# -*- coding: utf-8 -*-
from numpy import append, arange, array, exp, floor, nan, tensordot
from numpy.version import version as np_version
from pandas_ta._typing import DictLike, Int, IntFloat
from pandas import Series
from pandas_ta.utils import strided_window, v_offset, v_pos_default, v_series



def alma(
    close: Series, length: Int = None,
    sigma: IntFloat = None, dist_offset: IntFloat = None,
    offset: Int = None, **kwargs: DictLike
) -> Series:
    """Arnaud Legoux Moving Average

    This indicator attempts to reduce lag with Gaussian smoothing.

    Sources:
        * [prorealcode](https://www.prorealcode.com/prorealtime-indicators/alma-arnaud-legoux-moving-average/)
        * [sierrachart](https://www.sierrachart.com/index.php?page=doc/StudiesReference.php&ID=475&Name=Moving_Average_-_Arnaud_Legoux)

    Parameters:
        close (Series): ```close``` Series
        length (int): The period. Default: ```9```
        sigma (float): Smoothing value. Default ```6.0```
        dist_offset (float): Distribution offset, range ```[0, 1]```.
            Default ```0.85```
        offset (int): Post shift. Default: ```0```

    Other Parameters:
        fillna (value): ```pd.DataFrame.fillna(value)```

    Returns:
        (Series): 1 column
    """
    # Validate
    length = v_pos_default(length, 9)
    close = v_series(close, length)

    if close is None:
        return

    sigma = v_pos_default(sigma, 6.0)

    if isinstance(dist_offset, float) and 0 <= dist_offset <= 1:
        offset_ = float(dist_offset)
    else:
        offset_ = 0.85

    offset = v_offset(offset)

    # Calculate
    np_close = close.to_numpy()
    x = arange(length)
    k = floor(offset_ * (length - 1))
    weights = exp(-0.5 * ((sigma / length) * (x - k)) ** 2)
    weights /= weights.sum()

    if np_version >= "1.20.0":
        from numpy.lib.stride_tricks import sliding_window_view
        window = sliding_window_view(np_close, length)
    else:
        window = strided_window(np_close, length)
    result = append(array([nan] * (length - 1)),
                    tensordot(window, weights, axes=1))
    alma = Series(result, index=close.index)

    # Offset
    if offset != 0:
        alma = alma.shift(offset)

    # Fill
    if "fillna" in kwargs:
        alma.fillna(kwargs["fillna"], inplace=True)

    # Name and Category
    alma.name = f"ALMA_{length}_{sigma}_{offset_}"
    alma.category = "overlap"

    return alma