forked from D-Net/dnet-hadoop
mergin with branch beta
This commit is contained in:
commit
a38f0f5ea7
|
|
@ -6,7 +6,7 @@
|
|||
<parent>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<artifactId>dhp-build</artifactId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
|
||||
<artifactId>dhp-build-assembly-resources</artifactId>
|
||||
|
|
|
|||
|
|
@ -6,7 +6,7 @@
|
|||
<parent>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<artifactId>dhp-build</artifactId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
|
||||
<artifactId>dhp-build-properties-maven-plugin</artifactId>
|
||||
|
|
|
|||
|
|
@ -5,7 +5,7 @@
|
|||
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<artifactId>dhp-code-style</artifactId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
|
||||
<packaging>jar</packaging>
|
||||
|
||||
|
|
@ -47,12 +47,16 @@
|
|||
<groupId>org.apache.maven.plugins</groupId>
|
||||
<artifactId>maven-site-plugin</artifactId>
|
||||
<version>3.9.1</version>
|
||||
<configuration>
|
||||
<skip>true</skip>
|
||||
</configuration>
|
||||
</plugin>
|
||||
</plugins>
|
||||
</pluginManagement>
|
||||
</build>
|
||||
|
||||
<properties>
|
||||
|
||||
<project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
|
||||
<dhp.site.stage.path>sftp://dnet-hadoop@static-web.d4science.org/dnet-hadoop</dhp.site.stage.path>
|
||||
</properties>
|
||||
|
|
|
|||
|
|
@ -4,7 +4,7 @@
|
|||
<parent>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<artifactId>dhp</artifactId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<artifactId>dhp-build</artifactId>
|
||||
<packaging>pom</packaging>
|
||||
|
|
|
|||
|
|
@ -5,7 +5,7 @@
|
|||
<parent>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<artifactId>dhp</artifactId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
<relativePath>../pom.xml</relativePath>
|
||||
|
||||
</parent>
|
||||
|
|
|
|||
|
|
@ -4,7 +4,7 @@
|
|||
<parent>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<artifactId>dhp-actionmanager</artifactId>
|
||||
|
||||
|
|
|
|||
|
|
@ -4,7 +4,7 @@
|
|||
<parent>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<artifactId>dhp-aggregation</artifactId>
|
||||
<build>
|
||||
|
|
|
|||
|
|
@ -83,7 +83,7 @@
|
|||
--conf spark.sql.warehouse.dir=${sparkSqlWarehouseDir}
|
||||
</spark-opts>
|
||||
<arg>--workingPath</arg><arg>${workingPath}/COCI</arg>
|
||||
<arg>--outputPath</arg><arg>${workingPath}/COCI_JSON</arg>
|
||||
<arg>--outputPath</arg><arg>${workingPath}/COCI_JSON/</arg>
|
||||
<arg>--delimiter</arg><arg>${delimiter}</arg>
|
||||
<arg>--inputFile</arg><arg>${inputFileCoci}</arg>
|
||||
</spark>
|
||||
|
|
|
|||
|
|
@ -72,7 +72,6 @@ class DataciteToOAFTest extends AbstractVocabularyTest {
|
|||
|
||||
spark.read.load(targetPath).printSchema();
|
||||
|
||||
|
||||
val result: Dataset[Oaf] = spark.read.load(targetPath).as[Oaf]
|
||||
|
||||
result
|
||||
|
|
|
|||
|
|
@ -3,7 +3,7 @@
|
|||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
|
||||
|
|
|
|||
|
|
@ -1,11 +1,9 @@
|
|||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<project xmlns="http://maven.apache.org/POM/4.0.0"
|
||||
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
|
||||
xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
|
||||
<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
|
||||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
|
||||
|
|
|
|||
|
|
@ -3,7 +3,7 @@
|
|||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
<artifactId>dhp-dedup-openaire</artifactId>
|
||||
|
|
|
|||
|
|
