enrico.ottonello
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Merge pull request 'Beta stats wf updated' (#332) from antonis.lempesis/dnet-hadoop:beta into beta 

Reviewed-on: D-Net/dnet-hadoop#332
This commit is contained in:
Claudio Atzori 2023-10-10 09:35:32 +02:00
parent ef833840c3 17586f0ff8
commit 4e6fccf4f6
24 changed files with 1838 additions and 592 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
dhp-workflows/dhp-stats-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-monitor/oozie_app/config-default.xml Normal file
View File

@ -0,0 +1,30 @@
<configuration>
<property>
<name>jobTracker</name>
<value>${jobTracker}</value>
</property>
<property>
<name>nameNode</name>
<value>${nameNode}</value>
</property>
<property>
<name>oozie.use.system.libpath</name>
<value>true</value>
</property>
<property>
<name>oozie.action.sharelib.for.spark</name>
<value>spark2</value>
</property>
<property>
<name>hive_metastore_uris</name>
<value>thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083</value>
</property>
<property>
<name>hive_jdbc_url</name>
<value>jdbc:hive2://iis-cdh5-test-m3.ocean.icm.edu.pl:10000/;UseNativeQuery=1;?spark.executor.memory=22166291558;spark.yarn.executor.memoryOverhead=3225;spark.driver.memory=15596411699;spark.yarn.driver.memoryOverhead=1228</value>
</property>
<property>
<name>oozie.wf.workflow.notification.url</name>
<value>{serviceUrl}/v1/oozieNotification/jobUpdate?jobId=$jobId%26status=$status</value>
</property>
</configuration>

75
dhp-workflows/dhp-stats-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-monitor/oozie_app/copyDataToImpalaCluster.sh Normal file
View File

@ -0,0 +1,75 @@
export PYTHON_EGG_CACHE=/home/$(whoami)/.python-eggs
export link_folder=/tmp/impala-shell-python-egg-cache-$(whoami)
if ! [ -L $link_folder ]
then
rm -Rf "$link_folder"
ln -sfn ${PYTHON_EGG_CACHE}${link_folder} ${link_folder}
fi
#export HADOOP_USER_NAME=$2
function copydb() {
export HADOOP_USER="dimitris.pierrakos"
export HADOOP_USER_NAME='dimitris.pierrakos'
db=$1
FILE=("hive_wf_tmp_"$RANDOM)
hdfs dfs -mkdir hdfs://impala-cluster-mn1.openaire.eu:8020/tmp/$FILE/
# change ownership to impala
# hdfs dfs -conf /etc/impala_cluster/hdfs-site.xml -chmod -R 777 /tmp/$FILE/${db}.db
hdfs dfs -conf /etc/impala_cluster/hdfs-site.xml -chmod -R 777 /tmp/$FILE/
# copy the databases from ocean to impala
echo "copying $db"
hadoop distcp -Dmapreduce.map.memory.mb=6144 -pb hdfs://nameservice1/user/hive/warehouse/${db}.db hdfs://impala-cluster-mn1.openaire.eu:8020/tmp/$FILE/
hdfs dfs -conf /etc/impala_cluster/hdfs-site.xml -chmod -R 777 /tmp/$FILE/${db}.db
# drop tables from db
for i in `impala-shell -i impala-cluster-dn1.openaire.eu -d ${db} --delimited -q "show tables"`;
do
`impala-shell -i impala-cluster-dn1.openaire.eu -d ${db} -q "drop table $i;"`;
done
# drop views from db
for i in `impala-shell -i impala-cluster-dn1.openaire.eu -d ${db} --delimited -q "show tables"`;
do
`impala-shell -i impala-cluster-dn1.openaire.eu -d ${db} -q "drop view $i;"`;
done
# delete the database
impala-shell -i impala-cluster-dn1.openaire.eu -q "drop database if exists ${db} cascade";
# create the databases
impala-shell -i impala-cluster-dn1.openaire.eu -q "create database ${db}";
impala-shell -q "INVALIDATE METADATA"
echo "creating schema for ${db}"
for (( k = 0; k < 5; k ++ )); do
for i in `impala-shell -d ${db} --delimited -q "show tables"`;
do
impala-shell -d ${db} --delimited -q "show create table $i";
done | sed 's/"$/;/' | sed 's/^"//' | sed 's/[[:space:]]\date[[:space:]]/`date`/g' | impala-shell --user $HADOOP_USER_NAME -i impala-cluster-dn1.openaire.eu -c -f -
done
# load the data from /tmp in the respective tables
echo "copying data in tables and computing stats"
for i in `impala-shell -i impala-cluster-dn1.openaire.eu -d ${db} --delimited -q "show tables"`;
do
impala-shell -i impala-cluster-dn1.openaire.eu -d ${db} -q "load data inpath '/tmp/$FILE/${db}.db/$i' into table $i";
impala-shell -i impala-cluster-dn1.openaire.eu -d ${db} -q "compute stats $i";
done
# deleting the remaining directory from hdfs
hdfs dfs -conf /etc/impala_cluster/hdfs-site.xml -rm -R /tmp/$FILE/${db}.db
}
MONITOR_DB=$1
#HADOOP_USER_NAME=$2
copydb $MONITOR_DB'_institutions'
copydb $MONITOR_DB

29
dhp-workflows/dhp-stats-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-monitor/oozie_app/finalizeImpalaCluster.sh Normal file
View File

@ -0,0 +1,29 @@
export PYTHON_EGG_CACHE=/home/$(whoami)/.python-eggs
export link_folder=/tmp/impala-shell-python-egg-cache-$(whoami)
if ! [ -L $link_folder ]
then
rm -Rf "$link_folder"
ln -sfn ${PYTHON_EGG_CACHE}${link_folder} ${link_folder}
fi
function createShadowDB() {
SOURCE=$1
SHADOW=$2
# drop views from db
for i in `impala-shell -i impala-cluster-dn1.openaire.eu -d ${SHADOW} --delimited -q "show tables"`;
do
`impala-shell -i impala-cluster-dn1.openaire.eu -d ${SHADOW} -q "drop view $i;"`;
done
impala-shell -i impala-cluster-dn1.openaire.eu -q "drop database ${SHADOW} CASCADE";
impala-shell -i impala-cluster-dn1.openaire.eu -q "create database if not exists ${SHADOW}";
# impala-shell -i impala-cluster-dn1.openaire.eu -d ${SHADOW} -q "show tables" | sed "s/^/drop view if exists ${SHADOW}./" | sed "s/$/;/" | impala-shell -i impala-cluster-dn1.openaire.eu -f -
impala-shell -i impala-cluster-dn1.openaire.eu -d ${SOURCE} -q "show tables" --delimited | sed "s/\(.*\)/create view ${SHADOW}.\1 as select * from ${SOURCE}.\1;/" | impala-shell -i impala-cluster-dn1.openaire.eu -f -
}
MONITOR_DB=$1
MONITOR_DB_SHADOW=$2
createShadowDB $MONITOR_DB'_institutions' $MONITOR_DB'_institutions_shadow'
createShadowDB $MONITOR_DB $MONITOR_DB'_shadow'

54
dhp-workflows/dhp-stats-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-monitor/oozie_app/monitor.sh Normal file
View File

@ -0,0 +1,54 @@
export PYTHON_EGG_CACHE=/home/$(whoami)/.python-eggs
export link_folder=/tmp/impala-shell-python-egg-cache-$(whoami)
if ! [ -L $link_folder ]
then
rm -Rf "$link_folder"
ln -sfn ${PYTHON_EGG_CACHE}${link_folder} ${link_folder}
fi
export SOURCE=$1
export TARGET=$2
export SHADOW=$3
export SCRIPT_PATH=$4
export SCRIPT_PATH2=$5
export SCRIPT_PATH2=$6
export HIVE_OPTS="-hiveconf mapred.job.queue.name=analytics -hiveconf hive.spark.client.connect.timeout=120000ms -hiveconf hive.spark.client.server.connect.timeout=300000ms -hiveconf spark.executor.memory=19166291558 -hiveconf spark.yarn.executor.memoryOverhead=3225 -hiveconf spark.driver.memory=11596411699 -hiveconf spark.yarn.driver.memoryOverhead=1228"
export HADOOP_USER_NAME="oozie"
echo "Getting file from " $4
hdfs dfs -copyToLocal $4
echo "Getting file from " $5
hdfs dfs -copyToLocal $5
echo "Getting file from " $6
hdfs dfs -copyToLocal $6
#update Institutions DB
cat updateMonitorDB_institutions.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_institutions/g1" > foo
hive $HIVE_OPTS -f foo
cat updateMonitorDB.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_institutions/g1" > foo
hive $HIVE_OPTS -f foo
echo "Hive shell finished"
echo "Updating shadow monitor insitutions database"
hive -e "drop database if exists ${SHADOW}_institutions cascade"
hive -e "create database if not exists ${SHADOW}_institutions"
hive $HIVE_OPTS --database ${2}_institutions -e "show tables" | grep -v WARN | sed "s/\(.*\)/create view ${SHADOW}_institutions.\1 as select * from ${2}_institutions.\1;/" > foo
hive -f foo
echo "Shadow db monitor insitutions ready!"
#update Monitor DB
cat updateMonitorDBAll.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2/g1" > foo
hive $HIVE_OPTS -f foo
echo "Hive shell finished"
echo "Updating shadow monitor database"
hive -e "drop database if exists ${SHADOW} cascade"
hive -e "create database if not exists ${SHADOW}"
hive $HIVE_OPTS --database ${2} -e "show tables" | grep -v WARN | sed "s/\(.*\)/create view ${SHADOW}.\1 as select * from ${2}.\1;/" > foo
hive -f foo
echo "Shadow db monitor insitutions ready!"

