From 4b7c8dff2dec9fae75f73dca36b2f3c87de4441f Mon Sep 17 00:00:00 2001 From: antleb Date: Thu, 14 Oct 2021 18:53:35 +0300 Subject: [PATCH 1/2] fetching affiliated results for 4 orgs in monitor. fixed affiliated orgs in stats db --- .../oa/graph/stats/oozie_app/scripts/step20-createMonitorDB.sql | 2 +- .../eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step7.sql | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDB.sql b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDB.sql index 9ea50d4886..d39af37d5f 100644 --- a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDB.sql +++ b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDB.sql @@ -17,7 +17,7 @@ create table TARGET.result as union all select * from SOURCE.result r where exists (select 1 from SOURCE.result_concepts rc where rc.id=r.id) union all - select * from SOURCE.result r where exists (select 1 from SOURCE.result_projects rp join SOURCE.project p on p.id=rp.project join SOURCE.project_organizations po on po.id=p.id where rp.id=r.id and po.organization in ( + select * from SOURCE.result r where exists (select 1 from SOURCE.result_organization ro where ro.id=r.id and ro.organization in ( 'openorgs____::759d59f05d77188faee99b7493b46805', 'openorgs____::b84450f9864182c67b8611b5593f4250', 'openorgs____::d41cf6bd4ab1b1362a44397e0b95c975', diff --git a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step7.sql b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step7.sql index b3cbc9b419..d461bd343e 100644 --- a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step7.sql +++ b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step7.sql @@ -126,7 +126,7 @@ FROM ${stats_db_name}.otherresearchproduct_topics; CREATE TABLE ${stats_db_name}.result_organization AS SELECT substr(r.target, 4) AS id, substr(r.source, 4) AS organization FROM ${openaire_db_name}.relation r -WHERE r.reltype = 'resultOrganization' +WHERE r.reltype = 'resultOrganization' and r.relclass='hasAuthorInstitution' and r.subreltype='affiliation' and r.datainfo.deletedbyinference = false; CREATE TABLE ${stats_db_name}.result_projects AS From 41ecb1eb61e2d7a4285d60749c1ee0146b3436f9 Mon Sep 17 00:00:00 2001 From: antleb Date: Fri, 15 Oct 2021 13:42:55 +0300 Subject: [PATCH 2/2] invalidating medatadata before context thingies --- .../eu/dnetlib/dhp/oa/graph/stats/oozie_app/contexts.sh | 2 ++ 1 file changed, 2 insertions(+) diff --git a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/contexts.sh b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/contexts.sh index 6d42ab13d4..2a40eb9204 100644 --- a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/contexts.sh +++ b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/contexts.sh @@ -26,6 +26,8 @@ hdfs dfs -copyFromLocal concepts.csv ${TMP} hdfs dfs -chmod -R 777 ${TMP} echo "Creating and populating impala tables" +impala-shell -q "invalidate metadata" +impala-shell -d ${TARGET_DB} -q "invalidate metadata" impala-shell -q "create table ${TARGET_DB}.context (id string, name string) row format delimited fields terminated by ','" impala-shell -q "create table ${TARGET_DB}.category (context string, id string, name string) row format delimited fields terminated by ','" impala-shell -q "create table ${TARGET_DB}.concept (category string, id string, name string) row format delimited fields terminated by ','"