# Uses a dictionary to implement a word count,
# so that the user can see which words
# occur the most in the book Moby-Dick.

# minimum number of occurrences needed to be printed
OCCURRENCES = 2000

def intro():
    print("This program displays the most")
    print("frequently occurring words from")
    print("the book Moby Dick.")
    print()

# Reads book and returns a dictionary of (word, count) pairs.
def count_all_words(file):
    # read the book into a dictionary
    word_counts = {}
    for word in file.read().lower().split():
        word = word
        if word in word_counts:
            word_counts[word] += 1   # seen before
        else:
            word_counts[word] = 1    # never seen before
    return word_counts

# Returns a list of (count, word) tuples of all words in the
# given (word, count) dictionary that occur at least 2000 times.
def most_common_words(word_counts):
    most_common = []
    for (word, count) in word_counts.items():
        if count > OCCURRENCES:
            most_common.append((count, word))
    return most_common

# Displays top words that occur most in the dictionary.
def print_results(word_counts):
    wordlist = most_common_words(word_counts)
    wordlist.sort(reverse = True)
    for count, word in wordlist:
        print(word, "occurs", count, "times.")

def main():
    intro()
    with open("mobydick.txt") as file:
        word_counts = count_all_words(file)
        print_results(word_counts)

main()