@ -3,7 +3,7 @@
|
|||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
|
||||
|
|
|
|||
|
|
@ -3,7 +3,7 @@
|
|||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
|
||||
|
|
|
|||
|
|
@ -446,16 +446,12 @@ case object Crossref2Oaf {
|
|||
case "10.13039/501100000781" =>
|
||||
generateSimpleRelationFromAward(funder, "corda_______", extractECAward)
|
||||
generateSimpleRelationFromAward(funder, "corda__h2020", extractECAward)
|
||||
case "10.13039/100000001" =>
|
||||
generateSimpleRelationFromAward(funder, "nsf_________", a => a)
|
||||
case "10.13039/501100001665" =>
|
||||
generateSimpleRelationFromAward(funder, "anr_________", a => a)
|
||||
case "10.13039/501100002341" =>
|
||||
generateSimpleRelationFromAward(funder, "aka_________", a => a)
|
||||
case "10.13039/100000001" => generateSimpleRelationFromAward(funder, "nsf_________", a => a)
|
||||
case "10.13039/501100001665" => generateSimpleRelationFromAward(funder, "anr_________", a => a)
|
||||
case "10.13039/501100002341" => generateSimpleRelationFromAward(funder, "aka_________", a => a)
|
||||
case "10.13039/501100001602" =>
|
||||
generateSimpleRelationFromAward(funder, "aka_________", a => a.replace("SFI", ""))
|
||||
case "10.13039/501100000923" =>
|
||||
generateSimpleRelationFromAward(funder, "arc_________", a => a)
|
||||
generateSimpleRelationFromAward(funder, "sfi_________", a => a.replace("SFI", ""))
|
||||
case "10.13039/501100000923" => generateSimpleRelationFromAward(funder, "arc_________", a => a)
|
||||
case "10.13039/501100000038" =>
|
||||
val targetId = getProjectId("nserc_______", "1e5e62235d094afd01cd56e65112fc63")
|
||||
queue += generateRelation(sourceId, targetId, ModelConstants.IS_PRODUCED_BY)
|
||||
|
|
@ -468,14 +464,10 @@ case object Crossref2Oaf {
|
|||
val targetId = getProjectId("cihr________", "1e5e62235d094afd01cd56e65112fc63")
|
||||
queue += generateRelation(sourceId, targetId, ModelConstants.IS_PRODUCED_BY)
|
||||
queue += generateRelation(targetId, sourceId, ModelConstants.PRODUCES)
|
||||
case "10.13039/501100002848" =>
|
||||
generateSimpleRelationFromAward(funder, "conicytf____", a => a)
|
||||
case "10.13039/501100003448" =>
|
||||
generateSimpleRelationFromAward(funder, "gsrt________", extractECAward)
|
||||
case "10.13039/501100010198" =>
|
||||
generateSimpleRelationFromAward(funder, "sgov________", a => a)
|
||||
case "10.13039/501100004564" =>
|
||||
generateSimpleRelationFromAward(funder, "mestd_______", extractECAward)
|
||||
case "10.13039/501100002848" => generateSimpleRelationFromAward(funder, "conicytf____", a => a)
|
||||
case "10.13039/501100003448" => generateSimpleRelationFromAward(funder, "gsrt________", extractECAward)
|
||||
case "10.13039/501100010198" => generateSimpleRelationFromAward(funder, "sgov________", a => a)
|
||||
case "10.13039/501100004564" => generateSimpleRelationFromAward(funder, "mestd_______", extractECAward)
|
||||
case "10.13039/501100003407" =>
|
||||
generateSimpleRelationFromAward(funder, "miur________", a => a)
|
||||
val targetId = getProjectId("miur________", "1e5e62235d094afd01cd56e65112fc63")
|
||||
|
|
@ -487,15 +479,11 @@ case object Crossref2Oaf {
|
|||
"irb_hr______",
|
||||
a => a.replaceAll("Project No.", "").replaceAll("HRZZ-", "")
|
||||
)
|
||||
case "10.13039/501100006769" =>
|
||||
generateSimpleRelationFromAward(funder, "rsf_________", a => a)
|
||||
case "10.13039/501100001711" =>
|
||||
generateSimpleRelationFromAward(funder, "snsf________", snsfRule)
|
||||
case "10.13039/501100004410" =>
|
