Question

我需要转换以下数据：

EmployeeId     DataCategory     DataValue     StartDate   
----------     ------------     ----------   ---------- 
    2          'OfficeCode'         121        '03-01-12'
    2          'ManagerName'      'Steven'     '02-04-12'
    2          'State'              'MA'       '04-05-12'
    5          'OfficeCode'         133        '04-01-12'
    5          'ManagerName'      'Marcus'     '05-04-12'
    5          'State'              'WA'       '01-05-12'
    6          'ManagerName'      'Steven'     '07-04-12'
    6          'State'              'RI'       '06-05-12'
    7          'State'              'CA'       '08-08-12'

为：

EmployeeId      OfficeCode      ManagerName    State      OfficeCodeStartDate
----------     ------------     ----------   ----------   -------------------
    2               121          'Steven'       'MA'         '03-01-12'
    5               133          'Marcus'       'WA'         '04-01-12'
    6               null         'Steven'       'RI'           null
    7               null          null          'CA'           null

我可以这样转动：

select EmployeeId, OfficeCode, ManagerName, State
from
(
  select EmployeeId, DataCategory, DataValue
  from emp
) src
pivot
(
  max(DataValue)
  for DataCategory in (OfficeCode, ManagerName, State)
) piv

但是，我还需要DataCategory OfficeCode的StartDate（忽略任何其他类别的开始日期）。你能帮我用pivot / unpivot来达到预期的效果吗？除非有必要，否则我试图避免加入/加入联盟。

Answer 1

三种方式：一种使用UNION，一种使用JOIN，一种不使用PIVOT。选择你最喜欢的那个（根据查询计划，最后一个是最快的）。

/*
CREATE TABLE emp (
    EmployeeId INT,
    DataCategory VARCHAR(50),
    PRIMARY KEY (EmployeeID, DataCategory),
    DataValue VARCHAR(50) NOT NULL,
    StartDate VARCHAR(10) NOT NULL
)

INSERT INTO emp VALUES
('2','OfficeCode','121','03-01-12'),
('2','ManagerName','Steven','02-04-12'),
('2','State','MA','04-05-12'),
('5','OfficeCode','133','04-01-12'),
('5','ManagerName','Marcus','05-04-12'),
('5','State','WA','01-05-12'),
('6','ManagerName','Steven','07-04-12'),
('6','State','RI','06-05-12'),
('7','State','CA','08-08-12')
*/

select EmployeeId, OfficeCode, ManagerName, State, OfficeCodeStartDate
from
(
  select EmployeeId, DataCategory, DataValue
  from emp
  union all
  select EmployeeId, 'OfficeCodeStartDate' AS DataCategory, StartDate AS DataValue
  from emp
  WHERE DataCategory='OfficeCode'
) src
pivot
(
  max(DataValue)
  for DataCategory in (OfficeCode, ManagerName, State, OfficeCodeStartDate)
) piv

select piv.EmployeeId, OfficeCode, ManagerName, State, emp.StartDate AS OfficeCodeStartDate
from
(
  select EmployeeId, DataCategory, DataValue
  from emp
) src
pivot
(
  max(DataValue)
  for DataCategory in (OfficeCode, ManagerName, State)
) piv
LEFT JOIN emp ON emp.EmployeeId=piv.EmployeeId AND emp.DataCategory='OfficeCode'

SELECT EmployeeId,
    MIN(CASE WHEN DataCategory='OfficeCode' THEN DataValue END) AS OfficeCode,
    MIN(CASE WHEN DataCategory='ManagerName' THEN DataValue END) AS ManagerName,
    MIN(CASE WHEN DataCategory='State' THEN DataValue END) AS State,
    MIN(CASE WHEN DataCategory='OfficeCode' THEN StartDate END) AS OfficeCodeStartDate
FROM emp
GROUP BY EmployeeId

多列上的Pivot / Unpivot表

1 个答案: