X-Git-Url: https://git.ao2.it/tweeper.git/blobdiff_plain/6e627479745f13ec573e052e975bf23905f5b064..d38cac35c8e0aed0a223533bb54e46958c693212:/tweeper.php?ds=inline diff --git a/tweeper.php b/tweeper.php old mode 100755 new mode 100644 index 58c1642..45bc786 --- a/tweeper.php +++ b/tweeper.php @@ -20,14 +20,73 @@ date_default_timezone_set('UTC'); -function formatDate($timestamp) +define('USER_AGENT', 'Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0'); + +function str_to_gmdate($date) +{ + $timestamp = strtotime($date); + return epoch_to_gmdate($timestamp); +} + +function epoch_to_gmdate($timestamp) { return gmdate('D, d M Y H:i:s', $timestamp) . ' GMT'; } -class Tweeper { +function get_url_info($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; + } +} - private $USER_AGENT = "Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0"; + +class Tweeper { public function __construct($stylesheet) { $stylesheet_contents = $this->get_contents($stylesheet); @@ -47,7 +106,7 @@ class Tweeper { 'method' => "GET", 'header' => join(array( "Accept-language: en\r\n", - "User-Agent: {$this->USER_AGENT}\r\n" + "User-Agent: {" + USER_AGENT + "}\r\n" )) ) ); @@ -78,26 +137,61 @@ function usage($argv) if (php_sapi_name() != 'cli') $usage = htmlentities("{$_SERVER['SCRIPT_NAME']}?src_url="); else - $usage = "{$argv[0]} \n"; + $usage = "{$argv[0]} [-h|--help] \n"; return "usage: $usage"; } -if (isset($_GET['src_url'])) { - $src_url = $_GET['src_url']; -} else if (isset($argv[1])) { - $src_url = $argv[1]; -} else { - die(usage($argv)); +function parse_options_cli($argv, $argc) +{ + $options = array(); + + if ($argc < 2) + return $options; + + $cli_options = getopt("h", array("help")); + foreach ($cli_options as $opt => $val) { + switch ($opt) { + case 'h': + case 'help': + echo usage($argv); + exit(0); + default: + die(usage($argv)); + } + } + + $options['src_url'] = $argv[count($cli_options) + 1]; + + return $options; +} + +function parse_options_query_string() +{ + $options = array(); + + if (isset($_GET['src_url'])) + $options['src_url'] = $_GET['src_url']; + + return $options; } -$url = parse_url($src_url); + +if (php_sapi_name() != 'cli') + $options = parse_options_query_string(); +else + $options = parse_options_cli($argv, $argc); + +if (!isset($options['src_url'])) + die(usage($argv)); + +$url = parse_url($options['src_url']); if (FALSE === $url || empty($url["host"])) - die("Invalid url: $url\n"); + die("Invalid url: ${options['src_url']}\n"); -$stylesheet = "rss_converter_" . $url["host"] . ".xsl"; +$stylesheet = __DIR__ . "/rss_converter_" . $url["host"] . ".xsl"; if (FALSE === file_exists($stylesheet)) die("Conversion to RSS not supported: {$url["host"]}\n"); $tweeper = new Tweeper($stylesheet); -echo $tweeper->tweep($src_url); +echo $tweeper->tweep($options['src_url']);