X-Git-Url: https://git.ao2.it/tweeper.git/blobdiff_plain/1e04fc440d06b3871575bae5902167282fd1476e..110456377908327064a91a48e1f4c2c810b848fa:/tweeper.php diff --git a/tweeper.php b/tweeper.php index cd2c3ba..11e062b 100644 --- a/tweeper.php +++ b/tweeper.php @@ -20,6 +20,8 @@ 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); @@ -35,12 +37,13 @@ function get_url_info($url) { $ch = curl_init($url); curl_setopt_array($ch, array( - CURLOPT_HEADER => TRUE, // get the header - CURLOPT_NOBODY => TRUE, // don't donwload body - CURLOPT_FOLLOWLOCATION => TRUE, // the magic sauce + CURLOPT_HEADER => TRUE, + CURLOPT_NOBODY => TRUE, + CURLOPT_FOLLOWLOCATION => TRUE, // follow http redirects to get the real URL CURLOPT_RETURNTRANSFER => TRUE, - CURLOPT_SSL_VERIFYHOST => FALSE, // suppress certain SSL errors + CURLOPT_SSL_VERIFYHOST => FALSE, CURLOPT_SSL_VERIFYPEER => FALSE, + CURLOPT_USERAGENT => USER_AGENT, )); curl_exec($ch); $url_info = curl_getinfo($ch);; @@ -49,25 +52,42 @@ function get_url_info($url) return $url_info; } -function generate_enclosure($url) { +function generate_enclosure($url) +{ $url_info = get_url_info($url); - $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); + 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 $USER_AGENT = "Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0"; - public function __construct($stylesheet) { $stylesheet_contents = $this->get_contents($stylesheet); @@ -86,7 +106,7 @@ class Tweeper { 'method' => "GET", 'header' => join(array( "Accept-language: en\r\n", - "User-Agent: {$this->USER_AGENT}\r\n" + "User-Agent: {" + USER_AGENT + "}\r\n" )) ) ); @@ -122,25 +142,47 @@ function usage($argv) return "usage: $usage"; } -if (isset($_GET['src_url'])) { - $src_url = $_GET['src_url']; -} else if (isset($argv[1])) { - if ($argv[1] == "-h" || $argv[1] == "--help") { +function parse_options_cli($argv, $argc) +{ + $options = array(); + + if ($argc < 2) + return $options; + + if ($argv[1] == "-h" || $argv[1] == "--help") echo usage($argv); - return 0; - } - $src_url = $argv[1]; -} else { - die(usage($argv)); + else + $options['src_url'] = $argv[1]; + + return $options; } -$url = parse_url($src_url); +function parse_options_query_string() +{ + $options = array(); + + if (isset($_GET['src_url'])) + $options['src_url'] = $_GET['src_url']; + + return $options; +} + + +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: $src_url\n"); + die("Invalid url: ${options['src_url']}\n"); $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']);