"""
Threshold Network for 3-input XNOR Gate

XNOR3(a,b,c) = 1 when even number of inputs are 1 (0 or 2)
Built as: XOR(XNOR(a,b), c)
"""

import torch
from safetensors.torch import load_file


class ThresholdXNOR3:
    def __init__(self, weights_dict):
        self.w = weights_dict

    def __call__(self, a, b, c):
        # First XNOR: a XNOR b (1 when a=b)
        inp1 = torch.tensor([float(a), float(b)])
        n1 = int((inp1 * self.w['xnor1.layer1.n1.weight']).sum() + self.w['xnor1.layer1.n1.bias'] >= 0)
        n2 = int((inp1 * self.w['xnor1.layer1.n2.weight']).sum() + self.w['xnor1.layer1.n2.bias'] >= 0)

        h1 = torch.tensor([float(n1), float(n2)])
        xnor_ab = int((h1 * self.w['xnor1.layer2.weight']).sum() + self.w['xnor1.layer2.bias'] >= 0)

        # Second XOR: xnor_ab XOR c
        inp2 = torch.tensor([float(xnor_ab), float(c)])
        n3 = int((inp2 * self.w['xor2.layer1.n1.weight']).sum() + self.w['xor2.layer1.n1.bias'] >= 0)
        n4 = int((inp2 * self.w['xor2.layer1.n2.weight']).sum() + self.w['xor2.layer1.n2.bias'] >= 0)

        h2 = torch.tensor([float(n3), float(n4)])
        out = int((h2 * self.w['xor2.layer2.weight']).sum() + self.w['xor2.layer2.bias'] >= 0)

        return float(out)

    @classmethod
    def from_safetensors(cls, path="model.safetensors"):
        return cls(load_file(path))


if __name__ == "__main__":
    weights = load_file("model.safetensors")
    model = ThresholdXNOR3(weights)

    print("3-input XNOR Gate Truth Table:")
    print("-" * 30)
    correct = 0
    for a in [0, 1]:
        for b in [0, 1]:
            for c in [0, 1]:
                out = int(model(a, b, c))
                # XNOR3 = even parity = NOT XOR3
                expected = 1 - (a ^ b ^ c)
                status = "OK" if out == expected else "FAIL"
                if out == expected:
                    correct += 1
                print(f"XNOR3({a}, {b}, {c}) = {out}  [{status}]")
    print(f"\nTotal: {correct}/8 correct")