Merge pull request #70 from DiSSCo/feature/iiif-er

Update EntityRelationship + IIIF mediaType

pom.xml

@@ -20,6 +20,7 @@
     <commons-beanutils.version>1.9.4</commons-beanutils.version>
     <jaxb2-maven-plugin.version>3.1.0</jaxb2-maven-plugin.version>
     <commons-compress.version>1.27.1</commons-compress.version>
+    <netty-common.version>4.1.115.Final</netty-common.version>
     <jakarta.activation-api.version>2.1.3</jakarta.activation-api.version>
     <jakarta.xml.bind-api.version>4.0.2</jakarta.xml.bind-api.version>
     <mockito-inline.version>5.2.0</mockito-inline.version>
@@ -87,8 +88,15 @@
       <groupId>org.gbif</groupId>
       <artifactId>dwca-io</artifactId>
       <version>${dwca-io.version}</version>
+      <!-- Needed to exclude this dep from dwca-io because of CVE-2024-47554 -->
+      <exclusions>
+        <exclusion>
+          <groupId>commons-io</groupId>
+          <artifactId>commons-io</artifactId>
+        </exclusion>
+      </exclusions>
     </dependency>
-    <!--    Needed to overwrite gbif dep with security issue in this lib -->
+    <!-- Needed to overwrite gbif dep with security issue in this lib -->
     <dependency>
       <groupId>commons-beanutils</groupId>
       <artifactId>commons-beanutils</artifactId>
@@ -100,6 +108,12 @@
       <artifactId>commons-compress</artifactId>
       <version>${commons-compress.version}</version>
     </dependency>
+    <!-- Necessary to mitigate CVE-2024-47535 in spring-boot 3.3.5 (18-11-2024) -->
+    <dependency>
+      <groupId>io.netty</groupId>
+      <artifactId>netty-common</artifactId>
+      <version>${netty-common.version}</version>
+    </dependency>
     <dependency>
       <groupId>org.codehaus.mojo</groupId>
       <artifactId>jaxb2-maven-plugin</artifactId>

src/main/java/eu/dissco/core/translator/service/DwcaService.java

@@ -137,7 +137,7 @@ public void getSpecimenData(Set<String> ids, Archive archive, AtomicInteger proc
         var specimenData = dwcaRepository.getCoreRecords(batch, getTableName(archive.getCore()));
         log.info("Got specimen batch: {}", batch.size());
         addExtensionsToSpecimen(archive, batch, specimenData);
-        log.info("Start translation and publishing of batch: {}", specimenData.values().size());
+        log.info("Start translation and publishing of batch: {}", specimenData.size());
         processDigitalSpecimen(specimenData.values(), optionalEmlData, processedRecords);
       }
     } catch (ReachedMaximumLimitException e) {

src/main/java/eu/dissco/core/translator/terms/BaseDigitalObjectDirector.java

@@ -2,7 +2,6 @@
 
 import static eu.dissco.core.translator.domain.AgentRoleType.COLLECTOR;
 import static eu.dissco.core.translator.domain.AgentRoleType.CREATOR;
-import static eu.dissco.core.translator.domain.AgentRoleType.DATA_TRANSLATOR;
 import static eu.dissco.core.translator.domain.AgentRoleType.GEOREFERENCER;
 import static eu.dissco.core.translator.domain.AgentRoleType.IDENTIFIER;
 import static eu.dissco.core.translator.domain.AgentRoleType.RIGHTS_OWNER;
@@ -15,14 +14,13 @@
 import static eu.dissco.core.translator.domain.RelationshipType.HAS_SOURCE_SYSTEM_ID;
 import static eu.dissco.core.translator.domain.RelationshipType.HAS_URL;
 import static eu.dissco.core.translator.schema.Agent.Type.SCHEMA_PERSON;
-import static eu.dissco.core.translator.schema.Agent.Type.SCHEMA_SOFTWARE_APPLICATION;
 import static eu.dissco.core.translator.terms.utils.AgentsUtils.addAgent;
+import static eu.dissco.core.translator.terms.utils.EntityRelationshipUtils.addEntityRelationship;
 
 import com.fasterxml.jackson.databind.JsonNode;
 import com.fasterxml.jackson.databind.ObjectMapper;
 import eu.dissco.core.translator.component.OrganisationNameComponent;
 import eu.dissco.core.translator.component.SourceSystemComponent;
-import eu.dissco.core.translator.domain.RelationshipType;
 import eu.dissco.core.translator.exception.OrganisationException;
 import eu.dissco.core.translator.exception.UnknownPhysicalSpecimenIdType;
 import eu.dissco.core.translator.properties.FdoProperties;
@@ -235,7 +233,6 @@
 import eu.dissco.core.translator.terms.specimen.stratigraphy.lithostratigraphic.Group;
 import eu.dissco.core.translator.terms.specimen.stratigraphy.lithostratigraphic.LithostratigraphicTerms;
 import eu.dissco.core.translator.terms.specimen.stratigraphy.lithostratigraphic.Member;
-import java.time.Instant;
 import java.util.ArrayList;
 import java.util.List;
 import lombok.RequiredArgsConstructor;
@@ -362,38 +359,32 @@ private String getOrganisationName(String organisationId) throws OrganisationExc
   private List<EntityRelationship> assembleDigitalSpecimenEntityRelationships(
       DigitalSpecimen ds) {
     var relationships = new ArrayList<EntityRelationship>();
-    relationships.add(getEntityRelationship(HAS_ORGANISATION_ID, ds.getOdsOrganisationID()));
-    relationships.add(getEntityRelationship(HAS_SOURCE_SYSTEM_ID, ds.getOdsSourceSystemID()));
-    relationships.add(getEntityRelationship(HAS_FDO_TYPE, fdoProperties.getDigitalSpecimenType()));
+    relationships.add(addEntityRelationship(HAS_ORGANISATION_ID, ds.getOdsOrganisationID(),
+        fdoProperties.getApplicationName(), fdoProperties.getApplicationPID()));
+    relationships.add(addEntityRelationship(HAS_SOURCE_SYSTEM_ID, ds.getOdsSourceSystemID(),
+        fdoProperties.getApplicationName(), fdoProperties.getApplicationPID()));
+    relationships.add(addEntityRelationship(HAS_FDO_TYPE, fdoProperties.getDigitalSpecimenType(),
+        fdoProperties.getApplicationName(), fdoProperties.getApplicationPID()));
     if (ds.getOdsPhysicalSpecimenIDType().equals(OdsPhysicalSpecimenIDType.RESOLVABLE)) {
       relationships.add(
-          getEntityRelationship(HAS_PHYSICAL_IDENTIFIER, ds.getOdsPhysicalSpecimenID()));
+          addEntityRelationship(HAS_PHYSICAL_IDENTIFIER, ds.getOdsPhysicalSpecimenID(),
+              fdoProperties.getApplicationName(), fdoProperties.getApplicationPID()));
     }
     if (ds.getDctermsLicense() != null && ds.getDctermsLicense().startsWith("http")) {
-      relationships.add(getEntityRelationship(HAS_LICENSE, ds.getDctermsLicense()));
+      relationships.add(addEntityRelationship(HAS_LICENSE, ds.getDctermsLicense(),
+          fdoProperties.getApplicationName(), fdoProperties.getApplicationPID()));
     }
     if (ds.getOdsHasCitations() != null) {
       for (Citation citation : ds.getOdsHasCitations()) {
         if (citation.getId() != null && citation.getId().startsWith("http")) {
-          relationships.add(getEntityRelationship(HAS_REFERENCE, citation.getId()));
+          relationships.add(addEntityRelationship(HAS_REFERENCE, citation.getId(),
+              fdoProperties.getApplicationName(), fdoProperties.getApplicationPID()));
         }
       }
     }
     return relationships;
   }
 
