Data format converter for Chris' fast_align: https://github.com/clab/fast_align

fast_align_data_formatter.py

#!/usr/bin/env python

"""
Data format converter for Chris' fast_align: https://github.com/clab/fast_align
For the input format, please see the README of his code.
"""

import itertools
import sys

def main():
    if len(sys.argv) != 3:
        print 'Usage: %s fr en' % (sys.argv[0])
        sys.exit()
    fr = sys.argv[1]
    en = sys.argv[2]

    # TODO: error check for the input data.
    for i, lines in enumerate(itertools.izip(*[file(f) for f in [fr, en]]), 1):
        print '%s ||| %s' % (lines[0].rstrip(), lines[1].rstrip())

if __name__ == '__main__':
    main()