将记录拆分为具有上限和下限的块？

Question

我有一个 table (Db2) 有几百万条记录。过去table中删除了很多记录，因此ROW_NUMBER != PRIMARY KEY。例如如果主键从1开始生成，那么id为1000的记录按主键排序可能是第756条记录。

我有一个需求，需要用下面的方式展示数据。例如

Chunk   LowerLimitId   UpperLimitId

Chunk1  1               1098       // n = 1000 records between 1 and 1098

Chunk2  1099            2225       // n = 1000 records between 1099 and 2225

Chunk3  2226            3352       // n = 1000 records between 2226 and 3352

等等....

基本上LowerLimit 和Upper Limit 之间的总记录必须是1000。

即

"select count(*) from Table where id between 1 and 1098" = 1000

"select count(*) from Table where id between 1099 and 2225" = 1000

"select count(*) from Table where id between 2226 and 3352" = 1000 and so on

是否有任何 db2 子句或函数可以实现此目的？我不是要完整的查询。只是实现这一目标的方向。

Answer 1

您可以借助 analytic 函数计算行号 row_number 并将数据拆分成块。

示例查询

with t(id) as (
  select *
  from ( values (1), (3), (5), (6), (20), (13), (25), (23) )
)
, a as (
  select
    t.id
    , trunc((row_number() over(order by id asc) - 1) / 3) as grp
  from t
)
select
  concat('Chunk', grp + 1) as chunk
  , min(id) as LowerLimitId
  , max(id) as UpperLimitId
from a
group by grp

 CHUNK | LOWERLIMITID | UPPERLIMITID
-----: | -----------: | -----------:
Chunk1 |            1 |            5
Chunk2 |            6 |           20
Chunk3 |           23 |           25

db<>fiddle here

Answer 2

@astentx 的回答已经很好了。我只是对其进行了改进以适应实际需求。

with t(id) as (
  select *
  from ( values (1), (3), (5), (6), (20), (13), (25), (23) )
)
, a as (
  select 
    t.id,
    trunc((row_number() over(order by id asc) - 1) / 3) as grp
  from t
)
select
    a.grp,
    MIN(a.id) as startId,
    MAX(a.id) as endId
from a
group by a.grp

Sql Fiddle

Answer 3

您可以使用 Windows 函数来执行此操作。为每一行获取 row_number，然后您将能够 select 块。 select离子可以动态或静态完成。

Row_Number() over order by primaryKeyColumn ....

如果您可以 post 带有数据的 table 脚本，我可以 post 一个完整的答案。
@bluelurker

将记录拆分为具有上限和下限的块？

Split records in chunks with upper and lower limit?

sql

db2