-  private EntityRelationship getEntityRelationship(RelationshipType relationshipType,
-      String relatedResource) {
-    var entityRelationship = new EntityRelationship()
-        .withType("ods:EntityRelationship")
-        .withDwcRelationshipOfResource(relationshipType.getName())
-        .withDwcRelatedResourceID(relatedResource)
-        .withDwcRelationshipEstablishedDate(java.util.Date.from(Instant.now()));
-    entityRelationship.setOdsHasAgents(addAgent(entityRelationship.getOdsHasAgents(), fdoProperties.getApplicationName(),
-        fdoProperties.getApplicationPID(), DATA_TRANSLATOR, SCHEMA_SOFTWARE_APPLICATION));
-    return entityRelationship;
-  }
-
   private List<eu.dissco.core.translator.schema.Identifier> assembleIdentifiers(JsonNode data) {
     var identifiers = new ArrayList<Identifier>();
     for (String identifierTerm : identifierTerms) {
@@ -461,7 +452,8 @@ protected Identification createIdentification(JsonNode data, boolean dwc) {
             termMapper.retrieveTerm(new VerbatimIdentification(), data, dwc))
         .withOdsHasTaxonIdentifications(List.of(mappedTaxonIdentification))
         .withOdsHasCitations(assembleIdentificationCitations(data, dwc));
-    identification.setOdsHasAgents(addAgent(identification.getOdsHasAgents(), termMapper.retrieveTerm(new IdentifiedBy(), data, dwc),
+    identification.setOdsHasAgents(addAgent(identification.getOdsHasAgents(),
+        termMapper.retrieveTerm(new IdentifiedBy(), data, dwc),
         termMapper.retrieveTerm(new IdentifiedByID(), data, dwc), IDENTIFIER, SCHEMA_PERSON));
     return identification;
   }
@@ -590,8 +582,9 @@ private List<Event> assembleEventTerms(JsonNode data, boolean dwc) {
         .withDwcVitality(termMapper.retrieveTerm(new Vitality(), data, dwc))
         .withOdsHasLocation(location)
         .withOdsHasAssertions(assertions);
-    event.setOdsHasAgents(addAgent(event.getOdsHasAgents(), termMapper.retrieveTerm(new RecordedBy(), data, dwc),
-        termMapper.retrieveTerm(new RecordedByID(), data, dwc), COLLECTOR, SCHEMA_PERSON));
+    event.setOdsHasAgents(
+        addAgent(event.getOdsHasAgents(), termMapper.retrieveTerm(new RecordedBy(), data, dwc),
+            termMapper.retrieveTerm(new RecordedByID(), data, dwc), COLLECTOR, SCHEMA_PERSON));
     return List.of(event);
   }
 
@@ -714,19 +707,24 @@ public DigitalMedia assembleDigitalMedia(boolean dwc, JsonNode mediaRecord,
   private List<EntityRelationship> assembleDigitalMediaEntityRelationships(
       DigitalMedia digitalMedia) {
     var relationships = new ArrayList<EntityRelationship>();
-    relationships.add(getEntityRelationship(HAS_URL, digitalMedia.getAcAccessURI()));
+    relationships.add(addEntityRelationship(HAS_URL, digitalMedia.getAcAccessURI(),
+        fdoProperties.getApplicationName(), fdoProperties.getApplicationPID()));
     relationships.add(
-        getEntityRelationship(HAS_ORGANISATION_ID, digitalMedia.getOdsOrganisationID()));
+        addEntityRelationship(HAS_ORGANISATION_ID, digitalMedia.getOdsOrganisationID(),
+            fdoProperties.getApplicationName(), fdoProperties.getApplicationPID()));
     relationships.add(
-        getEntityRelationship(HAS_FDO_TYPE, fdoProperties.getDigitalMediaType()));
+        addEntityRelationship(HAS_FDO_TYPE, fdoProperties.getDigitalMediaType(),
+            fdoProperties.getApplicationName(), fdoProperties.getApplicationPID()));
     if (digitalMedia.getDctermsRights() != null && digitalMedia.getDctermsRights()
         .startsWith("http")) {
       relationships.add(
-          getEntityRelationship(HAS_LICENSE, digitalMedia.getDctermsRights()));
+          addEntityRelationship(HAS_LICENSE, digitalMedia.getDctermsRights(),
+              fdoProperties.getApplicationName(), fdoProperties.getApplicationPID()));
     }
     if (digitalMedia.getDctermsSource() != null && digitalMedia.getDctermsSource()
         .startsWith("http")) {
-      relationships.add(getEntityRelationship(HAS_SOURCE, digitalMedia.getDctermsSource()));
+      relationships.add(addEntityRelationship(HAS_SOURCE, digitalMedia.getDctermsSource(),
+          fdoProperties.getApplicationName(), fdoProperties.getApplicationPID()));
     }
     return relationships;
   }

