HIVE SQL 根据主键去重并实现其余字段分组聚合-Toy模板网

这篇具有很好参考价值的文章主要介绍了HIVE SQL 根据主键去重并实现其余字段分组聚合。希望对大家有所帮助。如果存在错误或未考虑完全的地方，请大家不吝赐教，您也可以点击"举报违法"按钮提交疑问。

相同个人id下所有字段按时间顺序补位，取首个不为空值
HIVE SQL 根据主键去重并实现其余字段分组聚合,hive,sql,hadoop

--数据建表
drop table if exists db.tb_name;
create table if not exists db.tb_name
( id string
	,name string
	,tele string
	,email string
	,`date` string
)
;
insert overwrite table db.tb_name
values 
("32001","张三","23456789",null,"2023-07-18")
,("32001",null,null,"23456789@163.com","2023-07-19")
,("32002","李四",null,"23456876@qq.com","2023-07-18")
,("32003","王二",null,null,"2023-07-18")
,("32003",null,"9876789",null,"2023-06-18")
,("32003",null,null,"9876789@gmail.com","2023-07-18")
,("32004","刘五","987456798",null,"2023-07-18")
,("32004","刘七","1987456798",null,"2023-07-20")
;

--distribute by 分区排序：类似MR中partition，进行分区，结合sort by使用

drop table if exists db.tb_name_new;
create table if not exists db.tb_name_new as
select id 
	,collect_list(`name`)[0] as `name`
	,collect_list(`tele`)[0] as `tele`
	,collect_list(`email`)[0] as `email`
	,collect_list(`date`)[0] as `date`
from 
(
select id
	,name
	,tele
	,email
	,`date`
from 
db.tb_name
distribute by id 
sort by id,`date` desc
) t 
group by id
;

*注意：此处是取的首个不为空（即不为null）的字段，所以在实际使用过程中应提前将空字符串转为null值。文章来源地址https://www.toymoban.com/news/detail-595313.html

--剔除字符串中的不可见字符，若该字段中均为不可见字符或该字段为空字符串，则转为空
case when length(regexp_replace(col_name,'[\\x00-\\x08\\x0B-\\x0C\\x0E-\\x1F]+|\\s+',''))>0 
	then regexp_replace(col_name,'[\\x00-\\x08\\x0B-\\x0C\\x0E-\\x1F]+|\\s+','')
	else null end as new_col_name