only enclosify certain mimetypes, use same user agent

[tweeper.git] / tweeper.php

diff --git a/tweeper.php b/tweeper.php
old mode 100755 (executable)
new mode 100644 (file)
index a275a64..c0db320
--- a/tweeper.php
+++ b/tweeper.php
@@ -20,14 +20,73 @@
 
 date_default_timezone_set('UTC');
 
 
 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';
 }
 
 {
   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);
 
   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",
         'method' => "GET",
         'header' => join(array(
           "Accept-language: en\r\n",

-          "User-Agent: {$this->USER_AGENT}\r\n"
+          "User-Agent: {" + USER_AGENT + "}\r\n"

         ))
       )
     );
         ))
       )
     );


@@ -86,6 +145,10 @@ function usage($argv)
 if (isset($_GET['src_url'])) {
   $src_url = $_GET['src_url'];
 } else if (isset($argv[1])) {
 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;
+  }

   $src_url = $argv[1];
 } else {
   die(usage($argv));
   $src_url = $argv[1];
 } else {
   die(usage($argv));


@@ -93,16 +156,11 @@ if (isset($_GET['src_url'])) {
 
 $url = parse_url($src_url);
 if (FALSE === $url || empty($url["host"]))
 
 $url = parse_url($src_url);
 if (FALSE === $url || empty($url["host"]))

-  die("Invalid url\n");
-
-$stylesheets = array(
-  "twitter.com" => "twitter_user_timeline2rss.xsl",
-);
-
-if (FALSE === array_key_exists($url["host"], $stylesheets))
-  die("Unsupported host\n");
+  die("Invalid url: $src_url\n");

 
 

-$stylesheet = $stylesheets[$url["host"]];
+$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);
 echo $tweeper->tweep($src_url);