||||
generateSimpleRelationFromAward(funder, "tubitakf____", a => a)
|
||||
case "10.10.13039/100004440" =>
|
||||
generateSimpleRelationFromAward(funder, "wt__________", a => a)
|
||||
case "10.13039/501100006769" => generateSimpleRelationFromAward(funder, "rsf_________", a => a)
|
||||
case "10.13039/501100001711" => generateSimpleRelationFromAward(funder, "snsf________", snsfRule)
|
||||
case "10.13039/501100004410" => generateSimpleRelationFromAward(funder, "tubitakf____", a => a)
|
||||
case "10.13039/100004440" =>
|
||||
generateSimpleRelationFromAward(funder, "wt__________", a => a)
|
||||
val targetId = getProjectId("wt__________", "1e5e62235d094afd01cd56e65112fc63")
|
||||
queue += generateRelation(sourceId, targetId, ModelConstants.IS_PRODUCED_BY)
|
||||
queue += generateRelation(targetId, sourceId, ModelConstants.PRODUCES)
|
||||
|
|
@ -516,6 +504,7 @@ case object Crossref2Oaf {
|
|||
case "CONICYT, Programa de Formación de Capital Humano Avanzado" =>
|
||||
generateSimpleRelationFromAward(funder, "conicytf____", extractECAward)
|
||||
case "Wellcome Trust Masters Fellowship" =>
|
||||
generateSimpleRelationFromAward(funder, "wt__________", a => a)
|
||||
val targetId = getProjectId("wt__________", "1e5e62235d094afd01cd56e65112fc63")
|
||||
queue += generateRelation(sourceId, targetId, ModelConstants.IS_PRODUCED_BY)
|
||||
queue += generateRelation(targetId, sourceId, ModelConstants.PRODUCES)
|
||||
|
|
|
|||
|
|
@ -3,7 +3,7 @@
|
|||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
|
||||
|
|
@ -51,7 +51,7 @@
|
|||
<dependency>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<artifactId>dhp-aggregation</artifactId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
<scope>compile</scope>
|
||||
</dependency>
|
||||
|
||||
|
|
|
|||
|
|
@ -274,7 +274,6 @@
|
|||
</spark-opts>
|
||||
<arg>--sourcePath</arg><arg>${sourcePath}/publication</arg>
|
||||
<arg>--preparedInfoPath</arg><arg>${workingDir}/publication</arg>
|
||||
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
|
||||
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
|
||||
<arg>--outputPath</arg><arg>${outputPath}/publication</arg>
|
||||
</spark>
|
||||
|
|
@ -304,7 +303,6 @@
|
|||
</spark-opts>
|
||||
<arg>--sourcePath</arg><arg>${sourcePath}/dataset</arg>
|
||||
<arg>--preparedInfoPath</arg><arg>${workingDir}/dataset</arg>
|
||||
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
|
||||
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
|
||||
<arg>--outputPath</arg><arg>${outputPath}/dataset</arg>
|
||||
</spark>
|
||||
|
|
@ -334,7 +332,6 @@
|
|||
</spark-opts>
|
||||
<arg>--sourcePath</arg><arg>${sourcePath}/otherresearchproduct</arg>
|
||||
<arg>--preparedInfoPath</arg><arg>${workingDir}/otherresearchproduct</arg>
|
||||
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
|
||||
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
|
||||
<arg>--outputPath</arg><arg>${outputPath}/otherresearchproduct</arg>
|
||||
</spark>
|
||||
|
|
@ -364,7 +361,6 @@
|
|||
</spark-opts>
|
||||
<arg>--sourcePath</arg><arg>${sourcePath}/software</arg>
|
||||
<arg>--preparedInfoPath</arg><arg>${workingDir}/software</arg>
|
||||
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
|
||||
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
|
||||
<arg>--outputPath</arg><arg>${outputPath}/software</arg>
|
||||
</spark>
|
||||
|
|
|
|||
|
|
@ -3,7 +3,7 @@
|
|||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
|
||||
|
|
|
|||
|
|
@ -4,8 +4,6 @@ package eu.dnetlib.dhp.oa.graph.hostedbymap.model.doaj;
