I want to be able to use Rust to spawn a child shell, then repeatedly pass it arbitrary commands and process their outputs. I have found plenty of examples online showing me
read_to_string is documented as
Read all bytes until EOF in this source
Thus, it is waiting until all the input is done, which will never happen until the shell is closed. You can fix this by reading a set amount of data from the output. Here's an example where I removed all the nice error printing you had to show the core of the solution:
use std::process::{Command, Stdio};
use std::io::{BufRead, Write, BufReader};
fn main() {
let mut child_shell = Command::new("/bin/bash")
.stdin(Stdio::piped())
.stdout(Stdio::piped())
.spawn()
.unwrap();
let child_in = child_shell.stdin.as_mut().unwrap();
let mut child_out = BufReader::new(child_shell.stdout.as_mut().unwrap());
let mut line = String::new();
loop {
child_in.write("ls\n".as_bytes()).unwrap();
child_out.read_line(&mut line).unwrap();
println!("{}", line);
}
}
Here, we use the BufRead trait to allow reading from the input until we have read one line worth. We then print that out and continue on our loop. Of course, there's more than one line of output per line of input, so this will just have more and more waiting to be read.
In your real code, you will need to have to figure out when to stop reading. This could be really easy if you have fixed-size responses, or really hard if you are trying to deal with a human-interactive program.
Be careful about using child_shell.stdin
or stdout
directly, without an Option::as_ref
, Option::as_mut
or Option::take
. Using stdin
or stdout
directly will move that item out of the Child
structure, leaving the Child
partially valid. You would no longer be able to call wait
or kill
on it for example.
On an unrelated note, you don't need to call trait methods like Error::description(&why)
. You can just say why.description()
.