A tiny chatbot inspired by https://codegolf.stackexchange.com/questions/20914/who-is-this-chatbot?noredirect=1&lq=1

chatbot.py

from collections import defaultdict
from random import choices, choice
import json
import re

word_map = defaultdict(lambda: defaultdict(int))
start_map = defaultdict(lambda: defaultdict(int))


def save(filename):
    with open(filename, 'w') as file:
        json.dump({'word_map': word_map, 'start_map': start_map}, file,
                  indent=4)


def load(filename):
    with open(filename, 'r') as file:
        maps = json.load(file)
    for first, map in maps['word_map'].items():
        for second, count in map.items():
            word_map[first][second] += count
    for first, map in maps['start_map'].items():
        for second, count in map.items():
            start_map[first][second] += count


def train_set(filename):
    with open(filename) as training_set:
        last_word = None
        for line in training_set.readlines():
            line = re.findall(r"[\w']+|[.,!?;]", line)
            if not line:
                continue
            process(last_word, line)
            last_word = line[-1]


def process(my_last_word, response):
    response = response + [None]
    word = response[0]
    start_map[my_last_word][word] += 1
    for follow in response[1:]:
        word_map[word][follow] += 1
        word = follow


def respond(their_last_word):
    if their_last_word in start_map:
        response_set = start_map[their_last_word]
    else:
        response_set = choice(list(word_map.values()))
    response = []
    while True:
        word = choices(list(response_set.keys()),
                       list(response_set.values()))[0]
        if word is None:
            return response
        response.append(word)
        response_set = word_map[word]


def merge(tokens):
    if not tokens:
        return ''
    output = tokens[0]
    for token in tokens[1:]:
        if token in '.,!?;':
            output += token
        else:
            output += ' ' + token
    return output


if __name__ == "__main__":
    my_last = None
    print('To save: >>filename')
    print('To load: <<filename')
    print('To train from text file: !!filename')
    while True:
        raw = input('HUMAN: ').lstrip()

        if raw.startswith('>>'):
            save(raw[2:])
            continue
        elif raw.startswith('<<'):
            load(raw[2:])
            continue
        elif raw.startswith('!!'):
            train_set(raw[2:])
            continue

        human_says = re.findall(r"[\w']+|[.,!?;]", raw)
        process(my_last, human_says)

        their_last = None
        if human_says:
            their_last = human_says[-1]

        response = respond(their_last)

        my_last = None
        if response:
            my_last = response[-1]

        print('MACHINE: ', merge(response))