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

added result_orcid, result_project provenance, issn in datasources

This commit is contained in:
Antonis Lempesis 2021-07-28 12:28:04 +03:00
parent 1a28a69cac
commit 4a9741825d
5 changed files with 22 additions and 39 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step13.sql
View File

@ -57,12 +57,14 @@ UNION ALL
SELECT * FROM ${stats_db_name}.software_sources
UNION ALL
SELECT * FROM ${stats_db_name}.otherresearchproduct_sources;
--
-- ANALYZE TABLE ${stats_db_name}.publication_sources COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.publication_sources COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.dataset_sources COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.dataset_sources COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.software_sources COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.software_sources COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_sources COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_sources COMPUTE STATISTICS FOR COLUMNS;
create table ${stats_db_name}.result_orcid as
select distinct res.id, regexp_replace(res.orcid, 'http://orcid.org/' ,'') as orcid
from (
SELECT substr(res.id, 4) as id, auth_pid.value as orcid
FROM ${openaire_db_name}.result res
LATERAL VIEW explode(author) a as auth
LATERAL VIEW explode(auth.pid) ap as auth_pid
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

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

@ -33,13 +33,4 @@ select * from ${stats_db_name}.dataset_refereed
union all
select * from ${stats_db_name}.software_refereed
union all
select * from ${stats_db_name}.otherresearchproduct_refereed;
--
-- ANALYZE TABLE ${stats_db_name}.publication_refereed COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.publication_refereed COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.dataset_refereed COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.dataset_refereed COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.software_refereed COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.software_refereed COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_refereed COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_refereed COMPUTE STATISTICS FOR COLUMNS;
select * from ${stats_db_name}.otherresearchproduct_refereed;

2
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step6.sql
View File

@ -13,7 +13,7 @@ WHERE r.reltype = 'projectOrganization'
and r.datainfo.deletedbyinference = false;
CREATE TABLE ${stats_db_name}.project_results AS
SELECT substr(r.target, 4) AS id, substr(r.source, 4) AS result
SELECT substr(r.target, 4) AS id, substr(r.source, 4) AS result, r.datainfo.provenanceaction.classname as provenance
FROM ${openaire_db_name}.relation r
WHERE r.reltype = 'resultProject'
and r.datainfo.deletedbyinference = false;

9
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step7.sql
View File

@ -130,12 +130,7 @@ WHERE r.reltype = 'resultOrganization'
and r.datainfo.deletedbyinference = false;
CREATE TABLE ${stats_db_name}.result_projects AS
select pr.result AS id, pr.id AS project, datediff(p.enddate, p.startdate) AS daysfromend
select pr.result AS id, pr.id AS project, datediff(p.enddate, p.startdate) AS daysfromend, pr.provenance as provenance
FROM ${stats_db_name}.result r
JOIN ${stats_db_name}.project_results pr ON r.id = pr.result
JOIN ${stats_db_name}.project_tmp p ON p.id = pr.id;
-- ANALYZE TABLE ${stats_db_name}.result_organization COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.result_organization COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.result_projects COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.result_projects COMPUTE STATISTICS FOR COLUMNS;
JOIN ${stats_db_name}.project_tmp p ON p.id = pr.id;

19
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step8.sql
View File

@ -17,7 +17,9 @@ CREATE TABLE ${stats_db_name}.datasource_tmp
`latitude` STRING,
`longitude` STRING,
`websiteurl` STRING,
`compatibility` STRING
`compatibility` STRING,
issn_printed STRING,
issn_online STRING
) CLUSTERED BY (id) INTO 100 buckets stored AS orc tblproperties ('transactional' = 'true');
-- Insert statement that takes into account the piwik_id of the openAIRE graph
@ -32,7 +34,9 @@ SELECT substr(d1.id, 4) AS id,
d1.latitude.value AS latitude,
d1.longitude.value AS longitude,
d1.websiteurl.value AS websiteurl,
d1.openairecompatibility.classid AS compatibility
d1.openairecompatibility.classid AS compatibility,
d1.journal.issnprinted AS issn_printed,
d1.journal.issnonline AS issn_online
FROM ${openaire_db_name}.datasource d1
LEFT OUTER JOIN
(SELECT id, split(originalidd, '\\:')[1] as piwik_id
@ -97,13 +101,4 @@ where d.datainfo.deletedbyinference = false;
CREATE OR REPLACE VIEW ${stats_db_name}.datasource_results AS
SELECT datasource AS id, id AS result
FROM ${stats_db_name}.result_datasources;
-- ANALYZE TABLE ${stats_db_name}.datasource_tmp COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.datasource_tmp COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.datasource_languages COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.datasource_languages COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.datasource_oids COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.datasource_oids COMPUTE STATISTICS FOR COLUMNS;
-- ANALYZE TABLE ${stats_db_name}.datasource_organizations COMPUTE STATISTICS;
-- ANALYZE TABLE ${stats_db_name}.datasource_organizations COMPUTE STATISTICS FOR COLUMNS;
FROM ${stats_db_name}.result_datasources;