phoenix的SQL查询

一.NULL值的处理——以FLUX字段为例
1.当遇到聚合时，可能为null值，可以用COALESCE(FLUX,0)将null值赋值为0

select URL,sum(COALESCE(FLUX,0)) as sf,sum(COALESCE(COUNT,0)) as sc from URLDAY 
where  DT>='20200510000000' and DT<='20200517000000'  
group by URL 
order by  sf desc 
limit 10

2.排序时，如果有null值，倒序会排在数值前面，此时可以将COALESCE(FLUX,0)放在order by 的之后，让排序正常。

select URL,sum(FLUX) as sf,sum(COUNT) as sc from URLDAY 
where  DT>='20200510000000' and DT<='20200517000000'  
group by URL 
order by  COALESCE(sf,0) desc 
limit 10

3.如果在where条件里过滤null值，则需要FLUX is not null

select URL,sum(FLUX) as sf,sum(COUNT) as sc from URLDAY 
where  DT>='20200510000000' and DT<='20200517000000'  and FLUX is not null
group by URL 
order by  sf desc 
limit 10

二.查询
1.打开sqlline

/usr/lib/phoenix-4.14.1/bin/sqlline.py 10.21.69.1,10.21.69.2,10.21.69.3:2181:/hbase98

sqlline

2.查看表信息

!describe CHANNEL5MINSPEED_2005

describe

3.退出sqlline

!exit

exit

三.重点函数
1.字符串相关
a.字符串截取函数
SUBSTR(字符串，起始位置，长度)

select SUBSTR(DT,1,10) from CHANNEL5MINSPEED_2005;

substr

b.去除空格
去除前后空格 TRIM(" a ")
去除最左空格 LTRIM(" b")
去除最右空格 RTRIM("b ")
c.填充字符
左边填充：LPAD（字符串，字符数，填充字符）
不要举一反三，没有RPAD，hhhhh

select LPAD(DT,20,'0') from CHANNEL5MINSPEED_2005;

LPAD

d.字符长度
LENGTH（“aaaa”）

select LENGTH(DT) from CHANNEL5MINSPEED_2005;

LENGTH

e.正则匹配
正则截取
REGEXP_SUBSTR（字符串，正则表达式，选取第几个）。省略第三个参数时，默认取第一个

正则替换
REGEXP_REPLACE（字符串，正则表达式，替换字符或字符串）

正则分割
REGEXP_SPLIT(字符串, 分割符)

重点：当分割字符相连时，会出现null值。

select REGEXP_SPLIT(DT,'5') from CHANNEL5MINSPEED_2005;
select REGEXP_SPLIT(DT,'0') from CHANNEL5MINSPEED_2005;

REGEXP_SPLIT