#!/usr/bin/env python

# word frequency in a text
# tested with Python24    vegaseat    25aug2005



def printf(format, *args):
	"""Python version of printf() in C Language"""
	print format % args, 


def tokenizeString(mySentence):
	"""creates a list of words separated at whitespaces"""
	wordList = mySentence.split()
	return wordList
	
	
def cleanString(myWordList, myCharacter2Remove):
	newWordList = []
	for word1 in myWordList:
    		word2 = removeCharacter(word1, myCharacter2Remove)
#		newWordList.append(word2.lower())
		newWordList.append(word2)
	return newWordList


def changeToLower(myWordList):
	newWordList = []
	for word in myWordList:
		newWordList.append(word.lower())
	return newWordList

	
def changeToUpper(myWordList):
	newWordList = []
	for word in myWordList:
		newWordList.append(word.upper())
	return newWordList
	
	
def removeCharacter(myWord, myCharacter2Remove):
	"""removes the specified characters in each word"""
#	newWord = ""
#	for pos in range(len(myWord)):
#		myChar = myWord[pos]
#		if myChar in myCharacter2Remove:
#			pass
#		else:
#			newWord += myChar
#	return newWord
	for pos in range(len(myCharacter2Remove)):
		myWord = myWord.replace(myCharacter2Remove[pos], '')		
	return myWord
	

def countFrequency(myList):
	"""creates a wordfrequency dictionary."""
	myDictionary = {}
	for myWord in myList:
		if len(myWord) != 0:
    			myDictionary[myWord] = myDictionary.get(myWord, 0) + 1
	return myDictionary


	
def getSortedKey(myDictionary):
	"""creates a list of keys and sorts the list"""
	keyList = myDictionary.keys()
	keyList.sort()
	return keyList

	

def printDictionary(myKeyList, myDictionary):
 	printf("\n\nFrequency of each word in the sentence (sorted):\n")
#	for key in myKeyList:
#  		printf("%-10s %d", key, myDictionary[key])
	printf("{")
	for key in myKeyList:
  		printf("'%s':%d, ", key, myDictionary[key])
	printf("}")
		
def printByLength(myKeyList, myDictionary, myLengthLimit):
 	printf("\n\nFrequency of each word (whose length >= %d) in the sentence (sorted):\n", myLengthLimit)
#	for key in myKeyList:
#  		printf("%-10s %d", key, myDictionary[key])
	printf("{")
	for key in myKeyList:
		if len(key) >= myLengthLimit:
  			printf("'%s':%d, ", key, myDictionary[key])
	printf("}")

def printByFrequency(myKeyList, myDictionary, myFrequencyLimit):
 	printf("\n\nFrequency of each word (whose frequency >= %d) in the sentence (sorted):\n", myFrequencyLimit)
#	for key in myKeyList:
#  		printf("%-10s %d", key, myDictionary[key])
	printf("{")
	for key in myKeyList:
		myFrequency = myDictionary[key]
		if myFrequency >= myFrequencyLimit:
  			printf("'%s':%d, ", key, myFrequency)
	printf("}")
		
#===================================

if __name__ == '__main__':
	myCharacter2Remove = [",", ".", "!", "?", ";", "'", '"']
	mySentence = """In this project, we implement a python script that computes frequency of 
each string in a given sentence. We will exclude some specific characters 
such as ",", ".", "!", "?", ";", "'", and '"' in a string. 
We assume that a sentence is already given in the script that you are implementing"""

	printf("\nSentence = ")
	printf('"%s"', mySentence)
	
	myWordList = tokenizeString(mySentence)
	
	myWordList = cleanString(myWordList, myCharacter2Remove)

	myWordList = changeToLower(myWordList)
	
	myDictionary = countFrequency(myWordList)
	
	myKeyList = getSortedKey(myDictionary)
	
	printDictionary(myKeyList, myDictionary)

	printByFrequency(myKeyList, myDictionary, 2)

	printByLength(myKeyList, myDictionary, 10)


	myWordList = changeToUpper(myWordList)
	
	myDictionary = countFrequency(myWordList)
	
	myKeyList = getSortedKey(myDictionary)
	
	printDictionary(myKeyList, myDictionary)

	printByFrequency(myKeyList, myDictionary, 2)

	printByLength(myKeyList, myDictionary, 10)


	printf("\n\n")