Merge pull request 'XML serialisation of instances with the same URLs - 2nd round' (#204) from instance_group_by_url into beta

Reviewed-on: D-Net/dnet-hadoop#204

dhp-workflows/dhp-graph-provision/src/main/java/eu/dnetlib/dhp/oa/provision/model/XmlInstance.java

@@ -12,6 +12,7 @@ import scala.Serializable;
 public class XmlInstance implements Serializable {
 
 	public static final AccessRight UNKNOWN_ACCESS_RIGHT;
+	public static final Qualifier UNKNOWN_REVIEW_LEVEL;
 
 	static {
 		UNKNOWN_ACCESS_RIGHT = new AccessRight();
@@ -19,6 +20,12 @@ public class XmlInstance implements Serializable {
 		UNKNOWN_ACCESS_RIGHT.setClassname(ModelConstants.UNKNOWN);
 		UNKNOWN_ACCESS_RIGHT.setSchemeid(ModelConstants.DNET_ACCESS_MODES);
 		UNKNOWN_ACCESS_RIGHT.setSchemename(ModelConstants.DNET_ACCESS_MODES);
+
+		UNKNOWN_REVIEW_LEVEL = new Qualifier();
+		UNKNOWN_REVIEW_LEVEL.setClassid("0000");
+		UNKNOWN_REVIEW_LEVEL.setClassname(ModelConstants.UNKNOWN);
+		UNKNOWN_ACCESS_RIGHT.setSchemeid(ModelConstants.DNET_REVIEW_LEVELS);
+		UNKNOWN_REVIEW_LEVEL.setSchemename(ModelConstants.DNET_REVIEW_LEVELS);
 	}
 
 	private String url;
@@ -50,7 +57,7 @@ public class XmlInstance implements Serializable {
 	// typed results
 	private String processingchargecurrency;
 
-	private Set<Qualifier> refereed = Sets.newHashSet();; // peer-review status
+	private Qualifier refereed; // peer-review status
 
 	public String getUrl() {
 		return url;
@@ -148,11 +155,11 @@ public class XmlInstance implements Serializable {
 		this.processingchargecurrency = processingchargecurrency;
 	}
 
-	public Set<Qualifier> getRefereed() {
+	public Qualifier getRefereed() {
 		return refereed;
 	}
 
-	public void setRefereed(Set<Qualifier> refereed) {
+	public void setRefereed(Qualifier refereed) {
 		this.refereed = refereed;
 	}
 }

dhp-workflows/dhp-graph-provision/src/main/java/eu/dnetlib/dhp/oa/provision/utils/XmlRecordFactory.java

@@ -31,6 +31,7 @@ import org.dom4j.Node;
 import org.dom4j.io.OutputFormat;
 import org.dom4j.io.SAXReader;
 import org.dom4j.io.XMLWriter;
+import org.json4s.Xml;
 
 import com.fasterxml.jackson.databind.ObjectMapper;
 import com.google.common.base.Joiner;
@@ -1208,14 +1209,8 @@ public class XmlRecordFactory implements Serializable {
 
 					if (instance.getRefereed() != null) {
 						fields
-							.addAll(
+							.add(
-								instance
+								XmlSerializationUtils.mapQualifier("refereed", instance.getRefereed()));
-									.getRefereed()
-									.stream()
-									.filter(Objects::nonNull)
-									.filter(r -> !r.isBlank())
-									.map(r -> XmlSerializationUtils.mapQualifier("refereed", r))
-									.collect(Collectors.toList()));
 					}
 					if (instance.getProcessingchargeamount() != null
 						&& isNotBlank(instance.getProcessingchargeamount())) {
@@ -1359,13 +1354,20 @@ public class XmlRecordFactory implements Serializable {
 					.map(Instance::getAccessright)
 					.min(new AccessRightComparator<AccessRight>())
 					.orElse(XmlInstance.UNKNOWN_ACCESS_RIGHT));
+		instance
+			.setRefereed(
+				instances
+					.stream()
+					.map(Pair::getValue)
+					.map(i -> Optional.ofNullable(i.getRefereed()).orElse(XmlInstance.UNKNOWN_REVIEW_LEVEL))
+					.min(new RefereedComparator())
+					.orElse(XmlInstance.UNKNOWN_REVIEW_LEVEL));
 
 		instances.forEach(p -> {
 			final Instance i = p.getRight();
 			instance.getCollectedfrom().add(i.getCollectedfrom());
 			instance.getHostedby().add(i.getHostedby());
 			instance.getInstancetype().add(i.getInstancetype());
-			instance.getRefereed().add(i.getRefereed());
 			instance
 				.setProcessingchargeamount(
 					Optional.ofNullable(i.getProcessingchargeamount()).map(apc -> apc.getValue()).orElse(null));

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

@@ -66,7 +66,7 @@ public class XmlRecordFactoryTest {
 		assertEquals("doi", doc.valueOf("//instance/alternateidentifier/@classid"));
 		assertEquals("10.5689/LIB.2018.2853550", doc.valueOf("//instance/alternateidentifier/text()"));
 
-		assertEquals(3, doc.selectNodes("//instance").size());
+		assertEquals(2, doc.selectNodes("//instance").size());
 
 		assertEquals("1721.47", doc.valueOf("//processingchargeamount/text()"));
 		assertEquals("EUR", doc.valueOf("//processingchargecurrency/text()"));

dhp-workflows/dhp-graph-provision/src/test/resources/eu/dnetlib/dhp/oa/provision/publication.json

@@ -453,6 +453,12 @@
         "schemeid": "dnet:dataCite_resource",
         "schemename": "dnet:dataCite_resource"
       },
+      "refereed": {
+        "classid": "0001",
+        "classname": "peerReviewed",
+        "schemeid": "dnet:review_levels",
+        "schemename": "dnet:review_levels"
+      },
       "license": {
         "dataInfo": {
           "deletedbyinference": false,
@@ -585,6 +591,12 @@
         "schemeid": "dnet:dataCite_resource",
         "schemename": "dnet:dataCite_resource"
       },
+      "refereed": {
+        "classid": "0000",
+        "classname": "UNKNOWN",
+        "schemeid": "dnet:review_levels",
+        "schemename": "dnet:review_levels"
+      },
       "license": {
         "dataInfo": {
           "deletedbyinference": false,
@@ -717,6 +729,12 @@
         "schemeid": "dnet:dataCite_resource",
         "schemename": "dnet:dataCite_resource"
       },
+      "refereed": {
+        "classid": "0002",
+        "classname": "nonPeerReviewed",
+        "schemeid": "dnet:review_levels",
+        "schemename": "dnet:review_levels"
+      },
       "license": {
         "dataInfo": {
           "deletedbyinference": false,
@@ -849,6 +867,12 @@
         "schemeid": "dnet:dataCite_resource",
         "schemename": "dnet:dataCite_resource"
       },
+      "refereed": {
+        "classid": "0001",
+        "classname": "peerReviewed",
+        "schemeid": "dnet:review_levels",
+        "schemename": "dnet:review_levels"
+      },
       "license": {
         "dataInfo": {
           "deletedbyinference": false,
@@ -866,7 +890,7 @@
         "value": ""
       },
       "url": [
-        "http://dx.doi.org/10.1109/TED.2018.2853552",
+        "http://dx.doi.org/10.1109/TED.2018.2853551",
         "http://dx.doi.org/10.1109/TED.2018.2853554"
       ]
     },