Question

我想从我的数据中排除重复项，除非副本是新分组/系列的一部分。

CREATE TABLE #MYTEMPTABLE
(CHILD VARCHAR(2), PARENT VARCHAR(2),YEAR INT)
INSERT INTO #MYTEMPTABLE (CHILD, PARENT,YEAR)
VALUES 
('1B','1A',2014),
('1A','1A',2014),
('2B','2A',2014),
('2A','2A',2014),
('3A','3A',2014),
('3B','3A',2014),
 also would have ('3B','3B',2014)
('3C','3B',2014),
('4A','4A',2014),
('4B','4A',2014),
 and ('4B','4B',2014)
('4C','4B',2014)

CREATE TABLE #MYTEMPTABLE2
(CHILD VARCHAR(2), PARENT VARCHAR(2),YEAR INT)

INSERT INTO #MYTEMPTABLE2  (CHILD, PARENT,YEAR)
VALUES 
('1A','1A',2015),
('1C', '1A',2015),
('2B','2A',2015),
('2A', '2A',2015),
('3A','3A',2015),
('3B','3A',2015),
  ('3B','3B',2014)
('3C','3B',2015),
('4A','4A',2015),
('4B','4A',2015),
  ('4B','4B',2015),
('4D','4B',2015)

我现在想要的是什么：
      1B，1A
      1A，1A
      2B，2A
      2A，2A
      3A，3A
      3B，3A
      3B，3B
      3C，3B
      4A，4A
      4B，4A
      4B，4B
      4C，4B
      1A，1A
      1C，1A
      4B，4B
      4D，4B

Answer 1

您可以将查询编写为：

--Step2: Get all data from first Table
SELECT PARENT,CHILD
FROM #MYTEMPTABLE
UNION ALL -- to allow valid duplicates
SELECT PARENT,CHILD
FROM #MYTEMPTABLE2 
WHERE PARENT IN (
-- Step1:Get parents which have a new Seq in second table
SELECT PARENT
FROM #MYTEMPTABLE2 T
WHERE NOT EXISTS ( SELECT * FROM #MYTEMPTABLE T1 
WHERE T.PARENT = T1.PARENT AND T.CHILD = T1.CHILD 
AND T.YEAR <> T1.YEAR)
)

Answer 2

这是一个艰难的。我不是百分之百肯定它是对的，但可能需要大学学位数学家和sql guru在一起做到这一点。它似乎根据您的要求正确显示。我简化并解释了下面的查询：

See SQL Fiddle

With MYTEMPTABLEHierarchy AS (
      Explode the entire Hierarchy for #MYTEMPTABLE. i.e. work out which parent and child relationship is on which hierarchical level
    ),

MYTEMPTABLEHierarchy2 AS (
      Explode the entire Hierarchy for #MYTEMPTABLE2. i.e. work out which parent and child relationship is on which hierarchical level
    ),

EqualHierarchy AS (
      Take the rows that are the same in both hierarchies.
    ),

DifferentHierarchy AS (
      Take the rows that are different in both hierarchies.
    ),

DuplicateAncestorHierarchy AS (
    Use the parent rows that have a different hierarchy and find all of their parents which become the duplicates.
)
Select * from EqualHierarchy
UNION ALL
Select * from DifferentHierarchy
UNION ALL
Select * from DuplicateAncestorHierarchy
Order by HLevel, PARENT, CHILD

SQL：比较父级层次结构，如果重复则排除

2 个答案: