Major - P3 It should support recursive generation of arrays e.g. generates arrays of arrays. It should leverage existing infrastructure of random generators in Cost Model project defined in random_generator.py.
Possible design:
class ArrayRandomDistribution (RandomDistribution): def __init__(self, lengths_distr: RandomDistribution, value_distr: RandomDistribution): """ArrayRandomDistribution inherits RandomDistribution, therefore, it allows recursive arrays generation since value_distr can be ArrayRandomDistribution or DocumentRandomDistribution""" self.lengths_distr = lengths_distr self.value_distr = value_distr def generate(self, size: int): arrays = [] lengths = self.lengths_distr.generate(size) for length in lengths: values = self.value_distr.generate(length) arrays.append(values) return arrays
Example of generating arrays of arrays of strings:
internal_length_distribution = RandomDistribution.normal(RangeGenerator(DataType.INT, 10, 25)) internal_values_distribution = RandomDistribution.uniform(RangeGenerator(DataType.STRING, "abc", "zzzz")) internal_array_distribution = ArrayRandomDistribution(internal_length_distribution, internal_values_distribution) length_distribution = RandomDistribution.choice(values=[10, 20], weights=[50, 50]) array_distribution = RandomDistribution.array(length_distribution , internal_array_distribution)