On Sun, 26 Jul 2009, Mambo Knave wrote:

> 0- access to a secure web site with userID/Passwd.
>
> 1- download from it a html page
> 2- change some content in it (buttons/menus)
> 3- resubmit it to the same web site
> 4- receive back the html response
> 5- convert the html response to text
>
> Step 0 must occur ONCE only. Steps 1-5 will occur hundreds of times and must
> be fast!
>
> What would be the easiest way to implement a scheme/program like that?

Use LiveHTTPHeaders to record the full HTTP stream of actions. Then repeat
that with (lib)curl.

> Is using LibCurl the proper approach?

Sure. Not the only one available, but certainly suitable.

> Can Curl command line do the same?

Yes, scripted of course.

-- 
  / daniel.haxx.se