离线数仓-7-数据仓库开发DIM层设计要点-拉链表同步&装载脚本

离线数仓-7-数据仓库开发DIM层设计要点-拉链表同步&装载脚本
- 一、DIM层维度模型设计要点
- - 6.用户维度表 -拉链表
  - - 1.用户维度表前期梳理
    - 2.用户维度表 DDL表设计分析
    - 3.用户维度表加载数据分析
    - - 1.拉链表首日装载数据SQL
      - 2.拉链表每日装载数据SQL
      - 1.拉链表每日装载数据 -第一种思路
        2.拉链表每日装载数据 -第二种思路
  - 7.DIM层维度模型-数据装载脚本
  - - 1.首日装载脚本
    - 2.每日装载脚本

离线数仓-7-数据仓库开发DIM层设计要点-拉链表同步&装载脚本

一、DIM层维度模型设计要点

6.用户维度表 -拉链表

拉链表回顾：
- 拉链表是什么：记录每条信息的生命周期，一旦一条记录的生命周期结束，就重新开始一条新的记录，并把当前日期放入生效开始日期。
- 拉链表意义：在于能够更加高效的保存维度信息的历史状态。
- 拉链表适合的场景：缓慢变化维，业务表格数据变化不频繁。
- 拉链表如何使用：分析的日期大于等于开始日期并且分析的日期小于等于结束日期。
- 之前文档汇总： https://blog.csdn.net/weixin_38136584/article/details/129137583?spm=1001.2014.3001.5501

1.用户维度表前期梳理

从业务系统寻找用户维度表的主维表和相关维表
- 1.获取用户相关表格如下：
  ods_user_info_inc
- 2.分析与之关联的每个表格中的具体字段，抽离出来“用户维度表”所需字段
- 3.确定这些表格中，谁是主维表。粒度小的是主维表，用户表user，所以此表作为主维表。
- 4.主维表确定以后，
  - 之前规则：每日全量表主维表对应的一行代表什么信息，维度表一行就代表对应信息；
  - 但是此处，用户表做为主维表，同时用户表也是一张拉链表，拉链表规定，每行是一个状态记录，所以这里每行代表一个用户状态。

2.用户维度表 DDL表设计分析

拉链表的开始时间和结束时间需要注意
拉链表按照什么做分区
- hive中对表进行分区的意义：分区后，一个表中，不同数据存放在不同路径，后面查询时，使用分区作为过滤条件，极大减少了查询的数据量级，查询效率会更高，便于将来查询。
- 对于拉链表来说，将来查询场景需要分析，拉链表也是一张维度表，
  - 维度表查询，一方面是获取全量最新，另一方面是获取历史上一份数据
  - 对于查询频繁方向来看，拉链表对于查询最新数据是比较频繁的。
  - 所以，就上面思考，设计拉链表的时候，需要思考：怎样加快查询最新一天拉链表的查询速度。
分区规划的设计如下图：
- dt最大的时间分区里面，存放全量最新的用户数据。
- dt之前的分区，存放当日过期的用户数据。
具体sql如下：

DROP TABLE IF EXISTS dim_user_zip;
CREATE EXTERNAL TABLE dim_user_zip
(
    `id`           STRING COMMENT '用户id',
    `login_name`   STRING COMMENT '用户名称',
    `nick_name`    STRING COMMENT '用户昵称',
    `name`         STRING COMMENT '用户姓名',
    `phone_num`    STRING COMMENT '手机号码',
    `email`        STRING COMMENT '邮箱',
    `user_level`   STRING COMMENT '用户等级',
    `birthday`     STRING COMMENT '生日',
    `gender`       STRING COMMENT '性别',
    `create_time`  STRING COMMENT '创建时间',
    `operate_time` STRING COMMENT '操作时间',
    `start_date`   STRING COMMENT '开始日期',
    `end_date`     STRING COMMENT '结束日期'
) COMMENT '用户表'
    PARTITIONED BY (`dt` STRING)
    STORED AS ORC
    LOCATION '/warehouse/gmall/dim/dim_user_zip/'
    TBLPROPERTIES ('orc.compress' = 'snappy');

