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

reintroduced the hostedby patching to the datacite records

This commit is contained in:
Claudio Atzori 2022-01-21 09:15:13 +01:00
parent 42e8f76778
commit b37bc277c4
2 changed files with 33 additions and 1019 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1021
dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/datacite/hostedBy_map.json
View File

File diff suppressed because it is too large Load Diff

31
dhp-workflows/dhp-aggregation/src/main/scala/eu/dnetlib/dhp/datacite/DataciteToOAFTransformation.scala
View File

@ -19,11 +19,33 @@ import java.time.chrono.ThaiBuddhistDate
import java.time.format.DateTimeFormatter
import java.util.{Date, Locale}
import scala.collection.JavaConverters._
import scala.io.{Codec, Source}
object DataciteToOAFTransformation {
case class HostedByMapType(
openaire_id: String,
datacite_name: String,
official_name: String,
similarity: Option[Float]
) {}
val mapper = new ObjectMapper()
val unknown_repository: HostedByMapType = HostedByMapType(
ModelConstants.UNKNOWN_REPOSITORY_ORIGINALID,
ModelConstants.UNKNOWN_REPOSITORY.getValue,
ModelConstants.UNKNOWN_REPOSITORY.getValue,
Some(1.0f)
)
val hostedByMap: Map[String, HostedByMapType] = {
val s = Source.fromInputStream(getClass.getResourceAsStream("hostedBy_map.json")).mkString
implicit lazy val formats: DefaultFormats.type = org.json4s.DefaultFormats
lazy val json: org.json4s.JValue = parse(s)
json.extract[Map[String, HostedByMapType]]
}
/** This method should skip record if json contains invalid text
* defined in gile datacite_filter
*
@ -534,12 +556,9 @@ object DataciteToOAFTransformation {
if (client.isDefined) {
instance.setHostedby(
OafMapperUtils.keyValue(
generateDSId(ModelConstants.UNKNOWN_REPOSITORY_ORIGINALID),
ModelConstants.UNKNOWN_REPOSITORY.getValue
)
)
val hb = hostedByMap.getOrElse(client.get.toUpperCase(), unknown_repository)
instance.setHostedby(OafMapperUtils.keyValue(generateDSId(hb.openaire_id), hb.official_name))
instance.setCollectedfrom(DATACITE_COLLECTED_FROM)
instance.setUrl(List(s"https://dx.doi.org/$doi").asJava)
instance.setAccessright(access_rights_qualifier)