如何仅在没有子查询的 UNION ALL 中按一个查询排序

Question

我正在使用 UNION ALL 从不相关的数据创建一个列表，以便在 SQL Server 2008 中保存到 Excel。只有一个 SELECT 语句将具有 ORDER BY，这就是我的问题出现的地方以下错误信息

'The multi-part identifier "p.Category" could not be bound.'

这是一个简化的工作代码示例，我更喜欢下面的第二个，但在取消注释第一个 SELECT 查询时会中断：

CREATE TABLE #People(Category varchar(50), [Name] varchar(50));

INSERT INTO #People VALUES(1, 'Steve');
INSERT INTO #People VALUES(99, 'Anna');
INSERT INTO #People VALUES(2, 'Sarah'); 

-- Unrelated data SELECT query
--SELECT 'Best Employee' AS Matrix
--     , 'Alan' AS [Name]
--UNION ALL

SELECT CASE
           WHEN p.Category = 1
           THEN 'Very Active'
           WHEN p.Category = 2
           THEN 'Active '
           ELSE 'Departed'
       END AS Matrix
     , p.name AS [Name]
FROM #People p
ORDER BY p.Category;

DROP TABLE #People;

经过一些研究，我发现我可以通过以下方式重写查询，这将在这个小示例中为我提供所需的结果，但在我的实际案例中我认为这不是理想的结果。有没有办法给我正确的结果，而无需像下面的工作示例中那样使用子查询？

CREATE TABLE #People(Category varchar(50), [Name]  varchar(50));

INSERT INTO #People VALUES(1, 'Steve');
INSERT INTO #People VALUES(99, 'Anna');
INSERT INTO #People VALUES(2, 'Sarah'); 

SELECT *
FROM
(
    -- Unrelated data SELECT query
    SELECT 'Best Employee' AS Matrix
         , 'Alan' AS [Name]
         , '' AS Category

    UNION ALL

    SELECT CASE
               WHEN p.Category = 1
               THEN 'Very Active'
               WHEN p.Category = 2
               THEN 'Active '
               ELSE 'Departed'
           END Matrix
         , p.name AS [Name]
         , p.Category Category
    FROM #People p
) x
ORDER BY x.Category;

DROP TABLE #People;

Answer 1

ORDER BY 不仅仅适用于一个SELECT，它适用于整个结果集，适用于所有行都经过UNIONed 之后的整个结果。因此，您需要在UNION ALL 的每个SELECT 中包含将用于排序的列。

例如，以下是错误的语法，因为ORDER BY 必须在UNION 的最后一个SELECT 之后。

SELECT CASE
           WHEN p.Category = 1
           THEN 'Very Active'
           WHEN p.Category = 2
           THEN 'Active '
           ELSE 'Departed'
       END AS Matrix
     , p.name AS [Name]
FROM #People p
ORDER BY p.Category

UNION ALL

SELECT 'Best Employee' AS Matrix
     , 'Alan' AS [Name]

问题中的第二个示例可以用更简单的形式编写，没有子查询：

SELECT CASE
           WHEN p.Category = 1
           THEN 'Very Active'
           WHEN p.Category = 2
           THEN 'Active '
           ELSE 'Departed'
       END AS Matrix
     , p.name AS [Name]
     , p.Category
FROM #People p

UNION ALL

SELECT 'Best Employee' AS Matrix
     , 'Alan' AS [Name]
     , NULL AS Category  --- or '' AS Category

ORDER BY Category;

在任何情况下，都必须在UNION 的每个SELECT 语句中添加Category 列。我不知道你怎么能避免它。

Answer 2

如果您的问题是您的排序导致 2 个表混淆，您可以添加一列，以便始终首先显示第一个选择结果集。

CREATE TABLE #People(Category varchar(50), [Name]  varchar(50));

INSERT INTO #People VALUES(1, 'Steve');
INSERT INTO #People VALUES(99, 'Anna');
INSERT INTO #People VALUES(2, 'Sarah'); 

SELECT Matrix, Name, Category
FROM
(
    -- Unrelated data SELECT query
    SELECT 'Best Employee' AS Matrix
         , 'Alan' AS [Name]
         , '' AS Category
         , 1 as Ord

UNION ALL

    SELECT CASE
               WHEN p.Category = 1
               THEN 'Very Active'
               WHEN p.Category = 2
               THEN 'Active '
               ELSE 'Departed'
           END Matrix
         , p.name AS [Name]
         , p.Category Category,
         2 as Ord
    FROM #People p
) x
ORDER BY x.Ord --,Additional Order by columns