Hive中的表分析函数接受零个或多个输入，然后产生多列或多行输出。

1.explode函数

explode函数以array类型数据输入，然后对数组中的数据进行迭代，返回多行结果，一行一个数组元素值

ARRAY函数是将一列输入转换成一个数组输出。

hive (jimdb)> SELECT ARRAY(1,2,3) FROM dual;
OK
_c0
[1,2,3]
Time taken: 0.448 seconds, Fetched: 1 row(s)

SELECT explode(array(1,2,3)) AS element;

Time taken: 0.327 seconds, Fetched: 3 row(s)

创建一张测试表单：

CREATE TABLE udtf_test(name STRING,subordinates ARRAY<STRING>)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY '\t'
COLLECTION ITEMS TERMINATED BY ',';

hive (jimdb)> select * from udtf_test;
OK
udtf_test.name udtf_test.subordinates
jim5 ["james","datacloase"]
jim4 ["james","datacloase"]
jim3 ["james","datacloase"]
jim2 ["james","datacloase"]
jim ["james","datacloase"]
Time taken: 0.348 seconds, Fetched: 5 row(s)

我执行下面这条语句，希望将字段subordinates拆分开，新生成一列，但是语句报错：

select name,explode(subordinates) from udtf_test;

hive (jimdb)> select name,explode(subordinates) from udtf_test;
FAILED: SemanticException [Error 10081]: UDTF's are not supported outside the SELECT clause, nor nested in expressions

Hive的表生成函数只是生成了一种数据的展示方式，而无法在表中产生一个其他的列。

因此这块需要使用LATERAL VIEW功能来进行处理。LATERAL VIEW将explode生成的结果当做一个视图来处理。

SELECT name, sub 
FROM udtf_test
LATERAL VIEW explode(subordinates) subView AS sub;

在这里LATERAL VIEW 是将 explode结果转换成一个视图subView,在视图中的单列列名定义为sub，然后在查询的时候引用这个列名就能够查到。

hive (jimdb)> SELECT name, sub 
> FROM udtf_test
> LATERAL VIEW explode(subordinates) subView AS sub;
OK
name sub
jim5 james
jim5 datacloase
jim4 james
jim4 datacloase
jim3 james
jim3 datacloase
jim2 james
jim2 datacloase
jim james
jim datacloase
Time taken: 0.399 seconds, Fetched: 10 row(s)

创建一张测试表：

drop table test1;
create table test1(name string,phonenumber string)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY '\t';

--需求是过滤掉该表中电话号码中0-9的某个数字在电话号码中出现6次及以上的号码，然后将正常的号码返回。

hive (jimdb)> select * from test1;
OK
test1.name test1.phonenumber
'jim he' '18191512076'
'xiaosong' '18392988059'
'jingxianghua' '18118818818'
'donghualing' '17191919999'

执行语句如下：

SELECT c.name,c.phonenumber
FROM 
(SELECT dd.name,dd.phonenumber,MAX(dd.cn) 
FROM (SELECT d.name,d.phonenumber,d.m, COUNT(*) cn
FROM (SELECT name,phonenumber,m FROM test1 LATERAL VIEW explode(split(phonenumber,'')) n AS m) d 
GROUP BY d.name,d.phonenumber,d.m) dd
GROUP BY dd.name,dd.phonenumber HAVING MAX(dd.cn) <=5) c;

hive (jimdb)> SELECT c.name,c.phonenumber
> FROM 
> (SELECT dd.name,dd.phonenumber,MAX(dd.cn) 
> FROM (SELECT d.name,d.phonenumber,d.m, COUNT(*) cn
> FROM (SELECT name,phonenumber,m FROM test1 LATERAL VIEW explode(split(phonenumber,'')) n AS m) d 
> GROUP BY d.name,d.phonenumber,d.m) dd
> GROUP BY dd.name,dd.phonenumber HAVING MAX(dd.cn) <=5) c;
Query ID = hadoop_20180611200632_14d3d30b-e64f-4aee-a7ca-fffa66049890
Total jobs = 1
Launching Job 1 out of 1
Number of reduce tasks not specified. Estimated from input data size: 1
In order to change the average load for a reducer (in bytes):
set hive.exec.reducers.bytes.per.reducer=<number>
In order to limit the maximum number of reducers:
set hive.exec.reducers.max=<number>
In order to set a constant number of reducers:
set mapreduce.job.reduces=<number>
Job running in-process (local Hadoop)
2018-06-11 20:06:35,732 Stage-1 map = 100%, reduce = 100%
Ended Job = job_local1118441439_0004
MapReduce Jobs Launched: 
Stage-Stage-1: HDFS Read: 3004 HDFS Write: 0 SUCCESS
Total MapReduce CPU Time Spent: 0 msec
OK
c.name c.phonenumber
'jim he' '18191512076'
'xiaosong' '18392988059'
Time taken: 2.872 seconds, Fetched: 2 row(s)