lsmyrnaios
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

Added fields: totalcost, fundedamount, currency, in project table

This commit is contained in:
dimitrispie 2022-10-27 16:44:26 +03:00
parent 2c0c3f1806
commit 5df9c63963
3 changed files with 13 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step11.sql
View File

@ -42,7 +42,9 @@ SELECT p.id,
CASE WHEN prr2.id IS NULL THEN 0 ELSE prr2.dp END AS delayedpubs, CASE WHEN prr2.id IS NULL THEN 0 ELSE prr2.dp END AS delayedpubs,
p.callidentifier, p.callidentifier,
p.code, p.code,
p.totalcost p.totalcost,
p.fundedamount,
p.currency
FROM ${stats_db_name}.project_tmp p FROM ${stats_db_name}.project_tmp p
LEFT JOIN (SELECT pr.id, count(distinct pr.result) AS np LEFT JOIN (SELECT pr.id, count(distinct pr.result) AS np
FROM ${stats_db_name}.project_results pr FROM ${stats_db_name}.project_results pr

8
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step13.sql
View File

@ -59,7 +59,7 @@ UNION ALL
SELECT * FROM ${stats_db_name}.otherresearchproduct_sources; SELECT * FROM ${stats_db_name}.otherresearchproduct_sources;
create table ${stats_db_name}.result_orcid STORED AS PARQUET as CREATE TABLE IF NOT EXISTS ${stats_db_name}.result_orcid STORED AS PARQUET as
select distinct res.id, regexp_replace(res.orcid, 'http://orcid.org/' ,'') as orcid select distinct res.id, regexp_replace(res.orcid, 'http://orcid.org/' ,'') as orcid
from ( from (
SELECT substr(res.id, 4) as id, auth_pid.value as orcid SELECT substr(res.id, 4) as id, auth_pid.value as orcid
@ -69,7 +69,7 @@ from (
LATERAL VIEW explode(auth.pid.qualifier.classid) apt as author_pid_type LATERAL VIEW explode(auth.pid.qualifier.classid) apt as author_pid_type
WHERE res.datainfo.deletedbyinference = FALSE and res.datainfo.invisible = FALSE and author_pid_type = 'orcid') as res; WHERE res.datainfo.deletedbyinference = FALSE and res.datainfo.invisible = FALSE and author_pid_type = 'orcid') as res;
create table ${stats_db_name}.result_result stored as parquet as CREATE TABLE IF NOT EXISTS ${stats_db_name}.result_result stored as parquet as
select substr(rel.source, 4) as source, substr(rel.target, 4) as target, relclass, subreltype select substr(rel.source, 4) as source, substr(rel.target, 4) as target, relclass, subreltype
from ${openaire_db_name}.relation rel from ${openaire_db_name}.relation rel
join ${openaire_db_name}.result r1 on rel.source=r1.id join ${openaire_db_name}.result r1 on rel.source=r1.id
@ -82,7 +82,7 @@ where reltype='resultResult'
and r2.resulttype.classname != 'other' and r2.resulttype.classname != 'other'
and rel.datainfo.deletedbyinference=false and rel.datainfo.invisible = FALSE; and rel.datainfo.deletedbyinference=false and rel.datainfo.invisible = FALSE;
create table ${stats_db_name}.result_citations_oc stored as parquet as CREATE TABLE IF NOT EXISTS ${stats_db_name}.result_citations_oc stored as parquet as
select substr(target, 4) as id, count(distinct substr(source, 4)) as citations select substr(target, 4) as id, count(distinct substr(source, 4)) as citations
from ${openaire_db_name}.relation rel from ${openaire_db_name}.relation rel
join ${openaire_db_name}.result r1 on rel.source=r1.id join ${openaire_db_name}.result r1 on rel.source=r1.id
@ -97,7 +97,7 @@ where relClass='Cites' and rel.datainfo.provenanceaction.classid = 'sysimport:cr
and rel.datainfo.deletedbyinference=false and rel.datainfo.invisible = FALSE and rel.datainfo.deletedbyinference=false and rel.datainfo.invisible = FALSE
group by substr(target, 4); group by substr(target, 4);
create table ${stats_db_name}.result_references_oc stored as parquet as CREATE TABLE IF NOT EXISTS ${stats_db_name}.result_references_oc stored as parquet as
select substr(source, 4) as id, count(distinct substr(target, 4)) as references select substr(source, 4) as id, count(distinct substr(target, 4)) as references
from ${openaire_db_name}.relation rel from ${openaire_db_name}.relation rel
join ${openaire_db_name}.result r1 on rel.source=r1.id join ${openaire_db_name}.result r1 on rel.source=r1.id

8
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step6.sql
View File

@ -48,7 +48,9 @@ CREATE TABLE ${stats_db_name}.project_tmp
delayedpubs INT, delayedpubs INT,
callidentifier STRING, callidentifier STRING,
code STRING, code STRING,
totalcost FLOAT totalcost FLOAT,
fundedamount FLOAT,
currency STRING
) CLUSTERED BY (id) INTO 100 buckets stored AS orc tblproperties ('transactional' = 'true'); ) CLUSTERED BY (id) INTO 100 buckets stored AS orc tblproperties ('transactional' = 'true');
INSERT INTO ${stats_db_name}.project_tmp INSERT INTO ${stats_db_name}.project_tmp
@ -72,7 +74,9 @@ SELECT substr(p.id, 4) AS id,
0 AS delayedpubs, 0 AS delayedpubs,
p.callidentifier.value AS callidentifier, p.callidentifier.value AS callidentifier,
p.code.value AS code, p.code.value AS code,
p.totalcost AS totalcost p.totalcost AS totalcost,
p.fundedamount AS fundedamount,
p.currency.value AS currency
FROM ${openaire_db_name}.project p FROM ${openaire_db_name}.project p
WHERE p.datainfo.deletedbyinference = false and p.datainfo.invisible=false; WHERE p.datainfo.deletedbyinference = false and p.datainfo.invisible=false;