# -*- coding: utf-8 -*-
# @Author: Gree
# @Date:   2021-06-02 14:35:10
# @Last Modified by:   Gree
# @Last Modified time: 2021-06-02 14:38:01


import re
import pandas as pd


class SportsCheck:
    """体育语料自动化分类的检查"""
    def sports_check(self, row):
        """
        sports_check 函数:
        input:
        output: generator
        features: 体育语料自动化分类的检查
        step1: 检查 sports 的语料字段 domain、intent、response_text 是否正确 ✅
        """
        # 获取 domain
        domain = row['domain']
        # 获取 query
        query = row['query']
        # 获取 intent
        intent = row['intent']
        # 获取 response_text
        response_text = row['response_text']

        # 捕获异常
        try:
            # 正则表达式匹配数据规律
            query_result = re.search(r'.*?(赛事|比赛|nba|cba|中超|詹姆斯|科比|乔丹|国安|英超|切尔西|对阵|阿森纳|塞尔利亚人|冠军|足球|湖人|决赛|巴斯坦人|广州恒大|女排|火箭|联赛|皇马|梅西|雷霆|拜仁|曼联|奥运会|森林狼|凯尔特人|世界杯|欧洲杯|公牛队|活塞队|老鹰|步行者|尤文图斯|赢了|比分|阿斯特拉|多少分|杜兰特|篮网|热火|猛龙|灰熊|欧联杯|骑士|热刺|对手|皇家马德里|队|巴萨|美洲杯|利物浦|vs|ac米兰).*', query)

        except Exception as e:
            print("The error of getting query_result in the module of sports_check():", e)


        # 捕获异常
        try:
            # 条件判断
            if query_result is not None:
                row['domain_is_right'] = 'yes'

                if 'search' in intent:
                    row['intent_is_right'] = 'yes'
                else:
                    row['reply_is_right'] = 'no'

                try:
                    if response_text is not None:
                        row['response_is_right'] = 'yes'
                    else:
                        row['response_is_right'] = 'no'
                except:
                    pass

                # 生成器
                yield {
                    'date_time': row['date_time'],
                    'request_id': row['request_id'],
                    'mac_wifi': row['mac_wifi'],
                    'user_id': row['user_id'],
                    'query': query,
                    'domain': domain,
                    'intent': intent,
                    'response_text': response_text,
                    'domain_is_right': row['domain_is_right'],
                    'intent_is_right': row['intent_is_right'],
                    'response_is_right': row['response_is_right']
                }

            else:

                # 生成器
                yield {
                    'date_time': row['date_time'],
                    'request_id': row['request_id'],
                    'mac_wifi': row['mac_wifi'],
                    'user_id': row['user_id'],
                    'query': query,
                    'domain': domain,
                    'intent': intent,
                    'response_text': response_text,
                    'domain_is_right': "",
                    'intent_is_right': "",
                    'response_is_right': ""
                }

        except Exception as e:
            print("The error of getting generator in the module of sports_check():", e)