Add initial support for scraping Pump.io activity streams

[tweeper.git] / tweeper.php

diff --git a/tweeper.php b/tweeper.php
old mode 100755 (executable)
new mode 100644 (file)
index 2e27a0d..3b8fb2d
--- a/tweeper.php
+++ b/tweeper.php
@@ -20,6 +20,17 @@
 
 date_default_timezone_set('UTC');
 
 
 date_default_timezone_set('UTC');
 

+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';
+}
+

 class Tweeper {
 
   private $USER_AGENT = "Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0";
 class Tweeper {
 
   private $USER_AGENT = "Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0";


@@ -31,6 +42,7 @@ class Tweeper {
     $xslDoc->loadXML($stylesheet_contents);
 
     $this->xsltProcessor = new XSLTProcessor();
     $xslDoc->loadXML($stylesheet_contents);
 
     $this->xsltProcessor = new XSLTProcessor();

+    $this->xsltProcessor->registerPHPFunctions();

     $this->xsltProcessor->importStylesheet($xslDoc);
   }
 
     $this->xsltProcessor->importStylesheet($xslDoc);
   }
 


@@ -67,20 +79,35 @@ 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>");

   else
   else

-    $usage = "{$argv[0]} <screen_name>\n";
+    $usage = "{$argv[0]} <src_url>\n";

 
 

-  die("usage: $usage");
+  return "usage: $usage";

 }
 
 }
 

-$tweeper = new Tweeper('twitter_user_timeline2rss.xsl');
+if (isset($_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));
+}
+
+$url = parse_url($src_url);
+if (FALSE === $url || empty($url["host"]))
+  die("Invalid url: $url\n");
+
+$stylesheet = __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);
+echo $tweeper->tweep($src_url);