BigQuery SQL 连接多列，但如果两者都不匹配，则应添加 0 行

Question

我正在尝试进行如下连接：

给定表 1：

+-----------+------+-------+
| date      | name | count |
+-----------+------+-------+
| 1.1.2020. | A    | 5     |
+-----------+------+-------+
| 1.1.2020. | B    | 2     |
+-----------+------+-------+
| 1.1.2020. | C    | 3     |
+-----------+------+-------+

和表 2：

+-----------+------+-------+
| date      | name | count |
+-----------+------+-------+
| 1.1.2020. | B    | 1     |
+-----------+------+-------+
| 1.1.2020. | C    | 4     |
+-----------+------+-------+
| 1.1.2020. | D    | 6     |
+-----------+------+-------+

我想按日期和姓名加入他们，但如果姓名不在其中一张表中，最终表应该仍然有它，计数为 0。

结果表将是

+-----------+------+---------+---------+
| date      | name | count_1 | count_2 |
+-----------+------+---------+---------+
| 1.1.2020. | A    | 5       | 0       |
+-----------+------+---------+---------+
| 1.1.2020. | B    | 2       | 1       |
+-----------+------+---------+---------+
| 1.1.2020. | C    | 3       | 4       |
+-----------+------+---------+---------+
| 1.1.2020. | D    | 0       | 6       |
+-----------+------+---------+---------+

无论我尝试哪种连接类型，它只匹配名称在两个表中的那些行，所以我只会在此处获取这两行

+-----------+------+---------+---------+
| date      | name | count_1 | count_2 |
+-----------+------+---------+---------+
| 1.1.2020. | B    | 2       | 1       |
+-----------+------+---------+---------+
| 1.1.2020. | C    | 3       | 4       |
+-----------+------+---------+---------+

这是我尝试过的查询之一，还有左连接、右连接等。

select * from 
    table_1 full outer join table_2 using(date, name)

谢谢！

Answer 1

考虑以下方法

select date, name, 
  sum(if(col = 'count_1', count, 0)) count_1,
  sum(if(col = 'count_2', count, 0)) count_2
from (
  select *, 'count_1' col from table_1 union all 
  select *, 'count_2' from table_2 
)
group by date, name        

如果应用于您问题中的样本数据 - 输出是

Answer 2

考虑另一种方法 - 使用 PIVOT 运算符减少冗长

select *
from (
  select *, 'count_1' col from table_1 union all 
  select *, 'count_2' from table_2 
)
pivot (sum(count) for col in ('count_1', 'count_2')) 

如果应用于您问题中的样本数据 - 输出是

Answer 3

您的查询应该有效，但我将其表述为：

select date, name, coalesce(t1.count, 0) as count_1,
       coalesce(t2.count, 0) as count_2
from table_1 t1 full outer join
     table_2 t2
     using (date, name);

这会将附加计数返回为0，而不是NULL。