138
dhp-workflows/dhp-stats-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-monitor/oozie_app/scripts/updateMonitorDB.sql Normal file
View File

@ -0,0 +1,138 @@
INSERT INTO TARGET.result select * from TARGET.result_new;
ANALYZE TABLE TARGET.result COMPUTE STATISTICS;
INSERT INTO TARGET.result_citations select * from SOURCE.result_citations orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_citations COMPUTE STATISTICS;
INSERT INTO TARGET.result_references_oc select * from SOURCE.result_references_oc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_references_oc COMPUTE STATISTICS;
INSERT INTO TARGET.result_classifications select * from SOURCE.result_classifications orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_classifications COMPUTE STATISTICS;
INSERT INTO TARGET.result_apc select * from SOURCE.result_apc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_apc COMPUTE STATISTICS;
INSERT INTO TARGET.result_concepts select * from SOURCE.result_concepts orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_concepts COMPUTE STATISTICS;
INSERT INTO TARGET.result_datasources select * from SOURCE.result_datasources orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_datasources COMPUTE STATISTICS;
INSERT INTO TARGET.result_fundercount select * from SOURCE.result_fundercount orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_fundercount COMPUTE STATISTICS;
INSERT INTO TARGET.result_gold select * from SOURCE.result_gold orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_gold COMPUTE STATISTICS;
INSERT INTO TARGET.result_greenoa select * from SOURCE.result_greenoa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_greenoa COMPUTE STATISTICS;
INSERT INTO TARGET.result_languages select * from SOURCE.result_languages orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_languages COMPUTE STATISTICS;
INSERT INTO TARGET.result_licenses select * from SOURCE.result_licenses orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_licenses COMPUTE STATISTICS;
INSERT INTO TARGET.result_oids select * from SOURCE.result_oids orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_oids COMPUTE STATISTICS;
INSERT INTO TARGET.result_organization select * from SOURCE.result_organization orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_organization COMPUTE STATISTICS;
INSERT INTO TARGET.result_peerreviewed select * from SOURCE.result_peerreviewed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_peerreviewed COMPUTE STATISTICS;
INSERT INTO TARGET.result_pids select * from SOURCE.result_pids orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_pids COMPUTE STATISTICS;
INSERT INTO TARGET.result_projectcount select * from SOURCE.result_projectcount orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_projectcount COMPUTE STATISTICS;
INSERT INTO TARGET.result_projects select * from SOURCE.result_projects orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_projects COMPUTE STATISTICS;
INSERT INTO TARGET.result_refereed select * from SOURCE.result_refereed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_refereed COMPUTE STATISTICS;
INSERT INTO TARGET.result_sources select * from SOURCE.result_sources orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_sources COMPUTE STATISTICS;
INSERT INTO TARGET.result_topics select * from SOURCE.result_topics orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_topics COMPUTE STATISTICS;
INSERT INTO TARGET.result_fos select * from SOURCE.result_fos orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_fos COMPUTE STATISTICS;
INSERT INTO TARGET.result_accessroute select * from SOURCE.result_accessroute orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_accessroute COMPUTE STATISTICS;
create or replace view TARGET.foo1 as select * from SOURCE.result_result rr where rr.source in (select id from TARGET.result_new);
create or replace view TARGET.foo2 as select * from SOURCE.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;
ANALYZE TABLE TARGET.result_result COMPUTE STATISTICS;
-- indicators
-- Sprint 1 ----
INSERT INTO TARGET.indi_pub_green_oa select * from SOURCE.indi_pub_green_oa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_green_oa COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_grey_lit select * from SOURCE.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;
INSERT INTO TARGET.indi_pub_doi_from_crossref select * from SOURCE.indi_pub_doi_from_crossref orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_doi_from_crossref COMPUTE STATISTICS;
-- Sprint 2 ----
INSERT INTO TARGET.indi_result_has_cc_licence select * from SOURCE.indi_result_has_cc_licence orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_has_cc_licence COMPUTE STATISTICS;
INSERT INTO TARGET.indi_result_has_cc_licence_url select * from SOURCE.indi_result_has_cc_licence_url orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_has_cc_licence_url COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_has_abstract select * from SOURCE.indi_pub_has_abstract orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_has_abstract COMPUTE STATISTICS;
INSERT INTO TARGET.indi_result_with_orcid select * from SOURCE.indi_result_with_orcid orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_with_orcid COMPUTE STATISTICS;
---- Sprint 3 ----
INSERT INTO TARGET.indi_funded_result_with_fundref select * from SOURCE.indi_funded_result_with_fundref orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_funded_result_with_fundref COMPUTE STATISTICS;
---- Sprint 4 ----
INSERT INTO TARGET.indi_pub_diamond select * from SOURCE.indi_pub_diamond orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_diamond COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_in_transformative select * from SOURCE.indi_pub_in_transformative orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_in_transformative COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_closed_other_open select * from SOURCE.indi_pub_closed_other_open orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_closed_other_open COMPUTE STATISTICS;
---- Sprint 5 ----
INSERT INTO TARGET.indi_result_no_of_copies select * from SOURCE.indi_result_no_of_copies orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_no_of_copies COMPUTE STATISTICS;
---- Sprint 6 ----
INSERT INTO TARGET.indi_pub_hybrid_oa_with_cc select * from SOURCE.indi_pub_hybrid_oa_with_cc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_hybrid_oa_with_cc COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_bronze_oa select * from SOURCE.indi_pub_bronze_oa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_bronze_oa COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_downloads select * from SOURCE.indi_pub_downloads orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_downloads_datasource select * from SOURCE.indi_pub_downloads_datasource orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads_datasource COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_downloads_year select * from SOURCE.indi_pub_downloads_year orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads_year COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_downloads_datasource_year select * from SOURCE.indi_pub_downloads_datasource_year orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads_datasource_year COMPUTE STATISTICS;
---- Sprint 7 ----
INSERT INTO TARGET.indi_pub_gold_oa select * from SOURCE.indi_pub_gold_oa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_gold_oa COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_hybrid select * from SOURCE.indi_pub_hybrid orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_hybrid COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_has_preprint select * from SOURCE.indi_pub_has_preprint orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_has_preprint COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_in_subscribed select * from SOURCE.indi_pub_in_subscribed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_in_subscribed COMPUTE STATISTICS;
INSERT INTO TARGET.indi_result_with_pid select * from SOURCE.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;
INSERT INTO TARGET.indi_impact_measures select * from SOURCE.indi_impact_measures orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_impact_measures COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_interdisciplinarity select * from SOURCE.indi_pub_interdisciplinarity orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_interdisciplinarity COMPUTE STATISTICS;
DROP TABLE IF EXISTS TARGET.result_new;

150
dhp-workflows/dhp-stats-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-monitor/oozie_app/scripts/updateMonitorDBAll.sql Normal file
View File

