make_data.py

import random

import twitter_grab

SCREEN_NAMES = [
                'GineokwKoenig',
                'RealNichelle',
                'TheRealNimoy',
                'WilliamShatner',
                'GeorgeTakei',
                'levarburton',
                'jonathansfrakes',
                'BrentSpiner',
                'SirPatStew',
                'wilw'
                ]

NUM_TWEETS_PER_SCREEN_NAME = 110

OUTPUT_FILENAME = '1100_tweets.txt'

def main():
    with open(OUTPUT_FILENAME, 'w') as output_file: 
        for screen_name in SCREEN_NAMES:
            query = twitter_grab.QueryBuilder()
            query.screen_name = screen_name
            query.include_entities = True
            query.count = NUM_TWEETS_PER_SCREEN_NAME
            
            tweets = query.get_tweets()
            for tweet in tweets:
                record = twitter_grab.make_record_from_tweet(tweet)
                sorted_record = sorted(record.items())
                record_string = ' '.join(
                    unicode(val).encode('ascii', 'ignore') 
                    for key, val in sorted_record)
                output_file.write(record_string + '\n')
            print 'Finished with {0}'.format(screen_name)
                
    with open(OUTPUT_FILENAME) as reopened_file:
        ordered_records = reopened_file.readlines()
        random.shuffle(ordered_records)
        randomized_records = ordered_records

    with open(OUTPUT_FILENAME, 'w') as final_write:
        for record in randomized_records:
            final_write.write(record)

if __name__ == '__main__':
    main()