import gensim
from gensim.models import word2vec
import logging
import jieba
import os
import numpy as np


def cut_txt(old_file):
    import jieba
    global cut_file     # 分词之后保存的文件名
    cut_file = old_file + '_cut.txt'

    try:
        fi = open(old_file, 'r', encoding='utf-8')
    except BaseException as e:  # 因BaseException是所有错误的基类，用它可以获得所有错误类型
        print(Exception, ":", e)    # 追踪错误详细信息

    text = fi.read()  # 获取文本内容
    new_text = jieba.cut(text, cut_all=False)  # 精确模式
    str_out = ' '.join(new_text).replace('，', '').replace('。', '').replace('？', '').replace('！', '') \
        .replace('“', '').replace('”', '').replace('：', '').replace('…', '').replace('（', '').replace('）', '') \
        .replace('—', '').replace('《', '').replace('》', '').replace('、', '').replace('‘', '') \
        .replace('’', '')     # 去掉标点符号
    fo = open(cut_file, 'w', encoding='utf-8')
    fo.write(str_out)
def model_train(train_file_name, save_model_file):  # model_file_name为训练语料的路径,save_model为保存模型名
    from gensim.models import word2vec
    import gensim
    import logging
    # 模型训练，生成词向量
    logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)
    sentences = word2vec.Text8Corpus(train_file_name)  # 加载语料
    model = gensim.models.Word2Vec(sentences, size=200,min_count=1)  # 训练skip-gram模型; 默认window=5
    model.save(save_model_file)
    model.wv.save_word2vec_format(save_model_name + ".bin", binary=True)   # 以二进制类型保存模型以便重用

if __name__=='__main__':
    cut_txt('fenci.txt')  # 须注意文件必须先另存为utf-8编码格式
    save_model_name = 'fenci.model'
    if not os.path.exists(save_model_name):     # 判断文件是否存在
        model_train(cut_file, save_model_name)
    else:
        print('此训练模型已经存在，不用再次训练')

    # 加载已训练好的模型
    model = word2vec.Word2Vec.load(save_model_name)

    y1 = model.similarity("何太冲", "张无忌")
    print(u"何太冲和张无忌的相似度为：", y1)

    y2 = model.most_similar("张无忌", topn=10)  # 10个最相关的
    print(u"和张无忌最相关的词有：\n")
    for item in y2:
        print(item[0], item[1])