Merge pull request #43 from DiSSCo/feature/additional-abcd-mapping

Add additional mapping ABCD
DiSSCo · Oct 18, 2023 · 2e92f98 · 2e92f98
2 parents 8355b14 + 4081818
commit 2e92f98
Show file tree

Hide file tree

Showing 10 changed files with 55 additions and 26 deletions.
diff --git a/src/main/java/eu/dissco/core/translator/service/BioCaseService.java b/src/main/java/eu/dissco/core/translator/service/BioCaseService.java
@@ -69,7 +69,7 @@ public class BioCaseService implements WebClientService {
   private static final List<String> allowedBasisOfRecord = List.of("PRESERVEDSPECIMEN", "FOSSIL",
       "OTHER", "ROCK", "MINERAL", "METEORITE", "FOSSILSPECIMEN", "LIVINGSPECIMEN", "MATERIALSAMPLE",
       "FOSSIL SPECIMEN", "ROCKSPECIMEN", "ROCK SPECIMEN", "MINERALSPECIMEN", "MINERAL SPECIMEN",
-      "METEORITESPECIMEN", "METEORITE SPECIMEN");
+      "METEORITESPECIMEN", "METEORITE SPECIMEN", "HERBARIUM SHEET", "HERBARIUMSHEET", "DRIED");
 
   private final ObjectMapper mapper;
   private final WebClientProperties webClientProperties;
@@ -203,7 +203,7 @@ private void processUnit(DataSet dataset, Unit unit)
       }
     } else {
       log.info("Record with record basis: {} and id: {} is ignored ", unit.getRecordBasis(),
-          unit.getUnitGUID());
+          unit.getUnitID());
     }
   }
 

diff --git a/src/main/java/eu/dissco/core/translator/terms/BaseDigitalObjectDirector.java b/src/main/java/eu/dissco/core/translator/terms/BaseDigitalObjectDirector.java
@@ -553,7 +553,7 @@ private String getNormalisedPhysicalSpecimenId(OdsPhysicalSpecimenIdType physica
         || physicalSpecimenIdType.equals(OdsPhysicalSpecimenIdType.RESOLVABLE)) {
       return physicalSpecimenId;
     } else {
-      var minifiedSourceSystemId = sourceSystemId.substring(sourceSystemId.indexOf('/') + 1);
+      var minifiedSourceSystemId = sourceSystemId.substring(sourceSystemId.lastIndexOf('/') + 1);
       return physicalSpecimenId + ":" + minifiedSourceSystemId;
     }
   }

diff --git a/src/main/java/eu/dissco/core/translator/terms/BiocaseDigitalObjectDirector.java b/src/main/java/eu/dissco/core/translator/terms/BiocaseDigitalObjectDirector.java
@@ -34,6 +34,9 @@ private static List<String> identifierTerms() {
     list.add("abcd:unitIDNumeric");
     list.add("abcd:unitGUID");
     list.add("abcd:recordURI");
+    // For now just look at the two first accession numbers as a shortcut
+    list.add("abcd:specimenUnit/accessions/accessionDateAndAccessionCatalogueAndAccessionNumber/0/value");
+    list.add("abcd:specimenUnit/accessions/accessionDateAndAccessionCatalogueAndAccessionNumber/1/value");
     return list;
   }
 

diff --git a/src/main/java/eu/dissco/core/translator/terms/specimen/BasisOfRecord.java b/src/main/java/eu/dissco/core/translator/terms/specimen/BasisOfRecord.java
@@ -7,7 +7,8 @@
 public class BasisOfRecord extends Term {
 
   public static final String TERM = DWC_PREFIX + "basisOfRecord";
-
+  private static final List<String> PRESERVED_SPECIMEN_ALTERNATIVES = List.of("HERBARIUM SHEET",
+      "HERBARIUMSHEET", "DRIED");
   private final List<String> dwcaTerms = List.of(TERM);
   private final List<String> abcdTerms = List.of("abcd:recordBasis");
 
@@ -18,7 +19,11 @@ public String retrieveFromDWCA(JsonNode unit) {
 
   @Override
   public String retrieveFromABCD(JsonNode unit) {
-    return super.searchJsonForTerm(unit, abcdTerms);
+    var result = super.searchJsonForTerm(unit, abcdTerms);
+    if (result != null && PRESERVED_SPECIMEN_ALTERNATIVES.contains(result.toUpperCase())) {
+      return "Preserved specimen";
+    }
+    return result;
   }
 
   @Override

diff --git a/src/main/java/eu/dissco/core/translator/terms/specimen/Preparations.java b/src/main/java/eu/dissco/core/translator/terms/specimen/Preparations.java
@@ -8,7 +8,7 @@ public class Preparations extends Term {
 
   public static final String TERM = DWC_PREFIX + "preparations";
   private final List<String> dwcaTerms = List.of("dwc:preparations");
-  private final List<String> abcdTerms = List.of("abcd:kindOfUnit/0/value");
+  private final List<String> abcdTerms = List.of("abcd:kindOfUnit/0/value", "abcd:specimenUnit/preparations/preparation/0/preparationType/value");
 
   @Override
   public String retrieveFromDWCA(JsonNode unit) {

diff --git a/src/main/java/eu/dissco/core/translator/terms/specimen/identification/IdentifiedBy.java b/src/main/java/eu/dissco/core/translator/terms/specimen/identification/IdentifiedBy.java
@@ -8,7 +8,7 @@ public class IdentifiedBy extends Term {
 
   public static final String TERM = DWC_PREFIX + "identifiedBy";
 
