Index: index.php
===================================================================
RCS file: /cvs/drupal/drupal/index.php,v
retrieving revision 1.96
diff -u -r1.96 index.php
--- index.php	20 Sep 2008 20:22:23 -0000	1.96
+++ index.php	12 Dec 2008 02:33:49 -0000
@@ -41,3 +41,10 @@
 }
 
 drupal_page_footer();
+
+require_once DRUPAL_ROOT . '/includes/browser/browser.inc';
+$browser = Browser::getInstance();
+$request = $browser->get('http://google.com');
+if ($request) {
+  file_put_contents('output.html', $request['content']);
+}
Index: includes/browser/wrapper.inc
===================================================================
RCS file: includes/browser/wrapper.inc
diff -N includes/browser/wrapper.inc
--- /dev/null	1 Jan 1970 00:00:00 -0000
+++ includes/browser/wrapper.inc	1 Jan 1970 00:00:00 -0000
@@ -0,0 +1,30 @@
+<?php
+// $Id$
+
+/**
+ * @file
+ * Text web browser for Drupal.
+ */
+
+interface HttpWrapper {
+  public function open();
+  
+  public function close();
+  
+  public function setUserAgent($agent);
+  
+  /**
+   * Make an HTTP GET request.
+   *
+   * @param $path
+   *   Drupal path or URL to request.
+   * @param $headers
+   *   An array containing additional HTTP request headers, each formatted as
+   *   "name: value".
+   * @return
+   *   Array of request information.
+   */
+  public function get($url, array $headers = array());
+  
+  public function post();
+}
Index: includes/browser/stream.inc
===================================================================
RCS file: includes/browser/stream.inc
diff -N includes/browser/stream.inc
--- /dev/null	1 Jan 1970 00:00:00 -0000
+++ includes/browser/stream.inc	1 Jan 1970 00:00:00 -0000
@@ -0,0 +1,3 @@
+<?php
+
+?>
Index: includes/browser/curl.inc
===================================================================
RCS file: includes/browser/curl.inc
diff -N includes/browser/curl.inc
--- /dev/null	1 Jan 1970 00:00:00 -0000
+++ includes/browser/curl.inc	1 Jan 1970 00:00:00 -0000
@@ -0,0 +1,127 @@
+<?php
+// $Id$
+
+/**
+ * @file
+ * Curl implementation of for the browser.
+ */
+
+class HttpWrapper_curl implements HttpWrapper {
+  /**
+   * The handle of the current cURL connection.
+   *
+   * @var resource
+   */
+  protected $handle;
+  
+  /**
+   * The current cookie file used by cURL.
+   *
+   * We do not reuse the cookies in further runs, so we do not need a file
+   * but we still need cookie handling, so we set the jar to NULL.
+   */
+  protected $cookieFile = NULL;
+  
+  protected $url;
+  
+  protected $headers;
+  
+  protected $content;
+  
+  public function Browser_curl() {
+    
+  }
+  
+  public function open() {
+    if (!isset($this->handle)) {
+      $this->handle = curl_init();
+      curl_setopt_array($this->handle, $this->getDefaultOptions());
+    }
+  }
+  
+  /**
+   * Close the cURL handler and unset the handler.
+   */
+  public function close() {
+    if (isset($this->handle)) {
+      curl_close($this->handle);
+      unset($this->handle);
+    }
+  }
+  
+  protected function getDefaultOptions() {
+    global $base_url;
+    return array(
+      CURLOPT_COOKIEJAR => $this->cookieFile,
+      CURLOPT_URL => $base_url,
+      CURLOPT_FOLLOWLOCATION => TRUE,
+      CURLOPT_RETURNTRANSFER => TRUE,
+      CURLOPT_SSL_VERIFYPEER => FALSE,  // Required to make the tests run on https://
+      CURLOPT_SSL_VERIFYHOST => FALSE,  // Required to make the tests run on https://
+      CURLOPT_HEADERFUNCTION => array(&$this, 'headerCallback'),
+    );
+  }
+  
+  public function setUserAgent($agent) {
+//    var_dump($this->handle);
+//    curl_setopt($this->handle, CURLOPT_USERAGENT, $agent);
+  }
+  
+  public function get($url, array $headers = array()) {
+    $this->execute(array(
+      CURLOPT_HTTPGET => TRUE,
+      CURLOPT_URL => $url,
+      CURLOPT_NOBODY => FALSE,
+      CURLOPT_HTTPHEADER => $headers,
+    ));
+
+    return $this->buildRequest();
+  }
+  
+  public function post() {
+    
+  }
+  
+  protected function buildRequest() {
+    if ($this->content) {
+      return array(
+        'url' => $this->url,
+        'headers' => $this->headers,
+        'content' => $this->content,
+      );
+    }
+    return FALSE;
+  }
+  
+  /**
+   * Performs a cURL exec with the specified options after calling curlConnect().
