implemented scholix Flat mapping

dhp-common/src/main/scala/eu/dnetlib/dhp/sx/graph/scholix/ScholixFlat.java

@@ -0,0 +1,158 @@
+package eu.dnetlib.dhp.sx.graph.scholix;
+
+import java.util.List;
+
+public class ScholixFlat {
+    private String identifier;
+    private String relationType;
+    private String sourceId;
+    private String sourceType;
+    private String sourceSubType;
+    private List<String> sourcePid;
+    private List<String> sourcePidType;
+    private List<String> sourcePublisher;
+    private String targetId;
+    private String targetType;
+    private String targetSubType;
+    private List<String> targetPid;
+    private List<String> targetPidType;
+    private List<String> targetPublisher;
+    private List<String> linkProviders;
+    private String publicationDate;
+    private String blob;
+
+    public String getIdentifier() {
+        return identifier;
+    }
+
+    public void setIdentifier(String identifier) {
+        this.identifier = identifier;
+    }
+
+    public String getRelationType() {
+        return relationType;
+    }
+
+    public void setRelationType(String relationType) {
+        this.relationType = relationType;
+    }
+    public String getSourceId() {
+        return sourceId;
+    }
+
+    public void setSourceId(String sourceId) {
+        this.sourceId = sourceId;
+    }
+
+    public String getSourceType() {
+        return sourceType;
+    }
+
+    public void setSourceType(String sourceType) {
+        this.sourceType = sourceType;
+    }
+
+    public String getSourceSubType() {
+        return sourceSubType;
+    }
+
+    public void setSourceSubType(String sourceSubType) {
+        this.sourceSubType = sourceSubType;
+    }
+
+    public List<String> getSourcePid() {
+        return sourcePid;
+    }
+
+    public void setSourcePid(List<String> sourcePid) {
+        this.sourcePid = sourcePid;
+    }
+
+    public List<String> getSourcePidType() {
+        return sourcePidType;
+    }
+
+    public void setSourcePidType(List<String> sourcePidType) {
+        this.sourcePidType = sourcePidType;
+    }
+
+    public List<String> getSourcePublisher() {
+        return sourcePublisher;
+    }
+
+    public void setSourcePublisher(List<String> sourcePublisher) {
+        this.sourcePublisher = sourcePublisher;
+    }
+
+    public String getTargetId() {
+        return targetId;
+    }
+
+    public void setTargetId(String targetId) {
+        this.targetId = targetId;
+    }
+
+    public String getTargetType() {
+        return targetType;
+    }
+
+    public void setTargetType(String targetType) {
+        this.targetType = targetType;
+    }
+
+    public String getTargetSubType() {
+        return targetSubType;
+    }
+
+    public void setTargetSubType(String targetSubType) {
+        this.targetSubType = targetSubType;
+    }
+
+    public List<String> getTargetPid() {
+        return targetPid;
+    }
+
+    public void setTargetPid(List<String> targetPid) {
+        this.targetPid = targetPid;
+    }
+
+    public List<String> getTargetPidType() {
+        return targetPidType;
+    }
+
+    public void setTargetPidType(List<String> targetPidType) {
+        this.targetPidType = targetPidType;
+    }
+
+    public List<String> getTargetPublisher() {
+        return targetPublisher;
+    }
+
+    public void setTargetPublisher(List<String> targetPublisher) {
+        this.targetPublisher = targetPublisher;
+    }
+
+    public List<String> getLinkProviders() {
+        return linkProviders;
+    }
+
+    public void setLinkProviders(List<String> linkProviders) {
+        this.linkProviders = linkProviders;
+    }
+
+    public String getPublicationDate() {
+        return publicationDate;
+    }
+
+    public void setPublicationDate(String publicationDate) {
+        this.publicationDate = publicationDate;
+    }
+
+    public String getBlob() {
+        return blob;
+    }
+
+    public void setBlob(String blob) {
+        this.blob = blob;
+    }
+}

