generate

import collections
import pickle
import random
import argparse

parser = argparse.ArgumentParser()
parser.add_argument('way_to_model', '--model ', type=str,
                    help='The way to directory with model')
parser.add_argument('way_to_output_file', "--output", type=str,
                    help='The way to output file')
parser.add_argument('first_word', '--seed', type=str,
                    help='The starting word')
parser.add_argument('length_of_text', '--length', type=int,
                    help='The length of the generated text')
parser.add_argument('Help', '--help', action=store_true,
                    help='Show the utility reference')
args = parser.parse_args()


def frequency_choice(frequencies):
    total = 0
    for key in frequencies:
        total += frequencies.get(key)
    r = random.uniform(0, total)
    up_to = 0
    for key in frequencies:
        if up_to + frequencies.get(key) >= r:
            return key
        up_to += frequencies.get(key)


def generate_text(way_to_model, first_word, length_of_text,
                  way_to_output_file):
    dictionary = collections.OrderedDict()
    if (first_word is None):
        list_of_keys = dictionary.keys()
        num = random.randint(0, len(list_of_keys))
        first_word = list_of_keys[num]
    with open('{}/model.pkl'.format(way_to_output_file), 'rb') as model:
        dictionary = pickle.load(model)
    num_of_words = length_of_text
    new_word = first_word
    space = " "
    # next_words_list = collections.OrderedDict()
    with open('{}/generated_text.txt'.format(way_to_output_file),
              'w', encoding="utf-8") as f:
        for i in range(num_of_words):
            next_words_dict = dictionary.get(new_word)
            f.write(str(new_word) + space)
            new_word = None
            while (new_word is None):
                new_word = frequency_choice(next_words_dict)
            next_words_dict.clear()


if (args.Help):
    print('''Help :
    It is the utility for generate text on the basis of model.
    You should specify the directory with model file :
    --model <dir_name> .
    Also you can specify a directory in which it
    is required to keep generated text or use the stdout:
    --model <dir_name> . Stdout use by default.
    If you want to choice the seed-word, you should use --seed:
    --seed <your_word>
    else it will be chosen accidentally.
    And you should input the lenght of generated text:
    --lengs <natural_num>.
    SUDDENLY use --help to see this text.''')
if (args.way_to_output_file is None):
    args.way_to_output_file = 'stdout_output.txt'
generate_text(args.way_to_model, args.first_word,
              args.length_of_text, args.way_to_output_file)
if (args.way_to_output_file is None):
    with open('{}/generated_text.txt'.format(args.way_to_output_file),
              'r', encoding="utf-8") as f:
        for line in f:
            print(line)