Write output corruption when spawning pwsh

[segevfiner]

import time
from winpty import PtyProcess

proc = PtyProcess.spawn(["pwsh"])
time.sleep(1)
print(proc.read(1024), end='')
proc.write("e")
time.sleep(1)
print(proc.read(1024), end='')
time.sleep(5)
proc.close(force=True)

(With starship prompt installed)

I'm getting the following:

PowerShell 7.2.1
Copyright (c) Microsoft Corporation.

https://aka.ms/powershell
Type 'help' to get help.

Loading personal and system profiles took 732ms.

…\qt-embedded-terminal on  master [!?] is 📦 v0.1.0 via ⬢ via 🐍 v3.10.2 (venv)
 
❯ e
Display all 131 possibilities? (y or n) _

That Display all 131 possibilities shouldn't be there, it's like someone pressed tab, but at no point in the code did I send a tab character, it seems like the string is corrupted on output, missing null terminator perhaps? Or encoding error?

[segevfiner]

Yep. Right here: https://github.com/andfoy/winpty-rs/blob/3b08c65244630fb5c1b7772b0295f317e92e98b9/src/pty/base.rs#L444-L454

encode_wide does not add a null terminator, but you are calling WideCharToMultiByte with -1 which tells it to treat it as a null terminated string. Undefined Behavior! as the Rust books like to proclaim. 😝

EDIT: But you did manually add a null terminator, so that's not the issue... time to keep digging...

EDIT 2: So actually caused by writing the nul byte through to the PTY. I will send a PR.

@andfoy

[segevfiner]

The fix was merged. Thanks @andfoy. But winpty-rs and this module still need to be released for the fix to be picked up. (Just an FYI in case anyone hits this).

P.S. This module seems to have not committed the Cargo.lock which means it will build with whatever happens to be the latest matching versions of winpty-rs and other libraries. I think this is a case where you should commit the Cargo.lock for reproducible builds (As this is a cdylib case), just remember to update it when needed.

[ccordoba12]

@andfoy is the maintainer of this project as well (as you can see in the commit history).

[segevfiner]

@andfoy is the maintainer of this project as well (as you can see in the commit history).

Yeah. I know. It's up to him when to make a new release of this.