记一道有趣的sql题

有一张运单表：dwd_biz_waybill_td，该表的主键是way_bill_id，并且有如下字段：

way_bill_id（运单表主键），shiping_date（下单日期，时间格式为yyyy-MM-dd），payment_customer_id（付款客户），damaged_degree_type（是否严重破损，为枚举值。1为是，0为否），is_throw_away（是否丢失，为枚举值。1为是，0为否），is_pickup_overtime（是否取超时，为枚举值。1为是，0为否）。需求为：

求连续俩周严重破损的客户，丢失的客户、取超时的客户。

思路：

步骤一：

求每天严重破损、取超时的客户

sql 复制代码

with v_customer_votes_base_data as (  
    select
        waybill_td.payment_customer_id,
        waybill_td.shiping_date,
	    CASE 
          when shiping_date>= date_format(date_trunc('week',current_date),'%Y-%m-%d')  then  'cur_week'
	      when cast(date_add('day',-7,cast(date_format(date_trunc('week',current_date),'%Y-%m-%d') as date )) as varchar)<=shiping_date
               and shiping_date<cast(date_format(date_trunc('week',current_date),'%Y-%m-%d') as varchar)  then  'last_week'
          else null
        end as shiping_date_week,
        sum(if(damaged_degree_type = 1,1,0)) as payment_serious_damage_votes ,     /**严重破损**/
        sum(if(is_throw_away = 1,1,0)) as payment_throw_away_votes ,  /**丢失票(排除退栏单)**/
        sum(if(is_pickup_overtime = 1,1,0)) as payment_order_pickup_overdue_votes ,/*订单取超时数*/
    from dwd_biz_waybill_td as waybill_td
	where waybill_td.shiping_date >= cast(date_add('day',-7,cast(date_format(date_trunc('week',current_date),'%Y-%m-%d') as date )) as varchar)
	and waybill_td.shiping_date<= date_format(current_date,'%Y-%m-%d')
    group by waybill_td.payment_customer_id,waybill_td.shiping_date
),

第二步，求本周与上周的相关指标

sql 复制代码

 /*客户连续俩周的相关指标*/
v_customer_votes_result as (  
    select payment_customer_id,
	       shiping_date_week,                                                                /*本周还是上周*/
    	   sum(payment_serious_damage_votes) as payment_serious_damage_votes ,               /*严重破损*/
           sum(payment_throw_away_votes) as payment_throw_away_votes,                        /*丢失票(排除退栏单)*/
           sum(payment_order_pickup_overdue_votes) as payment_order_pickup_overdue_votes,    /*订单取超时数*/
           sum(payment_inner_overdue_votes) as payment_inner_overdue_votes                   /*内因超时票（内因）*/
    from v_customer_votes_base_data
    where shiping_date_week is not null
    group by payment_customer_id,shiping_date_week
)

第三部，将本周与上周的指标打平成一行

sql 复制代码

 /*客户连续俩周的相关指标*/
v_customer_votes_result_two_week as (  
    select payment_customer_id,
           sum(if(shiping_date_week='cur_week' and payment_serious_damage_votes>0,
	               1,
			           0)
			     ) as payment_serious_damage_votes,
           sum(if(shiping_date_week='last_week' and payment_serious_damage_votes>0,
		             1,
			           0)
			     ) as payment_serious_damage_votes_last_week,
    	     sum(if(shiping_date_week='cur_week' and payment_throw_away_votes>0,
		               1,
			             0)
			     ) as payment_throw_away_votes,
    	     sum(if(shiping_date_week='last_week' and payment_throw_away_votes>0,
		               1,
			             0)
			     ) as payment_throw_away_votes_last_week,
    	     sum(if(shiping_date_week='cur_week' and payment_order_pickup_overdue_votes>0,
		               1,
			             0)
			     ) as payment_order_pickup_overdue_votes,
    	     sum(if(shiping_date_week='last_week' and payment_order_pickup_overdue_votes>0,
		               1,
			             0)
			     ) as payment_order_pickup_overdue_votes_last_week
    from v_customer_votes_result
    group by payment_customer_id
)

第四步：计算结果：

sql 复制代码

select 
    if(payment_serious_damage_votes>0 and payment_serious_damage_votes_last_week>0,1,0) as is_serious_damage_tw,    /*是否【连续2周】严重破损*/  
    if(payment_throw_away_votes>0 and payment_throw_away_votes_last_week>0,1,0) as is_serious_damage_tw   /*是否【连续2周】丢失*/
    if(payment_order_pickup_overdue_votes>0 and payment_order_pickup_overdue_votes_last_week>0,1,0) as is_serious_damage_tw,    /*是否【连续2周】取超时*/
from v_customer_votes_result_two_week