Merge branch 'master' of code-repo.d4science.org:D-Net/dhp-schemas

README.md

@@ -2,10 +2,11 @@ Description of the project
 --------------------------
 This project defines **object schemas** of the OpenAIRE main entities and the relationships that intercur among them. 
 Namely it defines the model for 
-  
+
--  **research product (result)** which subclasses in publication, dataset, other research product, software
+- the graph internal representation, defined under the package `eu.dnetlib.dhp.schema.oaf`
--  **data source** object describing the data provider (institutional repository, aggregators, cris systems)
+- the public graph dump representations, defined under the package `eu.dnetlib.dhp.schema.dump.oaf`
--  **organization** research bodies managing a data source or participating to a research project
+- the scholexplorer content representation, defined under the package `eu.dnetlib.dhp.schema.sx`
--  **project** research project
+- the contents acquired from the netadata aggregation subsystem, defined under the package `eu.dnetlib.dhp.schema.mdstore`
+- the ORCID common schemas, defined under the package `eu.dnetlib.dhp.schema.orcid`
  
 Te serialization of such objects (data store files) are used to pass data between workflow nodes in the processing pipeline.

pom.xml

@@ -5,7 +5,7 @@
     <groupId>eu.dnetlib.dhp</groupId>
     <artifactId>dhp-schemas</artifactId>
     <packaging>jar</packaging>
-    <version>2.6.14-SNAPSHOT</version>
+    <version>2.7.14-SNAPSHOT</version>
 
     <licenses>
         <license>

src/main/java/eu/dnetlib/dhp/schema/common/ModelConstants.java

@@ -55,54 +55,82 @@ public class ModelConstants {
 	public static final String SOFTWARE_RESULTTYPE_CLASSID = "software";
 	public static final String ORP_RESULTTYPE_CLASSID = "other";
 
-	public static final String RESULT_RESULT = "resultResult";
+	public static final String RESULT_RESULT = "resultResult"; // relType
 	/**
 	 * @deprecated Use {@link ModelConstants#RELATIONSHIP} instead.
 	 */
 	@Deprecated
-	public static final String PUBLICATION_DATASET = "publicationDataset";
+	public static final String PUBLICATION_DATASET = "publicationDataset"; // subreltype
-	public static final String IS_RELATED_TO = "isRelatedTo";
-	public static final String SUPPLEMENT = "supplement";
-	public static final String IS_SUPPLEMENT_TO = "isSupplementTo";
-	public static final String IS_SUPPLEMENTED_BY = "isSupplementedBy";
-	public static final String PART = "part";
-	public static final String IS_PART_OF = "isPartOf";
-	public static final String HAS_PART = "hasPart";
-	public static final String RELATIONSHIP = "relationship";
-	public static final String CITATION = "citation";
-	public static final String CITES = "cites";
-	public static final String IS_CITED_BY = "isCitedBy";
-	public static final String REVIEW = "review"; // subreltype
-	public static final String REVIEWS = "reviews";
-	public static final String IS_REVIEWED_BY = "isReviewedBy";
 
-	public static final String RESULT_PROJECT = "resultProject";
+	public static final String SUPPLEMENT = "supplement"; // subreltype
-	public static final String OUTCOME = "outcome";
+	public static final String IS_SUPPLEMENT_TO = "IsSupplementTo";
+	public static final String IS_SUPPLEMENTED_BY = "IsSupplementedBy";
+
+	public static final String PART = "part"; // subreltype
+	public static final String IS_PART_OF = "IsPartOf";
+	public static final String HAS_PART = "HasPart";
+
+	public static final String RELATIONSHIP = "relationship"; // subreltype
+
+	public static final String IS_RELATED_TO = "isRelatedTo";
+	public static final String IS_IDENTICAL_TO = "IsIdenticalTo";
+
+	public static final String REFERENCES = "References";
+	public static final String IS_REFERENCED_BY = "IsReferencedBy";
+	public static final String CONTINUES = "Continues";
+	public static final String IS_CONTINUED_BY = "IsContinuedBy";
+	public static final String DOCUMENTS = "Documents";
+	public static final String IS_DOCUMENTED_BY = "IsDocumentedBy";
+	public static final String IS_SOURCE_OF = "IsSourceOf";
+	public static final String IS_DERIVED_FROM = "IsDerivedFrom";
+	public static final String COMPILES = "Compiles";
+	public static final String IS_COMPILED_BY = "IsCompiledBy";
+
+	public static final String CITATION = "citation"; // subreltype
+	public static final String CITES = "Cites";
+	public static final String IS_CITED_BY = "IsCitedBy";
+
+	public static final String REVIEW = "review"; // subreltype
+	public static final String REVIEWS = "Reviews";
+	public static final String IS_REVIEWED_BY = "IsReviewedBy";
+
+	public static final String VERSION = "version"; // subreltype
+	public static final String IS_VERSION_OF = "IsVersionOf";
+	public static final String HAS_VERSION = "HasVersion";
+	public static final String IS_PREVIOUS_VERSION_OF = "IsPreviousVersionOf";
+	public static final String IS_NEW_VERSION_OF = "IsNewVersionOf";
+	public static final String IS_VARIANT_FORM_OF = "IsVariantFormOf";
+	public static final String IS_ORIGINAL_FORM_OF = "IsOriginalFormOf";
+	public static final String IS_OBSOLETED_BY = "IsObsoletedBy";
+	public static final String OBSOLETES = "Obsoletes";
+
+	public static final String RESULT_PROJECT = "resultProject"; // relType
+	public static final String OUTCOME = "outcome"; // subreltype
 	public static final String IS_PRODUCED_BY = "isProducedBy";
 	public static final String PRODUCES = "produces";
 
