X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=application%2FLinkUtils.php;h=4df5c0cababf247357d89ebd002e4b49a7910cf6;hb=ed7e1be12d65bdb1b924c7efb6a84fd591192c6c;hp=d8dc8b5e01a2dc01112cdb5b874626012c650d10;hpb=f66a1990e5d93a6f302ce594968e5e717b93da72;p=github%2Fshaarli%2FShaarli.git diff --git a/application/LinkUtils.php b/application/LinkUtils.php index d8dc8b5e..4df5c0ca 100644 --- a/application/LinkUtils.php +++ b/application/LinkUtils.php @@ -1,60 +1,90 @@ (.*?)!is', $html, $matches)) { - return trim(str_replace("\n", ' ', $matches[1])); - } - return false; + $isRedirected = false; + /** + * cURL callback function for CURLOPT_WRITEFUNCTION (called during the download). + * + * While downloading the remote page, we check that the HTTP code is 200 and content type is 'html/text' + * Then we extract the title and the charset and stop the download when it's done. + * + * @param resource $ch cURL resource + * @param string $data chunk of data being downloaded + * + * @return int|bool length of $data or false if we need to stop the download + */ + return function(&$ch, $data) use ($curlGetInfo, &$charset, &$title, &$isRedirected) { + $responseCode = $curlGetInfo($ch, CURLINFO_RESPONSE_CODE); + if (!empty($responseCode) && in_array($responseCode, [301, 302])) { + $isRedirected = true; + return strlen($data); + } + if (!empty($responseCode) && $responseCode !== 200) { + return false; + } + // After a redirection, the content type will keep the previous request value + // until it finds the next content-type header. + if (! $isRedirected || strpos(strtolower($data), 'content-type') !== false) { + $contentType = $curlGetInfo($ch, CURLINFO_CONTENT_TYPE); + } + if (!empty($contentType) && strpos($contentType, 'text/html') === false) { + return false; + } + if (!empty($contentType) && empty($charset)) { + $charset = header_extract_charset($contentType); + } + if (empty($charset)) { + $charset = html_extract_charset($data); + } + if (empty($title)) { + $title = html_extract_title($data); + } + // We got everything we want, stop the download. + if (!empty($responseCode) && !empty($contentType) && !empty($charset) && !empty($title)) { + return false; + } + + return strlen($data); + }; } /** - * Determine charset from downloaded page. - * Priority: - * 1. HTTP headers (Content type). - * 2. HTML content page (tag ). - * 3. Use a default charset (default: UTF-8). + * Extract title from an HTML document. * - * @param array $headers HTTP headers array. - * @param string $htmlContent HTML content where to look for charset. - * @param string $defaultCharset Default charset to apply if other methods failed. + * @param string $html HTML content where to look for a title. * - * @return string Determined charset. + * @return bool|string Extracted title if found, false otherwise. */ -function get_charset($headers, $htmlContent, $defaultCharset = 'utf-8') +function html_extract_title($html) { - if ($charset = headers_extract_charset($headers)) { - return $charset; - } - - if ($charset = html_extract_charset($htmlContent)) { - return $charset; + if (preg_match('!
(This is used in description to keep text formatting). + * + * @param string $text input text. + * + * @return string formatted text. + */ +function space2nbsp($text) +{ + return preg_replace('/(^| ) /m', '$1 ', $text); +} + +/** + * Format Shaarli's description + * + * @param string $description shaare's description. + * @param string $redirector if a redirector is set, use it to gerenate links. + * @param bool $urlEncode Use `urlencode()` on the URL after the redirector or not. + * @param string $indexUrl URL to Shaarli's index. + + * @return string formatted description. + */ +function format_description($description, $redirector = '', $urlEncode = true, $indexUrl = '') { + return nl2br(space2nbsp(hashtag_autolink(text2clickable($description, $redirector, $urlEncode), $indexUrl))); +} + +/** + * Generate a small hash for a link. + * + * @param DateTime $date Link creation date. + * @param int $id Link ID. + * + * @return string the small hash generated from link data. + */ +function link_small_hash($date, $id) +{ + return smallHash($date->format(LinkDB::LINK_DATE_FORMAT) . $id); +}