"""
Print the frequency of the (most frequent) words in a named text file.

Author: Rodney Topor
Last updated: 6 August 2011
"""

from re import split

frequencies = {} # word->frequency dictionary


def count(filename,frequency=False,number=0):
    """
    Print the frequency of the words in filename.
    
    Print in alphabetical order or frequency order depending on frequency.
    If printing by frequency and number > 0, print only the number
    most frequent words.
    """
    # make dictionary empty
    frequencies.clear()
    # read filename
    try:
        # add words in each line to dictionary
        for line in open(filename, "r"):
            process(line)
    except IOError: # basic error checking
        print 'IO Error in %s' % filename
        return
    if frequency:  
        # print the word frequencies by value
        pairs = [(count,word) for (word,count) in frequencies.iteritems()]
        pairs.sort()    # standard list processing function
        pairs.reverse() # ditto
        if number > 0:
            pairs = pairs[:number]
        for count, word in pairs:
            print "%s: %d" % (word, count)
    else: # order by key
        # print the word frequencies alphabetically
        pairs = frequencies.items() # list of (key,value) pairs in dictionary
        pairs.sort() 
        for word, count in pairs:
            print "%s: %d" % (word, count)


def process(line):
    """
    Add the number of occurrences of each word in line to the dictionary.
    """
    # split line into (alphanumeric) words 
    # (\w=alphanumeric character, \W=nonalphanumeric character)
    line = line.strip() # remove leading and trailing white space
    words = split('\W+', line)
    for word in words:
        if word != "":
            word = word.lower()
            # increment count for word in dictionary
            if frequencies.has_key(word):
                frequencies[word] = frequencies[word] + 1
            else:
                frequencies[word] = 1