import sys
from urllib import urlopen
from collections import defaultdict # requires python2.5
from random import choice
"""
A tool to generate a suite of lessons to learn an alphabet.

 get en2ru.matches at http://ruphus.com/svn/translit/matches/en2ru.matches
"""
#en2ru = open('en2ru.matches').read().decode('utf-8').splitlines()
#en2ru = open('easycountries-en2ru.txt').read().decode('utf-8').splitlines()
lexicon = sys.argv[1] or 'en2ru.matches'
language = sys.argv[2] or 'russian'

en2ru = open(lexicon).read().decode('utf-8').splitlines()

en2ru = [line.strip().split('\t') for line in en2ru if len(line.strip().split('\t')) == 2] 
ru = [r for e,r in en2ru]
en = [e for e,r in en2ru]

def freq(sequence):
 fq = defaultdict(int)
 for element in sequence: fq[element] += 1
 return fq

russianfq = freq(''.join(ru))

byfreq = sorted(russianfq.keys(), key=lambda letter: russianfq[letter], reverse = True)

patterns = [set(byfreq[:i]) for i in range(len(byfreq))]

levels= defaultdict(list)

for i, pattern in enumerate(patterns):  
 if i > 0:
  levels[i] = pattern

scores = {}

for word in ru:
 for i,pattern in levels.items(): 
  if set(word).issubset(pattern):
   scores[word] = (pattern, i)
   break

pages = defaultdict(list)

for word, (pattern, score) in scores.items():
 pages[score].append((pattern, word))

for k in range(len(pages)):
  for pattern, words in pages[k]:
    print ''.join(sorted(pattern))
    print ' '.join(words)

"""
for i in range(25):
  if pages[i]:
    pattern, words = pages[i]
    print ''.join(sorted(pattern)) 
    print ' '.join(sorted(words, key=len))
    print

class Page:
  def __init__(self, letters, words):
    self.letters = letters
    self.words = words
    self.template = template
    
  def 

"""