使用 INSERT-SELECT 插入新记录时出现问题

Question

尝试仅插入新记录。基于新的 UserID 或现有的 UserID 和新的 LicenseState。使我的 WHERE 子句正确时遇到问题。

示例代码如下：

CREATE TABLE #SourceData
(
    UserID  varchar(30),
    LicenseState  varchar(2),
    LicenseNumber  varchar(10),
    LicenseYear  int
)

CREATE TABLE #Person
(
  PersonID  int,
  LastName  varchar(30),
  FirstName  varchar(30),
  DomainAccountName  varchar(50)
)

CREATE TABLE #License
(
  LicenseID  int IDENTITY(1,1) NOT NULL,
  PersonID  int,
  LicenseState  varchar(2),
  LicenseNumber  varchar(10),
  LicenseYear  int
)

-- Load #SourceData
INSERT INTO #SourceData (UserID, LicenseState, LicenseNumber, LicenseYear)
VALUES ('gwashington', 'CA', '11111', 1999);

INSERT INTO #SourceData (UserID, LicenseState, LicenseNumber, LicenseYear)
VALUES ('gwashington', 'TX', '9999', 2001);

INSERT INTO #SourceData (UserID, LicenseState, LicenseNumber, LicenseYear)
VALUES ('jadams', 'CA', '22222', 2000);

INSERT INTO #SourceData (UserID, LicenseState, LicenseNumber, LicenseYear)
VALUES ('tjefferson', 'TX', '44444', 2002);

INSERT INTO #SourceData (UserID, LicenseState, LicenseNumber, LicenseYear)
VALUES ('alincoln', 'IL', '55555', 2012);

-- Load #Person
INSERT INTO #Person (PersonID, LastName, FirstName, DomainAccountName)
VALUES (1, 'Washington', 'George', 'gwashington');

INSERT INTO #Person (PersonID, LastName, FirstName, DomainAccountName)
VALUES (2, 'Adams', 'John', 'jadams');

INSERT INTO #Person (PersonID, LastName, FirstName, DomainAccountName)
VALUES (3, 'Jefferson', 'Thomas', 'tjefferson');

INSERT INTO #Person (PersonID, LastName, FirstName, DomainAccountName)
VALUES (4, 'Lincoln', 'Abraham', 'alincoln');

-- Load #License
INSERT INTO #License (PersonID, LicenseState, LicenseNumber, LicenseYear)
VALUES (1, 'CA', '11111', 1999);

INSERT INTO #License (PersonID, LicenseState, LicenseNumber, LicenseYear)
VALUES (2, 'CA', '22222', 2000);

INSERT INTO #License (PersonID, LicenseState, LicenseNumber, LicenseYear)
VALUES (3, 'TX', '44444', 2002);

INSERT INTO #License (PersonID, LicenseState, LicenseNumber, LicenseYear)
VALUES (3, 'OH', '33333', 2008);

我尝试了以下方法：

INSERT INTO #License (PersonID, LicenseState, LicenseNumber, LicenseYear)
SELECT
  person.PersonID,
  srce.LicenseState,
  srce.LicenseNumber,
  srce.LicenseYear
FROM #Person person
LEFT JOIN #SourceData srce
ON person.DomainAccountName = srce.UserID
LEFT JOIN #License lic
ON person.PersonID = lic.PersonID
WHERE lic.PersonID IS NULL
  OR (lic.PersonID IS NOT NULL AND lic.LicenseState <> srce.LicenseState)

或

WHERE lic.PersonID IS NULL
  OR (lic.PersonID IS NOT NULL AND lic.LicenseState NOT IN ('CA','OH','DC','WA','IL','NY','DE','CO','HI','NJ','PA','VA','FL','TX','CT'))

这是#SourceData 中的数据：

UserID      LicenseState    LicenseNumber   LicenseYear
gwashington CA              11111           1999
gwashington TX              9999            2001
jadams      CA              22222           2000
tjefferson  TX              44444           2002
alincoln    IL              55555           2012

这是#Person 中的数据：

PersonID    LastName    FirstName   DomainAccountName
1           Washington  George      gwashington
2           Adams       John        jadams
3           Jefferson   Thomas      tjefferson
4           Lincoln     Abraham     alincoln

这是插入前#License 中的数据：

LicenseID   PersonID    LicenseState    LicenseNumber   LicenseYear
1           1           CA              11111           1999
2           2           CA              22222           2000
3           3           TX              44444           2002
4           3           OH              33333           2008

这是所需的输出：

LicenseID   PersonID    LicenseState    LicenseNumber   LicenseYear
1           1           CA              11111           1999
2           2           CA              22222           2000
3           3           TX              44444           2002
4           3           OH              33333           2008
5           1           TX              9999            2001
6           4           IL              55555           2012

因此，应该添加 gwashington 和 TX 以及 alincoln 和 IL。我可以添加 alincoln 和 IL，但 gwashington 和 TX 没有。此外，INSERT-SELECT 会继续插入记录，尽管它们已经被插入。它应该只包含尚未从 #SourceData 加载的新记录到 #License。

感谢您的帮助。

Answer 1

看看这是否适合你：

INSERT INTO #License    
SELECT p.PersonID,    
       s.LicenseState,  
       s.LicenseNumber,   
       s.LicenseYear    
FROM   #SourceData s    
       JOIN #Person p    
         ON s.UserID = p.DomainAccountName    
       LEFT JOIN #License l    
              ON l.PersonID = p.PersonID    
                 AND l.LicenseState = s.LicenseState    
WHERE  l.PersonID IS NULL     

Answer 2

有更好的方法可以做到这一点，但在无法测试任何 atm 的情况下，解决方法可能有助于小型/中型表大小。

如果我正确解释数据，以下应该给出结果表：

WITH BASE AS ( 
SELECT
  CONCAT(lic.PersonID, lic.LicenseState) AS IDxLicense 
  FROM #License lic
)

SELECT
  person.PersonID,
  srce.LicenseState,
  srce.LicenseNumber,
  srce.LicenseYear
FROM #Person person
LEFT JOIN #SourceData srce
ON person.DomainAccountName = srce.UserID
LEFT JOIN #License lic
ON person.PersonID = lic.PersonID 
WHERE (CONCAT(person.PersonID, srce.LicenseState) NOT IN (SELECT IDxLicense FROM BASE))