Hive字符串转为复杂格式数据

1、字符串转为map
str_to_map(text[, delimiter1, delimiter2])
使用两个分隔符将文本拆分为键值对。 Delimiter1将文本分成K-V对，Delimiter2分割每个K-V对。对于delimiter1默认分隔符是','，对于delimiter2默认分隔符是'='。
示例：

select str_to_map('aaa:11&bbb:22', '&', ':');
select str_to_map('aaa:11&bbb:22', '&', ':')['aaa'];

select str_to_map('device_ds:2&uid_cnt:1','&',',')  --键值分割不到，值会出现Null

综合使用示范：

select a1.appkey,a1.appsource,index_key,index_value
from tab_sum a1 
lateral view explode(str_to_map(concat('device_ds:',a1.device_ds_cnt,'&','uid_cnt:',a1.uid_cnt),'&',':')) mid_list_tab as index_key,index_value;

2、字符串转为array
分割字符串函数: split
语法:  split(string str, stringpat)
返回值:  array
说明: 按照pat字符串分割str，会返回分割后的字符串数组
举例：

select split('aaa:11:bbb:22',':');
["aaa","11","bbb","22"]
select split('aaa:11:bbb:22',':')[0];
aaa

3、字符字段去重汇总转成array
collect_set函数：该函数的作用是将某字段的值进行去重汇总，产生Array类型字段。

drop table if exists xxxxx_tabletest;
CREATE TABLE xxxxx_tabletest(
  id string, 
  name string)
ROW FORMAT SERDE 
  'org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe' 
WITH SERDEPROPERTIES ( 
  'field.delim'=',', 
  'line.delim'='\n', 
  'serialization.format'=',');
insert into xxxxx_tabletest(id,name)
values
('1','A'),
('1','C'),
('1','B'),
('2','B'),
('2','C'),
('2','D'),
('3','B'),
('3','C'),
('3','D');
select id,collect_set(name) from xxxxx_tabletest group by id;
OK
1       ["A","C","B"]
2       ["B","C","D"]
3       ["B","C","D"]
Time taken: 36.966 seconds, Fetched: 3 row(s)