linq sql 一例特殊故障及处理

故障现象

数据结构如下：
＜Waybills＞waybill<－－>＜Clients＞sender＜－－＝＝>＜ClientsAddrs＞senderAddrs[]

from w in _context.Waybills.Where(o => o.FlowID == 2 && o.SenderStationID == localStationId) 
join c in _context.Clients.Include(o => o.clientAddrs) on w.RecipientID equals c.Id 
select new { w.Id, c.Name, w.RecipientID, w.RecipientAddrNo, c.clientAddrs[w.RecipientAddrNo].CityId, c.clientAddrs[w.RecipientAddrNo].Address, w.SenderStationID }
).ToList(); ;

生成的sql如下

SELECT [w].[Id], [c].[Name], [w].[RecipientID], [w].[RecipientAddrNo], 
	  ( SELECT [c0].[CityId] 
    FROM [ClientsAddr] AS [c0]
    WHERE [c].[Id] = [c0].[ClientId]
    ORDER BY (select 1)//=order  by id
    OFFSET [w].[RecipientAddrNo] ROWS FETCH NEXT 1 ROWS ONLY)as cityId,
	(select [c1].Address 
    FROM [ClientsAddr] AS [c1]
    WHERE [c].[Id] = [c1].[ClientId]
    ORDER BY (select 1)//=order by id
    OFFSET [w].[RecipientAddrNo] ROWS FETCH NEXT 1 ROWS ONLY)as address
FROM [Waybills] AS [w] 
INNER JOIN [Clients] AS [c] ON [w].[RecipientID] = [c].[Id] where (w.FlowID=2 and w.SenderStationID=1)

上述代码能直接运行，且结果正确。但认真看上面的代码，其中：

 ( SELECT [c0].[CityId] 
    FROM [ClientsAddr] AS [c0]
    WHERE [c].[Id] = [c0].[ClientId]
    ORDER BY (select 1)
    OFFSET [w].[RecipientAddrNo] ROWS FETCH NEXT 1 ROWS ONLY)as cityId,
    //只选择了<Client> ClientsAddr  库中，第RecipientAddrNo行数据的CityID字段；
	(select [c1].Address 
    FROM [ClientsAddr] AS [c1]
    WHERE [c].[Id] = [c1].[ClientId]
    ORDER BY (select 1)
    OFFSET [w].[RecipientAddrNo] ROWS FETCH NEXT 1 ROWS ONLY)as address
      //只选择了<Client> ClientsAddr  库中，第RecipientAddrNo行数据的Address字段；

其实是一ClientsAddr 库中同一条记录，且这两个字段(cityId和Address)均来自ClientsAddr,并满足以下条件

 (SELECT  [c0].[CityId] as cityId,[c0].[Address] as address
    FROM [ClientsAddr] AS [c0]
    WHERE [c].[Id] = [c0].[ClientId]
   and [c0].Id2=[w].[RecipientAddrNo]
   )

那么，能不能将两个子查询 合并为一个？但把两个字段合并时，出现错误：
消息 116，级别 16，状态 1，第 7 行
当没有用 EXISTS 引入子查询时，在选择列表中只能指定一个表达式。

原因分析

由于本例中，首先是Waybills库与收货人（Recipient）库Clients 进行一对一连接（on RecipientID=Clients.ID)，再由Client 与ClientsAddr进行一对多条件连接（通过条件 [c].[Id] = [c0].[ClientId] and [c0].Id2=[w].[RecipientAddrNo]，在EntityFramwork中，依据sql 语言要求，直接将linq 转化为

SELECT [w].[Id], [c].[Name], [w].[RecipientID], [w].[RecipientAddrNo] , 
	   (SELECT  [c0].[CityId] as cityId,[c0].[Address] as address
    FROM [ClientsAddr] AS [c0]
    WHERE [c].[Id] = [c0].[ClientId]
   and [c0].Id2=[w].[RecipientAddrNo]
   ),
   (SELECT  [c0].[Address] as address
    FROM [ClientsAddr] AS [c0]
    WHERE [c].[Id] = [c0].[ClientId]
   and [c0].Id2=[w].[RecipientAddrNo]
   )

FROM [Waybills] AS [w] 
INNER JOIN [Clients] AS [c] ON [w].[RecipientID] = [c].[Id] where (w.FlowID=2 and w.SenderStationID=1)

因在sqlserver 查询时，select a.f1…,b.f2… from a inner join b on a.fa=b.fb where exisits (select * from c where …) 或 select a.f1…,b.f2… from a inner join b on a.fa=b.fb where a.fc in (select * from c…) 多字段 select 子句只能出现 exists 或in 后面，因此 就只能将查询CityID 和Address两个字段的子句，分开为两个子句了。
这虽然是语法规定，但和想按照上述思路，一次查询一个子句能生成的结果，两个子句肯定效率要低。其中深层次的原因，也请数据库高手，在评论留言中说明。