Hive 中的复合数据结构简介以及一些函数的用法说明

目前 hive 支持的复合数据类型有以下几种：

arrays: ARRAY<\data_type> (Note: negative values and non-constant expressions are allowed as of Hive 0.14.)
maps: MAP<\primitive_type, data_type> (Note: negative values and non-constant expressions are allowed as of Hive 0.14.)
structs: STRUCT<\col_name : data_type [COMMENT col_comment], …>
union: UNIONTYPE<\data_type, data_type, …> (Note: Only available starting with Hive 0.7.0.)

一、map、struct、array 这3种的用法：

1、Array的使用

创建数据库表，以array作为数据类型

create table  person(name string,work_locations array<string>)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY '\t'
COLLECTION ITEMS TERMINATED BY ',';

数据
biansutao beijing,shanghai,tianjin,hangzhou
linan changchu,chengdu,wuhan
入库数据
LOAD DATA LOCAL INPATH '/home/hadoop/person.txt' OVERWRITE INTO TABLE person;
查询
hive> select * from person;
biansutao       ["beijing","shanghai","tianjin","hangzhou"]
linan   ["changchu","chengdu","wuhan"]
Time taken: 0.355 seconds
hive> select name from person;
linan
biansutao
Time taken: 12.397 seconds
hive> select work_locations[0] from person;
changchu
beijing
Time taken: 13.214 seconds
hive> select work_locations from person;   
["changchu","chengdu","wuhan"]
["beijing","shanghai","tianjin","hangzhou"]
Time taken: 13.755 seconds
hive> select work_locations[3] from person;
NULL
hangzhou
Time taken: 12.722 seconds
hive> select work_locations[4] from person;
NULL
NULL
Time taken: 15.958 seconds

2、Map 的使用

创建数据库表
create table score(name string, score map<string,int>)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY '\t'
COLLECTION ITEMS TERMINATED BY ','
MAP KEYS TERMINATED BY ':';
要入库的数据
biansutao '数学':80,'语文':89,'英语':95
jobs '语文':60,'数学':80,'英语':99
入库数据
LOAD DATA LOCAL INPATH '/home/hadoop/score.txt' OVERWRITE INTO TABLE score;
查询
hive> select * from score;
biansutao       {"数学":80,"语文":89,"英语":95}
jobs    {"语文":60,"数学":80,"英语":99}
Time taken: 0.665 seconds
hive> select name from score;
jobs
biansutao
Time taken: 19.778 seconds
hive> select t.score from score t;
{"语文":60,"数学":80,"英语":99}
{"数学":80,"语文":89,"英语":95}
Time taken: 19.353 seconds
hive> select t.score['语文'] from score t;
60
89
Time taken: 13.054 seconds
hive> select t.score['英语'] from score t;
99
95
Time taken: 13.769 seconds

3、Struct 的使用

创建数据表
CREATE TABLE test(id int,course struct<course:string,score:int>)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY '\t'
COLLECTION ITEMS TERMINATED BY ',';
数据
1 english,80
2 math,89
3 chinese,95
入库
LOAD DATA LOCAL INPATH '/home/hadoop/test.txt' OVERWRITE INTO TABLE test;
查询
hive> select * from test;
OK
1       {"course":"english","score":80}
2       {"course":"math","score":89}
3       {"course":"chinese","score":95}
Time taken: 0.275 seconds
hive> select course from test;
{"course":"english","score":80}
{"course":"math","score":89}
{"course":"chinese","score":95}
Time taken: 44.968 seconds
select t.course.course from test t; 
english
math
chinese
Time taken: 15.827 seconds
hive> select t.course.score from test t;
80
89
95
Time taken: 13.235 seconds

4、数据组合

CREATE TABLE union_test(foo UNIONTYPE<int, double, array<string>, struct<a:int,b:string>>);
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}


create table test1(id int,a MAP<STRING,ARRAY<STRING>>)
row format delimited fields terminated by '\t' 
collection items terminated by ','
MAP KEYS TERMINATED BY ':';
1 english:80,90,70
2 math:89,78,86
3 chinese:99,100,82
LOAD DATA LOCAL INPATH '/home/hadoop/test1.txt' OVERWRITE INTO TABLE test1;

二、hive中的一些不常见函数的用法：

常见的函数就不废话了，和标准sql类似，下面我们要聊到的基本是HQL里面专有的函数，

hive里面的函数大致分为如下几种：Built-in、Misc.、UDF、UDTF、UDAF

我们就挑几个标准SQL里没有，但是在HIVE SQL在做统计分析常用到的来说吧。

1、array_contains （Collection Functions）

这是内置的对集合进行操作的函数，用法举例：

create EXTERNAL table IF NOT EXISTS userInfo (id int,sex string, age int, name string, 
email string,sd string, ed string)  ROW FORMAT DELIMITED FIELDS TERMINATED BY '\t' location '/hive/dw';

select * from (select * from userinfo where sex='male' and 

!array_contains(split('1,2,3,4,5',','),cast(id as string))) tb1 where tb1.age < 30;

