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

more pervasive use of constants from ModelConstants, especially for ORCID

Browse Source
master
Claudio Atzori 3 years ago
parent ea9b00ce56
commit 6cbda49112
12 changed files with 33 additions and 35 deletions
Show Stats Download Patch File Download Diff File

3
dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/ConversionUtils.java
Unescape Escape View File

@ -26,6 +26,7 @@ import eu.dnetlib.broker.objects.OaBrokerRelatedDatasource;
import eu.dnetlib.broker.objects.OaBrokerRelatedPublication;
import eu.dnetlib.broker.objects.OaBrokerRelatedSoftware;
import eu.dnetlib.broker.objects.OaBrokerTypedValue;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.Author;
import eu.dnetlib.dhp.schema.oaf.Dataset;
import eu.dnetlib.dhp.schema.oaf.Datasource;
@ -144,7 +145,7 @@ public class ConversionUtils {
.filter(pid -> pid != null)
.filter(pid -> pid.getQualifier() != null)
.filter(pid -> pid.getQualifier().getClassid() != null)
.filter(pid -> pid.getQualifier().getClassid().equalsIgnoreCase("orcid"))
.filter(pid -> pid.getQualifier().getClassid().equalsIgnoreCase(ModelConstants.ORCID))
.map(pid -> pid.getValue())
.map(pid -> cleanOrcid(pid))
.filter(StringUtils::isNotBlank)

8
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/DoiBoostMappingUtil.scala
Unescape Escape View File

@ -5,6 +5,7 @@ import eu.dnetlib.dhp.schema.oaf.{DataInfo, Dataset, Field, Instance, KeyValue,
import eu.dnetlib.dhp.utils.DHPUtils
import org.apache.commons.lang3.StringUtils
import com.fasterxml.jackson.databind.ObjectMapper
import eu.dnetlib.dhp.schema.common.ModelConstants
import org.json4s
import org.json4s.DefaultFormats
import org.json4s.jackson.JsonMethods.parse
@ -28,7 +29,6 @@ object DoiBoostMappingUtil {
//STATIC STRING
val MAG = "microsoft"
val MAG_NAME = "Microsoft Academic Graph"
val ORCID = "orcid"
val ORCID_PENDING = "orcid_pending"
val CROSSREF = "Crossref"
val UNPAYWALL = "UnpayWall"
@ -37,8 +37,6 @@ object DoiBoostMappingUtil {
val doiBoostNSPREFIX = "doiboost____"
val OPENAIRE_PREFIX = "openaire____"
val SEPARATOR = "::"
val DNET_LANGUAGES = "dnet:languages"
val PID_TYPES = "dnet:pid_types"
val invalidName = List(",", "none none", "none, none", "none &na;", "(:null)", "test test test", "test test", "test", "&na; &na;")
@ -326,8 +324,8 @@ object DoiBoostMappingUtil {
def createORIDCollectedFrom(): KeyValue = {
val cf = new KeyValue
cf.setValue(ORCID)
cf.setKey("10|" + OPENAIRE_PREFIX + SEPARATOR + DHPUtils.md5(ORCID.toLowerCase))
cf.setValue(ModelConstants.ORCID_DS)
cf.setKey("10|" + OPENAIRE_PREFIX + SEPARATOR + DHPUtils.md5(ModelConstants.ORCID))
cf
}

2
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/crossref/Crossref2Oaf.scala
Unescape Escape View File

@ -87,7 +87,7 @@ case object Crossref2Oaf {
//MAPPING Crossref DOI into PID
val doi: String = (json \ "DOI").extract[String]
result.setPid(List(createSP(doi, "doi", PID_TYPES)).asJava)
result.setPid(List(createSP(doi, "doi", ModelConstants.DNET_PID_TYPES)).asJava)
//MAPPING Crossref DOI into OriginalId
//and Other Original Identifier of dataset like clinical-trial-number

4
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/mag/MagDataModel.scala
Unescape Escape View File

@ -188,7 +188,7 @@ case object ConversionUtil {
val authors = inputParams._2
val pub = new Publication
pub.setPid(List(createSP(paper.Doi.toLowerCase, "doi", PID_TYPES)).asJava)
pub.setPid(List(createSP(paper.Doi.toLowerCase, "doi", ModelConstants.DNET_PID_TYPES)).asJava)
pub.setOriginalId(List(paper.PaperId.toString, paper.Doi.toLowerCase).asJava)
//Set identifier as 50|doiboost____::md5(DOI)
@ -247,7 +247,7 @@ case object ConversionUtil {
val description = inputParams._2
val pub = new Publication
pub.setPid(List(createSP(paper.Doi.toLowerCase, "doi", PID_TYPES)).asJava)
pub.setPid(List(createSP(paper.Doi.toLowerCase, "doi", ModelConstants.DNET_PID_TYPES)).asJava)
pub.setOriginalId(List(paper.PaperId.toString, paper.Doi.toLowerCase).asJava)
//Set identifier as 50 | doiboost____::md5(DOI)

