===============================================================================
create table psn
(
    id int,
    name string,
    likes array,
    address map
)
partitioned by (age int)
row format delimited
fields terminated by '\t'
collection items terminated by '-'
map keys terminated by ':'
lines terminated by '\n'; 
====================================================================================
hive> load data local inpath '/root/a.txt' overwrite into table psn partition(age=10);
Loading data to table default.psn partition (age=10)
OK
Time taken: 3.817 seconds
=================================================================================
hive> select * from psn;
OK
1   zhang3  ["sing","tennis","running"] {"beijing":"daxing"}    10
2   li4 ["sing","pingpong","swim"]  {"shanghai":"baoshan"}  10
3   wang5   ["read","joke","football"]  {"guangzou":"baiyun"}   10
==============================================================================

需求：
    一次性统计每种爱好出现了多少次，每个城市出现了多少次，每个区出现多少次。

分析：
    这个需求有点像hive实现wordcount案例，或者说它就是两个wc案例的聚合，只不过现在这个不用split。
    在wc案例中，我们使用explode完美地解决了一列记录wc操作。
    但是在hive中的udtf函数(split/explode)中，select子句只能单独出现一个udtf函数，且udtf函数不能与其它字段和函数一并使用。
    #####只能select explode(..) from emp;
    #####不能select explode(..), explode(..) from emp;
    #####不能select id,explode(..) from emp;
    这就会造成对于一些复杂逻辑就会出现无法处理的问题，就比如以上这个两列记录的wc操作。
    这时候就需要用到lateral view了,它可以将udtf函数产生的多行结果组织成一张虚拟表。

===================================================================================
hive> select count(distinct c1),count(distinct c2),count(distinct c3)from psn 
    >lateral view explode(likes)t1 as c1 
    >lateral view explode(address)t2 as c2,c3;

#####t1和t2为经过udtf函数产生的虚拟表的表名，c1/c2/c3为字段别名
#####数组经过explode会产生一列数据，map集合产生两列。

Hadoop job information for Stage-1: number of mappers: 1; number of reducers: 1
2019-04-24 22:59:16,471 Stage-1 map = 0%,  reduce = 0%
2019-04-24 22:59:25,681 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 1.76 sec
2019-04-24 22:59:36,268 Stage-1 map = 100%,  reduce = 100%, Cumulative CPU 4.15 sec
MapReduce Total cumulative CPU time: 4 seconds 150 msec
Ended Job = job_1556088929464_0004
MapReduce Jobs Launched: 
Stage-Stage-1: Map: 1  Reduce: 1   Cumulative CPU: 4.15 sec   HDFS Read: 14429 HDFS Write: 105 SUCCESS
Total MapReduce CPU Time Spent: 4 seconds 150 msec
OK
8   3   3
Time taken: 35.986 seconds, Fetched: 1 row(s)
=============================================================================

文章名称：hive中的lateralview结合udtf函数的使用解决生产问题
网站网址：http://azwzsj.com/article/ieghgd.html