hive学习笔记之二:复杂数据类型

欢迎访问我的GitHub

这里分类和汇总了欣宸的全部原创(含配套源码):https://github.com/zq2599/blog_demos

《hive学习笔记》系列导航

  1. 基本数据类型
  2. 复杂数据类型
  3. 内部表和外部表
  4. 分区表
  5. 分桶
  6. HiveQL基础
  7. 内置函数
  8. Sqoop
  9. 基础UDF
  10. 用户自定义聚合函数(UDAF)
  11. UDTF

本篇概览

  • 作为《hive学习笔记》的第二篇,前面咱们了解了基本类型,本篇要学习的是复杂数据类型;
  • 复杂数据类型一共有四种:
  1. ARRAY:数组
  2. MAP:键值对
  3. STRUCT:命名字段集合
  4. UNION:从几种数据类型中指明选择一种,UNION的值必须于这些数据类型之一完全匹配;
  • 接下来逐个学习;

准备环境

  1. 确保hadoop已经启动;
  2. 进入hive控制台的交互模式;
  3. 执行以下命令,使查询结果中带有字段名:
set hive.cli.print.header=true;

ARRAY

  1. 创建名为t2的表,只有person和friends两个字段,person是字符串类型,friends是数组类型,通过文本文件导入数据时,person和friends之间的分隔符是竖线,friends内部的多个元素之间的分隔符是逗号,注意声明分隔符的语法:
create table if not exists t2(
person string,
friends array<string>
)
row format delimited 
fields terminated by '|'
collection items terminated by ',';  
  1. 创建文本文件002.txt,内容如下,可见只有两条记录,第一条person字段值为tom,friends字段里面有三个元素,用逗号分隔:
tom|tom_friend_0,tom_friend_1,tom_friend_2
jerry|jerry_friend_0,jerry_friend_1,jerry_friend_2,jerry_friend_3,jerry_friend_4,jerry_friend_5
  1. 执行以下语句,从本地的002.txt文件导入数据到t2表:
load data local inpath '/home/hadoop/temp/202010/25/002.txt' into table t2;
  1. 查看全部数据:
hive> select * from t2;
OK
t2.person	t2.friends
tom	["tom_friend_0","tom_friend_1","tom_friend_2"]
jerry	["jerry_friend_0","jerry_friend_1","jerry_friend_2","jerry_friend_3","jerry_friend_4","jerry_friend_5"]
Time taken: 0.052 seconds, Fetched: 2 row(s)
  1. 查询friends中的某个元素的SQL:
select person, friends[0], friends[3] from t2;  

执行结果如下,第一条记录没有friends[3],显示为NULL:

hive> select person, friends[0], friends[3] from t2; 
OK
person	_c1	_c2
tom	tom_friend_0	NULL
jerry	jerry_friend_0	jerry_friend_3
Time taken: 0.052 seconds, Fetched: 2 row(s)
  1. 数组元素中是否包含某值的SQL:
select person, array_contains(friends, 'tom_friend_0') from t2;

执行结果如下,第一条记录friends数组中有tom_friend_0,显示为true,第二条记录不包含,就显示false:

hive> select person, array_contains(friends, 'tom_friend_0') from t2;
OK
person	_c1
tom	true
jerry	false
Time taken: 0.061 seconds, Fetched: 2 row(s)
  1. 第一条记录的friends数组中有三个元素,借助LATERAL VIEW语法可以把这三个元素拆成三行,SQL如下:
select t.person, single_friend
from (
    select person, friends 
    from  t2 where person='tom'
) t LATERAL VIEW explode(t.friends) v as single_friend;

执行结果如下,可见数组中的每个元素都能拆成单独一行:

OK
t.person	single_friend
tom	tom_friend_0
tom	tom_friend_1
tom	tom_friend_2
Time taken: 0.058 seconds, Fetched: 3 row(s)
  • 以上就是数组的基本操作,接下来是键值对;

MAP,建表,导入数据

  • 接下来打算创建名为t3的表,只有person和address两个字段,person是字符串类型,address是MAP类型,通过文本文件导入数据时,对分隔符的定义如下:
  1. person和address之间的分隔符是竖线
  2. address内部有多个键值对,它们的分隔符是逗号
  3. 而每个键值对的键和值的分隔符是冒号
  • 满足上述要求的建表语句如下所示:
create table if not exists t3(
person string,
address map<string, string>
)
row format delimited 
fields terminated by '|'
collection items terminated by ',' 
map keys terminated by ':';
  • 创建文本文件003.txt,可见用了三种分隔符来分隔字段、MAP中的多个元素、每个元素键和值:
tom|province:guangdong,city:shenzhen
jerry|province:jiangsu,city:nanjing
  • 导入003.txt的数据到t3表:
load data local inpath '/home/hadoop/temp/202010/25/003.txt' into table t3;

