从同一表SQL获取不同列的最新和最早的值

问题描述

我有一张可选择日期的桌子。我希望能够应用WHERE语句返回一组，然后从一个列返回最早的日期，而从另一列返回最晚的日期。这是一个示例表：

ID  StartDate   EndDate     Person   Subject
1   01/03/2010  03/03/2010   Paul    Math
2   12/05/2010  22/05/2010   Steve   Science
3   04/03/2010  08/03/2010   Paul    English

所以我想返回Person ='Paul'的所有记录。但是返回类似（最早的）StartDate = 01/03/2010（来自记录ID 1）和（最新的）EndDate = 08/03/2010（来自记录ID 3）和Subject =来自最新的EndDate（来自记录3）的东西。我需要为保罗提供上述数据和Id的两行。需要所有与person ='Paul'

的行

预期结果：

ID  StartDate   EndDate Person  Subject
1   2010-03-01  2010-03-08  Paul    English
3   2010-03-01  2010-03-08  Paul    English

您是否在以下SQL查询中看到有关性能（数百万条记录）的任何问题：

 Select PT3.ID,PT4.EarliestStartDate AS StartDate,PT4.EndDate,PT4.Person,PT4.Subject from Data AS PT3 
Join( SELECT Top 1 with ties * FROM Data AS PT  
Join( SELECT PT1.Person as Person1,MIN(PT1.StartDate ) as EarliestStartDate FROM Data AS PT1 where Person ='Paul' group by PT1.Person)  AS PT2 ON  PT.Person =PT2.Person1WHERE Person  ='Paul'
  Order By Row_Number() over (Partition By PT.Person Order By PT.EndDate desc)) AS PT4 ON PT3.Person = PT4.Person

解决方法

这可能有效：

首先执行group by，以获取每个人所需的日期。
将分组移动到公用表表达式（cte）。
重新加入完整的数据集以获取所有行。
对于每个结果记录，请选择带有cross apply的相应主题。
可选：过滤所需人员（where子句）。

样本数据

create table Data
(
  ID int,StartDate date,EndDate date,Person nvarchar(10),Subject nvarchar(10)
);

insert into Data (ID,StartDate,EndDate,Person,Subject) values
(1,'2010-03-01','2010-03-03','Paul','Math'),(2,'2010-05-12','2010-05-22','Steve','Science'),(3,'2010-03-04','2010-03-08','English');

解决方案

with cte as
(
  select min(d.StartDate) as StartDate,max(d.EndDate) as EndDate,d.Person
  from Data d
  group by d.Person
)
select d.Id,c.StartDate,c.EndDate,c.Person,x.Subject
from cte c
join data d
  on d.Person = c.Person
cross apply ( select top 1 d2.Subject
              from Data d2
              where d2.Person = c.Person
                and d2.EndDate = c.EndDate ) x
where d.Person = 'Paul'
order by d.Person;

Fiddle来查看实际效果。

这应该为您提供正确的值。

DECLARE @values TABLE (
    Id INT,StartDate DATETIME2,EndDate DATETIME2,Person NVARCHAR(MAX),Subject NVARCHAR(MAX))

INSERT INTO @values VALUES 
(1,'2101-03-04','English')

;WITH sort AS (
    SELECT
        Person,MIN(StartDate) OVER(PARTITION BY Person) StartDate,MAX(EndDate) OVER(PARTITION BY Person) EndDate,Subject,ROW_NUMBER() OVER(PARTITION BY Person ORDER BY EndDate DESC) rownum
    FROM @values v)

SELECT
    Person,Subject
FROM sort
WHERE rownum = 1

它将为您提供以下结果：

Person  MinStartDate    MaxEndDate   Subject
Paul    2010-03-01      2010-03-08   English
Steve   2010-05-12      2010-05-22   Science

如果您只关心选择Pauls数据，这应该就足够了：

SELECT 
   Person,MIN(startDate) AS Earliest,MAX(EndDate) AS Latest,(
      SELECT TOP 1 Subject 
      FROM Data
      WHERE Person = 'Paul'
      ORDER BY EndDate DESC
   ) AS Subject
FROM Data WHERE Person = 'Paul'

如果您想选择所有人，这应该起作用：

SELECT 
   d1.Person,MIN(d1.startDate) AS Earliest,MAX(d1.EndDate) AS Latest,(
      SELECT TOP 1 d2.Subject 
      FROM Data d2
      WHERE d2.Person = d1.Person
      ORDER BY EndDate DESC
   ) AS Subject
FROM Data d1
GROUP BY d1.Person

编辑：您似乎希望对表中的每一行都有结果，而不是每个人只有一个，但是总要显示最早和最新的日期。这是你的追求吗？

SELECT 
   d1.Id,d1.Person,(
      SELECT MIN(d1.StartDate) AS Earliest
      FROM Data d2
      WHERE d2.Person = d1.Person
   ) AS Earliest,(
      SELECT MAX(d1.EndDate) AS Latest
      FROM Data d2
      WHERE d2.Person = d1.Person
   ) AS Latest,(
      SELECT TOP 1 d2.Subject 
      FROM Data d2
      WHERE d2.Person = d1.Person
      ORDER BY EndDate DESC
   ) AS Subject
FROM Data d1

group-by having having having self-join sql sql sql-server