Random sampling of non-overlapping substrings of length k

Question

Given a string of length n, how would I (pseudo)randomly sample m substrings of size k such that none of the sampled substrings overlap? Most of my sc

Answer 1

This is a recursive approach in Python. At each step, randomly select from among the remaining partitions of the string, then randomly select a substring of length k from the chosen partition. Replace this partition with the split of the partition on the substring chosen. Filter out partitions of length smaller than k, and repeat. The list of substrings returns when there are m of them, or there are no partitions left with length greater than or equal to k.

import random

def f(l, k, m, result=[]):
    if len(result) == m or len(l) == 0:
        return result
    else:
        if isinstance(l, str):
            l = [l]
        part_num = random.randint(0, len(l)-1)
        partition = l[part_num]
        start = random.randint(0, len(partition)-k)
        result.append(partition[start:start+k])
        l.remove(partition)
        l.extend([partition[:start], partition[start+k:]])
        return f([part for part in l if len(part) >= k], k, m, result)