src/main/java/eu/dissco/core/translator/terms/BiocaseDigitalObjectDirector.java

@@ -24,7 +24,8 @@ public class BiocaseDigitalObjectDirector extends BaseDigitalObjectDirector {
   public BiocaseDigitalObjectDirector(ObjectMapper mapper, TermMapper termMapper,
       OrganisationNameComponent rorComponent, SourceSystemComponent sourceSystemComponent,
       FdoProperties fdoProperties) {
-    super(mapper, termMapper, rorComponent, sourceSystemComponent, fdoProperties, identifierTerms());
+    super(mapper, termMapper, rorComponent, sourceSystemComponent, fdoProperties,
+        identifierTerms());
   }
 
   private static List<String> identifierTerms() {

src/main/java/eu/dissco/core/translator/terms/DwcaDigitalObjectDirector.java

@@ -25,7 +25,8 @@ public class DwcaDigitalObjectDirector extends BaseDigitalObjectDirector {
   public DwcaDigitalObjectDirector(ObjectMapper mapper, TermMapper termMapper,
       OrganisationNameComponent rorComponent, SourceSystemComponent sourceSystemComponent,
       FdoProperties fdoProperties) {
-    super(mapper, termMapper, rorComponent, sourceSystemComponent, fdoProperties, identifierTerms());
+    super(mapper, termMapper, rorComponent, sourceSystemComponent, fdoProperties,
+        identifierTerms());
   }
 
   private static List<String> identifierTerms() {
@@ -52,7 +53,7 @@ protected List<Citation> assembleSpecimenCitations(JsonNode data, boolean dwc) {
       var references = data.get(EXTENSION).get("gbif:Reference");
       for (int i = 0; i < references.size(); i++) {
         var citationJson = references.get(i);
-        if (citationJson.properties().size() <= 1){
+        if (citationJson.properties().size() <= 1) {
           log.debug("Skipping citation with only one property: {}", citationJson);
         } else {
           citations.add(createCitation(citationJson, dwc));

src/main/java/eu/dissco/core/translator/terms/media/MediaType.java

@@ -12,9 +12,11 @@ public class MediaType extends Term {
   private static final String STILL_IMAGE = "Still_Image";
   private static final String SOUND = "Sound";
   private static final String MOVING_IMAGE = "Moving_Image";
+  private static final String INTERACTIVE_RESOURCE = "Interactive_Resource";
   private final List<String> dwcaTerms = List.of(TERM, "dc:type");
   private final List<String> imageFormats = List.of("IMAGE/JPG", "JPG", "IMAGE/JPEG",
       "JPEG", "IMAGE/PNG", "PNG", "IMAGE/TIF", "TIF");
+  private final List<String> interactiveResourceFormats = List.of("APPLICATION/JSON");
 
