使用 SubString 和 lpad/rpad 替換數字

Question

有沒有辦法用 Hive 中的 substring 替換數字。

scenario:

1. I need to check 0's from right to left and as soon as i find 0 before any digit then i need to replace all the trailing 0's by 9.
2. In output at-least 3  digit should be there before 9.
3. In Input if 2 or less digits are available in input then I need to skip some 0's and make sure that at-least 3 digits are there before 9.
4. If more than 3 digits are available before trailing 0's then only need to replace 0.No need to replace digits.

見下表

input     output

123000    123999
120000    120999
123400    123499
101010    101019

我嘗試使用以下查詢，它按預期工作。（ Hive 加入 CTE ）

with mytable as (
select '123000' as input 
union all 
select '120000' as input 
union all 
select '123400' as input 
union all 
select '101010' as input
) 
select input,lpad(concat(splitted[0], translate(splitted[1],'0','9')),6,0) as output 
from (
select input, split(regexp_replace(input,'(\\d{3,}?)(0+)$','$1|$2'),'\\|') splitted from mytable )s;

但是在我超過 500 行的實際查詢中，很難為單個列調整此邏輯（使用 CTE）。 所以想知道是否有任何方法可以僅使用 lpad/rpad 和 substring/length 來實現相同的目標，並且可以通過添加函數而不使用 CTE 查詢來實現。

so let say if length of digits before trailing 0's is less than 6 then can skip the substring 
from (input,1,6) and will replace the remaining 0's and if  length of digits before trailing 0's is  6 
or more then 6 then just keep digits as it is and replace remaining trailing 0's by 9.

請建議。

我的實際查詢看起來像。

with mytable as
(
select lpad(input,13,9) as output from mytable where code='00'
union
select output  from mytable where code='01'
)
select t1.*,m1.output from table1 t1 , mytable m1 where  
(t1.card='00' and substr(t1.low,1,13)<=m1.low and m1.output <= substr(t1.output,1,13) and m1.card='00' )
or
(t1.card='01' and substr(t1.low,1,16)<=m1.low and m1.output <=  substr(t1.output,1,16) and m1.card='01' )

我想為 2nd output 替換上述邏輯，其中 code=01 在聯合查詢中。

Answer 1

您可以使用單個查詢代替 UNION 來做同樣的事情：

而不是這個

select lpad(output,13,9) as output from mytable where code='00'
union
select output  from mytable where code='01'

用這個

select distinct
       case code when '00' then lpad(output,13,9)
                 when '01' then output  
        end output  
  from mytable 

如果需要過濾代碼， where code in ('00','01')

Answer 2

它現在工作了。我修改了我的查詢如下。

with mytable as
(
select lpad(input,13,9) as output from mytable where code='00'
union
select lpad(concat(split(regexp_replace('(\\d{6,}?)(0+)$','$1|$2'),'\\|') [0], 
translate(split(regexp_replace(input,'(\\d{6,}?)(0+)$','$1|$2'),'\\|')[1],'0','9')),16,0 )
output  from mytable where code='01'
)
select t1.*,m1.output from table1 t1 , mytable m1 where  
(t1.card='00' and substr(t1.low,1,13)<=m1.low 
 and m1.output <= substr(t1.output,1,13) and m1.card='00' )
 or
(t1.card='01' and substr(t1.low,1,16)<=m1.low 
and output <=  substr(output,1,16) and m1.card='01')