X-Git-Url: https://git.ao2.it/tweeper.git/blobdiff_plain/ae7d50a3877b17b61b5cdf32cb4e4943162ab004..66097964b2c241e20eae5af87fb4a14fd2c1d889:/src/Tweeper.php?ds=inline

diff --git a/src/Tweeper.php b/src/Tweeper.php
index 877e882..7b2654d 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 (Windows NT 6.1; WOW64; rv:64.0) Gecko/20100101 Firefox/64.0";
+  private static $userAgent = "Mozilla/5.0 (Linux; U; Android 4.3; en-us; SM-N900T Build/JSS15J)";
   private static $maxConnectionTimeout = 5;
   private static $maxConnectionRetries = 5;
 
@@ -53,11 +53,14 @@ class Tweeper {
    *   Enables showing multimedia content (images, videos) directly in the
    *   item description (enabled by default). Only some stylesheets supports
    *   this functionality (twitter, instagram, dilbert).
+   * @param bool $verbose_output
+   *   Enables showing non-fatal errors like XML parsing errors.
    */
-  public function __construct($generate_enclosure = FALSE, $show_usernames = TRUE, $show_multimedia = TRUE) {
+  public function __construct($generate_enclosure = FALSE, $show_usernames = TRUE, $show_multimedia = TRUE, $verbose_output = TRUE) {
     $this->generate_enclosure = $generate_enclosure;
     $this->show_usernames = $show_usernames;
     $this->show_multimedia = $show_multimedia;
+    $this->verbose_output = $verbose_output;
   }
 
   /**
@@ -84,6 +87,46 @@ 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, $delim = ' ') {
@@ -123,8 +166,6 @@ class Tweeper {
       CURLOPT_FOLLOWLOCATION => TRUE,
       CURLOPT_COOKIEFILE => "",
       CURLOPT_RETURNTRANSFER => TRUE,
-      CURLOPT_SSL_VERIFYHOST => FALSE,
-      CURLOPT_SSL_VERIFYPEER => FALSE,
       CURLOPT_HTTPHEADER => array('Accept-language: en'),
       CURLOPT_USERAGENT => Tweeper::$userAgent,
     ));
@@ -146,8 +187,6 @@ class Tweeper {
       // Follow http redirects to get the real URL.
       CURLOPT_FOLLOWLOCATION => TRUE,
       CURLOPT_RETURNTRANSFER => TRUE,
-      CURLOPT_SSL_VERIFYHOST => FALSE,
-      CURLOPT_SSL_VERIFYPEER => FALSE,
       CURLOPT_USERAGENT => Tweeper::$userAgent,
     ));
 
@@ -343,8 +382,10 @@ class Tweeper {
       $xmlDoc->loadHTML($html);
     }
 
-    foreach (libxml_get_errors() as $xml_error) {
-      Tweeper::logXmlError($xml_error);
+    if ($this->verbose_output) {
+      foreach (libxml_get_errors() as $xml_error) {
+        Tweeper::logXmlError($xml_error);
+      }
     }
     libxml_clear_errors();
     libxml_use_internal_errors($xml_errors_value);