#! /usr/bin/env python
# -*- coding: UTF-8 -*-

import os

consonants = u"कखगघङचछजझञटठडढणतथदधनपफबभमयरलवशषसह"
vowels = u"अआइईउऊऋएऐओऔ"
joiners = u"ािीुूृेैोौ्ँंः"
othersymbols = u"ॐ"

# file and folder path
textfolder = "./nepaliarticles/" # folder in which files are kept
#resultfile = "freqency.txt" # file in which frequency is updates

filelist = os.listdir(textfolder)


# module definition
# this module calculates the freq of given str in all the files and updates the result to resultfile
def calculateAndUpdate(findString):
	count = 0
	for f in filelist:
		in_file = open(textfolder+f, "r")
		text = in_file.read()
		in_file.close()
		
		count += text.count(findString.encode('utf-8'))

	print findString, count
	
	out_file = open(resultfile, "a")
	out_file.write(findString.encode('utf-8')+"\t"+str(count)+"\n")
	out_file.close()



# calculate consonants
resultfile = "frequency-consonants.txt"
for i in range(len(consonants)):
	calculateAndUpdate(consonants[i])


# calculate vowels
resultfile = "frequency-vowels.txt"
for i in range(len(vowels)):
	calculateAndUpdate(vowels[i])

# calculate joiners
resultfile = "frequency-joiners.txt"
for i in range(len(joiners)):
	calculateAndUpdate(joiners[i])

# calculate othersymbols
resultfile = "frequency-othersymbols.txt"
for i in range(len(othersymbols)):
	calculateAndUpdate(othersymbols[i])

# calculate consonant+joiner
resultfile = "frequency-consonant-joiner.txt"
for i in range(len(consonants)):
	for j in range(len(joiners)):
		calculateAndUpdate(consonants[i]+joiners[j])


# calculate consonant+consonant
resultfile = "frequency-consonant-consonant.txt"
for i in range(len(consonants)):
	for j in range(len(consonants)):
		calculateAndUpdate(consonants[i]+consonants[j])