src/Tweeper.php: set User-Agent to impersonate a Google crawler

[tweeper.git] / src / Tweeper.php

diff --git a/src/Tweeper.php b/src/Tweeper.php
index 10e07e3..4a32f56 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, $delim = ' ') {
@@ -157,7 +117,7 @@ class Tweeper {
   /**
    * Get the contents from a URL.
    */
-  private static function getUrlContents($url) {
+  private static function getUrlContents($url, $user_agent = NULL) {
     $ch = curl_init($url);
     curl_setopt_array($ch, [
       CURLOPT_HEADER => FALSE,
@@ -167,7 +127,7 @@ class Tweeper {
       CURLOPT_COOKIEFILE => "",
       CURLOPT_RETURNTRANSFER => TRUE,
       CURLOPT_HTTPHEADER => ['Accept-language: en'],
-      CURLOPT_USERAGENT => Tweeper::$userAgent,
+      CURLOPT_USERAGENT => isset($user_agent) ? $user_agent : Tweeper::$userAgent,
     ]);
     $contents = Tweeper::curlExec($ch);
     curl_close($ch);
@@ -178,7 +138,7 @@ class Tweeper {
   /**
    * Get the headers from a URL.
    */
-  private static function getUrlInfo($url) {
+  private static function getUrlInfo($url, $user_agent = NULL) {
     $ch = curl_init($url);
     curl_setopt_array($ch, [
       CURLOPT_HEADER => TRUE,
@@ -187,7 +147,7 @@ class Tweeper {
       // Follow http redirects to get the real URL.
       CURLOPT_FOLLOWLOCATION => TRUE,
       CURLOPT_RETURNTRANSFER => TRUE,
-      CURLOPT_USERAGENT => Tweeper::$userAgent,
+      CURLOPT_USERAGENT => isset($user_agent) ? $user_agent : Tweeper::$userAgent,
     ]);
 
     $ret = Tweeper::curlExec($ch);
@@ -403,7 +363,7 @@ 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;