[dmh2000] - a telnet console in your Win32 application

daveh at dmh2000 dot com

[home]

If you are used to programming in a Unix/Linux environment, then you know that you can SSH or telnet into a remote system, get a command shell, run interactive console programs, give keyboard input and observe their output. No big deal. But in Windows, its not so easy. There is no real support for remote command line access in the same sense. What is a developer to do?

One solution : put a telnet console right into your application. The basic telnet protocol makes it easy to include a simple telnet server as part of your application. This will let you telnet into the program remotely and interact with it. In this case you don't get a full command shell. You get the capability to read character input and write character output from your program. What you do with it is up to you.

All you need to do in the application is open a TCP stream socket, bind it to a user port, and wait for a connection. Once you get a connection from a telnet client on some remote machine (or even the local machine) you read and write characters. The telnet protocol supports various options and negotiations, but you can ignore those if you like and most clients will still work (I haven't found any that don't).

Application notes

• take into account is how the telnet client handles keyboard input. Some clients send keyboard input to the application on a keystroke by keystroke basis. Others by default use a line mode. Most clients have options that you can set to force whichever mode you want. Its up to your application how you want to work. Line mode is nice because the client side operator can do line editing before sending the characters. On the other hand if you need to respond to individual characters, then character mode may be the way to go.
• most telnet clients support ANSI/VT100 escape codes to clear screen, position cursor, and scroll windows. You can make a nice interactive interface using those codes. Remember that since you are the server, you don't have to parse those codes. You just send them to the client to control the client screen. It is easy to code up a set of #define macros to let you inject the control sequences into your output.
• The structure of the application is sort of like what you would have with any program that accepts keyboard input and gives console output. One additional requirement is an outer loop to accept connections from the telnet client, then an inner loop that processes the connection. You need this outer loop if you want the clients to be able to connect, disconnect and reconnect at will. In a normal Win32 console program that takes direct console input, you really don't have to worry about the input failing, or if it does, consider it some sort of fatal error. In a telnet server application the inner loop has to handle the case where the telnet client disconnects, which can happen at any time. This means that after any input you must check if the client disconnected and break to the outer loop to wait for the next connection. Take a look at the example program for the structure. See the example below.

Here is downloadable code for a library supporting the inclusion of a telnet server in your application, along with a simple example. It is written for Win32 sockets, although it would be easy to port to Linux or say, vxWorks sockets.

program structure example

	/*  ================================ */
	/*  outer loop waits for connections */
	/*  ================================ */
	for(;;) {
		/*  wait for a connection */
		status = telnet_connect(t,&addr,&port);
		if (status != 0) {
			/*  accept socket failed for some reason, quit completely */
			break;
		}
		printf("connected to %08lx:%d\n",addr,port);

		/*  ================================ */
		/*  inner loop runs the application  */
		/*  ================================ */
		for(;;) {

			/*  get one character, blocking (or poll with telnet_poll) */
			status = telnet_get(t,&ch,1);
			if (status == TELNET_LOSTCOMM) {
				/*  lost connection, break to outer loop and try to reconnect */
				printf("lost connection 1\n");
				break;
			}
			
			/** got a character, do something with it */

			/*  write something out in printf syntax */
			status = telnet_printf(t,"\n\rreceived a character : %c\n\r",ch);

			/*  write a string */
			status = telnet_puts(t,"hello world\n\r");

			/*  you can check status on output functions, but it isn't necessary. */
			/*  the loss of connection is always detected on the telnet_get or telnet_poll call */
		}

		/*  close the client connection */
		telnet_disconnect(t);
	}