Question

Previously, I have used Tk/Expect along with plink (PuTTY) to automate remote shell operations. However, Expect sometimes fails if different shells have different prompts, or if prompt characters appear in non-prompt shell output (!)

I want to know if there is a way to tell if a remote shell is expecting input, either through plink, or some third party library. I am okay programming at the sockets layer if necessary.

Answer 1

No, there is no way to reliably detect a shell prompt at the level you’re working at. Some alternative options might be:

• Always run commands which will print some recognizable output when they complete — e.g, doSomething ; echo uniqueStringThatMeansThatWeAreDone — then search for that string.
• Make your automation script set the prompt to something it’ll recognize when it logs in. If you don’t know what shell you’re using, just exec the one you want.

• Use a SSH library which can execute commands directly, rather than trying to script a shell. I’m not familiar with what’s available in Tcl, but Perl has Net::SSH::Perl, which will let you do things like:

  my $ssh = Net::SSH::Perl->new("remotehost");
  $ssh->login("user", "pass");
  my $output = $ssh->cmd("doSomething");