4th Order Markov Chain Generator

markov4.py

import random

by_word = True
text = open("test.txt").read().replace("  ", " ")

if by_word:
    text = text.split(" ")

freq = {}

p3 = ' '
p2 = ' '
p1 = ' '
p = ' '
for c in text:
    if not p3 in freq:
        freq[p3] = {}
    if not p2 in freq[p3]:
        freq[p3][p2] = {}
    if not p1 in freq[p3][p2]:
        freq[p3][p2][p1] = {}
    if not p in freq[p3][p2][p1]:
        freq[p3][p2][p1][p] = {}
    if not c in freq[p3][p2][p1][p]:
        freq[p3][p2][p1][p][c] = 0
    freq[p3][p2][p1][p][c] += 1
    p1 = p
    p = c

#print(freq)

s = ""
c3 = random.choice(list(freq.keys()))
c2 = random.choice(list(freq[c3].keys()))
c1 = random.choice(list(freq[c3][c2].keys()))
c = random.choice(list(freq[c3][c2][c1].keys()))
for i in range(0, 2000):
    t = sum(list(freq[c3][c2][c1][c].values()))
    r = random.randint(0, t)

    for l, v in freq[c3][c2][c1][c].items():
        if r < v:
            s += l + (" " if by_word else "")
            c1 = c
            c = l
            break
        else:
            r -= v

print(s)