Question

我正在尝试了解表格运算符APPLY。

以下是示例：

CREATE TABLE #y ( Name char(8), hoursWorked int);
GO
INSERT INTO #y VALUES ('jim',4);
INSERT INTO #y VALUES ('michael',40);
INSERT INTO #y VALUES ('raj',1000);
INSERT INTO #y VALUES ('jason',7);
INSERT INTO #y VALUES ('tim',50);
GO

CREATE TABLE #x ( Name char(8),game char(8), NumBets int);
GO
INSERT INTO #x VALUES ('jim','chess',4);
INSERT INTO #x VALUES ('jim','BG',10);
INSERT INTO #x VALUES ('jim','draughts',100);
INSERT INTO #x VALUES ('jim','football',5);
INSERT INTO #x VALUES ('michael','chess',40);
INSERT INTO #x VALUES ('michael','BG',7);
INSERT INTO #x VALUES ('michael','draughts',65);
INSERT INTO #x VALUES ('michael','football',50);
INSERT INTO #x VALUES ('raj','chess',400);
INSERT INTO #x VALUES ('raj','BG',70);
INSERT INTO #x VALUES ('raj','draughts',650);
INSERT INTO #x VALUES ('tim','draughts',60000);
GO

SELECT  y.Name, 
        y.hoursWorked,
        x.game,
        x.NumBets
FROM    #y y
        OUTER APPLY 
          (
          SELECT TOP 2 *
          FROM   #x
          WHERE  Name = y.Name 
          ORDER BY NumBets
        ) x
ORDER BY  y.Name,
        x.NumBets DESC;

我的主要障碍是在使用APPLY时理解所以我想知道使用standard sql中实现的sql-server 2005得到与上面相同的结果有多难？ APPLY是否会使查询更短或更易读？如果此示例显示使用APPLY没有大的优势，那么使用APPLY有什么优势的明显示例是什么？

Answer 1

首先 - 使用apply，您可以调用表值函数，其中参数值取自您查询的表格，如下所示：

select
    t1.col3, -- column from table
    f1.col1  -- column from function
from table1 as t1
    left outer join table2 as t2 on t2.col1 = t1.col1
    outer apply dbo.function1(t1.col1, t2.col2) as f1

或粉碎xml列

select
    t1.col3,
    t.c.value('@value', 'int') as value
from table1 as t1
    -- table1.col1 is xml iike <Data @Value="...">...</Data>
    outer apply t1.col1.nodes('Data') as t(c)

从我的经验来看，apply在您必须进行预先计算时非常有用：

select
    t1.col3,
    a1.col1,  --calculated value
    a2.col1   -- another calculated value, first one was used
from table1 as t1
    outer apply (select t1.col1 * 5 as col1) as a1
    outer apply (select a1.col1 - 4 as col1) as a2

使用apply的另一个示例是 unpivot 操作：

select
    t1.col1, c.name, c.value
from table1 as t1
    outer apply (
        select 'col1', t1.col1 union all
        select 'col2', t1.col2
    ) as c(name, value)

最后，这里的您的查询是根据SQL 2005实现的，而不使用 apply ：

;with cte as (
    select
        y.Name, 
        y.hoursWorked,
        x.game,
        x.NumBets,
        row_number() over(partition by x.Name order by x.NumBets) as row_num
    from y
        left outer join x on x.Name = y.Name
)
select Name, hoursWorked, game, NumBets
from cte
where row_num <= 2
order by Name, NumBets desc

请参阅sql fiddle示例

何时使用表运算符APPLY

1 个答案: