src/Tweeper.php: fix rendering Instagram images in some feed readers

[tweeper.git] / src / Tweeper.php

diff --git a/src/Tweeper.php b/src/Tweeper.php
index 9dc1381..2d3870f 100644 (file)
--- a/src/Tweeper.php
+++ b/src/Tweeper.php
@@ -36,7 +36,7 @@ date_default_timezone_set('UTC');
  */
 class Tweeper {
 
  */
 class Tweeper {
 

-  private static $userAgent = "Mozilla/5.0";
+  private static $userAgent = "APIs-Google (+https://developers.google.com/webmasters/APIs-Google.html)";

   private static $maxConnectionTimeout = 5;
   private static $maxConnectionRetries = 5;
 
   private static $maxConnectionTimeout = 5;
   private static $maxConnectionRetries = 5;
 


@@ -111,6 +111,17 @@ class Tweeper {
       }
     } while (curl_errno($ch) == CURLE_OPERATION_TIMEDOUT && ++$attempt < Tweeper::$maxConnectionRetries);
 
       }
     } while (curl_errno($ch) == CURLE_OPERATION_TIMEDOUT && ++$attempt < Tweeper::$maxConnectionRetries);
 

+    $response_code = curl_getinfo($ch, CURLINFO_RESPONSE_CODE);
+    if (FALSE === $response_code) {
+      trigger_error(curl_error($ch), E_USER_WARNING);
+      return FALSE;
+    }
+
+    if ($response_code >= 400) {
+      trigger_error("HTTP reponse code $response_code", E_USER_WARNING);
+      return FALSE;
+    }
+

     return $ret;
   }
 
     return $ret;
   }
 


@@ -223,6 +234,34 @@ class Tweeper {
   }
 
   /**
   }
 
   /**

+   * Generate a data URL.
+   */
+  public static function generateDataURL($url) {
+    $supported_content_types = [
+      "image/gif",
+      "image/jpeg",
+      "image/png",
+    ];
+
+    $url_info = Tweeper::getUrlInfo($url);
+    if (FALSE === $url_info) {
+      trigger_error("Failed to retrieve info for URL: " . $url, E_USER_WARNING);
+      return '';
+    }
+
+    $supported = in_array($url_info['content_type'], $supported_content_types);
+    if (!$supported) {
+      trigger_error("Unsupported data URL type \"" . $url_info['content_type'] . "\" for URL: " . $url_info['url'], E_USER_WARNING);
+      return '';
+    }
+
+    $base64Data = base64_encode(file_get_contents($url));
+    $dataURL = 'data: ' . $url_info['content_type'] . ';base64,' . $base64Data;
+
+    return $dataURL;
+  }
+
+  /**

    * Mimic the message from libxml.c::php_libxml_ctx_error_level()
    */
   private static function logXmlError($error) {
    * Mimic the message from libxml.c::php_libxml_ctx_error_level()
    */
   private static function logXmlError($error) {


@@ -301,13 +340,19 @@ class Tweeper {
 
     $data = json_decode($matches[1], $assoc = TRUE);
 
 
     $data = json_decode($matches[1], $assoc = TRUE);
 

-    // The "qe" object contains elements which will result in invalid XML
-    // element names, so remove it.
+    // Remove items that may contain elements which can result in invalid XML
+    // element names (for example names starting with a number).

     unset($data["qe"]);
     unset($data["qe"]);

-
-    // The "knobs" object contains elements with undefined namespaces, so
-    // remove it to silence an error message.
-    unset($data["knobs"]);
+    unset($data["knobx"]);
+    unset($data["to_cache"]);
+
+    // Stop here in case Instagram redirected to the login page, this can
+    // happen when too many consecutive requests have been made from the same
+    // IP.
+    if (array_key_exists("LoginAndSignupPage", $data["entry_data"])) {
+      trigger_error("Cannot open instagram page: redirected to Login page.\n", E_USER_WARNING);
+      return NULL;
+    }

 
     $json = json_encode($data);
 
 
     $json = json_encode($data);
 


@@ -336,6 +381,9 @@ class Tweeper {
     $get_xml_host_method = 'getXml' . Tweeper::toUpperCamelCase($host, '.');
     if (method_exists($this, $get_xml_host_method)) {
       $xml_data = call_user_func_array([$this, $get_xml_host_method], [$html]);
     $get_xml_host_method = 'getXml' . Tweeper::toUpperCamelCase($host, '.');
     if (method_exists($this, $get_xml_host_method)) {
       $xml_data = call_user_func_array([$this, $get_xml_host_method], [$html]);

+      if (NULL === $xml_data) {
+        return NULL;
+      }

       $xmlDoc->loadXML($xml_data);
     }
     else {
       $xmlDoc->loadXML($xml_data);
     }
     else {