Hive常见内置函数及其使用

函数分类

HIVE CLI命令

显示当前会话有多少函数可用
SHOW FUNCTIONS;

显示函数的描写叙述信息
DESC FUNCTION concat;

显示函数的扩展描写叙述信息
DESC FUNCTION EXTENDED concat;

简单函数

函数的计算粒度为单条记录。
关系运算
数学运算
逻辑运算
数值计算
类型转换
日期函数
条件函数
字符串函数
统计函数

聚合函数

函数处理的数据粒度为多条记录。
sum()—求和
count()—求数据量
avg()—求平均直
distinct—求不同值数
min—求最小值
max—求最人值

集合函数

复合类型构建
复杂类型訪问
复杂类型长度

特殊函数

窗体函数

应用场景
用于分区排序
动态Group By
Top N
累计计算
层次查询

Windowing functions

leadlagFIRST_VALUELAST_VALUE
1
2
3
4

分析函数

Analytics functions

RANKROW_NUMBERDENSE_RANKCUME_DISTPERCENT_RANKNTILE
1
2
3
4
5
6

混合函数

java_method(class,method [,arg1 [,arg2])reflect(class,method [,arg1 [,arg2..]])hash(a1 [,a2...])
1
2
3

UDTF

lateralView: LATERAL VIEW udtf(expression) tableAlias AS columnAlias (',' columnAlias)*  fromClause: FROM baseTable (lateralView)*  
1
2

ateral view用于和split, explode等UDTF一起使用，它可以将一行数据拆成多行数据，在此基础上可以对拆分后的数据进行聚合。lateral view首先为原始表的每行调用UDTF，UTDF会把一行拆分成一或者多行。lateral view再把结果组合。产生一个支持别名表的虚拟表。

经常使用函数Demo：

create table employee(id string,money double,type string)row format delimited fields terminated by '\t' lines terminated by '\n' stored as textfile;load data local inpath '/liguodong/hive/data' into table employee;select * from employee;优先级依次为NOT AND ORselect id,money from employee where (id='1001' or id='1002') and money='100';
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

cast类型转换

select cast(1.5 as int);
1

if推断

if(con,'','');hive (default)> select if(2>1,'YES','NO');YES
1
2
3
4

case when con then '' when con then '' else  '' end (''里面类型要一样)select case when id='1001' then 'v1001' when id='1002' then 'v1002' else 'v1003' end from employee;
1
2
3

get_json_object

get_json_object(json 解析函数，用来处理json，必须是json格式)select get_json_object('{"name":"jack","age":"20"}','$.name');
1
2

URL解析函数

parse_url(string urlString, string partToExtract [, string keyToExtract])select parse_url('http://facebook.com/path1/p.php?k1=v1&k2=v2#Ref1', 'HOST') fromemployee limit 1;
1
2
3
4

字符串连接函数： concat
语法: concat(string A, string B…)
返回值: string
说明：返回输入字符串连接后的结果，支持随意个输入字符串
举例：

hive> select concat('abc','def’,'gh') from lxw_dual;abcdefgh
1
2

带分隔符字符串连接函数： concat_ws
语法: concat_ws(string SEP, string A, string B…)
返回值: string
说明：返回输入字符串连接后的结果， SEP 表示各个字符串间的分隔符

concat_ws(string SEP, array<string>)举例：hive> select concat_ws(',','abc','def','gh') from lxw_dual;abc,def,gh
1
2
3
4
5

列出该字段全部不反复的值，相当于去重collect_set(id)  //返回的是数组列出该字段全部的值。列出来不去重 collect_list(id)   //返回的是数组select collect_set(id) from taborder;
1
2
3
4
5
6
7

求和sum(money)统计列数count(*)select sum(num),count(*) from taborder;
1
2
3
4
5
6
7

窗体函数

first_value(第一行值)first_value(money) over (partition by id order by money)select ch,num,first_value(num) over (partition by ch order by num) from taborder;
1
2
3
4
5

rows between 1 preceding and 1 following (当前行以及当前行的前一行与后一行)hive (liguodong)> select ch,num,first_value(num) over (partition by ch order by num ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) from taborder;
1
2
3
4

last_value 最后一行值hive (liguodong)> select ch,num,last_value(num) over (partition by ch) from taborder;
1
2
3

 lead 去当前行后面的第二行的值 lead(money,2) over (order by money) lag  去当前行前面的第二行的值 lag(money,2) over (order by money) ``` ``` select ch, num, lead(num,2) over (order by num) from taborder;
1
2
3
4
5
6
7
8
9
10
11

 select ch, num, lag(num,2) over (order by num) from taborder;
1

rank排名

rank() over(partition by id order by money)
1

select ch, num, rank() over(partition by ch order by num) as rank from taborder;
1

select ch, num, dense_rank() over(partition by ch order by num) as dense_rank from taborder;
1

cume_dist

cume_dist (同样值的最大行号/行数)cume_dist() over (partition by id order by money)percent_rank (同样值的最小行号-1)/(行数-1)第一个总是从0開始percent_rank() over (partition by id order by money) select ch,num,cume_dist() over (partition by ch order by num) as cume_dist, percent_rank() over (partition by ch order by num) as percent_rank from taborder;
1
2
3
4
5
6
7
8
9
10
11
12

 ntile分片  ntile(2) over (order by money desc)  分两份  select ch,num,ntile(2) over (order by num desc) from taborder;
1
2
3
4

混合函数

select id,java_method("java.lang,Math","sqrt",cast(id as double)) as sqrt from hiveTest;
1

UDTF

 select id,adid  from employee  lateral view explode(split(type,'B')) tt as adid; explode 把一列转成多行hive (liguodong)>  select id,adid                >  from hiveDemo                >  lateral view explode(split(str,',')) tt as adid;
1
2
3
4
5
6
7
8

正則表達式
使用正則表達式的函数
regexp_replace(string subject A,string B,string C)
regexp_extract(string subject,string pattern,int index)

hive> select regexp_replace('foobar', 'oo|ar', '') from lxw_dual;fbhive> select regexp_replace('979|7.10.80|8684', '.*\\|(.*)',1) from hiveDemo limit 1;
1
2
3
4

hive> select regexp_replace('979|7.10.80|8684', '(.*?)\\|(.*)',1) from hiveDemo limit 1;
1