elab_mvp/mvp.py

from mysql_db import MysqlDB
from excel_util import ExcelUtil


class Mvp:
    """
     ce mvp 答题数据统计
     城市特例 北京市，上海市， 重庆市，天津市
    """

    age_dict = {
        '00-04年生': '00后',
        '05-09年生': '05后',
        '50-59年生': '50后',
        '60-69年生': '60后',
        '70-74年生': '70后',
        '75-79年生': '75后',
        '80-84年生': '80后',
        '85-89年生': '85后',
        '90-94年生': '90后',
        '95-99年生': '95后'
    }

    crowd = ['A', 'B', 'C', 'D', 'E', 'F']

    # 获取答题记录中城市列表
    sql_1 = 'select city from f_t_daren_score_2 group by city'

    # 获取父选项和父题id
    sql_2 = 'select a.id, a.content, b.id, b.name from bq_option a left join bq_question b on a.question_id = b.id ' \
            'where a.serial_number = %s and b.serial_number = %s and a.status = b.status = 1 '

    # 获取答题人的年龄段集合
    sql_4 = 'select nld from f_t_daren_score_2 group by nld'

    # 根据城市，年龄段，人群分类统计答题记录数
    sql_5 = 'select group_type, COUNT(uuid) from f_t_daren_score_2 where (city = %s or province = %s) and nld ' \
            '= %s and uuid in %s group by group_type '

    # 根据父选项获取子选项id列表
    sql_6 = 'SELECT c.id, c.sub_question_id, c.content FROM bq_sub_option c WHERE c.father_id in (SELECT a.id FROM ' \
            'bq_option a ' \
            'LEFT JOIN bq_question b ON a.question_id = b.id WHERE a.serial_number = %s AND b.serial_number = %s ' \
            'and a.status = 1 and b.status = 1) and c.status = 1 '

    # 根据子题id获取包含子题id的测试
    sql_7 = 'select group_type from bq_testcase where status = 1 and FIND_IN_SET(%s, question_ids)'

    # 根据子选项id统计答题数
    sql_8 = 'SELECT count(1) FROM f_t_daren_score_2 a LEFT JOIN d_shangju_tiku_02 b ON a.sub_question_id = ' \
            'b.sub_question_id AND (a.score  = b.score or a.score = b.sub_option_id) and a.testcase_id = ' \
            'b.testcase_id WHERE b.sub_option_id in %s' \
            'and (a.city = %s or a.province = %s) and a.nld = %s and a.uuid in %s'

    # 获取一个uuid下答题的子选项id列表
    sql_10 = 'select  DISTINCT uuid, GROUP_CONCAT(DISTINCT b.sub_option_id)  from f_t_daren_score_2 a left join ' \
             'd_shangju_tiku_02 b on a.sub_question_id = b.sub_question_id and (a.score = b.score or a.score = ' \
             'b.sub_option_id) where a.status = ' \
             'b.status = 1 group by uuid '

    # 向表mvp_crowd_info插入数据
    sql_11 = 'insert into mvp_crowd_info(age_area, city_name, crowd_type, status) values(%s, %s, %s, 1)'

    # 向表mvp_crowd_info_behavior中插入数据
    sql_12 = 'insert into mvp_crowd_info_behavior(crowd_info_id, behavioral_interest, standard_value, status) values(' \
             '%s, %s, ' \
             '%s, 1) '

    # 向表mvp_crowd_info_module中插入数据
    sql_13 = 'insert into mvp_crowd_info_module(crowd_info_id, module_name, standard_value, status) values (%s, %s, ' \
             '%s, 1) '

    sql_14 = 'select a.id, a.age_area, a.city_name, a.crowd_type from mvp_crowd_info a where a.status = 1'

