【问题标题】：SQL Server 2014: How to include header and detail in csv exportSQL Server 2014：如何在 csv 导出中包含标题和详细信息
【发布时间】：2019-07-29 11:45:52
【问题描述】：

我正在尝试生成一个脚本来导出 csv 文件以显示标题记录和相应的详细记录。

作为一个例子，输出需要是

20138000100012, H,  2013800010, 03.04.2017, WK, 001,    2,  21.05.1984, NULL,   C
20138000100012, D,  INSURE, 0.71
20138000100012, D,  RENT,   98.58
20138000200060, H,  2013800020, 02.04.2018, WK, 006,    0,  25.05.2009, NULL,   C
20138000200060, D,  INSURE, 0.80
20138000200060, D,  RENT,   98.58
20138000500052, D,  LAUNDR, 1.29

我尝试使用下面的 T-SQL 脚本来生成它：

select * 
from 
    (select 
         cast(dbo.[RE-TENANCY].[TENANCY-REF] as nvarchar) AS [0],
         'H' as [RowLevel],
         substring(convert(varchar(10),dbo.[RE-TENANCY].[TENANCY-REF]), 1, 10) AS [1],
         -- convert(varchar,dbo.[IH-PLACE-CHG].[START-DATE],104) AS [  ]
         format(dbo.[IH-PLACE-CHG].[START-DATE], 'dd.MM.yyyy') AS [2],
         dbo.[RE-TENANCY].[RENT-GROUP] AS [3], 
         cast(substring( dbo.[RE-TENANCY].[TENANCY-REF],11,3) as nvarchar) AS [4]
, cast(substring( dbo.[RE-TENANCY].[TENANCY-REF],14,1) as nvarchar) AS [5]
, cast(format(dbo.[RE-TENANCY].[TNCY-START], 'dd.MM.yyyy') as nvarchar) AS [6]
, cast(format(dbo.[RE-TENANCY].[TNCY-END], 'dd.MM.yyyy') as nvarchar) AS [7]
, substring(dbo.[RE-TENANCY].[TNCY-STATUS],1,1) AS [8]
    FROM            
        dbo.[RE-EX-RAISE-DEBIT-DTL] 
    INNER JOIN
        dbo.[RE-TENANCY] ON dbo.[RE-EX-RAISE-DEBIT-DTL].REFERENCE = dbo.[RE-TENANCY].[TENANCY-REF] 
    INNER JOIN
        dbo.[IH-PLACE-CHG] ON substring(dbo.[RE-TENANCY].[TENANCY-REF],1,10) = dbo.[IH-PLACE-CHG].[PLACE-REF]

    union

select cast(dbo.[RE-TENANCY].[TENANCY-REF] as nvarchar) AS [0]
,'D' as [RowLevel]

, dbo.[RE-EX-RAISE-DEBIT-DTL].[ACCOUNT-CODE] AS [1]
, convert(varchar,[dbo].[RE-EX-RAISE-DEBIT-DTL].[CHARGE-AMT]) AS [2]
, '' AS [3]
, '' AS [4]
, '' AS [5]
, '' AS [6]
, '' AS [7]
, '' AS [8]
FROM            dbo.[RE-EX-RAISE-DEBIT-DTL] INNER JOIN
                         dbo.[RE-TENANCY] ON dbo.[RE-EX-RAISE-DEBIT-DTL].REFERENCE = dbo.[RE-TENANCY].[TENANCY-REF] INNER JOIN
                         dbo.[IH-PLACE-CHG] ON substring(dbo.[RE-TENANCY].[TENANCY-REF],1,10) = dbo.[IH-PLACE-CHG].[PLACE-REF]