MAP,查询

  1. 查看全部数据:
hive> select * from t3;
OK
t3.person	t3.address
tom	{"province":"guangdong","city":"shenzhen"}
jerry	{"province":"jiangsu","city":"nanjing"}
Time taken: 0.075 seconds, Fetched: 2 row(s)
  1. 查看MAP中的某个key,语法是field[“xxx”]
hive> select person, address["province"] from t3;
OK
person	_c1
tom	guangdong
jerry	jiangsu
Time taken: 0.075 seconds, Fetched: 2 row(s)
  1. 使用if函数,下面的SQL是判断address字段中是否有"street"键,如果有就显示对应的值,没有就显示filed street not exists
select person, 
if(address['street'] is null, "filed street not exists", address['street']) 
from t3;

输出如下,由于address字段只有provincecity两个键,因此会显示filed street not exists

OK
tom	filed street not exists
jerry	filed street not exists
Time taken: 0.087 seconds, Fetched: 2 row(s)
  1. 使用explode将address字段的每个键值对展示成一行:
hive> select explode(address) from t3;
OK
province	guangdong
city	shenzhen
province	jiangsu
city	nanjing
Time taken: 0.081 seconds, Fetched: 4 row(s)
  1. 上面的explode函数只能展示address字段,如果还要展示其他字段就要继续LATERAL VIEW语法,如下,可见前面的数组展开为一个字段,MAP展开为两个字段,分别是key和value:
select t.person, address_key, address_value
from (
    select person, address 
    from  t3 where person='tom'
) t LATERAL VIEW explode(t.address) v as  address_key, address_value;

结果如下:

OK
tom	province	guangdong
tom	city	shenzhen
Time taken: 0.118 seconds, Fetched: 2 row(s)
  1. size函数可以查看MAP中键值对的数量:
hive> select person, size(address) from t3;
OK
tom	2
jerry	2
Time taken: 0.082 seconds, Fetched: 2 row(s)

STRUCT

  1. STRUCT是一种记录类型,它封装了一个命名的字段集合,里面有很多属性,新建名为t4的表,其info字段就是STRUCT类型,里面有age和city两个属性,person和info之间的分隔符是竖线,info内部的多个元素之间的分隔符是逗号,注意声明分隔符的语法:
create table if not exists t4(
person string,
info struct<age:int, city:string>
)
row format delimited 
fields terminated by '|'
collection items terminated by ',';
  1. 准备好名为004.txt的文本文件,内容如下:
tom|11,shenzhen
jerry|12,nanjing
  1. 加载004.txt的数据到t4表:
load data local inpath '/home/hadoop/temp/202010/25/004.txt' into table t4;
  1. 查看t4的所有数据:
hive> select * from t4;
OK
tom	{"age":11,"city":"shenzhen"}
jerry	{"age":12,"city":"nanjing"}
Time taken: 0.063 seconds, Fetched: 2 row(s)
  1. 查看指定字段,用filedname.xxx语法:
hive> select person, info.city from t4;
OK
tom	shenzhen
jerry	nanjing
Time taken: 0.141 seconds, Fetched: 2 row(s)

UNION

  • 最后一种是UNIONTYPE,这是从几种数据类型中指明选择一种,由于UNIONTYPE数据的创建设计到UDF(create_union),这里先不展开了,先看看建表语句:
CREATE TABLE union_test(foo UNIONTYPE<int, double, array<string>, struct<a:int,b:string>>);
  • 查询结果:
SELECT foo FROM union_test;

{0:1}
{1:2.0}
{2:["three","four"]}
{3:{"a":5,"b":"five"}}
{2:["six","seven"]}
{3:{"a":8,"b":"eight"}}
{0:9}
{1:10.0}
  • 至此,hive的基础数据类型和复杂数据类型咱们都实际操作过一遍了,接下来的文章将展开更多hive知识,期待与您共同进步;

欢迎关注公众号:程序员欣宸

微信搜索「程序员欣宸」,我是欣宸,期待与您一同畅游Java世界…

©️2020 CSDN 皮肤主题: 书香水墨 设计师:CSDN官方博客 返回首页