fix(windows): force UTF-8 encoding for all spawned child processes

- Set UTF-8 environment variables (PYTHONUTF8, PYTHONIOENCODING, LANG,
  LC_ALL) on all child processes via process.rs helpers
- Configure PTY terminals per shell flavor: chcp 65001 for cmd.exe,
  [Console]::OutputEncoding for PowerShell, LANG=C.UTF-8 for Git Bash
- Use /K and -NoExit for interactive shells to avoid nested processes

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
xintaofei
2026-04-12 18:47:31 +08:00
parent 843cf8df19
commit 25def31a23
2 changed files with 56 additions and 3 deletions

View File

@@ -13,6 +13,7 @@ pub fn configure_std_command(command: &mut Command) -> &mut Command {
{
use std::os::windows::process::CommandExt;
command.creation_flags(CREATE_NO_WINDOW);
set_utf8_env(command);
}
command
@@ -33,11 +34,47 @@ pub fn configure_tokio_command(
#[cfg(windows)]
{
command.creation_flags(CREATE_NO_WINDOW);
set_utf8_env(command);
}
command
}
/// Hint child processes to produce UTF-8 output on Windows.
///
/// Sets environment variables recognised by common runtimes (Python, MSYS2/Git
/// Bash, .NET console apps). Not all programs honour these, but they cover the
/// most frequent sources of mojibake in practice.
#[cfg(windows)]
fn set_utf8_env<C: SetEnv>(command: &mut C) {
// Python
command.env("PYTHONUTF8", "1");
command.env("PYTHONIOENCODING", "utf-8");
// MSYS2 / Git-for-Windows / POSIX-layer tools
command.env("LANG", "C.UTF-8");
command.env("LC_ALL", "C.UTF-8");
}
/// Abstraction over the `.env()` method shared by std and tokio Command types.
#[cfg(windows)]
trait SetEnv {
fn env(&mut self, key: &str, val: &str) -> &mut Self;
}
#[cfg(windows)]
impl SetEnv for Command {
fn env(&mut self, key: &str, val: &str) -> &mut Self {
Command::env(self, key, val)
}
}
#[cfg(windows)]
impl SetEnv for tokio::process::Command {
fn env(&mut self, key: &str, val: &str) -> &mut Self {
tokio::process::Command::env(self, key, val)
}
}
#[cfg(windows)]
fn maybe_windows_cmd_shim(program: &OsStr) -> Option<OsString> {
let path = Path::new(program);

View File

@@ -92,11 +92,19 @@ fn detect_windows_shell_flavor(shell: &str) -> WindowsShellFlavor {
fn configure_shell_command(cmd: &mut CommandBuilder, shell: &str, initial_command: Option<&str>) {
#[cfg(target_os = "windows")]
{
// Force UTF-8 output for all Windows shell flavors
cmd.env("PYTHONUTF8", "1");
cmd.env("PYTHONIOENCODING", "utf-8");
match detect_windows_shell_flavor(shell) {
WindowsShellFlavor::Cmd => {
if let Some(command) = initial_command {
cmd.env("CODEG_CMD", command);
cmd.args(["/D", "/S", "/C", "%CODEG_CMD%"]);
// Set UTF-8 code page before running the actual command
cmd.args(["/D", "/S", "/C", "chcp 65001 >nul & %CODEG_CMD%"]);
} else {
// /K runs the command then stays open for interactive use
cmd.args(["/D", "/S", "/K", "chcp 65001 >nul"]);
}
}
WindowsShellFlavor::PowerShell => {
@@ -106,16 +114,24 @@ fn configure_shell_command(cmd: &mut CommandBuilder, shell: &str, initial_comman
"-NoLogo",
"-NoProfile",
"-Command",
"$ErrorActionPreference = 'Stop'; Invoke-Expression $env:CODEG_CMD",
"[Console]::OutputEncoding = [System.Text.Encoding]::UTF8; $ErrorActionPreference = 'Stop'; Invoke-Expression $env:CODEG_CMD",
]);
} else {
cmd.args(["-NoLogo", "-NoProfile"]);
// -NoExit runs the command then stays open for interactive use
cmd.args([
"-NoLogo",
"-NoProfile",
"-NoExit",
"-Command",
"[Console]::OutputEncoding = [System.Text.Encoding]::UTF8; $host.UI.RawUI.WindowTitle = 'codeg'",
]);
}
}
WindowsShellFlavor::Posix => {
cmd.env("TERM", "xterm-256color");
cmd.env("COLORTERM", "truecolor");
cmd.env("TERM_PROGRAM", "codeg");
cmd.env("LANG", "C.UTF-8");
if let Some(command) = initial_command {
cmd.env("CODEG_CMD", command);
cmd.args(["-l", "-i", "-c", "eval \"$CODEG_CMD\""]);