Bgbot — a MediaWiki robot Code

--- a/trunk/browser.php
+++ b/trunk/browser.php
@@ -1,6 +1,4 @@
 <?php
-# $Id$
-#
 #  browser utils
 #
 #  Copyright (C) 2004 Borislav Manolov
@@ -20,46 +18,51 @@
 #  59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
 #  http://www.gnu.org/copyleft/gpl.html
 #
-#  Author: Borislav Manolov <b.manolov at web dot de>
-#          http://purl.oclc.org/NET/borislav/
-#
-#  This program uses portions of 
+#  Author: Borislav Manolov <b.manolov at gmail dot com>
+#          http://purl.org/NET/borislav/
+#
+#  This program uses portions of
 #    Snoopy - the PHP net client
 #    Author: Monte Ohrt <monte@ispi.net>
 #    Copyright (c): 1999-2000 ispi, all rights reserved
 #    Version: 1.01
 #    http://snoopy.sourceforge.net/
 #############################################################################
+require_once('utils.php');
 
 class Browser {
 
-  var $host    = '';        // host for connection
-  var $agent   = 'Mozilla/5.0 (PHPBrowser)'; // user agent
-  var $cookies = array();   // cookies
-  var $print_cookies = false; // whether to print cookies
+	var $host    = '';        // host for connection
+	var $agent   = 'Mozilla/5.0 (PHPBrowser)'; // user agent
+	var $cookies = array();   // cookies
+	var $print_cookies = false; // whether to print cookies
 	var $cookies_file = 'cookies.txt';   // cookies
-  
-  var $content = '';        // content returned from server
-  var $headers = array();   // headers returned from server
-
-  var $error        = '';   // error messages
-  var $conn_timeout = 120;   // timeout for socket connection
-  var $is_redirect = false; // true if the fetched page is a redirect
-
-  var $fetch_method  = 'GET';     // fetch method
-  var $submit_method = 'POST';    // submit method
-  var $http_version  = 'HTTP/1.1';// http version
-  var $content_type  = array(     // content types
-    'text' => 'application/x-www-form-urlencoded',
-    'binary' => 'multipart/form-data'
-  );
-  var $mime_boundary = ''; // MIME boundary for binary submission
+
+	# data for basic HTTP Authentication
+	var $user = '';
+	var $pass = '';
+
+	var $content = '';        // content returned from server
+	var $headers = array();   // headers returned from server
+
+	var $error        = '';   // error messages
+	var $conn_timeout = 120;   // timeout for socket connection
+	var $is_redirect = false; // true if the fetched page is a redirect
+
+	var $fetch_method  = 'GET';     // fetch method
+	var $submit_method = 'POST';    // submit method
+	var $http_version  = 'HTTP/1.1';// http version
+	var $content_type  = array(     // content types
+		'text' => 'application/x-www-form-urlencoded',
+		'binary' => 'multipart/form-data'
+	);
+	var $mime_boundary = ''; // MIME boundary for binary submission
 
 
   # constructor
   # $params - assoc array (name => value)
   # return nothing
-  function Browser($params) {
+  function Browser($params = array()) {
     settype($params, 'array');
     foreach ( $params as $field => $value ) {
       if ( isset($this->$field) ) {
@@ -73,9 +76,10 @@
 
   # fetch a page
   # $uri - location of the page
+	# $do_auth:boolean - add an authentication header
   # return true by success
-  function fetch($uri) {
-    return $this->make_request($uri, $this->fetch_method);
+  function fetch($uri, $do_auth = false) {
+    return $this->make_request($uri, $this->fetch_method, '', '', $do_auth);
   }
 
 
@@ -84,8 +88,9 @@
   # $vars - assoc array with form fields and their values
   # $file - assoc array (field name => file name)
   #         set only by upload
+	# $do_auth:boolean - add an authentication header
   # return true by success
-  function submit( $uri, $vars, $file = array() ) {
+  function submit( $uri, $vars, $file = array(), $do_auth = false ) {
     $postdata = '';
     if ( empty($file) ) {
       foreach ( $vars as $key => $val ) {
@@ -110,8 +115,8 @@
       $base_name = basename($file_name);
 
       $postdata .= '--'. $this->mime_boundary ."\r\n";
-      $postdata .= '"Content-Disposition: form-data; name="'.
-                   $field_name . '"; filename="' . $base_name . "\"\r\n\r\n";
+      $postdata .= 'Content-Disposition: form-data; name="'. $field_name .
+				'"; filename="' . $base_name . "\"\r\n\r\n";
       $postdata .= $file_content . "\r\n";
       $postdata .= '--'. $this->mime_boundary ."--\r\n";
     }
@@ -121,7 +126,7 @@
                   : $this->content_type['binary'] ;
 
     return $this->make_request($uri, $this->submit_method,
-                               $content_type, $postdata);
+                               $content_type, $postdata, $do_auth);
   }
 
