X-Git-Url: https://git.ao2.it/tweeper.git/blobdiff_plain/f82b9f8a651307d7c6ba52c2d1d7da3dcc696189..f10331c2001ba9ac72b8ccede428a98dcd9282e8:/tweeper.php diff --git a/tweeper.php b/tweeper.php index bdba04d..94d91c3 100644 --- a/tweeper.php +++ b/tweeper.php @@ -20,74 +20,10 @@ date_default_timezone_set('UTC'); -define('USER_AGENT', 'Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0'); - -function str_to_gmdate($date) -{ - $timestamp = strtotime($date); - return epoch_to_gmdate($timestamp); -} - -function epoch_to_gmdate($timestamp) -{ - return gmdate('D, d M Y H:i:s', $timestamp) . ' GMT'; -} - -function get_url_info($url) -{ - $ch = curl_init($url); - curl_setopt_array($ch, array( - CURLOPT_HEADER => TRUE, - CURLOPT_NOBODY => TRUE, - CURLOPT_FOLLOWLOCATION => TRUE, // follow http redirects to get the real URL - CURLOPT_RETURNTRANSFER => TRUE, - CURLOPT_SSL_VERIFYHOST => FALSE, - CURLOPT_SSL_VERIFYPEER => FALSE, - CURLOPT_USERAGENT => USER_AGENT, - )); - curl_exec($ch); - $url_info = curl_getinfo($ch);; - curl_close($ch); - - return $url_info; -} - -function generate_enclosure($url) -{ - $url_info = get_url_info($url); - - switch($url_info['content_type']) { - case "video/avi": - case "image/gif": - case "image/jpeg": - case "audio/x-midi": - case "audio/mpeg": - case "video/mpeg": - case "audio/vorbis": - case "application/ogg": - case "audio/webm": - case "audio/aac": - case "audio/mp4": - case "audio/wav": - case "video/mp4": - case "video/ogg": - $dom = new DomDocument(); - $enc = $dom->createElement('enclosure'); - $enc->setAttribute( 'url', $url_info['url']); - $enc->setAttribute( 'length', $url_info['download_content_length']); - $enc->setAttribute( 'type', $url_info['content_type']); - - $dom->appendChild($enc); - - return $dom->saveXML($enc); - default: - return; - } -} - - class Tweeper { + private static $USER_AGENT = "Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0"; + public function __construct($stylesheet, $generate_enclosure = FALSE) { $stylesheet_contents = $this->get_contents($stylesheet); @@ -100,14 +36,25 @@ class Tweeper { $this->xsltProcessor->importStylesheet($xslDoc); } - private function get_contents($uri) { + public static function epoch_to_gmdate($timestamp) + { + return gmdate('D, d M Y H:i:s', $timestamp) . ' GMT'; + } + + public static function str_to_gmdate($date) + { + $timestamp = strtotime($date); + return Tweeper::epoch_to_gmdate($timestamp); + } + + private static function get_contents($uri) { # https://www.wjsams.com/c/docs/Wiki/Php-HowToSetUserAgentOnFileGetContents $opts = array( 'http' => array( 'method' => "GET", 'header' => join(array( "Accept-language: en\r\n", - "User-Agent: {" + USER_AGENT + "}\r\n" + "User-Agent: {" + Tweeper::$USER_AGENT + "}\r\n" )) ) ); @@ -117,8 +64,65 @@ class Tweeper { return $contents; } + private static function get_info($url) + { + $ch = curl_init($url); + curl_setopt_array($ch, array( + CURLOPT_HEADER => TRUE, + CURLOPT_NOBODY => TRUE, + CURLOPT_FOLLOWLOCATION => TRUE, // follow http redirects to get the real URL + CURLOPT_RETURNTRANSFER => TRUE, + CURLOPT_SSL_VERIFYHOST => FALSE, + CURLOPT_SSL_VERIFYPEER => FALSE, + CURLOPT_USERAGENT => Tweeper::$USER_AGENT, + )); + curl_exec($ch); + $url_info = curl_getinfo($ch); + curl_close($ch); + + return $url_info; + } + + public static function generate_enclosure($url) + { + $supported_content_types = array( + "application/ogg", + "audio/aac", + "audio/mp4", + "audio/mpeg", + "audio/vorbis", + "audio/wav", + "audio/webm", + "audio/x-midi", + "image/gif", + "image/jpeg", + "video/avi", + "video/mp4", + "video/mpeg", + "video/ogg", + ); + + $url_info = Tweeper::get_info($url); + + $supported = in_array($url_info['content_type'], $supported_content_types); + if (!$supported) { + error_log("Unsupported enclosure content type \"" . $url_info['content_type'] . "\" for URL: " . $url_info['url']); + return ''; + } + + $dom = new DomDocument(); + $enc = $dom->createElement('enclosure'); + $enc->setAttribute('url', $url_info['url']); + $enc->setAttribute('length', $url_info['download_content_length']); + $enc->setAttribute('type', $url_info['content_type']); + + $dom->appendChild($enc); + + return $dom->saveXML($enc); + } + public function tweep($uri) { - $html = $this->get_contents($uri); + $html = Tweeper::get_contents($uri); $xmlDoc = new DOMDocument(); $xmlDoc->loadHTML($html);