@ -0,0 +1,150 @@
DROP TABLE IF EXISTS TARGET.result_new;
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____::4d4051b56708688235252f1d8fddb8c1', --Iscte - Instituto Universitário de Lisboa
'openorgs____::ab4ac74c35fa5dada770cf08e5110fab' -- Universidade Católica Portuguesa
) )) foo;
INSERT INTO TARGET.result select * from TARGET.result_new;
ANALYZE TABLE TARGET.result_new COMPUTE STATISTICS;
INSERT INTO TARGET.result select * from TARGET.result_new;
ANALYZE TABLE TARGET.result COMPUTE STATISTICS;
INSERT INTO TARGET.result_citations select * from SOURCE.result_citations orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_citations COMPUTE STATISTICS;
INSERT INTO TARGET.result_references_oc select * from SOURCE.result_references_oc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_references_oc COMPUTE STATISTICS;
INSERT INTO TARGET.result_classifications select * from SOURCE.result_classifications orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_classifications COMPUTE STATISTICS;
INSERT INTO TARGET.result_apc select * from SOURCE.result_apc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_apc COMPUTE STATISTICS;
INSERT INTO TARGET.result_concepts select * from SOURCE.result_concepts orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_concepts COMPUTE STATISTICS;
INSERT INTO TARGET.result_datasources select * from SOURCE.result_datasources orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_datasources COMPUTE STATISTICS;
INSERT INTO TARGET.result_fundercount select * from SOURCE.result_fundercount orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_fundercount COMPUTE STATISTICS;
INSERT INTO TARGET.result_gold select * from SOURCE.result_gold orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_gold COMPUTE STATISTICS;
INSERT INTO TARGET.result_greenoa select * from SOURCE.result_greenoa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_greenoa COMPUTE STATISTICS;
INSERT INTO TARGET.result_languages select * from SOURCE.result_languages orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_languages COMPUTE STATISTICS;
INSERT INTO TARGET.result_licenses select * from SOURCE.result_licenses orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_licenses COMPUTE STATISTICS;
INSERT INTO TARGET.result_oids select * from SOURCE.result_oids orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_oids COMPUTE STATISTICS;
INSERT INTO TARGET.result_organization select * from SOURCE.result_organization orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_organization COMPUTE STATISTICS;
INSERT INTO TARGET.result_peerreviewed select * from SOURCE.result_peerreviewed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_peerreviewed COMPUTE STATISTICS;
INSERT INTO TARGET.result_pids select * from SOURCE.result_pids orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_pids COMPUTE STATISTICS;
INSERT INTO TARGET.result_projectcount select * from SOURCE.result_projectcount orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_projectcount COMPUTE STATISTICS;
INSERT INTO TARGET.result_projects select * from SOURCE.result_projects orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_projects COMPUTE STATISTICS;
INSERT INTO TARGET.result_refereed select * from SOURCE.result_refereed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_refereed COMPUTE STATISTICS;
INSERT INTO TARGET.result_sources select * from SOURCE.result_sources orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_sources COMPUTE STATISTICS;
INSERT INTO TARGET.result_topics select * from SOURCE.result_topics orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_topics COMPUTE STATISTICS;
INSERT INTO TARGET.result_fos select * from SOURCE.result_fos orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_fos COMPUTE STATISTICS;
INSERT INTO TARGET.result_accessroute select * from SOURCE.result_accessroute orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.result_accessroute COMPUTE STATISTICS;
create or replace view TARGET.foo1 as select * from SOURCE.result_result rr where rr.source in (select id from TARGET.result_new);
create or replace view TARGET.foo2 as select * from SOURCE.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;
ANALYZE TABLE TARGET.result_result COMPUTE STATISTICS;
-- indicators
-- Sprint 1 ----
INSERT INTO TARGET.indi_pub_green_oa select * from SOURCE.indi_pub_green_oa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_green_oa COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_grey_lit select * from SOURCE.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;
INSERT INTO TARGET.indi_pub_doi_from_crossref select * from SOURCE.indi_pub_doi_from_crossref orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_doi_from_crossref COMPUTE STATISTICS;
-- Sprint 2 ----
INSERT INTO TARGET.indi_result_has_cc_licence select * from SOURCE.indi_result_has_cc_licence orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_has_cc_licence COMPUTE STATISTICS;
INSERT INTO TARGET.indi_result_has_cc_licence_url select * from SOURCE.indi_result_has_cc_licence_url orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_has_cc_licence_url COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_has_abstract select * from SOURCE.indi_pub_has_abstract orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_has_abstract COMPUTE STATISTICS;
INSERT INTO TARGET.indi_result_with_orcid select * from SOURCE.indi_result_with_orcid orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_with_orcid COMPUTE STATISTICS;
---- Sprint 3 ----
INSERT INTO TARGET.indi_funded_result_with_fundref select * from SOURCE.indi_funded_result_with_fundref orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_funded_result_with_fundref COMPUTE STATISTICS;
---- Sprint 4 ----
INSERT INTO TARGET.indi_pub_diamond select * from SOURCE.indi_pub_diamond orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_diamond COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_in_transformative select * from SOURCE.indi_pub_in_transformative orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_in_transformative COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_closed_other_open select * from SOURCE.indi_pub_closed_other_open orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_closed_other_open COMPUTE STATISTICS;
---- Sprint 5 ----
INSERT INTO TARGET.indi_result_no_of_copies select * from SOURCE.indi_result_no_of_copies orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_no_of_copies COMPUTE STATISTICS;
---- Sprint 6 ----
INSERT INTO TARGET.indi_pub_hybrid_oa_with_cc select * from SOURCE.indi_pub_hybrid_oa_with_cc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_hybrid_oa_with_cc COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_bronze_oa select * from SOURCE.indi_pub_bronze_oa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_bronze_oa COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_downloads select * from SOURCE.indi_pub_downloads orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_downloads_datasource select * from SOURCE.indi_pub_downloads_datasource orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads_datasource COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_downloads_year select * from SOURCE.indi_pub_downloads_year orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads_year COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_downloads_datasource_year select * from SOURCE.indi_pub_downloads_datasource_year orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads_datasource_year COMPUTE STATISTICS;
---- Sprint 7 ----
INSERT INTO TARGET.indi_pub_gold_oa select * from SOURCE.indi_pub_gold_oa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_gold_oa COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_hybrid select * from SOURCE.indi_pub_hybrid orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_hybrid COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_has_preprint select * from SOURCE.indi_pub_has_preprint orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_has_preprint COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_in_subscribed select * from SOURCE.indi_pub_in_subscribed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_in_subscribed COMPUTE STATISTICS;
INSERT INTO TARGET.indi_result_with_pid select * from SOURCE.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;
INSERT INTO TARGET.indi_impact_measures select * from SOURCE.indi_impact_measures orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_impact_measures COMPUTE STATISTICS;
INSERT INTO TARGET.indi_pub_interdisciplinarity select * from SOURCE.indi_pub_interdisciplinarity orig where exists (select 1 from TARGET.result_new r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_interdisciplinarity COMPUTE STATISTICS;
DROP TABLE IF EXISTS TARGET.result_new;

12
dhp-workflows/dhp-stats-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-monitor/oozie_app/scripts/updateMonitorDB_institutions.sql Normal file
View File

@ -0,0 +1,12 @@
DROP TABLE IF EXISTS TARGET.result_new;
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____::4d4051b56708688235252f1d8fddb8c1', --Iscte - Instituto Universitário de Lisboa
'openorgs____::ab4ac74c35fa5dada770cf08e5110fab' -- Universidade Católica Portuguesa
) )) foo;
INSERT INTO TARGET.result select * from TARGET.result_new;
ANALYZE TABLE TARGET.result_new COMPUTE STATISTICS;

110
dhp-workflows/dhp-stats-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-monitor/oozie_app/workflow.xml Normal file
View File

@ -0,0 +1,110 @@
<workflow-app name="Stats Monitor Update" xmlns="uri:oozie:workflow:0.5">
<parameters>
<property>
<name>stats_db_name</name>
<description>the target stats database name</description>
</property>
<property>
<name>monitor_db_name</name>
<description>the target monitor db name</description>
</property>
<property>
<name>monitor_db_shadow_name</name>
<description>the name of the shadow monitor db</description>
</property>
<property>
<name>hive_metastore_uris</name>
<description>hive server metastore URIs</description>
</property>
<property>
<name>hive_jdbc_url</name>
<description>hive server jdbc url</description>
</property>
<property>
<name>hive_timeout</name>
<description>the time period, in seconds, after which Hive fails a transaction if a Hive client has not sent a hearbeat. The default value is 300 seconds.</description>
</property>
<property>
<name>hadoop_user_name</name>
<description>user name of the wf owner</description>
</property>
</parameters>
<global>
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<configuration>
<property>
<name>hive.metastore.uris</name>
<value>${hive_metastore_uris}</value>
</property>
<property>
<name>hive.txn.timeout</name>
<value>${hive_timeout}</value>
</property>
<property>
<name>mapred.job.queue.name</name>
<value>analytics</value>
</property>
</configuration>
</global>
<start to="resume_from"/>
<decision name="resume_from">
<switch>
<case to="Step1-updateMonitorDB">${wf:conf('resumeFrom') eq 'Step1-updateMonitorDB'}</case>
<case to="Step2-copyDataToImpalaCluster">${wf:conf('resumeFrom') eq 'Step2-copyDataToImpalaCluster'}</case>
<case to="Step3-finalizeImpalaCluster">${wf:conf('resumeFrom') eq 'Step3-finalizeImpalaCluster'}</case>
<default to="Step1-updateMonitorDB"/>
</switch>
</decision>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<action name="Step1-updateMonitorDB">
<shell xmlns="uri:oozie:shell-action:0.1">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<exec>monitor.sh</exec>
<argument>${stats_db_name}</argument>
<argument>${monitor_db_name}</argument>
<argument>${monitor_db_shadow_name}</argument>
<argument>${wf:appPath()}/scripts/updateMonitorDB_institutions.sql</argument>
<argument>${wf:appPath()}/scripts/updateMonitorDB.sql</argument>
<argument>${wf:appPath()}/scripts/updateMonitorDBAll.sql</argument>
<file>monitor.sh</file>
</shell>
<ok to="Step2-copyDataToImpalaCluster"/>
<error to="Kill"/>
</action>
<action name="Step2-copyDataToImpalaCluster">
<shell xmlns="uri:oozie:shell-action:0.1">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<exec>copyDataToImpalaCluster.sh</exec>
<argument>${monitor_db_name}</argument>
<argument>${hadoop_user_name}</argument>
<file>copyDataToImpalaCluster.sh</file>
</shell>
<ok to="Step3-finalizeImpalaCluster"/>
<error to="Kill"/>
</action>
<action name="Step3-finalizeImpalaCluster">
<shell xmlns="uri:oozie:shell-action:0.1">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<exec>finalizeImpalaCluster.sh</exec>
<argument>${monitor_db_name}</argument>
<argument>${monitor_db_shadow_name}</argument>
<file>finalizeImpalaCluster.sh</file>
</shell>
<ok to="End"/>
<error to="Kill"/>
</action>
<end name="End"/>
</workflow-app>

8
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/copyDataToImpalaCluster.sh
View File

@ -24,13 +24,13 @@ function copydb() {
# drop tables from db
for i in `impala-shell --user $HADOOP_USER_NAME -i impala-cluster-dn1.openaire.eu -d ${db} --delimited -q "show tables"`;
do
`impala-shell -i impala-cluster-dn1.openaire.eu -d -d ${db} -q "drop table $i;"`;
`impala-shell -i impala-cluster-dn1.openaire.eu -d ${db} -q "drop table $i;"`;
done
# drop views from db
for i in `impala-shell --user $HADOOP_USER_NAME -i impala-cluster-dn1.openaire.eu -d ${db} --delimited -q "show tables"`;
do
`impala-shell -i impala-cluster-dn1.openaire.eu -d -d ${db} -q "drop view $i;"`;
`impala-shell -i impala-cluster-dn1.openaire.eu -d ${db} -q "drop view $i;"`;
done
# delete the database
@ -82,12 +82,12 @@ copydb $USAGE_STATS_DB
copydb $PROD_USAGE_STATS_DB
copydb $EXT_DB
copydb $STATS_DB
#copydb $MONITOR_DB
copydb $MONITOR_DB
copydb $OBSERVATORY_DB
copydb $MONITOR_DB'_funded'
copydb $MONITOR_DB'_institutions'
copydb $MONITOR_DB'_RIs_tail'
copydb $MONITOR_DB'_ris_tail'
contexts="knowmad::other dh-ch::other enermaps::other gotriple::other neanias-atmospheric::other rural-digital-europe::other covid-19::other aurora::other neanias-space::other north-america-studies::other north-american-studies::other eutopia::other"
for i in ${contexts}

