hive使用struct、map与array类型字段

hive支持struct,map,array三种集合类型

struct

与 C语言 、 golang 中的struct类似,可以通过 . 语法来访问

定义一个包含struct字段的表

create table test2(field1 struct<name:string,age:int> comment "test field") row format delimited fields terminated by "," collection items terminated by ":";

其中 row format delimited fields terminated by "," 表示字段之间用","分割

collection items terminated by ":"; 字段中各个元素项之间用":"分割

然后导入数据,可以使用 sql 语句和文件的方式,首先是sql语句

insert into test2(field1)values(named_struct("name",'zhangsan',"age",25));
insert into test2(field1)values(named_struct("name","lisi","age",23));

从文件导入,新建一个test.txt,内容如下

name:wangwu|age:27
name:zhaoliu|age:28

使用 load data 语句从本地导入

load data local inpath '/Users/zhangsheng/hive/note/hive/test.txt' into table test2;

使用 . 语法查询数据

-- 查询所有数据
select * from test2;
-- 查询name字段
select field1.name from test2;
-- 查询name为zhangsan的记录
select field1 from test2 where field1.name = "zhangsan";

结果如下

result1.png

result2.png

result3.png

map

同struct, map 的用法基本相似,以下为创建一个map类型字段

create table test3(field2 map<string,string>) row format delimited fields terminated by ',' collection items terminated by "|" map keys terminated by ":";

map keys terminated by ":"表示键值对之间用":"来分割

同样的,先使用sql方式插入

insert into test3(field2)values(str_to_map("name:zhangsan,age:25")),(str_to_map("name:lisi,age:23"));

再导入文件,文件内容如下:

load data local inpath  '/Users/zhangsheng/hive/note/hive/test.txt' into table test3;

map的访问通过 map[key] 的形式进行,测试一下

select * from test3;
select field2["name"] as name,field2["age"] as age from test3;
select * from test3 where field2["age"] > 25;

结果如下

result4.png

result5.png

result6.png

array

与所有支持数组的编程语言一致,使用 array[下标] 的方式访问,创建一个 array 类型字段

create table test4(field4 array<string>);

插入数据

insert into test4(field4)values(array("zhangsan","lisi","wangwu"));
insert into test4(field4)values(`array`("lily","bob","alice"));
insert into test4(field4)values(`array`("A","B","C"));

查询测试

select * from test4;
select field4[0] from test4;
select * from test4 where field4[0]="zhangsan";

结果如下

result7.png

result8.png

result9.png