    def __init__(self, path=None):
        self.shangju_db = MysqlDB('shangju')
        self.marketing_db = MysqlDB('bi_report')
        # self.shangju_db.truncate('mvp_standard_score')
        self.tag_data = ExcelUtil(file_name=path).init_mvp_data()
        self.crowd_info = ExcelUtil(file_name=path, sheet_name='选项-人群分类对应表').init_crowd_info()
        self.citys = self.init_city()
        self.age = self.init_age()
        self.people_sub_option_ids = self.marketing_db.select(self.sql_10)
        self.crowd_contain_sub_option_ids = self.get_crowd_contain_sub_option_ids()
        self.module_scores = ExcelUtil(file_name='set-behavior-tag.xlsx', sheet_name='算法关系表').init_module_info()
        # self.scores_tag = ExcelUtil(file_name='行为与模块分值汇总.xlsx', sheet_name='行为').init_scores()
        # self.score_module = ExcelUtil(file_name='行为与模块分值汇总.xlsx', sheet_name='模块').init_scores()
        self.scores_tag = None
        self.score_module = None

    def close(self):
        self.shangju_db.close()
        self.marketing_db.close()

    def init_city(self):
        """
            获取答题数据中的城市。
        :return:
        """
        citys = ['宁波市', '上海市', '苏州市', '无锡市', '宁波市']
        # citys_info = self.marketing_db.select(self.sql_1)
        # citys.extend([x[0] for x in citys_info if x[0] is not None])
        return citys

    def query_behavioral_info(self, city=None, age=None, crowd=None):
        """
            查询行为兴趣信息
        :return:
        """
        # datas = []
        # for key in self.tag_data.keys():
        #     values = self.tag_data[key]
        #     for value in values:
        #         question = value[0].split('-')[0]
        #         option = value[0].split('-')[1]
        #         corr = value[1]
        #         data = self.shangju_db.select(self.sql_2, [option, question])
        #         if len(data) > 0:
        #             print([question, option, data[0][3], data[0][1], key, corr])
        #             datas.append([question, option, data[0][3], data[0][1], key, corr])
        # self.shangju_db.truncate('mvp_question_classification')
        # self.shangju_db.add_some(self.sql_3, datas)
        scores_behavioral = self.city_age_crowd(city, age, crowd)
        # scores_module = self.module_score(crowd, city, age, scores_behavioral['score'])
        # result = {'行为兴趣分值': scores_behavioral['score'], '模块分值': scores_module}
        print('update finished!!!')
        return scores_behavioral

    def module_score(self, crowd, city, age, scores):
        """
            模块分数计算
            城市 年龄 人群分类 模块名称 分数
        :return:
        """
        import json
        print(json.dumps(scores, ensure_ascii=False))
        modules = self.module_scores[crowd]
        result = []
        for key in modules.keys():
            values = modules[key]
            module_name = key
            score = 0
            for value in values:
                behavioral_name = value[0]
                weight = float(value[2])
                standard_score = [x[4] for x in scores if x[2] == behavioral_name]
                if len(standard_score) > 0:
                    score += standard_score[0] * weight
            result.append([city, age, crowd, module_name, score])
        return result

    # def insert_data(self, scores_behavioral, scores_module):
    def insert(self):
        """
            计算数据写入数据库中，供接口查看
        :return:
        """
        infos = []
        for city in ['上海市', '宁波市', '苏州市', '杭州市', ' 无锡市']:
            for age in ['50-59年生', '60-69年生', '70-74年生', '75-79年生', '80-84年生', '85-89年生', '90-94年生', '95-99年生', '00'
                                                                                                                '-04年生',
                        '05-09年生']:
                for c_type in ['A', 'B', 'C', 'D', 'E', 'F']:
                    age_area = self.age_dict.get(age)
                    if age_area:
                        infos.append([age_area, city, c_type])
        self.shangju_db.add_some(self.sql_11, infos)

    def query_data(self):
        ids = self.shangju_db.select(self.sql_14)
        return ids

    def shanghai_85_module_score_insert(self):
        """
            上海市，85后模块分数计算
        :return:
        """
        result = []
        for crowd in self.crowd:
            modules = self.module_scores[crowd]
            for key in modules.keys():
                values = modules[key]
                module_name = key
                score = 0
                for value in values:
                    behavioral_name = value[0]
                    weight = float(value[2])
                    # standard_score = [x[4] for x in scores if x[2] == behavioral_name]
                    standard_score = float(value[1])
                    if standard_score is not None:
                        score += standard_score * weight
                result.append(['上海市', '85后', crowd, module_name, score])
        return {'score': result, 'data': self.module_scores}

    def tag_module_score_insert(self):
        """
            标签模块分数写入数据库
        :return:
        """
        ids = self.query_data()
        insert_data = []
        insert_data_1 = []
        for tag, module in zip(self.scores_tag, self.score_module):
            city = tag[0]
            age = tag[1]
            crowd = tag[2]
            tag_name = tag[3]
            tag_score = tag[4]

            city_2 = module[0]
            age_2 = module[1]
            crowd_2 = module[2]
            module_name_2 = module[3]
            module_score_2 = module[4]

            for id in ids:
                city_1 = id[2]
                age_1 = id[1]
                crowd_1 = id[3]
                id_1 = id[0]
                if city == city_1 and self.age_dict[age] == age_1 and crowd == crowd_1:
                    insert_data.append([id_1, tag_name, tag_score])
                if city_2 == city_1 and self.age_dict[age_2] == age_1 and crowd_2 == crowd_1:
                    insert_data_1.append([id_1, module_name_2, module_score_2])
        self.shangju_db.add_some(self.sql_12, insert_data)
        self.shangju_db.add_some(self.sql_13, insert_data_1)

    def init_age(self):
        """
           获取答题数据中的年龄
        """
        age_info = self.marketing_db.select(self.sql_4)
        # print([x[0] for x in age_info])
        return [x[0] for x in age_info if x[0] is not None]

    def city_age_crowd(self, city=None, age=None, crowd=None):
        data_start = []
        result = []
        module_scores = []
        if city is not None and age is not None and crowd is not None:
            print('获取指定城市，年龄段，人群类型的数据...')
            people_uuids = self.get_people_uuid_by_type(crowd)
            behavior_data = None
            if len(people_uuids) > 0:
                print('{}-{}-{}'.format(city, age, crowd))
                datas = self.behavior_tag_init(city, age, people_uuids)
                data_start.append(datas)
                all_data, behavior_data_1 = self.calculation_standard_score(datas, city, age, crowd)
                result.append(all_data)
                behavior_data = behavior_data_1
            if behavior_data:
                module_scores.extend(self.module_score(crowd, city, age, behavior_data))
            pass
        else:
            print('获取所有case的数据...')
            # for city in self.citys:
            # for city in [city]:
            for age in self.age:
                for crowd_type in self.crowd:
                    if age == '85-89年生' and city == '上海市':
                        print('上海市85后数据导入人工值，无需计算...')
                        pass
                    else:
                        # print(' {}{}'.format(city, age))
                        people_uuids = self.get_people_uuid_by_type(crowd_type)
                        behavior_data = None
                        if len(people_uuids) > 0:
                            print('{}-{}-{}'.format(city, age, crowd_type))
                            datas = self.behavior_tag_init(city, age, people_uuids)
                            data_start.append(datas)
                            all_data, behavior_data_1 = self.calculation_standard_score(datas, city, age, crowd)
                            result.append(all_data)
                            behavior_data = behavior_data_1
                        if behavior_data:
                           module_scores.extend(self.module_score(crowd_type, city, age, behavior_data))
        # return result
        # data_list = []
        # for e in data_start:
        #     for key in e.keys():
        #         values = e[key]
        #         for sub_e in values:
        #             ele = [key]
        #             ele.extend(sub_e)
        #             data_list.append(ele)
        #     pass
        return {'tag_score': result, 'module_score': module_scores}
        # return {'score': result, 'data': data_list}

    def behavior_tag_init(self, city, age, people_uuids):
        result = {}
        self.group_type_count = self.marketing_db.select(self.sql_5, [city, city, age, people_uuids])
        # 表名
        for key in self.tag_data:
            values = self.tag_data[key]
            result_sub = {}
            # 标签
            for key_tag_name in values.keys():
                questions = values[key_tag_name]
                elements = []
                for value in questions:
                    question = value[0].split('-')[0]
                    option = value[0].split('-')[1]
                    corr = value[1]
                    fz, fm = self.molecular_value(question, option, city, age, people_uuids)
                    if fm == 0:
                        c = 0
                    else:
                        c = fz / fm
                    elements.append([question, option, corr, fz, fm, c])
                result_sub[key_tag_name] = elements
            result[key] = self.indicator_calculation_d_e(result_sub)
        return result

