words_frequency_count.py

#!/usr/bin/env python
import sys
import logging


logger = logging.getLogger('wordsfrequency')


class WrongInputError(Exception):
    pass


def count_words_frequency(text):
    """ Counts words frequency in text.
    Returns tuple, structure is illustrated by example - ((word, frequency), <...>).
    Tuple is ordered by frequency in desc order.
    Word with equal frequency ordered by alphabet.

    Arguments:
    text -- string
    """
    words = text.split()
    stats = dict()
    for word in words:
        word = word.lower()
        if word not in stats:
            stats[word] = 0
        stats[word] += 1
    # prepare stats for ordering
    output = list(stats.items())
    # python sort is stable, so we need to sort stats by key in alphabetical order
    output.sort(key=lambda item: item[0])
    # and then sort it by frequency (elements with equal frequency
    # will not be changed, because sort is stable)
    output.sort(key=lambda item: item[1], reverse=True)
    return tuple(output)


if __name__ == "__main__":
    file_data = None
    try:
        if(len(sys.argv) > 1):
            source = sys.argv[1]
            file_data = open(source).read()
        else:
            raise WrongInputError("Cant find file path in script arguments")
    except IOError:
        logger.error("""Error: cant read the file.\n\nAborting.""")
        sys.exit()
    except WrongInputError:
        logger.error("""Error: Please, provide real path to the file as first argument.\n\n"""\
                     """For example: python words_frequency_count.py ./source.txt\n\n"""\
                     """Aborting.""")
        sys.exit()

    if file_data:
        buf = []
        stats = count_words_frequency(file_data)
        for row in stats:
            buf.append('{}:{}'.format(*row))
        print("\n".join(buf))