alessia.bardi
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Merge pull request 'master' (#111) from antonis.lempesis/dnet-hadoop:master into master 

Reviewed-on: D-Net/dnet-hadoop#111
This commit is contained in:
Claudio Atzori 2021-05-28 14:09:12 +02:00
parent 338327171d 168edcbde3
commit a4cfabdbc6
8 changed files with 336 additions and 75 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/workflow.xml
View File

@ -16,6 +16,14 @@
<name>monitor_db_production_name</name> <name>monitor_db_production_name</name>
<description>the name of the monitor public database</description> <description>the name of the monitor public database</description>
</property> </property>
<property>
<name>observatory_db_name</name>
<description>the monitor database name</description>
</property>
<property>
<name>observatory_db_production_name</name>
<description>the name of the monitor public database</description>
</property>
<property> <property>
<name>stats_tool_api_url</name> <name>stats_tool_api_url</name>
<description>The url of the API of the stats tool. Is used to trigger the cache promote.</description> <description>The url of the API of the stats tool. Is used to trigger the cache promote.</description>
@ -77,6 +85,19 @@
<argument>${monitor_db_production_name}</argument> <argument>${monitor_db_production_name}</argument>
<file>updateProductionViews.sh</file> <file>updateProductionViews.sh</file>
</shell> </shell>
<ok to="updateObservatoryViews"/>
<error to="Kill"/>
</action>
<action name="updateObservatoryViews">
<shell xmlns="uri:oozie:shell-action:0.1">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<exec>updateProductionViews.sh</exec>
<argument>${observatory_db_name}</argument>
<argument>${observatory_db_production_name}</argument>
<file>updateProductionViews.sh</file>
</shell>
<ok to="promoteCache"/> <ok to="promoteCache"/>
<error to="Kill"/> <error to="Kill"/>
</action> </action>

28
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/observatory.sh Normal file
View File

@ -0,0 +1,28 @@
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
echo "Getting file from " $4
hdfs dfs -copyToLocal $4
echo "Creating observatory database"
impala-shell -q "drop database if exists ${TARGET} cascade"
impala-shell -q "create database if not exists ${TARGET}"
impala-shell -d ${SOURCE} -q "show tables" --delimited | sed "s/\(.*\)/create view ${TARGET}.\1 as select * from ${SOURCE}.\1;/" | impala-shell -f -
cat step21-createObservatoryDB.sql | sed s/SOURCE/$1/g | sed s/TARGET/$2/g1 | impala-shell -f -
echo "Impala shell finished"
echo "Updating shadow observatory database"
impala-shell -q "create database if not exists ${SHADOW}"
impala-shell -d ${SHADOW} -q "show tables" --delimited | sed "s/^/drop view if exists ${SHADOW}./" | sed "s/$/;/" | impala-shell -f -
impala-shell -d ${TARGET} -q "show tables" --delimited | sed "s/\(.*\)/create view ${SHADOW}.\1 as select * from ${TARGET}.\1;/" | impala-shell -f -
echo "Shadow db ready!"

32
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step12.sql
View File

@ -45,35 +45,3 @@ FROM ${stats_db_name}.dataset
UNION ALL UNION ALL
SELECT *, bestlicence AS access_mode SELECT *, bestlicence AS access_mode
FROM ${stats_db_name}.otherresearchproduct; FROM ${stats_db_name}.otherresearchproduct;
-------------------------------------------------------------------------------
-- To see with Antonis if the following is needed and where it should be placed
-------------------------------------------------------------------------------
CREATE TABLE ${stats_db_name}.numbers_country AS
SELECT org.country AS country, count(distinct rd.datasource) AS datasources, count(distinct r.id) AS publications
FROM ${stats_db_name}.result r,
${stats_db_name}.result_datasources rd,
${stats_db_name}.datasource d,
${stats_db_name}.datasource_organizations dor,
${stats_db_name}.organization org
WHERE r.id = rd.id
AND rd.datasource = d.id
AND d.id = dor.id
AND dor.organization = org.id
AND r.type = 'publication'
AND r.bestlicence = 'Open Access'
GROUP BY org.country;
-- ANALYZE TABLE ${stats_db_name}.datasource COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.datasource COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.publication COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.publication COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.dataset COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.dataset COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.software COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.software COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.otherresearchproduct COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.otherresearchproduct COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.numbers_country COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.numbers_country COMPUTE STATISTICS FOR COLUMNS;

31
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16.sql
View File

@ -59,33 +59,4 @@ from result_gold
union all union all
select distinct r.id, false as gold select distinct r.id, false as gold
from ${stats_db_name}.result r from ${stats_db_name}.result r
where r.id not in (select id from result_gold); where r.id not in (select id from result_gold);
-- shortcut result-country through the organization affiliation
create table ${stats_db_name}.result_affiliated_country as
select r.id as id, o.country as country
from ${stats_db_name}.result r
join ${stats_db_name}.result_organization ro on ro.id=r.id
join ${stats_db_name}.organization o on o.id=ro.organization
where o.country is not null and o.country!='';
-- shortcut result-country through datasource of deposition
create table ${stats_db_name}.result_deposited_country as
select r.id as id, o.country as country
from ${stats_db_name}.result r
join ${stats_db_name}.result_datasources rd on rd.id=r.id
join ${stats_db_name}.datasource d on d.id=rd.datasource
join ${stats_db_name}.datasource_organizations dor on dor.id=d.id
join ${stats_db_name}.organization o on o.id=dor.organization
where o.country is not null and o.country!='';
-- ANALYZE TABLE ${stats_db_name}.result_peerreviewed COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.result_peerreviewed COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.result_greenoa COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.result_greenoa COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.result_gold COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.result_gold COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.result_affiliated_country COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.result_affiliated_country COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.result_deposited_country COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.result_deposited_country COMPUTE STATISTICS FOR COLUMNS;

5
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_5.sql
View File

@ -52,7 +52,4 @@ LEFT OUTER JOIN ${stats_db_name}.result_gold gold on gold.id=r.id;
drop table if exists ${stats_db_name}.result; drop table if exists ${stats_db_name}.result;
drop view if exists ${stats_db_name}.result; drop view if exists ${stats_db_name}.result;
create table ${stats_db_name}.result stored as parquet as select * from ${stats_db_name}.result_tmp; create table ${stats_db_name}.result stored as parquet as select * from ${stats_db_name}.result_tmp;
drop table ${stats_db_name}.result_tmp; drop table ${stats_db_name}.result_tmp;
--
-- ANALYZE TABLE ${stats_db_name}.result COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.result COMPUTE STATISTICS FOR COLUMNS;

6
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDB.sql
View File

@ -19,9 +19,6 @@ create table TARGET.result as
select * from SOURCE.result r where exists (select 1 from SOURCE.result_concepts rc where rc.id=r.id) ) foo; select * from SOURCE.result r where exists (select 1 from SOURCE.result_concepts rc where rc.id=r.id) ) foo;
compute stats TARGET.result; compute stats TARGET.result;
create table TARGET.result_affiliated_country as select * from SOURCE.result_affiliated_country rac where exists (select 1 from TARGET.result r where r.id=rac.id);
compute stats TARGET.result_affiliated_country;
create table TARGET.result_citations as select * from SOURCE.result_citations orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_citations as select * from SOURCE.result_citations orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.result_citations; compute stats TARGET.result_citations;
@ -34,9 +31,6 @@ compute stats TARGET.result_concepts;
create table TARGET.result_datasources as select * from SOURCE.result_datasources orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_datasources as select * from SOURCE.result_datasources orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.result_datasources; compute stats TARGET.result_datasources;
create table TARGET.result_deposited_country as select * from SOURCE.result_deposited_country orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.result_deposited_country;
create table TARGET.result_fundercount as select * from SOURCE.result_fundercount orig where exists (select 1 from TARGET.result r where r.id=orig.id); create table TARGET.result_fundercount as select * from SOURCE.result_fundercount orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.result_fundercount; compute stats TARGET.result_fundercount;