dhp-common/src/main/scala/eu/dnetlib/dhp/sx/graph/scholix/ScholixUtils.scala

@@ -1,14 +1,15 @@
 package eu.dnetlib.dhp.sx.graph.scholix
 
-import eu.dnetlib.dhp.schema.oaf.{Publication, Relation, Result, StructuredProperty}
+import eu.dnetlib.dhp.schema.oaf.{Dataset, OtherResearchProduct, Publication, Relation, Result, Software, StructuredProperty}
 import eu.dnetlib.dhp.schema.sx.scholix._
-import eu.dnetlib.dhp.schema.sx.summary.{CollectedFromType, SchemeValue, ScholixSummary, Typology}
+import eu.dnetlib.dhp.schema.sx.summary.{AuthorPid, CollectedFromType, SchemeValue, ScholixSummary, Typology}
 import eu.dnetlib.dhp.utils.DHPUtils
 import org.apache.spark.sql.expressions.Aggregator
 import org.apache.spark.sql.{Encoder, Encoders}
 import org.json4s
 import org.json4s.DefaultFormats
 import org.json4s.jackson.JsonMethods.parse
+
 import scala.collection.JavaConverters._
 import scala.io.Source
 
@@ -59,6 +60,36 @@ object ScholixUtils extends Serializable {
     }
   }
 
+  def flattenizeScholix(input: Scholix, json: String): ScholixFlat = {
+
+    val flat: ScholixFlat = new ScholixFlat
+    flat.setIdentifier(input.getIdentifier)
+    flat.setRelationType(input.getRelationship.getName)
+    if (input.getSource != null && input.getSource.getIdentifier != null) {
+      flat.setSourceId(input.getSource.getDnetIdentifier)
+      flat.setSourcePid(input.getSource.getIdentifier.asScala.map(p => p.getIdentifier).distinct.toList.asJava)
+      flat.setSourcePidType(input.getSource.getIdentifier.asScala.map(p => p.getSchema).distinct.toList.asJava)
+      flat.setSourceType(input.getSource.getObjectType)
+      flat.setSourceSubType(input.getSource.getObjectSubType)
+    } else return null
+    if (input.getSource.getPublisher != null)
+      flat.setSourcePublisher(input.getSource.getPublisher.asScala.map(p => p.getName).toList.asJava)
+    if (input.getTarget != null && input.getTarget.getIdentifier != null) {
+      flat.setTargetId(input.getTarget.getDnetIdentifier)
+      flat.setTargetPid(input.getTarget.getIdentifier.asScala.map(p => p.getIdentifier).distinct.toList.asJava)
+      flat.setTargetPidType(input.getTarget.getIdentifier.asScala.map(p => p.getSchema).distinct.toList.asJava)
+      flat.setTargetType(input.getTarget.getObjectType)
+      flat.setTargetSubType(input.getTarget.getObjectSubType)
+    } else return null
+    if (input.getTarget.getPublisher != null)
+      flat.setTargetPublisher(input.getTarget.getPublisher.asScala.map(p => p.getName).distinct.toList.asJava)
+    flat.setPublicationDate(input.getPublicationDate)
+    if (input.getLinkprovider != null)
+      flat.setLinkProviders(input.getLinkprovider.asScala.map(l => l.getName).distinct.toList.asJava)
+    flat.setBlob(json);
+    flat
+  }
+
   def inverseRelationShip(rel: ScholixRelationship): ScholixRelationship = {
     new ScholixRelationship(rel.getInverse, rel.getSchema, rel.getName)
 
@@ -232,7 +263,14 @@ object ScholixUtils extends Serializable {
 
     if (summaryObject.getAuthor != null && !summaryObject.getAuthor.isEmpty) {
       val l: List[ScholixEntityId] =
-        summaryObject.getAuthor.asScala.map(a => new ScholixEntityId(a, null)).toList
+        summaryObject.getAuthor.asScala.map(a => {
+          if (a.getORCID != null)
+            new ScholixEntityId(
+              a.getFullname,
+              List(new ScholixIdentifier(a.getORCID, "ORCID", s"https://orcid.org/${a.getORCID}")).asJava
+            )
+          else new ScholixEntityId(a.getFullname, null)
+        }).toList
       if (l.nonEmpty)
         r.setCreator(l.asJava)
     }
@@ -377,10 +415,13 @@ object ScholixUtils extends Serializable {
     if (persistentIdentifiers.isEmpty)
       return null
     s.setLocalIdentifier(persistentIdentifiers.asJava)
-    if (r.isInstanceOf[Publication])
+    r match {
-      s.setTypology(Typology.publication)
+      case _: Publication => s.setTypology(Typology.publication)
-    else
+      case _: Dataset => s.setTypology(Typology.dataset)
-      s.setTypology(Typology.dataset)
+      case _: Software => s.setTypology(Typology.software)
+      case _: OtherResearchProduct => s.setTypology(Typology.otherresearchproduct)
+      case _ =>
+    }
 
     s.setSubType(r.getInstance().get(0).getInstancetype.getClassname)
 
