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

Switch to Plain View
Pig, mail # user - Some optimization advices


Copy link to this message
-
Some optimization advices
Jerome Pierson 2013-01-31, 17:19
Hi There,

I am a beginner, I achieved something, but I guess I could have done
better. Let me explain.
(Pig 0.10)

My data is DESCRIBE as :

  xmlToTuple: {(node_attr_id: int,node_attr_lon:
chararray,node_attr_lat: chararray,tag: {(tag_attr_k:
chararray,tag_attr_v: chararray)})}
and DUMP like this :

((100312088,45.2745669,-12.7776222,{(created_by,JOSM)}))
((100948454,45.2620946,-12.7849171,))
((100948519,45.2356985,-12.7707014,{(created_by,JOSM)}))
((704398904,45.2416667,-13.0058333,{(lat,-13.00583333),(lon,45.24166667)}))
((1230941976,45.0743117,-12.6888807,{(place,village)}))
((1230941977,45.0832807,-12.6810328,{(name,Mtsahara)}))
((1976927219,45.2272263,-12.7794359,))
((1751057677,45.2216163,-12.7825896,{(amenity,fast_food),(name,Brochetterie)}))
((1751057678,45.2216953,-12.7829678,{(amenity,fast_food),(name,Brochetterie)}))
((100948360,45.2338541,-12.7762230,{(amenity,ferry_terminal)}))
((362795028,45.2086809,-12.8062991,{(amenity,fuel),(operator,Total)}))

I want to extract the record which have a certain value for the
tag_attr_k field. For example, give me the record where there is a
tag_attr_k = amesity ? That should be :

(100948360,-12.7762230,45.2338541,{(amenity,ferry_terminal)})
(362795028,-12.8062991,45.2086809,{(operator,Total),(amenity,fuel)})
(1751057677,-12.7825896,45.2216163,{(amenity,fast_food),(name,Brochetterie)})
(1751057678,-12.7829678,45.2216953,{(amenity,fast_food),(name,Brochetterie)})

So (node_attr_id, node_attr_lat , node_attr_lon,{(tag_attr_k,
tag_attr_v)...(tag_attr_k,tag_attr_v)}

I ended up with this script.
...
XmlTag = foreach xmlToTuple GENERATE FLATTEN ($0); --removed top
including level bag
XmlTag2 = foreach XmlTag GENERATE $0 as id, $1 as lon, $2 as lat,
FLATTEN (tag) as (key, value); --flatten the bag of tags
XmlTag3 =  FILTER XmlTag2 BY key == 'amenity'; -- get all the records
with amenity tags
XmlTag4 = JOIN XmlTag3 BY id, XmlTag2 BY id; -- re-build records with
all tags containing amenity tag
XmlTag7 = foreach XmlTag4 GENERATE $0 as id,$1 as lon, $2 as lat,$8 as
key, $9 as value; -- re-build records : removing redundant field
XmlTag5 = GROUP XmlTag7 BY (id,lat,lon); -- re-build records : grouping
redundant records
XmlTag8 = foreach XmlTag5 { --rebuild records : id,lat,long
{(key,value)...(key,value)}
         tag = foreach XmlTag7 GENERATE  key, value;
     GENERATE group.id,group.lat,group.lon,tag;
};

Using this variable:

xmlToTuple: {(node_attr_id: int,node_attr_lon: chararray,node_attr_lat:
chararray,tag: {(tag_attr_k: chararray,tag_attr_v: chararray)})}
XmlTag: {null::node_attr_id: int,null::node_attr_lon:
chararray,null::node_attr_lat: chararray,null::tag: {(tag_attr_k:
chararray,tag_attr_v: chararray)}}
XmlTag2: {id: int,lon: chararray,lat: chararray,key: chararray,value:
chararray}
XmlTag3: {id: int,lon: chararray,lat: chararray,key: chararray,value:
chararray}
XmlTag4: {XmlTag3::id: int,XmlTag3::lon: chararray,XmlTag3::lat:
chararray,XmlTag3::key: chararray,XmlTag3::value: chararray,XmlTag2::id:
int,XmlTag2::lon: chararray,XmlTag2::lat: chararray,XmlTag2::key:
chararray,XmlTag2::value: chararray}
XmlTag7: {id: int,lon: chararray,lat: chararray,key: chararray,value:
chararray}
XmlTag5: {group: (id: int,lat: chararray,lon: chararray),XmlTag7: {(id:
int,lon: chararray,lat: chararray,key: chararray,value: chararray)}}
XmlTag8: {id: int,lat: chararray,lon: chararray,tag: {(key:
chararray,value: chararray)}}
I guess this not very straightforward and can be largely optimized.
Please give me some hints ?

Regards,
J�r�me
+
Cheolsoo Park 2013-01-31, 19:45
+
Jonathan Coveney 2013-01-31, 23:27
+
Jerome Pierson 2013-02-05, 16:06
+
Cheolsoo Park 2013-02-05, 22:13
+
Jerome Person 2013-02-05, 22:57
+
Prashant Kommireddi 2013-02-05, 23:10
+
Jerome Person 2013-02-06, 10:00
+
Cheolsoo Park 2013-02-06, 16:41
+
Jerome Person 2013-02-06, 16:55
+
psic 2013-02-05, 22:57