fix: prefer .ps1 over .cmd shim on Windows

[fengmk2]

Summary

Spawning a node_modules/.bin/*.cmd shim on Windows forces the OS to start cmd.exe, which intercepts Ctrl+C with a "Terminate batch job (Y/N)?" prompt and leaves the terminal in a corrupt state (backspace prints ^H, Ctrl+C prints ^C, input becomes sluggish).

When which() resolves a .cmd shim under node_modules/.bin/ and the system has pwsh.exe or powershell.exe available, the planner substitutes the sibling .ps1 path so the task graph records the script that actually runs. At spawn time, any .ps1 program is invoked via powershell.exe -File, so Ctrl+C propagates cleanly without the intermediate cmd.exe hop. If no PowerShell host is available, the substitution is skipped — the original .cmd path is preserved so the task still runs (matching pre-PR behavior).

Architecture

The logic is split across two layers so task metadata and cache keys reflect the real script, while the PowerShell host hop stays a spawn-time concern. Both layers consult the same POWERSHELL_HOST cache so plan-time and spawn-time decisions can never disagree.

Plan layer — vite_task_plan::ps1_shim

• powershell_host() -> Option<&'static Arc<AbsolutePath>> — cached via LazyLock, prefers pwsh.exe over powershell.exe, None when neither is on PATH (or not on Windows). Owned by the plan crate but exposed as a pub accessor so the spawn layer can reuse the same cache.
• prefer_ps1_sibling(AbsolutePathBuf) -> AbsolutePathBuf runs inside which() right after which::which_in resolves a binary. Gated on powershell_host().is_some(): if no host is available, the .cmd is kept untouched so we don't hand CreateProcess a .ps1 it can't execute.
• On Windows with host + sibling .ps1: substitutes node_modules/.bin/vite.cmd with node_modules/.bin/vite.ps1. SpawnCommand.program_path and SpawnFingerprint reference the .ps1 directly; cache keys stay portable (workspace-relative).
• .cmd/.bin/node_modules comparisons are case-insensitive.
• No-op on non-Windows (compile-time const fn passthrough).

Spawn layer — vite_task::session::execute::powershell

• wrap_ps1_with_powershell(&program, &args) -> Option<(program, args)>.
• If the program ends in .ps1, wraps the spawn as powershell.exe -NoProfile -NoLogo -ExecutionPolicy Bypass -File <ps1> <args>, using the same host as the plan layer.
• Option return lets the caller reuse the original Arcs on the passthrough path — no clones for non-.ps1 spawns.

Scope (what is not touched)

The plan-time substitution fires only when all of the following hold:

1. Extension is .cmd (case-insensitive), and
2. The file lives under node_modules/.bin/ (case-insensitive path-component check), and
3. A sibling .ps1 exists, and
4. pwsh.exe or powershell.exe is on PATH.

Only .cmd is considered because npm/pnpm/yarn's cmd-shim (the tool that generates the triplets) emits .cmd/.ps1/POSIX — never .bat.

Everything else is left alone:

• .exe / .com / .bat binaries
• System tools (C:\Windows\System32\where.cmd, chcp.com, etc.)
• User-installed .cmd wrappers outside node_modules/.bin/
• Any .cmd without a sibling .ps1
• Stripped-down Windows installs with no PowerShell host on PATH (fall back to .cmd)

Closes voidzero-dev/vite-plus#1176

Test plan

• cargo clippy --workspace --all-targets clean on macOS
• cargo test --workspace passes on macOS — 4 ps1_shim tests + 2 powershell tests run cross-platform via injected host
• Windows CI green
• Manual verification on Windows: reproduce with https://github.com/Curtion/report-vite-plus-1 after bumping vite-task in vite-plus; Ctrl+C during vp run dev should leave the terminal clean
• Manual verification: confirm pwsh.exe is preferred when both it and powershell.exe are on PATH
• Manual verification: on a Windows box with PATH stripped of PowerShell, confirm .cmd is still used and tasks still run (regression guard for the Cursor-flagged "no host" case)

[fengmk2]

@cursor review

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 4f14ebe. Configure here.}

[fengmk2]

before

after

[fengmk2]

@cursor review

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 3e1fc38. Configure here.}

[fengmk2]

@cursor review

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 26021ca. Configure here.}

[fengmk2]

@cursor review

[fengmk2]

@cursor review

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit be08530. Configure here.}