 
@@ -130,8 +135,9 @@
   # $request_method - GET / POST
   # $content_type - content type (for POST submission)
   # $postdata - data (for POST submission)
+	# $do_auth:boolean - add an authentication header based on $this->user and $this->pass
   # return true if the request succeeded, false otherwise
-  function make_request($uri, $request_method, $content_type = '', $postdata = '') {
+  function make_request($uri, $request_method, $content_type = '', $postdata = '', $do_auth = false) {
     $uri_parts = parse_url($uri);
     if ( $uri_parts['scheme'] != 'http') { // not a valid protocol
       $this->error = "Invalid protocol: $uri_parts[scheme]";
@@ -156,12 +162,18 @@
     if ( empty($path) ) { $path = '/'; }
 		$headers = "$request_method $path $this->http_version\r\n" .
 			"User-Agent: $this->agent\r\nHost: $this->host\r\nAccept: */*\r\n";
+
+	if ($do_auth) {
+		$headers .= 'Authorization: Basic '.
+			base64_encode($this->user.':'.$this->pass) . "\r\n";
+	}
+
     if ( isset($this->cookies[$this->host]) ) {
 			$cookie_headers .= 'Cookie: ';
       foreach ($this->cookies[$this->host] as $cookie_name => $cookie_data) {
         $cookie_headers .= $cookie_name .'='. urlencode($cookie_data[0]) .'; ';
       }
-			# add $cookie_headers w/o last 2 chars + "\r\n"
+			# add $cookie_headers w/o last 2 chars
 			$headers .= substr($cookie_headers, 0, -2) . "\r\n";
     }
 
@@ -209,45 +221,45 @@
 
 	# read cookies from file
 	function read_cookies() {
-		$curr_time = time();
-		$lines = file($this->cookies_file);
-		foreach ($lines as $line) {
-			$line = trim($line);
-			if ( empty($line) ) { continue; }  
-			list($host, $cookie_expire, $cookie_name, $cookie_val) = explode("\t", $line);
-			# add cookie if not expired
-			if ($curr_time < $cookie_expire) {
-				$this->cookies[$host][$cookie_name] = array($cookie_val, $cookie_expire);
+		if (file_exists($this->cookies_file)) {
+			$curr_time = time();
+			$lines = file($this->cookies_file);
+			foreach ($lines as $line) {
+				$line = trim($line);
+				if ( empty($line) ) { continue; }
+				list($host, $cookie_expire, $cookie_name, $cookie_val) = explode("\t", $line);
+				# add cookie if not expired
+				if ($curr_time < $cookie_expire) {
+					$this->cookies[$host][$cookie_name] = array($cookie_val, $cookie_expire);
+				}
 			}
-		}
-		# write not expired cookies back to file
-		$cookies_str = '';
-		foreach ($this->cookies as $host => $cookie_data) {
-			foreach ($cookie_data as $cookie_name => $cookie_subdata) {
-				$cookies_str .= "$host\t$cookie_subdata[1]\t$cookie_name\t$cookie_subdata[0]\n";
+			# write not expired cookies back to file
+			$cookies_str = '';
+			foreach ($this->cookies as $host => $cookie_data) {
+				foreach ($cookie_data as $cookie_name => $cookie_subdata) {
+					$cookies_str .= "$host\t$cookie_subdata[1]\t$cookie_name\t$cookie_subdata[0]\n";
+				}
 			}
 		}
 		my_fwrite($this->cookies_file, $cookies_str, 'w');
 	}
 
-  # set cookies for a redirection
-  function set_cookies() {
+	# set cookies
+	function set_cookies() {
 		$len = count($this->headers);
 		$cookies_str = '';
-    for ($i = 0; $i < $len; $i++) {
-      if (preg_match('/^Set-Cookie:\s+([^=]+)=([^;]+);\s+expires=([^;]+)/i', 
+		for ($i = 0; $i < $len; $i++) {
+			if (preg_match('/^Set-Cookie:\s+([^=]+)=([^;]+);\s+(expires=([^;]+))?/i',
 					$this->headers[$i], $matches)) {
-				$exp_time = strtotime($matches[3]);
+				$exp_time = isset($matches[4]) ? strtotime($matches[4]) : time() + 60*60*24*30;
 				$cookies_str .= "$this->host\t$exp_time\t$matches[1]\t$matches[2]\n";
-        $this->cookies[$this->host][$matches[1]] = $matches[2];
-        if ( $this->print_cookies ) {
-          echo "$matches[1] = $matches[2]; expires at $matches[3]\n";
-        }
-      }
-    }
+				$this->cookies[$this->host][$matches[1]] = array($matches[2], $matches[4]);
+				if ( $this->print_cookies ) {
+					echo "$matches[1] = $matches[2]; expires at $matches[4]\n";
+				}
+			}
+		}
 		my_fwrite($this->cookies_file, $cookies_str);
-  }
+	}
 
 } // end of class Browser
-
-?>