|
|||
import java.io.Serializable;
|
||||
import java.util.List;
|
||||
|
||||
import com.fasterxml.jackson.annotation.JsonIgnore;
|
||||
|
||||
public class BibJson implements Serializable {
|
||||
private Editorial editorial;
|
||||
private PidScheme pid_scheme;
|
||||
|
|
|
|||
|
|
@ -42,10 +42,7 @@ import java.io.IOException;
|
|||
import java.sql.Array;
|
||||
import java.sql.ResultSet;
|
||||
import java.sql.SQLException;
|
||||
import java.util.ArrayList;
|
||||
import java.util.Arrays;
|
||||
import java.util.Date;
|
||||
import java.util.List;
|
||||
import java.util.*;
|
||||
import java.util.function.Consumer;
|
||||
import java.util.function.Function;
|
||||
import java.util.function.Predicate;
|
||||
|
|
@ -311,7 +308,13 @@ public class MigrateDbEntitiesApplication extends AbstractMigrationApplication i
|
|||
ds.setContentpolicies(prepareListOfQualifiers(rs.getArray("contentpolicies")));
|
||||
ds.setConsenttermsofuse(rs.getBoolean("consenttermsofuse"));
|
||||
ds.setFulltextdownload(rs.getBoolean("fulltextdownload"));
|
||||
ds.setConsenttermsofusedate(rs.getDate("consenttermsofusedate").toString());
|
||||
ds
|
||||
.setConsenttermsofusedate(
|
||||
Optional
|
||||
.ofNullable(
|
||||
rs.getDate("consenttermsofusedate"))
|
||||
.map(c -> c.toString())
|
||||
.orElse(null));
|
||||
|
||||
return Arrays.asList(ds);
|
||||
} catch (final Exception e) {
|
||||
|
|
|
|||
|
|
@ -3,7 +3,7 @@
|
|||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
|
||||
|
|
|
|||
|
|
@ -3,7 +3,7 @@
|
|||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
<artifactId>dhp-stats-promote</artifactId>
|
||||
|
|
|
|||
|
|
@ -3,7 +3,7 @@
|
|||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
<artifactId>dhp-stats-update</artifactId>
|
||||
|
|
|
|||
|
|
@ -12,6 +12,8 @@ and (ri.accessright = 'Open Access'
|
|||
or ri.accessright = 'Embargo' or ri.accessright = 'Open Source')) tmp
|
||||
on p.id= tmp.id;
|
||||
|
||||
compute stats indi_pub_green_oa;
|
||||
|
||||
create table indi_pub_grey_lit stored as parquet as
|
||||
select distinct p.id, coalesce(grey_lit, 0) as grey_lit
|
||||
from publication p
|
||||
|
|
@ -22,6 +24,8 @@ join result_classifications rt on rt.id = p.id
|
|||
where rt.type not in ('Article','Part of book or chapter of book','Book','Doctoral thesis','Master thesis','Data Paper', 'Thesis', 'Bachelor thesis', 'Conference object') and
|
||||
not exists (select 1 from result_classifications rc where type ='Other literature type' and rc.id=p.id)) tmp on p.id=tmp.id;
|
||||
|
||||
compute stats indi_pub_grey_lit;
|
||||
|
||||
create table indi_pub_doi_from_crossref stored as parquet as
|
||||
select distinct p.id, coalesce(doi_from_crossref, 0) as doi_from_crossref
|
||||
from publication p
|
||||
|
|
@ -31,6 +35,7 @@ join datasource d on d.id = ri.collectedfrom
|
|||
where pidtype='Digital Object Identifier' and d.name ='Crossref') tmp
|
||||
on tmp.id=p.id;
|
||||
|
||||
compute stats indi_pub_doi_from_crossref;
|
||||
---- Sprint 2 ----
|
||||
create table indi_result_has_cc_licence stored as parquet as
|
||||
select distinct r.id, (case when lic='' or lic is null then 0 else 1 end) as has_cc_license
|
||||
|
|
@ -40,6 +45,8 @@ join result_licenses as license on license.id = r.id
|
|||
where lower(license.type) LIKE '%creativecommons.org%' OR lower(license.type) LIKE '%cc-%') tmp
|
||||
on r.id= tmp.id;
|
||||
|
||||
compute stats indi_result_has_cc_licence;
|
||||
|
||||
create table indi_result_has_cc_licence_url stored as parquet as
