X-Git-Url: https://git.ao2.it/tweeper.git/blobdiff_plain/af103c976dd4992d79e9d9a71837aecff30d6e9c..b922824bc561f7f3e31c6f9962d96e9084497ced:/src/Tweeper.php

diff --git a/src/Tweeper.php b/src/Tweeper.php
index 5b9f235..f1d579f 100644
--- a/src/Tweeper.php
+++ b/src/Tweeper.php
@@ -6,7 +6,7 @@ namespace Tweeper;
  * @file
  * Tweeper - a Twitter to RSS web scraper.
  *
- * Copyright (C) 2013-2018  Antonio Ospite <ao2@ao2.it>
+ * Copyright (C) 2013-2020  Antonio Ospite <ao2@ao2.it>
  *
  * This program is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -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 = "Mozilla/5.0";
   private static $maxConnectionTimeout = 5;
   private static $maxConnectionRetries = 5;
 
@@ -157,18 +157,18 @@ class Tweeper {
   /**
    * 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($ch, array(
+    curl_setopt_array($ch, [
       CURLOPT_HEADER => FALSE,
       CURLOPT_CONNECTTIMEOUT => Tweeper::$maxConnectionTimeout,
       // Follow http redirects to get the real URL.
       CURLOPT_FOLLOWLOCATION => TRUE,
       CURLOPT_COOKIEFILE => "",
       CURLOPT_RETURNTRANSFER => TRUE,
-      CURLOPT_HTTPHEADER => array('Accept-language: en'),
-      CURLOPT_USERAGENT => Tweeper::$userAgent,
-    ));
+      CURLOPT_HTTPHEADER => ['Accept-language: en'],
+      CURLOPT_USERAGENT => isset($user_agent) ? $user_agent : Tweeper::$userAgent,
+    ]);
     $contents = Tweeper::curlExec($ch);
     curl_close($ch);
 
@@ -178,17 +178,17 @@ 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($ch, array(
+    curl_setopt_array($ch, [
       CURLOPT_HEADER => TRUE,
       CURLOPT_NOBODY => TRUE,
       CURLOPT_CONNECTTIMEOUT => Tweeper::$maxConnectionTimeout,
       // Follow http redirects to get the real URL.
       CURLOPT_FOLLOWLOCATION => TRUE,
       CURLOPT_RETURNTRANSFER => TRUE,
-      CURLOPT_USERAGENT => Tweeper::$userAgent,
-    ));
+      CURLOPT_USERAGENT => isset($user_agent) ? $user_agent : Tweeper::$userAgent,
+    ]);
 
     $ret = Tweeper::curlExec($ch);
     if (FALSE === $ret) {
@@ -209,7 +209,7 @@ class Tweeper {
    * Generate an RSS <enclosure/> element.
    */
   public static function generateEnclosure($url) {
-    $supported_content_types = array(
+    $supported_content_types = [
       "application/octet-stream",
       "application/ogg",
       "application/pdf",
@@ -228,7 +228,7 @@ class Tweeper {
       "video/mp4",
       "video/mpeg",
       "video/ogg",
-    );
+    ];
 
     $url_info = Tweeper::getUrlInfo($url);
     if (FALSE === $url_info) {
@@ -310,13 +310,13 @@ class Tweeper {
 
     $encoder = new XmlEncoder();
     $normalizer = new ObjectNormalizer();
-    $serializer = new Serializer(array($normalizer), array($encoder));
+    $serializer = new Serializer([$normalizer], [$encoder]);
 
-    $serializer_options = array(
+    $serializer_options = [
       'xml_encoding' => "UTF-8",
       'xml_format_output' => TRUE,
       'xml_root_node_name' => $root_node_name,
-    );
+    ];
 
     $xml_data = $serializer->serialize($data, 'xml', $serializer_options);
     if (!$xml_data) {
@@ -375,7 +375,7 @@ class Tweeper {
     // If there is a host-specific method to get the XML data, use it!
     $get_xml_host_method = 'getXml' . Tweeper::toUpperCamelCase($host, '.');
     if (method_exists($this, $get_xml_host_method)) {
-      $xml_data = call_user_func_array(array($this, $get_xml_host_method), array($html));
+      $xml_data = call_user_func_array([$this, $get_xml_host_method], [$html]);
       $xmlDoc->loadXML($xml_data);
     }
     else {
@@ -403,7 +403,7 @@ class Tweeper {
       return NULL;
     }
 
-    $stylesheet_contents = Tweeper::getUrlContents($stylesheet);
+    $stylesheet_contents = file_get_contents($stylesheet);
     if (FALSE === $stylesheet_contents) {
       trigger_error("Cannot open $stylesheet", E_USER_WARNING);
       return NULL;
@@ -434,7 +434,7 @@ class Tweeper {
 
     if (TRUE === $validate_scheme) {
       $scheme = $url["scheme"];
-      if (!in_array($scheme, array("http", "https"))) {
+      if (!in_array($scheme, ["http", "https"])) {
         trigger_error("unsupported scheme: $scheme", E_USER_WARNING);
         return NULL;
       }
@@ -455,7 +455,15 @@ class Tweeper {
       return NULL;
     }
 
-    $html = Tweeper::getUrlContents($src_url);
+    // Override User-Agent for twitter.com to force it to serve the mobile UI.
+    if ($host == "twitter.com") {
+      $user_agent = "Mozilla/5.0 (Linux; U; Android 4.3; en-us; SM-N900T Build/JSS15J)";
+    }
+    else {
+      $user_agent = NULL;
+    }
+
+    $html = Tweeper::getUrlContents($src_url, $user_agent);
     if (FALSE === $html) {
       trigger_error("Failed to retrieve $src_url", E_USER_WARNING);
       return NULL;
@@ -463,7 +471,7 @@ class Tweeper {
 
     $preprocess_html_host_method = 'preprocessHtml' . Tweeper::toUpperCamelCase($host, '.');
     if (method_exists($this, $preprocess_html_host_method)) {
-      $html = call_user_func_array(array($this, $preprocess_html_host_method), array($html));
+      $html = call_user_func_array([$this, $preprocess_html_host_method], [$html]);
     }
 
     $xmlDoc = $this->htmlToXml($html, $host);