added result_orcid, result_project provenance, issn in datasources

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

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

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

@@ -34,12 +34,3 @@ 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;

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

@@ -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;

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

@@ -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;

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

@@ -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
@@ -98,12 +102,3 @@ 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;