lsmyrnaios
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

using ModelConstants

This commit is contained in:
Claudio Atzori 2021-05-14 11:29:49 +02:00
parent 0358ae16ce
commit ca3f3a7687
2 changed files with 34 additions and 34 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/SparkGenerateDoiBoost.scala
View File

@ -160,17 +160,17 @@ object SparkGenerateDoiBoost {
val r:Relation = new Relation val r:Relation = new Relation
r.setSource(pub.getId) r.setSource(pub.getId)
r.setTarget(affId) r.setTarget(affId)
r.setRelType("resultOrganization") r.setRelType(ModelConstants.RESULT_ORGANIZATION)
r.setRelClass("hasAuthorInstitution") r.setRelClass(ModelConstants.HAS_AUTHOR_INSTITUTION)
r.setSubRelType("affiliation") r.setSubRelType(ModelConstants.AFFILIATION)
r.setDataInfo(pub.getDataInfo) r.setDataInfo(pub.getDataInfo)
r.setCollectedfrom(List(DoiBoostMappingUtil.createMAGCollectedFrom()).asJava) r.setCollectedfrom(List(DoiBoostMappingUtil.createMAGCollectedFrom()).asJava)
val r1:Relation = new Relation val r1:Relation = new Relation
r1.setTarget(pub.getId) r1.setTarget(pub.getId)
r1.setSource(affId) r1.setSource(affId)
r1.setRelType("resultOrganization") r1.setRelType(ModelConstants.RESULT_ORGANIZATION)
r1.setRelClass("isAuthorInstitutionOf") r1.setRelClass(ModelConstants.IS_AUTHOR_INSTITUTION_OF)
r1.setSubRelType("affiliation") r1.setSubRelType(ModelConstants.AFFILIATION)
r1.setDataInfo(pub.getDataInfo) r1.setDataInfo(pub.getDataInfo)
r1.setCollectedfrom(List(DoiBoostMappingUtil.createMAGCollectedFrom()).asJava) r1.setCollectedfrom(List(DoiBoostMappingUtil.createMAGCollectedFrom()).asJava)
List(r, r1) List(r, r1)

56
dhp-workflows/dhp-doiboost/src/main/java/eu/dnetlib/doiboost/crossref/Crossref2Oaf.scala
View File

@ -121,10 +121,10 @@ case object Crossref2Oaf {
// TITLE // TITLE
val mainTitles = for {JString(title) <- json \ "title" if title.nonEmpty} yield createSP(title, "main title", "dnet:dataCite_title") val mainTitles = for {JString(title) <- json \ "title" if title.nonEmpty} yield createSP(title, "main title", ModelConstants.DNET_DATACITE_TITLE)
val originalTitles = for {JString(title) <- json \ "original-title" if title.nonEmpty} yield createSP(title, "alternative title", "dnet:dataCite_title") val originalTitles = for {JString(title) <- json \ "original-title" if title.nonEmpty} yield createSP(title, "alternative title", ModelConstants.DNET_DATACITE_TITLE)
val shortTitles = for {JString(title) <- json \ "short-title" if title.nonEmpty} yield createSP(title, "alternative title", "dnet:dataCite_title") val shortTitles = for {JString(title) <- json \ "short-title" if title.nonEmpty} yield createSP(title, "alternative title", ModelConstants.DNET_DATACITE_TITLE)
val subtitles = for {JString(title) <- json \ "subtitle" if title.nonEmpty} yield createSP(title, "subtitle", "dnet:dataCite_title") val subtitles = for {JString(title) <- json \ "subtitle" if title.nonEmpty} yield createSP(title, "subtitle", ModelConstants.DNET_DATACITE_TITLE)
result.setTitle((mainTitles ::: originalTitles ::: shortTitles ::: subtitles).asJava) result.setTitle((mainTitles ::: originalTitles ::: shortTitles ::: subtitles).asJava)
// DESCRIPTION // DESCRIPTION
@ -136,11 +136,11 @@ case object Crossref2Oaf {
result.setSource(sourceList.asJava) result.setSource(sourceList.asJava)
//RELEVANT DATE Mapping //RELEVANT DATE Mapping
val createdDate = generateDate((json \ "created" \ "date-time").extract[String], (json \ "created" \ "date-parts").extract[List[List[Int]]], "created", "dnet:dataCite_date") val createdDate = generateDate((json \ "created" \ "date-time").extract[String], (json \ "created" \ "date-parts").extract[List[List[Int]]], "created", ModelConstants.DNET_DATACITE_DATE)
val postedDate = generateDate((json \ "posted" \ "date-time").extractOrElse[String](null), (json \ "posted" \ "date-parts").extract[List[List[Int]]], "available", "dnet:dataCite_date") val postedDate = generateDate((json \ "posted" \ "date-time").extractOrElse[String](null), (json \ "posted" \ "date-parts").extract[List[List[Int]]], "available", ModelConstants.DNET_DATACITE_DATE)
val acceptedDate = generateDate((json \ "accepted" \ "date-time").extractOrElse[String](null), (json \ "accepted" \ "date-parts").extract[List[List[Int]]], "accepted", "dnet:dataCite_date") val acceptedDate = generateDate((json \ "accepted" \ "date-time").extractOrElse[String](null), (json \ "accepted" \ "date-parts").extract[List[List[Int]]], "accepted", ModelConstants.DNET_DATACITE_DATE)
val publishedPrintDate = generateDate((json \ "published-print" \ "date-time").extractOrElse[String](null), (json \ "published-print" \ "date-parts").extract[List[List[Int]]], "published-print", "dnet:dataCite_date") val publishedPrintDate = generateDate((json \ "published-print" \ "date-time").extractOrElse[String](null), (json \ "published-print" \ "date-parts").extract[List[List[Int]]], "published-print", ModelConstants.DNET_DATACITE_DATE)
val publishedOnlineDate = generateDate((json \ "published-online" \ "date-time").extractOrElse[String](null), (json \ "published-online" \ "date-parts").extract[List[List[Int]]], "published-online", "dnet:dataCite_date") val publishedOnlineDate = generateDate((json \ "published-online" \ "date-time").extractOrElse[String](null), (json \ "published-online" \ "date-parts").extract[List[List[Int]]], "published-online", ModelConstants.DNET_DATACITE_DATE)
val issuedDate = extractDate((json \ "issued" \ "date-time").extractOrElse[String](null), (json \ "issued" \ "date-parts").extract[List[List[Int]]]) val issuedDate = extractDate((json \ "issued" \ "date-time").extractOrElse[String](null), (json \ "issued" \ "date-parts").extract[List[List[Int]]])
if (StringUtils.isNotBlank(issuedDate)) { if (StringUtils.isNotBlank(issuedDate)) {
@ -155,7 +155,7 @@ case object Crossref2Oaf {
val subjectList:List[String] = (json \ "subject").extractOrElse[List[String]](List()) val subjectList:List[String] = (json \ "subject").extractOrElse[List[String]](List())
if (subjectList.nonEmpty) { if (subjectList.nonEmpty) {
result.setSubject(subjectList.map(s=> createSP(s, "keywords", "dnet:subject_classification_typologies")).asJava) result.setSubject(subjectList.map(s=> createSP(s, "keywords", ModelConstants.DNET_SUBJECT_TYPOLOGIES)).asJava)
} }
@ -193,7 +193,7 @@ case object Crossref2Oaf {
instance.setAccessright(getRestrictedQualifier()) instance.setAccessright(getRestrictedQualifier())
result.setInstance(List(instance).asJava) result.setInstance(List(instance).asJava)
instance.setInstancetype(OafUtils.createQualifier(cobjCategory.substring(0, 4), cobjCategory.substring(5), ModelConstants.DNET_PUBLICATION_RESOURCE, ModelConstants.DNET_PUBLICATION_RESOURCE)) instance.setInstancetype(OafUtils.createQualifier(cobjCategory.substring(0, 4), cobjCategory.substring(5), ModelConstants.DNET_PUBLICATION_RESOURCE, ModelConstants.DNET_PUBLICATION_RESOURCE))
result.setResourcetype(OafUtils.createQualifier(cobjCategory.substring(0, 4),"dnet:dataCite_resource")) result.setResourcetype(OafUtils.createQualifier(cobjCategory.substring(0, 4),ModelConstants.DNET_DATA_CITE_RESOURCE))
instance.setCollectedfrom(createCrossrefCollectedFrom()) instance.setCollectedfrom(createCrossrefCollectedFrom())
if (StringUtils.isNotBlank(issuedDate)) { if (StringUtils.isNotBlank(issuedDate)) {
@ -293,9 +293,9 @@ case object Crossref2Oaf {
val r = new Relation val r = new Relation
r.setSource(sourceId) r.setSource(sourceId)
r.setTarget(targetId) r.setTarget(targetId)
r.setRelType("resultProject") r.setRelType(ModelConstants.RESULT_PROJECT)
r.setRelClass(relClass) r.setRelClass(relClass)