@@ -393,7 +434,20 @@ object ScholixUtils extends Serializable {
     }
 
     if (r.getAuthor != null && !r.getAuthor.isEmpty) {
-      val authors: List[String] = r.getAuthor.asScala.map(a => a.getFullname).toList
+      val authors: List[AuthorPid] = r.getAuthor.asScala
+        .map(a => {
+          var ORCID: String = null
+          if (a.getPid != null) {
+            val result = a.getPid.asScala.find(p =>
+              p.getQualifier != null && p.getQualifier.getClassid != null && p.getQualifier.getClassid.toLowerCase
+                .contains("orcid")
+            )
+            if (result.isDefined)
+              ORCID = result.get.getValue
+          }
+          new AuthorPid(a.getFullname, ORCID)
+        })
+        .toList
       if (authors.nonEmpty)
         s.setAuthor(authors.asJava)
     }

dhp-common/src/test/java/eu/dnetlib/dhp/sx/graph/scholix/ScholixFlatTest.java

@@ -0,0 +1,42 @@
+
+package eu.dnetlib.dhp.sx.graph.scholix;
+
+import static org.junit.jupiter.api.Assertions.*;
+
+import java.io.*;
+import java.util.zip.GZIPInputStream;
+
+import org.junit.jupiter.api.Test;
+
+import com.fasterxml.jackson.databind.ObjectMapper;
+
+import eu.dnetlib.dhp.schema.sx.scholix.Scholix;
+
+public class ScholixFlatTest {
+
+	@Test
+	public void flattenScholixTest() throws IOException {
+		final ObjectMapper mapper = new ObjectMapper();
+		InputStream gzipStream = new GZIPInputStream(getClass().getResourceAsStream("scholix_records.gz"));
+		Reader decoder = new InputStreamReader(gzipStream, "UTF-8");
+		BufferedReader buffered = new BufferedReader(decoder);
+		String line;
+		FileWriter myWriter = new FileWriter("/Users/sandro/Downloads/records");
+		while ((line = buffered.readLine()) != null) {
+			final Scholix s = mapper.readValue(line, Scholix.class);
+			final ScholixFlat flat = ScholixUtils.flattenizeScholix(s, line);
+			assertNotNull(s);
+			assertNotNull(flat);
+			assertEquals(s.getIdentifier(), flat.getIdentifier());
+			assertEquals(s.getRelationship().getName(), flat.getRelationType());
+			assertEquals(s.getSource().getObjectType(), flat.getSourceType());
+			assertEquals(s.getSource().getObjectSubType(), flat.getSourceSubType());
+			myWriter.write(mapper.writeValueAsString(flat));
+			myWriter.write("\n");
+
+		}
+
+		myWriter.close();
+
+	}
+}

dhp-workflows/dhp-graph-provision/src/main/java/eu/dnetlib/dhp/sx/provision/ScholixFlat.java

@@ -1,194 +1,190 @@
+
 package eu.dnetlib.dhp.sx.provision;
 
-
-import com.fasterxml.jackson.core.JsonProcessingException;
-import com.fasterxml.jackson.databind.ObjectMapper;
-import eu.dnetlib.dhp.schema.sx.scholix.Scholix;
-import eu.dnetlib.dhp.schema.sx.scholix.ScholixResource;
-
 import java.util.ArrayList;
 import java.util.List;
 
+import com.fasterxml.jackson.core.JsonProcessingException;
+import com.fasterxml.jackson.databind.ObjectMapper;
+
+import eu.dnetlib.dhp.schema.sx.scholix.Scholix;
+import eu.dnetlib.dhp.schema.sx.scholix.ScholixResource;
+
 public class ScholixFlat {
-    private static ObjectMapper MAPPER = new ObjectMapper();
+	private static ObjectMapper MAPPER = new ObjectMapper();
-    private List<String> linkProvider= new ArrayList<>();
+	private List<String> linkProvider = new ArrayList<>();
 
-    private String publicationDate;
+	private String publicationDate;
 
-    private List<String> sourceLinkPublisher = new ArrayList<>();
+	private List<String> sourceLinkPublisher = new ArrayList<>();
-    private List<String> targetLinkPublisher = new ArrayList<>();
+	private List<String> targetLinkPublisher = new ArrayList<>();
 
