + * If the title from the fetched content comes from a PDF, then its very possible that the character encoding is not + * UTF-8. This methods tries to identify the character encoding and translate the title to UTF-8. + * + * @param $title + * + * @return string (maybe contains invalid UTF-8 character) + */ + private function convertPdfEncodingToUTF8($title) + { + // first try UTF-8 because its easier to detect its present/absence + foreach (['UTF-8', 'UTF-16BE', 'WINDOWS-1252'] as $encoding) { + if (mb_check_encoding($title, $encoding)) { + return mb_convert_encoding($title, 'UTF-8', $encoding); + } + } + + return $title; + } + + /** + * Remove invalid UTF-8 characters from the given string. + * + * @param string $rawText + * + * @return string + */ + private function sanitizeUTF8Text($rawText) + { + if (mb_check_encoding($rawText, 'UTF-8')) { + return $rawText; + } + + return iconv('UTF-8', 'UTF-8//IGNORE', $rawText); + } + + /** + * 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) + { + $this->updateOriginUrl($entry, $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']; 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); - $entry->setTitle($title); $entry->setContent($html); - $entry->setHttpStatus(isset($content['status']) ? $content['status'] : ''); - - $entry->setLanguage(isset($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); - if (false !== $domainName) { - $entry->setDomainName($domainName); + 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']) && $this->storeArticleHeaders) { + $entry->setHeaders($content['all_headers']); } - if (isset($content['open_graph']['og_image']) && $content['open_graph']['og_image']) { - $entry->setPreviewPicture($content['open_graph']['og_image']); + if (!empty($content['date'])) { + $this->updatePublishedAt($entry, $content['date']); } - // if content is an image define as a preview too - if (isset($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) { - $entry->setPreviewPicture($content['url']); + 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' => $url, + 'entry_url' => $content['url'], 'error_msg' => $e->getMessage(), ]); } - - return $entry; } /** - * Assign some tags to an entry. + * Update the origin_url field when a redirection occurs + * This field is set if it is empty and new url does not match ignore list. * - * @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 + * @param Entry $entry + * @param string $url */ - public function assignTagsToEntry(Entry $entry, $tags, array $entitiesReady = []) + private function updateOriginUrl(Entry $entry, $url) { - if (!is_array($tags)) { - $tags = explode(',', $tags); - } + if (!empty($url) && $entry->getUrl() !== $url) { + $parsed_entry_url = parse_url($entry->getUrl()); + $parsed_content_url = parse_url($url); - // keeps only Tag entity from the "not yet flushed entities" - $tagsNotYetFlushed = []; - foreach ($entitiesReady as $entity) { - if ($entity instanceof Tag) { - $tagsNotYetFlushed[$entity->getLabel()] = $entity; - } - } + $diff_ec = array_diff_assoc($parsed_entry_url, $parsed_content_url); + $diff_ce = array_diff_assoc($parsed_content_url, $parsed_entry_url); - foreach ($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); - } - } + $diff = array_merge($diff_ec, $diff_ce); + $diff_keys = array_keys($diff); + sort($diff_keys); - // only add the tag on the entry if the relation doesn't exist - if (false === $entry->getTags()->contains($tagEntity)) { - $entry->addTag($tagEntity); + switch ($diff_keys) { + case ['path']: + if (($parsed_entry_url['path'] . '/' === $parsed_content_url['path']) // diff is trailing slash, we only replace the url of the entry + || ($url === urldecode($entry->getUrl()))) { // we update entry url if new url is a decoded version of it, see EntryRepository#findByUrlAndUserId + $entry->setUrl($url); + } + break; + case ['scheme']: + $entry->setUrl($url); + break; + case ['fragment']: + case ['query']: + case ['fragment', 'query']: + // noop + break; + default: + if (empty($entry->getOriginUrl())) { + $entry->setOriginUrl($entry->getUrl()); + } + $entry->setUrl($url); + break; } } } /** - * 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 * @@ -167,6 +366,6 @@ 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']); } }