diff --git a/dhp-workflows/dhp-monitor-update/pom.xml b/dhp-workflows/dhp-monitor-update/pom.xml new file mode 100644 index 000000000..ca0bb9837 --- /dev/null +++ b/dhp-workflows/dhp-monitor-update/pom.xml @@ -0,0 +1,32 @@ + + + + dhp-workflows + eu.dnetlib.dhp + 1.2.4-SNAPSHOT + + 4.0.0 + dhp-monitor-update + + + org.apache.spark + spark-core_2.11 + + + org.apache.spark + spark-sql_2.11 + + + + + + pl.project13.maven + git-commit-id-plugin + 2.1.11 + + false + + + + + diff --git a/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/scripts/createMonitorDB.sql b/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/scripts/createMonitorDB.sql index 63085e7a1..ccbe8cab8 100644 --- a/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/scripts/createMonitorDB.sql +++ b/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/scripts/createMonitorDB.sql @@ -1,5 +1,6 @@ DROP TABLE IF EXISTS TARGET.result_new; +<<<<<<< HEAD create table TARGET.result_new stored as parquet as select distinct * from ( select * from SOURCE.result r where exists (select 1 from SOURCE.result_organization ro where ro.id=r.id and ro.organization in ( @@ -76,18 +77,103 @@ ANALYZE TABLE TARGET.result_topics COMPUTE STATISTICS; INSERT INTO TARGET.result_fos select * from TARGET.result_fos orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); ANALYZE TABLE TARGET.result_fos COMPUTE STATISTICS; +======= +create table TARGET.result_new as + select distinct * from ( + select * from SOURCE.result r where exists (select 1 from SOURCE.result_organization ro where ro.id=r.id and ro.organization in ( +-- 'openorgs____::b8b8ca674452579f3f593d9f5e557483', -- University College Cork +-- 'openorgs____::38d7097854736583dde879d12dacafca' -- Brown University + 'openorgs____::57784c9e047e826fefdb1ef816120d92' --Arts et Métiers ParisTech + ) )) foo; + +COMPUTE STATS TARGET.result_new; + +INSERT INTO TARGET.result select * from TARGET.result_new; +COMPUTE STATS TARGET.result; + +INSERT INTO TARGET.result_citations select * from TARGET.result_citations orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_citations; + +INSERT INTO TARGET.result_references_oc select * from TARGET.result_references_oc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_references_oc; + +INSERT INTO TARGET.result_citations_oc select * from TARGET.result_citations_oc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_citations_oc; + +INSERT INTO TARGET.result_classifications select * from TARGET.result_classifications orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_classifications; + +INSERT INTO TARGET.result_apc select * from TARGET.result_apc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_apc; + +INSERT INTO TARGET.result_concepts select * from TARGET.result_concepts orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_concepts; + +INSERT INTO TARGET.result_datasources select * from TARGET.result_datasources orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_datasources; + +INSERT INTO TARGET.result_fundercount select * from TARGET.result_fundercount orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_fundercount; + +INSERT INTO TARGET.result_gold select * from TARGET.result_gold orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_gold; + +INSERT INTO TARGET.result_greenoa select * from TARGET.result_greenoa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_greenoa; + +INSERT INTO TARGET.result_languages select * from TARGET.result_languages orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_languages; + +INSERT INTO TARGET.result_licenses select * from TARGET.result_licenses orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_licenses; + +INSERT INTO TARGET.result_oids select * from TARGET.result_oids orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_oids; + +INSERT INTO TARGET.result_organization select * from TARGET.result_organization orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_organization; + +INSERT INTO TARGET.result_peerreviewed select * from TARGET.result_peerreviewed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_peerreviewed; + +INSERT INTO TARGET.result_pids select * from TARGET.result_pids orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_pids; + +INSERT INTO TARGET.result_projectcount select * from TARGET.result_projectcount orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_projectcount; + +INSERT INTO TARGET.result_projects select * from TARGET.result_projects orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_projects; + +INSERT INTO TARGET.result_refereed select * from TARGET.result_refereed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_refereed; + +INSERT INTO TARGET.result_sources select * from TARGET.result_sources orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_sources; + +INSERT INTO TARGET.result_topics select * from TARGET.result_topics orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_topics; + +INSERT INTO TARGET.result_fos select * from TARGET.result_fos orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.result_fos; +>>>>>>> beta create view TARGET.foo1 as select * from TARGET.result_result rr where rr.source in (select id from TARGET.result_new); create view TARGET.foo2 as select * from TARGET.result_result rr where rr.target in (select id from TARGET.result_new); INSERT INTO TARGET.result_result select distinct * from (select * from TARGET.foo1 union all select * from TARGET.foo2) foufou; drop view TARGET.foo1; drop view TARGET.foo2; +<<<<<<< HEAD ANALYZE TABLE TARGET.result_result COMPUTE STATISTICS; +======= +COMPUTE STATS TARGET.result_result; +>>>>>>> beta -- indicators -- Sprint 1 ---- INSERT INTO TARGET.indi_pub_green_oa select * from TARGET.indi_pub_green_oa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +<<<<<<< HEAD ANALYZE TABLE TARGET.indi_pub_green_oa COMPUTE STATISTICS; INSERT INTO TARGET.indi_pub_grey_lit select * from TARGET.indi_pub_grey_lit orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); ANALYZE TABLE TARGET.indi_pub_grey_lit COMPUTE STATISTICS; @@ -138,6 +224,58 @@ INSERT INTO TARGET.indi_pub_in_subscribed select * from TARGET.indi_pub_in_subsc ANALYZE TABLE TARGET.indi_pub_in_subscribed COMPUTE STATISTICS; INSERT INTO TARGET.indi_result_with_pid select * from TARGET.indi_result_with_pid orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); ANALYZE TABLE TARGET.indi_result_with_pid COMPUTE STATISTICS; +======= +COMPUTE STATS TARGET.indi_pub_green_oa; +INSERT INTO TARGET.indi_pub_grey_lit select * from TARGET.indi_pub_grey_lit orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_pub_grey_lit; +INSERT INTO TARGET.indi_pub_doi_from_crossref select * from TARGET.indi_pub_doi_from_crossref orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_pub_doi_from_crossref; +-- Sprint 2 ---- +INSERT INTO TARGET.indi_result_has_cc_licence select * from TARGET.indi_result_has_cc_licence orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_result_has_cc_licence; +INSERT INTO TARGET.indi_result_has_cc_licence_url select * from TARGET.indi_result_has_cc_licence_url orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_result_has_cc_licence_url; +INSERT INTO TARGET.indi_pub_has_abstract select * from TARGET.indi_pub_has_abstract orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_pub_has_abstract; +INSERT INTO TARGET.indi_result_with_orcid select * from TARGET.indi_result_with_orcid orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_result_with_orcid; +---- Sprint 3 ---- +INSERT INTO TARGET.indi_funded_result_with_fundref select * from TARGET.indi_funded_result_with_fundref orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_funded_result_with_fundref; +---- Sprint 4 ---- +INSERT INTO TARGET.indi_pub_diamond select * from TARGET.indi_pub_diamond orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_pub_diamond; +INSERT INTO TARGET.indi_pub_in_transformative select * from TARGET.indi_pub_in_transformative orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_pub_in_transformative; +INSERT INTO TARGET.indi_pub_closed_other_open select * from TARGET.indi_pub_closed_other_open orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_pub_closed_other_open; +---- Sprint 5 ---- +INSERT INTO TARGET.indi_result_no_of_copies select * from TARGET.indi_result_no_of_copies orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_result_no_of_copies; +---- Sprint 6 ---- +INSERT INTO TARGET.indi_pub_hybrid_oa_with_cc select * from TARGET.indi_pub_hybrid_oa_with_cc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_pub_hybrid_oa_with_cc; +INSERT INTO TARGET.indi_pub_downloads select * from TARGET.indi_pub_downloads orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id); +COMPUTE STATS TARGET.indi_pub_downloads; +INSERT INTO TARGET.indi_pub_downloads_datasource select * from TARGET.indi_pub_downloads_datasource orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id); +COMPUTE STATS TARGET.indi_pub_downloads_datasource; +INSERT INTO TARGET.indi_pub_downloads_year select * from TARGET.indi_pub_downloads_year orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id); +COMPUTE STATS TARGET.indi_pub_downloads_year; +INSERT INTO TARGET.indi_pub_downloads_datasource_year select * from TARGET.indi_pub_downloads_datasource_year orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id); +COMPUTE STATS TARGET.indi_pub_downloads_datasource_year; +---- Sprint 7 ---- +INSERT INTO TARGET.indi_pub_gold_oa select * from TARGET.indi_pub_gold_oa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_pub_gold_oa; +INSERT INTO TARGET.indi_pub_hybrid select * from TARGET.indi_pub_hybrid orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_pub_hybrid; + +INSERT INTO TARGET.indi_pub_has_preprint select * from TARGET.indi_pub_has_preprint orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_pub_has_preprint; +INSERT INTO TARGET.indi_pub_in_subscribed select * from TARGET.indi_pub_in_subscribed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_pub_in_subscribed; +INSERT INTO TARGET.indi_result_with_pid select * from TARGET.indi_result_with_pid orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); +COMPUTE STATS TARGET.indi_result_with_pid; +>>>>>>> beta --create table TARGET.indi_datasets_gold_oa stored as parquet as select * from SOURCE.indi_datasets_gold_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id); --compute stats TARGET.indi_datasets_gold_oa; --create table TARGET.indi_software_gold_oa stored as parquet as select * from SOURCE.indi_software_gold_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id); diff --git a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step15_5.sql b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step15_5.sql index 61c0726ff..beec31082 100644 --- a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step15_5.sql +++ b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step15_5.sql @@ -29,12 +29,12 @@ select rcount.pid, sum(case when rcount.type='publication' then rcount.count els from rcount group by rcount.pid; -create view ${stats_db_name}.rndexpenditure as select * from ${external_stats_db_name}.rndexpediture; -create view ${stats_db_name}.rndgdpexpenditure as select * from ${external_stats_db_name}.rndgdpexpenditure; -create view ${stats_db_name}.doctoratestudents as select * from ${external_stats_db_name}.doctoratestudents; -create view ${stats_db_name}.totalresearchers as select * from ${external_stats_db_name}.totalresearchers; -create view ${stats_db_name}.totalresearchersft as select * from ${external_stats_db_name}.totalresearchersft; -create view ${stats_db_name}.hrrst as select * from ${external_stats_db_name}.hrrst; +create view ${stats_db_name}.rndexpenditure as select * from stats_ext.rndexpediture; +create view ${stats_db_name}.rndgdpexpenditure as select * from stats_ext.rndgdpexpenditure; +create view ${stats_db_name}.doctoratestudents as select * from stats_ext.doctoratestudents; +create view ${stats_db_name}.totalresearchers as select * from stats_ext.totalresearchers; +create view ${stats_db_name}.totalresearchersft as select * from stats_ext.totalresearchersft; +create view ${stats_db_name}.hrrst as select * from stats_ext.hrrst; create table ${stats_db_name}.result_instance stored as parquet as select distinct r.* diff --git a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/workflow.xml b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/workflow.xml index e9453d7b1..8d2e56380 100644 --- a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/workflow.xml @@ -102,7 +102,6 @@ ${wf:conf('resumeFrom') eq 'Step16_5'} ${wf:conf('resumeFrom') eq 'Step19-finalize'} ${wf:conf('resumeFrom') eq 'step20-createMonitorDB'} - ${wf:conf('resumeFrom') eq 'step20-createMonitorDB-post'} ${wf:conf('resumeFrom') eq 'step21-createObservatoryDB-pre'} ${wf:conf('resumeFrom') eq 'step21-createObservatoryDB'} ${wf:conf('resumeFrom') eq 'step21-createObservatoryDB-post'}