259
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step21-createObservatoryDB.sql Normal file
View File

@ -0,0 +1,259 @@
create table TARGET.result_affiliated_country stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa,
r.peer_reviewed, r.type, c.code as ccode, c.name as cname
from SOURCE.result r
join SOURCE.result_organization ro on ro.id=r.id
join SOURCE.organization o on o.id=ro.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, c.code, c.name;
create table TARGET.result_affiliated_year stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa, r.peer_reviewed, r.type, r.year
from SOURCE.result r
join SOURCE.result_organization ro on ro.id=r.id
join SOURCE.organization o on o.id=ro.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, r.year;
create table TARGET.result_affiliated_year_country stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa,
r.peer_reviewed, r.type, r.year, c.code as ccode, c.name as cname
from SOURCE.result r
join SOURCE.result_organization ro on ro.id=r.id
join SOURCE.organization o on o.id=ro.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, r.year, c.code, c.name;
create table TARGET.result_affiliated_datasource stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa, r.peer_reviewed, r.type, d.name as dname
from SOURCE.result r
join SOURCE.result_organization ro on ro.id=r.id
join SOURCE.organization o on o.id=ro.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
left outer join SOURCE.result_datasources rd on rd.id=r.id
left outer join SOURCE.datasource d on d.id=rd.datasource
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, d.name;
create table TARGET.result_affiliated_datasource_country stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa,
r.peer_reviewed, r.type, d.name as dname, c.code as ccode, c.name as cname
from SOURCE.result r
join SOURCE.result_organization ro on ro.id=r.id
join SOURCE.organization o on o.id=ro.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
left outer join SOURCE.result_datasources rd on rd.id=r.id
left outer join SOURCE.datasource d on d.id=rd.datasource
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, d.name, c.code, c.name;
create table TARGET.result_affiliated_organization stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa,
r.peer_reviewed, r.type, o.name as oname
from SOURCE.result r
join SOURCE.result_organization ro on ro.id=r.id
join SOURCE.organization o on o.id=ro.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, o.name;
create table TARGET.result_affiliated_organization_country stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa,
r.peer_reviewed, r.type, o.name as oname, c.code as ccode, c.name as cname
from SOURCE.result r
join SOURCE.result_organization ro on ro.id=r.id
join SOURCE.organization o on o.id=ro.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, o.name, c.code, c.name;
create table TARGET.result_affiliated_funder stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa, r.peer_reviewed, r.type, p.funder as pfunder
from SOURCE.result r
join SOURCE.result_organization ro on ro.id=r.id
join SOURCE.organization o on o.id=ro.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
join SOURCE.result_projects rp on rp.id=r.id
join SOURCE.project p on p.id=rp.project
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, p.funder;
create table TARGET.result_affiliated_funder_country stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa,
r.peer_reviewed, r.type, p.funder as pfunder, c.code as ccode, c.name as cname
from SOURCE.result r
join SOURCE.result_organization ro on ro.id=r.id
join SOURCE.organization o on o.id=ro.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
join SOURCE.result_projects rp on rp.id=r.id
join SOURCE.project p on p.id=rp.project
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, p.funder, c.code, c.name;
create table TARGET.result_deposited_country stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa,
r.peer_reviewed, r.type, c.code as ccode, c.name as cname
from SOURCE.result r
join SOURCE.result_datasources rd on rd.id=r.id
join SOURCE.datasource d on d.id=rd.datasource and d.type in ('Institutional Repository','Data Repository', 'Repository', 'Publication Repository')
join SOURCE.datasource_organizations dor on dor.id=d.id
join SOURCE.organization o on o.id=dor.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, c.code, c.name;
create table TARGET.result_deposited_year stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa, r.peer_reviewed, r.type, r.year
from SOURCE.result r
join SOURCE.result_datasources rd on rd.id=r.id
join SOURCE.datasource d on d.id=rd.datasource and d.type in ('Institutional Repository','Data Repository', 'Repository', 'Publication Repository')
join SOURCE.datasource_organizations dor on dor.id=d.id
join SOURCE.organization o on o.id=dor.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, r.year;
create table TARGET.result_deposited_year_country stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa,
r.peer_reviewed, r.type, r.year, c.code as ccode, c.name as cname
from SOURCE.result r
join SOURCE.result_datasources rd on rd.id=r.id