10
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/finalizeImpalaCluster.sh
View File

@ -13,7 +13,7 @@ function createShadowDB() {
# drop views from db
for i in `impala-shell -i impala-cluster-dn1.openaire.eu -d ${SHADOW} --delimited -q "show tables"`;
do
`impala-shell -i impala-cluster-dn1.openaire.eu -d -d ${SHADOW} -q "drop view $i;"`;
`impala-shell -i impala-cluster-dn1.openaire.eu -d ${SHADOW} -q "drop view $i;"`;
done
impala-shell -i impala-cluster-dn1.openaire.eu -q "drop database ${SHADOW} CASCADE";
@ -36,13 +36,13 @@ createShadowDB $MONITOR_DB $MONITOR_DB_SHADOW
createShadowDB $OBSERVATORY_DB $OBSERVATORY_DB_SHADOW
createShadowDB USAGE_STATS_DB USAGE_STATS_DB_SHADOW
createShadowDB $MONITOR_DB'_funded' $MONITOR_DB'_funded_shadow'
createShadowDB $MONITOR_DB'_institutions' $MONITOR_DB'_institutions_shadow'
createShadowDB $MONITOR_DB'_RIs_tail' $MONITOR_DB'_RIs_tail_shadow'
createShadowDB $MONITOR_DB'_funded' $MONITOR_DB_SHADOW'_shadow_funded'
createShadowDB $MONITOR_DB'_institutions' $MONITOR_DB_SHADOW'_shadow_institutions'
createShadowDB $MONITOR_DB'_ris_tail' $MONITOR_DB_SHADOW'_shadow_ris_tail'
contexts="knowmad::other dh-ch::other enermaps::other gotriple::other neanias-atmospheric::other rural-digital-europe::other covid-19::other aurora::other neanias-space::other north-america-studies::other north-american-studies::other eutopia::other"
for i in ${contexts}
do
tmp=`echo "$i" | sed 's/'-'/'_'/g' | sed 's/'::'/'_'/g'`
createShadowDB ${MONITOR_DB}'_'${tmp} ${MONITOR_DB}'_'${tmp}'_shadow'
createShadowDB ${MONITOR_DB}'_'${tmp} ${MONITOR_DB_SHADOW}'_shadow_'${tmp}
done

22
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/monitor.sh Normal file → Executable file
View File

@ -14,6 +14,7 @@ export SCRIPT_PATH2=$5
export SCRIPT_PATH3=$6
export SCRIPT_PATH4=$7
export SCRIPT_PATH5=$8
export SCRIPT_PATH6=$9
export HIVE_OPTS="-hiveconf mapred.job.queue.name=analytics -hiveconf hive.spark.client.connect.timeout=120000ms -hiveconf hive.spark.client.server.connect.timeout=300000ms -hiveconf spark.executor.memory=19166291558 -hiveconf spark.yarn.executor.memoryOverhead=3225 -hiveconf spark.driver.memory=11596411699 -hiveconf spark.yarn.driver.memoryOverhead=1228"
export HADOOP_USER_NAME="oozie"
@ -33,12 +34,19 @@ hdfs dfs -copyToLocal $7
echo "Getting file from " $8
hdfs dfs -copyToLocal $8
echo "Getting file from " $9
hdfs dfs -copyToLocal $9
echo "Creating monitor database"
cat step20-createMonitorDBAll.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2/g1" > foo
hive $HIVE_OPTS -f foo
cat step20-createMonitorDB_funded.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_funded/g1" > foo
hive $HIVE_OPTS -f foo
cat step20-createMonitorDB.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_funded/g1" > foo
hive $HIVE_OPTS -f foo
#
cat step20-createMonitorDB_institutions.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_institutions/g1" > foo
hive $HIVE_OPTS -f foo
cat step20-createMonitorDB.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_institutions/g1" > foo
@ -56,14 +64,20 @@ do
hive $HIVE_OPTS -f foo
done
cat step20-createMonitorDB_RIs_tail.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_RIs_tail/g1" | sed "s/CONTEXTS/\"'knowmad::other','dh-ch::other', 'enermaps::other', 'gotriple::other', 'neanias-atmospheric::other', 'rural-digital-europe::other', 'covid-19::other', 'aurora::other', 'neanias-space::other', 'north-america-studies::other', 'north-american-studies::other', 'eutopia::other'\"/g" > foo
cat step20-createMonitorDB_RIs_tail.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_ris_tail/g1" | sed "s/CONTEXTS/\"'knowmad::other','dh-ch::other', 'enermaps::other', 'gotriple::other', 'neanias-atmospheric::other', 'rural-digital-europe::other', 'covid-19::other', 'aurora::other', 'neanias-space::other', 'north-america-studies::other', 'north-american-studies::other', 'eutopia::other'\"/g" > foo
hive $HIVE_OPTS -f foo
cat step20-createMonitorDB.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_RIs_tail/g1" > foo
cat step20-createMonitorDB.sql | sed "s/SOURCE/$1/g" | sed "s/TARGET/$2_ris_tail/g1" > foo
hive $HIVE_OPTS -f foo
echo "Hive shell finished"
echo "Updating shadow monitor all database"
hive -e "drop database if exists ${SHADOW} cascade"
hive -e "create database if not exists ${SHADOW}"
hive $HIVE_OPTS --database ${2} -e "show tables" | grep -v WARN | sed "s/\(.*\)/create view ${SHADOW}.\1 as select * from ${2}.\1;/" > foo
hive -f foo
echo "Updated shadow monitor all database"
echo "Updating shadow monitor funded database"
hive -e "drop database if exists ${SHADOW}_funded cascade"
hive -e "create database if not exists ${SHADOW}_funded"

0
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step13.sql Normal file → Executable file
View File

11
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step15.sql
View File

@ -37,8 +37,15 @@ select * from ${stats_db_name}.otherresearchproduct_refereed;
create table if not exists ${stats_db_name}.indi_impact_measures STORED AS PARQUET as
select substr(id, 4) as id, measures_ids.id impactmetric, cast(measures_ids.unit.value[0] as double) score,
cast(measures_ids.unit.value[0] as decimal(6,3)) score_dec, measures_ids.unit.value[1] class
cast(measures_ids.unit.value[0] as decimal(6,3)) score_dec, measures_ids.unit.value[1] impact_class
from ${openaire_db_name}.result lateral view explode(measures) measures as measures_ids
where measures_ids.id!='views' and measures_ids.id!='downloads';
ANALYZE TABLE indi_impact_measures COMPUTE STATISTICS;
create table if not exists ${stats_db_name}.result_apc_affiliations STORED AS PARQUET as
select distinct substr(rel.target,4) id, substr(rel.source,4) organization, o.legalname.value name,
cast(rel.properties[0].value as double) apc_amount,
rel.properties[1].value apc_currency
from ${openaire_db_name}.relation rel
join ${openaire_db_name}.organization o on o.id=rel.source
join ${openaire_db_name}.result r on r.id=rel.target
where rel.subreltype = 'affiliation' and rel.datainfo.deletedbyinference = false and size(rel.properties)>0;

1
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step15_5.sql
View File

@ -35,6 +35,7 @@ create or replace view ${stats_db_name}.doctoratestudents as select * from stats
create or replace view ${stats_db_name}.totalresearchers as select * from stats_ext.totalresearchers;
create or replace view ${stats_db_name}.totalresearchersft as select * from stats_ext.totalresearchersft;
create or replace view ${stats_db_name}.hrrst as select * from stats_ext.hrrst;
create or replace view ${stats_db_name}.graduatedoctorates as select * from stats_ext.graduatedoctorates;
create table if not exists ${stats_db_name}.result_instance stored as parquet as
select distinct r.*

1283
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16-createIndicatorsTables.sql
View File