   @Override
   public String retrieveFromDWCA(JsonNode unit) {
@@ -51,6 +53,8 @@ public String retrieveFromABCD(JsonNode unit) {
       format = format.toUpperCase();
       if (imageFormats.contains(format)) {
         return STILL_IMAGE;
+      } else if (interactiveResourceFormats.contains(format)) {
+        return INTERACTIVE_RESOURCE;
       } else {
         log.warn("Unable to determine media type of digital media object");
         return null;

src/main/java/eu/dissco/core/translator/terms/specimen/DatasetID.java

@@ -5,6 +5,7 @@
 import java.util.List;
 
 public class DatasetID extends Term {
+
   public static final String TERM = DWC_PREFIX + "datasetID";
 
   private final List<String> dwcaTerms = List.of(TERM);

src/main/java/eu/dissco/core/translator/terms/specimen/DynamicProperties.java

@@ -5,6 +5,7 @@
 import java.util.List;
 
 public class DynamicProperties extends Term {
+
   public static final String TERM = DWC_PREFIX + "dynamicProperties";
 
   private final List<String> dwcaTerms = List.of(TERM);

src/main/java/eu/dissco/core/translator/terms/specimen/OrganismID.java

@@ -5,6 +5,7 @@
 import java.util.List;
 
 public class OrganismID extends Term {
+
   public static final String TERM = DWC_PREFIX + "organismID";
 
   private final List<String> dwcaTerms = List.of(TERM);

src/main/java/eu/dissco/core/translator/terms/specimen/OrganismName.java

@@ -5,6 +5,7 @@
 import java.util.List;
 
 public class OrganismName extends Term {
+
   public static final String TERM = DWC_PREFIX + "organismName";
 
   private final List<String> dwcaTerms = List.of(TERM);

src/main/java/eu/dissco/core/translator/terms/specimen/OrganismRemarks.java

@@ -5,6 +5,7 @@
 import java.util.List;
 
 public class OrganismRemarks extends Term {
+
   public static final String TERM = DWC_PREFIX + "organismRemarks";
 
   private final List<String> dwcaTerms = List.of(TERM);

src/main/java/eu/dissco/core/translator/terms/specimen/OrganismScope.java

@@ -5,6 +5,7 @@
 import java.util.List;
 
 public class OrganismScope extends Term {
+
   public static final String TERM = DWC_PREFIX + "organismScope";
 
   private final List<String> dwcaTerms = List.of(TERM);

src/main/java/eu/dissco/core/translator/terms/specimen/SpecimenName.java

@@ -9,7 +9,8 @@ public class SpecimenName extends Term {
 
   public String calculate(DigitalSpecimen ds) {
     var acceptedIdentification = retrieveAcceptedIdentification(ds);
-    if (acceptedIdentification != null && acceptedIdentification.getOdsHasTaxonIdentifications() != null
+    if (acceptedIdentification != null
+        && acceptedIdentification.getOdsHasTaxonIdentifications() != null
         && !acceptedIdentification.getOdsHasTaxonIdentifications().isEmpty()) {
       if (acceptedIdentification.getOdsHasTaxonIdentifications().get(0)
           .getDwcScientificName() != null) {

src/main/java/eu/dissco/core/translator/terms/specimen/TopicDiscipline.java

@@ -19,7 +19,8 @@ public class TopicDiscipline extends Term {
   public OdsTopicDiscipline calculate(DigitalSpecimen ds) {
     var basisOfRecord = ds.getDwcBasisOfRecord();
     var acceptedIdentification = retrieveAcceptedIdentification(ds);
-    if (acceptedIdentification != null && acceptedIdentification.getOdsHasTaxonIdentifications() != null
+    if (acceptedIdentification != null
+        && acceptedIdentification.getOdsHasTaxonIdentifications() != null
         && !acceptedIdentification.getOdsHasTaxonIdentifications().isEmpty()) {
       return getDiscipline(basisOfRecord,
           acceptedIdentification.getOdsHasTaxonIdentifications().get(0).getDwcKingdom());

