自动从行生成列（没有 IF）

Question

我的问题与How to simulate a pivot table with BigQuery? 类似。但是想象一下你有不可预测的行数，所以你不能在 IF 语句中列出它们。 那么有没有办法从一列中获取 DISTINCT 值并从中创建列列表？

Answer 1

下面的非常简单的例子只是为了展示方法：

假设你有 temp.example 表：

select * from
(select '2015-08-01' as day, 1 as category, 11 as volume),
(select '2015-08-01' as day, 2 as category, 21 as volume),
(select '2015-08-01' as day, 3 as category, 31 as volume),
(select '2015-08-02' as day, 1 as category, 101 as volume),
(select '2015-08-02' as day, 2 as category, 201 as volume),
(select '2015-08-03' as day, 1 as category, 301 as volume),
(select '2015-08-03' as day, 2 as category, 302 as volume),
(select '2015-08-03' as day, 4 as category, 304 as volume)

并且，假设您想要构建如下所示的查询，但事先不知道您有多少不同的类别

select 
   day,
   sum(if(category = 1, volume, 0)) as category_1,
   sum(if(category = 2, volume, 0)) as category_2,
   sum(if(category = 3, volume, 0)) as category_3,
   sum(if(category = 4, volume, 0)) as category_4
from temp.example
group by day
order by day

下面是执行此操作的 GBQ 代码

select 'select day, ' + 
   group_concat_unquoted('sum(if(category = ' + string(category) + ', volume, 0)) as category_' + string(category)) 
   + ' from temp.example group by day order by day'
from (select category from temp.example group by category order by category)

这个查询的输出是一个查询，如果你运行它，它将为你生成数据透视

天 category_1 category_2 category_3 category_4 2015-08-01 11 21 31 0 2015-08-02 101 201 0 0 2015-08-03 301 302 0 304

Answer 2

是的，确实有可能，但您需要使用自己的编程语言，因为您无法使用 SQL 查询语言生成 SQL 语法。