From 2753044d13da0465b8b9061e70252ed6ac69a325 Mon Sep 17 00:00:00 2001
From: Claudio Atzori <claudio.atzori@isti.cnr.it>
Date: Thu, 11 Jan 2024 16:28:26 +0100
Subject: [PATCH] refined mapping for the extraction of the original resource
 type

---
 .../dhp/oa/graph/raw/OafToOafMapper.java      | 30 +++++++++++++++++--
 .../dhp/oa/graph/raw/OdfToOafMapper.java      | 19 ++++++++----
 2 files changed, 42 insertions(+), 7 deletions(-)

diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OafToOafMapper.java b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OafToOafMapper.java
index a63296d18..eee518353 100644
--- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OafToOafMapper.java
+++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OafToOafMapper.java
@@ -11,6 +11,7 @@ import java.util.List;
 import java.util.Set;
 import java.util.stream.Collectors;
 
+import org.apache.commons.lang3.ObjectUtils;
 import org.apache.commons.lang3.StringUtils;
 import org.dom4j.Document;
 import org.dom4j.Element;
@@ -27,6 +28,15 @@ import eu.dnetlib.dhp.schema.oaf.utils.ModelHardLimits;
 
 public class OafToOafMapper extends AbstractMdRecordToOafMapper {
 
+	private static Set<String> DC_TYPE_PUBLICATION_VERSION = new HashSet<>();
+
+	static {
+		DC_TYPE_PUBLICATION_VERSION.add("info:eu-repo/semantics/submittedVersion");
+		DC_TYPE_PUBLICATION_VERSION.add("info:eu-repo/semantics/acceptedVersion");
+		DC_TYPE_PUBLICATION_VERSION.add("info:eu-repo/semantics/publishedVersion");
+		DC_TYPE_PUBLICATION_VERSION.add("info:eu-repo/semantics/updatedVersion");
+	}
+
 	public OafToOafMapper(final VocabularyGroup vocs, final boolean invisible, final boolean shouldHashId,
 		final boolean forceOrginalId) {
 		super(vocs, invisible, shouldHashId, forceOrginalId);
@@ -192,24 +202,40 @@ public class OafToOafMapper extends AbstractMdRecordToOafMapper {
 	/**
 	 * The Dublin Core element dc:type can be repeated, but we need to base our mapping on a single value
 	 * So this method tries to give precedence to the COAR resource type, when available. Otherwise, it looks for the
-	 * openaire's info:eu-repo type, and as last resort picks the 1st type text available
+	 * openaire's info:eu-repo type, but excluding the following
+	 *
+	 * info:eu-repo/semantics/draft
+	 * info:eu-repo/semantics/submittedVersion
+	 * info:eu-repo/semantics/acceptedVersion
+	 * info:eu-repo/semantics/publishedVersion
+	 * info:eu-repo/semantics/updatedVersion
+	 *
+	 * Then, it picks the 1st dc:type text available and, in case there is no dc:type element, as last resort it tries
+	 * to extract the type from the dr:CobjCategory element
+	 *
+	 * Examples:
 	 *
 	 *     	<dc:type>http://purl.org/coar/resource_type/c_5794</dc:type>
 	 *     	<dc:type>info:eu-repo/semantics/article</dc:type>
 	 *     	<dc:type>Conference article</dc:type>
+	 *     	<dr:CobjCategory type="publication">0006</dr:CobjCategory>
 	 *
 	 * @param doc the input document
 	 * @return the chosen resource type
 	 */
 	@Override
 	protected String findOriginalType(Document doc) {
-		return (String) doc
+		final String dcType = (String) doc
 			.selectNodes("//dc:type")
 			.stream()
 			.map(o -> "" + ((Node) o).getText().trim())
+			.filter(t -> !DC_TYPE_PUBLICATION_VERSION.contains(t))
 			.sorted(new OriginalTypeComparator())
 			.findFirst()
 			.orElse(null);
+
+		final String drCobjCategory = doc.valueOf("//dr:CobjCategory/text()");
+		return ObjectUtils.firstNonNull(dcType, drCobjCategory);
 	}
 
 	@Override
diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java
index e63b01a00..08529125c 100644
--- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java
+++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/OdfToOafMapper.java
@@ -221,27 +221,36 @@ public class OdfToOafMapper extends AbstractMdRecordToOafMapper {
 	}
 
 	/**
-	 * The Datacite element
+	 * Extracts the resource type from The Datacite element
 	 *
-	 * <datacite:resourceType xs:anyURI="http://purl.org/coar/resource_type/c_6501">journal article</datacite:resourceType>
+	 * <datacite:resourceType
+	 * 		anyURI="http://purl.org/coar/resource_type/c_6501"
+	 * 		uri="http://purl.org/coar/resource_type/c_6501"
+	 * 	    resourceTypeGeneral="Dataset">journal article</datacite:resourceType>
 	 *
 	 * @param doc the input document
 	 * @return the chosen resource type
 	 */
 	@Override
 	protected String findOriginalType(Document doc) {
-		return Optional
+		final String resourceType = Optional
 			.ofNullable(
 				(Element) doc
 					.selectSingleNode(
 						"//*[local-name()='metadata']/*[local-name() = 'resource']/*[local-name() = 'resourceType']"))
 			.map(element -> {
-				final String resourceTypeURI = element.attributeValue("anyURI");
+				final String resourceTypeURI = element.attributeValue("uri");
+				final String resourceTypeAnyURI = element.attributeValue("anyURI");
 				final String resourceTypeTxt = element.getText();
+				final String resourceTypeGeneral = element.attributeValue("resourceTypeGeneral");
 
-				return ObjectUtils.firstNonNull(resourceTypeURI, resourceTypeTxt);
+				return ObjectUtils
+					.firstNonNull(resourceTypeURI, resourceTypeAnyURI, resourceTypeTxt, resourceTypeGeneral);
 			})
 			.orElse(null);
+
+		final String drCobjCategory = doc.valueOf("//dr:CobjCategory/text()");
+		return ObjectUtils.firstNonNull(resourceType, drCobjCategory);
 	}
 
 	@Override