3.用户维度表加载数据分析

对于拉链表来说，只做增量同步即可。
拉链表具体装载过程如下：
- 1.拉链表最开始加载的处理:全量同步数据到拉链表，并且添加开始时间和结束时间
  - 开始日期：以什么日期为准？获取不到开始时间就以拉链表最初同步数据的时间为准。
  - 结束时间：添加最大时间即可
- 2.拉链表第二天加载数据的处理：
  - 1.基于binlog来获取当天用户全量表相关更新操作日志，进行分析
  - 2.获取当天的用户变化表，并添加上开始时间和结束时间
  - 3.当天用户变化表和前一天全量的用户拉链表进行合并，得到最新的全量用户拉链表。
- 3.拉链表后面的加载数据跟第二天一致。
数据流向：基于拉链表具体装载过程的分析，需要书写两个加载过程，分别是第一天和后面的。

1.拉链表首日装载数据SQL

隐私信息的脱敏，手机号，用户名等等

insert overwrite table dim_user_zip partition (dt='9999-12-31')
select
    data.id,
    data.login_name,
    data.nick_name,
    md5(data.name),
    md5(data.phone_num),
    md5(data.email),
    data.user_level,
    data.birthday,
    data.gender,
    data.create_time,
    data.operate_time,
    '2020-06-14' start_date,
    '9999-12-31' end_date
from ods_user_info_inc
where dt='2020-06-14'
and type='bootstrap-insert';

2.拉链表每日装载数据SQL

1.拉链表每日装载数据 -第一种思路

第一种SQL思路具体过程
- 1.获取子查询前一日全量数据
- 2.获取子查询当日新增及变化数据
- 3.使用full join 将新表和旧表两部分数据关联
- 4.使用CET将两个子查询连接起来，为后面共享使用提供便利
- 5.然后使用if判断，数据更新到全量最新分区还是前一天的分区
  - 1.判断是否为当日全量最新数据，fulljoin之后，如果新表id不为空，取新表所有字段，否则取老表所有字段，并更新为最新的分区，整合完毕数据，即为需要当日全量最新数据，更新到时间最大分区
  - 2.判断是否为过期数据，fulljoin之后，新旧表格两边id都不为空，那么对应数据就是过期数据，需要更新到前一天分区里面，同时结束时间也需要更新为前一天的时间。
- 5.将两部分查询到的数据进行uninon all操作；
- 6.使用hive动态分区的概念，使用一个insert命令，将不同数据发往不同分区。
前一日全量数据SQL

 select
            id,
            login_name,
            nick_name,
            name,
            phone_num,
            email,
            user_level,
            birthday,
            gender,
            create_time,
            operate_time,
            start_date,
            end_date
        from dim_user_zip
        where dt='9999-12-31'

当日新增及变化数据的SQL，需要将每天变化的最终数据获取到，因为是binlog数据，会记录一天中某条记录变化的各种状态，可能一天内会出现多条数据，只获取最后的状态数据即可，所以这里需要进行判断操作。分组TopN操作：开窗函数+过滤。

select
            id,
            login_name,
            nick_name,
            md5(name) name,
            md5(phone_num) phone_num,
            md5(email) email,
            user_level,
            birthday,
            gender,
            create_time,
            operate_time,
            '2020-06-15' start_date,
            '9999-12-31' end_date
        from
        (
            select
                data.id,
                data.login_name,
                data.nick_name,
                data.name,
                data.phone_num,
                data.email,
                data.user_level,
                data.birthday,
                data.gender,
                data.create_time,
                data.operate_time,
                row_number() over (partition by data.id order by ts desc) rn
            from ods_user_info_inc
            where dt='2020-06-15'
        )t1
        where rn=1

判断是否为当日全量最新数据

with
tmp as
(
    select
        old.id old_id,
        old.login_name old_login_name,
        old.nick_name old_nick_name,
        old.name old_name,
        old.phone_num old_phone_num,
        old.email old_email,
        old.user_level old_user_level,
        old.birthday old_birthday,
        old.gender old_gender,
        old.create_time old_create_time,
        old.operate_time old_operate_time,
        old.start_date old_start_date,
        old.end_date old_end_date,
        new.id new_id,
        new.login_name new_login_name,
        new.nick_name new_nick_name,
        new.name new_name,
        new.phone_num new_phone_num,
        new.email new_email,
        new.user_level new_user_level,
        new.birthday new_birthday,
        new.gender new_gender,
        new.create_time new_create_time,
        new.operate_time new_operate_time,
        new.start_date new_start_date,
        new.end_date new_end_date
    from
    (
        select
            id,
            login_name,
            nick_name,
            name,
            phone_num,
            email,
            user_level,
            birthday,
            gender,
            create_time,
            operate_time,
            start_date,
            end_date
        from dim_user_zip
        where dt='9999-12-31'
    )old
    full outer join
    (
        select
            id,
            login_name,
            nick_name,
            md5(name) name,
            md5(phone_num) phone_num,
            md5(email) email,
            user_level,
            birthday,
            gender,
            create_time,
            operate_time,
            '2020-06-15' start_date,
            '9999-12-31' end_date
        from
        (
            select
                data.id,
                data.login_name,
                data.nick_name,
                data.name,
                data.phone_num,
                data.email,
                data.user_level,
                data.birthday,
                data.gender,
                data.create_time,
                data.operate_time,
                row_number() over (partition by data.id order by ts desc) rn
            from ods_user_info_inc
            where dt='2020-06-15'
        )t1
        where rn=1
    )new
    on old.id=new.id
)
select
    if(new_id is not null,new_id,old_id),
    if(new_id is not null,new_login_name,old_login_name),
    if(new_id is not null,new_nick_name,old_nick_name),
    if(new_id is not null,new_name,old_name),
    if(new_id is not null,new_phone_num,old_phone_num),
    if(new_id is not null,new_email,old_email),
    if(new_id is not null,new_user_level,old_user_level),
    if(new_id is not null,new_birthday,old_birthday),
    if(new_id is not null,new_gender,old_gender),
    if(new_id is not null,new_create_time,old_create_time),
    if(new_id is not null,new_operate_time,old_operate_time),
    if(new_id is not null,new_start_date,old_start_date),
    if(new_id is not null,new_end_date,old_end_date),
    if(new_id is not null,new_end_date,old_end_date) dt
from tmp

判断是否为过期数据

with
tmp as
(
    select
        old.id old_id,
        old.login_name old_login_name,
        old.nick_name old_nick_name,
        old.name old_name,
        old.phone_num old_phone_num,
        old.email old_email,
        old.user_level old_user_level,
        old.birthday old_birthday,
        old.gender old_gender,
        old.create_time old_create_time,
        old.operate_time old_operate_time,
        old.start_date old_start_date,
        old.end_date old_end_date,
        new.id new_id,
        new.login_name new_login_name,
        new.nick_name new_nick_name,
        new.name new_name,
        new.phone_num new_phone_num,
        new.email new_email,
        new.user_level new_user_level,
        new.birthday new_birthday,
        new.gender new_gender,
        new.create_time new_create_time,
        new.operate_time new_operate_time,
        new.start_date new_start_date,
        new.end_date new_end_date
    from
    (
        select
            id,
            login_name,
            nick_name,
            name,
            phone_num,
            email,
            user_level,
            birthday,
            gender,
            create_time,
            operate_time,
            start_date,
            end_date
        from dim_user_zip
        where dt='9999-12-31'
    )old
    full outer join
    (
        select
            id,
            login_name,
            nick_name,
            md5(name) name,
            md5(phone_num) phone_num,
            md5(email) email,
            user_level,
            birthday,
            gender,
            create_time,
            operate_time,
            '2020-06-15' start_date,
            '9999-12-31' end_date
        from
        (
            select
                data.id,
                data.login_name,
                data.nick_name,
                data.name,
                data.phone_num,
                data.email,
                data.user_level,
                data.birthday,
                data.gender,
                data.create_time,
                data.operate_time,
                row_number() over (partition by data.id order by ts desc) rn
            from ods_user_info_inc
            where dt='2020-06-15'
        )t1
        where rn=1
    )new
    on old.id=new.id
)
select
    old_id,
    old_login_name,
    old_nick_name,
    old_name,
    old_phone_num,
    old_email,
    old_user_level,
    old_birthday,
    old_gender,
    old_create_time,
    old_operate_time,
    old_start_date,
    cast(date_add('2020-06-15',-1) as string) old_end_date,
    cast(date_add('2020-06-15',-1) as string) dt
from tmp

