日期成列的行 - SQL Server 2008

Question

我需要一些关于 PIVOT 表或其他东西的帮助才能以我需要的方式获得结果。

我有一张这样的桌子。

+---------------+--------------+----------+-----------+---------------+-----------------------------------------+-----------------+---------------+------------------+----------+-------+------------+--------------+------------+-------------------+----------------+----------+-------------------+
|     Client_Id |  Project_Id  | Hotel_Id | Room_Type | Room_Category |             Allotment_Date              | Number_Of_Rooms | Number_Booked | Number_Available | Overbook | Price | Dep_Amount | Full_Payment | Admin_Only | HotelAllotment_Id | Price_Excl_VAT | VAT_Code | Charge_Dep_Amount |
+---------------+--------------+----------+-----------+---------------+-----------------------------------------+-----------------+---------------+------------------+----------+-------+------------+--------------+------------+-------------------+----------------+----------+-------------------+
|     DEFAULT   |     TEMPLATE |     2423 |       276 |               |                     2010-12-05 00:00:00 |           99999 |             1 |            99998 |        0 | 33000 |          0 |            1 |          0 |            279611 |              0 |          |                 0 |
|     DEFAULT   |     TEMPLATE |     2423 |       276 |               |                     2010-12-06 00:00:00 |           99999 |             1 |            99998 |        0 | 33000 |          0 |            1 |          0 |            279612 |              0 |          |                 0 |
|     DEFAULT   |     TEMPLATE |     2423 |       276 |               |                     2010-12-07 00:00:00 |           99999 |             1 |            99998 |        0 | 33000 |          0 |            1 |          0 |            279613 |              0 |          |                 0 |
|     DEFAULT   |     TEMPLATE |     2424 |       276 |               |                     2010-12-05 00:00:00 |           99999 |             1 |            99998 |        0 | 22000 |      22000 |            0 |          0 |            279590 |              0 |          |                 0 |
|     DEFAULT   |     TEMPLATE |     2424 |       276 |               |                     2010-12-06 00:00:00 |           99999 |             1 |            99998 |        0 | 22000 |      22000 |            0 |          0 |            279591 |              0 |          |                 0 |
|     DEFAULT   |     TEMPLATE |     2424 |       276 |               |                     2010-12-07 00:00:00 |           99999 |             1 |            99998 |        0 | 22000 |      22000 |            0 |          0 |            279592 |              0 |          |                 0 |
+---------------+--------------+----------+-----------+---------------+-----------------------------------------+-----------------+---------------+------------------+----------+-------+------------+--------------+------------+-------------------+----------------+----------+-------------------+

我需要这样显示数据，以获取列中的日期和每天预订的数量。

+---------------+--------------+----------+-----------+---------------+------------+------------+--------------+
|     Client_Id |  Project_Id  | Hotel_Id | Room_Type | Room_Category | 2010-12-05 | 2010-12-06 | 2010-12-07   |
+---------------+--------------+----------+-----------+---------------+------------+------------+--------------+
|     DEFAULT   |     TEMPLATE |     2423 |       276 |               |          1 |          1 |            1 |
|     DEFAULT   |     TEMPLATE |     2424 |       276 |               |          1 |          1 |            1 |
+---------------+--------------+----------+-----------+---------------+------------+------------+--------------+

我需要将其按 Hotel_Id、Room_Type 和 Room_Category（如果有）分组

我需要它是动态的，因为日期可能会改变。

我尝试使用简单的数据透视表，但没有成功。

任何帮助都会很棒。

Answer 1

编辑：关于这个“解决方案”的进一步讨论表明，谨慎的做法是进一步详细说明为什么这不应该被用作可靠的代码。除了像SELECT * 这样明显的 NEVER EVER 之外，沿着这条路走只会超过编译时间限制。假设我们用以下 CTE 填充了 @t_Date 表变量（而且方式更好）：

    -- Thanks @AaronBertrand!
    WITH cte_Date(DateVal) AS (
        SELECT TOP (10000) DATEADD(DAY, ROW_NUMBER() 
                   OVER (ORDER BY s1.object_id), '19991231')
        FROM sys.all_objects AS s1
        CROSS JOIN sys.all_objects AS s2
    )
    INSERT INTO @t_Date ( DateVal )
    SELECT  DateVal
    FROM    cte_Date;

不可避免地，我们会在某个时候遇到如下消息：

消息 8632，第 17 级，状态 2，第 2 行

内部错误：已达到表达式服务限制。请在您的查询中寻找可能复杂的表达式，并尝试简化它们。

请注意，即使是错误消息也可以总结为，正如 @bluefeet 和 @Lamak 或多或少所说的那样，“不要在数据层这样做。”

所以，正如帖子所说：

@bluefeet 和@Lamak 是 StackOverflow 上两个最优秀的数据库人员，因此您需要注意他们的话。如果你真的需要测试这个，使用@bluefeet 的答案，因为你收到一个编译时错误，你可以从你的查询中解脱出来，并通过以下示例解决这个问题。为了任何可能需要接手您的工作职责的人，请不要将这样的代码移动到生产环境中。我发布这个 hack-fest 解决方法只是为了让您可以测试您正在研究的任何理论，并通过适当的步骤跟进以使您的数据层更适合部署。

