Question

所以，我有一张表

 user_id, product, price
  12, foobar, 1.2
  12, foo, 2.2
  13, baz, 2
  14, biz, 20
  14, buzz, 21
  and so on..

我想按用户ID分组，基本上存储为以下架构

schema:
  user_id int, 
  product_prices string

，结果是

12\t foobar,1.2\tfoo,2.2
13\t baz,2
14\t biz,20\tbuzz,21

等等

我如何在hive（sqlish）中执行此操作？

Answer 1

可以使用CTE，CONCAT和concat_ws

完成

尝试以下代码

with temp as
( select user_id, CONCAT(product , ',' , cast(price as String)) res from your_table)

SELECT user_id, concat_ws("\t",collect_set(res))
FROM temp  
GROUP BY user_id;

步骤1 CONCAT产品和价格与＆＃34;，＆＃34;

步骤2将concat_ws用于＆＃34; \ t＆＃34;到CONCAT数组。

Answer 2

select      user_id
           ,concat_ws('\t',collect_list(concat_ws(',',product,cast(price as string))))
from        mytable
group by    user_id
;

+----+----------------------+
| 12 | foobar,1.2   foo,2.2 |
| 13 | baz,2                |
| 14 | biz,20   buzz,21     |
+----+----------------------+

...如果字典（地图）真的是你想要的：

select      user_id
           ,str_to_map(concat_ws(',',collect_list(concat_ws(':',product,cast(price as string)))))
from        mytable
group by    user_id
;

+----+------------------------------+
| 12 | {"foobar":"1.2","foo":"2.2"} |
| 13 | {"baz":"2"}                  |
| 14 | {"biz":"20","buzz":"21"}     |
+----+------------------------------+

groupby并将其存储为配置单元

2 个答案: