SQL 从多个表中选择分组依据答案

【问题标题】：SQL Select from multiple tables Group BySQL 从多个表中选择分组依据
【发布时间】：2016-09-08 17:53:59
【问题描述】：

我正在从 SQL 数据库构建数据提取。我有以下 SQL 查询，它从多个表中选择特定列。有问题的关系是这样的：我有一个 ConveyancingCondition 表，它有一个外键到一个 ConveyancingDetail 表，它有一个外键到 OTPSale 表。 OTPSale 中的记录可能有一个 ConveyancingDetail 记录和多个 ConveyancingCondition 记录。

我需要单独选择 ConveyancingCondition 字段，并将它们组合成一个字符串。

当我执行选择时，我收到同一个 OTPSale 的多条记录。我知道我需要按conveyancingDetail 分组，但我得到分组错误“选择列表中的列无效，因为它既不包含在聚合函数或GROUP BY 子句中。”。我选择的字段太多，无法将它们全部添加到组中（至少 50 个不同的列）。

这样做的正确方法是什么？

select distinct
    --other fields from other tables
     (select FORMAT(conveyancingCondition.DateRequired, 'dd/MM/yyyy')
     where conveyancingCondition.Name = 'Deposit') 'Deposit Date Required',      
     (select FORMAT(conveyancingCondition.DateRequired, 'dd/MM/yyyy')       
     (select FORMAT(conveyancingCondition.DateRequired, 'dd/MM/yyyy')
     where conveyancingCondition.Name = 'Subject') 'Subject To Sale Date Required',     
     (select FORMAT(conveyancingCondition.DateRequired, 'dd/MM/yyyy')
     where conveyancingCondition.Name = 'Guarant') 'Guarantee Date Required',
    ISNULL(stuff((SELECT ( ',' + conveyancingCondition.DateRequired) FROM otp.ConveyancingCondition cd where (cd.ConveyancingDetailId = conveyancingDetail.Id AND cd.IsApplicable=1) FOR XML PATH( '' ) ), 1, 1, ''), NULL) as 'ConditionDates'

from otp.OTPSale    
left join otp.ConveyancingDetail conveyancingDetail 
on conveyancingDetail.Id = otp.OTPSale.ConveyancingDetailId
left join otp.ConveyancingCondition conveyancingCondition 
on conveyancingCondition.ConveyancingDetailId = conveyancingDetail.Id
WHERE otp.OTPSale.IsTransferRequired= 1 
--group by conveyancingDetail.Id

数据如下：

OTPSale

Id ConveyancingDetail IsTransferRequired
1  1                  1
2  2                  1

ConveyancingDetail

Id
1
2
3

转让条件

Id  ConveyancingDetailId    Name    IsApplicable    DateRequired
1   1                       Deposit 1               2016-09-12      
2   1                       Bond    1               2016-09-26      
3   1                       Subject 1               2016-09-26        
4   1                       Guarant 1               2016-09-30      
5   1                       Other   0               NULL    
6   2                       Deposit 1               2016-09-15      
7   2                       Bond    1               2016-09-16      
8   2                       Subject 1               2016-09-17        
9   2                       Guarant 1               2016-09-18      
10  2                       Other   0               NULL

这是我所期望的

OTPSaleId   Deposit Date Required   Bond Date Required  Subject Date Required   Guarantee Date Required     Other   ConditionDates

1           2016-09-12              2016-09-26          2016-09-26              2016-09-30              NULL    2016/09/22, 2016/09/26, 2016/09/26, 2016/09/30
2           2016-09-15              2016-09-16          2016-09-17              2016-09-18              NULL    2016/09/16, 2016/09/16, 2016/09/17, 2016/09/18

【问题讨论】：

这适用于哪个 RDBMS？请添加一个标签以指定您使用的是mysql、postgresql、sql-server、oracle 还是db2 - 或者完全是其他东西。
哎呀，忘了，sql server - 添加在标签中

标签： sql sql-server group-by concatenation

【解决方案1】：

尝试条件聚合

select conveyancingDetail.Id
    --other fields from other tables
     max(case conveyancingCondition.Name when 'Deposit' then FORMAT(conveyancingCondition.DateRequired, 'dd/MM/yyyy') end) [Deposit Date Required],      
     max(case conveyancingCondition.Name when 'Subject' then FORMAT(conveyancingCondition.DateRequired, 'dd/MM/yyyy') end) [Subject To Sale Date Required],     
     --...

from otp.OTPSale    
left join otp.ConveyancingDetail conveyancingDetail 
on conveyancingDetail.Id = otp.OTPSale.ConveyancingDetailId
left join otp.ConveyancingCondition conveyancingCondition 
on conveyancingCondition.ConveyancingDetailId = conveyancingDetail.Id
WHERE otp.OTPSale.IsTransferRequired= 1 
group by conveyancingDetail.Id

【讨论】：