+   *
+   * @param $options
+   *   Custom cURL options.
+   * @return
+   *   Content returned from the exec.
+   */
+  protected function execute($options) {
+    $this->open();
+
+    curl_setopt_array($this->handle, $options);
+    $this->content = curl_exec($this->handle);
+    $this->url = curl_getinfo($this->handle, CURLINFO_EFFECTIVE_URL);
+    // $this->headers should be filled by headerCallback.
+  }
+  
+  /**
+   * Reads headers and stores in $headers array.
+   *
+   * @param $curlHandler
+   *   The cURL handler.
+   * @param $header
+   *   An header.
+   * @return 
+   *   The string length of the header. (required by cURL)
+   */
+  protected function headerCallback($handler, $header) {
+    $this->headers[] = $header;
+    return strlen($header);
+  }
+}
Index: includes/browser/browser.inc
===================================================================
RCS file: includes/browser/browser.inc
diff -N includes/browser/browser.inc
--- /dev/null	1 Jan 1970 00:00:00 -0000
+++ includes/browser/browser.inc	1 Jan 1970 00:00:00 -0000
@@ -0,0 +1,121 @@
+<?php
+// $Id$
+
+/**
+ * @file
+ * Text web browser for Drupal.
+ */
+
+/*
+ * TODO
+ * 
+ * Hooks for get, post events.
+ * HTTP authentication
+ * Allow custom headers
+ * Decide on parsing which will allow for meta refresh suppose
+ * Deal with Drupal-Assertion callback stuff.
+ * 
+ * Look at: http://pear.php.net/package/HTTP_Request2/download
+ */
+
+class Browser {
+  /**
+   * The DrupalBrowser instance.
+   *
+   * @var DrupalBrowser
+   */
+  protected static $browser;
+  
+  protected $wrapper;
+  
+  protected $additional_headers;
+  
+  /**
+   * The URL currently loaded in the browser.
+   *
+   * @var string
+   */
+  protected $url;
+  
+  /**
+   * The headers of the page currently loaded in the internal browser.
+   *
+   * @var Array
+   */
+  protected $headers;
+
+  /**
+   * The content of the page currently loaded in the internal browser.
+   *
+   * @var string
+   */
+  protected $content;
+  
+  protected function Browser() {
+    // TODO use variable to switch between curl and stream.
+    require_once './includes/browser/wrapper.inc';
+    require_once './includes/browser/curl.inc';
+    $this->wrapper = new HttpWrapper_curl();
+    
+    $this->setUserAgent('Drupal (+http://drupal.org/)');
+  }
+  
+  final public static function getInstance() {
+    if (!isset(self::$browser)) {
+      self::$browser = new Browser();
+    }
+    return self::$browser;
+  }
+  
+  public function setUserAgent($agent) {
+    $this->additional_headers['User-Agent'] = $agent;
+//    $this->wrapper->setUserAgent($agent);
+  }
+  
+  public function getUserAgent() {
+    return $this->additional_headers['User-Agent'];
+  }
+  
+  /**
+   * Retrieves a Drupal path or an absolute path.
+   *
+   * @param $path
+   *   Drupal path or URL to load into browser.
+   * @param $options
+   *   Options to be forwarded to url().
+   * @param $headers
+   *   An array containing additional HTTP request headers, each formatted as
+   *   "name: value".
+   * @return
+   *   The retrieved HTML string, also available as $this->getContent().
+   */
+  public function get($path, array $options = array(), array $headers = array()) {
+    $request = $this->wrapper->get($path, $options, $headers);
+    
+    // TODO Error check, look for meta refresh, etc.
+    $this->setState($request['url'], $request['headers'], $request['content']);
+    return $request;
+  }
+  
+  public function post() {
+    
+  }
+  
+  /**
+   * Gets the current raw HTML of the last requested page.
+   * 
+   * @return
+   *   Raw HTML of last requested page.
+   */
+  public function getContent() {
+    return $this->content;
+  }
+  
+  protected function setState($url, $headers, $content) {
+    $this->url = $url;
+    $this->headers = $headers;
+    $this->content = $content;
+    
+    module_invoke_all('browser_request', self::$browser); // TODO decide on hooks
+  }
+}