"""
 python program to count words
 Jim's office, Oct 4 2011


 # NOT DEBUGGED YET

"""

import string

def only_lower(word):
    """ Return copy of word with only a - z characters.
        >>> only_lower('hello.')
        'hello'
    """
    lower_only = filter(lambda x: 'a' <= x 'z', list(word))
    return string.join(lower_only, '')

def count_words(words):
    counts = {}
    for word in words:
        counts[word] = counts.get(word, 0) + 1
    return counts

def count_pairs(words):
    count2 = {}
    for i in range(len(words)-1):
        pair = (words[i], words[i+1])
        count2[pair] = count2.get(pair, 0) + 1
    return count2
    
def count_n(words, n):
	ncount = {}
	for i in range(len(words-n):
		for j in range(words[i], words[i+n]):
			group = []
			group.append(j)
		ncount[group] = ncount.get(group, 0) +1
	return ncount
			
def output():
	print "Counted N-grams in " + filename + "."
	print "-----------------------------------"
	print "| N-gram        |    Count        |"
	print "-----------------------------------"
	for key, value in counts.items():
		print key + "|" + value + "\n"
	for key, value in count2.items():
		print key + "|" + value + "\n"
	for key, value in ncount.items():
		print key + "|" + value + "\n"

def file_to_words(filename=None, file_content=None):
    """Read in a file, return a list of the words in it.
       >>> file_to_words(file_content="One two three..")
       ['one', 'two', 'three']                                        
    """
    if filename:
        file = open(filename)
        text = file.read()          # slurp all of text into one big string.
    elif file_content:
        text = file_content
    words = string.split(text)  # split words on whitespace
    print words
    words = map(lambda x: x.lower(), words)
    print words
    words = map(only_lower, words)
    print words
    return words

def main():
    # put your top level "do stuff" here


if __name__ == "__main__":
    import doctest
    doctest.testmod()
    #main()            # uncomment me once things are working