Harmonise stratigraphy DD-179

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

@@ -2,17 +2,7 @@
 
 
 import com.fasterxml.jackson.databind.JsonNode;
-
 import efg.DataSets.DataSet;
-import eu.dissco.core.translator.terms.specimen.CollectingNumber;
-import eu.dissco.core.translator.terms.specimen.Collector;
-import eu.dissco.core.translator.terms.specimen.DatasetId;
-import eu.dissco.core.translator.terms.specimen.DateCollected;
-import eu.dissco.core.translator.terms.specimen.Modified;
-import eu.dissco.core.translator.terms.specimen.ObjectType;
-import eu.dissco.core.translator.terms.specimen.PhysicalSpecimenCollection;
-import eu.dissco.core.translator.terms.specimen.SpecimenName;
-import eu.dissco.core.translator.terms.specimen.TypeStatus;
 import java.util.List;
 import lombok.extern.slf4j.Slf4j;
 import org.gbif.dwc.ArchiveFile;
@@ -29,38 +19,56 @@ public String retrieveFromDWCA(ArchiveFile archiveFile, Record rec) {
     return "";
   }
 
-  protected String searchDWCAForTerm(ArchiveFile archiveFile, Record rec, List<String> originalTerms){
+  protected String searchDWCAForTerm(ArchiveFile archiveFile, Record rec,
+      List<String> originalTerms) {
     for (var originalTerm : originalTerms) {
-      if (archiveFile.getField(originalTerm) != null){
+      if (archiveFile.getField(originalTerm) != null) {
         var value = rec.value(archiveFile.getField(originalTerm).getTerm());
-        if (value != null){
+        if (value != null) {
           return value;
         }
       }
     }
-    log.info("Term not found in any of these search fields: {}", originalTerms);
+    log.debug("Term not found in any of these search fields: {}", originalTerms);
     return null;
   }
 
-  protected String searchAbcdForTerm(JsonNode attributes, List<String> originalTerms){
+  protected String searchAbcdForTerm(JsonNode attributes, List<String> originalTerms) {
     for (var originalTerm : originalTerms) {
-      if (attributes.get(originalTerm) != null){
+      if (attributes.get(originalTerm) != null) {
         return attributes.get(originalTerm).asText();
       }
     }
-    log.info("Term not found in any of these search fields: {}", originalTerms);
+    log.debug("Term not found in any of these search fields: {}", originalTerms);
     return null;
   }
 
+  protected String combineABCDTerms(JsonNode unit, List<String> abcdTerms) {
+    var builder = new StringBuilder();
+    for (var abcdTerm : abcdTerms) {
+      if (unit.get(abcdTerm) != null) {
+        if (builder.length() != 0) {
+          builder.append(" | ");
+        }
+        builder.append(unit.get(abcdTerm).asText());
+      }
+    }
+    if (builder.isEmpty()) {
+      return null;
+    } else {
+      return builder.toString();
+    }
+  }
+
   public abstract String getTerm();
 
   public String retrieveFromABCD(JsonNode unit) {
-    log.info("No specific attributes retrieve specified for field: {}", getTerm());
+    log.debug("No specific attributes retrieve specified for field: {}", getTerm());
     return null;
   }
 
   public String retrieveFromABCD(DataSet datasets) {
-    log.info("No specific attributes retrieve specified for field: {}", getTerm());
+    log.debug("No specific attributes retrieve specified for field: {}", getTerm());
     return null;
   }
 }

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

@@ -24,6 +24,22 @@
 import eu.dissco.core.translator.terms.specimen.location.Locality;
 import eu.dissco.core.translator.terms.specimen.location.StateProvince;
 import eu.dissco.core.translator.terms.specimen.location.WaterBody;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.biostratigraphic.HighestBiostratigraphicZone;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.biostratigraphic.LowestBiostratigraphicZone;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.chronostratigraphic.EarliestAgeOrLowestStage;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.chronostratigraphic.EarliestEonOrLowestEonothem;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.chronostratigraphic.EarliestEpochOrLowestSeries;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.chronostratigraphic.EarliestEraOrLowestErathem;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.chronostratigraphic.EarliestPeriodOrLowestSystem;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.chronostratigraphic.LatestAgeOrHighestStage;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.chronostratigraphic.LatestEonOrHighestEonothem;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.chronostratigraphic.LatestEpochOrHighestSeries;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.chronostratigraphic.LatestEraOrHighestErathem;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.chronostratigraphic.LatestPeriodOrHighestSystem;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.lithostratigraphic.Bed;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.lithostratigraphic.Formation;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.lithostratigraphic.Group;
+import eu.dissco.core.translator.terms.specimen.stratigraphy.lithostratigraphic.Member;
 import java.util.ArrayList;
 import java.util.List;
 import lombok.RequiredArgsConstructor;
@@ -56,6 +72,27 @@ public static List<Term> locationTerms() {
     return list;
   }
 
+  private static List<Term> stratigraphyTerms() {
+    var list = new ArrayList<Term>();
+    list.add(new EarliestAgeOrLowestStage());
+    list.add(new EarliestEonOrLowestEonothem());
+    list.add(new EarliestEpochOrLowestSeries());
+    list.add(new EarliestEraOrLowestErathem());
+    list.add(new EarliestPeriodOrLowestSystem());
+    list.add(new LatestAgeOrHighestStage());
+    list.add(new LatestEonOrHighestEonothem());
+    list.add(new LatestEpochOrHighestSeries());
+    list.add(new LatestEraOrHighestErathem());
+    list.add(new LatestPeriodOrHighestSystem());
+    list.add(new Bed());
+    list.add(new Formation());
+    list.add(new Group());
+    list.add(new Member());
+    list.add(new HighestBiostratigraphicZone());
+    list.add(new LowestBiostratigraphicZone());
+    return list;
+  }
+
   public static List<Term> harmonisedTerms() {
     var list = new ArrayList<Term>();
     list.add(new SpecimenName());
@@ -68,6 +105,7 @@ public static List<Term> harmonisedTerms() {
     list.add(new Collector());
     list.add(new TypeStatus());
     list.addAll(locationTerms());
+    list.addAll(stratigraphyTerms());
     return list;
   }
 

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

@@ -22,21 +22,20 @@ public String retrieveFromDWCA(ArchiveFile archiveFile, Record rec) {
 
   @Override
   public String retrieveFromABCD(JsonNode unit) {
-    StringBuilder builder = new StringBuilder();
-    combinePossibleValues(unit, builder);
-    if (builder.length() == 0){
+    var value = combinePossibleValues(unit);
+    if (value == null){
       if (unit.get("abcd:gathering/agents/gatheringAgentsText") != null){
-        builder.append(unit.get("abcd:gathering/agents/gatheringAgentsText").asText());
-        return builder.toString();
+        return unit.get("abcd:gathering/agents/gatheringAgentsText").asText();
       } else {
         return null;
       }
     } else {
-      return builder.toString();
+      return value;
     }
   }
 
-  private void combinePossibleValues(JsonNode unit, StringBuilder builder) {
+  private String combinePossibleValues(JsonNode unit) {
+    var builder = new StringBuilder();
     var iterateOverElements = true;
     var numberFound = 0;
     while (iterateOverElements) {
@@ -59,6 +58,11 @@ private void combinePossibleValues(JsonNode unit, StringBuilder builder) {
         iterateOverElements = false;
       }
     }
+    if (builder.length() != 0){
+      return builder.toString();
+    } else {
+      return null;
+    }
   }
 
 

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

@@ -13,15 +13,12 @@ public class TypeStatus extends Term {
   private final List<String> dwcaTerms = List.of(TERM);
 
   // Pick the first TypeStatus from ABCD
-  private final List<String> abcdTermsTypeStatus =
+  private final List<String> abcdTerms =
       List.of(
-          "abcd:specimenUnit/nomenclaturalTypeDesignations/nomenclaturalTypeDesignation/0/typeStatus");
-  private final List<String> abcdTermsTypeName =
-      List.of(
-          "abcd:specimenUnit/nomenclaturalTypeDesignations/nomenclaturalTypeDesignation/0/typifiedName/fullScientificNameString");
-  private final List<String> abcdTermsCitation =
-      List.of(
-          "abcd:specimenUnit/nomenclaturalTypeDesignations/nomenclaturalTypeDesignation/0/nomenclaturalReference/titleCitation");
+          "abcd:specimenUnit/nomenclaturalTypeDesignations/nomenclaturalTypeDesignation/0/typeStatus",
+          "abcd:specimenUnit/nomenclaturalTypeDesignations/nomenclaturalTypeDesignation/0/typifiedName/fullScientificNameString",
+          "abcd:specimenUnit/nomenclaturalTypeDesignations/nomenclaturalTypeDesignation/0/nomenclaturalReference/titleCitation"
+          );
 
