D-Net
/
dnet-hadoop
14
5
Fork 11
Code Issues 14 Pull Requests 13 Projects 2 Releases 2 Wiki Activity

Merge pull request 'Update step15_5.sql' (#313) from antonis.lempesis/dnet-hadoop:beta into beta 

Reviewed-on: #313
This commit is contained in:
Claudio Atzori 2023-06-21 10:26:16 +02:00
parent 85e016df17 74cb060bfe
commit 35e42a86ed
1 changed files with 11 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step15_5.sql
View File

@ -8,14 +8,14 @@ left outer join ${stats_db_name}.result_projects rp on rp.id=r.id
left outer join ${stats_db_name}.project p on p.id=rp.project left outer join ${stats_db_name}.project p on p.id=rp.project
group by r.id; group by r.id;
create table ${stats_db_name}.result_fundercount STORED AS PARQUET as create table if not exists ${stats_db_name}.result_fundercount STORED AS PARQUET as
select r.id, count(distinct p.funder) as count select r.id, count(distinct p.funder) as count
from ${stats_db_name}.result r from ${stats_db_name}.result r
left outer join ${stats_db_name}.result_projects rp on rp.id=r.id left outer join ${stats_db_name}.result_projects rp on rp.id=r.id
left outer join ${stats_db_name}.project p on p.id=rp.project left outer join ${stats_db_name}.project p on p.id=rp.project
group by r.id; group by r.id;
create table ${stats_db_name}.project_resultcount STORED AS PARQUET as create table if not exists ${stats_db_name}.project_resultcount STORED AS PARQUET as
with rcount as ( with rcount as (
select p.id as pid, count(distinct r.id) as `count`, r.type as type select p.id as pid, count(distinct r.id) as `count`, r.type as type
from ${stats_db_name}.project p from ${stats_db_name}.project p
@ -29,14 +29,14 @@ select rcount.pid, sum(case when rcount.type='publication' then rcount.count els
from rcount from rcount
group by rcount.pid; group by rcount.pid;
create view ${stats_db_name}.rndexpenditure as select * from stats_ext.rndexpediture; create or replace view ${stats_db_name}.rndexpenditure as select * from stats_ext.rndexpediture;
create view ${stats_db_name}.rndgdpexpenditure as select * from stats_ext.rndgdpexpenditure; create or replace view ${stats_db_name}.rndgdpexpenditure as select * from stats_ext.rndgdpexpenditure;
create view ${stats_db_name}.doctoratestudents as select * from stats_ext.doctoratestudents; create or replace view ${stats_db_name}.doctoratestudents as select * from stats_ext.doctoratestudents;
create view ${stats_db_name}.totalresearchers as select * from stats_ext.totalresearchers; create or replace view ${stats_db_name}.totalresearchers as select * from stats_ext.totalresearchers;
create view ${stats_db_name}.totalresearchersft as select * from stats_ext.totalresearchersft; create or replace view ${stats_db_name}.totalresearchersft as select * from stats_ext.totalresearchersft;
create view ${stats_db_name}.hrrst as select * from stats_ext.hrrst; create or replace view ${stats_db_name}.hrrst as select * from stats_ext.hrrst;
create table ${stats_db_name}.result_instance stored as parquet as create table if not exists ${stats_db_name}.result_instance stored as parquet as
select distinct r.* select distinct r.*
from ( from (
select substr(r.id, 4) as id, inst.accessright.classname as accessright, inst.accessright.openaccessroute as accessright_uw, substr(inst.collectedfrom.key, 4) as collectedfrom, select substr(r.id, 4) as id, inst.accessright.classname as accessright, inst.accessright.openaccessroute as accessright_uw, substr(inst.collectedfrom.key, 4) as collectedfrom,
@ -44,7 +44,7 @@ from (
from ${openaire_db_name}.result r lateral view explode(r.instance) instances as inst lateral view explode(inst.pid) pids as p) r from ${openaire_db_name}.result r lateral view explode(r.instance) instances as inst lateral view explode(inst.pid) pids as p) r
join ${stats_db_name}.result res on res.id=r.id; join ${stats_db_name}.result res on res.id=r.id;
create table ${stats_db_name}.result_apc STORED AS PARQUET as create table if not exists ${stats_db_name}.result_apc STORED AS PARQUET as
select r.id, r.amount, r.currency select r.id, r.amount, r.currency
from ( from (
select substr(r.id, 4) as id, cast(inst.processingchargeamount.value as float) as amount, inst.processingchargecurrency.value as currency select substr(r.id, 4) as id, cast(inst.processingchargeamount.value as float) as amount, inst.processingchargecurrency.value as currency
@ -52,4 +52,4 @@ from (
join ${stats_db_name}.result res on res.id=r.id join ${stats_db_name}.result res on res.id=r.id
where r.amount is not null; where r.amount is not null;
create view ${stats_db_name}.issn_gold_oa_dataset as select * from ${external_stats_db_name}.issn_gold_oa_dataset; create or replace view ${stats_db_name}.issn_gold_oa_dataset as select * from ${external_stats_db_name}.issn_gold_oa_dataset;