-    private String sourceDnetIdentifier ;
+	private String sourceDnetIdentifier;
-    private String targetDnetIdentifier ;
+	private String targetDnetIdentifier;
-    private List<String> sourcePids = new ArrayList<>();
+	private List<String> sourcePids = new ArrayList<>();
-    private List<String> sourcePidTypes = new ArrayList<>();
+	private List<String> sourcePidTypes = new ArrayList<>();
-    private List<String> targetPids = new ArrayList<>();
+	private List<String> targetPids = new ArrayList<>();
-    private List<String> targetPidTypes = new ArrayList<>();
+	private List<String> targetPidTypes = new ArrayList<>();
 
-    private String json;
+	private String json;
 
+	public void addLinkProvider(final String providerName) {
+		addStringToList(providerName, this.linkProvider);
+	}
 
+	public void addSourceLinkPublisher(final String linkPublisher) {
+		addStringToList(linkPublisher, sourceLinkPublisher);
 
-    public void addLinkProvider(final String providerName) {
+	}
-        addStringToList(providerName, this.linkProvider);
-    }
 
-    public void addSourceLinkPublisher(final String linkPublisher) {
+	public void addTargetLinkPublisher(final String linkPublisher) {
-        addStringToList(linkPublisher, sourceLinkPublisher);
+		addStringToList(linkPublisher, targetLinkPublisher);
 
-    }
+	}
-    public void addTargetLinkPublisher(final String linkPublisher) {
-        addStringToList(linkPublisher, targetLinkPublisher);
 
-    }
+	public void addSourcePid(final String pid) {
+		addStringToList(pid, sourcePids);
+	}
 
-    public void addSourcePid(final String pid) {
+	public void addSourcePidType(final String pidType) {
-        addStringToList(pid, sourcePids);
+		addStringToList(pidType, sourcePidTypes);
-    }
+	}
 
-    public void addSourcePidType(final String pidType) {
+	public void addTargetPidType(final String pidType) {
-        addStringToList(pidType, sourcePidTypes);
+		addStringToList(pidType, targetPidTypes);
-    }
+	}
 
-    public void addTargetPidType(final  String pidType) {
+	public void addTargetPid(final String pid) {
-        addStringToList(pidType, targetPidTypes);
+		addStringToList(pid, targetPids);
-    }
+	}
 
+	public void addStringToList(final String s, final List<String> l) {
+		if (l != null && !l.contains(s))
+			l.add(s);
+	}
 
-    public void addTargetPid(final String pid) {
+	public String getSourceDnetIdentifier() {
-        addStringToList(pid, targetPids);
+		return sourceDnetIdentifier;
-    }
+	}
 
-    public void  addStringToList(final String s, final List<String>l ) {
+	public void setSourceDnetIdentifier(String sourceDnetIdentifier) {
-        if (l!= null && !l.contains(s))
+		this.sourceDnetIdentifier = sourceDnetIdentifier;
-            l.add(s);
+	}
-    }
 
-    public String getSourceDnetIdentifier() {
+	public String getTargetDnetIdentifier() {
-        return sourceDnetIdentifier;
+		return targetDnetIdentifier;
-    }
+	}
 
-    public void setSourceDnetIdentifier(String sourceDnetIdentifier) {
+	public void setTargetDnetIdentifier(String targetDnetIdentifier) {
-        this.sourceDnetIdentifier = sourceDnetIdentifier;
+		this.targetDnetIdentifier = targetDnetIdentifier;
-    }
+	}
 
-    public String getTargetDnetIdentifier() {
+	public List<String> getSourcePids() {
-        return targetDnetIdentifier;
+		return sourcePids;
-    }
+	}
 
-    public void setTargetDnetIdentifier(String targetDnetIdentifier) {
+	public void setSourcePids(List<String> sourcePids) {
-        this.targetDnetIdentifier = targetDnetIdentifier;
+		this.sourcePids = sourcePids;
-    }
+	}
 
-    public List<String> getSourcePids() {
+	public List<String> getSourcePidTypes() {
-        return sourcePids;
+		return sourcePidTypes;
-    }
+	}
 
-    public void setSourcePids(List<String> sourcePids) {
+	public void setSourcePidTypes(List<String> sourcePidTypes) {
-        this.sourcePids = sourcePids;
+		this.sourcePidTypes = sourcePidTypes;
-    }
+	}
 
-    public List<String> getSourcePidTypes() {
+	public List<String> getTargetPids() {
-        return sourcePidTypes;
+		return targetPids;
-    }
+	}
 