|
||||
select distinct r.id, case when lic_host='' or lic_host is null then 0 else 1 end as has_cc_license_url
|
||||
from result r
|
||||
|
|
@ -49,16 +56,21 @@ join result_licenses as license on license.id = r.id
|
|||
WHERE lower(parse_url(license.type, "HOST")) = "creativecommons.org") tmp
|
||||
on r.id= tmp.id;
|
||||
|
||||
compute stats indi_result_has_cc_licence_url;
|
||||
|
||||
create table indi_pub_has_abstract stored as parquet as
|
||||
select distinct publication.id, coalesce(abstract, 1) has_abstract
|
||||
from publication;
|
||||
|
||||
compute stats indi_pub_has_abstract;
|
||||
|
||||
create table indi_result_with_orcid stored as parquet as
|
||||
select distinct r.id, coalesce(has_orcid, 0) as has_orcid
|
||||
from result r
|
||||
left outer join (select id, 1 as has_orcid from result_orcid) tmp
|
||||
on r.id= tmp.id;
|
||||
|
||||
compute stats indi_result_with_orcid;
|
||||
|
||||
---- Sprint 3 ----
|
||||
create table indi_funded_result_with_fundref stored as parquet as
|
||||
|
|
@ -68,6 +80,8 @@ left outer join (select distinct id, 1 as fundref from project_results
|
|||
where provenance='Harvested') tmp
|
||||
on r.id= tmp.id;
|
||||
|
||||
compute stats indi_funded_result_with_fundref;
|
||||
|
||||
create table indi_result_org_country_collab stored as parquet as
|
||||
with tmp as
|
||||
(select o.id as id, o.country , ro.id as result,r.type from organization o
|
||||
|
|
@ -79,6 +93,8 @@ join tmp as o2 on o1.result=o2.result
|
|||
where o1.id<>o2.id and o1.country<>o2.country
|
||||
group by o1.id, o1.type,o2.country;
|
||||
|
||||
compute stats indi_result_org_country_collab;
|
||||
|
||||
create table indi_result_org_collab stored as parquet as
|
||||
with tmp as
|
||||
(select o.id, ro.id as result,r.type from organization o
|
||||
|
|
@ -90,6 +106,8 @@ join tmp as o2 on o1.result=o2.result
|
|||
where o1.id<>o2.id
|
||||
group by o1.id, o2.id, o1.type;
|
||||
|
||||
compute stats indi_result_org_collab;
|
||||
|
||||
create table indi_funder_country_collab stored as parquet as
|
||||
with tmp as (select funder, project, country from organization_projects op
|
||||
join organization o on o.id=op.id
|
||||
|
|
@ -101,6 +119,8 @@ join tmp as f2 on f1.project=f2.project
|
|||
where f1.country<>f2.country
|
||||
group by f1.funder, f2.country, f1.country;
|
||||
|
||||
compute stats indi_funder_country_collab;
|
||||
|
||||
create table indi_result_country_collab stored as parquet as
|
||||
with tmp as
|
||||
(select country, ro.id as result,r.type from organization o
|
||||
|
|
@ -112,6 +132,8 @@ join tmp as o2 on o1.result=o2.result
|
|||
where o1.country<>o2.country
|
||||
group by o1.country, o2.country, o1.type;
|
||||
|
||||
compute stats indi_result_country_collab;
|
||||
|
||||
---- Sprint 4 ----
|
||||
create table indi_pub_diamond stored as parquet as
|
||||
select distinct pd.id, coalesce(in_diamond_journal, 0) as in_diamond_journal
|
||||
|
|
@ -123,6 +145,8 @@ join stats_ext.plan_s_jn ps where (ps.issn_print=d.issn_printed and ps.issn_onli
|
|||
and (ps.journal_is_in_doaj=true or ps.journal_is_oa=true) and ps.has_apc=false) tmp
|
||||
on pd.id=tmp.id;
|
||||
|
||||
compute stats indi_pub_diamond;
|
||||
|
||||
create table indi_pub_hybrid stored as parquet as
|
||||
select distinct pd.id, coalesce(is_hybrid, 0) as is_hybrid
|
||||
from publication_datasources pd
|
||||
|
|
@ -133,6 +157,8 @@ join stats_ext.plan_s_jn ps where (ps.issn_print=d.issn_printed and ps.issn_onli
|
|||