-  private final List<String> dwcaTerms = List.of(TERM);
+  private final List<String> dwcaTerms = List.of(TERM, "dwc:identifiedByID");
 
   private final List<String> abcdTerms = List.of("identifiers/identifier/0/personName/fullName");
 

diff --git a/src/main/java/eu/dissco/core/translator/terms/specimen/identification/taxonomy/Genus.java b/src/main/java/eu/dissco/core/translator/terms/specimen/identification/taxonomy/Genus.java
@@ -9,7 +9,8 @@ public class Genus extends AbstractTaxonomy {
 
   private final List<String> dwcaTerms = List.of(TERM);
   private final List<String> abcdTerms = List.of(
-      "result/taxonIdentified/scientificName/nameAtomised/botanical/genusOrMonomial");
+      "result/taxonIdentified/scientificName/nameAtomised/botanical/genusOrMonomial",
+      "result/taxonIdentified/scientificName/nameAtomised/zoological/genusOrMonomial");
 
   @Override
   public String retrieveFromDWCA(JsonNode unit) {

diff --git a/...java/eu/dissco/core/translator/terms/specimen/identification/taxonomy/ScientificName.java b/...java/eu/dissco/core/translator/terms/specimen/identification/taxonomy/ScientificName.java
@@ -11,7 +11,9 @@ public class ScientificName extends AbstractTaxonomy {
   private final List<String> dwcaTerms = List.of("dwc:scientificName");
   private final List<String> abcdTerms = List.of(
       "result/taxonIdentified/scientificName/fullScientificNameString",
-      "result/mineralRockIdentified/classifiedName/fullScientificNameString");
+      "result/mineralRockIdentified/classifiedName/fullScientificNameString",
+      "result/taxonIdentified/informalNameString/value",
+      "result/taxonIdentified/scientificName/nameAtomised/zoological/namedIndividual");
 
   @Override
   public String retrieveFromDWCA(JsonNode unit) {

diff --git a/src/test/java/eu/dissco/core/translator/terms/specimen/BasisOfRecordTest.java b/src/test/java/eu/dissco/core/translator/terms/specimen/BasisOfRecordTest.java
@@ -3,15 +3,29 @@
 import static eu.dissco.core.translator.TestUtils.MAPPER;
 import static org.assertj.core.api.Assertions.assertThat;
 
+import java.util.stream.Stream;
 import org.junit.jupiter.api.Test;
 import org.junit.jupiter.api.extension.ExtendWith;
+import org.junit.jupiter.params.ParameterizedTest;
+import org.junit.jupiter.params.provider.Arguments;
+import org.junit.jupiter.params.provider.MethodSource;
 import org.mockito.junit.jupiter.MockitoExtension;
 
 @ExtendWith(MockitoExtension.class)
 class BasisOfRecordTest {
 
   private final BasisOfRecord basisOfRecord = new BasisOfRecord();
 
+  public static Stream<Arguments> prepareABCDRecordBasis() {
+    return Stream.of(
+        Arguments.of("PreservedSpecimen", "PreservedSpecimen"),
+        Arguments.of("RockSpecimen", "RockSpecimen"),
+        Arguments.of("HerbariumSheet", "Preserved specimen"),
+        Arguments.of("Dried", "Preserved specimen"),
+        Arguments.of(null, null)
+    );
+  }
+
   @Test
   void testRetrieveFromDWCA() {
     // Given
@@ -25,32 +39,21 @@ void testRetrieveFromDWCA() {
     assertThat(result).isEqualTo("LivingSpecimen");
   }
 
-  @Test
-  void testRetrieveFromDWCANotLiving() {
-    // Given
-    var unit = MAPPER.createObjectNode();
-    unit.put("dwc:basisOfRecord", "PreservedSpecimen");
-
-    // When
-    var result = basisOfRecord.retrieveFromDWCA(unit);
-
-    // Then
-    assertThat(result).isEqualTo("PreservedSpecimen");
-  }
-
-  @Test
-  void testRetrieveFromABCD() {
+  @ParameterizedTest
+  @MethodSource("prepareABCDRecordBasis")
+  void testRetrieveFromABCD(String recordBasis, String expected) {
     // Given
     var unit = MAPPER.createObjectNode();
-    unit.put("abcd:recordBasis", "RockSpecimen");
+    unit.put("abcd:recordBasis", recordBasis);
 
     // When
     var result = basisOfRecord.retrieveFromABCD(unit);
 
     // Then
-    assertThat(result).isEqualTo("RockSpecimen");
+    assertThat(result).isEqualTo(expected);
   }
 
+
   @Test
   void testGetTerm() {
     // When

diff --git a/...test/java/eu/dissco/core/translator/terms/specimen/identification/taxonomy/GenusTest.java b/...test/java/eu/dissco/core/translator/terms/specimen/identification/taxonomy/GenusTest.java
@@ -53,6 +53,21 @@ void testRetrieveFromABCD() {
     assertThat(result).isEqualTo("Arrabidaea");
   }
 
+  @Test
+  void testRetrieveFromABCDZoology() {
+    // Given
+    var unit = MAPPER.createObjectNode();
+    unit.put(
+        "result/taxonIdentified/scientificName/nameAtomised/zoological/genusOrMonomial",
+        "Alchemilla");
+
+    // When
+    var result = genus.retrieveFromABCD(unit);
+
+    // Then
+    assertThat(result).isEqualTo("Alchemilla");
+  }
+
   @Test
   void testGetTerm() {
     // When