hive的重要操作

1、trans_array实现

       select 
            b1.item_id,
            b2.product_info
        from(
            select 
                product_id as product_list
            from table_1
            where product_id is not null and length(product_id)>1
        )b1
        LATERAL VIEW EXPLODE(split(b1.product_list, '###')) b2 AS product_info      

2、group by 字符串连接操作

select 
        user_id
        ,concat_ws(',', collect_set(leaf_name)) as leaf_names
        ,concat_ws(',', collect_list(leaf_name)) as leaf_names
        ,count(*) as cnt 
from table_1
group by user_id 

主意:collect_set 返回不重复的集合；collect_list返回重复的集合

另外concat_ws也可以不在group by中使用，比如：

select concat_ws(',', 'a', 'b', 'c') from dual
# 输出：a,b,c

concat是用在非goup by 场景，如：

select  concat('a', '
', 'b', '	', 'c',  ',',  'd'))

#测试
select 
    --replace(info, '	', '')
    regexp_replace(info, '	', '')
    ,regexp_replace(info, '(	|
|###|  )', '')
from(
    select concat('a', '
', 'b', '	', 'c', '	', '测试###a    c,华尔街') as info
)a

参考： https://blog.csdn.net/changzoe/article/details/81181820