   @Override
   public String retrieveFromDWCA(ArchiveFile archiveFile, Record rec) {
@@ -30,10 +27,7 @@ public String retrieveFromDWCA(ArchiveFile archiveFile, Record rec) {
 
   @Override
   public String retrieveFromABCD(JsonNode unit) {
-    var status = super.searchAbcdForTerm(unit, abcdTermsTypeStatus);
-    var typeName = super.searchAbcdForTerm(unit, abcdTermsTypeName);
-    var citation = super.searchAbcdForTerm(unit, abcdTermsCitation);
-    return status + " | " + typeName + " | " + citation;
+    return super.combineABCDTerms(unit, abcdTerms);
   }
 
   @Override

src/main/java/eu/dissco/core/translator/terms/specimen/stratigraphy/biostratigraphic/HighestBiostratigraphicZone.java

@@ -0,0 +1,34 @@
+package eu.dissco.core.translator.terms.specimen.stratigraphy.biostratigraphic;
+
+import com.fasterxml.jackson.databind.JsonNode;
+import eu.dissco.core.translator.terms.Term;
+import java.util.List;
+import org.gbif.dwc.ArchiveFile;
+import org.gbif.dwc.record.Record;
+
+public class HighestBiostratigraphicZone extends Term {
+
+  public static final String TERM = DWC_PREFIX + "highestBiostratigraphicZone";
+
+  private final List<String> dwcaTerms = List.of(TERM);
+  private final List<String> abcdTerms = List.of(
+      "abcd-efg:earthScienceSpecimen/unitStratigraphicDetermination/biostratigraphicAttributionsType/biostratigraphicAttribution/0/zonalFossilType",
+      "abcd-efg:earthScienceSpecimen/unitStratigraphicDetermination/biostratigraphicAttributionsType/biostratigraphicAttribution/0/fossilZoneName",
+      "abcd-efg:earthScienceSpecimen/unitStratigraphicDetermination/biostratigraphicAttributionsType/biostratigraphicAttribution/0/fossilSubzoneName");
+
+  @Override
+  public String retrieveFromDWCA(ArchiveFile archiveFile, Record rec) {
+    return super.searchDWCAForTerm(archiveFile, rec, dwcaTerms);
+  }
+
+  @Override
+  public String retrieveFromABCD(JsonNode unit) {
+    return super.combineABCDTerms(unit, abcdTerms);
+  }
+
+  @Override
+  public String getTerm() {
+    return TERM;
+  }
+
+}

src/main/java/eu/dissco/core/translator/terms/specimen/stratigraphy/biostratigraphic/LowestBiostratigraphicZone.java

@@ -0,0 +1,34 @@
+package eu.dissco.core.translator.terms.specimen.stratigraphy.biostratigraphic;
+
+import com.fasterxml.jackson.databind.JsonNode;
+import eu.dissco.core.translator.terms.Term;
+import java.util.List;
+import org.gbif.dwc.ArchiveFile;
+import org.gbif.dwc.record.Record;
+
+public class LowestBiostratigraphicZone extends Term {
+
+  public static final String TERM = DWC_PREFIX + "lowestBiostratigraphicZone";
+
+  private final List<String> dwcaTerms = List.of(TERM);
+  private final List<String> abcdTerms = List.of(
+      "abcd-efg:earthScienceSpecimen/unitStratigraphicDetermination/biostratigraphicAttributionsType/biostratigraphicAttribution/0/zonalFossilType",
+      "abcd-efg:earthScienceSpecimen/unitStratigraphicDetermination/biostratigraphicAttributionsType/biostratigraphicAttribution/0/fossilZoneName",
+      "abcd-efg:earthScienceSpecimen/unitStratigraphicDetermination/biostratigraphicAttributionsType/biostratigraphicAttribution/0/fossilSubzoneName");
+
+  @Override
+  public String retrieveFromDWCA(ArchiveFile archiveFile, Record rec) {
+    return super.searchDWCAForTerm(archiveFile, rec, dwcaTerms);
+  }
+
+  @Override
+  public String retrieveFromABCD(JsonNode unit) {
+    return super.combineABCDTerms(unit, abcdTerms);
+  }
+
+  @Override
+  public String getTerm() {
+    return TERM;
+  }
+
+}

src/main/java/eu/dissco/core/translator/terms/specimen/stratigraphy/chronostratigraphic/AbstractChronoStratigraphy.java

