#!/usr/bin/env python
"""Export RGB / 255 BCHW -> Softmax probabilities [1, C], in training class order.

Training normalization is embedded. No resize, pseudo boxes or background gating.
"""
from __future__ import annotations

import argparse
import copy
import json
import sys
from pathlib import Path

import numpy as np
import torch
import yaml
from torch import nn

sys.path.insert(0, str(Path(__file__).resolve().parents[1]))
from models import build_model  # noqa: E402


def ordered_names(names) -> list[str]:
    if isinstance(names, dict):
        indexed = {int(k): v for k, v in names.items()}
        if set(indexed) != set(range(len(indexed))):
            raise ValueError("Class IDs must be contiguous from zero")
        names = [indexed[i] for i in range(len(indexed))]
    if not isinstance(names, list) or not names or not all(isinstance(n, str) for n in names):
        raise ValueError("Expected a nonempty class-name list or indexed mapping")
    if len(set(names)) != len(names):
        raise ValueError("Duplicate class names")
    return names


def input_size(config: dict) -> list[int]:
    size = config["data"]["image_size"]
    size = [size, size] if isinstance(size, int) else list(size)
    if len(size) != 2 or any(not isinstance(v, int) or v < 1 for v in size):
        raise ValueError(f"Invalid training image_size: {size}")
    return size


def preprocessing(config: dict, size: int | list[int]) -> float | None:
    shape = [size, size] if isinstance(size, int) else list(size)
    transforms = config["data"]["val_transforms"]
    sequence = [t["name"] for t in transforms]
    if sequence not in (["Resize", "ToTensor", "PerImageStandardize"],
                        ["ToTensor", "PerImageStandardize"], ["Resize", "ToTensor"], ["ToTensor"]):
        raise ValueError(f"Unsupported validation preprocessing: {sequence}")
    eps = None
    for transform in transforms:
        if transform["name"] == "Resize" and list(transform["size"]) != shape:
            raise ValueError("Training size differs from export input")
        if transform["name"] == "PerImageStandardize":
            eps = float(transform.get("eps", 1e-5))
            if eps <= 0:
                raise ValueError("Standardization eps must be positive")
    return eps


class ClassifierProbabilities(nn.Module):
    def __init__(self, classifier: nn.Module, eps: float | None = None):
        super().__init__()
        self.classifier, self.eps = classifier, eps

    def normalize(self, x: torch.Tensor) -> torch.Tensor:
        if self.eps is None:
            return x
        dims = (1, 2, 3)
        mean = x.mean(dim=dims, keepdim=True)
        # Match training torch.std(correction=1), separately for each image.
        variance = ((x - mean) ** 2).sum(dim=dims, keepdim=True) / (x[0].numel() - 1)
        return (x - mean) / (torch.sqrt(variance) + self.eps)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        return self.classifier(self.normalize(x)).softmax(dim=1)


def main() -> int:
    import openvino as ov

    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument("--checkpoint", default="runs/classify/best.pth")
    parser.add_argument("--classes", help="Training class order: classes.txt or eval_report.json")
    parser.add_argument("--fp32-weights", action="store_true", help="Keep full-precision weights")
    parser.add_argument("--out", required=True, help="New directory; existing directories are rejected")
    args = parser.parse_args()
    checkpoint, out = Path(args.checkpoint), Path(args.out)
    if out.exists():
        raise FileExistsError(f"Refusing to overwrite {out}")
    state = torch.load(checkpoint, map_location="cpu", weights_only=False)
    config = state["config"]
    names = config["data"].get("classes")
    if args.classes:
        path = Path(args.classes)
        supplied = (json.loads(path.read_text())["classes"] if path.suffix == ".json"
                    else path.read_text().splitlines())
        if names and ordered_names(supplied) != ordered_names(names):
            raise ValueError("--classes contradicts checkpoint class order")
        names = supplied
    if not names:
        report = checkpoint.parent / "eval_report.json"
        if not report.exists():
            raise ValueError("No recorded class order; provide --classes (do not guess folder order)")
        names = json.loads(report.read_text())["classes"]
    names = ordered_names(names)
    size = input_size(config)
    eps = preprocessing(config, size)
    model_config = copy.deepcopy(config["model"])
    model_config["backbone"]["pretrained"] = False
    model_config["init_from"] = None
    model_config["num_classes"] = len(names)
    model = build_model(model_config).eval()
    model.load_state_dict(state["model"], strict=True)
    wrapper = ClassifierProbabilities(model, eps).eval()
    shape = [1, model_config["backbone"].get("in_channels", 3), *size]
    torch.manual_seed(42)
    example = torch.rand(shape)
    converted = ov.convert_model(wrapper, example_input=example, input=shape)
    converted.input().get_tensor().set_names({"images"})
    converted.output().get_tensor().set_names({"probabilities"})
    converted.set_rt_info("classify", ["model_info", "task"])
    converted.set_rt_info(json.dumps(dict(enumerate(names))), ["model_info", "names"])
    out.mkdir(parents=True)
    arch = model_config["backbone"].get("arch", model_config["backbone"]["name"])
    xml = out / f"{arch}.xml"
    ov.save_model(converted, str(xml), compress_to_fp16=not args.fp32_weights)
    compiled = ov.Core().compile_model(str(xml), "CPU", {"PERFORMANCE_HINT": "LATENCY"})
    errors = []
    with torch.inference_mode():
        for sample in (example, torch.zeros_like(example), torch.ones_like(example)):
            expected = wrapper(sample).numpy()
            actual = compiled(sample.numpy())[compiled.output()]
            assert actual.shape == (1, len(names))
            assert np.isfinite(actual).all() and (actual >= 0).all() and (actual <= 1).all()
            np.testing.assert_allclose(actual.sum(axis=1), 1, atol=1e-5)
            np.testing.assert_allclose(actual, expected, atol=0.02, rtol=0.02)
            assert actual.argmax(axis=1).tolist() == expected.argmax(axis=1).tolist()
            errors.append(float(np.max(np.abs(expected - actual))))
    metadata = {"task": "classify", "names": dict(enumerate(names)), "imgsz": size,
                "channels": shape[1], "layout": "BCHW", "input_range": [0, 1],
                "output": "softmax_probabilities", "input_shape": shape,
                "output_shape": [1, len(names)], "standardize_eps": eps,
                "weight_storage": "fp32" if args.fp32_weights else "fp16"}
    (out / "metadata.yaml").write_text(yaml.safe_dump(metadata, allow_unicode=True))
    (out / "classes.txt").write_text("\n".join(names) + "\n")
    summary = {"checkpoint": str(checkpoint.resolve()), **metadata, "max_abs_errors": errors}
    (out / "export.json").write_text(json.dumps(summary, ensure_ascii=False, indent=2))
    print(json.dumps({"model": str(xml), **summary}, ensure_ascii=False, indent=2))
    return 0


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