总结，第一种方式的最终SQL

with
tmp as
(
    select
        old.id old_id,
        old.login_name old_login_name,
        old.nick_name old_nick_name,
        old.name old_name,
        old.phone_num old_phone_num,
        old.email old_email,
        old.user_level old_user_level,
        old.birthday old_birthday,
        old.gender old_gender,
        old.create_time old_create_time,
        old.operate_time old_operate_time,
        old.start_date old_start_date,
        old.end_date old_end_date,
        new.id new_id,
        new.login_name new_login_name,
        new.nick_name new_nick_name,
        new.name new_name,
        new.phone_num new_phone_num,
        new.email new_email,
        new.user_level new_user_level,
        new.birthday new_birthday,
        new.gender new_gender,
        new.create_time new_create_time,
        new.operate_time new_operate_time,
        new.start_date new_start_date,
        new.end_date new_end_date
    from
    (
        select
            id,
            login_name,
            nick_name,
            name,
            phone_num,
            email,
            user_level,
            birthday,
            gender,
            create_time,
            operate_time,
            start_date,
            end_date
        from dim_user_zip
        where dt='9999-12-31'
    )old
    full outer join
    (
        select
            id,
            login_name,
            nick_name,
            md5(name) name,
            md5(phone_num) phone_num,
            md5(email) email,
            user_level,
            birthday,
            gender,
            create_time,
            operate_time,
            '2020-06-15' start_date,
            '9999-12-31' end_date
        from
        (
            select
                data.id,
                data.login_name,
                data.nick_name,
                data.name,
                data.phone_num,
                data.email,
                data.user_level,
                data.birthday,
                data.gender,
                data.create_time,
                data.operate_time,
                row_number() over (partition by data.id order by ts desc) rn
            from ods_user_info_inc
            where dt='2020-06-15'
        )t1
        where rn=1
    )new
    on old.id=new.id
)
insert overwrite table dim_user_zip partition(dt)
select
    if(new_id is not null,new_id,old_id),
    if(new_id is not null,new_login_name,old_login_name),
    if(new_id is not null,new_nick_name,old_nick_name),
    if(new_id is not null,new_name,old_name),
    if(new_id is not null,new_phone_num,old_phone_num),
    if(new_id is not null,new_email,old_email),
    if(new_id is not null,new_user_level,old_user_level),
    if(new_id is not null,new_birthday,old_birthday),
    if(new_id is not null,new_gender,old_gender),
    if(new_id is not null,new_create_time,old_create_time),
    if(new_id is not null,new_operate_time,old_operate_time),
    if(new_id is not null,new_start_date,old_start_date),
    if(new_id is not null,new_end_date,old_end_date),
    if(new_id is not null,new_end_date,old_end_date) dt
from tmp
union all
select
    old_id,
    old_login_name,
    old_nick_name,
    old_name,
    old_phone_num,
    old_email,
    old_user_level,
    old_birthday,
    old_gender,
    old_create_time,
    old_operate_time,
    old_start_date,
    cast(date_add('2020-06-15',-1) as string) old_end_date,
    cast(date_add('2020-06-15',-1) as string) dt
from tmp
where old_id is not null
and new_id is not null;

2.拉链表每日装载数据 -第二种思路

1.将两个子查询（前一日全量数据、当日新增及变化数据）使用union all连接起来
2.使用开窗函数，将整个数据进行开窗，
- rk结果=1的就写到全量表中，结束日期使用最大日期；
- rk结果=2的就写到前一天分区里面，结束日期处理为前一天时间
- hive中使用动态分区来实现数据写入到不同分区。
多个select 查询uinon all 后，再进行select的话，默认获取的是第一个select的所有字段。
第二种方式的SQL

insert overwrite table dim_user_zip partition(dt)
select 
			id,
            login_name,
            nick_name,
            name,
            phone_num,
            email,
            user_level,
            birthday,
            gender,
            create_time,
            operate_time,
            start_date,
            if(rk =2 ,data_sub('2020-06-15',1),end_date) end_date,
            if(rk =1 ,'9999-12-31',data_sub('2020-06-15',1)) dt
from (
select 
			id,
            login_name,
            nick_name,
            name,
            phone_num,
            email,
            user_level,
            birthday,
            gender,
            create_time,
            operate_time,
            start_date,
            end_date,
            rank() over (partition by id order by start_date desc) rk
from 
( select
            id,
            login_name,
            nick_name,
            name,
            phone_num,
            email,
            user_level,
            birthday,
            gender,
            create_time,
            operate_time,
            start_date,
            end_date
        from dim_user_zip
        where dt='9999-12-31'
 union all 
 select
            id,
            login_name,
            nick_name,
            md5(name) name,
            md5(phone_num) phone_num,
            md5(email) email,
            user_level,
            birthday,
            gender,
            create_time,
            operate_time,
            '2020-06-15' start_date,
            '9999-12-31' end_date
        from
        (
            select
                data.id,
                data.login_name,
                data.nick_name,
                data.name,
                data.phone_num,
                data.email,
                data.user_level,
                data.birthday,
                data.gender,
                data.create_time,
                data.operate_time,
                row_number() over (partition by data.id order by ts desc) rn
            from ods_user_info_inc
            where dt='2020-06-15'
        )t1
        where rn=1
	) t2
) t3

