Sql 从2个表中查找每条记录的唯一最接近的日期匹配

您可以使用一个相关子查询，在那里您可以 Select 最近的日期.

对于这种动态，您需要至少处理两次数据. 其基本思想是使用rownumber来确定两个日期的位置， Select 两个或更多的叉，然后简单地 Select 行中的下一个，这取决于它的行数.

这意味着表2中有足够的数据来填充这一列，否则您需要进行更多的编程，以确定在最后一个日期之后的下一个日期.

CREATE TABLE Table1 (
  ID INTEGER,
  JoiningDt DATETIME,
  DocNum VARCHAR(4)
);

INSERT INTO Table1
  (ID, JoiningDt, DocNum)
VALUES
  ('1', '05-Apr-2024', 'A123'),
  ('2', '06-Apr-2024', 'A123'),
  ('3', '04-Apr-2024', 'B123');

CREATE TABLE Table2 (
  DocNum VARCHAR(4),
  ClosestDt DATETIME
);

INSERT INTO Table2
  (DocNum, ClosestDt)
VALUES
  ('A123', '03-Apr-2024'),
  ('A123', '04-Apr-2024'),
  ('A123', '07-Apr-2024'),
  ('A123', '08-Apr-2024'),
  ('B123', '02-Apr-2024'),
  ('B123', '05-Apr-2024');

9 rows affected

WITH CTE1 As (
SELECT t1.ID, t1.JoiningDt, t1.DocNum,
  (SELECT TOP 1 ClosestDt FROM Table2 
  WHERE DocNum = t1.DocNum AND ClosestDt > t1.JoiningDt ORDER BY ClosestDt  ASC   ) ClosestDt
  , ROW_NUMBER() OVER(PARTITION BY DocNum,   (SELECT TOP 1 ClosestDt FROM Table2 
  WHERE DocNum = t1.DocNum AND ClosestDt > t1.JoiningDt ORDER BY ClosestDt  ASC   ) ORDER BY ID) rn
FROM Table1 t1
 )
SELECT ID, JoiningDt, DocNum, 
  CASE WHEN rn = 1 then ClosestDt ELSE
  (SELECT ClosestDt FROM Table2 
  WHERE DocNum = c1.DocNum AND ClosestDt > c1.JoiningDt ORDER BY ClosestDt  ASC   
  OFFSET c1.rn -1  ROWS FETCH NEXT 1 ROWS ONLY) END 
  FROM CTE1 c1

fiddle

另一种使用两个CTE的方法，这样子 Select 就不会运行两次，但它就像第一次只重写了

WITH CTE1 As (
SELECT t1.ID, t1.JoiningDt, t1.DocNum,
  (SELECT TOP 1 ClosestDt FROM Table2 
  WHERE DocNum = t1.DocNum AND ClosestDt > t1.JoiningDt ORDER BY ClosestDt  ASC   ) ClosestDt
FROM Table1 t1
 ), CTE2 AS (
  SELECT
       ID, JoiningDt, DocNum, ClosestDt
    , ROW_NUMBER() OVER(PARTITION BY DocNum,   ClosestDt ORDER BY ID) rn
  FROM CTE1
  )
SELECT ID, JoiningDt, DocNum, 
  CASE WHEN rn = 1 then ClosestDt ELSE
  (SELECT ClosestDt FROM Table2 
  WHERE DocNum = c1.DocNum AND ClosestDt > c1.JoiningDt ORDER BY ClosestDt  ASC   
  OFFSET c1.rn -1  ROWS FETCH NEXT 1 ROWS ONLY) END 
  FROM CTE2 c1

fiddle

最好在编程语言中使用cursorm是一种缓慢的方法，因为它是一种缓慢的方法，而且对于太多的行，这将花费很长的时间，所以您应该限制表1的行数，

DECLARE @ID int
DECLARE @JoiningDt DATETIME
DECLARE @DocNum VARCHAR(4)

DECLARE @MyTableVar TABLE (
    ID INT NOT NULL,
    JoiningDt datetime,
    DocNum VARCHAR(4),
    NearestDate DATETIME);


DECLARE db_cursor CURSOR FOR 
SELECT ID, JoiningDt, DocNum FROM Table1

OPEN db_cursor  
FETCH NEXT FROM db_cursor INTO @ID, @JoiningDt, @DocNum

WHILE @@FETCH_STATUS = 0  
BEGIN  
INSERT INTO @MyTableVar
SELECT TOP 1 @ID, @JoiningDt, @DocNum,ClosestDt   FROM table2
WHERE DocNum = @DocNum AND ClosestDt > @JoiningDt     
  AND NOT EXISTS (SELECT 1 FROM  @MyTableVar WHERE NearestDate  = ClosestDt)
  ORDER BY ClosestDt ASC; 

      FETCH NEXT FROM db_cursor INTO @ID, @JoiningDt, @DocNum
END 

CLOSE db_cursor  
DEALLOCATE db_cursor
SELECT * FROM @MyTableVar ORDER BY ID 

fiddle