#!/usr/bin/python
# -*- coding: UTF-8 -*-
import sys
import codecs   #可以以特定編碼開啟檔案
import jieba
import jieba.posseg as pseg
reload(sys)               #zzh說這種方法不好，不要再用了！！！  可是真的很好用啊 QAQ
sys.setdefaultencoding('utf-8')
import gensim

#model = gensim.models.Word2Vec.load("22620491.model")
model = gensim.models.KeyedVectors.load_word2vec_format('news_12g_baidubaike_20g_novel_90g_embedding_64.bin'
 
, binary=True)
word_vec = model.wv
del model     #把模型給word_vec，所以Model刪掉。
'''
print word_vec[u'難過']
'''
f = codecs.open("qinggancidanmu.txt",'r','utf-8')    #codecs包指定TXT開啟方式
lines = f.readlines()
doc = open('fenlei.txt', 'w')




for line in lines: #每一行彈幕
if lines.index(line) % 500 ==0:   #顯示跑到多少條資料
print 
 
lines.index(line)

    words=line.split(" ")
    words.pop(0)
    u = []
    for word in words:
            if word != "\r\n":    #去掉換行符，linux只用\n換行。win下用\r\n表示換行。反正\n不行就\r\n試試！
                #print type(word)
u.append(word)    #word_vec輸入必須要unicode才行。
le = [u'樂']
    ai = [u'哀']
    nu = [u'怒']
    jing = [u'驚'
 
]
    wu = [u'惡']
    try:
        l,a,n,j,w=word_vec.n_similarity(u, le),word_vec.n_similarity(u, ai),word_vec.n_similarity(u, nu),word_vec.n_similarity(u, jing),word_vec.n_similarity(u, wu)
        list=[l,a,n,j,w]
        #print list
doc.write(line.split(" ",1)[0]+" "+str(list.index(max(list)))+" "+line.split(" ",1)[1]+'\n')      # index記得 變成 str啊 ！！改了好半天!
except:
        doc.write(line.split(" ",1)[0]+" "+"-1"+" "+line.split(" ",1)[1]+'\n')
        continue
print("end")
f.close()
doc.close()