src/main/java/eu/dissco/core/translator/terms/utils/AgentsUtils.java

@@ -65,30 +65,39 @@ private static void handleMultipleAgents(
       List<Agent> agents, String agentValue, String agentId, AgentRoleType role, Type type) {
     var ids = new String[0];
     var agentValues = new String[0];
-    if (agentValue != null && (agentValue.contains("&") || agentValue.contains("|"))) {
-      agentValues = Arrays.stream(agentValue.split("[&|]")).map(String::trim).toArray(String[]::new);
+    if (needsParsing(agentValue)) {
+      agentValues = Arrays.stream(agentValue.split("[&|]")).map(String::trim)
+          .toArray(String[]::new);
     }
-    if (agentId != null && (agentId.contains("&") || agentId.contains("|"))) {
+    if (needsParsing(agentId)) {
       ids = Arrays.stream(agentId.split("[&|]")).map(String::trim).toArray(String[]::new);
     }
     if (agentValues.length == ids.length) {
       for (int i = 0; i < agentValues.length; i++) {
         constructAgent(agents, agentValues[i], ids[i], role, type);
       }
     } else if (agentValues.length > ids.length) {
-      log.warn(
-          "The number of agentValues values is greater than ids, ignoring ids for term: {} and agentId: {}",
-          agentValue, agentId);
+      if (ids.length != 0) {
+        log.warn(
+            "The number of agentValues values is greater than ids, ignoring ids for term: {} and agentId: {}",
+            agentValue, agentId);
+      }
       for (String agent : agentValues) {
         constructAgent(agents, agent, null, role, type);
       }
     } else {
-      log.warn(
-          "The number of ids is greater than agentValue, ignoring agentValue values for term: {} and agentId: {}",
-          agentValue, agentId);
+      if (agentValues.length != 0) {
+        log.warn(
+            "The number of ids is greater than agentValue, ignoring agentValue values for term: {} and agentId: {}",
+            agentValue, agentId);
+      }
       for (String idValue : ids) {
         constructAgent(agents, null, idValue, role, type);
       }
     }
   }
+
+  private static boolean needsParsing(String value) {
+    return value != null && (value.contains("&") || value.contains("|"));
+  }
 }

src/main/java/eu/dissco/core/translator/terms/utils/EntityRelationshipUtils.java

@@ -0,0 +1,47 @@
+package eu.dissco.core.translator.terms.utils;
+
+import static eu.dissco.core.translator.domain.AgentRoleType.DATA_TRANSLATOR;
+import static eu.dissco.core.translator.schema.Agent.Type.SCHEMA_SOFTWARE_APPLICATION;
+import static eu.dissco.core.translator.terms.utils.AgentsUtils.addAgent;
+
+import eu.dissco.core.translator.domain.RelationshipType;
+import eu.dissco.core.translator.schema.EntityRelationship;
+import java.net.URI;
+import java.net.URISyntaxException;
+import java.time.Instant;
+import java.util.Date;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+public class EntityRelationshipUtils {
+
+  private EntityRelationshipUtils() {
+    // This is a Utility class
+  }
+
+  public static EntityRelationship addEntityRelationship(
+      RelationshipType relationshipType, String relatedResource, String agentName, String agentId) {
+    if (relatedResource == null) {
+      log.warn("Related resource for type {} is null. Skipping entity relationship creation.",
+          relationshipType.getName());
+      return null;
+    }
+    var entityRelationship = new EntityRelationship()
+        .withType("ods:EntityRelationship")
+        .withDwcRelationshipOfResource(relationshipType.getName())
+        .withDwcRelatedResourceID(relatedResource)
+        .withDwcRelationshipEstablishedDate(Date.from(Instant.now()));
+    entityRelationship.setOdsHasAgents(
+        addAgent(entityRelationship.getOdsHasAgents(), agentName, agentId, DATA_TRANSLATOR,
+            SCHEMA_SOFTWARE_APPLICATION));
+    if (relatedResource.startsWith("http")) {
+      try {
+        entityRelationship.setOdsRelatedResourceURI(new URI(relatedResource));
+      } catch (URISyntaxException e) {
+        log.warn("Could not create URI for related resource: " + relatedResource);
+      }
+    }
+    return entityRelationship;
+  }
+
+}