words.py

#! /usr/bin/env python
# -*- coding: utf-8 -*-
# generate json data from txt file

import re

def main():
    words = []
    file = open("words.txt")
    while True:
        line = file.readline()
        if not line:
            break
        line = str.strip(line)
        if line and line[0] != '#':
            fields = re.split(' ', line)
            words.append(fields[0])
    file.close()
    for i, word in enumerate(words):
        j = i + 1
        while j < len(words):
            if word == words[j]:
                del words[j]
            j = j + 1
    print "({"
    print "  'word': ["
    for i, word in enumerate(words):
        if i == len(words) - 1:
            print "    '%s'" % (word)
        else:
            print "    '%s'," % (word)
    print "  ],"
    print "  'length': %d" % (len(words))
    print "})"

if __name__ == '__main__':
    main()