-    public void setSourcePidTypes(List<String> sourcePidTypes) {
+	public void setTargetPids(List<String> targetPids) {
-        this.sourcePidTypes = sourcePidTypes;
+		this.targetPids = targetPids;
-    }
+	}
 
-    public List<String> getTargetPids() {
+	public List<String> getTargetPidTypes() {
-        return targetPids;
+		return targetPidTypes;
-    }
+	}
 
-    public void setTargetPids(List<String> targetPids) {
+	public void setTargetPidTypes(List<String> targetPidTypes) {
-        this.targetPids = targetPids;
+		this.targetPidTypes = targetPidTypes;
-    }
+	}
 
-    public List<String> getTargetPidTypes() {
+	public List<String> getSourceLinkPublisher() {
-        return targetPidTypes;
+		return sourceLinkPublisher;
-    }
+	}
 
-    public void setTargetPidTypes(List<String> targetPidTypes) {
+	public void setSourceLinkPublisher(List<String> sourceLinkPublisher) {
-        this.targetPidTypes = targetPidTypes;
+		this.sourceLinkPublisher = sourceLinkPublisher;
-    }
+	}
 
-    public List<String> getSourceLinkPublisher() {
+	public List<String> getTargetLinkPublisher() {
-        return sourceLinkPublisher;
+		return targetLinkPublisher;
-    }
+	}
 
-    public void setSourceLinkPublisher(List<String> sourceLinkPublisher) {
+	public void setTargetLinkPublisher(List<String> targetLinkPublisher) {
-        this.sourceLinkPublisher = sourceLinkPublisher;
+		this.targetLinkPublisher = targetLinkPublisher;
-    }
+	}
 
-    public List<String> getTargetLinkPublisher() {
+	public List<String> getLinkProvider() {
-        return targetLinkPublisher;
+		return linkProvider;
-    }
+	}
 
-    public void setTargetLinkPublisher(List<String> targetLinkPublisher) {
+	public void setLinkProvider(List<String> linkProvider) {
-        this.targetLinkPublisher = targetLinkPublisher;
+		this.linkProvider = linkProvider;
-    }
+	}
 
+	public String getPublicationDate() {
+		return publicationDate;
+	}
 
-    public List<String> getLinkProvider() {
+	public void setPublicationDate(String publicationDate) {
-        return linkProvider;
+		this.publicationDate = publicationDate;
-    }
+	}
 
-    public void setLinkProvider(List<String> linkProvider) {
+	public String getJson() {
-        this.linkProvider = linkProvider;
+		return json;
-    }
+	}
 
-    public String getPublicationDate() {
+	public void setJson(String json) {
-        return publicationDate;
+		this.json = json;
-    }
+	}
 
-    public void setPublicationDate(String publicationDate) {
+	public static ScholixFlat fromScholix(final Scholix scholix) throws JsonProcessingException {
-        this.publicationDate = publicationDate;
+		if (scholix == null || scholix.getSource() == null || scholix.getTarget() == null)
-    }
+			return null;
+		final ScholixFlat flat = new ScholixFlat();
+		if (scholix.getLinkprovider() != null)
+			scholix.getLinkprovider().forEach(l -> flat.addLinkProvider(l.getName()));
 
-    public String getJson() {
+		flat.setPublicationDate(scholix.getPublicationDate());
-        return json;
-    }
 
-    public void setJson(String json) {
+		final ScholixResource source = scholix.getSource();
-        this.json = json;
+		flat.setSourceDnetIdentifier(source.getDnetIdentifier());
-    }
+		if (source.getIdentifier() != null) {
+			source.getIdentifier().forEach(i -> {
+				flat.addSourcePid(i.getIdentifier());
+				flat.addSourcePidType(i.getSchema());
+			});
+		}
+		if (source.getPublisher() != null) {
+			source.getPublisher().forEach(p -> flat.addSourceLinkPublisher(p.getName()));
+		}
 
