chatsimulator.py

import re
from functools import reduce
from collections import defaultdict
from random import randint

usr_dict = defaultdict(list)
data = []

# Parse chat logs
for filename in ['94.html', '61.html', '36.html']:
    file_data = re.findall(r'<span class="user">(.*?)</span>.*?<p>(.*?)<\/p>',
                           reduce(str.__add__,
                                  reversed(open(filename).readlines())))
    data += list(reduce(list.__add__, map(lambda x: list(map(lambda y: (x[0], y), x[1].split())), file_data)))

# Generate markov chain
last_msg = data[0]
for entry in data[1:]:
    usr_dict[last_msg].append(entry)
    last_msg = entry

last_sender = None
last_word = None
while True:
    if last_word is None:
        keys = list(usr_dict.keys())
        last_word = keys[randint(0, len(keys))]

    # Draw next word from pool
    msg_pool = usr_dict[last_word]
    last_word = msg_pool[randint(0, len(msg_pool)-1)]
    
    # Check whether sender has changed
    if last_word[0] != last_sender:
        # Print sender name
        print('\n\n'+last_word[0])
        last_sender = last_word[0]
  
    print(last_word[1] + ' ', end='')