7.DIM层维度模型-数据装载脚本

1.首日装载脚本

数仓上线第一天，执行一次即可

#!/bin/bash

APP=gmall

if [ -n "$2" ] ;then
   do_date=$2
else 
   echo "请传入日期参数"
   exit
fi 

dim_user_zip="
insert overwrite table ${APP}.dim_user_zip partition (dt='9999-12-31')
select
    data.id,
    data.login_name,
    data.nick_name,
    md5(data.name),
    md5(data.phone_num),
    md5(data.email),
    data.user_level,
    data.birthday,
    data.gender,
    data.create_time,
    data.operate_time,
    '$do_date' start_date,
    '9999-12-31' end_date
from ${APP}.ods_user_info_inc
where dt='$do_date'
and type='bootstrap-insert';
"

dim_sku_full="
with
sku as
(
    select
        id,
        price,
        sku_name,
        sku_desc,
        weight,
        is_sale,
        spu_id,
        category3_id,
        tm_id,
        create_time
    from ${APP}.ods_sku_info_full
    where dt='$do_date'
),
spu as
(
    select
        id,
        spu_name
    from ${APP}.ods_spu_info_full
    where dt='$do_date'
),
c3 as
(
    select
        id,
        name,
        category2_id
    from ${APP}.ods_base_category3_full
    where dt='$do_date'
),
c2 as
(
    select
        id,
        name,
        category1_id
    from ${APP}.ods_base_category2_full
    where dt='$do_date'
),
c1 as
(
    select
        id,
        name
    from ${APP}.ods_base_category1_full
    where dt='$do_date'
),
tm as
(
    select
        id,
        tm_name
    from ${APP}.ods_base_trademark_full
    where dt='$do_date'
),
attr as
(
    select
        sku_id,
        collect_set(named_struct('attr_id',attr_id,'value_id',value_id,'attr_name',attr_name,'value_name',value_name)) attrs
    from ${APP}.ods_sku_attr_value_full
    where dt='$do_date'
    group by sku_id
),
sale_attr as
(
    select
        sku_id,
        collect_set(named_struct('sale_attr_id',sale_attr_id,'sale_attr_value_id',sale_attr_value_id,'sale_attr_name',sale_attr_name,'sale_attr_value_name',sale_attr_value_name)) sale_attrs
    from ${APP}.ods_sku_sale_attr_value_full
    where dt='$do_date'
    group by sku_id
)
insert overwrite table ${APP}.dim_sku_full partition(dt='$do_date')
select
    sku.id,
    sku.price,
    sku.sku_name,
    sku.sku_desc,
    sku.weight,
    sku.is_sale,
    sku.spu_id,
    spu.spu_name,
    sku.category3_id,
    c3.name,
    c3.category2_id,
    c2.name,
    c2.category1_id,
    c1.name,
    sku.tm_id,
    tm.tm_name,
    attr.attrs,
    sale_attr.sale_attrs,
    sku.create_time
from sku
left join spu on sku.spu_id=spu.id
left join c3 on sku.category3_id=c3.id
left join c2 on c3.category2_id=c2.id
left join c1 on c2.category1_id=c1.id
left join tm on sku.tm_id=tm.id
left join attr on sku.id=attr.sku_id
left join sale_attr on sku.id=sale_attr.sku_id;
"

dim_province_full="
insert overwrite table ${APP}.dim_province_full partition(dt='$do_date')
select
    province.id,
    province.name,
    province.area_code,
    province.iso_code,
    province.iso_3166_2,
    region_id,
    region_name
from
(
    select
        id,
        name,
        region_id,
        area_code,
        iso_code,
        iso_3166_2
    from ${APP}.ods_base_province_full
    where dt='$do_date'
)province
left join
(
    select
        id,
        region_name
    from ${APP}.ods_base_region_full
    where dt='$do_date'
)region
on province.region_id=region.id;
"