File diff suppressed because it is too large Load Diff

117
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDB.sql
View File

@ -88,85 +88,88 @@ create view if not exists TARGET.doctoratestudents as select * from SOURCE.docto
create view if not exists TARGET.totalresearchers as select * from SOURCE.totalresearchers;
create view if not exists TARGET.totalresearchersft as select * from SOURCE.totalresearchersft;
create view if not exists TARGET.hrrst as select * from SOURCE.hrrst;
create view if not exists TARGET.graduatedoctorates as select * from SOURCE.graduatedoctorates;
create table TARGET.result_citations stored as parquet as select * from SOURCE.result_citations orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_citations COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_citations COMPUTE STATISTICS;
create table TARGET.result_references_oc stored as parquet as select * from SOURCE.result_references_oc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_references_oc COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_references_oc COMPUTE STATISTICS;
create table TARGET.result_citations_oc stored as parquet as select * from SOURCE.result_citations_oc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_citations_oc COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_citations_oc COMPUTE STATISTICS;
create table TARGET.result_classifications stored as parquet as select * from SOURCE.result_classifications orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_classifications COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_classifications COMPUTE STATISTICS;
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);
ANALYZE TABLE TARGET.result_apc COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_apc COMPUTE STATISTICS;
create table TARGET.result_concepts stored as parquet as select * from SOURCE.result_concepts orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_concepts COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_concepts COMPUTE STATISTICS;
create table TARGET.result_datasources stored as parquet as select * from SOURCE.result_datasources orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_datasources COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_datasources COMPUTE STATISTICS;
create table TARGET.result_fundercount stored as parquet as select * from SOURCE.result_fundercount orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_fundercount COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_fundercount COMPUTE STATISTICS;
create table TARGET.result_gold stored as parquet as select * from SOURCE.result_gold orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_gold COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_gold COMPUTE STATISTICS;
create table TARGET.result_greenoa stored as parquet as select * from SOURCE.result_greenoa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_greenoa COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_greenoa COMPUTE STATISTICS;
create table TARGET.result_languages stored as parquet as select * from SOURCE.result_languages orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_languages COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_languages COMPUTE STATISTICS;
create table TARGET.result_licenses stored as parquet as select * from SOURCE.result_licenses orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_licenses COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_licenses COMPUTE STATISTICS;
create table TARGET.licenses_normalized STORED AS PARQUET as select * from SOURCE.licenses_normalized;
ANALYZE TABLE TARGET.licenses_normalized COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.licenses_normalized COMPUTE STATISTICS;
create table TARGET.result_oids stored as parquet as select * from SOURCE.result_oids orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_oids COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_oids COMPUTE STATISTICS;
create table TARGET.result_organization stored as parquet as select * from SOURCE.result_organization orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_organization COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_organization COMPUTE STATISTICS;
create table TARGET.result_peerreviewed stored as parquet as select * from SOURCE.result_peerreviewed orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_peerreviewed COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_peerreviewed COMPUTE STATISTICS;
create table TARGET.result_pids stored as parquet as select * from SOURCE.result_pids orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_pids COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_pids COMPUTE STATISTICS;
create table TARGET.result_projectcount stored as parquet as select * from SOURCE.result_projectcount orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_projectcount COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_projectcount COMPUTE STATISTICS;
create table TARGET.result_projects stored as parquet as select * from SOURCE.result_projects orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_projects COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_projects COMPUTE STATISTICS;
create table TARGET.result_refereed stored as parquet as select * from SOURCE.result_refereed orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_refereed COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_refereed COMPUTE STATISTICS;
create table TARGET.result_sources stored as parquet as select * from SOURCE.result_sources orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_sources COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_sources COMPUTE STATISTICS;
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);
ANALYZE TABLE TARGET.result_topics COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_topics COMPUTE STATISTICS;
create table TARGET.result_fos stored as parquet as select * from SOURCE.result_fos orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_fos COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_fos COMPUTE STATISTICS;
create table TARGET.result_accessroute stored as parquet as select * from SOURCE.result_accessroute orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.result_accessroute COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_accessroute COMPUTE STATISTICS;
create table TARGET.result_orcid stored as parquet as select * from SOURCE.result_orcid orig where exists (select 1 from TARGET.result r where r.id=orig.id);
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;
drop view TARGET.foo1;
drop view TARGET.foo2;
ANALYZE TABLE TARGET.result_result COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result_result COMPUTE STATISTICS;
-- datasources
create view if not exists TARGET.datasource as select * from SOURCE.datasource;
@ -175,7 +178,7 @@ create view if not exists TARGET.datasource_organizations as select * from SOURC
create view if not exists TARGET.datasource_sources as select * from SOURCE.datasource_sources;
create table TARGET.datasource_results stored as parquet as select id as result, datasource as id from TARGET.result_datasources;
ANALYZE TABLE TARGET.datasource_results COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.datasource_results COMPUTE STATISTICS;
-- organizations
create view if not exists TARGET.organization as select * from SOURCE.organization;
@ -193,28 +196,28 @@ create view if not exists TARGET.project_classification as select * from SOURCE.
create view if not exists TARGET.project_organization_contribution as select * from SOURCE.project_organization_contribution;
create table TARGET.project_results stored as parquet as select id as result, project as id from TARGET.result_projects;
ANALYZE TABLE TARGET.project_results COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.project_results COMPUTE STATISTICS;
-- indicators
-- Sprint 1 ----
create table TARGET.indi_pub_green_oa stored as parquet as select * from SOURCE.indi_pub_green_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_green_oa COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_green_oa COMPUTE STATISTICS;
create table TARGET.indi_pub_grey_lit stored as parquet as select * from SOURCE.indi_pub_grey_lit orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_grey_lit COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_grey_lit COMPUTE STATISTICS;
create table TARGET.indi_pub_doi_from_crossref stored as parquet as select * from SOURCE.indi_pub_doi_from_crossref orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_doi_from_crossref COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_doi_from_crossref COMPUTE STATISTICS;
-- Sprint 2 ----
create table TARGET.indi_result_has_cc_licence stored as parquet as select * from SOURCE.indi_result_has_cc_licence orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_has_cc_licence COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_result_has_cc_licence COMPUTE STATISTICS;
create table TARGET.indi_result_has_cc_licence_url stored as parquet as select * from SOURCE.indi_result_has_cc_licence_url orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_has_cc_licence_url COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_result_has_cc_licence_url COMPUTE STATISTICS;
create table TARGET.indi_pub_has_abstract stored as parquet as select * from SOURCE.indi_pub_has_abstract orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_has_abstract COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_has_abstract COMPUTE STATISTICS;
create table TARGET.indi_result_with_orcid stored as parquet as select * from SOURCE.indi_result_with_orcid orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_with_orcid COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_result_with_orcid COMPUTE STATISTICS;
---- Sprint 3 ----
create table TARGET.indi_funded_result_with_fundref stored as parquet as select * from SOURCE.indi_funded_result_with_fundref orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_funded_result_with_fundref COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_funded_result_with_fundref COMPUTE STATISTICS;
create view TARGET.indi_result_org_collab as select * from SOURCE.indi_result_org_collab;
create view TARGET.indi_result_org_country_collab as select * from SOURCE.indi_result_org_country_collab;
create view TARGET.indi_project_collab_org as select * from SOURCE.indi_project_collab_org;
@ -223,32 +226,32 @@ create view TARGET.indi_funder_country_collab as select * from SOURCE.indi_funde
create view TARGET.indi_result_country_collab as select * from SOURCE.indi_result_country_collab;
---- Sprint 4 ----
create table TARGET.indi_pub_diamond stored as parquet as select * from SOURCE.indi_pub_diamond orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_diamond COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_diamond COMPUTE STATISTICS;
create table TARGET.indi_pub_in_transformative stored as parquet as select * from SOURCE.indi_pub_in_transformative orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_in_transformative COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_in_transformative COMPUTE STATISTICS;
create table TARGET.indi_pub_closed_other_open stored as parquet as select * from SOURCE.indi_pub_closed_other_open orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_closed_other_open COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_closed_other_open COMPUTE STATISTICS;
---- Sprint 5 ----
create table TARGET.indi_result_no_of_copies stored as parquet as select * from SOURCE.indi_result_no_of_copies orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_no_of_copies COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_result_no_of_copies COMPUTE STATISTICS;
---- Sprint 6 ----
create table TARGET.indi_pub_hybrid_oa_with_cc stored as parquet as select * from SOURCE.indi_pub_hybrid_oa_with_cc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_hybrid_oa_with_cc COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_hybrid_oa_with_cc COMPUTE STATISTICS;
create table TARGET.indi_pub_bronze_oa stored as parquet as select * from SOURCE.indi_pub_bronze_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_bronze_oa COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_bronze_oa COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads stored as parquet as select * from SOURCE.indi_pub_downloads orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_downloads COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads_datasource stored as parquet as select * from SOURCE.indi_pub_downloads_datasource orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads_datasource COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_downloads_datasource COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads_year stored as parquet as select * from SOURCE.indi_pub_downloads_year orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads_year COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_downloads_year COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads_datasource_year stored as parquet as select * from SOURCE.indi_pub_downloads_datasource_year orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
ANALYZE TABLE TARGET.indi_pub_downloads_datasource_year COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_downloads_datasource_year COMPUTE STATISTICS;
---- Sprint 7 ----
create table TARGET.indi_pub_gold_oa stored as parquet as select * from SOURCE.indi_pub_gold_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_gold_oa COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_gold_oa COMPUTE STATISTICS;
create table TARGET.indi_pub_hybrid stored as parquet as select * from SOURCE.indi_pub_hybrid orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_hybrid COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_hybrid COMPUTE STATISTICS;
create view TARGET.indi_org_fairness as select * from SOURCE.indi_org_fairness;
create view TARGET.indi_org_fairness_pub_pr as select * from SOURCE.indi_org_fairness_pub_pr;
create view TARGET.indi_org_fairness_pub_year as select * from SOURCE.indi_org_fairness_pub_year;
@ -259,12 +262,22 @@ create view TARGET.indi_org_findable as select * from SOURCE.indi_org_findable;
create view TARGET.indi_org_openess as select * from SOURCE.indi_org_openess;
create view TARGET.indi_org_openess_year as select * from SOURCE.indi_org_openess_year;
create table TARGET.indi_pub_has_preprint stored as parquet as select * from SOURCE.indi_pub_has_preprint orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_has_preprint COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_has_preprint COMPUTE STATISTICS;
create table TARGET.indi_pub_in_subscribed stored as parquet as select * from SOURCE.indi_pub_in_subscribed orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_in_subscribed COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_in_subscribed COMPUTE STATISTICS;
create table TARGET.indi_result_with_pid stored as parquet as select * from SOURCE.indi_result_with_pid orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_result_with_pid COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_result_with_pid COMPUTE STATISTICS;
create table TARGET.indi_impact_measures stored as parquet as select * from SOURCE.indi_impact_measures orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_impact_measures COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_impact_measures COMPUTE STATISTICS;
create table TARGET.indi_pub_interdisciplinarity stored as parquet as select * from SOURCE.indi_pub_interdisciplinarity orig where exists (select 1 from TARGET.result r where r.id=orig.id);
ANALYZE TABLE TARGET.indi_pub_interdisciplinarity COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.indi_pub_interdisciplinarity COMPUTE STATISTICS;
create table TARGET.result_apc_affiliations stored as parquet as select * from SOURCE.result_apc_affiliations orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_apc_affiliations COMPUTE STATISTICS;
create table TARGET.indi_is_project_result_after stored as parquet as select * from SOURCE.indi_is_project_result_after orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
create table TARGET.indi_is_funder_plan_s stored as parquet as select * from SOURCE.indi_is_funder_plan_s orig where exists (select 1 from TARGET.result r where r.id=orig.id);
create view TARGET.indi_funder_fairness as select * from SOURCE.indi_funder_fairness;
create view TARGET.indi_funder_openess as select * from SOURCE.indi_funder_openess;
create view TARGET.indi_funder_findable as select * from SOURCE.indi_funder_findable;
create view TARGET.indi_ris_fairness as select * from SOURCE.indi_ris_fairness;
create view TARGET.indi_ris_openess as select * from SOURCE.indi_ris_openess;
create view TARGET.indi_ris_findable as select * from SOURCE.indi_ris_findable;