    def molecular_value(self, queston, option, city, age, people_uuids):
        # 获取当前父选项包含的子选项id和子题id列表
        result = self.shangju_db.select(self.sql_6, [option, queston])
        sub_option_ids = []
        group_types = []
        for rt in result:
            sub_option_id, sub_question_id, content = rt[0], rt[1], rt[2]
            grouptypes = self.shangju_db.select(self.sql_7, [sub_question_id])
            for g_t in grouptypes:
                if g_t[0] not in group_types:
                    group_types.append(g_t[0])
            sub_option_ids.append(sub_option_id)
        # 计算子选项在答题记录中的点击数
        sub_options_count = 0
        if len(sub_option_ids) > 0:
            result_1 = self.marketing_db.select(self.sql_8, [sub_option_ids, city, city, age, people_uuids])
            sub_options_count = result_1[0][0]
        # 计算父选项包含的子选项对应的子题所在的测试gt包含的点击数。
        denominator_value = 0
        for info in self.group_type_count:
            if info[0] in group_types:
                denominator_value += info[1]
        return sub_options_count, denominator_value

    def indicator_calculation_d_e(self, data):
        result = {}
        for key in data.keys():
            values = data[key]
            c_list = []
            for x in values:
                _x = x[5]
                if _x is not None and x != 0:
                    c_list.append(_x)
            fm_list = [x[4] for x in values]
            sum_c = sum(fm_list)
            if len(c_list) == 0:
                min_c = 0
            else:
                min_c = min(c_list)
            elements = []
            for value in values:
                _value = []
                c = value[5]
                if sum_c == 0:
                    d = 0
                else:
                    d = c / sum_c
                e = c - min_c
                _value.extend(value)
                _value.append(d)
                _value.append(e)
                elements.append(_value)
            result[key] = elements
        return result

    def calculation_standard_score(self, datas, city, age, crowd_type):
        scores = {}
        for key_tag_type in datas.keys():
            print(key_tag_type)
            tag_type_data = datas[key_tag_type]
            scores_sub = []
            for key_tag in tag_type_data.keys():
                key_tag_data = tag_type_data[key_tag]
                print(key_tag)
                print('     父题序号 父选项序号 相关系系数 分子值 分母值 百分比 人数权重 偏离值')
                values = [x[5] for x in key_tag_data]
                min_c = min(values)
                f = min_c
                for value in key_tag_data:
                    print('     {}'.format(value))
                    if value[2] is not None and value[7] is not None:
                        f += float(value[2] * value[7])
                print('     标准分：{}'.format(f))
                scores_sub.append([city, age, key_tag, crowd_type, f])
            scores[key_tag_type] = scores_sub
            # self.shangju_db.add_some(self.sql_9, scores)
        return scores, scores['用户画像-行为兴趣']

    def get_crowd_people(self):
        result = {}
        for type in self.crowd:
            uuids = self.get_people_uuid_by_type(type)
            result[type] = len(uuids)
        return result

    def get_people_uuid_by_type(self, type):
        uuids = []
        type_sub_option_ids = self.crowd_contain_sub_option_ids[type]
        for people in self.people_sub_option_ids:
            uuid = people[0]
            sub_option_ids = list(map(int, str(people[1]).split(',')))
            # list(set(a).intersection(set(b)))
            if len(list(set(sub_option_ids).intersection(set(type_sub_option_ids)))) > 0 and uuid not in uuids:
                uuids.append(uuid)
        return uuids

    def get_crowd_contain_sub_option_ids(self):
        """
            获取ABCDEF人群包含的子选项id
        :return:
        """
        infos = {}
        for key in self.crowd_info.keys():
            values = self.crowd_info[key]
            sub_option_ids = []
            for value in values:
                if value is not None:
                    vals = str(value).split('-')
                    option, question = vals[1], vals[0]
                    query_result = self.shangju_db.select(self.sql_6, [option, question])
                    for qr in query_result:
                        sub_option_id, sub_question_id, content = qr[0], qr[1], qr[2]
                        sub_option_ids.append(int(sub_option_id))
            infos[key] = sub_option_ids
        print(infos)
        return infos