dim_coupon_full="
insert overwrite table ${APP}.dim_coupon_full partition(dt='$do_date')
select
    id,
    coupon_name,
    coupon_type,
    coupon_dic.dic_name,
    condition_amount,
    condition_num,
    activity_id,
    benefit_amount,
    benefit_discount,
    case coupon_type
        when '3201' then concat('满',condition_amount,'元减',benefit_amount,'元')
        when '3202' then concat('满',condition_num,'件打',10*(1-benefit_discount),'折')
        when '3203' then concat('减',benefit_amount,'元')
    end benefit_rule,
    create_time,
    range_type,
    range_dic.dic_name,
    limit_num,
    taken_count,
    start_time,
    end_time,
    operate_time,
    expire_time
from
(
    select
        id,
        coupon_name,
        coupon_type,
        condition_amount,
        condition_num,
        activity_id,
        benefit_amount,
        benefit_discount,
        create_time,
        range_type,
        limit_num,
        taken_count,
        start_time,
        end_time,
        operate_time,
        expire_time
    from ${APP}.ods_coupon_info_full
    where dt='$do_date'
)ci
left join
(
    select
        dic_code,
        dic_name
    from ${APP}.ods_base_dic_full
    where dt='$do_date'
    and parent_code='32'
)coupon_dic
on ci.coupon_type=coupon_dic.dic_code
left join
(
    select
        dic_code,
        dic_name
    from ${APP}.ods_base_dic_full
    where dt='$do_date'
    and parent_code='33'
)range_dic
on ci.range_type=range_dic.dic_code;
"

dim_activity_full="
insert overwrite table ${APP}.dim_activity_full partition(dt='$do_date')
select
    rule.id,
    info.id,
    activity_name,
    rule.activity_type,
    dic.dic_name,
    activity_desc,
    start_time,
    end_time,
    create_time,
    condition_amount,
    condition_num,
    benefit_amount,
    benefit_discount,
    case rule.activity_type
        when '3101' then concat('满',condition_amount,'元减',benefit_amount,'元')
        when '3102' then concat('满',condition_num,'件打',10*(1-benefit_discount),'折')
        when '3103' then concat('打',10*(1-benefit_discount),'折')
    end benefit_rule,
    benefit_level
from
(
    select
        id,
        activity_id,
        activity_type,
        condition_amount,
        condition_num,
        benefit_amount,
        benefit_discount,
        benefit_level
    from ${APP}.ods_activity_rule_full
    where dt='$do_date'
)rule
left join
(
    select
        id,
        activity_name,
        activity_type,
        activity_desc,
        start_time,
        end_time,
        create_time
    from ${APP}.ods_activity_info_full
    where dt='$do_date'
)info
on rule.activity_id=info.id
left join
(
    select
        dic_code,
        dic_name
    from ${APP}.ods_base_dic_full
    where dt='$do_date'
    and parent_code='31'
)dic
on rule.activity_type=dic.dic_code;
"

case $1 in
"dim_user_zip")
    hive -e "$dim_user_zip"
;;
"dim_sku_full")
    hive -e "$dim_sku_full"
;;
"dim_province_full")
    hive -e "$dim_province_full"
;;
"dim_coupon_full")
    hive -e "$dim_coupon_full"
;;
"dim_activity_full")
    hive -e "$dim_activity_full"
;;
"all")
    hive -e "$dim_user_zip$dim_sku_full$dim_province_full$dim_coupon_full$dim_activity_full"
;;
esac