292
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDBAll.sql Normal file
View File

@ -0,0 +1,292 @@
drop database if exists TARGET cascade;
create database if not exists TARGET;
create view if not exists TARGET.category as select * from SOURCE.category;
create view if not exists TARGET.concept as select * from SOURCE.concept;
create view if not exists TARGET.context as select * from SOURCE.context;
create view if not exists TARGET.country as select * from SOURCE.country;
create view if not exists TARGET.countrygdp as select * from SOURCE.countrygdp;
create view if not exists TARGET.creation_date as select * from SOURCE.creation_date;
create view if not exists TARGET.funder as select * from SOURCE.funder;
create view if not exists TARGET.fundref as select * from SOURCE.fundref;
create view if not exists TARGET.rndexpenditure as select * from SOURCE.rndexpediture;
create view if not exists TARGET.rndgdpexpenditure as select * from SOURCE.rndgdpexpenditure;
create view if not exists TARGET.doctoratestudents as select * from SOURCE.doctoratestudents;
create view if not exists TARGET.totalresearchers as select * from SOURCE.totalresearchers;
create view if not exists TARGET.totalresearchersft as select * from SOURCE.totalresearchersft;
create view if not exists TARGET.hrrst as select * from SOURCE.hrrst;
create view if not exists TARGET.graduatedoctorates as select * from SOURCE.graduatedoctorates;
create table TARGET.result stored as parquet as
select distinct * from (
select * from SOURCE.result r where exists (select 1 from SOURCE.result_projects rp join SOURCE.project p on rp.project=p.id where rp.id=r.id)
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_organization ro where ro.id=r.id and ro.organization in (
'openorgs____::b84450f9864182c67b8611b5593f4250', --"Athena Research and Innovation Center In Information Communication & Knowledge Technologies', --ARC"
'openorgs____::d41cf6bd4ab1b1362a44397e0b95c975', --National Research Council
'openorgs____::d2a09b9d5eabb10c95f9470e172d05d2', --??? Not exists ??
'openorgs____::d169c7407dd417152596908d48c11460', --Masaryk University
'openorgs____::1ec924b1759bb16d0a02f2dad8689b21', --University of Belgrade
'openorgs____::0ae431b820e4c33db8967fbb2b919150', --University of Helsinki
'openorgs____::759d59f05d77188faee99b7493b46805', --University of Minho
'openorgs____::cad284878801b9465fa51a95b1d779db', --Universidad Politécnica de Madrid
'openorgs____::eadc8da90a546e98c03f896661a2e4d4', --University of Göttingen
'openorgs____::c0286313e36479eff8676dba9b724b40', --National and Kapodistrian University of Athens
-- 'openorgs____::c80a8243a5e5c620d7931c88d93bf17a', --Université Paris Diderot
'openorgs____::c08634f0a6b0081c3dc6e6c93a4314f3', --Bielefeld University
'openorgs____::6fc85e4a8f7ecaf4b0c738d010e967ea', --University of Southern Denmark
'openorgs____::3d6122f87f9a97a99d8f6e3d73313720', --Humboldt-Universität zu Berlin
'openorgs____::16720ada63d0fa8ca41601feae7d1aa5', --TU Darmstadt
'openorgs____::ccc0a066b56d2cfaf90c2ae369df16f5', --KU Leuven
'openorgs____::4c6f119632adf789746f0a057ed73e90', --University of the Western Cape
'openorgs____::ec3665affa01aeafa28b7852c4176dbd', --Rudjer Boskovic Institute
'openorgs____::5f31346d444a7f06a28c880fb170b0f6', --Ghent University
'openorgs____::2dbe47117fd5409f9c61620813456632', --University of Luxembourg
'openorgs____::6445d7758d3a40c4d997953b6632a368', --National Institute of Informatics (NII)
'openorgs____::b77c01aa15de3675da34277d48de2ec1', -- Valencia Catholic University Saint Vincent Martyr
'openorgs____::7fe2f66cdc43983c6b24816bfe9cf6a0', -- Unviersity of Warsaw
'openorgs____::15e7921fc50d9aa1229a82a84429419e', -- University Of Thessaly
'openorgs____::11f7919dadc8f8a7251af54bba60c956', -- Technical University of Crete
'openorgs____::84f0c5f5dbb6daf42748485924efde4b', -- University of Piraeus
'openorgs____::4ac562f0376fce3539504567649cb373', -- University of Patras
'openorgs____::3e8d1f8c3f6cd7f418b09f1f58b4873b', -- Aristotle University of Thessaloniki
'openorgs____::3fcef6e1c469c10f2a84b281372c9814', -- World Bank
'openorgs____::1698a2eb1885ef8adb5a4a969e745ad3', -- École des Ponts ParisTech
'openorgs____::e15adb13c4dadd49de4d35c39b5da93a', -- Nanyang Technological University
'openorgs____::4b34103bde246228fcd837f5f1bf4212', -- Autonomous University of Barcelona
'openorgs____::72ec75fcfc4e0df1a76dc4c49007fceb', -- McMaster University
'openorgs____::51c7fc556e46381734a25a6fbc3fd398', -- University of Modena and Reggio Emilia
'openorgs____::235d7f9ad18ecd7e6dc62ea4990cb9db', -- Bilkent University
'openorgs____::31f2fa9e05b49d4cf40a19c3fed8eb06', -- Saints Cyril and Methodius University of Skopje
'openorgs____::db7686f30f22cbe73a4fde872ce812a6', -- University of Milan
'openorgs____::b8b8ca674452579f3f593d9f5e557483', -- University College Cork
'openorgs____::38d7097854736583dde879d12dacafca', -- Brown University
'openorgs____::57784c9e047e826fefdb1ef816120d92', --Arts et Métiers ParisTech
'openorgs____::2530baca8a15936ba2e3297f2bce2e7e', -- University of Cape Town
'openorgs____::d11f981828c485cd23d93f7f24f24db1', -- Technological University Dublin
'openorgs____::5e6bf8962665cdd040341171e5c631d8', -- Delft University of Technology
'openorgs____::846cb428d3f52a445f7275561a7beb5d', -- University of Manitoba
'openorgs____::eb391317ed0dc684aa81ac16265de041', -- Universitat Rovira i Virgili
'openorgs____::66aa9fc2fceb271423dfabcc38752dc0', -- Lund University
'openorgs____::3cff625a4370d51e08624cc586138b2f', -- IMT Atlantique
'openorgs____::c0b262bd6eab819e4c994914f9c010e2', -- National Institute of Geophysics and Volcanology
'openorgs____::1624ff7c01bb641b91f4518539a0c28a', -- Vrije Universiteit Amsterdam
'openorgs____::4d4051b56708688235252f1d8fddb8c1', --Iscte - Instituto Universitário de Lisboa
'openorgs____::ab4ac74c35fa5dada770cf08e5110fab', -- Universidade Católica Portuguesa
'openorgs____::4d4051b56708688235252f1d8fddb8c1', -- Iscte - Instituto Universitário de Lisboa
'openorgs____::5d55fb216b14691cf68218daf5d78cd9', -- Munster Technological University
'openorgs____::0fccc7640f0cb44d5cd1b06b312a06b9', -- Cardiff University
'openorgs____::8839b55dae0c84d56fd533f52d5d483a' -- Leibniz Institute of Ecological Urban and Regional Development
) )) foo;
--ANALYZE TABLE TARGET.result COMPUTE STATISTICS;
create view if not exists TARGET.category as select * from SOURCE.category;
create view if not exists TARGET.concept as select * from SOURCE.concept;
create view if not exists TARGET.context as select * from SOURCE.context;
create view if not exists TARGET.country as select * from SOURCE.country;
create view if not exists TARGET.countrygdp as select * from SOURCE.countrygdp;
create view if not exists TARGET.creation_date as select * from SOURCE.creation_date;
create view if not exists TARGET.funder as select * from SOURCE.funder;
create view if not exists TARGET.fundref as select * from SOURCE.fundref;
create view if not exists TARGET.rndexpenditure as select * from SOURCE.rndexpediture;
create view if not exists TARGET.rndgdpexpenditure as select * from SOURCE.rndgdpexpenditure;
create view if not exists TARGET.doctoratestudents as select * from SOURCE.doctoratestudents;
create view if not exists TARGET.totalresearchers as select * from SOURCE.totalresearchers;
create view if not exists TARGET.totalresearchersft as select * from SOURCE.totalresearchersft;
create view if not exists TARGET.hrrst as select * from SOURCE.hrrst;
--create view if not exists TARGET.graduatedoctorates as select * from SOURCE.graduatedoctorates;
create table TARGET.result_citations stored as parquet as select * from SOURCE.result_citations orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_citations COMPUTE STATISTICS;
create table TARGET.result_references_oc stored as parquet as select * from SOURCE.result_references_oc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_references_oc COMPUTE STATISTICS;
create table TARGET.result_citations_oc stored as parquet as select * from SOURCE.result_citations_oc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_citations_oc COMPUTE STATISTICS;
create table TARGET.result_classifications stored as parquet as select * from SOURCE.result_classifications orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_classifications COMPUTE STATISTICS;
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);
--ANALYZE TABLE TARGET.result_apc COMPUTE STATISTICS;
create table TARGET.result_concepts stored as parquet as select * from SOURCE.result_concepts orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_concepts COMPUTE STATISTICS;
create table TARGET.result_datasources stored as parquet as select * from SOURCE.result_datasources orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_datasources COMPUTE STATISTICS;
create table TARGET.result_fundercount stored as parquet as select * from SOURCE.result_fundercount orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_fundercount COMPUTE STATISTICS;
create table TARGET.result_gold stored as parquet as select * from SOURCE.result_gold orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_gold COMPUTE STATISTICS;
create table TARGET.result_greenoa stored as parquet as select * from SOURCE.result_greenoa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_greenoa COMPUTE STATISTICS;
create table TARGET.result_languages stored as parquet as select * from SOURCE.result_languages orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_languages COMPUTE STATISTICS;
create table TARGET.result_licenses stored as parquet as select * from SOURCE.result_licenses orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_licenses COMPUTE STATISTICS;
create table TARGET.licenses_normalized STORED AS PARQUET as select * from SOURCE.licenses_normalized;
--ANALYZE TABLE TARGET.licenses_normalized COMPUTE STATISTICS;
create table TARGET.result_oids stored as parquet as select * from SOURCE.result_oids orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_oids COMPUTE STATISTICS;
create table TARGET.result_organization stored as parquet as select * from SOURCE.result_organization orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_organization COMPUTE STATISTICS;
create table TARGET.result_peerreviewed stored as parquet as select * from SOURCE.result_peerreviewed orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_peerreviewed COMPUTE STATISTICS;
create table TARGET.result_pids stored as parquet as select * from SOURCE.result_pids orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_pids COMPUTE STATISTICS;
create table TARGET.result_projectcount stored as parquet as select * from SOURCE.result_projectcount orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_projectcount COMPUTE STATISTICS;