r.setSubRelType("outcome") r.setSubRelType(ModelConstants.OUTCOME)
r.setCollectedfrom(List(cf).asJava) r.setCollectedfrom(List(cf).asJava)
r.setDataInfo(di) r.setDataInfo(di)
r.setLastupdatetimestamp(ts) r.setLastupdatetimestamp(ts)
@ -310,8 +310,8 @@ case object Crossref2Oaf {
funder.award.get.map(extractField).filter(a => a!= null && a.nonEmpty).foreach( funder.award.get.map(extractField).filter(a => a!= null && a.nonEmpty).foreach(
award => { award => {
val targetId = getProjectId(nsPrefix, DHPUtils.md5(award)) val targetId = getProjectId(nsPrefix, DHPUtils.md5(award))
queue += generateRelation(sourceId, targetId , "isProducedBy") queue += generateRelation(sourceId, targetId , ModelConstants.IS_PRODUCED_BY)
queue += generateRelation(targetId , sourceId, "produces") queue += generateRelation(targetId , sourceId, ModelConstants.PRODUCES)
} }
) )
} }
@ -342,22 +342,22 @@ case object Crossref2Oaf {
case "10.13039/501100001602" => generateSimpleRelationFromAward(funder, "aka_________", a => a.replace("SFI", "")) case "10.13039/501100001602" => generateSimpleRelationFromAward(funder, "aka_________", a => a.replace("SFI", ""))
case "10.13039/501100000923" => generateSimpleRelationFromAward(funder, "arc_________", a => a) case "10.13039/501100000923" => generateSimpleRelationFromAward(funder, "arc_________", a => a)
case "10.13039/501100000038"=> val targetId = getProjectId("nserc_______" , "1e5e62235d094afd01cd56e65112fc63") case "10.13039/501100000038"=> val targetId = getProjectId("nserc_______" , "1e5e62235d094afd01cd56e65112fc63")
queue += generateRelation(sourceId, targetId, "isProducedBy" ) queue += generateRelation(sourceId, targetId, ModelConstants.IS_PRODUCED_BY)
queue += generateRelation(targetId, sourceId, "produces" ) queue += generateRelation(targetId, sourceId, ModelConstants.PRODUCES)
case "10.13039/501100000155"=> val targetId = getProjectId("sshrc_______" , "1e5e62235d094afd01cd56e65112fc63") case "10.13039/501100000155"=> val targetId = getProjectId("sshrc_______" , "1e5e62235d094afd01cd56e65112fc63")
queue += generateRelation(sourceId,targetId, "isProducedBy" ) queue += generateRelation(sourceId,targetId, ModelConstants.IS_PRODUCED_BY)
queue += generateRelation(targetId,sourceId, "produces" ) queue += generateRelation(targetId,sourceId, ModelConstants.PRODUCES)
case "10.13039/501100000024"=> val targetId = getProjectId("cihr________" , "1e5e62235d094afd01cd56e65112fc63") case "10.13039/501100000024"=> val targetId = getProjectId("cihr________" , "1e5e62235d094afd01cd56e65112fc63")
queue += generateRelation(sourceId,targetId, "isProducedBy" ) queue += generateRelation(sourceId,targetId, ModelConstants.IS_PRODUCED_BY)
queue += generateRelation(targetId,sourceId, "produces" ) queue += generateRelation(targetId,sourceId, ModelConstants.PRODUCES)
case "10.13039/501100002848" => generateSimpleRelationFromAward(funder, "conicytf____", a => a) case "10.13039/501100002848" => generateSimpleRelationFromAward(funder, "conicytf____", a => a)
case "10.13039/501100003448" => generateSimpleRelationFromAward(funder, "gsrt________", extractECAward) case "10.13039/501100003448" => generateSimpleRelationFromAward(funder, "gsrt________", extractECAward)
case "10.13039/501100010198" => generateSimpleRelationFromAward(funder, "sgov________", a=>a) case "10.13039/501100010198" => generateSimpleRelationFromAward(funder, "sgov________", a=>a)
case "10.13039/501100004564" => generateSimpleRelationFromAward(funder, "mestd_______", extractECAward) case "10.13039/501100004564" => generateSimpleRelationFromAward(funder, "mestd_______", extractECAward)
case "10.13039/501100003407" => generateSimpleRelationFromAward(funder, "miur________", a=>a) case "10.13039/501100003407" => generateSimpleRelationFromAward(funder, "miur________", a=>a)
