workspace/tests/test_external_checkpoint_verify.py

from __future__ import annotations

import subprocess
import sys
from pathlib import Path

from scripts.verify_external_checkpoint import build_manifest


def test_external_checkpoint_manifest_ready_when_required_files_exist(tmp_path: Path) -> None:
    checkpoint = tmp_path / "checkpoint"
    checkpoint.mkdir()
    (checkpoint / "config.json").write_text("{}", encoding="utf-8")
    (checkpoint / "model.safetensors").write_bytes(b"weights")

    manifest = build_manifest(
        checkpoint,
        model_family="smolvla",
        repo_id="lerobot/smolvla_base",
        revision="abc",
        required_files=["config.json", "model.safetensors"],
    )

    assert manifest["ready"] is True
    assert manifest["missing_required_files"] == []
    assert manifest["file_count"] == 2
    assert manifest["total_bytes"] == len("{}") + len(b"weights")
    model_row = next(row for row in manifest["files"] if row["path"] == "model.safetensors")
    assert len(model_row["sha256"]) == 64


def test_external_checkpoint_cli_reports_missing_required_files(tmp_path: Path) -> None:
    checkpoint = tmp_path / "checkpoint"
    checkpoint.mkdir()
    (checkpoint / "config.json").write_text("{}", encoding="utf-8")
    out = tmp_path / "manifest.json"

    result = subprocess.run(
        [
            sys.executable,
            "scripts/verify_external_checkpoint.py",
            "--checkpoint",
            str(checkpoint),
            "--out",
            str(out),
            "--model-family",
            "smolvla",
            "--required-file",
            "config.json",
            "--required-file",
            "model.safetensors",
        ],
        capture_output=True,
        text=True,
    )

    assert result.returncode == 2
    assert "model.safetensors" in result.stdout
    assert out.exists()