create table TARGET.result_projects stored as parquet as select * from SOURCE.result_projects orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_projects COMPUTE STATISTICS;
create table TARGET.result_refereed stored as parquet as select * from SOURCE.result_refereed orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_refereed COMPUTE STATISTICS;
create table TARGET.result_sources stored as parquet as select * from SOURCE.result_sources orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_sources COMPUTE STATISTICS;
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);
--ANALYZE TABLE TARGET.result_topics COMPUTE STATISTICS;
create table TARGET.result_fos stored as parquet as select * from SOURCE.result_fos orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_fos COMPUTE STATISTICS;
create table TARGET.result_accessroute stored as parquet as select * from SOURCE.result_accessroute orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_accessroute COMPUTE STATISTICS;
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;
drop view TARGET.foo1;
drop view TARGET.foo2;
--ANALYZE TABLE TARGET.result_result COMPUTE STATISTICS;
-- datasources
create view if not exists TARGET.datasource as select * from SOURCE.datasource;
create view if not exists TARGET.datasource_oids as select * from SOURCE.datasource_oids;
create view if not exists TARGET.datasource_organizations as select * from SOURCE.datasource_organizations;
create view if not exists TARGET.datasource_sources as select * from SOURCE.datasource_sources;
create table TARGET.datasource_results stored as parquet as select id as result, datasource as id from TARGET.result_datasources;
--ANALYZE TABLE TARGET.datasource_results COMPUTE STATISTICS;
-- organizations
create view if not exists TARGET.organization as select * from SOURCE.organization;
create view if not exists TARGET.organization_datasources as select * from SOURCE.organization_datasources;
create view if not exists TARGET.organization_pids as select * from SOURCE.organization_pids;
create view if not exists TARGET.organization_projects as select * from SOURCE.organization_projects;
create view if not exists TARGET.organization_sources as select * from SOURCE.organization_sources;
-- projects
create view if not exists TARGET.project as select * from SOURCE.project;
create view if not exists TARGET.project_oids as select * from SOURCE.project_oids;
create view if not exists TARGET.project_organizations as select * from SOURCE.project_organizations;
create view if not exists TARGET.project_resultcount as select * from SOURCE.project_resultcount;
create view if not exists TARGET.project_classification as select * from SOURCE.project_classification;
create view if not exists TARGET.project_organization_contribution as select * from SOURCE.project_organization_contribution;
create table TARGET.project_results stored as parquet as select id as result, project as id from TARGET.result_projects;
--ANALYZE TABLE TARGET.project_results COMPUTE STATISTICS;
-- indicators
-- Sprint 1 ----
create table TARGET.indi_pub_green_oa stored as parquet as select * from SOURCE.indi_pub_green_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_green_oa COMPUTE STATISTICS;
create table TARGET.indi_pub_grey_lit stored as parquet as select * from SOURCE.indi_pub_grey_lit orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_grey_lit COMPUTE STATISTICS;
create table TARGET.indi_pub_doi_from_crossref stored as parquet as select * from SOURCE.indi_pub_doi_from_crossref orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_doi_from_crossref COMPUTE STATISTICS;
-- Sprint 2 ----
create table TARGET.indi_result_has_cc_licence stored as parquet as select * from SOURCE.indi_result_has_cc_licence orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_result_has_cc_licence COMPUTE STATISTICS;
create table TARGET.indi_result_has_cc_licence_url stored as parquet as select * from SOURCE.indi_result_has_cc_licence_url orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_result_has_cc_licence_url COMPUTE STATISTICS;
create table TARGET.indi_pub_has_abstract stored as parquet as select * from SOURCE.indi_pub_has_abstract orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_has_abstract COMPUTE STATISTICS;
create table TARGET.indi_result_with_orcid stored as parquet as select * from SOURCE.indi_result_with_orcid orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_result_with_orcid COMPUTE STATISTICS;
---- Sprint 3 ----
create table TARGET.indi_funded_result_with_fundref stored as parquet as select * from SOURCE.indi_funded_result_with_fundref orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_funded_result_with_fundref COMPUTE STATISTICS;
create view TARGET.indi_result_org_collab as select * from SOURCE.indi_result_org_collab;
create view TARGET.indi_result_org_country_collab as select * from SOURCE.indi_result_org_country_collab;
create view TARGET.indi_project_collab_org as select * from SOURCE.indi_project_collab_org;
create view TARGET.indi_project_collab_org_country as select * from SOURCE.indi_project_collab_org_country;
create view TARGET.indi_funder_country_collab as select * from SOURCE.indi_funder_country_collab;
create view TARGET.indi_result_country_collab as select * from SOURCE.indi_result_country_collab;
---- Sprint 4 ----
create table TARGET.indi_pub_diamond stored as parquet as select * from SOURCE.indi_pub_diamond orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_diamond COMPUTE STATISTICS;
create table TARGET.indi_pub_in_transformative stored as parquet as select * from SOURCE.indi_pub_in_transformative orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_in_transformative COMPUTE STATISTICS;
create table TARGET.indi_pub_closed_other_open stored as parquet as select * from SOURCE.indi_pub_closed_other_open orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_closed_other_open COMPUTE STATISTICS;
---- Sprint 5 ----
create table TARGET.indi_result_no_of_copies stored as parquet as select * from SOURCE.indi_result_no_of_copies orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_result_no_of_copies COMPUTE STATISTICS;
---- Sprint 6 ----
create table TARGET.indi_pub_hybrid_oa_with_cc stored as parquet as select * from SOURCE.indi_pub_hybrid_oa_with_cc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_hybrid_oa_with_cc COMPUTE STATISTICS;
create table TARGET.indi_pub_bronze_oa stored as parquet as select * from SOURCE.indi_pub_bronze_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_bronze_oa COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads stored as parquet as select * from SOURCE.indi_pub_downloads orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
--ANALYZE TABLE TARGET.indi_pub_downloads COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads_datasource stored as parquet as select * from SOURCE.indi_pub_downloads_datasource orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
--ANALYZE TABLE TARGET.indi_pub_downloads_datasource COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads_year stored as parquet as select * from SOURCE.indi_pub_downloads_year orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
--ANALYZE TABLE TARGET.indi_pub_downloads_year COMPUTE STATISTICS;
create table TARGET.indi_pub_downloads_datasource_year stored as parquet as select * from SOURCE.indi_pub_downloads_datasource_year orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);
--ANALYZE TABLE TARGET.indi_pub_downloads_datasource_year COMPUTE STATISTICS;
---- Sprint 7 ----
create table TARGET.indi_pub_gold_oa stored as parquet as select * from SOURCE.indi_pub_gold_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_gold_oa COMPUTE STATISTICS;
create table TARGET.indi_pub_hybrid stored as parquet as select * from SOURCE.indi_pub_hybrid orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_hybrid COMPUTE STATISTICS;
create view TARGET.indi_org_fairness as select * from SOURCE.indi_org_fairness;
create view TARGET.indi_org_fairness_pub_pr as select * from SOURCE.indi_org_fairness_pub_pr;
create view TARGET.indi_org_fairness_pub_year as select * from SOURCE.indi_org_fairness_pub_year;
create view TARGET.indi_org_fairness_pub as select * from SOURCE.indi_org_fairness_pub;
create view TARGET.indi_org_fairness_year as select * from SOURCE.indi_org_fairness_year;
create view TARGET.indi_org_findable_year as select * from SOURCE.indi_org_findable_year;
create view TARGET.indi_org_findable as select * from SOURCE.indi_org_findable;
create view TARGET.indi_org_openess as select * from SOURCE.indi_org_openess;
create view TARGET.indi_org_openess_year as select * from SOURCE.indi_org_openess_year;
create table TARGET.indi_pub_has_preprint stored as parquet as select * from SOURCE.indi_pub_has_preprint orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_has_preprint COMPUTE STATISTICS;
create table TARGET.indi_pub_in_subscribed stored as parquet as select * from SOURCE.indi_pub_in_subscribed orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_in_subscribed COMPUTE STATISTICS;
create table TARGET.indi_result_with_pid stored as parquet as select * from SOURCE.indi_result_with_pid orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_result_with_pid COMPUTE STATISTICS;
create table TARGET.indi_impact_measures stored as parquet as select * from SOURCE.indi_impact_measures orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_impact_measures COMPUTE STATISTICS;
create table TARGET.indi_pub_interdisciplinarity stored as parquet as select * from SOURCE.indi_pub_interdisciplinarity orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.indi_pub_interdisciplinarity COMPUTE STATISTICS;
create table TARGET.result_apc_affiliations stored as parquet as select * from SOURCE.result_apc_affiliations orig where exists (select 1 from TARGET.result r where r.id=orig.id);
--ANALYZE TABLE TARGET.result_apc_affiliations COMPUTE STATISTICS;
create table TARGET.indi_is_project_result_after stored as parquet as select * from SOURCE.indi_is_project_result_after orig where exists (select 1 from TARGET.result r where r.id=orig.id);
create table TARGET.indi_is_funder_plan_s stored as parquet as select * from SOURCE.indi_is_funder_plan_s orig where exists (select 1 from TARGET.result r where r.id=orig.id);
create view TARGET.indi_funder_fairness as select * from SOURCE.indi_funder_fairness;
create view TARGET.indi_funder_openess as select * from SOURCE.indi_funder_openess;
create view TARGET.indi_funder_findable as select * from SOURCE.indi_funder_findable;
create view TARGET.indi_ris_fairness as select * from SOURCE.indi_ris_fairness;
create view TARGET.indi_ris_openess as select * from SOURCE.indi_ris_openess;
create view TARGET.indi_ris_findable as select * from SOURCE.indi_ris_findable;

