Question

我正在寻找一些帮助，将非equi连接查询转换为仅支持equi连接查询的hive。

我尝试过这样做，并且很高兴知道如果我把它们弄好了，请让我知道我的错误以及实现同样的正确方法。

A）

SELECT 
  a.time_p,
  count(distinct a.consumerid)
from tableA a
join tableB
  on a.consumerid = tableB.consumerid
  and tableB.time_p != a.time_p
group by a.time_p

我试过了::

SELECT time_p as period, count(distinct consumerid) as count
FROM(
SELECT 
  a.time_p,
  a.consumerid,
  CASE
        WHEN tableB.time_p != a.time_p THEN 1
        ELSE null END as testcase
from tableA a
join tableB
  on a.consumerid = tableB.consumerid  
)t WHERE testcase is not null
group by time_p

由于

Answer 1

你可以有不平等条件 - 只是不作为连接条件：

SELECT a.time_p,count(distinct a.consumerid)
from tableA a
join tableB b
ON a.consumerid = b.consumerid
WHERE b.time_p != a.time_p
group by a.time_p;

加入后将应用WHERE条件。

将带有非equi连接的sql转换为hive

1 个答案: