pig:group by之后的其它统计方法一

 --测试Top N后的其它统计
 A = LOAD '/TraceParser/blackcore/' USING PigStorage() as (lk_id:chararray,host:chararray);
 --DUMP A;

 B = GROUP A BY lk_id;
 B = FOREACH B GENERATE group as lk_id,COUNT($1) as amount;
 --DUMP B;

 --TOP N的数据集
 C = ORDER B BY amount DESC;
 C = LIMIT C 5;
 --DUMP C;

 --除TOP N后的数据集归并为其它
 D = JOIN B BY lk_id LEFT OUTER,C BY lk_id;
 --DESCRIBE D;
 --DUMP D;
 E = FILTER D BY C::lk_id is null OR C::lk_id == '';
 --DUMP E;
 E = FOREACH E GENERATE B::lk_id,B::amount;
 --DESCRIBE E;
 --SUM的用法？

 --DUMP E;
 --DUMP A;
 H = JOIN E BY B::lk_id,A BY lk_id;
 DESCRIBE H;
 --H: {E::B::lk_id: chararray,E::B::amount: long,A::lk_id: chararray,A::host: chararray}
 --DUMP H;
 I = GROUP H all;
 I = FOREACH I GENERATE '其它',COUNT($1) as amount;
 DUMP I;