Interacting with a non-terminating program expecting input

jgaz · n00b Joined: 14 Feb 2021 Posts: 49

Okay, this is a weird one I admit...

I got the itch to mess around with BASIC games that are older than I am! Why? I don't know. A lot of these are fairly simple simulation games with some randomized element to them. They're purely text-based, give or take some ASCII art. Besides their simulationist nature, the other thing they have in common is a game loop that requires human input. If you grew up in the US in the 1980's, think Lemonade Stand on the Apple II. If you grew up across the pond during the same period, I think it was Granny's Garden on the Beeb for you.

My goal here is for me to try to map out optimal choices for all sets of conditions within certain BASIC games. I think I'll start with Hamurabi as it's actually kind of interesting. The major variables are population count, acres of land, bushels of grain, and if some disaster was triggered during the previous round. No, this isn't a homework assignment... this is 100% ADHD-driven curiosity, to be perfectly honest. So, normally I'd just use grep, awk or some Perl to extract values and dump that to CSV for further analysis. The problem is, I can't really leverage them with a program that is by design in a loop awaiting input forever.

I very dimly recall dev-tcltk/expect was written to solve this problem. I think I knew some BBS guys who used it to manage their modem pool back in the day. I've read the documentation and I follow a lot of it, but not all of it. I get how I can look for a string, such as a password prompt, and insert the anticipated value. I don't understand how to extract some value and throw that into a CSV, and then make a decision. Also, I don't know TCL so it's probably not an optimal choice. I suspect there is probably now a BASH-y or Perl-ish way to solve this same set of problems. At the risk of sounding dim-witted, I'm not connecting the dots. Can someone please point me in the right direction(s)?

For technical context, I'm compiling these to amd64 executables with guru::dev-lang/fbc. Some compile without issues, a few need a small patch to compile in FreeBASIC's QuickBASIC emulation mode. So, what's the fix? If I can't just use a pipe, how do I capture output? I assume I'll have to do some virtual terminal tricks here for simulated human interactions, what's the best way to do that? I get that once I get this working, I'll have to patch the BASIC code further to allow passing forced combinations of values at runtime for efficiency. That part, at least, I understand.

halcon · l33t Joined: 15 Dec 2019 Posts: 686

Hello jqaz,

sublogic · Posted: Sun Jul 06, 2025 12:35 am Post subject:

For what it's worth, there is also a dev-python/pexpect . https://pexpect.readthedocs.io/en/stable/. I have no experience with it.
_________________
The practical unit of "Learning Experience" is the milli-Gentoo.

Genone · Posted: Sun Jul 06, 2025 9:26 am Post subject:

Key question for this is if those games (and/or the fbc compiler) use the standard UNIX IO interface (stdout, stderr and stdin) or less abstract interfaces like termio. Regular IO manipulation tools only support the first case, the second case is more tricky as most likely you'll have to simulate individual keypresses.

If they use the standard IO interfaces and the issue is only the interactive nature I'd look into using something like pythons POpen interface or fifo nodes (aka named pipes) and redirect stdin, e.g.