2、get_json_object （Misc. Functions）

测试数据：

first {"store":{"fruit":[{"weight":8,"type":"apple"},{"weight":9,"type":"pear"}],
"bicycle":{"price":19.951,"color":"red1"}},"email":"amy@only_for_json_udf_test.net","owner":"amy1"} third
first {"store":{"fruit":[{"weight":9,"type":"apple"},{"weight":91,"type":"pear"}],
"bicycle":{"price":19.952,"color":"red2"}},"email":"amy@only_for_json_udf_test.net","owner":"amy2"} third
first {"store":{"fruit":[{"weight":10,"type":"apple"},{"weight":911,"type":"pear"}],
"bicycle":{"price":19.953,"color":"red3"}},"email":"amy@only_for_json_udf_test.net","owner":"amy3"} third

create external table if not exists t_json(
f1 string, f2 string, f3 string) 
row format delimited fields TERMINATED BY ' ' 
location '/test/json'
select get_json_object(t_json.f2, '$.owner') from t_json;
SELECT * from t_json where get_json_object(t_json.f2, '$.store.fruit[0].weight') = 9;
SELECT get_json_object(t_json.f2, '$.non_exist_key') FROM t_json;
这里尤其要注意UDTF的问题，官方文档有说明：

json_tuple
A new json_tuple() UDTF is introduced in hive 0.7. It takes a set of names (keys) and a JSON string,
 and returns a tuple of values using one function. This is much more efficient than calling GET_JSON_OBJECT to retrieve more than one key from a single JSON string. 
which is what JSON_TUPLE is for. As JSON_TUPLE is a UDTF, you will need to use the LATERAL VIEW syntax in order to achieve the same goal.

For example,

select a.timestamp, get_json_object(a.appevents, '$.eventid'), get_json_object(a.appenvets, '$.eventname') from log a;
should be changed to

select a.timestamp, b.*
from log a lateral view json_tuple(a.appevent, 'eventid', 'eventname') b as f1, f2;
UDTF(User-Defined Table-Generating Functions)  用来解决 输入一行输出多行(On-to-many maping) 的需求。  

通过Lateral view可以方便的将UDTF得到的行转列的结果集合在一起提供服务，因为直接在SELECT使用UDTF会存在限制，
即仅仅能包含单个字段，不光是多个UDTF，仅仅单个UDTF加上其他字段也是不可以，
hive提示在UDTF中仅仅能有单一的表达式。如下：
hive> select my_test(“abcef:aa”) as qq,’abcd’ from sunwg01;
FAILED: Error in semantic analysis: Only a single expression in the SELECT clause is supported with UDTF’s

使用Lateral view可以实现上面的需求，Lateral view语法如下：
lateralView: LATERAL VIEW udtf(expression) tableAlias AS columnAlias (‘,’ columnAlias)*
fromClause: FROM baseTable (lateralView)*
hive> create table sunwg ( a array, b array )
> ROW FORMAT DELIMITED
> FIELDS TERMINATED BY ‘\t’
> COLLECTION ITEMS TERMINATED BY ‘,’;
OK
Time taken: 1.145 seconds
hive> load data local inpath ‘/home/hjl/sunwg/sunwg.txt’ overwrite into table sunwg;
Copying data from file:/home/hjl/sunwg/sunwg.txt
Loading data to table sunwg
OK
Time taken: 0.162 seconds
hive> select * from sunwg;
OK
[10,11] ["tom","mary"]
[20,21] ["kate","tim"]
Time taken: 0.069 seconds
hive>
> SELECT a, name
> FROM sunwg LATERAL VIEW explode(b) r1 AS name;
OK
[10,11] tom
[10,11] mary
[20,21] kate
[20,21] tim
Time taken: 8.497 seconds

hive> SELECT id, name
> FROM sunwg LATERAL VIEW explode(a) r1 AS id
> LATERAL VIEW explode(b) r2 AS name;
OK
10 tom
10 mary
11 tom
11 mary
20 kate
20 tim
21 kate
21 tim
Time taken: 9.687 seconds

3、parse_url_tuple

测试数据：

url1 http://facebook.com/path1/p.php?k1=v1&k2=v2#Ref1
url2 https://cwiki.apache.org/confluence/display/Hive/LanguageManual+UDF#LanguageManualUDF-getjsonobject
url3 https://www.google.com.hk/#hl=zh-CN&newwindow=1&safe=strict&q=hive+translate+example&oq=hive+translate+example
&gs_l=serp.j4.8.serp0Hs&bav=on.

