X-Git-Url: https://git.ao2.it/tweeper.git/blobdiff_plain/7678d6017c992b1e1d210c9294644ac153d28cb9..c754e06af5f84dda20441887efb4346a1729d01c:/tweeper.php diff --git a/tweeper.php b/tweeper.php index 3b8fb2d..e9345cc 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); @@ -31,11 +33,67 @@ 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) +{ + $supported_content_types = array( + "video/avi", + "image/gif", + "image/jpeg", + "audio/x-midi", + "audio/mpeg", + "video/mpeg", + "audio/vorbis", + "application/ogg", + "audio/webm", + "audio/aac", + "audio/mp4", + "audio/wav", + "video/mp4", + "video/ogg", + ); + + $url_info = get_url_info($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 ''; + } - private $USER_AGENT = "Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0"; + $dom = new DomDocument(); + $enc = $dom->createElement('enclosure'); + $enc->setAttribute( 'url', $url_info['url']); + $enc->setAttribute( 'length', $url_info['download_content_length']); + $enc->setAttribute( 'type', $url_info['content_type']); - public function __construct($stylesheet) { + $dom->appendChild($enc); + + return $dom->saveXML($enc); +} + + +class Tweeper { + + public function __construct($stylesheet, $generate_enclosure = FALSE) { $stylesheet_contents = $this->get_contents($stylesheet); $xslDoc = new DOMDocument(); @@ -43,6 +101,7 @@ class Tweeper { $this->xsltProcessor = new XSLTProcessor(); $this->xsltProcessor->registerPHPFunctions(); + $this->xsltProcessor->setParameter('', 'generateEnclosure', $generate_enclosure); $this->xsltProcessor->importStylesheet($xslDoc); } @@ -53,7 +112,7 @@ class Tweeper { 'method' => "GET", 'header' => join(array( "Accept-language: en\r\n", - "User-Agent: {$this->USER_AGENT}\r\n" + "User-Agent: {" + USER_AGENT + "}\r\n" )) ) ); @@ -82,32 +141,73 @@ class Tweeper { function usage($argv) { if (php_sapi_name() != 'cli') - $usage = htmlentities("{$_SERVER['SCRIPT_NAME']}?src_url="); + $usage = htmlentities("{$_SERVER['SCRIPT_NAME']}?src_url=&generate_enclosure=<0|1>"); else - $usage = "{$argv[0]} \n"; + $usage = "{$argv[0]} [-e|-h|--help] \n"; return "usage: $usage"; } -if (isset($_GET['src_url'])) { - $src_url = $_GET['src_url']; -} else if (isset($argv[1])) { - if ($argv[1] == "-h" || $argv[1] == "--help") { - echo usage($argv); - return 0; +function parse_options_cli($argv, $argc) +{ + $options = array( + 'generate_enclosure' => FALSE + ); + + if ($argc < 2) + return $options; + + $cli_options = getopt("eh", array("help")); + foreach ($cli_options as $opt => $val) { + switch ($opt) { + case 'e': + $options['generate_enclosure'] = TRUE; + break; + case 'h': + case 'help': + echo usage($argv); + exit(0); + default: + die(usage($argv)); + } } - $src_url = $argv[1]; -} else { - die(usage($argv)); + + $options['src_url'] = $argv[count($cli_options) + 1]; + + return $options; +} + +function parse_options_query_string() +{ + $options = array( + 'generate_enclosure' => FALSE + ); + + if (isset($_GET['src_url'])) + $options['src_url'] = $_GET['src_url']; + + if (isset($_GET['generate_enclosure'])) + $options['generate_enclosure'] = $_GET['generate_enclosure'] == 1; + + 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 = __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); +$tweeper = new Tweeper($stylesheet, $options['generate_enclosure']); +echo $tweeper->tweep($options['src_url']);