merged from dhp_oaf_model

dhp-common/src/main/java/eu/dnetlib/dhp/schema/oaf/CleaningFunctions.java

@@ -13,7 +13,7 @@ import eu.dnetlib.dhp.schema.common.ModelConstants;
 
 public class CleaningFunctions {
 
-	public static final String DOI_PREFIX_REGEX = "^.*10\\.";
+	public static final String DOI_PREFIX_REGEX = "(^10\\.|\\/10.)";
 	public static final String ORCID_PREFIX_REGEX = "^http(s?):\\/\\/orcid\\.org\\/";
 	public static final String CLEANING_REGEX = "(?:\\n|\\r|\\t)";
 
@@ -277,7 +277,7 @@ public class CleaningFunctions {
 
 			// TODO add cleaning for more PID types as needed
 			case "doi":
-				pid.setValue(value.toLowerCase().replaceAll(DOI_URL_PREFIX_REGEX, ""));
+				pid.setValue(value.toLowerCase().replaceAll(DOI_PREFIX_REGEX, "10."));
 				break;
 		}
 		return pid;

dhp-common/src/main/java/eu/dnetlib/dhp/schema/oaf/OafMapperUtils.java

@@ -9,9 +9,9 @@ import java.util.function.Function;
 import java.util.function.Predicate;
 import java.util.stream.Collectors;
 
+import eu.dnetlib.dhp.schema.common.AccessRightComparator;
 import org.apache.commons.lang3.StringUtils;
 
-import eu.dnetlib.dhp.schema.common.LicenseComparator;
 import eu.dnetlib.dhp.schema.common.ModelSupport;
 import eu.dnetlib.dhp.utils.DHPUtils;
 
@@ -327,10 +327,10 @@ public class OafMapperUtils {
 
 	protected static Qualifier getBestAccessRights(final List<Instance> instanceList) {
 		if (instanceList != null) {
-			final Optional<Qualifier> min = instanceList
+			final Optional<AccessRight> min = instanceList
 				.stream()
 				.map(i -> i.getAccessright())
-				.min(new LicenseComparator());
+				.min(new AccessRightComparator<>());
 
 			final Qualifier rights = min.isPresent() ? min.get() : new Qualifier();
 

dhp-common/src/main/java/eu/dnetlib/dhp/schema/oaf/utils/IdentifierFactory.java

@@ -8,6 +8,11 @@ import java.util.Objects;
 import java.util.Optional;
 import java.util.stream.Collectors;
 
+import com.google.common.collect.HashBiMap;
+import com.google.common.collect.Maps;
+import com.google.common.collect.Sets;
+import eu.dnetlib.dhp.schema.common.ModelConstants;
+import eu.dnetlib.dhp.schema.oaf.KeyValue;
 import org.apache.commons.lang3.StringUtils;
 
 import eu.dnetlib.dhp.schema.oaf.CleaningFunctions;
@@ -30,6 +35,13 @@ public class IdentifierFactory implements Serializable {
 
 	public static final int ID_PREFIX_LEN = 12;
 
+	public static final HashBiMap<String, String> PID_AUTHORITY = HashBiMap.create(2);
+
+	static {
+		PID_AUTHORITY.put(ModelConstants.CROSSREF_ID, "Crossref");
+		PID_AUTHORITY.put(ModelConstants.DATACITE_ID, "Datacite");
+	}
+
 	/**
 	 * Creates an identifier from the most relevant PID (if available) in the given entity T. Returns entity.id
 	 * when no PID is available
@@ -43,6 +55,14 @@ public class IdentifierFactory implements Serializable {
 			return entity.getId();
 		}
 
+		if (Optional.ofNullable(
+				entity.getCollectedfrom())
+				.map(c -> c.stream()
+						.noneMatch(cf -> PID_AUTHORITY.containsKey(cf.getKey()) || PID_AUTHORITY.containsValue(cf.getValue())))
+				.orElse(true)) {
+			return entity.getId();
+		}
+
 		Map<String, List<StructuredProperty>> pids = entity
 			.getPid()
 			.stream()

dhp-common/src/test/java/eu/dnetlib/dhp/schema/oaf/utils/IdentifierFactoryTest.java

@@ -23,7 +23,7 @@ public class IdentifierFactoryTest {
 	public void testCreateIdentifierForPublication() throws IOException {
 
 		verifyIdentifier(
-			"publication_doi1.json", "50|doi_________::" + DHPUtils.md5("10.1016/j.cmet.2011.03.013"), true);
+			"publication_doi1.json", "50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f", false);
 		verifyIdentifier(
 			"publication_doi2.json", "50|doi_________::" + DHPUtils.md5("10.1016/j.cmet.2010.03.013"), true);
 		verifyIdentifier("publication_pmc1.json", "50|pmc_________::" + DHPUtils.md5("21459329"), true);

dhp-common/src/test/resources/eu/dnetlib/dhp/schema/oaf/utils/publication_doi2.json

@@ -1 +1 @@
-{"id":"50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f","pid":[{"qualifier":{"classid":"doi"},"value":"10.1016/j.cmet.2011.03.013"},{"qualifier":{"classid":"doi"},"value":"10.1016/j.cmet.2010.03.013"},{"qualifier":{"classid":"urn"},"value":"urn:nbn:nl:ui:29-f3ed5f9e-edf6-457e-8848-61b58a4075e2"},{"qualifier":{"classid":"scp-number"},"value":"79953761260"},{"qualifier":{"classid":"pmc"},"value":"21459329"}]}
+{"id":"50|DansKnawCris::0829b5191605bdbea36d6502b8c1ce1f","pid":[{"qualifier":{"classid":"doi"},"value":"10.1016/j.cmet.2010.03.013"},{"qualifier":{"classid":"urn"},"value":"urn:nbn:nl:ui:29-f3ed5f9e-edf6-457e-8848-61b58a4075e2"},{"qualifier":{"classid":"scp-number"},"value":"79953761260"},{"qualifier":{"classid":"pmc"},"value":"21459329"}],"collectedfrom":[{"key":"10|openaire____::081b82f96300b6a6e3d282bad31cb6e2","value":"Crossref"}]}