Make get_contents() a static method

[tweeper.git] / tweeper.php

diff --git a/tweeper.php b/tweeper.php
index cd2c3ba..6cdf8cf 100644 (file)
--- 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);;
@@ -49,14 +52,38 @@ function get_url_info($url)
   return $url_info;
 }
 
-function generate_enclosure($url) {
+function generate_enclosure($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",
+  );
+
   $url_info = get_url_info($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 '';
+  }
+
   $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']);
+  $enc->setAttribute('url', $url_info['url']);
+  $enc->setAttribute('length', $url_info['download_content_length']);
+  $enc->setAttribute('type', $url_info['content_type']);
 
   $dom->appendChild($enc);
 
@@ -66,9 +93,7 @@ function generate_enclosure($url) {
 
 class Tweeper {
 
-  private $USER_AGENT = "Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0";
-
-  public function __construct($stylesheet) {
+  public function __construct($stylesheet, $generate_enclosure = FALSE) {
     $stylesheet_contents = $this->get_contents($stylesheet);
 
     $xslDoc = new DOMDocument();
@@ -76,17 +101,18 @@ class Tweeper {
 
     $this->xsltProcessor = new XSLTProcessor();
     $this->xsltProcessor->registerPHPFunctions();
+    $this->xsltProcessor->setParameter('', 'generateEnclosure', $generate_enclosure);
     $this->xsltProcessor->importStylesheet($xslDoc);
   }
 
-  private function get_contents($uri) {
+  private static function get_contents($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"
+          "User-Agent: {" + USER_AGENT + "}\r\n"
         ))
       )
     );
@@ -97,7 +123,7 @@ class Tweeper {
   }
 
   public function tweep($uri) {
-    $html = $this->get_contents($uri);
+    $html = Tweeper::get_contents($uri);
 
     $xmlDoc = new DOMDocument();
     $xmlDoc->loadHTML($html);
@@ -115,32 +141,73 @@ class Tweeper {
 function usage($argv)
 {
   if (php_sapi_name() != 'cli')
-    $usage = htmlentities("{$_SERVER['SCRIPT_NAME']}?src_url=<src_url>");
+    $usage = htmlentities("{$_SERVER['SCRIPT_NAME']}?src_url=<src_url>&generate_enclosure=<0|1>");
   else
-    $usage = "{$argv[0]} <src_url>\n";
+    $usage = "{$argv[0]} [-e|-h|--help] <src_url>\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;
 }
 
-$url = parse_url($src_url);
+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: $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);
+$tweeper = new Tweeper($stylesheet, $options['generate_enclosure']);
+echo $tweeper->tweep($options['src_url']);