30
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/orcidnodoi/oaf/PublicationToOaf.java
Unescape Escape View File

@ -30,7 +30,6 @@ public class PublicationToOaf implements Serializable {
static Logger logger = LoggerFactory.getLogger(PublicationToOaf.class);
public static final String ORCID = "ORCID";
public final static String orcidPREFIX = "orcid_______";
public static final String OPENAIRE_PREFIX = "openaire____";
public static final String SEPARATOR = "::";
@ -69,7 +68,9 @@ public class PublicationToOaf implements Serializable {
private static Map<String, Pair<String, String>> datasources = new HashMap<String, Pair<String, String>>() {
{
put(ORCID.toLowerCase(), new Pair<>(ORCID, OPENAIRE_PREFIX + SEPARATOR + "orcid"));
put(
ModelConstants.ORCID,
new Pair<>(ModelConstants.ORCID_DS, OPENAIRE_PREFIX + SEPARATOR + ModelConstants.ORCID));
}
};
@ -102,8 +103,6 @@ public class PublicationToOaf implements Serializable {
}
}
public static final String PID_TYPES = "dnet:pid_types";
public Oaf generatePublicationActionsFromJson(final String json) {
try {
if (parsedPublications != null) {
@ -138,8 +137,8 @@ public class PublicationToOaf implements Serializable {
mapQualifier(
"sysimport:actionset:orcidworks-no-doi",
"sysimport:actionset:orcidworks-no-doi",
"dnet:provenanceActions",
"dnet:provenanceActions"));
ModelConstants.DNET_PROVENANCE_ACTIONS,
ModelConstants.DNET_PROVENANCE_ACTIONS));
publication.setDataInfo(dataInfo);
publication.setLastupdatetimestamp(new Date().getTime());
@ -159,7 +158,9 @@ public class PublicationToOaf implements Serializable {
publication
.getExternalReference()
.add(
convertExtRef(extId, classid, classname, "dnet:pid_types", "dnet:pid_types"));
convertExtRef(
extId, classid, classname, ModelConstants.DNET_PID_TYPES,
ModelConstants.DNET_PID_TYPES));
}
});
@ -505,24 +506,21 @@ public class PublicationToOaf implements Serializable {
private KeyValue createCollectedFrom() {
KeyValue cf = new KeyValue();
cf.setValue(ORCID);
cf.setValue(ModelConstants.ORCID_DS);
cf.setKey("10|" + OPENAIRE_PREFIX + SEPARATOR + "806360c771262b4d6770e7cdf04b5c5a");
return cf;
}
private KeyValue createHostedBy() {
KeyValue hb = new KeyValue();
hb.setValue("Unknown Repository");
hb.setKey("10|" + OPENAIRE_PREFIX + SEPARATOR + "55045bd2a65019fd8e6741a755395c8c");
return hb;
return ModelConstants.UNKNOWN_REPOSITORY;
}
private StructuredProperty mapAuthorId(String orcidId) {
final StructuredProperty sp = new StructuredProperty();
sp.setValue(orcidId);
final Qualifier q = new Qualifier();
q.setClassid(ORCID.toLowerCase());
q.setClassname(ORCID.toLowerCase());
q.setClassid(ModelConstants.ORCID);
q.setClassname(ModelConstants.ORCID_CLASSNAME);
q.setSchemeid(ModelConstants.DNET_PID_TYPES);
q.setSchemename(ModelConstants.DNET_PID_TYPES);
sp.setQualifier(q);
@ -535,8 +533,8 @@ public class PublicationToOaf implements Serializable {
mapQualifier(
"sysimport:crosswalk:entityregistry",
"Harvested",
"dnet:provenanceActions",
"dnet:provenanceActions"));
ModelConstants.DNET_PROVENANCE_ACTIONS,
ModelConstants.DNET_PROVENANCE_ACTIONS));
sp.setDataInfo(dataInfo);
return sp;
}

3
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/uw/UnpayWallToOAF.scala
Unescape Escape View File

