Web page and cFP application.

[Khalid]

Hi,

I would like to input some parameters to my application -- running on a cFP controller -- using a web page from the cFP's embedded Web Server.

Here are the details: let's say I have a LabVIEW-built application (.exe) running on my cFP. The cFP controller has a Web Server. When enabled, it publishes any HTML documents from its root folder. I can browse this page from any other remote machine with access to the cFP. I don't need LabVIEW run-time engine to do this.

So far so good. I would like to have a web page with a button on it. When this button on the web page is pressed from a remote machine, I want to pass this as a boolean input to one of my VIs in the running application.

To complicate things further, the remote machines will not have the LabVIEW run-time installed. So, Remote Panels are ruled out.

If the application was running on a PC (instead of the cFP), I could have used the CGI VIs from the Internet toolkit to pass the boolean input from the web page to the VI. But I can't use CGI VIs on cFP.

What options do I have for doing something like this?

Any input (pun intended) will be much appreciated.

-Khalid

[Matthew_Williams]

Two suggestions, neither particularly elegant:

1) In one of our cFP application we built a small web server that was just smart enough to
listen for our expected URL requests and reply (plus a default error page).  This gives
you complete control over how the application responds to any request so passing out
a boolean value when a certain link (button) is pressed is easy.  The disadvantage is
the reduced functionality of the web server and increased complexity of handling pages.

2) A 'wrapper' that just listens for a specific URL but otherwise passes any requests
through to the built in web server.  Listen on ip.address:8080 for the URL of your button
(hyperlink).  When it sees that specific URL respond to the application, otherwise pass
the request unaltered to the server (ip.address:80 or 127.0.0.1:80 should also work).
You will also have to handle the reply side of the conversation in the wrapper.

Hopefully someone can suggest simpler approaches!

Matt

[Khalid]

Thanks for the response Matthew. I like the ideas you have presented.

So to build this custom "Web Server" or even the wrapper, am I right in understanding you as to just have a TCP listener on port 8080? Any broad tips on building this would be appreciated.

Thank you and best regards,

-Khalid

[Matthew_Williams]

Take a look at the attached.  tcp_fcn.vi is the basic listener; pass in port 8080 in this case.  Same
.vi is used to send the reponse.  Your application will have to properly parse the http request and
response; don't forget to handle a generic 'not found' case 😉

Error_x_timeout.vi just passes through any errors except timeouts.

Matt

[Khalid]

Hi Matt,

Thanks for sharing the listener code. I am able to succesfully monitor any URL requests on a given port from a web page. I then parse the URL and get the values into my other VIs fine. Thanks for all the help so far!

However, I have one small problem: when I request the dummy URL from a web page, my browser is -- religiously -- waiting to display this URL. How do I respond to this? What would be a generic HTTP response? Should I be responding on a different port than that I am listening on (8000)? I guess I somehow need to send back a custom error page. Would appreciate if you can point me in the right direction for this as well.

Thanks much!

-Khalid

[Khalid]

Hi Matt,

After some head-scratching, and googling, I stumbled onto W3C's HHTP reference document. I wouldn't call it the most lucid document I have read. In any case, I figured out that I got to send a response of the following format:

HTTP/1.0 200 OK
(more header stuff)
CRLF
[Message Body]

I experimented with the above, and noticed that even though I sent the response right-away, the browser would not see the response until I closed the connection. Is this how it is supposed to be -- that each response is to be ended with a close-connection?

I am having a tough time figuring out the rest of the header for the HTTP response. Would appreciate if you could share the format of simple, working header.

Also, I was wondering if I can use the Status code of 307 ("Temporary Redirect") instead of the 200 ("OK") and direct the browser to the actual Web Server on the cFP (port 80). This way I don't have to respond with a page of my own. I however couldn't find where I would specify the forwarded address and port with the 307. Is this is even possible?

Thanks for any input.

Regards,

-Khalid

[Matthew_Williams]

Hi Khalid.  Attached is a response string exactly as it is submitted to the tcp_fcn.vi 'send' function.
You do have to close the connection with this approach.  There may be a way to avoid closing but
this met our needs 😉

I haven't tried any re-directs but I don't see why they wouldn't work once you get the HTTP syntax
figured out.  The W3C site is a good resource but it can be difficult to figure out exactly what you need.

Hope this helps.

Matt

[Matthew_Williams]

The browser will expect your response to be from port 8000 and to
the 'from' port of the original request.  Closing the connection on the
server side (as you figured out) will cause the transaction to
complete.  I had forgotten about the close myself; sorry about that.

Matt

Message Edited by Matthew Williams on 09-26-2006 08:20 AM

[Khalid]

Hi Matt,

Thanks for the help! I am able to succesfully respond back, actually I redirect to the built-in Web Server on port 80, using the 307 status code and the 'Location' entity.

This is pretty neat -- controlling a cFP app from a remote browser without LabVIEW RunTime Engine!

Thanks again.

-Khalid

[Matthew_Williams]

Cool!  Thanks for the update on the re-direct; now I know for certain.  Glad it worked out.

Matt