join SOURCE.datasource d on d.id=rd.datasource and d.type in ('Institutional Repository','Data Repository', 'Repository', 'Publication Repository')
join SOURCE.datasource_organizations dor on dor.id=d.id
join SOURCE.organization o on o.id=dor.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, r.year, c.code, c.name;
create table TARGET.result_deposited_datasource stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa,
r.peer_reviewed, r.type, d.name as dname
from SOURCE.result r
join SOURCE.result_datasources rd on rd.id=r.id
join SOURCE.datasource d on d.id=rd.datasource and d.type in ('Institutional Repository','Data Repository', 'Repository', 'Publication Repository')
join SOURCE.datasource_organizations dor on dor.id=d.id
join SOURCE.organization o on o.id=dor.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, d.name;
create table TARGET.result_deposited_datasource_country stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa,
r.peer_reviewed, r.type, d.name as dname, c.code as ccode, c.name as cname
from SOURCE.result r
join SOURCE.result_datasources rd on rd.id=r.id
join SOURCE.datasource d on d.id=rd.datasource and d.type in ('Institutional Repository','Data Repository', 'Repository', 'Publication Repository')
join SOURCE.datasource_organizations dor on dor.id=d.id
join SOURCE.organization o on o.id=dor.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, d.name, c.code, c.name;
create table TARGET.result_deposited_organization stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa, r.peer_reviewed, r.type, o.name as oname
from SOURCE.result r
join SOURCE.result_datasources rd on rd.id=r.id
join SOURCE.datasource d on d.id=rd.datasource and d.type in ('Institutional Repository','Data Repository', 'Repository', 'Publication Repository')
join SOURCE.datasource_organizations dor on dor.id=d.id
join SOURCE.organization o on o.id=dor.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, o.name;
create table TARGET.result_deposited_organization_country stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa,
r.peer_reviewed, r.type, o.name as oname, c.code as ccode, c.name as cname
from SOURCE.result r
join SOURCE.result_datasources rd on rd.id=r.id
join SOURCE.datasource d on d.id=rd.datasource and d.type in ('Institutional Repository','Data Repository', 'Repository', 'Publication Repository')
join SOURCE.datasource_organizations dor on dor.id=d.id
join SOURCE.organization o on o.id=dor.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, o.name, c.code, c.name;
create table TARGET.result_deposited_funder stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa,
r.peer_reviewed, r.type, p.funder as pfunder
from SOURCE.result r
join SOURCE.result_datasources rd on rd.id=r.id
join SOURCE.datasource d on d.id=rd.datasource and d.type in ('Institutional Repository','Data Repository', 'Repository', 'Publication Repository')
join SOURCE.datasource_organizations dor on dor.id=d.id
join SOURCE.organization o on o.id=dor.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
join SOURCE.result_projects rp on rp.id=r.id
join SOURCE.project p on p.id=rp.project
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, p.funder;
create table TARGET.result_deposited_funder_country stored as parquet as
select count(distinct r.id) as total, r.green, r.gold, case when rl.type is not null then true else false end as licence,
case when pids.pid is not null then true else false end as pid, case when r.access_mode in ('Open Access', 'Open Source') then true else false end as oa,
r.peer_reviewed, r.type, p.funder as pfunder, c.code as ccode, c.name as cname
from SOURCE.result r
join SOURCE.result_datasources rd on rd.id=r.id
join SOURCE.datasource d on d.id=rd.datasource and d.type in ('Institutional Repository','Data Repository', 'Repository', 'Publication Repository')
join SOURCE.datasource_organizations dor on dor.id=d.id
join SOURCE.organization o on o.id=dor.organization
join SOURCE.country c on c.code=o.country and c.continent_name='Europe'
join SOURCE.result_projects rp on rp.id=r.id
join SOURCE.project p on p.id=rp.project
left outer join SOURCE.result_licenses rl on rl.id=r.id
left outer join SOURCE.result_pids pids on pids.id=r.id
group by r.green, r.gold, licence, pid, oa, r.peer_reviewed, r.type, p.funder, c.code, c.name;
compute stats TARGET.result_affiliated_country;
compute stats TARGET.result_affiliated_year;
compute stats TARGET.result_affiliated_year_country;
compute stats TARGET.result_affiliated_datasource;
compute stats TARGET.result_affiliated_datasource_country;
compute stats TARGET.result_affiliated_organization;
compute stats TARGET.result_affiliated_organization_country;
compute stats TARGET.result_affiliated_funder;
compute stats TARGET.result_affiliated_funder_country;
compute stats TARGET.result_deposited_country;
compute stats TARGET.result_deposited_year;
compute stats TARGET.result_deposited_year_country;
compute stats TARGET.result_deposited_datasource;
compute stats TARGET.result_deposited_datasource_country;
compute stats TARGET.result_deposited_organization;
compute stats TARGET.result_deposited_organization_country;
compute stats TARGET.result_deposited_funder;
compute stats TARGET.result_deposited_funder_country;