create external table if not exists t_url(f1 string, f2 string) 
row format delimited fields TERMINATED BY ' ' 
location '/test/url';
SELECT f1, b.* FROM t_url LATERAL VIEW parse_url_tuple(f2, 'HOST', 'PATH', 'QUERY', 'QUERY:k1') b as host, path, query, query_id;
结果：

url1 facebook.com /path1/p.php k1=v1&k2=v2 v1
url2 cwiki.apache.org /confluence/display/Hive/LanguageManual+UDF NULL NULL
url3 www.google.com.hk / NULL NULL

4、explode

explode 是一个 hive 内置的表生成函数：Built-in Table-Generating Functions (UDTF)，

主要是解决 1 to N 的问题，即它可以把一行输入拆成多行，
比如一个 array 的每个元素拆成一行，作为一个虚表输出。它有如下需要注意的地方：

Using the syntax "SELECT udtf(col) AS colAlias..." has a few limitations:
No other expressions are allowed in SELECT
SELECT pageid, explode(adid_list) AS myCol... is not supported
UDTF's can't be nested
SELECT explode(explode(adid_list)) AS myCol... is not supported
GROUP BY / CLUSTER BY / DISTRIBUTE BY / SORT BY is not supported
SELECT explode(adid_list) AS myCol ... GROUP BY myCol is not supported
从上面的原理与语法上可知，

select 列中不能 udtf 和其它非 udtf 列混用，

udtf 不能嵌套，

不支持 GROUP BY / CLUSTER BY / DISTRIBUTE BY / SORT BY

还有 select 中出现的 udtf 一定需要列别名，否则会报错：

SELECT explode(myCol) AS myNewCol FROM myTable;
SELECT explode(myMap) AS (myMapKey, myMapValue) FROM myMapTable;
SELECT explode(myCol) AS pos, myNewCol FROM myTable;

5、lateral view

lateral view 是Hive中提供给UDTF的conjunction，
它可以解决UDTF不能添加额外的select列的问题。
当我们想对hive表中某一列进行split之后，想对其转换成1 to N的模式，即一行转多列。
hive不允许我们在UDTF函数之外，再添加其它select语句。

如下，我们想将登录某个游戏的用户id放在一个字段user_ids里，对每一行数据用UDTF后输出多行。

select game_id, explode(split(user_ids,'\\[\\[\\[')) as user_id   from login_game_log  where dt='2014-05-15' ;
FAILED: Error in semantic analysis: UDTF's are not supported outside the SELECT clause, nor nested in expressions。
提示语法分析错误，UDTF不支持函数之外的select 语句，如果我们想支持怎么办呢？接下来就是Lateral View 登场的时候了。

Lateral view 其实就是用来和像类似explode这种UDTF函数联用的。
lateral view 会将UDTF生成的结果放到一个虚拟表中，然后这个虚拟表（1 to N）会和输入行即每个game_id进行join 
来达到连接UDTF外的select字段的目的（源表和拆分的虚表按行做行内 1 join N 的直接连接），
这也是为什么 LATERAL VIEW udtf(expression) 后面需要表别名和列别名的原因。

Lateral View Syntax

lateralView: LATERAL VIEW udtf(expression) tableAlias AS columnAlias (',' columnAlias)*

fromClause: FROM baseTable (lateralView)*

可以看出，可以在2个地方用Lateral view：

在udtf前面用

在from baseTable后面用

例如：

pageid adid_list

front_page   [1, 2, 3]

contact_page [3, 4, 5]

SELECT pageid, adid
FROM pageAds LATERAL VIEW explode(adid_list) adTable AS adid;


pageid               adid

front_page         1

front_page         2

front_page         3

contact_page     3

contact_page     4

contact_page     5

From语句后可以跟多个Lateral View。

A FROM clause can have multiple LATERAL VIEW clauses. 
Subsequent LATERAL VIEWS can reference columns from any of the tables appearing to the left of the LATERAL VIEW.

给定数据：

Array<int> col1     Array<string> col2

[1, 2]                       [a", "b", "c"]

[3, 4]                       [d", "e", "f"]

转换目标：

想同时把第一列和第二列拆开，类似做笛卡尔乘积。



我们可以这样写：

SELECT myCol1, myCol2 FROM baseTable
LATERAL VIEW explode(col1) myTable1 AS myCol1
LATERAL VIEW explode(col2) myTable2 AS myCol2;


还有一种情况，如果UDTF转换的Array是空的怎么办呢？

在Hive0.12里面会支持outer关键字，如果UDTF的结果是空，默认会被忽略输出。

如果加上outer关键字，则会像left outer join 一样，还是会输出select出的列，而UDTF的输出结果是NULL。

总结：

Lateral View通常和UDTF一起出现，为了解决UDTF不允许在select字段的问题。

Multiple Lateral View可以实现类似笛卡尔乘积。

Outer关键字可以把不输出的UDTF的空结果，输出成NULL，防止丢失数据。