#!/usr/bin/env python
"""
count.py - Count Kalusa Words
Author: Sean B. Palmer, inamidst.com
"""

import sys, re

r_word = re.compile(r"[A-Za-z']+")
r_comment = re.compile(r'\[[^\]]+\]')

def parse(s): 
   s = r_comment.sub(' ', s)
   return [word.lower().strip("'") for word in r_word.findall(s)]

def main(): 
   corpus, bound = (sys.argv[1:] + ['100'])[:2]
   bound = int(bound)

   freq = {}
   f = open(corpus)
   for line in f: 
      line = line.strip(' \t\r\n')
      if line.count('\t') != 3: continue

      num, kal, eng, cq = line.split('\t')
      cq = float(cq)
      if cq < bound: continue

      for word in parse(kal): 
         try: freq[word] += 1
         except KeyError: 
            freq[word] = 1
   f.close()

   for word in sorted(freq.iterkeys()): 
      print word, '(%s)' % freq[word]

if __name__ == '__main__': 
   main()