Use cURL for Tweeper::get_contents() too
[tweeper.git] / tweeper.php
old mode 100755 (executable)
new mode 100644 (file)
index 2e27a0d..14c3a7a
@@ -22,37 +22,107 @@ date_default_timezone_set('UTC');
 
 class Tweeper {
 
-  private $USER_AGENT = "Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0";
+  private static $USER_AGENT = "Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0";
 
-  public function __construct($stylesheet) {
+  public function __construct($stylesheet, $generate_enclosure = FALSE) {
     $stylesheet_contents = $this->get_contents($stylesheet);
 
     $xslDoc = new DOMDocument();
     $xslDoc->loadXML($stylesheet_contents);
 
     $this->xsltProcessor = new XSLTProcessor();
+    $this->xsltProcessor->registerPHPFunctions();
+    $this->xsltProcessor->setParameter('', 'generateEnclosure', $generate_enclosure);
     $this->xsltProcessor->importStylesheet($xslDoc);
   }
 
-  private function get_contents($uri) {
-    # https://www.wjsams.com/c/docs/Wiki/Php-HowToSetUserAgentOnFileGetContents
-    $opts = array(
-      'http' => array(
-        'method' => "GET",
-        'header' => join(array(
-          "Accept-language: en\r\n",
-          "User-Agent: {$this->USER_AGENT}\r\n"
-        ))
-      )
-    );
+  public static function epoch_to_gmdate($timestamp)
+  {
+    return gmdate('D, d M Y H:i:s', $timestamp) . ' GMT';
+  }
+
+  public static function str_to_gmdate($date)
+  {
+    $timestamp = strtotime($date);
+    return Tweeper::epoch_to_gmdate($timestamp);
+  }
+
+  private static function get_contents($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);
 
-    $context = stream_context_create($opts);
-    $contents = file_get_contents($uri, false, $context);
     return $contents;
   }
 
+  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 $url_info;
+  }
+
+  public static function generate_enclosure($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",
+    );
+
+    $url_info = Tweeper::get_info($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 '';
+    }
+
+    $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);
+  }
+
   public function tweep($uri) {
-    $html = $this->get_contents($uri);
+    $html = Tweeper::get_contents($uri);
 
     $xmlDoc = new DOMDocument();
     $xmlDoc->loadHTML($html);
@@ -67,20 +137,76 @@ class Tweeper {
   }
 }
 
-if (isset($_GET['screen_name'])) {
-  $screen_name = $_GET['screen_name'];
-} else if (isset($argv[1])) {
-  $screen_name = $argv[1];
-} else {
-  if (isset($_SERVER['SCRIPT_NAME']))
-    $usage = htmlentities("{$_SERVER['SCRIPT_NAME']}?screen_name=<screen_name>");
+function usage($argv)
+{
+  if (php_sapi_name() != 'cli')
+    $usage = htmlentities("{$_SERVER['SCRIPT_NAME']}?src_url=<src_url>&generate_enclosure=<0|1>");
   else
-    $usage = "{$argv[0]} <screen_name>\n";
+    $usage = "{$argv[0]} [-e|-h|--help] <src_url>\n";
 
-  die("usage: $usage");
+  return "usage: $usage";
 }
 
-$tweeper = new Tweeper('twitter_user_timeline2rss.xsl');
+function parse_options_cli($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));
+    }
+  }
+
+  $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;
+}
+
+
+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: ${options['src_url']}\n");
+
+$stylesheet = "file://" . __DIR__ . "/rss_converter_" . $url["host"] . ".xsl";
+if (FALSE === file_exists($stylesheet))
+  die("Conversion to RSS not supported: {$url["host"]}\n");
 
-$src_uri = 'https://twitter.com/' . $screen_name;
-echo $tweeper->tweep($src_uri);
+$tweeper = new Tweeper($stylesheet, $options['generate_enclosure']);
+echo $tweeper->tweep($options['src_url']);