"""Run checkpoint-backed SatlasNet inference on a test NPZ."""

import argparse
import importlib.util
from pathlib import Path
import numpy as np
import torch
import yaml

ROOT = Path(__file__).resolve().parents[1]


def main():
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument("--config", type=Path, default=ROOT / "conf/config.yaml"); parser.add_argument("--data", type=Path)
    parser.add_argument("--checkpoint", type=Path); parser.add_argument("--output-dir", type=Path)
    parser.add_argument("--device", choices=("auto", "cpu", "cuda"), default="auto"); args = parser.parse_args()
    config = yaml.safe_load(args.config.read_text()); checkpoint_path = args.checkpoint or ROOT / config["paths"]["checkpoint"]
    if not checkpoint_path.is_file(): raise FileNotFoundError(f"checkpoint not found: {checkpoint_path}")
    spec = importlib.util.spec_from_file_location("satlaspretrain", ROOT / "model/satlaspretrain.py")
    module = importlib.util.module_from_spec(spec); spec.loader.exec_module(module)
    checkpoint = torch.load(checkpoint_path, map_location="cpu", weights_only=False)
    model = module.SatlasPretrain(**config["model"]); model.load_state_dict(checkpoint["model"])
    use_cuda = torch.cuda.is_available() and args.device != "cpu"
    if args.device == "cuda" and not use_cuda: raise RuntimeError("CUDA requested but unavailable")
    device = torch.device("cuda" if use_cuda else "cpu"); model.to(device).eval()
    data_path = args.data or ROOT / config["data"]["root"] / "test.npz"; archive = np.load(data_path)
    module.validate_npz(archive, config)
    with torch.inference_mode():
        outputs = model(torch.from_numpy(archive["highres_images"]).to(device),
                        torch.from_numpy(archive["lowres_images"]).to(device),
                        torch.from_numpy(archive["valid_highres_times"]).to(device),
                        torch.from_numpy(archive["valid_lowres_times"]).to(device))
    predictions = {}
    for name, value in outputs.items():
        if name in ("segmentation", "property", "classification"): value = value.softmax(1)
        elif name != "regression": value = value.sigmoid()
        predictions[name] = value.cpu().numpy().astype(np.float32)
    output = args.output_dir or ROOT / config["paths"]["inference_dir"]; output.mkdir(parents=True, exist_ok=True)
    np.savez_compressed(output / "predictions.npz", **predictions, checkpoint=np.asarray(str(checkpoint_path)),
                        sample_ids=archive["sample_ids"],
                        source=archive["source"],
                        protocol=archive["protocol"] if "protocol" in archive else np.asarray("provided_npz"))
    print(f"inference={output / 'predictions.npz'} checkpoint={checkpoint_path}")


if __name__ == "__main__": main()