Source code for storx.hash_m

"""Contains functionality to calculate hash values."""
import hashlib
import random
import numpy as np
import xarray as xr



def hash_array(array: xr.DataArray, nsample=3):
    """
    Return a hash for the given array created.

    The algorithm converts randomly selected array elements to the string
    representation of the underlying bytedata which is used in a sha512
    algorithm.
    """
    seed = 5772156649015328606

    random.seed(a=seed, version=2)
    # Select nsample random flat indices within the array size
    selected_inds = [random.randint(0, array.size) for i in range(nsample)]

    # The corresponding value in the multidimensional array for each selected
    # index
    selected_vals = np.zeros(nsample)
    for j in range(nsample):
        # Unravel flat index to multidimensional index
        unraveled_inds = np.array(np.unravel_index(selected_inds[j],
                                                   array.shape))
        # Create dictionary that can by used by xarray index selection
        attrs = {array.dims[i]:unraveled_inds[i] \
                    for i in range(len(array.dims))}
        selected_vals[j] = array.isel(attrs)

    hasher = hashlib.new("sha512")
    hasher.update(np.ndarray.tobytes(selected_vals))

    return hasher.hexdigest()