source: src/compdic_utf8/text2fst.py @ f4bf33e

#!/usr/bin/env python
# -*- coding: utf-8 -*-
"""-----------------------------------------------------------------------------
# Name:         text2FST     
# Purpose:      A tool for translating a dictionary file into a OpenFST format.
#
# Author:      Krzysztof Szarzynski <szarznyski.wmi.amu.edu.pl>
#
# Created:     19/11/2012
# Copyright:   (c) UAM Text Tools 2012
# Licence:     Simplified BSD License
# Usage:
#       cat dictionary.dic | ./text2fst > output.fst
# Warning: the dictionary.dic file must be UTF8 _without_ BOM
# 
# TODO: Checking the BOM and removing it from the dictionary.file
-----------------------------------------------------------------------------"""

import sys
import locale

encoding = locale.getdefaultlocale()[1]

def prn(str):
    sys.stdout.write(str.encode(encoding))

begState = 0
endState = 1
eps = u"<eps>"
currentState = begState
for line in sys.stdin:
    prn(u"%s %s %s\n"%(begState, currentState, eps))
    line = line.decode('UTF-8')
    for letter in line:
        prn(u"%s %s %s\n"%(currentState, currentState+1, letter))
        currentState+=1
    prn(u"%s %s %s\n"%(currentState, endState, eps))
print endState