2
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDB_RIs.sql
View File

@ -12,4 +12,4 @@ create table TARGET.result stored as parquet as
-- join SOURCE.result
where rc.id=r.id and conc.category like CONTEXT)
) foo;
ANALYZE TABLE TARGET.result COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result COMPUTE STATISTICS;

2
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDB_RIs_tail.sql
View File

@ -12,4 +12,4 @@ create table TARGET.result stored as parquet as
-- join SOURCE.result
where rc.id=r.id and conc.category not in (CONTEXTS))
) foo;
ANALYZE TABLE TARGET.result COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result COMPUTE STATISTICS;

2
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDB_funded.sql
View File

@ -6,4 +6,4 @@ create table TARGET.result stored as parquet as
select * from SOURCE.result r where exists (select 1 from SOURCE.result_projects rp join SOURCE.project p on rp.project=p.id where rp.id=r.id)
) foo;
ANALYZE TABLE TARGET.result COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result COMPUTE STATISTICS;

12
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDB_institutions.sql
View File

@ -42,7 +42,7 @@ create table TARGET.result stored as parquet as
'openorgs____::31f2fa9e05b49d4cf40a19c3fed8eb06', -- Saints Cyril and Methodius University of Skopje
'openorgs____::db7686f30f22cbe73a4fde872ce812a6', -- University of Milan
'openorgs____::b8b8ca674452579f3f593d9f5e557483', -- University College Cork
'openorgs____::38d7097854736583dde879d12dacafca' -- Brown University
'openorgs____::38d7097854736583dde879d12dacafca', -- Brown University
'openorgs____::57784c9e047e826fefdb1ef816120d92', --Arts et Métiers ParisTech
'openorgs____::2530baca8a15936ba2e3297f2bce2e7e', -- University of Cape Town
'openorgs____::d11f981828c485cd23d93f7f24f24db1', -- Technological University Dublin
@ -52,7 +52,13 @@ create table TARGET.result stored as parquet as
'openorgs____::66aa9fc2fceb271423dfabcc38752dc0', -- Lund University
'openorgs____::3cff625a4370d51e08624cc586138b2f', -- IMT Atlantique
'openorgs____::c0b262bd6eab819e4c994914f9c010e2', -- National Institute of Geophysics and Volcanology
'openorgs____::1624ff7c01bb641b91f4518539a0c28a' -- Vrije Universiteit Amsterdam
'openorgs____::1624ff7c01bb641b91f4518539a0c28a', -- Vrije Universiteit Amsterdam
'openorgs____::4d4051b56708688235252f1d8fddb8c1', --Iscte - Instituto Universitário de Lisboa
'openorgs____::ab4ac74c35fa5dada770cf08e5110fab', -- Universidade Católica Portuguesa
'openorgs____::4d4051b56708688235252f1d8fddb8c1', -- Iscte - Instituto Universitário de Lisboa
'openorgs____::5d55fb216b14691cf68218daf5d78cd9', -- Munster Technological University
'openorgs____::0fccc7640f0cb44d5cd1b06b312a06b9', -- Cardiff University
'openorgs____::8839b55dae0c84d56fd533f52d5d483a' -- Leibniz Institute of Ecological Urban and Regional Development
))) foo;
ANALYZE TABLE TARGET.result COMPUTE STATISTICS;
--ANALYZE TABLE TARGET.result COMPUTE STATISTICS;

38
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step21-createObservatoryDB.sql
View File

@ -8,7 +8,7 @@ from ${stats_db_name}.result r
group by rl.id
) rln on rln.id=r.id;
ANALYZE TABLE ${observatory_db_name}.result_cc_licence COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_cc_licence COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_country stored as parquet as
select
@ -39,7 +39,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_country COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_affiliated_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_year stored as parquet as
select
@ -70,7 +70,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, r.year;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_year COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_affiliated_year COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_year_country stored as parquet as
select
@ -101,7 +101,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, r.year, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_year_country COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_affiliated_year_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_datasource stored as parquet as
select
@ -134,7 +134,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, d.name;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_datasource COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_affiliated_datasource COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_datasource_country stored as parquet as
select
@ -167,7 +167,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, d.name, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_datasource_country COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_affiliated_datasource_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_organization stored as parquet as
select
@ -198,7 +198,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, o.name;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_organization COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_affiliated_organization COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_organization_country stored as parquet as
select
@ -229,7 +229,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, o.name, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_organization_country COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_affiliated_organization_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_funder stored as parquet as
select
@ -262,7 +262,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, p.funder;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_funder COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_affiliated_funder COMPUTE STATISTICS;
create table ${observatory_db_name}.result_affiliated_funder_country stored as parquet as
select
@ -295,7 +295,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, p.funder, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_affiliated_funder_country COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_affiliated_funder_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_country stored as parquet as
select
@ -328,7 +328,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_deposited_country COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_deposited_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_year stored as parquet as
select
@ -361,7 +361,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, r.year;
ANALYZE TABLE ${observatory_db_name}.result_deposited_year COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_deposited_year COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_year_country stored as parquet as
select
@ -394,7 +394,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, r.year, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_deposited_year_country COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_deposited_year_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_datasource stored as parquet as
select
@ -427,7 +427,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, d.name;
ANALYZE TABLE ${observatory_db_name}.result_deposited_datasource COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_deposited_datasource COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_datasource_country stored as parquet as
select
@ -460,7 +460,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, d.name, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_deposited_datasource_country COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_deposited_datasource_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_organization stored as parquet as
select
@ -493,7 +493,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, o.name;
ANALYZE TABLE ${observatory_db_name}.result_deposited_organization COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_deposited_organization COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_organization_country stored as parquet as
select
@ -526,7 +526,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, o.name, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_deposited_organization_country COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_deposited_organization_country COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_funder stored as parquet as
select
@ -561,7 +561,7 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, p.funder;
ANALYZE TABLE ${observatory_db_name}.result_deposited_funder COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_deposited_funder COMPUTE STATISTICS;
create table ${observatory_db_name}.result_deposited_funder_country stored as parquet as
select
@ -596,4 +596,4 @@ group by r.green, r.gold, case when rl.type is not null then true else false end
case when r.access_mode in ('Open Access', 'Open Source') then true else false end, r.peer_reviewed, r.type, abstract,
cc_licence, r.authors > 1, rpc.count > 1, rfc.count > 1, p.funder, c.code, c.name;
ANALYZE TABLE ${observatory_db_name}.result_deposited_funder_country COMPUTE STATISTICS;
--ANALYZE TABLE ${observatory_db_name}.result_deposited_funder_country COMPUTE STATISTICS;

0
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step5.sql Normal file → Executable file
View File

16
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/workflow.xml
View File

@ -317,15 +317,12 @@
</action>
<action name="Step16-createIndicatorsTables">
<shell xmlns="uri:oozie:shell-action:0.1">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<exec>indicators.sh</exec>
<argument>${stats_db_name}</argument>
<argument>${external_stats_db_name}</argument>
<argument>${wf:appPath()}/scripts/step16-createIndicatorsTables.sql</argument>
<file>indicators.sh</file>
</shell>
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step16-createIndicatorsTables.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>external_stats_db_name=${external_stats_db_name}</param>
</hive2>
<ok to="Step16_1-definitions"/>
<error to="Kill"/>
</action>
@ -378,6 +375,7 @@
<argument>${wf:appPath()}/scripts/step20-createMonitorDB_institutions.sql</argument>
<argument>${wf:appPath()}/scripts/step20-createMonitorDB_RIs.sql</argument>
<argument>${wf:appPath()}/scripts/step20-createMonitorDB_RIs_tail.sql</argument>
<argument>${wf:appPath()}/scripts/step20-createMonitorDBAll.sql</argument>
<file>monitor.sh</file>
</shell>
<ok to="step21-createObservatoryDB-pre"/>