val targetId = getProjectId("miur________" , "1e5e62235d094afd01cd56e65112fc63") val targetId = getProjectId("miur________" , "1e5e62235d094afd01cd56e65112fc63")
queue += generateRelation(sourceId,targetId, "isProducedBy" ) queue += generateRelation(sourceId,targetId, ModelConstants.IS_PRODUCED_BY)
queue += generateRelation(targetId,sourceId, "produces" ) queue += generateRelation(targetId,sourceId, ModelConstants.PRODUCES)
case "10.13039/501100006588" | case "10.13039/501100006588" |
"10.13039/501100004488" => generateSimpleRelationFromAward(funder, "irb_hr______", a=>a.replaceAll("Project No.", "").replaceAll("HRZZ-","") ) "10.13039/501100004488" => generateSimpleRelationFromAward(funder, "irb_hr______", a=>a.replaceAll("Project No.", "").replaceAll("HRZZ-","") )
case "10.13039/501100006769"=> generateSimpleRelationFromAward(funder, "rsf_________", a=>a) case "10.13039/501100006769"=> generateSimpleRelationFromAward(funder, "rsf_________", a=>a)
@ -365,8 +365,8 @@ case object Crossref2Oaf {
case "10.13039/501100004410"=> generateSimpleRelationFromAward(funder, "tubitakf____", a =>a) case "10.13039/501100004410"=> generateSimpleRelationFromAward(funder, "tubitakf____", a =>a)
case "10.10.13039/100004440"=> generateSimpleRelationFromAward(funder, "wt__________", a =>a) case "10.10.13039/100004440"=> generateSimpleRelationFromAward(funder, "wt__________", a =>a)
case "10.13039/100004440"=> val targetId = getProjectId("wt__________" , "1e5e62235d094afd01cd56e65112fc63") case "10.13039/100004440"=> val targetId = getProjectId("wt__________" , "1e5e62235d094afd01cd56e65112fc63")
queue += generateRelation(sourceId,targetId, "isProducedBy" ) queue += generateRelation(sourceId,targetId, ModelConstants.IS_PRODUCED_BY)
queue += generateRelation(targetId,sourceId, "produces" ) queue += generateRelation(targetId,sourceId, ModelConstants.PRODUCES)
case _ => logger.debug("no match for "+funder.DOI.get ) case _ => logger.debug("no match for "+funder.DOI.get )
@ -376,7 +376,7 @@ case object Crossref2Oaf {
} else { } else {
funder.name match { funder.name match {
case "European Union’s Horizon 2020 research and innovation program" => generateSimpleRelationFromAward(funder, "corda__h2020", extractECAward) case "European Union’s Horizon 2020 research and innovation program" => generateSimpleRelationFromAward(funder, "corda__h2020", extractECAward)
case "European Union's" => case "European Union's" =>
generateSimpleRelationFromAward(funder, "corda__h2020", extractECAward) generateSimpleRelationFromAward(funder, "corda__h2020", extractECAward)
generateSimpleRelationFromAward(funder, "corda_______", extractECAward) generateSimpleRelationFromAward(funder, "corda_______", extractECAward)
@ -384,8 +384,8 @@ case object Crossref2Oaf {
"The French National Research Agency" => generateSimpleRelationFromAward(funder, "anr_________", a => a) "The French National Research Agency" => generateSimpleRelationFromAward(funder, "anr_________", a => a)
case "CONICYT, Programa de Formación de Capital Humano Avanzado" => generateSimpleRelationFromAward(funder, "conicytf____", extractECAward) case "CONICYT, Programa de Formación de Capital Humano Avanzado" => generateSimpleRelationFromAward(funder, "conicytf____", extractECAward)
case "Wellcome Trust Masters Fellowship" => val targetId = getProjectId("wt__________", "1e5e62235d094afd01cd56e65112fc63") case "Wellcome Trust Masters Fellowship" => val targetId = getProjectId("wt__________", "1e5e62235d094afd01cd56e65112fc63")
queue += generateRelation(sourceId, targetId, "isProducedBy" ) queue += generateRelation(sourceId, targetId, ModelConstants.IS_PRODUCED_BY )
queue += generateRelation(targetId, sourceId, "produces" ) queue += generateRelation(targetId, sourceId, ModelConstants.PRODUCES )
case _ => logger.debug("no match for "+funder.name ) case _ => logger.debug("no match for "+funder.name )
} }