Pig Lab3

load, foreach , 

3) dump :-
    
    to execute data flow.
   writes output into console.

    entire dataflow will be converted as single map reduce job.

   ex:

  grunt> dump res;

______________________________________________

4) store:-

    to execute data flow.
   writes output into disk (local/hdfs)

   grunt> store res into 'results1';
    -- tab is delimiter.
         
   grunt> store res into 'results2' 
          using PigStorage(',');

   load and store operators use Storage Methods.
_______________________________________
     
5) filter :-

      to create subsets , based on given criteria.

     (row filter, equivalant to 'where' clause of sql select statement )

 grunt> e1 = filter emp by (sex=='m');
grunt> dump e1

grunt> e1 = filter emp by (sex=='m' and dno==12)
>> ;
grunt> dump e1
   
___________________________________________

6) Limit :-

    to fetch top n number of tuples.

 grunt> top3 = limit emp 3;
grunt> dump top3

(101,vino,26000,m,11)
(102,Sri,25000,f,11)
(103,mohan,13000,m,13)

7) sample:-

     to create subsets, in random sample style.

(102,Sri,25000,f,11)
(104,lokitha,8000,f,12)
(203,ccc,10000,f,13)

grunt> rs = sample emp 0.5;
grunt> dump rs      

_________________________________________________

8) Aggregated functions in PigLatin.

   SUM(), AVG(), MAX(), MIN(), COUNT()

   
grunt> r = foreach emp generate SUM(sal) as tot;
grunt> dump r

   ABOVE statement will be failed during execution,.

    bcoz, AGGREGATED functions are applied only on inner bags.
   when you group data , inner bags will be produced.

_________________________________________

9) group: -
      to get inner bags foreach data group.
   based on grouping field.

grunt> describe emp;
emp: {id: int,name: chararray,sal: int,sex: chararray,dno: int}
grunt> -- select sex, sum(sal) from emp group by sex
grunt> e = foreach emp generate sex, sal;
grunt> bySex = group e by sex;
grunt> describe bySex
bySex: {group: chararray,e: {sex: chararray,sal: int}}
grunt> dump bySex

grunt> res = foreach bySex generate
>>         group as sex, SUM(e.sal) as tot;
grunt> describe res
res: {sex: chararray,tot: long}
grunt> store res into 'myhdfs1';

grunt> cat myhdfs1/part-r-00000
f       103000
m       125000
____________________________________


grunt> describe emp
emp: {id: int,name: chararray,sal: int,sex: chararray,dno: int}
grunt> ee = foreach emp generate dno, sal;
grunt> byDno = group ee by dno;
grunt> res = foreach byDno generate 
>>      group as dno, SUM(ee.sal) as tot;
grunt> store res into 'pdemo/res1';

grunt> ls pdemo/res1
hdfs://localhost/user/training/pdemo/res1/_logs <dir>
hdfs://localhost/user/training/pdemo/res1/part-r-00000<r 1>       28
grunt> cat pdemo/res1/part-r-00000
11      51000
12      48000
13      129000
grunt> 

grunt> -- single grouping and multiple aggregations
grunt> res1 = foreach bySex generate
>>    group as sex, 
>>     SUM(e.sal) as tot,
>>    AVG(e.sal) as avg,
>>    MAX(e.sal) as min,
>>    MIN(e.sal) as mn,
>>    COUNT(e) as cnt;
grunt> dump res1
(f,103000,20600.0,50000,8000,5)
(m,125000,25000.0,50000,6000,5)

_________________________________

grunt> -- multi grouping..
grunt> e = foreach emp generate dno, sex, sal;
grunt> grp = group e by dno, sex;
2016-06-09 19:28:35,893 [main] ERROR org.apache.pig.tools.grunt.Grunt - ERROR 1000: Error during parsing. Unrecognized alias sex
Details at logfile: /home/training/pig_1465437040017.log
grunt> 
  above statement is invalid.
  pig does not allow groping by multiple fields.

  solution:
   make multiple fields as a tuple field, and group it by tuple.

grunt> grp = group e by (dno, sex);
grunt> describe grp
grp: {group: (dno: int,sex: chararray),e: {dno: int,sex: chararray,sal: int}}
grunt> res = foreach grp generate 
>>     group.dno, group.sex, SUM(e.sal) as tot; 
grunt> dump res
(11,f,25000)
(11,m,26000)
(12,f,18000)
(12,m,30000)
(13,f,60000)
(13,m,69000)
_________________________________
grunt> -- select sum(sal) from emp;
grunt> -- old one
grunt> e = foreach emp generate 'ibm' as org, sal;
grunt> dump e; 

(ibm,26000)
(ibm,25000)
(ibm,13000)
(ibm,8000)
(ibm,6000)
(ibm,10000)
(ibm,30000)
(ibm,50000)
(ibm,10000)
(ibm,50000)

grunt> grp = group e by org;
grunt> res = foreach grp generate 
>>         SUM(e.sal) as tot;
grunt> dump res
(228000)

_____________________________________

 2nd one --- for entire column aggregation.

grunt> describe emp;
emp: {id: int,name: chararray,sal: int,sex: chararray,dno: int}
grunt> e = foreach emp generate sal;
grunt> grp = group e all;
grunt> dump grp

(all,{(26000),(25000),(13000),(8000),(6000),(10000),(30000),(50000),(10000),(50000)})

grunt> res = foreach grp generate 
>>     SUM(e.sal) as tot, 
>>    AVG(e.sal) as avg, MAX(e.sal) as max,
>>    MIN(e.sal) as min, COUNT(e) as cnt;
grunt> dump res
(228000,22800.0,50000,6000,10)
_____________________________________________
Assignment:

  describe emp--> id,name,sal,sex,dno

________________________________
    Marketing A 30
    marketing   B       40
     :
     :
    Fin A 50
     :
    Fin D       30
_________________________

____________________________________________________

Cogroup: