tweeper.php: support host-specific methods for preprocessing the HTML data

[tweeper.git] / tweeper.php

diff --git a/tweeper.php b/tweeper.php
index 9f09936..efc0fd6 100644 (file)
--- a/tweeper.php
+++ b/tweeper.php
@@ -32,12 +32,20 @@ class Tweeper {
 
   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);
+    if (FALSE === $timestamp) {
+      $timestamp = 0;
+    }
+
     return Tweeper::epoch_to_gmdate($timestamp);
   }
 
@@ -153,7 +161,7 @@ class Tweeper {
   private function load_stylesheet($host) {
     $stylesheet = "file://" . __DIR__ . "/rss_converter_" . $host . ".xsl";
     if (FALSE === file_exists($stylesheet)) {
-      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;
     }
 
@@ -201,6 +209,10 @@ class Tweeper {
     return $serializer->getSerializedData();
   }
 
+  private function get_xml_instagram_com($html) {
+    return $this->json_to_xml($html, '/window._sharedData = (.*);/', 'instagram');
+  }
+
   private function html_to_xml($html, $host) {
     $xmlDoc = new DOMDocument();
 
@@ -232,7 +244,10 @@ class Tweeper {
       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;
     }
@@ -242,7 +257,12 @@ class Tweeper {
       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;
     }
@@ -257,12 +277,18 @@ class Tweeper {
   }
 }
 
+function is_cli()
+{
+  return (substr(php_sapi_name(), 0, 3) == 'cli');
+}
+
 function usage($argv)
 {
-  if (php_sapi_name() != 'cli')
-    $usage = htmlentities("{$_SERVER['SCRIPT_NAME']}?src_url=<src_url>&generate_enclosure=<0|1>");
-  else
+  if (is_cli()) {
     $usage = "{$argv[0]} [-e|-h|--help] <src_url>\n";
+  } else {
+    $usage = htmlentities("{$_SERVER['SCRIPT_NAME']}?src_url=<src_url>&generate_enclosure=<0|1>");
+  }
 
   return "usage: $usage";
 }
@@ -313,12 +339,12 @@ function parse_options_query_string()
 }
 
 
-if (php_sapi_name() != 'cli') {
-  $options = parse_options_query_string();
-  $ERROR_STREAM = fopen('php://output', 'w');
-} else {
+if (is_cli()) {
   $options = parse_options_cli($argv, $argc);
   $ERROR_STREAM = fopen('php://stderr', 'w');
+} else {
+  $options = parse_options_query_string();
+  $ERROR_STREAM = fopen('php://output', 'w');
 }
 
 if (!isset($options['src_url'])) {