TODO: remove item about duplicated RSS items

[tweeper.git] / tweeper.php

diff --git a/tweeper.php b/tweeper.php
index 37b73f5..5e9d242 100644 (file)
--- a/tweeper.php
+++ b/tweeper.php
@@ -32,12 +32,20 @@ class Tweeper {
 
   public static function epoch_to_gmdate($timestamp)
   {
 
   public static function epoch_to_gmdate($timestamp)
   {

+    if (!is_numeric($timestamp) || is_nan($timestamp)) {
+      $timestamp = 0;
+    }
+

     return gmdate('D, d M Y H:i:s', $timestamp) . ' GMT';
   }
 
   public static function str_to_gmdate($date)
   {
     $timestamp = strtotime($date);
     return gmdate('D, d M Y H:i:s', $timestamp) . ' GMT';
   }
 
   public static function str_to_gmdate($date)
   {
     $timestamp = strtotime($date);

+    if (FALSE === $timestamp) {
+      $timestamp = 0;
+    }
+

     return Tweeper::epoch_to_gmdate($timestamp);
   }
 
     return Tweeper::epoch_to_gmdate($timestamp);
   }
 


@@ -153,7 +161,7 @@ class Tweeper {
   private function load_stylesheet($host) {
     $stylesheet = "file://" . __DIR__ . "/rss_converter_" . $host . ".xsl";
     if (FALSE === file_exists($stylesheet)) {
   private function load_stylesheet($host) {
     $stylesheet = "file://" . __DIR__ . "/rss_converter_" . $host . ".xsl";
     if (FALSE === file_exists($stylesheet)) {

-      trigger_error("Conversion to RSS not supported: $host", E_USER_ERROR);
+      trigger_error("Conversion to RSS not supported for $host ($stylesheet not found)", E_USER_ERROR);

       return NULL;
     }
 
       return NULL;
     }
 


@@ -205,6 +213,12 @@ class Tweeper {
     return $this->json_to_xml($html, '/window._sharedData = (.*);/', 'instagram');
   }
 
     return $this->json_to_xml($html, '/window._sharedData = (.*);/', 'instagram');
   }
 

+  private function preprocess_html_facebook_com($html) {
+    $html = str_replace('<!--', '', $html);
+    $html = str_replace('-->', '', $html);
+    return $html;
+  }
+

   private function html_to_xml($html, $host) {
     $xmlDoc = new DOMDocument();
 
   private function html_to_xml($html, $host) {
     $xmlDoc = new DOMDocument();
 


@@ -236,7 +250,10 @@ class Tweeper {
       return NULL;
     }
 
       return NULL;
     }
 

-    $xsltProcessor = $this->load_stylesheet($url["host"]);
+    // Strip the leading www. to be more forgiving on input URLs
+    $host = preg_replace('/^www\./', '', $url["host"]);
+
+    $xsltProcessor = $this->load_stylesheet($host);

     if (NULL === $xsltProcessor) {
       return NULL;
     }
     if (NULL === $xsltProcessor) {
       return NULL;
     }


@@ -246,7 +263,12 @@ class Tweeper {
       return NULL;
     }
 
       return NULL;
     }
 

-    $xmlDoc = $this->html_to_xml($html, $url["host"]);
+    $preprocess_html_host_method = 'preprocess_html_' . str_replace(".", "_", $host);
+    if (method_exists($this, $preprocess_html_host_method)) {
+      $html = call_user_func_array(array($this, $preprocess_html_host_method), array($html));
+    }
+
+    $xmlDoc = $this->html_to_xml($html, $host);

     if (NULL === $xmlDoc) {
       return NULL;
     }
     if (NULL === $xmlDoc) {
       return NULL;
     }