import numpy as np

from typing import AnyStr, List
from numpy.typing import ArrayLike

from spatz.sensors import Accelerometer, Gyroscope, IMU, CoordSystem
from spatz.transforms import Transform, GaussianNoise, DriftingBias
from spatz.simulations.data_source import DataSource
from spatz.dataset import Dataset
from spatz.logger import Logger


class WSEN_ISDS(IMU):
    pass


g = 9.81


class WSEN_ISDS_ACC(Accelerometer):
    def __init__(self, dataset: DataSource, logger: Logger, orientation=np.identity(3), offset=0, transforms: List[Transform] = []):
        super().__init__(dataset, logger, orientation, offset, transforms)

        self.__bias = DriftingBias(np.zeros(3), np.array([
            0.00113044 / g * 1000, 
            0.00108539 / g * 1000, 
            0.00127884 / g * 1000
        ]), 400)

        self.__constant_bias = np.random.normal(0, 0.81423, 3)

        self.__normal = GaussianNoise(0, np.array([
            0.0003330315865455515 / g * 1000, 
            0.00016874534484267122 / g * 1000, 
            0.0003885568325537318 / g * 100
        ]))

    def _get_name(self) -> AnyStr:
        return 'WSEN_ISDS_ACC'

    def _sensor_specific_effects(self, x: ArrayLike) -> ArrayLike:
        t = self._dataset.get_time()
        # Convert to milli-g.
        x = x / g * 1000

        # Apply noise to the true values.
        y = self.__constant_bias + self.__normal(t, self.__bias(t, x))
        noise = y - x

        # Log the chosen noise values.
        self._logger.write('acc_x_noise', noise[0], self._get_name())
        self._logger.write('acc_y_noise', noise[1], self._get_name())
        self._logger.write('acc_z_noise', noise[2], self._get_name())

        # The WSEN-ISDS accelerometer only measures acceleration between -16g and 16g. 
        y = np.clip(y, -16000, +16000)

        return y
    

class WSEN_ISDS_GYRO(Gyroscope):
    def __init__(self, dataset: Dataset, logger: Logger, orientation = np.identity(3), transforms: List[Transform] = []):
        super().__init__(dataset, logger, orientation, transforms)

        self.__bias = DriftingBias(np.zeros(3), np.array([0.00218 * 1000, 0.00105 * 1000, 0.00203 * 1000]), 400)
        self.__constant_bias = np.random.normal(0, 2*2000, 3)
        self.__normal = GaussianNoise(0, np.array([0.0049272 * 1000, 0.00557833 * 1000, 0.00407826 * 1000]))

    def _get_name(self) -> AnyStr:
        return 'WSEN_ISDS_GYRO'

    def _sensor_specific_effects(self, x: ArrayLike) -> ArrayLike:
        # Convert to milli-degrees per second.
        x = (x / np.pi) * 180 * 1000

        t = self._dataset.get_time()
        x = self.__constant_bias + self.__normal(t, self.__bias(t, x))

        # TODO: Noise model.

        self._log('ox', x[0])
        self._log('oy', x[1])
        self._log('oz', x[2])

        return x