为相同的序列元素分配多个连续的重复

Question

我正在使用 SQL 服务器 table，它看起来像这样：

diff	nom	rec_num
-7	abc	9
-6	abc	9
-5	abc	7
-4	abc	6
-3	abc	6
-2	abc	6
-1	abc	13
-7	fav	7
-6	fav	9
-5	fav	9
-4	fav	13
-3	fav	7
-2	fav	7
-1	fav	7

真实的table要大很多，这是样本。

我需要为 table 中的每个 rec_num 指定它们连续重复的次数。比如有一个序列5 5 4 4 4 3 5 5 5 5 1，那么工作的结果应该是2 2 3 3 3 1 4 4 4 4 1。我需要的是连续 的重复次数 ，而不是重复的总次数。

使用ROW_NUMBER的组合：

with cte as (
  select diff, nom, rec_num,
  row_number() over (partition by nom order by nom, diff desc) rn1,
  row_number() over (partition by nom, rec_num order by nom, diff desc) rn2
  from t)
  
 select diff, nom, rec_num,
   row_number() over (partition by nom, rn1 - rn2 order by nom, diff desc) as con_repeats
from cte

我得到了以下 table

diff	nom	rec_num	con_repeats
-1	abc	13	1
-2	abc	6	1
-3	abc	6	2
-4	abc	6	3
-5	abc	7	1
-6	abc	9	1
-7	abc	9	2
-1	fav	7	1
-2	fav	7	2
-3	fav	7	3
-4	fav	13	1
-5	fav	9	1
-6	fav	9	2
-7	fav	7	1

但我需要 table 看起来像这样：

diff	nom	rec_num	con_repeats
-1	abc	13	1
-2	abc	6	3
-3	abc	6	3
-4	abc	6	3
-5	abc	7	1
-6	abc	9	2
-7	abc	9	2
-1	fav	7	3
-2	fav	7	3
-3	fav	7	3
-4	fav	13	1
-5	fav	9	2
-6	fav	9	2
-7	fav	7	1

我该怎么做？

Answer 1

您只需使用 group by

即可完成此操作

with x as (
    select nom, rec_num, Count(*) n
    from t
    group by nom,rec_num
)
update t set t.con_repeats=x.n
from x
join t on t.nom=x.nom and t.rec_num=x.rec_num

见Fiddle

编辑

问题弄清楚后，需要一个不同的解决方案，这里使用window functions来识别重复值的孤岛，updatable CTE将每个孤岛组的max count应用到来源 table:

with groups as (
    select t.*,
        Dense_Rank() over (partition by nom order by (rn - rn2), rec_num) as grp,
        Row_Number() over (partition by nom, (rn - rn2), rec_num order by diff) as c
    from(
        select t.*, row_number() over (partition by nom order by diff) as rn, Row_Number() over (partition by nom, rec_num order by diff) as rn2
        from t
    )t
),
cnt as (
    select *, Max(c) over (partition by nom,grp) maxc
    from groups
)
update cnt set con_repeats=maxc;

select * from t;

看到这个New fiddle

Answer 2

-- Please see if this helps.

SELECT t1.*, repeats FROM tbl t1
INNER JOIN 
    (SELECT repeats, nom, rec_num FROM (
        (SELECT COUNT(*) repeats, nom, rec_num FROM tbl GROUP BY nom, rec_num))t
     ) t2
ON t2.nom = t1.nom and t2.rec_num = t1.rec_num
ORDER BY nom, diff DESC