如何使用子字符串查找最大字符

Question

这应该相对容易。我以索赔数据为生。我有以下假设的ClaimID：17060C048388A3

关于这一点，第14个字符空间中的A代表“调整”，下面的数字是已调整的时间数。因此，例如，此特定索赔已经调整了3次。不幸的是，ClaimID 17060C048388A1,17060C048388A2和17060C048388A3（所有三个调整显示）。在计算这些声明时，我如何仅选择最高调整以避免重复。并非所有声明都经过调整，因此它们并非都具有A并且不是全部15个字符。

所以我想我需要开始一个案例陈述如下：

SELECT DISTINCT TOP 10000 *,
    CASE WHEN CLAIMID LIKE '%A%' THEN ??????

如果ClaimID中有'A'，那么在ClaimID中'A'之后抓取下一个最高位的任何建议。因此，如果索赔有8个调整，我只想看到以A8结尾的那个。

先谢谢你们，格雷格

Answer 1

有了CTE：

with cte as(
  select
    claimid,
    case 
      when claimid like '%A%' then left(claimid, charindex('A', claimid) - 1) 
      else claimid
    end claim,
    case 
      when claimid like '%A%' then cast(right(claimid, len(claimid) - charindex('A', claimid)) as int)
      else 0
    end adjustment
  from tablename
)
select distinct top 10000 t.* 
from tablename t inner join (
  select 
    claim + 
    case max(adjustment) 
      when 0 then ''
      else 'A' + cast(max(adjustment) as varchar(10))
    end claimid 
  from cte
  group by claim
) g on g.claimid = t.claimid

见demo。 对于此示例数据：

> claimid        | value
> :------------- | :----
> 17060C048388A1 | a    
> 17060C048388A2 | b    
> 17060C04838811 | c    
> 17060C04838866 | d    
> 17060C04838855 | e    
> 17060C048388A3 | f    
> 17060C04838899 | g 

结果是：

> claimid        | value
> :------------- | :----
> 17060C04838811 | c    
> 17060C04838866 | d    
> 17060C04838855 | e    
> 17060C048388A3 | f    
> 17060C04838899 | g 

Answer 2

这将更简单;

select * from Table1 where
regexp_like("column1",substr("column1",1,instr("column1",'A'))||
            (select max(substr("column1",-1)) from Table1) );`

检查一下：http://sqlfiddle.com/#!4/cb16f/7

Answer 3

这对我有用（假设我没有注意到任何错误）。我从所有评论中抽取了例子并将它们放在一起，谢谢大家的回复。

SELECT DISTINCT SUBSTRING(CLAIMID, 1,13) + MAX(RIGHT(CLAIMID, 2)) AS 
MAX_CLAIM_ADJUSTMENT
FROM LA_TEMP.DBO.ITEM19
WHERE (SUBSTRING(CLAIMID, 14, 1) = 'A'
OR SUBSTRING(CLAIMID, 13, 1) = 'A')
GROUP BY left(CLAIMID, 13)
UNION
SELECT DISTINCT SUBSTRING(CLAIMID, 1,14) + MAX(RIGHT(CLAIMID, 1)) AS 
MAX_CLAIM_ADJUSTMENT
FROM LA_TEMP.DBO.ITEM19
WHERE (SUBSTRING(CLAIMID, 15, 1) = 'A'
OR SUBSTRING(CLAIMID, 14, 1) = 'A')
GROUP BY left(CLAIMID, 14)
ORDER BY MAX_CLAIM_ADJUSTMENT

我注意到A可以是第13或第14个字符。

Answer 4

我再次查看了您的问题并创建了以下解决方案：

SELECT base, MAX(adjustment) AS max_adjustment, base + IIF(MAX(adjustment) > 0, 'A', '') + MAX(adjustment) max_adjusted_claim
FROM (
  SELECT 
    IIF(PATINDEX('%A%', ClaimID) = 0, ClaimID, LEFT(ClaimID, PATINDEX('%A%', ClaimID)-1)) AS base, 
    REPLACE(REVERSE(LEFT(REVERSE(ClaimID), PATINDEX('%A%', REVERSE(ClaimID)))), 'A', '') AS adjustment
  FROM table_name
)t GROUP BY base

demo on dbfiddle.uk