Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Pig >> mail # user >> Manually build tuple from three group relations


Copy link to this message
-
RE: Manually build tuple from three group relations
You could use two rounds of the outer join/filter by null idiom. For example after the first round you would get allTermsMinusNonNumbers like this:

grunt> sh cat allTerms
aa
bb
cc
11
22
33
grunt> sh cat nonNumbers
cc
grunt> allTerms = load 'allTerms' as (term:chararray);
grunt> nonNumbers = load 'nonNumbers' as (term:chararray);
grunt> j1 = join allTerms by term left outer, nonNumbers by term;
grunt> allTermsMinusNonNumbers = filter j1 by nonNumbers::term is null;    
grunt>
grunt> dump allTermsMinusNonNumbers

(11,)
(22,)
(33,)
(aa,)

William F Dowling
Sr Technical Specialist, Software Engineering
Thomson Reuters
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] On Behalf Of John Conwell
Sent: Wednesday, July 06, 2011 6:28 PM
To: [EMAIL PROTECTED]
Subject: Manually build tuple from three group relations

I have a dataset where each tupple is a term.  I then do two filter
operations, to find all terms that have numbers, then all terms that dont
have numbers.

Oddly, there are some terms that dont fit into either group (not really sure
how).  So at this point I have 3 bags, all terms, tems with numbers, and
terms without numbers.

What I'm trying to find out is what terms are in the list of all terms, but
are not in either of the two filtered bags.  I thought I'd use the DIFF
function, but it only operates on different tuples in the same bag.  So
somehow I think I need to crate a new relation, that has three tuples at the
same level (row?).  Then I could use the DIFF function.

Any ideas?

The script I have so far is shown below...

   terms = LOAD 'terms' AS (term:chararray, min:float, max:float,
count:int);

terms = FOREACH terms GENERATE term;

--bag of all terms

allTerms = GROUP terms ALL;

--bag of terms without numbers

nonNumbers = FILTER terms BY NOT (term MATCHES '^.*[0-9].*$');

nonNumbers = GROUP nonNumbers ALL;

--bag of terms with numbers

withNumbers = FILTER terms BY (term MATCHES '^.*[0-9].*$');

withNumbers = GROUP withNumbers ALL;