@@ -0,0 +1,38 @@
+package eu.dissco.core.translator.terms.specimen.stratigraphy.chronostratigraphic;
+
+import com.fasterxml.jackson.databind.JsonNode;
+import eu.dissco.core.translator.terms.Term;
+import java.util.List;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+public abstract class AbstractChronoStratigraphy extends Term {
+
+  private static final String ABCD_DIVISION =
+      "abcd-efg:earthScienceSpecimen/unitStratigraphicDetermination/chronostratigraphicAttributions/chronostratigraphicAttribution/%s/chronoStratigraphicDivision";
+  private static final String ABCD_VALUE =
+      "abcd-efg:earthScienceSpecimen/unitStratigraphicDetermination/chronostratigraphicAttributions/chronostratigraphicAttribution/%s/chronostratigraphicName";
+
+  protected String searchABCDChronostratigraphy(JsonNode unit, List<String> divisionSearched) {
+    for (var divisionSearch : divisionSearched) {
+      var iterateOverElements = true;
+      var numberFound = 0;
+      while (iterateOverElements) {
+        var divisionNode = unit.get(String.format(ABCD_DIVISION, numberFound));
+        if (divisionNode != null) {
+          var division = divisionNode.asText();
+          if (division.equalsIgnoreCase(divisionSearch)
+              && unit.get(String.format(ABCD_VALUE, numberFound)) != null) {
+            return unit.get(String.format(ABCD_VALUE, numberFound)).asText();
+          }
+          numberFound++;
+        } else {
+          iterateOverElements = false;
+        }
+      }
+    }
+    log.debug("No stratigraphy found for division: {}", divisionSearched);
+    return null;
+  }
+
+}

src/main/java/eu/dissco/core/translator/terms/specimen/stratigraphy/chronostratigraphic/EarliestAgeOrLowestStage.java

@@ -0,0 +1,28 @@
+package eu.dissco.core.translator.terms.specimen.stratigraphy.chronostratigraphic;
+
+import com.fasterxml.jackson.databind.JsonNode;
+import java.util.List;
+import org.gbif.dwc.ArchiveFile;
+import org.gbif.dwc.record.Record;
+
+public class EarliestAgeOrLowestStage extends AbstractChronoStratigraphy {
+
+  public static final String TERM = DWC_PREFIX + "earliestAgeOrLowestStage";
+  private final List<String> dwcaTerms = List.of(TERM);
+
+
+  @Override
+  public String retrieveFromDWCA(ArchiveFile archiveFile, Record rec) {
+    return super.searchDWCAForTerm(archiveFile, rec, dwcaTerms);
+  }
+
+  @Override
+  public String retrieveFromABCD(JsonNode unit) {
+    return super.searchABCDChronostratigraphy(unit, List.of("SubStage", "Stage"));
+  }
+
+  @Override
+  public String getTerm() {
+    return TERM;
+  }
+}

src/main/java/eu/dissco/core/translator/terms/specimen/stratigraphy/chronostratigraphic/EarliestEonOrLowestEonothem.java

@@ -0,0 +1,29 @@
+package eu.dissco.core.translator.terms.specimen.stratigraphy.chronostratigraphic;
+
+import com.fasterxml.jackson.databind.JsonNode;
+import java.util.List;
+import org.gbif.dwc.ArchiveFile;
+import org.gbif.dwc.record.Record;
+
+public class EarliestEonOrLowestEonothem extends AbstractChronoStratigraphy {
+
+  public static final String TERM = DWC_PREFIX + "earliestEonOrLowestEonothem";
+  private final List<String> dwcaTerms = List.of(TERM);
+
+
+  @Override
+  public String retrieveFromDWCA(ArchiveFile archiveFile, Record rec) {
+    return super.searchDWCAForTerm(archiveFile, rec, dwcaTerms);
+  }
+
+  @Override
+  public String retrieveFromABCD(JsonNode unit) {
+    return super.searchABCDChronostratigraphy(unit, List.of("Eonothem"));
+  }
+
+  @Override
+  public String getTerm() {
+    return TERM;
+  }
+
+}