IF NOT EXISTS ( SELECT  1
                FROM    sys.objects
                WHERE   name = 'yourtable'
                    AND type = 'U' )
BEGIN
    --DROP TABLE dbo.yourtable;
    CREATE TABLE dbo.yourtable
    (    
        [Client_Id] varchar(7), 
        [Project_Id] varchar(8), 
        [Hotel_Id] int, 
        [Room_Type] int, 
        [Room_Category] int, 
        [Allotment_Date] datetime, 
        [Number_Of_Rooms] int, 
        [Number_Booked] int, 
        [Number_Available] int, 
        [Overbook] int, 
        [Price] int, 
        [Dep_Amount] int, 
        [Full_Payment] int, 
        [Admin_Only] int, 
        [HotelAllotment_Id] int, 
        [Price_Excl_VAT] int, 
        [VAT_Code] int, 
        [Charge_Dep_Amount] INT
    );

    DECLARE @t_Date    TABLE
    (
        DateVal        DATE
    );

    WITH cte_Date AS (
        SELECT  DateVal = CAST( GETDATE() AS DATE )
        UNION ALL
        SELECT  DateVal = DATEADD( DAY, -1, DateVal )
        FROM    cte_Date
        WHERE   DateVal > '2002-01-01'  
    )
    INSERT INTO @t_Date ( DateVal )
    SELECT  DateVal
    FROM    cte_Date
    OPTION ( MAXRECURSION 5000 );

    INSERT INTO dbo.yourtable( [Client_Id], [Project_Id], [Hotel_Id], [Room_Type], [Room_Category], [Allotment_Date], 
        [Number_Of_Rooms], [Number_Booked], [Number_Available], [Overbook], [Price], [Dep_Amount], [Full_Payment], 
        [Admin_Only], [HotelAllotment_Id], [Price_Excl_VAT], [VAT_Code], [Charge_Dep_Amount] )
    SELECT  'DEFAULT', 'TEMPLATE', 2423, 276, NULL, DateVal, 99999, 1, 99998, 0, 33000, 0, 1, 0, 279611, 0, NULL, 0
    FROM    @t_Date;
END;
GO

SELECT  COUNT( DISTINCT Allotment_date )
FROM    dbo.yourtable;

DECLARE @cols AS NVARCHAR(MAX),
        @query  AS NVARCHAR(MAX)

select @cols = STUFF((SELECT ',' + QUOTENAME(convert(varchar(10), allotment_date, 120)) 
                    from yourtable
                    group by allotment_date
                    order by allotment_date
            FOR XML PATH(''), TYPE
            ).value('.', 'NVARCHAR(MAX)') 
        ,1,1,'')

set @query = '
            SELECT  *
            FROM (    SELECT client_id, project_id, hotel_id,
                        room_type, room_category,' + @cols + ' 
                    from 
                    (
                      select client_id, project_id, hotel_id,
                        room_type, room_category,
                        allotment_date, number_booked
                      from yourtable
                    ) x
                    pivot 
                    (
                        sum(number_booked)
                        for allotment_date in (' + @cols + ')
                    ) p ) a;'

execute sp_executesql @query;
GO

Answer 2

在编写动态版本之前，您可以使用PIVOT 函数获取结果我建议先编写静态版本。

如果您的值数量有限，那么您可以将所有 dates 硬编码为列标题：

select client_id, project_id, hotel_id,
  room_type, room_category,
  [2010-12-05], [2010-12-06], [2010-12-07]
from 
(
  select client_id, project_id, hotel_id,
    room_type, room_category,
    allotment_date, number_booked
  from yourtable
) d
pivot
(
  sum(number_booked)
  for allotment_date in ([2010-12-05], [2010-12-06], [2010-12-07])
) p;

见SQL Fiddle with Demo。但是，如果您的值未知，那么您将需要创建一个要使用动态 SQL 执行的 sql 字符串。

DECLARE @cols AS NVARCHAR(MAX),
    @query  AS NVARCHAR(MAX)

select @cols = STUFF((SELECT ',' + QUOTENAME(convert(varchar(10), allotment_date, 120)) 
                    from yourtable
                    group by allotment_date
                    order by allotment_date
            FOR XML PATH(''), TYPE
            ).value('.', 'NVARCHAR(MAX)') 
        ,1,1,'')

set @query = 'SELECT client_id, project_id, hotel_id,
                room_type, room_category,' + @cols + ' 
            from 
            (
              select client_id, project_id, hotel_id,
                room_type, room_category,
                allotment_date, number_booked
              from yourtable
            ) x
            pivot 
            (
                sum(number_booked)
                for allotment_date in (' + @cols + ')
            ) p '

execute sp_executesql @query;

见SQL Fiddle with Demo。这些给出了结果：

| CLIENT_ID | PROJECT_ID | HOTEL_ID | ROOM_TYPE | ROOM_CATEGORY | 2010-12-05 | 2010-12-06 | 2010-12-07 |
|-----------|------------|----------|-----------|---------------|------------|------------|------------|
|   DEFAULT |   TEMPLATE |     2423 |       276 |        (null) |          1 |          1 |          1 |
|   DEFAULT |   TEMPLATE |     2424 |       276 |        (null) |          1 |          1 |          1 |