) AS HBfile order by cast(HBfile.[0] as bigint), HBfile.[6] desc

但此代码错误地生成了以下结果：

20138000100012, H,  2013800010, 03.04.2017, WK, 001,    2,  21.05.1984, NULL,   C
20138000100012, H,  2013800010, 04.04.2016, WK, 001,    2,  21.05.1984, NULL,   C
20138000100012, H,  2013800010, 02.04.2018, WK, 001,    2,  21.05.1984, NULL,   C
20138000100012, D,  INSURE, 0.71
20138000100012, D,  RENT,   98.58
20138000200060, H,  2013800020, 02.04.2018, WK, 006,    0,  25.05.2009, NULL,   C
20138000200060, H,  2013800020, 03.04.2017, WK, 006,    0,  25.05.2009, NULL,   C
20138000200060, H,  2013800020, 04.04.2016, WK, 006,    0,  25.05.2009, NULL,   C
20138000200060, D,  INSURE, 0.80
20138000200060, D,  RENT,   98.58
20138000500060, D,  LAUNDR, 1.29

任何想法，帮助表示赞赏。谢谢

【问题讨论】：

我会用 union all 代替 union。 Union 采取了删除重复项的不必要步骤。 [TENANCY-REF],RowLevel desc 的订单能否解决您的问题？
你有一个惊人的答案。为什么忽略它？？

标签： sql-server csv tsql ssis sql-server-2014

【解决方案1】：

使用 WHERE 子句过滤掉不需要的标题行。合意度似乎基于结果中的第一个日期列。

【讨论】：

【解决方案2】：

试图找出问题

由于重复在 Headers 部分，那么您应该查看 headers 查询：

select * 
from 
    (select 
         cast(dbo.[RE-TENANCY].[TENANCY-REF] as nvarchar) AS [0],
         'H' as [RowLevel],
         substring(convert(varchar(10),dbo.[RE-TENANCY].[TENANCY-REF]), 1, 10) AS [1],
         -- convert(varchar,dbo.[IH-PLACE-CHG].[START-DATE],104) AS [  ]
         format(dbo.[IH-PLACE-CHG].[START-DATE], 'dd.MM.yyyy') AS [2],
         dbo.[RE-TENANCY].[RENT-GROUP] AS [3], 
         cast(substring( dbo.[RE-TENANCY].[TENANCY-REF],11,3) as nvarchar) AS [4]
, cast(substring( dbo.[RE-TENANCY].[TENANCY-REF],14,1) as nvarchar) AS [5]
, cast(format(dbo.[RE-TENANCY].[TNCY-START], 'dd.MM.yyyy') as nvarchar) AS [6]
, cast(format(dbo.[RE-TENANCY].[TNCY-END], 'dd.MM.yyyy') as nvarchar) AS [7]
, substring(dbo.[RE-TENANCY].[TNCY-STATUS],1,1) AS [8]
    FROM            
        dbo.[RE-EX-RAISE-DEBIT-DTL] 
    INNER JOIN
        dbo.[RE-TENANCY] ON dbo.[RE-EX-RAISE-DEBIT-DTL].REFERENCE = dbo.[RE-TENANCY].[TENANCY-REF] 
    INNER JOIN
        dbo.[IH-PLACE-CHG] ON substring(dbo.[RE-TENANCY].[TENANCY-REF],1,10) = dbo.[IH-PLACE-CHG].[PLACE-REF]

此查询包含 2 个内部连接，这意味着如果连接不是在 1-1 关系上进行的，则生成的行可能会超过源计数。

检查您提供的输出后：

20138000100012, H,  2013800010, 03.04.2017, WK, 001,    2,  21.05.1984, NULL,   C
20138000100012, H,  2013800010, 04.04.2016, WK, 001,    2,  21.05.1984, NULL,   C
20138000100012, H,  2013800010, 02.04.2018, WK, 001,    2,  21.05.1984, NULL,   C
20138000100012, D,  INSURE, 0.71
20138000100012, D,  RENT,   98.58
20138000200060, H,  2013800020, 02.04.2018, WK, 006,    0,  25.05.2009, NULL,   C
20138000200060, H,  2013800020, 03.04.2017, WK, 006,    0,  25.05.2009, NULL,   C
20138000200060, H,  2013800020, 04.04.2016, WK, 006,    0,  25.05.2009, NULL,   C
20138000200060, D,  INSURE, 0.80
20138000200060, D,  RENT,   98.58
20138000500060, D,  LAUNDR, 1.29