2.每日装载脚本

数仓上线之后，以后每天都需要执行一遍

#!/bin/bash

APP=gmall

# 如果是输入的日期按照取输入日期；如果没输入日期取当前时间的前一天
if [ -n "$2" ] ;then
    do_date=$2
else 
    do_date=`date -d "-1 day" +%F`
fi

dim_user_zip="
set hive.exec.dynamic.partition.mode=nonstrict;
with
tmp as
(
    select
        old.id old_id,
        old.login_name old_login_name,
        old.nick_name old_nick_name,
        old.name old_name,
        old.phone_num old_phone_num,
        old.email old_email,
        old.user_level old_user_level,
        old.birthday old_birthday,
        old.gender old_gender,
        old.create_time old_create_time,
        old.operate_time old_operate_time,
        old.start_date old_start_date,
        old.end_date old_end_date,
        new.id new_id,
        new.login_name new_login_name,
        new.nick_name new_nick_name,
        new.name new_name,
        new.phone_num new_phone_num,
        new.email new_email,
        new.user_level new_user_level,
        new.birthday new_birthday,
        new.gender new_gender,
        new.create_time new_create_time,
        new.operate_time new_operate_time,
        new.start_date new_start_date,
        new.end_date new_end_date
    from
    (
        select
            id,
            login_name,
            nick_name,
            name,
            phone_num,
            email,
            user_level,
            birthday,
            gender,
            create_time,
            operate_time,
            start_date,
            end_date
        from ${APP}.dim_user_zip
        where dt='9999-12-31'
    )old
    full outer join
    (
        select
            id,
            login_name,
            nick_name,
            md5(name) name,
            md5(phone_num) phone_num,
            md5(email) email,
            user_level,
            birthday,
            gender,
            create_time,
            operate_time,
            '$do_date' start_date,
            '9999-12-31' end_date
        from
        (
            select
                data.id,
                data.login_name,
                data.nick_name,
                data.name,
                data.phone_num,
                data.email,
                data.user_level,
                data.birthday,
                data.gender,
                data.create_time,
                data.operate_time,
                row_number() over (partition by data.id order by ts desc) rn
            from ${APP}.ods_user_info_inc
            where dt='$do_date'
        )t1
        where rn=1
    )new
    on old.id=new.id
)
insert overwrite table ${APP}.dim_user_zip partition(dt)
select
    if(new_id is not null,new_id,old_id),
    if(new_id is not null,new_login_name,old_login_name),
    if(new_id is not null,new_nick_name,old_nick_name),
    if(new_id is not null,new_name,old_name),
    if(new_id is not null,new_phone_num,old_phone_num),
    if(new_id is not null,new_email,old_email),
    if(new_id is not null,new_user_level,old_user_level),
    if(new_id is not null,new_birthday,old_birthday),
    if(new_id is not null,new_gender,old_gender),
    if(new_id is not null,new_create_time,old_create_time),
    if(new_id is not null,new_operate_time,old_operate_time),
    if(new_id is not null,new_start_date,old_start_date),
    if(new_id is not null,new_end_date,old_end_date),
    if(new_id is not null,new_end_date,old_end_date) dt
from tmp
union all
select
    old_id,
    old_login_name,
    old_nick_name,
    old_name,
    old_phone_num,
    old_email,
    old_user_level,
    old_birthday,
    old_gender,
    old_create_time,
    old_operate_time,
    old_start_date,
    cast(date_add('$do_date',-1) as string) old_end_date,
    cast(date_add('$do_date',-1) as string) dt
from tmp
where old_id is not null
and new_id is not null;
"

dim_sku_full="
with
sku as
(
    select
        id,
        price,
        sku_name,
        sku_desc,
        weight,
        is_sale,
        spu_id,
        category3_id,
        tm_id,
        create_time
    from ${APP}.ods_sku_info_full
    where dt='$do_date'
),
spu as
(
    select
        id,
        spu_name
    from ${APP}.ods_spu_info_full
    where dt='$do_date'
),
c3 as
(
    select
        id,
        name,
        category2_id
    from ${APP}.ods_base_category3_full
    where dt='$do_date'
),
c2 as
(
    select
        id,
        name,
        category1_id
    from ${APP}.ods_base_category2_full
    where dt='$do_date'
),
c1 as
(
    select
        id,
        name
    from ${APP}.ods_base_category1_full
    where dt='$do_date'
),
tm as
(
    select
        id,
        tm_name
    from ${APP}.ods_base_trademark_full
    where dt='$do_date'
),
attr as
(
    select
        sku_id,
        collect_set(named_struct('attr_id',attr_id,'value_id',value_id,'attr_name',attr_name,'value_name',value_name)) attrs
    from ${APP}.ods_sku_attr_value_full
    where dt='$do_date'
    group by sku_id
),
sale_attr as
(
    select
        sku_id,
        collect_set(named_struct('sale_attr_id',sale_attr_id,'sale_attr_value_id',sale_attr_value_id,'sale_attr_name',sale_attr_name,'sale_attr_value_name',sale_attr_value_name)) sale_attrs
    from ${APP}.ods_sku_sale_attr_value_full
    where dt='$do_date'
    group by sku_id
)
insert overwrite table ${APP}.dim_sku_full partition(dt='$do_date')
select
    sku.id,
    sku.price,
    sku.sku_name,
    sku.sku_desc,
    sku.weight,
    sku.is_sale,
    sku.spu_id,
    spu.spu_name,
    sku.category3_id,
    c3.name,
    c3.category2_id,
    c2.name,
    c2.category1_id,
    c1.name,
    sku.tm_id,
    tm.tm_name,
    attr.attrs,
    sale_attr.sale_attrs,
    sku.create_time
from sku
left join spu on sku.spu_id=spu.id
left join c3 on sku.category3_id=c3.id
left join c2 on c3.category2_id=c2.id
left join c1 on c2.category1_id=c1.id
left join tm on sku.tm_id=tm.id
left join attr on sku.id=attr.sku_id
left join sale_attr on sku.id=sale_attr.sku_id;
"

