连续日期及连续天数

create table if not exists test_serialdate (

  rq  string comment '日期'

) stored as rcfile

;

insert into table test_serialdate

select '2019-01-01' as rq  union all

select '2019-01-02' as rq  union all

select '2019-01-05' as rq  union all

select '2019-01-06' as rq  union all

select '2019-01-08' as rq  union all

select '2019-01-09' as rq  union all

select '2019-01-10' as rq  union all

select '2019-01-11' as rq  union all

select '2019-01-17' as rq  union all

select '2019-01-18' as rq  ;

select b.gp,b.startdate,b.enddate,b.days,(case when b.gp = 0 then 0 else b.missingdays end)

from

(

select a.gp,min(a.rq) as startdate,max(a.rq) as enddate,

      (max(a.id1)-min(a.id1)+1) as days,

      max(datediff(a.rq,a.rq2)) as missingdays

from

(

  select ta.rq,

        datediff(ta.rq,'2019-01-01') as id1,    --距离初始日期的天数

        nvl(tb.id2,0) as id2,                  --比本日期小的天数

        tc.rq2,                                --比本日期小的最大日期

        nvl((datediff(ta.rq,'2019-01-01')-tb.id2),0) as gp  --比本日期小的缺失天数

  from test_serialdate ta

  left join

  ( --记录中比本日期小的数据量

    select t11.rq,count(1) as id2

    from test_serialdate t11

    inner join test_serialdate t12

    where t11.rq > t12.rq

    group by t11.rq

  ) tb

  on ta.rq = tb.rq

  left join

  ( --记录中比本日起小的最大日期

    select t21.rq,max(t22.rq) as rq2

    from test_serialdate t21

    inner join test_serialdate t22

    where t21.rq > t22.rq

    group by t21.rq

  ) tc

  on ta.rq = tc.rq

) a

group by a.gp

) b

;