def build_all_the_files(num_training,num_test,n):
chinese_sents,english_sents = ref.read_all_the_files(n)
with open('chinese_sents.json', 'wb') as fp:
json.dump(chinese_sents,fp)
build_occurence_dict(0,num_training)
test_ref = ref.build_chinese_word_sent_dict(num_training, num_test+num_training,{},n)
with open('test_ref_dict.json', 'wb') as fp:
json.dump(test_ref,fp)
train_data = ref.build_chinese_word_sent_dict(0,num_training,{},n)
with open('train_data.json', 'wb') as fp:
json.dump(train_data,fp)
build_cooccurrence_dict()
def build_all_the_files(num_training, num_test, n):
chinese_sents, english_sents = ref.read_all_the_files(n)
with open('chinese_sents.json', 'wb') as fp:
json.dump(chinese_sents, fp)
build_occurence_dict(0, num_training)
test_ref = ref.build_chinese_word_sent_dict(num_training,
num_test + num_training, {}, n)
with open('test_ref_dict.json', 'wb') as fp:
json.dump(test_ref, fp)
train_data = ref.build_chinese_word_sent_dict(0, num_training, {}, n)
with open('train_data.json', 'wb') as fp:
json.dump(train_data, fp)
build_cooccurrence_dict()
#!/usr/bin/env python
# -*- coding: utf-8 -*-
from microsofttranslator import Translator
# if this fails, do:
# sudo easy_install microsofttranslator
from ref_definitions2 import read_all_the_files
translator = Translator('finalproj', 'LZTVKNyxQEjmIUbMWp1HhkN4x9XkIbnT6fHhaJfLFmo=')
#print translator.translate('你好'.decode('utf8'), 'en', 'zh-CHT')
(chinese_sentences,english_sentences) = read_all_the_files()
of = open('bing_translations.txt', 'w')
for chinese_sent in chinese_sentences:
chinese_sent = ''.join([x for x in chinese_sent if x != ' '])
print >> of, chinese_sent
print >> of, translator.translate(chinese_sent.decode('utf8'), 'en', 'zh-CHT').encode('utf-8')
print >> of, '=========='
#!/usr/bin/env python
# -*- coding: utf-8 -*-
from microsofttranslator import Translator
# if this fails, do:
# sudo easy_install microsofttranslator
from ref_definitions2 import read_all_the_files
translator = Translator('finalproj',
'LZTVKNyxQEjmIUbMWp1HhkN4x9XkIbnT6fHhaJfLFmo=')
#print translator.translate('你好'.decode('utf8'), 'en', 'zh-CHT')
(chinese_sentences, english_sentences) = read_all_the_files()
of = open('bing_translations.txt', 'w')
for chinese_sent in chinese_sentences:
chinese_sent = ''.join([x for x in chinese_sent if x != ' '])
print >> of, chinese_sent
print >> of, translator.translate(chinese_sent.decode('utf8'), 'en',
'zh-CHT').encode('utf-8')
print >> of, '=========='
