features

All features following require ES with groovy script enabled.

• Distinct precision_threshold depends on ES or you can specify by second parameters.
  When you have lot of shards, 40000 consume too much memory.
• select,groupBy now support functions and field alias
• nested function is also available,eg. split(substring('newtype',0,3),'c')[0]
• Binary operation support now, eg. floor(substring(newtype,0,14)/100)/5)*5

pom.xml changes

Profile local and online are added. So we can choose profile we want in IDE eg. Intel Idea.
local make you run Application without dependency problem since the scope of some dependences is setted to provied.

MainTestSuite changes

Cause the unit test depend on ES instance, it make confuse sometimes that we just can not connect local ES instance just because the cluster_name is not consistent。

Settings settings = Settings.builder().put("client.transport.ignore_cluster_name",true).build();

functions support

• floor
• split
• trim
• log
• log10
• substring
• round
• sqrt
• concat_ws
• *
• /
• %

Unit Test

org.nlpcn.es4sql.SQLFunctionsTest

Example

check Example file:

org.nlpcn.es4sql.Test

SQLs:

SELECT newtype as nt  from  twitter2 

SELECT sum(num) as num2,newtype as nt  
from  twitter2 
group by nt  order by num2 

SELECT sum(num_d) as num2,split(newtype,',') as nt  
from  twitter2 
group by nt  
order by num2

SELECT sum(num_d) as num2,floor(num) as nt  
from  twitter2 
group by floor(num),newtype  
order by num2

SELECT split('newtype','b')[1] as nt,sum(num_d) as num2   
from  twitter2 
group by nt

SELECT split(substring('newtype',0,3),'c')[0] as nt,num_d   
from  twitter2 
group by nt

SELECT trim(newtype) as nt 
from  twitter2


SELECT floor(floor(substring(time,0,14)/100)/5)*5 as nt,
count(distinct(mid)) as cvalue 
FROM twitter2  
where ty='buffer' and day='20160815' and domain='baidu.com' 
group by nt 
order by cvalue