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 091b030..2d3870f 100644 (file)
--- a/src/Tweeper.php
+++ b/src/Tweeper.php
@@ -36,7 +36,7 @@ date_default_timezone_set('UTC');
  */
 class Tweeper {
 
-  private static $userAgent = "Mozilla/5.0 (Linux; U; Android 4.3; en-us; SM-N900T Build/JSS15J)";
+  private static $userAgent = "APIs-Google (+https://developers.google.com/webmasters/APIs-Google.html)";
   private static $maxConnectionTimeout = 5;
   private static $maxConnectionRetries = 5;
 
@@ -87,46 +87,6 @@ class Tweeper {
   }
 
   /**
-   * Convert Twitter mobile date to the date format expected in a RSS document.
-   */
-  public static function twitterToRssDate($date) {
-    // Twitter uses relative timestamps in minutes for recent tweets.
-    if (preg_match('/^(\d+)m$/', $date, $matches)) {
-      $timestamp = strtotime("+" . $matches[1] . " min", time());
-      if (FALSE === $timestamp) {
-        $timestamp = 0;
-      }
-    }
-    else {
-      /*
-       * In case the time is specified put it after the date,
-       * to make it recognized by strptime().
-       */
-      if (preg_match('/(.*) - (.*)/', $date, $matches)) {
-        $date = $matches[2] . " " . $matches[1];
-      }
-
-      $timestamp = strtotime($date);
-      if (FALSE === $timestamp) {
-        $timestamp = 0;
-      }
-
-      /*
-       * The twitter mobile UI usually only specifies the month and the day, so
-       * strtotime($date) may interpret the date as being in the future.
-       *
-       * If the date is in the future it is probably in the same day but in the
-       * previous year.
-       */
-      if ($timestamp > time()) {
-        $timestamp = strtotime('-1 years', $timestamp);
-      }
-    }
-
-    return Tweeper::epochToRssDate($timestamp);
-  }
-
-  /**
    * Convert string to UpperCamelCase.
    */
   public static function toUpperCamelCase($str, $delim = ' ') {
@@ -151,13 +111,24 @@ class Tweeper {
       }
     } while (curl_errno($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;
   }
 
   /**
    * Get the contents from a URL.
    */
-  private static function getUrlContents($url) {
+  private static function getUrlContents($url, $user_agent = NULL) {
     $ch = curl_init($url);
     curl_setopt_array($ch, [
       CURLOPT_HEADER => FALSE,
@@ -167,7 +138,7 @@ class Tweeper {
       CURLOPT_COOKIEFILE => "",
       CURLOPT_RETURNTRANSFER => TRUE,
       CURLOPT_HTTPHEADER => ['Accept-language: en'],
-      CURLOPT_USERAGENT => Tweeper::$userAgent,
+      CURLOPT_USERAGENT => isset($user_agent) ? $user_agent : Tweeper::$userAgent,
     ]);
     $contents = Tweeper::curlExec($ch);
     curl_close($ch);
@@ -178,7 +149,7 @@ class Tweeper {
   /**
    * Get the headers from a URL.
    */
-  private static function getUrlInfo($url) {
+  private static function getUrlInfo($url, $user_agent = NULL) {
     $ch = curl_init($url);
     curl_setopt_array($ch, [
       CURLOPT_HEADER => TRUE,
@@ -187,7 +158,7 @@ class Tweeper {
       // Follow http redirects to get the real URL.
       CURLOPT_FOLLOWLOCATION => TRUE,
       CURLOPT_RETURNTRANSFER => TRUE,
-      CURLOPT_USERAGENT => Tweeper::$userAgent,
+      CURLOPT_USERAGENT => isset($user_agent) ? $user_agent : Tweeper::$userAgent,
     ]);
 
     $ret = Tweeper::curlExec($ch);
@@ -263,6 +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) {
@@ -341,13 +340,19 @@ class Tweeper {
 
     $data = json_decode($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($data["qe"]);
-
-    // The "knobs" object contains elements with undefined namespaces, so
-    // remove it to silence an error message.
-    unset($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);
 
@@ -376,6 +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]);
+      if (NULL === $xml_data) {
+        return NULL;
+      }
       $xmlDoc->loadXML($xml_data);
     }
     else {