Merge pull request 'beta' (#218) from antonis.lempesis/dnet-hadoop:beta into beta

Reviewed-on: #218

dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDB.sql

@@ -107,11 +107,6 @@ compute stats TARGET.result_sources;
 create table TARGET.result_topics stored as parquet as select * from SOURCE.result_topics orig where exists (select 1 from TARGET.result r where r.id=orig.id);
 compute stats TARGET.result_topics;
 
-create table TARGET.result_apc stored as parquet as select * from SOURCE.result_apc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
-compute stats TARGET.result_apc;
-
-
-
 create view TARGET.foo1 as select * from SOURCE.result_result rr where rr.source in (select id from TARGET.result);
 create view TARGET.foo2 as select * from SOURCE.result_result rr where rr.target in (select id from TARGET.result);
 create table TARGET.result_result STORED AS PARQUET as select distinct * from (select * from TARGET.foo1 union all select * from TARGET.foo2) foufou;

dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step7.sql

@@ -127,6 +127,7 @@ CREATE TABLE ${stats_db_name}.result_organization STORED AS PARQUET AS
 SELECT substr(r.target, 4) AS id, substr(r.source, 4) AS organization
 FROM ${openaire_db_name}.relation r
 WHERE r.reltype = 'resultOrganization'
+  and r.target like '50|%'
   and r.datainfo.deletedbyinference = false and r.datainfo.invisible=false;
 
 CREATE TABLE ${stats_db_name}.result_projects STORED AS PARQUET AS

dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step8.sql

@@ -93,7 +93,7 @@ where d.datainfo.deletedbyinference=false and d.datainfo.invisible=false;
 CREATE TABLE ${stats_db_name}.datasource_organizations STORED AS PARQUET AS
 SELECT substr(r.target, 4) AS id, substr(r.source, 4) AS organization
 FROM ${openaire_db_name}.relation r
-WHERE r.reltype = 'datasourceOrganization' and r.datainfo.deletedbyinference = false and r.datainfo.invisible=false;
+WHERE r.reltype = 'datasourceOrganization' and r.datainfo.deletedbyinference = false and r.target like '20|%' and r.datainfo.invisible=false;
 
 -- datasource sources:
 -- where the datasource info have been collected from.