转置没有聚合的行和列答案

【问题标题】：Transpose rows and columns with no aggregate转置没有聚合的行和列
【发布时间】：2014-01-08 15:06:10
【问题描述】：

我有以下数据集

Account Contact

1   324324324
1   674323234
2   833343432
2   433243443
3   787655455
4   754327545
4   455435435
5   543544355
5   432455553
5   432433242
5   432432432

我想输出如下：

Account Contact1    Contact2    Contact3    Contact4

1   324324324   674323234       
2   833343432   433243443       
3   787655455           
4   754327545   455435435       
5   543544355   432455553   432433242   432432432

问题还在于我的帐户数量和联系人数量不固定

【问题讨论】：

标签： sql-server sql-server-2008 pivot

【解决方案1】：

如果您要应用 PIVOT 函数，您将需要使用聚合函数来获取结果，但您还需要使用像 row_number() 这样的窗口函数来为每个联系人生成唯一的序列帐户。

首先，您将查询您的数据，类似于：

select account, contact,
  'contact'
    + cast(row_number() over(partition by account
                              order by contact) as varchar(10)) seq
from yourtable

见SQL Fiddle with Demo。这将创建一个具有唯一序列的新列：

| ACCOUNT |   CONTACT |      SEQ |
|---------|-----------|----------|
|       1 | 324324324 | contact1 |
|       1 | 674323234 | contact2 |

如果您的列数有限，那么您可以对查询进行硬编码：

select account,
  contact1, contact2, contact3, contact4
from 
(
  select account, contact,
    'contact'
      + cast(row_number() over(partition by account
                                order by contact) as varchar(10)) seq
  from yourtable
) d
pivot
(
  max(contact)
  for seq in (contact1, contact2, contact3, contact4)
) piv;

见SQL Fiddle with Demo

如果您有未知数量的列，那么您将不得不使用动态 SQL：

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

select @cols = STUFF((SELECT ',' + QUOTENAME(seq) 
                    from
                    (
                      select 'contact'
                              + cast(row_number() over(partition by account
                                                        order by contact) as varchar(10)) seq
                      from yourtable
                    ) d
                    group by seq
                    order by seq
            FOR XML PATH(''), TYPE
            ).value('.', 'NVARCHAR(MAX)') 
        ,1,1,'')

set @query = 'SELECT account, ' + @cols + ' 
            from 
            (
                select account, contact,
                  ''contact''
                    + cast(row_number() over(partition by account
                                              order by contact) as varchar(10)) seq
                from yourtable
            ) x
            pivot 
            (
                max(contact)
                for seq in (' + @cols + ')
            ) p '

execute sp_executesql @query;

见SQL Fiddle with Demo。两者都会给你一个结果：

| ACCOUNT |  CONTACT1 |  CONTACT2 |  CONTACT3 |  CONTACT4 |
|---------|-----------|-----------|-----------|-----------|
|       1 | 324324324 | 674323234 |    (null) |    (null) |
|       2 | 433243443 | 833343432 |    (null) |    (null) |
|       3 | 787655455 |    (null) |    (null) |    (null) |
|       4 | 455435435 | 754327545 |    (null) |    (null) |
|       5 | 432432432 | 432433242 | 432455553 | 543544355 |

【讨论】：

【解决方案2】：

只是生成动态PIVOT的方式略有不同：

DECLARE @c INT;

SELECT TOP 1 @c = COUNT(*) 
  FROM dbo.YourTable
  GROUP BY Account 
  ORDER BY COUNT(*) DESC;

DECLARE @dc1 NVARCHAR(MAX) = N'', @dc2 NVARCHAR(MAX) = N'', @sql NVARCHAR(MAX);

SELECT @dc1 += ',Contact' + RTRIM(i), @dc2 += ',[Contact' + RTRIM(i) + ']'
  FROM (SELECT TOP (@c) i = number + 1 
  FROM master.dbo.spt_values WHERE type = N'P' ORDER BY number) AS x;

SET @sql = N'SELECT Account ' + @dc1 + 
  ' FROM (SELECT Account, Contact, rn = ''Contact'' 
      + RTRIM(ROW_NUMBER() OVER (PARTITION BY Account ORDER BY Contact))
    FROM dbo.YourTable) AS src PIVOT (MAX(Contact) FOR rn IN (' 
      + STUFF(@dc2, 1, 1, '') + ')) AS p;';

EXEC sp_executesql @sql;

SQLiddle demo

【讨论】：