29
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/workflow.xml
View File

@ -25,6 +25,14 @@
<name>monitor_db_shadow_name</name> <name>monitor_db_shadow_name</name>
<description>the name of the shadow monitor db</description> <description>the name of the shadow monitor db</description>
</property> </property>
<property>
<name>observatory_db_name</name>
<description>the target monitor db name</description>
</property>
<property>
<name>observatory_db_shadow_name</name>
<description>the name of the shadow monitor db</description>
</property>
<property> <property>
<name>stats_tool_api_url</name> <name>stats_tool_api_url</name>
<description>The url of the API of the stats tool. Is used to trigger the cache update.</description> <description>The url of the API of the stats tool. Is used to trigger the cache update.</description>
@ -305,11 +313,26 @@
<argument>${wf:appPath()}/scripts/step20-createMonitorDB.sql</argument> <argument>${wf:appPath()}/scripts/step20-createMonitorDB.sql</argument>
<file>monitor.sh</file> <file>monitor.sh</file>
</shell> </shell>
<ok to="Step21"/> <ok to="step21-createObservatoryDB"/>
<error to="Kill"/> <error to="Kill"/>
</action> </action>
<action name="Step21"> <action name="step21-createObservatoryDB">
<shell xmlns="uri:oozie:shell-action:0.1">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<exec>observatory.sh</exec>
<argument>${stats_db_name}</argument>
<argument>${observatory_db_name}</argument>
<argument>${observatory_db_shadow_name}</argument>
<argument>${wf:appPath()}/scripts/step21-createObservatoryDB.sql</argument>
<file>observatory.sh</file>
</shell>
<ok to="Step22"/>
<error to="Kill"/>
</action>
<action name="Step22">
<shell xmlns="uri:oozie:shell-action:0.1"> <shell xmlns="uri:oozie:shell-action:0.1">
<job-tracker>${jobTracker}</job-tracker> <job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node> <name-node>${nameNode}</name-node>
@ -322,4 +345,4 @@
</action> </action>
<end name="End"/> <end name="End"/>
</workflow-app> </workflow-app>