X-Git-Url: https://git.ao2.it/tweeper.git/blobdiff_plain/25c3e276f01838c8fa55e9ca59f01d2707392988..5333d3c8ec110a4349dfc3b56168a157afc70082:/src/Tweeper.php

diff --git a/src/Tweeper.php b/src/Tweeper.php
index 93ac9e0..22d16ab 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) 2013-2016  Antonio Ospite <ao2@ao2.it>
+ * Copyright (C) 2013-2018  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
@@ -36,13 +36,21 @@ date_default_timezone_set('UTC');
  */
 class Tweeper {
 
-  private static $userAgent = "Mozilla/5.0 (Windows NT 6.1; rv:22.0) Gecko/20130405 Firefox/22.0";
+  private static $userAgent = "Mozilla/5.0 (Windows NT 6.1; WOW64; rv:60.0) Gecko/20100101 Firefox/60.0";
 
   /**
-   * Constructor sets up {@link $generate_enclosure}.
+   * Create a new Tweeper object controlling optional settings.
+   *
+   * @param bool $generate_enclosure
+   *   Enables the creation of <enclosure/> elements (disabled by default).
+   * @param bool $show_usernames
+   *   Enables showing the username in front of the content for multi-user
+   *   sites (enabled by default). Only some stylesheets supports this
+   *   functionality (twitter, instagram, pump.io).
    */
-  public function __construct($generate_enclosure = FALSE) {
+  public function __construct($generate_enclosure = FALSE, $show_usernames = TRUE) {
     $this->generate_enclosure = $generate_enclosure;
+    $this->show_usernames = $show_usernames;
   }
 
   /**
@@ -116,7 +124,14 @@ class Tweeper {
       CURLOPT_SSL_VERIFYPEER => FALSE,
       CURLOPT_USERAGENT => Tweeper::$userAgent,
     ));
-    curl_exec($ch);
+
+    $ret = curl_exec($ch);
+    if (FALSE === $ret) {
+      trigger_error(curl_error($ch));
+      curl_close($ch);
+      return FALSE;
+    }
+
     $url_info = curl_getinfo($ch);
     if (FALSE === $url_info) {
       trigger_error(curl_error($ch));
@@ -152,6 +167,10 @@ class Tweeper {
     );
 
     $url_info = Tweeper::getUrlInfo($url);
+    if (FALSE === $url_info) {
+      error_log("Failed to retrieve info for URL: " . $url);
+      return '';
+    }
 
     $supported = in_array($url_info['content_type'], $supported_content_types);
     if (!$supported) {
@@ -249,7 +268,19 @@ class Tweeper {
       return NULL;
     }
 
-    return Tweeper::jsonToXml($matches[1], 'instagram');
+    $data = json_decode($matches[1], $assoc = TRUE);
+
+    // The "qe" object contains elements which will result in invalid XML
+    // element names, so remove it.
+    unset($data["qe"]);
+
+    // The "knobs" object contains elements with undefined namespaces, so
+    // remove it to silence an error message.
+    unset($data["knobs"]);
+
+    $json = json_encode($data);
+
+    return Tweeper::jsonToXml($json, 'instagram');
   }
 
   /**
@@ -300,6 +331,9 @@ class Tweeper {
     }
 
     $stylesheet_contents = Tweeper::getUrlContents($stylesheet);
+    if (FALSE === $stylesheet_contents) {
+      return NULL;
+    }
 
     $xslDoc = new DOMDocument();
     $xslDoc->loadXML($stylesheet_contents);
@@ -307,6 +341,7 @@ class Tweeper {
     $xsltProcessor = new XSLTProcessor();
     $xsltProcessor->registerPHPFunctions();
     $xsltProcessor->setParameter('', 'generate-enclosure', $this->generate_enclosure);
+    $xsltProcessor->setParameter('', 'show-usernames', $this->show_usernames);
     $xsltProcessor->importStylesheet($xslDoc);
 
     return $xsltProcessor;
@@ -315,21 +350,30 @@ class Tweeper {
   /**
    * Convert the site content to RSS.
    */
-  public function tweep($src_url) {
+  public function tweep($src_url, $host=NULL, $validate_scheme=TRUE) {
     $url = parse_url($src_url);
-    if (FALSE === $url || empty($url["host"])) {
+    if (FALSE === $url) {
       trigger_error("Invalid URL: $src_url", E_USER_ERROR);
       return NULL;
     }
 
-    $scheme = $url["scheme"];
-    if (!in_array($scheme, array("http", "https"))) {
-      trigger_error("unsupported scheme: $scheme", E_USER_ERROR);
-      return NULL;
+    if (TRUE === $validate_scheme) {
+      $scheme = $url["scheme"];
+      if (!in_array($scheme, array("http", "https"))) {
+        trigger_error("unsupported scheme: $scheme", E_USER_ERROR);
+        return NULL;
+      }
     }
 
-    // Strip the leading www. to be more forgiving on input URLs.
-    $host = preg_replace('/^www\./', '', $url["host"]);
+    // if the host is not given derive it from the URL
+    if (NULL === $host) {
+      if (empty($url["host"])) {
+        trigger_error("Invalid host in URL: $src_url", E_USER_ERROR);
+        return NULL;
+      }
+      // Strip the leading www. to be more forgiving on input URLs.
+      $host = preg_replace('/^www\./', '', $url["host"]);
+    }
 
     $xsltProcessor = $this->loadStylesheet($host);
     if (NULL === $xsltProcessor) {
@@ -352,11 +396,11 @@ class Tweeper {
     }
 
     $output = $xsltProcessor->transformToXML($xmlDoc);
-
     if (FALSE === $output) {
       trigger_error('XSL transformation failed.', E_USER_ERROR);
       return NULL;
     }
+
     return $output;
   }