python中文近义词工具_python根据词条生成同义词库

1.[代码][Python]代码

#!/usr/bin/python

# -*- coding: utf-8 -*-

#from __future__ import unicode_literals

import codecs

import sys

import re

def enmu_all_keys(key):

kl = []

if len(key)==0:

return kl

prefix = ''

for c in key:

prefix = prefix + c

if len(prefix) == 1:

continue

kl.append(prefix)

kl2 = enmu_all_keys(key[1:])

return kl + kl2

def main():

fh = codecs.open('place.txt',"r", "UTF-8")

lines = fh.readlines()

fh.close()

i = 0

ht = {}

for l in lines:

l = l.strip()

ht[l] = 1

for k in ht:

if len(k) == 1:

continue

subk = {}

kl = enmu_all_keys(k)

for sk in kl:

#print sk, sk != k ,ht.has_key(sk)

if sk != k and ht.has_key(sk):

subk[sk] = 1

ht[k] = subk

for k in ht:

if ht[k] != 1 and ht[k] != {}:

s = ''

for sk in ht[k]:

s = s + sk + ',';

print (k+'/'+s.strip(',')).encode('UTF-8')

else :

print (k+'/null').encode("UTF-8")

if __name__ == "__main__":

main()