X-Git-Url: https://git.ao2.it/tweeper.git/blobdiff_plain/d01598c2467cc1671bfc3c7028d250f3c0ca65bb..ad1a1aabaa6107083bc3a53d57d61e569daf2d36:/tweeper.php?ds=sidebyside diff --git a/tweeper.php b/tweeper.php index 74e9e2f..c2c5a04 100644 --- a/tweeper.php +++ b/tweeper.php @@ -20,110 +20,109 @@ date_default_timezone_set('UTC'); -define('USER_AGENT', 'Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0'); +class Tweeper { -function str_to_gmdate($date) -{ - $timestamp = strtotime($date); - return epoch_to_gmdate($timestamp); -} + private static $USER_AGENT = "Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0"; -function epoch_to_gmdate($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($ch); + public function __construct($stylesheet, $generate_enclosure = FALSE) { + $stylesheet_contents = $this->get_contents($stylesheet); - return $url_info; -} + $xslDoc = new DOMDocument(); + $xslDoc->loadXML($stylesheet_contents); -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", - ); + $this->xsltProcessor = new XSLTProcessor(); + $this->xsltProcessor->registerPHPFunctions(); + $this->xsltProcessor->setParameter('', 'generateEnclosure', $generate_enclosure); + $this->xsltProcessor->importStylesheet($xslDoc); + } - $url_info = get_url_info($url); + public static function epoch_to_gmdate($timestamp) + { + return gmdate('D, d M Y H:i:s', $timestamp) . ' GMT'; + } - $supported = in_array($url_info['content_type'], $supported_content_types); - if (!$supported) { - error_log("Unsupported enclosure content type \"" . $url_info['content_type'] . "\" for URL: " . $url_info['url']); - return ''; + public static function str_to_gmdate($date) + { + $timestamp = strtotime($date); + return Tweeper::epoch_to_gmdate($timestamp); } - $dom = new DomDocument(); - $enc = $dom->createElement('enclosure'); - $enc->setAttribute('url', $url_info['url']); - $enc->setAttribute('length', $url_info['download_content_length']); - $enc->setAttribute('type', $url_info['content_type']); + private static function get_contents($url) + { + $ch = curl_init($url); + curl_setopt_array($ch, array( + CURLOPT_HEADER => FALSE, + CURLOPT_RETURNTRANSFER => TRUE, + CURLOPT_SSL_VERIFYHOST => FALSE, + CURLOPT_SSL_VERIFYPEER => FALSE, + CURLOPT_HTTPHEADER => array('Accept-language: en'), + CURLOPT_USERAGENT => Tweeper::$USER_AGENT, + )); + $contents = curl_exec($ch); + curl_close($ch); - $dom->appendChild($enc); + return $contents; + } - return $dom->saveXML($enc); -} + 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 $url_info; + } -class Tweeper { + public static function generate_enclosure($url) + { + $supported_content_types = array( + "application/ogg", + "audio/aac", + "audio/mp4", + "audio/mpeg", + "audio/vorbis", + "audio/wav", + "audio/webm", + "audio/x-midi", + "image/gif", + "image/jpeg", + "video/avi", + "video/mp4", + "video/mpeg", + "video/ogg", + ); - public function __construct($stylesheet, $generate_enclosure = FALSE) { - $stylesheet_contents = $this->get_contents($stylesheet); + $url_info = Tweeper::get_info($url); - $xslDoc = new DOMDocument(); - $xslDoc->loadXML($stylesheet_contents); + $supported = in_array($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 ''; + } - $this->xsltProcessor = new XSLTProcessor(); - $this->xsltProcessor->registerPHPFunctions(); - $this->xsltProcessor->setParameter('', 'generateEnclosure', $generate_enclosure); - $this->xsltProcessor->importStylesheet($xslDoc); - } + $dom = new DomDocument(); + $enc = $dom->createElement('enclosure'); + $enc->setAttribute('url', $url_info['url']); + $enc->setAttribute('length', $url_info['download_content_length']); + $enc->setAttribute('type', $url_info['content_type']); - private function get_contents($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" - )) - ) - ); + $dom->appendChild($enc); - $context = stream_context_create($opts); - $contents = file_get_contents($uri, false, $context); - return $contents; + return $dom->saveXML($enc); } public function tweep($uri) { - $html = $this->get_contents($uri); + $html = Tweeper::get_contents($uri); $xmlDoc = new DOMDocument(); $xmlDoc->loadHTML($html); @@ -205,7 +204,7 @@ $url = parse_url($options['src_url']); if (FALSE === $url || empty($url["host"])) die("Invalid url: ${options['src_url']}\n"); -$stylesheet = __DIR__ . "/rss_converter_" . $url["host"] . ".xsl"; +$stylesheet = "file://" . __DIR__ . "/rss_converter_" . $url["host"] . ".xsl"; if (FALSE === file_exists($stylesheet)) die("Conversion to RSS not supported: {$url["host"]}\n");