解决方法
这个问题非常类似于
PIVOT rows to columns with more than 1 value returned,您需要将行中的字符串数据聚合到列中.我将修改该答案,以演示如何将数据转换为最终结果.
由于您正在聚合字符串值,因此您需要应用min()或max()聚合函数,但为了使最终结果显示多行,您需要强制执行多行.
为此,您需要使用row_number()为名称中的每个参数生成唯一的序列号.应用PIVOT函数时,此编号将用于分组,您将生成多行:
select Car,Truck,Bicycle from ( select vt.name,vp.parameter,row_number() over(partition by vt.name order by vt.id) seq from vehicle_types vt left join vehicle_parameters vp on vt.id = vp.vehicletype ) d pivot ( max(parameter) for name in (Car,Bicycle) ) piv;
这也可以使用带有CASE表达式的聚合函数编写:
select max(case when name = 'Car' then parameter end) Car,max(case when name = 'Truck' then parameter end) Truck,max(case when name = 'Bicycle' then parameter end) Bicycle from ( select vt.name,row_number() over(partition by vt.name order by vt.id) seq from vehicle_types vt left join vehicle_parameters vp on vt.id = vp.vehicletype ) d group by seq;
如果你有一个已知数量的名称,上面两个版本是很好的,那么你可以硬编码查询,但如果你没有,那么你将不得不使用动态sql:
DECLARE @cols AS NVARCHAR(MAX),@query AS NVARCHAR(MAX) select @cols = STUFF((SELECT ',' + QUOTENAME(name) from vehicle_types group by id,name order by id FOR XML PATH(''),TYPE ).value('.','NVARCHAR(MAX)'),1,'') set @query = 'SELECT ' + @cols + ' from ( select vt.name,row_number() over(partition by vt.name order by vt.id) seq from vehicle_types vt left join vehicle_parameters vp on vt.id = vp.vehicletype ) x pivot ( max(parameter) for name in (' + @cols + ') ) p ' execute sp_executesql @query;
见SQL Fiddle with Demo.所有版本都给出了结果:
| CAR | TRUCK | BICYCLE | |--------|------------------|------------| | make | maxload | frame | | year | hasconcretemixer | isroad | | engine | (null) | ismountain |