BioinstLab/gmass-demo

1"""Prepare a Hugging Face Space deployment bundle for the Gradio app."""2 3from __future__ import annotations4 5import argparse6import os7import shutil8import stat9from pathlib import Path10 11 12ROOT = Path(__file__).resolve().parents[1]13APP_DIR = ROOT / "app"14DEFAULT_OUTPUT_DIR = ROOT / "dist" / "hf_space"15COMBINED_RESULTS = ROOT / "data" / "eval_outputs" / "combined" / "all_models_scored.jsonl"16SOURCE_DIRS = ("configs", "core", "models", "probes", "scorer", "scripts", "translation")17SOURCE_FILES = ("run_bilingual_eval.py",)18 19 20def copy_file(src: Path, dst: Path) -> None:21    dst.parent.mkdir(parents=True, exist_ok=True)22    shutil.copy2(src, dst)23    print(f"copied {src.relative_to(ROOT)} -> {dst.relative_to(ROOT)}")24 25 26def copy_tree(src: Path, dst: Path) -> None:27    if dst.exists():28        remove_tree(dst)29    shutil.copytree(30        src,31        dst,32        ignore=shutil.ignore_patterns(33            "__pycache__",34            "*.pyc",35            "*.pyo",36            "logs",37            "*.log",38            "run_codespaces_app.sh",39        ),40    )41    print(f"copied {src.relative_to(ROOT)} -> {dst.relative_to(ROOT)}")42 43 44def handle_remove_readonly(func, path, _exc_info) -> None:45    os.chmod(path, stat.S_IWRITE)46    func(path)47 48 49def remove_tree(path: Path) -> None:50    shutil.rmtree(path, onerror=handle_remove_readonly)51 52 53def prepare_space_bundle(output_dir: Path, include_results: bool = False) -> Path:54    if output_dir.exists():55        remove_tree(output_dir)56    output_dir.mkdir(parents=True)57 58    copy_file(APP_DIR / "app.py", output_dir / "app.py")59    copy_file(APP_DIR / "gmass_app.py", output_dir / "gmass_app.py")60    copy_file(APP_DIR / "spaces_README.md", output_dir / "README.md")61    copy_file(APP_DIR / "spaces_requirements.txt", output_dir / "requirements.txt")62    if (ROOT / "LICENSE").exists():63        copy_file(ROOT / "LICENSE", output_dir / "LICENSE")64    for source_dir in SOURCE_DIRS:65        copy_tree(ROOT / source_dir, output_dir / source_dir)66    for source_file in SOURCE_FILES:67        copy_file(ROOT / source_file, output_dir / source_file)68    (output_dir / ".gitignore").write_text(69        "\n".join(70            [71                "__pycache__/",72                "*.py[cod]",73                ".env",74                ".env.*",75                "logs/",76                "*.log",77                "",78            ]79        ),80        encoding="utf-8",81    )82    print(f"wrote {output_dir.relative_to(ROOT) / '.gitignore'}")83 84    PUBLIC_METRICS = ROOT / "data" / "public_metrics" / "benchmark_summary.json"85    if PUBLIC_METRICS.exists():86        copy_file(PUBLIC_METRICS, output_dir / PUBLIC_METRICS.relative_to(ROOT))87 88    if include_results:89        if not COMBINED_RESULTS.exists():90            raise FileNotFoundError(91                f"Cannot include benchmark results; missing {COMBINED_RESULTS.relative_to(ROOT)}"92            )93        copy_file(COMBINED_RESULTS, output_dir / COMBINED_RESULTS.relative_to(ROOT))94 95    print("")96    print(f"Space bundle ready at: {output_dir}")97    print("Next: copy or push that directory to your Hugging Face Space repository.")98    return output_dir99 100 101def main() -> None:102    parser = argparse.ArgumentParser(description="Prepare the Hugging Face Space bundle.")103    parser.add_argument(104        "--output-dir",105        default=str(DEFAULT_OUTPUT_DIR),106        help="Output directory for the generated Space bundle.",107    )108    parser.add_argument(109        "--include-results",110        action="store_true",111        help="Include precomputed combined benchmark results if available.",112    )113    args = parser.parse_args()114 115    prepare_space_bundle(Path(args.output_dir).resolve(), include_results=args.include_results)116 117 118if __name__ == "__main__":119    main()120