-    public static ScholixFlat fromScholix(final Scholix scholix) throws JsonProcessingException {
+		final ScholixResource target = scholix.getSource();
-        if (scholix== null || scholix.getSource()==null || scholix.getTarget()== null)
+		flat.setTargetDnetIdentifier(target.getDnetIdentifier());
-            return null;
+		if (target.getIdentifier() != null) {
-        final ScholixFlat flat = new ScholixFlat();
+			target.getIdentifier().forEach(i -> {
-        if (scholix.getLinkprovider()!= null)
+				flat.addTargetPid(i.getIdentifier());
-            scholix.getLinkprovider().forEach(l ->flat.addLinkProvider(l.getName()));
+				flat.addTargetPidType(i.getSchema());
-
+			});
-        flat.setPublicationDate(scholix.getPublicationDate());
+		}
-
+		if (target.getPublisher() != null) {
-        final ScholixResource source = scholix.getSource();
+			target.getPublisher().forEach(p -> flat.addTargetLinkPublisher(p.getName()));
-        flat.setSourceDnetIdentifier(source.getDnetIdentifier());
+		}
-        if (source.getIdentifier()!= null) {
+		flat.setJson(MAPPER.writeValueAsString(scholix));
-            source.getIdentifier().forEach(i -> {
+		return flat;
-                flat.addSourcePid(i.getIdentifier());
+	}
-                flat.addSourcePidType(i.getSchema());
-            });
-        }
-        if (source.getPublisher()!= null) {
-            source.getPublisher().forEach(p -> flat.addSourceLinkPublisher(p.getName()));
-        }
-
-
-
-        final ScholixResource target = scholix.getSource();
-        flat.setTargetDnetIdentifier(target.getDnetIdentifier());
-        if (target.getIdentifier()!= null) {
-            target.getIdentifier().forEach(i -> {
-                flat.addTargetPid(i.getIdentifier());
-                flat.addTargetPidType(i.getSchema());
-            });
-        }
-        if (target.getPublisher()!= null) {
-            target.getPublisher().forEach(p -> flat.addTargetLinkPublisher(p.getName()));
-        }
-        flat.setJson(MAPPER.writeValueAsString(scholix));
-        return flat;
-    }
 
 }

dhp-workflows/dhp-graph-provision/src/test/java/eu/dnetlib/dhp/oa/provision/XmlRecordFactoryTest.java

@@ -56,7 +56,7 @@ public class XmlRecordFactoryTest {
 
 		assertNotNull(doc);
 
-		//System.out.println(doc.asXML());
+		// System.out.println(doc.asXML());
 
 		assertEquals("0000-0001-9613-6638", doc.valueOf("//creator[@rank = '1']/@orcid"));
 		assertEquals("0000-0001-9613-6639", doc.valueOf("//creator[@rank = '1']/@orcid_pending"));

dhp-workflows/dhp-graph-provision/src/test/scala/eu/dnetlib/dhp/sx/provision/ScholixFlatTest.scala

@@ -6,19 +6,17 @@ import org.junit.Before
 
 import org.junit.jupiter.api.{Test}
 
-class ScholixFlatTest{
+class ScholixFlatTest {
-
-
-  var spark:SparkSession = null
 
+  var spark: SparkSession = null
 
   def initSpark(): Unit = {
 
-    if (spark!= null)
+    if (spark != null)
       return
     println("SONO QUI")
     val conf = new SparkConf
-    conf.setAppName(getClass.getSimpleName )
+    conf.setAppName(getClass.getSimpleName)
     conf.setMaster("local[*]")
     conf.set("spark.driver.host", "localhost")
     conf.set("hive.metastore.local", "true")
@@ -27,26 +25,22 @@ class ScholixFlatTest{
     spark = SparkSession
       .builder()
       .appName(getClass.getSimpleName)
-    .config(conf)
+      .config(conf)
       .getOrCreate()
   }
 
-
-
   def after(): Unit = {
     spark.stop()
   }
 
-
   @Test
-  def testScholixConversion (): Unit = {
+  def testScholixConversion(): Unit = {
     initSpark()
     val p = getClass.getResource("/eu/dnetlib/dhp/sx/provision/scholix_dump.zip").getPath
 
     val t = spark.read.text(p).count
     println(s"total =$t")
 
-
   }
 
 }

pom.xml

@@ -807,7 +807,7 @@
 		<mockito-core.version>3.3.3</mockito-core.version>
 		<mongodb.driver.version>3.4.2</mongodb.driver.version>
 		<vtd.version>[2.12,3.0)</vtd.version>
-		<dhp-schemas.version>[3.16.0]</dhp-schemas.version>
+		<dhp-schemas.version>[3.16.1-SNAPSHOT]</dhp-schemas.version>
 		<dnet-actionmanager-api.version>[4.0.3]</dnet-actionmanager-api.version>
 		<dnet-actionmanager-common.version>[6.0.5]</dnet-actionmanager-common.version>
 		<dnet-openaire-broker-common.version>[3.1.6]</dnet-openaire-broker-common.version>