X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;ds=inline;f=src%2FWallabag%2FCoreBundle%2FHelper%2FContentProxy.php;h=f0d8c1b420c9ba5fa192eb43b4949d84d1f79b5a;hb=4a81360efcdfe4bab8d75f7227c9cf5bfd514189;hp=29259bbdb6ca57e0f66927a1993b029e4036ceec;hpb=8648f0c00534e8af83b2a5451269d79906db6c16;p=github%2Fwallabag%2Fwallabag.git diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php index 29259bbd..f0d8c1b4 100644 --- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php +++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php @@ -53,7 +53,7 @@ class ContentProxy if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) { $fetchedContent = $this->graby->fetchContent($url); - $fetchedContent['title'] = $this->sanitizeUTF8Text($fetchedContent['title']); + $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 @@ -66,29 +66,15 @@ class ContentProxy // so we'll be able to refetch it in the future $content['url'] = !empty($content['url']) ? $content['url'] : $url; - $this->stockEntry($entry, $content); - } - - /** - * Remove invalid UTF-8 characters from the given string in following steps: - * - try to interpret the given string as ISO-8859-1, convert it to UTF-8 and return it (if its valid) - * - simply remove every invalid UTF-8 character and return the result (https://stackoverflow.com/a/1433665) - * @param String $rawText - * @return string - */ - private function sanitizeUTF8Text($rawText) { - if (mb_check_encoding($rawText, 'utf-8')) { - return $rawText; // return because its valid utf-8 text - } - - // we assume that $text is encoded in ISO-8859-1 (and not the similar Windows-1252 or other encoding) - $convertedText = utf8_encode($rawText); - if (mb_check_encoding($convertedText, 'utf-8')) { - return $convertedText; + // 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); } - // last resort: simply remove invalid UTF-8 character because $rawText can have some every exotic encoding - return iconv("UTF-8", "UTF-8//IGNORE", $rawText); + $this->stockEntry($entry, $content); } /** @@ -199,6 +185,59 @@ class ContentProxy $entry->setTitle($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.