Python file to take a normal random point and select a place within a not-normal weighted set of regions

terrain_boundaries.py

import numpy as np

def weightedly_distribute_and_scale_noise_point(region_boundaries, region_weights, noise_point):
    """
        region_boundaries are n points within the region, starting with the min and ending at the max, forming n-1 buckets
        region_weights is a list of weights that should be length n-1, this will determine what % of points fall here
        noise_point should be between -1.0 and 1.0
    """
    
    normalised_noise_point = (noise_point + 1.0)/2.0
    
    weight_totals = float(sum(region_weights))
    
    min_point = region_boundaries[0]
    max_point = region_boundaries[-1]
    
    point_range = (max_point - min_point)
    
    weight_boundaries = [0.0]
    running_weight = 0.0
    for weight in region_weights:
        running_weight += weight/weight_totals
        weight_boundaries.append(running_weight)
        
    region_index = np.digitize([normalised_noise_point], weight_boundaries)[0]  # this tells you what region you are in, but not how far along
    
    
    wrmax = weight_boundaries[region_index]
    wrmin = weight_boundaries[region_index - 1]
    
    try:
        how_far_along_region = (normalised_noise_point - wrmin) / (wrmax - wrmin)
    except ZeroDivisionError:
        how_far_along_region = 0.0
        
    
    bmax = region_boundaries[region_index]
    bmin = region_boundaries[region_index - 1]
    
    output_point = bmin + (how_far_along_region * (bmax - bmin))
    
    return output_point
    
    
if __name__ == "__main__":
    
    boundaries = [-10000, -6250, -6000, -4000, -2000, 0, 1000, 2000, 10000]
    weights = [1, 2, 35, 20, 7, 30, 2, 3]
    
    for x in range(2001):
        p = (x/1000.0) - 1.0
        print("{}\t{}".format(p, weightedly_distribute_and_scale_noise_point(boundaries, weights, p)))