dim_province_full="
insert overwrite table ${APP}.dim_province_full partition(dt='$do_date')
select
    province.id,
    province.name,
    province.area_code,
    province.iso_code,
    province.iso_3166_2,
    region_id,
    region_name
from
(
    select
        id,
        name,
        region_id,
        area_code,
        iso_code,
        iso_3166_2
    from ${APP}.ods_base_province_full
    where dt='$do_date'
)province
left join
(
    select
        id,
        region_name
    from ${APP}.ods_base_region_full
    where dt='$do_date'
)region
on province.region_id=region.id;
"

dim_coupon_full="
insert overwrite table ${APP}.dim_coupon_full partition(dt='$do_date')
select
    id,
    coupon_name,
    coupon_type,
    coupon_dic.dic_name,
    condition_amount,
    condition_num,
    activity_id,
    benefit_amount,
    benefit_discount,
    case coupon_type
        when '3201' then concat('满',condition_amount,'元减',benefit_amount,'元')
        when '3202' then concat('满',condition_num,'件打',10*(1-benefit_discount),'折')
        when '3203' then concat('减',benefit_amount,'元')
    end benefit_rule,
    create_time,
    range_type,
    range_dic.dic_name,
    limit_num,
    taken_count,
    start_time,
    end_time,
    operate_time,
    expire_time
from
(
    select
        id,
        coupon_name,
        coupon_type,
        condition_amount,
        condition_num,
        activity_id,
        benefit_amount,
        benefit_discount,
        create_time,
        range_type,
        limit_num,
        taken_count,
        start_time,
        end_time,
        operate_time,
        expire_time
    from ${APP}.ods_coupon_info_full
    where dt='$do_date'
)ci
left join
(
    select
        dic_code,
        dic_name
    from ${APP}.ods_base_dic_full
    where dt='$do_date'
    and parent_code='32'
)coupon_dic
on ci.coupon_type=coupon_dic.dic_code
left join
(
    select
        dic_code,
        dic_name
    from ${APP}.ods_base_dic_full
    where dt='$do_date'
    and parent_code='33'
)range_dic
on ci.range_type=range_dic.dic_code;
"

dim_activity_full="
insert overwrite table ${APP}.dim_activity_full partition(dt='$do_date')
select
    rule.id,
    info.id,
    activity_name,
    rule.activity_type,
    dic.dic_name,
    activity_desc,
    start_time,
    end_time,
    create_time,
    condition_amount,
    condition_num,
    benefit_amount,
    benefit_discount,
    case rule.activity_type
        when '3101' then concat('满',condition_amount,'元减',benefit_amount,'元')
        when '3102' then concat('满',condition_num,'件打',10*(1-benefit_discount),'折')
        when '3103' then concat('打',10*(1-benefit_discount),'折')
    end benefit_rule,
    benefit_level
from
(
    select
        id,
        activity_id,
        activity_type,
        condition_amount,
        condition_num,
        benefit_amount,
        benefit_discount,
        benefit_level
    from ${APP}.ods_activity_rule_full
    where dt='$do_date'
)rule
left join
(
    select
        id,
        activity_name,
        activity_type,
        activity_desc,
        start_time,
        end_time,
        create_time
    from ${APP}.ods_activity_info_full
    where dt='$do_date'
)info
on rule.activity_id=info.id
left join
(
    select
        dic_code,
        dic_name
    from ${APP}.ods_base_dic_full
    where dt='$do_date'
    and parent_code='31'
)dic
on rule.activity_type=dic.dic_code;
"

case $1 in
"dim_user_zip")
    hive -e "$dim_user_zip"
;;
"dim_sku_full")
    hive -e "$dim_sku_full"
;;
"dim_province_full")
    hive -e "$dim_province_full"
;;
"dim_coupon_full")
    hive -e "$dim_coupon_full"
;;
"dim_activity_full")
    hive -e "$dim_activity_full"
;;
"all")
    hive -e "$dim_user_zip$dim_sku_full$dim_province_full$dim_coupon_full$dim_activity_full"
;;
esac