def request_sougou_cloud_word(self, role_num):
'''输入:数字串
输出:与该字符串相匹配的搜狗云词'''
import sogou_cloud_words
if isinstance(role_num, int):
role_num = str(role_num)
matched_sentence = sogou_cloud_words.get_cloud_words(role_num)[0]
return matched_sentence
# sg = SouGouCloudWord(module_path='E:\SVN\chocolate_ime\script\gen_update_words\gen_hot_words')
# print sg.get_pinyin('数字串')
# print sg.form_sentence_to_role_num('数字串')#7489424826
# sg.request_sougou_cloud_word('7489424826')
# if __name__ == '__main__':
# vnd = VarifyNgramData('E:\SVN\chocolate_ime\script\gen_update_words\gen_hot_words')
#print vnd.get_pinyin_list('谁知道')
# print vnd.get_input_role('yes')
# prefix, mapping_word_list = vnd.get_code_sentence('快快乐乐的')
# print prefix, mapping_word_list
# mulit_word_pinyin_dic = gen_multi_word_pinyin_dic()
# print mulit_word_pinyin_dic[u'量']
# gw = GenMappingWordLengh()
def sougou_cloud_word(self):
import sogou_cloud_words
src_data_path = r'E:\SVN\linguistic_model\N_gram\varify_sample_lvjun'
varify_sample_filename = os.path.join(src_data_path, 'forum_high_freq_sentence_role_num.txt')
des_path = r'E:\SVN\linguistic_model\N_gram\varify_sample_lvjun'
checkout_sample_filename_backward = os.path.join(des_path, 'forum_high_freq_sentence_sougou_checkout.txt')
with codecs.open(varify_sample_filename, encoding='utf-8') as f, \
codecs.open(checkout_sample_filename_backward, mode='wb', encoding='utf-8') as wf:
count = 0
for line in f.readlines():
mapping_sentence_role_num_list = []
count += 1
print count
splited_line = line.split()
sentence = splited_line[0]
key_str = splited_line[-1].strip().encode('utf-8')
matched_sentence = sogou_cloud_words.get_cloud_words(key_str)[0]
if matched_sentence == sentence:
mapping_sentence_role_num_list.append('%s\t%s\n'%(sentence, key_str))
wf.writelines(mapping_sentence_role_num_list)
def request_sougou_cloud_word(self, role_num):
'''输入:数字串
输出:与该字符串相匹配的搜狗云词'''
import sogou_cloud_words
if isinstance(role_num, int):
role_num = str(role_num)
matched_sentence = sogou_cloud_words.get_cloud_words(role_num)[0]
return matched_sentence
# sg = SouGouCloudWord(module_path='E:\SVN\chocolate_ime\script\gen_update_words\gen_hot_words')
# print sg.get_pinyin('数字串')
# print sg.form_sentence_to_role_num('数字串')#7489424826
# sg.request_sougou_cloud_word('7489424826')
# if __name__ == '__main__':
# vnd = VarifyNgramData('E:\SVN\chocolate_ime\script\gen_update_words\gen_hot_words')
#print vnd.get_pinyin_list('谁知道')
# print vnd.get_input_role('yes')
# prefix, mapping_word_list = vnd.get_code_sentence('快快乐乐的')
# print prefix, mapping_word_list
# mulit_word_pinyin_dic = gen_multi_word_pinyin_dic()
# print mulit_word_pinyin_dic[u'量']
# gw = GenMappingWordLengh()
def get_pinyin(sentence):
add_pinyin = AddPinyin()
pinyin_list = []
for word in sentence:
pinyin = add_pinyin.get_pinyin(word)
pinyin_list.append(pinyin)
pinyin_str = ''.join(pinyin_list)
role_num = ''.join([coding_map[letter] for letter in pinyin_str if letter.isalpha()])
return role_num
def post_to_ali(role_num):
url = 'http://ali_0000.baiwenbao.com:5000/%s'%role_num
html = urllib2.urlopen(url).read()
json_list = simplejson.loads(html)
print json_list[0]
def post_to_local(role_num):
url = 'http://127.0.0.1:5000/%s'%role_num
html = urllib2.urlopen(url).read()
json_list = simplejson.loads(html)
print json_list[0]
if __name__ == "__main__":
start_time = time.time()
sentence = u'学挖掘机到底哪家强中国山东找蓝翔'
role_num = get_pinyin(sentence)
# role_num = '945464983248432694224336335433826'
# role_num = 'jintianyourenqudapingpangqiuma'
post_to_ali(role_num)
# post_to_local(role_num)
print get_cloud_words(role_num)[0]