-	public static final String DATASOURCE_ORGANIZATION = "datasourceOrganization";
+	public static final String DATASOURCE_ORGANIZATION = "datasourceOrganization"; // relType
-	public static final String PROVISION = "provision";
+	public static final String PROVISION = "provision"; // subreltype
 	public static final String IS_PROVIDED_BY = "isProvidedBy";
 	public static final String PROVIDES = "provides";
 
-	public static final String PROJECT_ORGANIZATION = "projectOrganization";
+	public static final String PROJECT_ORGANIZATION = "projectOrganization"; // relType
-	public static final String PARTICIPATION = "participation";
+	public static final String PARTICIPATION = "participation"; // subreltype
 	public static final String HAS_PARTICIPANT = "hasParticipant";
 	public static final String IS_PARTICIPANT = "isParticipant";
 
-	public static final String RESULT_ORGANIZATION = "resultOrganization";
+	public static final String RESULT_ORGANIZATION = "resultOrganization"; // relType
-	public static final String AFFILIATION = "affiliation";
+	public static final String AFFILIATION = "affiliation"; // subreltype
 	public static final String IS_AUTHOR_INSTITUTION_OF = "isAuthorInstitutionOf";
 	public static final String HAS_AUTHOR_INSTITUTION = "hasAuthorInstitution";
 
-	public static final String ORG_ORG_RELTYPE = "organizationOrganization";
+	public static final String ORG_ORG_RELTYPE = "organizationOrganization"; // relType
 
-	public static final String DEDUP = "dedup";
+	public static final String DEDUP = "dedup"; // subreltype
 	public static final String MERGES = "merges";
 	public static final String IS_MERGED_IN = "isMergedIn";
 
-	public static final String SIMILARITY = "similarity";
+	public static final String SIMILARITY = "similarity"; // subreltype
 	public static final String IS_SIMILAR_TO = "isSimilarTo";
 
 	public static final String IS_DIFFERENT_FROM = "isDifferentFrom";

src/main/java/eu/dnetlib/dhp/schema/oaf/utils/CleaningFunctions.java

@@ -46,18 +46,27 @@ public class CleaningFunctions {
 	 * @return the PID containing the normalised value.
 	 */
 	public static StructuredProperty normalizePidValue(StructuredProperty pid) {
-		String value = Optional
+		pid.setValue(
-			.ofNullable(pid.getValue())
+				normalizePidValue(
-			.map(String::trim)
+					pid.getQualifier().getClassid(),
-			.orElseThrow(() -> new IllegalArgumentException("PID value cannot be empty"));
+					pid.getValue()));
-		switch (pid.getQualifier().getClassid()) {
 
-			// TODO add cleaning for more PID types as needed
-			case "doi":
-				pid.setValue(value.toLowerCase().replaceFirst(DOI_PREFIX_REGEX, DOI_PREFIX));
-				break;
-		}
 		return pid;
 	}
 
+	public static String normalizePidValue(String pidType, String pidValue) {
+		String value = Optional
+				.ofNullable(pidValue)
+				.map(String::trim)
+				.orElseThrow(() -> new IllegalArgumentException("PID value cannot be empty"));
+
+		switch (pidType) {
+
+			// TODO add cleaning for more PID types as needed
+			case "doi":
+				return value.toLowerCase().replaceFirst(DOI_PREFIX_REGEX, DOI_PREFIX);
+		}
+		return value;
+	}
+
 }

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

@@ -164,9 +164,7 @@ public class IdentifierFactory implements Serializable {
 					.stream()
 					// filter away PIDs provided by a DS that is not considered an authority for the
 					// given PID Type
-					.filter(p -> {
+					.filter(p -> shouldFilterPid(collectedFrom, p, mapHandles))
-						return shouldFilterPid(collectedFrom, p, mapHandles);
-					})
 					.map(CleaningFunctions::normalizePidValue)
 					.filter(CleaningFunctions::pidFilter))
 			.orElse(Stream.empty());
@@ -193,13 +191,17 @@ public class IdentifierFactory implements Serializable {
 	}
 
 	private static <T extends OafEntity> String idFromPid(T entity, StructuredProperty s, boolean md5) {
+		return idFromPid(ModelSupport.getIdPrefix(entity.getClass()), s.getQualifier().getClassid(), s.getValue(), md5);
+	}
+
+	public static String idFromPid(String numericPrefix, String pidType, String pidValue, boolean md5) {
 		return new StringBuilder()
-			.append(ModelSupport.getIdPrefix(entity.getClass()))
+				.append(numericPrefix)
-			.append(ID_PREFIX_SEPARATOR)
+				.append(ID_PREFIX_SEPARATOR)
-			.append(createPrefix(s.getQualifier().getClassid()))
+				.append(createPrefix(pidType))
-			.append(ID_SEPARATOR)
+				.append(ID_SEPARATOR)
-			.append(md5 ? md5(s.getValue()) : s.getValue())
+				.append(md5 ? md5(pidValue) : pidValue)
-			.toString();
+				.toString();
 	}
 
 	// create the prefix (length = 12)