Merge pull request 'Modification of Microsoft Academic Graph Mapping' (#435) from mag_only_doi into beta

Reviewed-on: D-Net/dnet-hadoop#435

dhp-workflows/dhp-aggregation/src/main/scala/eu/dnetlib/dhp/collection/mag/MagUtility.scala

@@ -79,23 +79,6 @@ object MagUtility extends Serializable {
   private val MAGCollectedFrom = keyValue(ModelConstants.MAG_ID, ModelConstants.MAG_NAME)
 
   private val MAGDataInfo: DataInfo = {
-    val di = new DataInfo
-    di.setDeletedbyinference(false)
-    di.setInferred(false)
-    di.setInvisible(false)
-    di.setTrust("0.9")
-    di.setProvenanceaction(
-      OafMapperUtils.qualifier(
-        ModelConstants.SYSIMPORT_ACTIONSET,
-        ModelConstants.SYSIMPORT_ACTIONSET,
-        ModelConstants.DNET_PROVENANCE_ACTIONS,
-        ModelConstants.DNET_PROVENANCE_ACTIONS
-      )
-    )
-    di
-  }
-
-  private val MAGDataInfoInvisible: DataInfo = {
     val di = new DataInfo
     di.setDeletedbyinference(false)
     di.setInferred(false)
@@ -111,8 +94,7 @@ object MagUtility extends Serializable {
     )
     di
   }
-
-  val datatypedict = Map(
+val datatypedict = Map(
     "bool"     -> BooleanType,
     "int"      -> IntegerType,
     "uint"     -> IntegerType,
@@ -453,7 +435,6 @@ object MagUtility extends Serializable {
 
       case "repository" =>
         result = new Publication()
-        result.setDataInfo(MAGDataInfoInvisible)
         qualifier(
           "0038",
           "Other literature type",
@@ -488,8 +469,7 @@ object MagUtility extends Serializable {
     }
 
     if (result != null) {
-      if (result.getDataInfo == null)
-        result.setDataInfo(MAGDataInfo)
+      result.setDataInfo(MAGDataInfo)
       val i = new Instance
       i.setInstancetype(tp)
       i.setInstanceTypeMapping(
@@ -512,7 +492,7 @@ object MagUtility extends Serializable {
       return null
 
     result.setCollectedfrom(List(MAGCollectedFrom).asJava)
-    val pidList = List(
+    var pidList = List(
       structuredProperty(
         paper.paperId.get.toString,
         qualifier(
@@ -525,7 +505,7 @@ object MagUtility extends Serializable {
       )
     )
 
-    result.setPid(pidList.asJava)
+
 
     result.setOriginalId(pidList.map(s => s.getValue).asJava)
 
@@ -618,10 +598,9 @@ object MagUtility extends Serializable {
     }
 
     val instance = result.getInstance().get(0)
-    instance.setPid(pidList.asJava)
-    if (paper.doi.orNull != null)
-      instance.setAlternateIdentifier(
-        List(
+
+    if (paper.doi.orNull != null) {
+      pidList = pidList ::: List(
           structuredProperty(
             paper.doi.get,
             qualifier(
@@ -632,8 +611,10 @@ object MagUtility extends Serializable {
             ),
             null
           )
-        ).asJava
-      )
+        )
+    }
+    instance.setPid(pidList.asJava)
+    result.setPid(pidList.asJava)
     instance.setUrl(paper.urls.get.asJava)
     instance.setHostedby(ModelConstants.UNKNOWN_REPOSITORY)
     instance.setCollectedfrom(MAGCollectedFrom)

dhp-workflows/dhp-aggregation/src/main/scala/eu/dnetlib/dhp/collection/mag/SparkMAGtoOAF.scala

@@ -35,9 +35,12 @@ class SparkMAGtoOAF(propertyPath: String, args: Array[String], log: Logger)
   def convertMAG(spark: SparkSession, magBasePath: String, mdStorePath: String): Unit = {
     import spark.implicits._
 
+
+
     spark.read
       .load(s"$magBasePath/mag_denormalized")
       .as[MAGPaper]
+      .filter(col("doi").isNotNull)
       .map(s => MagUtility.convertMAGtoOAF(s))
       .filter(s => s != null)
       .write

dhp-workflows/dhp-aggregation/src/test/scala/eu/dnetlib/dhp/collection/mag/MAGMappingTest.scala

@@ -3,13 +3,17 @@ package eu.dnetlib.dhp.collection.mag
 import com.fasterxml.jackson.databind.ObjectMapper
 import eu.dnetlib.dhp.schema.oaf.{Dataset, Publication, Result}
 import org.apache.spark.sql.SparkSession
+import org.apache.spark.sql.functions.col
 import org.junit.jupiter.api.Assertions._
 import org.junit.jupiter.api.Test
 
+
+
 class MAGMappingTest {
 
   val mapper = new ObjectMapper()
 
+
   def mappingTest(): Unit = {
 
     val spark = SparkSession
@@ -18,12 +22,12 @@ class MAGMappingTest {
       .master("local[*]")
       .getOrCreate()
 
-    val s = new SparkMagOrganizationAS(null, null, null)
-
-    s.generateAS(spark, "/home/sandro/Downloads/mag_test", "/home/sandro/Downloads/mag_AS")
-
+    val s = new SparkMAGtoOAF(null, null, null)
+    s.convertMAG(spark, "/Users/sandro/Downloads/", "/Users/sandro/Downloads/mag_OAF")
   }
 
+
+
   @Test
   def mappingMagType(): Unit = {