The best/intended way to interact with scryer-prolog as a subprocess.

[peschue]

I wonder what is the best way to run scryer-prolog as a subprocess, where the parent process can send instructions and parse results. I currently try this with Python.

If I use a TTY-capable library (pexpect) I get all kinds of TTY control characters.

My script (you can probably skip to read the script, the script output clearly shows my "problem") is

#!/usr/bin/env python3

import pexpect
import sys

file, query = sys.argv[1:]

p = pexpect.spawn("scryer-prolog", encoding='utf8')
p.setecho(False)

p.expect_exact("?-")

print(f"consulting '{file}'")
p.sendline(f"consult('{file}').")

out = p.readline()
print(f"read {repr(out)}")

out = p.readline()
print(f"read {repr(out)}")

p.expect_exact("?-")

print(f"query '{query}'")
p.sendline(query)
out = p.readline()
print(f"read {repr(out)}")

print(f"send CRLF")
p.send('\r\n')
out = p.readline()
print(f"read {repr(out)}")

p.expect_exact("?-")
print("got another query prompt")

and I call it as

./pycaller-pexpect.py foo.pl "foo(X)."

where foo.pl contains

foo(x).
foo(f(f)).

The idea is that the Python script consults the file and sends the query and later will process the output. The output of the above script is

consulting 'foo.pl'
read " \r\x1b[3Cconsult('foo.pl').\x1b[?2004l\r\n"
read '   true.\r\n'
query 'foo(X).'
read ' \r\x1b[3Cfoo(X).\x1b[?2004l\r\n'
send CRLF
read '   X = x.\r\n'
got another query prompt

Is there a configuration parameter to get rid of the lines with the control characters? (\x1b[...)

I also tried to use the subprocess library which does not provide a TTY to scryer-prolog. The script here (again you might skip reading the script) is

#!/usr/bin/python3 -u

import subprocess
import sys

file, query = sys.argv[1:]

p = subprocess.Popen("scryer-prolog", shell=True, encoding='utf8', stdin=subprocess.PIPE, stdout=subprocess.PIPE)

print(f"consulting file '{file}'")
p.stdin.write(f"consult('{file}').\n")
p.stdin.flush()

print("reading 1")
out = p.stdout.readline()
print(f"read '{repr(out)}'")

print("sending query '{query}'")
p.stdin.write(query + '\n')
p.stdin.flush()

print("reading 2")
out = p.stdout.readline()
print(f"read '{repr(out)}'")

The output indicates that scryer-prolog does not check if it has a TTY.

$ ./pycaller-subprocess.py foo.pl "foo(X)."
consulting file 'foo.pl'
reading 1
read '   true.\n'
sending query '{query}'
reading 2
thread 'main' panicked at 'failed to enable raw mode: IoError(Os { code: 25, kind: Other, message: "Inappropriate ioctl for device" })', src/prolog/machine/system_calls.rs:48:5
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
read '   X = x'

The first part - consulting 'foo.pl' gives a nice and clear output. But the query somehow fails.

So what is the preferred way to call scryer-prolog from another program, with or without a TTY?

[peschue]

The idea to use scryer-prolog like this I got from this issue: #225

[peschue]

Created issue for the panic: #1439