@emdzej/inpax-compiler

CLI for compiling INPA .IPS source files to .IPO bytecode. Thin wrapper around the @emdzej/inpax-compiler-core library — lexer / parser / semantic / codegen / writer pipeline.

Install / build

pnpm --filter @emdzej/inpax-compiler build

After build the binary is at dist/index.js and registered via bin so pnpm exec inpax-compile works inside the workspace. Or invoke the root-level shortcut from anywhere:

pnpm compile <input.ips> [options]
pnpm compile:debug <input.ips>     # node --inspect-brk for VS Code attach

Usage

inpax-compile myscript.ips                  # → myscript.ipo
inpax-compile myscript.ips -o build/        # → build/myscript.ipo
inpax-compile *.ips                         # batch mode
inpax-compile myscript.ips --encoding cp1252

The --encoding cp1252 flag is the right default for legacy BMW INPA sources written on Windows; UTF-8 is the default when the flag is omitted.

Subcommands

• inpax-compile <file> — compile one or more .ips files.
• inpax-compile new <name> — scaffold a starter .ips with inpainit / inpaexit / a single screen so a new script compiles cleanly.

CI / validation

The exit code is non-zero on parse / semantic / codegen errors. Pipe stderr to a logfile and use the standard exit code in CI:

inpax-compile MY_SCRIPT.IPS 2> compile-errors.log || exit 1

See also

• @emdzej/inpax-compiler-core — the underlying pipeline that does the actual work.
• AGENTS.md covers the bytecode format, INPACOMP.EXE compatibility notes, and the language reference.