and (ps.journal_is_in_doaj=false and ps.journal_is_oa=false)) tmp
|
||||
on pd.id=tmp.id;
|
||||
|
||||
compute stats indi_pub_hybrid;
|
||||
|
||||
create table indi_pub_in_transformative stored as parquet as
|
||||
select distinct pd.id, coalesce(is_transformative, 0) as is_transformative
|
||||
from publication pd
|
||||
|
|
@ -143,6 +169,8 @@ join stats_ext.plan_s_jn ps where (ps.issn_print=d.issn_printed and ps.issn_onli
|
|||
and ps.is_transformative_journal=true) tmp
|
||||
on pd.id=tmp.id;
|
||||
|
||||
compute stats indi_pub_in_transformative;
|
||||
|
||||
create table indi_pub_closed_other_open stored as parquet as
|
||||
select distinct ri.id, coalesce(pub_closed_other_open, 0) as pub_closed_other_open from result_instance ri
|
||||
left outer join
|
||||
|
|
@ -153,11 +181,12 @@ where d.type like '%Journal%' and ri.accessright='Closed Access' and
|
|||
(p.bestlicence='Open Access' or p.bestlicence='Open Source')) tmp
|
||||
on tmp.id=ri.id;
|
||||
|
||||
|
||||
compute stats indi_pub_closed_other_open;
|
||||
---- Sprint 5 ----
|
||||
create table indi_result_no_of_copies stored as parquet as
|
||||
select id, count(id) as number_of_copies from result_instance group by id;
|
||||
|
||||
compute stats indi_result_no_of_copies;
|
||||
---- Sprint 6 ----
|
||||
create table indi_pub_gold_oa stored as parquet as
|
||||
WITH gold_oa AS (
|
||||
|
|
@ -183,6 +212,8 @@ LEFT OUTER JOIN (
|
|||
JOIN issn on issn.id=pd.datasource
|
||||
JOIN gold_oa on issn.issn = gold_oa.issn) tmp ON pd.id=tmp.id;
|
||||
|
||||
compute stats indi_pub_gold_oa;
|
||||
|
||||
create table indi_datasets_gold_oa stored as parquet as
|
||||
WITH gold_oa AS (
|
||||
SELECT issn_l, journal_is_in_doaj, journal_is_oa, issn_1 as issn
|
||||
|
|
@ -210,6 +241,8 @@ LEFT OUTER JOIN (
|
|||
JOIN issn on issn.id=pd.datasource
|
||||
JOIN gold_oa on issn.issn = gold_oa.issn) tmp ON pd.id=tmp.id;
|
||||
|
||||
compute stats indi_datasets_gold_oa;
|
||||
|
||||
create table indi_software_gold_oa stored as parquet as
|
||||
WITH gold_oa AS (
|
||||
SELECT issn_l, journal_is_in_doaj, journal_is_oa, issn_1 as issn
|
||||
|
|
@ -237,6 +270,8 @@ LEFT OUTER JOIN (
|
|||
JOIN issn on issn.id=pd.datasource
|
||||
JOIN gold_oa on issn.issn = gold_oa.issn) tmp ON pd.id=tmp.id;
|
||||
|
||||
compute stats indi_software_gold_oa;
|
||||
|
||||
create table indi_org_findable stored as parquet as
|
||||
with result_with_pid as (
|
||||
select ro.organization organization, count(distinct rp.id) no_result_with_pid from result_organization ro
|
||||
|
|
@ -263,6 +298,8 @@ join result_with_pid_share on result_with_pid_share.organization=allresults.orga
|
|||
left outer join (
|
||||
select organization, abstract_share from result_with_abstract_share) tmp on tmp.organization=allresults.organization;
|
||||
|
||||
compute stats indi_org_findable;
|
||||
|
||||
create table indi_org_openess stored as parquet as
|
||||
WITH datasets_oa as (
|
||||
SELECT ro.organization, count(dg.id) no_oadatasets FROM indi_datasets_gold_oa dg
|
||||
|
|
@ -313,6 +350,8 @@ left outer join (
|
|||
left outer join (
|
||||
select organization,s from allsoftwaresshare) tmp1 on tmp1.organization=allpubsshare.organization;
|
||||
|
||||
compute stats indi_org_openess;
|
||||
|
||||
create table indi_pub_hybrid_oa_with_cc stored as parquet as
|
||||
WITH hybrid_oa AS (
|
||||
SELECT issn_l, journal_is_in_doaj, journal_is_oa, issn_print as issn
|
||||
|
|
@ -343,6 +382,8 @@ LEFT OUTER JOIN (
|
|||
JOIN indi_result_has_cc_licence cc on pd.id=cc.id
