From c02c7741c911b293fbb2c5ba131fb6087672253f Mon Sep 17 00:00:00 2001
From: antleb <antleb@di.uoa.gr>
Date: Tue, 10 Nov 2020 17:11:30 +0200
Subject: [PATCH] fixes in db creation

---
 .../stats/oozie_app/scripts/step16_6.sql      | 32 +++++++++++++++++++
 .../graph/stats/oozie_app/scripts/step19.sql  |  8 +++++
 2 files changed, 40 insertions(+)
 create mode 100644 dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_6.sql
 create mode 100644 dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step19.sql

diff --git a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_6.sql b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_6.sql
new file mode 100644
index 000000000..ced7bbc11
--- /dev/null
+++ b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_6.sql
@@ -0,0 +1,32 @@
+-------------------------------------------
+--- Extra tables, mostly used by indicators
+
+create table ${stats_db_name}.result_projectcount as
+select r.id, count(distinct p.id) as count
+from ${stats_db_name}.result r
+left outer join ${stats_db_name}.result_projects rp on rp.id=r.id
+left outer join ${stats_db_name}.project p on p.id=rp.project
+group by r.id;
+
+create table ${stats_db_name}.result_fundercount as
+select r.id, count(distinct p.funder) as count
+from ${stats_db_name}.result r
+left outer join ${stats_db_name}.result_projects rp on rp.id=r.id
+left outer join ${stats_db_name}.project p on p.id=rp.project
+group by r.id;
+
+create table ${stats_db_name}.project_resultcount as
+with rcount as (
+    select p.id as pid, count(distinct r.id) as `count`, r.type as type
+    from ${stats_db_name}.project p
+    left outer join ${stats_db_name}.result_projects rp on rp.project=p.id
+    left outer join ${stats_db_name}.result r on r.id=rp.id
+    group by r.type, p.id )
+select rcount.pid, sum(case when rcount.type='publication' then rcount.count else 0 end) as publications,
+    sum(case when rcount.type='dataset' then rcount.count else 0 end) as datasets,
+    sum(case when rcount.type='software' then rcount.count else 0 end) as software,
+    sum(case when rcount.type='other' then rcount.count else 0 end) as other
+from rcount
+group by rcount.pid;
+
+create view ${stats_db_name}.rndexpenditure as select * from stats_ext.rndexpediture
\ No newline at end of file
diff --git a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step19.sql b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step19.sql
new file mode 100644
index 000000000..34e48a18a
--- /dev/null
+++ b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step19.sql
@@ -0,0 +1,8 @@
+------------------------------------------------------
+------------------------------------------------------
+-- Impala table statistics - Needed to make the tables
+-- visible for impala
+------------------------------------------------------
+------------------------------------------------------
+
+INVALIDATE METADATA ${stats_db_name};