fix resolved url format, added alternate identifier from original pid

dhp-workflows/dhp-aggregation/src/main/scala/eu/dnetlib/dhp/bioschema/BioschemaModelConstants.scala

@@ -12,12 +12,17 @@ import java.util.regex.Pattern
   The following class are utility class used for the mapping from
   bioschema json datacite to OAF Schema
  */
+
 case class RelatedIdentifierType(
   relationType: String,
   relatedIdentifier: String,
   relatedIdentifierType: String
 ) {}
 
+case class AlternateIdentifierType(
+  alternateIdentifier: String
+) {}
+
 case class IdentifierType(
   identifier: String,
   identifierType: String
@@ -41,10 +46,10 @@ case class CreatorType(
 case class TitleType(title: Option[String], titleType: Option[String], lang: Option[String]) {}
 
 case class SubjectType(
-                        schemeURI: Option[String],
+  schemeURI: Option[String],
-                        value: Option[String],
+  value: Option[String],
-                        subjectScheme: Option[String]
+  subjectScheme: Option[String]
-                      ) {}
+) {}
 
 case class DescriptionType(descriptionType: Option[String], description: Option[String]) {}
 
@@ -70,9 +75,10 @@ object BioschemaModelConstants {
 
   val PED_PREFIX: String = "ped_________"
 
-  val resolvedURL: Map[String, String] = Map(
+  val resolvedURLPattern: Map[String, String] = Map(
-    "uniprot" -> "https://www.uniprot.org/uniprot/",
+    "https://identifiers.org/pubmed:"  -> "pubmed",
-    "pubmed"  -> "https://pubmed.ncbi.nlm.nih.gov/"
+    "http://purl.uniprot.org/uniprot/" -> "uniprot",
+    "https://identifiers.org/uniprot:" -> "uniprot"
   )
 
   val collectedFromMap: Map[String, KeyValue] = {

dhp-workflows/dhp-aggregation/src/main/scala/eu/dnetlib/dhp/bioschema/BioschemaToOAFTransformation.scala

@@ -11,9 +11,7 @@ import org.json4s.DefaultFormats
 import org.json4s.JsonAST.{JField, JObject, JString}
 import org.json4s.jackson.JsonMethods.parse
 
-import java.text.SimpleDateFormat
 import java.time.LocalDate
-import java.util.{Date, Locale}
 import scala.collection.JavaConverters._
 
 object BioschemaToOAFTransformation {
@@ -200,32 +198,42 @@ object BioschemaToOAFTransformation {
     if (result.getId == null)
       return List()
 
-    val alternativeIdentifierUrls: List[String] = for {
+    val alternativeIdentifierUrls: List[AlternateIdentifierType] = for {
       JObject(alternateIdentifiers)                               <- json \\ "alternateIdentifiers"
       JField("alternateIdentifier", JString(alternateIdentifier)) <- alternateIdentifiers
-      foundResolvedURLId = resolvedURL
+    } yield AlternateIdentifierType(alternateIdentifier)
-        .map(k => {
-          if (alternateIdentifier.contains(s"${k._1}:"))
-            k._1
-          else
-            null
-        })
-        .find(s => s != null)
-      alternativeIdentifierUrl = StringUtils.substringAfter(alternateIdentifier, s"${foundResolvedURLId.get}:")
-    } yield alternativeIdentifierUrl
 
-    alternativeIdentifierUrls.map(id => {
+    val alternativeIdentifierUrl: AlternateIdentifierType = alternativeIdentifierUrls.asJava.get(0)
-      var alternateIdentifier: StructuredProperty = null
+
-      alternateIdentifier = OafMapperUtils.structuredProperty(
+    val alternativeIdentifiers = resolvedURLPattern
-        id,
+      .map(pattern => {
-        "uniprot",
+        if (alternativeIdentifierUrl.alternateIdentifier.startsWith(s"${pattern._1}")) {
-        "uniprot",
+          val relatedId = StringUtils.substringAfter(alternativeIdentifierUrl.alternateIdentifier, s"${pattern._1}")
-        ModelConstants.DNET_PID_TYPES,
+          OafMapperUtils.structuredProperty(
-        ModelConstants.DNET_PID_TYPES,
+            relatedId,
-        dataInfo
+            pattern._2,
-      )
+            pattern._2,
-      instance.setAlternateIdentifier(List(alternateIdentifier).asJava)
+            ModelConstants.DNET_PID_TYPES,
-    })
+            ModelConstants.DNET_PID_TYPES,
+            dataInfo
+          )
+        } else
+          null
+      })
+      .find(s => s != null)
+      .get
+
+    val defaultAlternatedIdentifer: StructuredProperty = OafMapperUtils.structuredProperty(
+      pid,
+      datasourceKey,
+      datasourceKey,
+      ModelConstants.DNET_PID_TYPES,
+      ModelConstants.DNET_PID_TYPES,
+      dataInfo
+    )
+    var finalAlternativeIdentifiers: List[StructuredProperty] = List()
+    finalAlternativeIdentifiers = List(alternativeIdentifiers) ::: List(defaultAlternatedIdentifer)
+    instance.setAlternateIdentifier(finalAlternativeIdentifiers.asJava)
 
     if (exportLinks) {
       val rels: List[RelatedIdentifierType] = for {
@@ -272,21 +280,17 @@ object BioschemaToOAFTransformation {
 
         rel.setProperties(List(dateProps).asJava)
 
-        val foundResolvedURLId = resolvedURL
+        resolvedURLPattern
-          .map(k => {
+          .map(p => {
-            if (r.relatedIdentifier.contains(s"${k._1}:"))
+            if (r.relatedIdentifier.startsWith(s"${p._1}")) {
-              k._1
+              val relatedId = StringUtils.substringAfter(r.relatedIdentifier, s"${p._1}")
-            else
+              rel.setTarget(
+                DHPUtils.generateUnresolvedIdentifier(relatedId, p._2)
+              )
+            } else
               null
           })
-          .find(s => s != null);
+          .find(s => s != null)
-        if (foundResolvedURLId.nonEmpty) {
-          val relatedId = StringUtils.substringAfter(r.relatedIdentifier, s"${foundResolvedURLId.get}:")
-          rel.setTarget(s"${resolvedURL(foundResolvedURLId.get)}${relatedId}")
-        } else
-          rel.setTarget(
-            DHPUtils.generateUnresolvedIdentifier(r.relatedIdentifier, r.relatedIdentifierType)
-          )
         rel.setSource(id)
         rel.setCollectedfrom(List(collectedFromMap(datasourceKey)).asJava)
         rel.getCollectedfrom.asScala.map(c => c.getValue).toList