#!/usr/bin/env python
  
  import sys
  import codecs
  
  romanToScript = {}
  
  lexiconLocation = sys.argv[1]
  wordLexiconLoc = sys.argv[2]
  outLexiconLoc = sys.argv[3]
  
  # First create a map from the romanized to the script version
  lexicon = codecs.open(lexiconLocation + "callhome_arabic_lexicon_991012/ar_lex.v07", \
      encoding="iso-8859-6")
  for line in lexicon:
    lineComp = line.strip().split('\t')
    romanToScript[lineComp[0]] = lineComp[1]
  lexicon.close()
  
  # Now read the word lexicon and write out a script lexicon
  wordLexicon = open(wordLexiconLoc)
  outLexicon = codecs.open(outLexiconLoc, "w+", encoding="utf-8")
  for line in wordLexicon:
    lineComp = line.strip().split(" ")
    if lineComp[0] in romanToScript:
      lineComp[0] = romanToScript[lineComp[0]]
    outLexicon.write(" ".join(lineComp) + '
  ')
  
  wordLexicon.close()
  outLexicon.close()