storage.py 文件源码

def insert_into_reverse_dict(self, hash_val, text):
        """
        ????: ??????
        ????:
            @hash: ??text????
            @text: ??text
        ????: ??????????????20%?, ???????, ????????.
        """
        word_num = 0;
        weight_avg = 0;
        weight_total = 0;

        word_list = []
        weight_list = []

        # ????
        word_with_weight = jieba.analyse.extract_tags(text, withWeight=True)
        for word, weight in word_with_weight:
            word_num += 1;
            weight_total += float(weight);
        if word_num > 0:
            weight_avg = weight_total / word_num;
        for word, weight in word_with_weight:
            if weight < (self.rate * weight_avg):
                break
            word_list.append(word);
            weight_list.append(weight);

        # ???????
        list_len = len(word_list)
        key_list = self.gen_key_list(word_list, weight_list, list_len, self.word_max_len)
        for key in key_list:
            self.reverse_dict.add(key, 100, hash_val); # ????(key -> hash)