ngram.py

#!/usr/bin/env python3

from nltk.corpus import brown

PUNCTUATION = "!\"#$%'()*+,-./:;<=>?@[\\]^_`{|}~"
MIN_OCCURANCES = 20
NGRAM_LENGTH = 3

# Hold a count of words
countdict = {}

# Results in form (count, word 1, word 2, word 3)
results = list()

words = ("",) * NGRAM_LENGTH

# Itterate through every word
for word in brown.words():
    words = words[1:] + (word.lower(),)
    if words in countdict:
        countdict[words] = countdict[words] + 1
    else:
        countdict[words] = 1

# Itterate through results filtering out punctuation and counts smaller than MIN_OCCURANCES
for ngram in countdict:
    if countdict[ngram] > MIN_OCCURANCES:
        if all(word not in PUNCTUATION for word in ngram):
            results.append((countdict[ngram], ngram))

# Print results sorted by number of occurances
for result in sorted(results, key=lambda result: result[0]):
    print(result)