|
||||
where cc.has_cc_license=1) tmp on pd.id=tmp.id;
|
||||
|
||||
compute stats indi_pub_hybrid_oa_with_cc;
|
||||
|
||||
create table indi_pub_downloads stored as parquet as
|
||||
SELECT result_id, sum(downloads) no_dowloads from openaire_prod_usage_stats.usage_stats
|
||||
join publication on result_id=id
|
||||
|
|
@ -350,6 +391,8 @@ where downloads>0
|
|||
GROUP BY result_id
|
||||
order by no_dowloads desc;
|
||||
|
||||
compute stats indi_pub_downloads;
|
||||
|
||||
create table indi_pub_downloads_datasource stored as parquet as
|
||||
SELECT result_id, repository_id, sum(downloads) no_dowloads from openaire_prod_usage_stats.usage_stats
|
||||
join publication on result_id=id
|
||||
|
|
@ -357,15 +400,21 @@ where downloads>0
|
|||
GROUP BY result_id, repository_id
|
||||
order by result_id;
|
||||
|
||||
compute stats indi_pub_downloads_datasource;
|
||||
|
||||
create table indi_pub_downloads_year stored as parquet as
|
||||
SELECT result_id, substring(us.`date`, 1,4) as `year`, sum(downloads) no_dowloads from openaire_prod_usage_stats.usage_stats us
|
||||
join publication on result_id=id where downloads>0
|
||||
GROUP BY result_id, `year`
|
||||
order by `year` asc;
|
||||
|
||||
compute stats indi_pub_downloads_year;
|
||||
|
||||
create table indi_pub_downloads_datasource_year stored as parquet as
|
||||
SELECT result_id, substring(us.`date`, 1,4) as `year`, repository_id, sum(downloads) no_dowloads from openaire_prod_usage_stats.usage_stats us
|
||||
join publication on result_id=id
|
||||
where downloads>0
|
||||
GROUP BY result_id, repository_id, `year`
|
||||
order by `year` asc, result_id;
|
||||
|
||||
compute stats indi_pub_downloads_datasource_year;
|
||||
|
|
@ -3,7 +3,7 @@
|
|||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
<artifactId>dhp-usage-raw-data-update</artifactId>
|
||||
|
|
|
|||
|
|
@ -3,7 +3,7 @@
|
|||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
<artifactId>dhp-usage-stats-build</artifactId>
|
||||
|
|
|
|||
|
|
@ -3,7 +3,7 @@
|
|||
<parent>
|
||||
<artifactId>dhp-workflows</artifactId>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
</parent>
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
|
||||
|
|
|
|||
|
|
@ -6,7 +6,7 @@
|
|||
<parent>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<artifactId>dhp</artifactId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
<relativePath>../pom.xml</relativePath>
|
||||
</parent>
|
||||
|
||||
|
|
|
|||
6
pom.xml
6
pom.xml
|
|
@ -3,7 +3,7 @@
|
|||
<modelVersion>4.0.0</modelVersion>
|
||||
<groupId>eu.dnetlib.dhp</groupId>
|
||||
<artifactId>dhp</artifactId>
|
||||
<version>1.2.4-SNAPSHOT</version>
|
||||
<version>1.2.5-SNAPSHOT</version>
|
||||
<packaging>pom</packaging>
|
||||
|
||||
<licenses>
|
||||
|
|
@ -551,6 +551,9 @@
|
|||
<groupId>org.apache.maven.plugins</groupId>
|
||||
<artifactId>maven-site-plugin</artifactId>
|
||||
<version>3.9.1</version>
|
||||
<configuration>
|
||||
<skip>${dhp.site.skip}</skip>
|
||||
</configuration>
|
||||
</plugin>
|
||||
|
||||
<plugin>
|
||||
|
|
@ -791,6 +794,7 @@
|
|||
<dhp.spark.version>2.4.0.cloudera2</dhp.spark.version>
|
||||
<dhp.jackson.version>2.9.6</dhp.jackson.version>
|
||||
<dhp.commons.lang.version>3.5</dhp.commons.lang.version>
|
||||
<dhp.site.skip>true</dhp.site.skip>
|
||||
<dhp.guava.version>11.0.2</dhp.guava.version>
|
||||
<scala.version>2.11.12</scala.version>
|
||||
<junit-jupiter.version>5.6.1</junit-jupiter.version>
|
||||
|
|
|
|||
Loading…
Reference in New Issue