"""Refresh the Space's vendored assets from the canonical project sources.

A Hugging Face Space deploys only the contents of ``space/`` — it has no access
to the project's ``src/`` package or ``eval/`` split. So the demo ships:

* ``vendor/mundart/{__init__,eval,baseline}.py`` — verbatim copies of the tested,
  dependency-free LID modules (no re-implementation → no divergence risk; rerun
  this script after any change to the canonical modules);
* ``data/{lid_train,probe_gsw_de}.jsonl`` — copies of the committed eval split the
  app trains the baseline on and uses for the gsw→de alignment lookup.

Run from ``projects/03-mundart/``::

    .venv/bin/python space/build_assets.py
"""

from __future__ import annotations

import shutil
from pathlib import Path

_SPACE = Path(__file__).resolve().parent
_PROJECT = _SPACE.parent

#: Canonical pure modules → vendored package (copied verbatim).
_MODULES = ("__init__.py", "eval.py", "baseline.py")
#: Committed eval files the app needs at runtime.
_DATA = ("lid_train.jsonl", "probe_gsw_de.jsonl")


def main() -> int:
    src_pkg = _PROJECT / "src" / "mundart"
    dst_pkg = _SPACE / "vendor" / "mundart"
    dst_pkg.mkdir(parents=True, exist_ok=True)
    for name in _MODULES:
        shutil.copyfile(src_pkg / name, dst_pkg / name)
        print(f"vendored mundart/{name}")

    src_eval = _PROJECT / "eval"
    dst_data = _SPACE / "data"
    dst_data.mkdir(parents=True, exist_ok=True)
    for name in _DATA:
        shutil.copyfile(src_eval / name, dst_data / name)
        print(f"copied data/{name}")
    return 0


if __name__ == "__main__":
    raise SystemExit(main())