ewoks.py

from datetime import datetime
import os
import pickle
import sys

import markovify
import tweepy


CONFIG_KEYS = {
    'aniruel': ('', '', '', ''),
    'manolon': ('', '', '', ''),
    'neus': ('', '', '', ''),
    'toalla': ('', '', '', ''),
}

if len(sys.argv) != 3 or sys.argv[1] not in CONFIG_KEYS:
    print('Usage: %s <twitter> <corpus>' % sys.argv[0])
    print('Available Twitter tokens: ' + ', '.join(list(CONFIG_KEYS)))
    sys.exit(1)

_, config_key, corpus_filename = sys.argv

consumer_key, consumer_secret, access_token, access_token_secret = \
    CONFIG_KEYS[config_key]

model_filename = corpus_filename + '.pickle'
if os.path.exists(model_filename):
    with open(model_filename, 'rb') as fp:
        unpickled_model = pickle.load(fp)

    text_model = markovify.Text.from_dict(unpickled_model)
else:
    with open(corpus_filename) as fp:
        text = fp.read()

    if text.count('\n') > 2:
        text_model = markovify.NewlineText(text, state_size=3)
    else:
        text_model = markovify.Text(text, state_size=3)

    with open(model_filename, 'wb') as fp:
        pickle.dump(text_model.to_dict(), fp)

auth = tweepy.OAuthHandler(consumer_key, consumer_secret)
auth.set_access_token(access_token, access_token_secret)
api = tweepy.API(auth)

while True:
    sentence = text_model.make_short_sentence(140, tries=10000)
    if sentence is None:
        continue

    try:
        api.update_status(sentence)
    except tweepy.error.TweepError as e:
        if e.api_code == 187:  # status is a duplicate
            continue
        else:  # any other error
            raise
    else:
        break

print('[%s] %s -> %s' % (datetime.now(), config_key, sentence))