@ -1,5 +1,6 @@
package eu.dnetlib.doiboost.uw
import eu.dnetlib.dhp.schema.common.ModelConstants
import eu.dnetlib.dhp.schema.oaf.{Instance, Publication}
import org.json4s
import org.json4s.DefaultFormats
@ -32,7 +33,7 @@ object UnpayWallToOAF {
val is_oa = (json\ "is_oa").extract[Boolean]
val oaLocation:OALocation = (json \ "best_oa_location").extractOrElse[OALocation](null)
pub.setPid(List(createSP(doi, "doi", PID_TYPES)).asJava)
pub.setPid(List(createSP(doi, "doi", ModelConstants.DNET_PID_TYPES)).asJava)
pub.setId(generateIdentifier(pub, doi.toLowerCase))
pub.setCollectedfrom(List(createUnpayWallCollectedFrom()).asJava)

6
dhp-workflows/dhp-doiboost/src/test/resources/eu/dnetlib/dhp/doiboost/publication_merge
View File

File diff suppressed because one or more lines are too long

2
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/dump/Constants.java
Unescape Escape View File

@ -24,8 +24,6 @@ public class Constants {
public static String RESEARCH_INFRASTRUCTURE = "Research Infrastructure/Initiative";
public static String ORCID = "orcid";
static {
accessRightsCoarMap.put("OPEN", "c_abf2");
accessRightsCoarMap.put("RESTRICTED", "c_16ec");

2
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/dump/ResultMapper.java
Unescape Escape View File

@ -503,7 +503,7 @@ public class ResultMapper implements Serializable {
private static Pid getOrcid(List<StructuredProperty> p) {
for (StructuredProperty pid : p) {
if (pid.getQualifier().getClassid().equals(Constants.ORCID)) {
if (pid.getQualifier().getClassid().equals(ModelConstants.ORCID)) {
Optional<DataInfo> di = Optional.ofNullable(pid.getDataInfo());
if (di.isPresent()) {
return Pid

2
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/AbstractMdRecordToOafMapper.java
Unescape Escape View File

@ -68,7 +68,7 @@ public abstract class AbstractMdRecordToOafMapper {
protected static final String DATACITE_SCHEMA_KERNEL_3 = "http://datacite.org/schema/kernel-3";
protected static final String DATACITE_SCHEMA_KERNEL_3_SLASH = "http://datacite.org/schema/kernel-3/";
protected static final Qualifier ORCID_PID_TYPE = qualifier(
"ORCID", "Open Researcher and Contributor ID", DNET_PID_TYPES, DNET_PID_TYPES);
ModelConstants.ORCID_PENDING, ModelConstants.ORCID_CLASSNAME, DNET_PID_TYPES, DNET_PID_TYPES);
protected static final Qualifier MAG_PID_TYPE = qualifier(
"MAGIdentifier", "Microsoft Academic Graph Identifier", DNET_PID_TYPES, DNET_PID_TYPES);

3
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OafToOafMapper.java
Unescape Escape View File

@ -19,6 +19,7 @@ import com.google.common.collect.Lists;
import eu.dnetlib.dhp.common.PacePerson;
import eu.dnetlib.dhp.oa.graph.raw.common.VocabularyGroup;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.Author;
import eu.dnetlib.dhp.schema.oaf.DataInfo;
import eu.dnetlib.dhp.schema.oaf.Field;
@ -61,7 +62,7 @@ public class OafToOafMapper extends AbstractMdRecordToOafMapper {
author.setPid(new ArrayList<>());
if (StringUtils.isNotBlank(pid)) {
if (type.startsWith("ORCID")) {
if (type.toLowerCase().startsWith(ORCID)) {
final String cleanedId = pid
.replaceAll("http://orcid.org/", "")
.replaceAll("https://orcid.org/", "");

3
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java
Unescape Escape View File

@ -20,6 +20,7 @@ import com.google.common.collect.Lists;
import eu.dnetlib.dhp.common.PacePerson;
import eu.dnetlib.dhp.oa.graph.raw.common.VocabularyGroup;
import eu.dnetlib.dhp.schema.common.ModelConstants;
import eu.dnetlib.dhp.schema.oaf.Author;
import eu.dnetlib.dhp.schema.oaf.DataInfo;
import eu.dnetlib.dhp.schema.oaf.Field;
@ -98,7 +99,7 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
.replaceAll(" ", "")
.replaceAll("_", "");
if (type.startsWith("ORCID")) {
if (type.toLowerCase().startsWith(ModelConstants.ORCID)) {
final String cleanedId = id.replaceAll("http://orcid.org/", "").replaceAll("https://orcid.org/", "");
res.add(structuredProperty(cleanedId, ORCID_PID_TYPE, info));
} else if (type.startsWith("MAGID")) {

Write Preview
Loading…
Cancel
Save