[doiboost] merged workflow from branch beta

This commit is contained in:
Claudio Atzori 2023-06-28 09:15:37 +02:00
parent 5f32edd9bf
commit 288ec0b7d6
6 changed files with 1578 additions and 1542 deletions

View File

@ -554,7 +554,7 @@ public class PublicationToOaf implements Serializable {
private KeyValue createCollectedFrom() {
KeyValue cf = new KeyValue();
cf.setValue(ModelConstants.ORCID.toUpperCase());
cf.setKey("10|" + OPENAIRE_PREFIX + SEPARATOR + "806360c771262b4d6770e7cdf04b5c5a");
cf.setKey("10|" + OPENAIRE_PREFIX + SEPARATOR + "cd0f74b5955dc87fd0605745c4b49ee8");
return cf;
}

View File

@ -391,6 +391,28 @@ object DoiBoostMappingUtil {
di
}
def createSubject(value: String, classId: String, schemeId: String): Subject = {
val s = new Subject
s.setQualifier(OafMapperUtils.qualifier(classId, classId, schemeId, schemeId))
s.setValue(value)
s
}
def createSubject(
value: String,
classId: String,
className: String,
schemeId: String,
schemeName: String
): Subject = {
val s = new Subject
s.setQualifier(OafMapperUtils.qualifier(classId, className, schemeId, schemeName))
s.setValue(value)
s
}
def createSP(
value: String,
classId: String,

View File

@ -201,7 +201,7 @@ case object Crossref2Oaf {
if (subjectList.nonEmpty) {
result.setSubject(
subjectList.map(s => createSP(s, "keywords", ModelConstants.DNET_SUBJECT_TYPOLOGIES)).asJava
subjectList.map(s => createSubject(s, "keyword", ModelConstants.DNET_SUBJECT_TYPOLOGIES)).asJava
)
}
@ -370,10 +370,40 @@ case object Crossref2Oaf {
case dataset: Dataset => convertDataset(dataset)
}
val doisReference: List[String] = for {
JObject(reference_json) <- json \ "reference"
JField("DOI", JString(doi_json)) <- reference_json
} yield doi_json
if (doisReference != null && doisReference.nonEmpty) {
val citation_relations: List[Relation] = generateCitationRelations(doisReference, result)
resultList = resultList ::: citation_relations
}
resultList = resultList ::: List(result)
resultList
}
private def createCiteRelation(source: Result, targetPid: String, targetPidType: String): List[Relation] = {
val targetId = IdentifierFactory.idFromPid("50", targetPidType, targetPid, true)
val from = new Relation
from.setSource(source.getId)
from.setTarget(targetId)
from.setRelType(ModelConstants.RESULT_RESULT)
from.setRelClass(ModelConstants.CITES)
from.setSubRelType(ModelConstants.CITATION)
from.setCollectedfrom(source.getCollectedfrom)
from.setDataInfo(source.getDataInfo)
from.setLastupdatetimestamp(source.getLastupdatetimestamp)
List(from)
}
def generateCitationRelations(dois: List[String], result: Result): List[Relation] = {
dois.flatMap(d => createCiteRelation(result, d, "doi"))
}
def mappingFunderToRelations(
funders: List[mappingFunder],
sourceId: String,

View File

@ -2,7 +2,7 @@ package eu.dnetlib.doiboost.mag
import eu.dnetlib.dhp.schema.common.ModelConstants
import eu.dnetlib.dhp.schema.oaf.utils.IdentifierFactory
import eu.dnetlib.dhp.schema.oaf.{Instance, Journal, Publication, StructuredProperty}
import eu.dnetlib.dhp.schema.oaf.{Instance, Journal, Publication, StructuredProperty, Subject}
import eu.dnetlib.doiboost.DoiBoostMappingUtil
import eu.dnetlib.doiboost.DoiBoostMappingUtil._
import org.json4s
@ -210,8 +210,8 @@ case object ConversionUtil {
val className = "Microsoft Academic Graph classification"
val classid = "MAG"
val p: List[StructuredProperty] = fieldOfStudy.subjects.flatMap(s => {
val s1 = createSP(
val p: List[Subject] = fieldOfStudy.subjects.flatMap(s => {
val s1 = createSubject(
s.DisplayName,
classid,
className,
@ -219,10 +219,10 @@ case object ConversionUtil {
ModelConstants.DNET_SUBJECT_TYPOLOGIES
)
val di = DoiBoostMappingUtil.generateDataInfo(s.Score.toString)
var resList: List[StructuredProperty] = List(s1)
var resList: List[Subject] = List(s1)
if (s.MainType.isDefined) {
val maintp = s.MainType.get
val s2 = createSP(
val s2 = createSubject(
s.MainType.get,
classid,
className,
@ -232,7 +232,7 @@ case object ConversionUtil {
s2.setDataInfo(di)
resList = resList ::: List(s2)
if (maintp.contains(".")) {
val s3 = createSP(
val s3 = createSubject(
maintp.split("\\.").head,
classid,
className,

View File

@ -1,9 +1,14 @@
package eu.dnetlib.dhp.doiboost.crossref
import eu.dnetlib.dhp.schema.common.ModelConstants
import eu.dnetlib.dhp.schema.oaf._
import eu.dnetlib.dhp.utils.DHPUtils
import eu.dnetlib.doiboost.crossref.Crossref2Oaf
import org.codehaus.jackson.map.{ObjectMapper, SerializationConfig}
import org.json4s
import org.json4s.JsonAST.{JField, JObject, JString}
import org.json4s.{DefaultFormats, JValue}
import org.json4s.jackson.JsonMethods
import org.junit.jupiter.api.Assertions._
import org.junit.jupiter.api.Test
import org.slf4j.{Logger, LoggerFactory}
@ -109,6 +114,47 @@ class CrossrefMappingTest {
}
private def parseJson(input: String): JValue = {
implicit lazy val formats: DefaultFormats.type = org.json4s.DefaultFormats
lazy val json: json4s.JValue = JsonMethods.parse(input)
json
}
@Test
def testCitationRelations(): Unit = {
val json = Source
.fromInputStream(getClass.getResourceAsStream("/eu/dnetlib/doiboost/crossref/publication_license_embargo.json"))
.mkString
assertNotNull(json)
assertFalse(json.isEmpty)
val result: List[Oaf] = Crossref2Oaf.convert(json)
assertTrue(result.nonEmpty)
val j = parseJson(json)
val doisReference: List[String] = for {
JObject(reference_json) <- j \ "reference"
JField("DOI", JString(doi_json)) <- reference_json
} yield doi_json
val relationList: List[Relation] = result
.filter(s => s.isInstanceOf[Relation])
.map(r => r.asInstanceOf[Relation])
.filter(r => r.getSubRelType.equalsIgnoreCase(ModelConstants.CITATION))
assertNotNull(relationList)
assertFalse(relationList.isEmpty)
assertEquals(doisReference.size, relationList.size)
mapper.getSerializationConfig.enable(SerializationConfig.Feature.INDENT_OUTPUT)
relationList.foreach(p => println(mapper.writeValueAsString(p)))
}
@Test
def testEmptyTitle(): Unit = {
val json = Source