除了使用此命令生成的第四列之外，所有标题列都是相同的：

format(dbo.[IH-PLACE-CHG].[START-DATE], 'dd.MM.yyyy') AS [2]

解决方案

如果您只需要第一个值，那么您可以使用First_Value() 窗口函数：

FIRST_VALUE(format(dbo.[IH-PLACE-CHG].[START-DATE], 'dd.MM.yyyy')) OVER(PARTITION BY  cast(dbo.[RE-TENANCY].[TENANCY-REF] as nvarchar)) as [2]

您可以在标题查询中添加一个不同的运算符，整个 SQL 命令将是这样的：

select * 
from 
    (select Distinct 
         cast(dbo.[RE-TENANCY].[TENANCY-REF] as nvarchar) AS [0],
         'H' as [RowLevel],
         substring(convert(varchar(10),dbo.[RE-TENANCY].[TENANCY-REF]), 1, 10) AS [1],
         -- convert(varchar,dbo.[IH-PLACE-CHG].[START-DATE],104) AS [  ]
         FIRST_VALUE(format(dbo.[IH-PLACE-CHG].[START-DATE], 'dd.MM.yyyy')) OVER(PARTITION BY  cast(dbo.[RE-TENANCY].[TENANCY-REF] as nvarchar)) as [2],
         dbo.[RE-TENANCY].[RENT-GROUP] AS [3], 
         cast(substring( dbo.[RE-TENANCY].[TENANCY-REF],11,3) as nvarchar) AS [4]
, cast(substring( dbo.[RE-TENANCY].[TENANCY-REF],14,1) as nvarchar) AS [5]
, cast(format(dbo.[RE-TENANCY].[TNCY-START], 'dd.MM.yyyy') as nvarchar) AS [6]
, cast(format(dbo.[RE-TENANCY].[TNCY-END], 'dd.MM.yyyy') as nvarchar) AS [7]
, substring(dbo.[RE-TENANCY].[TNCY-STATUS],1,1) AS [8]
    FROM            
        dbo.[RE-EX-RAISE-DEBIT-DTL] 
    INNER JOIN
        dbo.[RE-TENANCY] ON dbo.[RE-EX-RAISE-DEBIT-DTL].REFERENCE = dbo.[RE-TENANCY].[TENANCY-REF] 
    INNER JOIN
        dbo.[IH-PLACE-CHG] ON substring(dbo.[RE-TENANCY].[TENANCY-REF],1,10) = dbo.[IH-PLACE-CHG].[PLACE-REF]

    union

select cast(dbo.[RE-TENANCY].[TENANCY-REF] as nvarchar) AS [0]
,'D' as [RowLevel]

, dbo.[RE-EX-RAISE-DEBIT-DTL].[ACCOUNT-CODE] AS [1]
, convert(varchar,[dbo].[RE-EX-RAISE-DEBIT-DTL].[CHARGE-AMT]) AS [2]
, '' AS [3]
, '' AS [4]
, '' AS [5]
, '' AS [6]
, '' AS [7]
, '' AS [8]
FROM            dbo.[RE-EX-RAISE-DEBIT-DTL] INNER JOIN
                         dbo.[RE-TENANCY] ON dbo.[RE-EX-RAISE-DEBIT-DTL].REFERENCE = dbo.[RE-TENANCY].[TENANCY-REF] INNER JOIN
                         dbo.[IH-PLACE-CHG] ON substring(dbo.[RE-TENANCY].[TENANCY-REF],1,10) = dbo.[IH-PLACE-CHG].[PLACE-REF]


) AS HBfile order by cast(HBfile.[0] as bigint), HBfile.[6] desc

【讨论】：