Merge branch 'main' into feature/remove-empty-objects

DiSSCo · Nov 25, 2024 · b36d9d4 · b36d9d4
2 parents cde73bb + ece746a
commit b36d9d4
Show file tree

Hide file tree

Showing 15 changed files with 332 additions and 80 deletions.
diff --git a/src/main/java/eu/dissco/core/translator/terms/BaseDigitalObjectDirector.java b/src/main/java/eu/dissco/core/translator/terms/BaseDigitalObjectDirector.java
@@ -194,6 +194,7 @@
 import eu.dissco.core.translator.terms.specimen.identification.taxonomy.Tribe;
 import eu.dissco.core.translator.terms.specimen.identification.taxonomy.VerbatimTaxonRank;
 import eu.dissco.core.translator.terms.specimen.identification.taxonomy.VernacularName;
+import eu.dissco.core.translator.terms.specimen.location.AbstractMeterTerm;
 import eu.dissco.core.translator.terms.specimen.location.Continent;
 import eu.dissco.core.translator.terms.specimen.location.Country;
 import eu.dissco.core.translator.terms.specimen.location.CountryCode;
@@ -279,6 +280,17 @@ public abstract class BaseDigitalObjectDirector {
   private final FdoProperties fdoProperties;
   private final List<String> identifierTerms;
 
+  private static Double stringToDouble(Term term, String value) {
+    try {
+      if (value != null) {
+        return Double.valueOf(value);
+      }
+    } catch (NumberFormatException ex) {
+      log.warn("Unable to parse value: {} to a double for term: {}", value, term.getTerm());
+    }
+    return null;
+  }
+
   public DigitalSpecimen assembleDigitalSpecimenTerm(JsonNode data, boolean dwc)
       throws OrganisationException, UnknownPhysicalSpecimenIdType {
     var ds = assembleDigitalSpecimenTerms(data, dwc);
@@ -524,7 +536,6 @@ protected Identification createIdentification(JsonNode data, boolean dwc) {
     return identification;
   }
 
-
   private List<Event> assembleEventTerms(JsonNode data, boolean dwc) {
     var geoReference = new Georeference()
         .withType("ods:Georeference")
@@ -600,29 +611,27 @@ private List<Event> assembleEventTerms(JsonNode data, boolean dwc) {
         .withDwcStateProvince(termMapper.retrieveTerm(new StateProvince(), data, dwc))
         .withDwcWaterBody(termMapper.retrieveTerm(new WaterBody(), data, dwc))
         .withDwcHigherGeography(termMapper.retrieveTerm(new HigherGeography(), data, dwc))
-        .withDwcMaximumDepthInMeters(parseToDouble(new MaximumDepthInMeters(), data, dwc))
         .withDwcMaximumDistanceAboveSurfaceInMeters(
             parseToDouble(new MaximumDistanceAboveSurfaceInMeters(), data, dwc))
-        .withDwcMaximumElevationInMeters(parseToDouble(new MaximumElevationInMeters(), data, dwc))
-        .withDwcMinimumDepthInMeters(parseToDouble(new MinimumDepthInMeters(), data, dwc))
         .withDwcMinimumDistanceAboveSurfaceInMeters(
             parseToDouble(new MinimumDistanceAboveSurfaceInMeters(), data, dwc))
-        .withDwcMinimumElevationInMeters(parseToDouble(new MinimumElevationInMeters(), data, dwc))
         .withDwcVerbatimDepth(termMapper.retrieveTerm(new VerbatimDepth(), data, dwc))
         .withDwcVerbatimElevation(termMapper.retrieveTerm(new VerbatimElevation(), data, dwc))
         .withDwcVerticalDatum(termMapper.retrieveTerm(new VerticalDatum(), data, dwc))
         .withDwcLocationAccordingTo(termMapper.retrieveTerm(new LocationAccordingTo(), data, dwc))
-        .withDwcLocationRemarks(termMapper.retrieveTerm(new LocationRemarks(), data, dwc));
+        .withDwcLocationRemarks(termMapper.retrieveTerm(new LocationRemarks(), data, dwc))
+        .withOdsHasGeoreference(geoReference)
+        .withOdsHasGeologicalContext(geologicalContext);
+    setMinMaxMeterField(new MinimumElevationInMeters(), location, data, dwc);
+    setMinMaxMeterField(new MaximumElevationInMeters(), location, data, dwc);
+    setMinMaxMeterField(new MinimumDepthInMeters(), location, data, dwc);
+    setMinMaxMeterField(new MaximumDepthInMeters(), location, data, dwc);
     if (!Objects.equals(geoReference, EMPTY_GEOREFERENCE)) {
       location.setOdsHasGeoreference(geoReference);
     }
     if (!Objects.equals(geologicalContext, EMPTY_GEOLOGICAL_CONTEXT)) {
       location.setOdsHasGeologicalContext(geologicalContext);
     }
-    setMinMaxMeterField(new MinimumElevationInMeters(), location, data, dwc);
-    setMinMaxMeterField(new MaximumElevationInMeters(), location, data, dwc);
-    setMinMaxMeterField(new MinimumDepthInMeters(), location, data, dwc);
-    setMinMaxMeterField(new MaximumDepthInMeters(), location, data, dwc);
     var assertions = new EventAssertions().gatherEventAssertions(mapper, data, dwc);
     var event = new Event()
         .withType("ods:Event")
@@ -664,6 +673,31 @@ private List<Event> assembleEventTerms(JsonNode data, boolean dwc) {
     return List.of(event);
   }
 
+  private void setMinMaxMeterField(AbstractMeterTerm term, Location location, JsonNode data,
+      boolean dwc) {
+    var sanitizedValue = termMapper.retrieveTerm(term, data, dwc);
+    var doubleElevation = stringToDouble(term, sanitizedValue);
+    if (doubleElevation == null && sanitizedValue != null) {
+      if (term instanceof MaximumElevationInMeters) {
+        location.setDwcVerbatimElevation(
+            location.getDwcVerbatimElevation() + "| unparseable value maximumElevationInMeters: "
+                + sanitizedValue);
+      } else if (term instanceof MinimumElevationInMeters) {
+        location.setDwcVerbatimElevation(
+            location.getDwcVerbatimElevation() + "| unparseable value minimumElevationInMeters: "
+                + sanitizedValue);
+      } else if (term instanceof MaximumDepthInMeters) {
+        location.setDwcVerbatimDepth(
+            location.getDwcVerbatimDepth() + "| unparseable value maximumDepthInMeters: "
+                + sanitizedValue);
+      } else if (term instanceof MinimumDepthInMeters) {
+        location.setDwcVerbatimDepth(
+            location.getDwcVerbatimDepth() + "| unparseable value minimumDepthInMeters: "
+                + sanitizedValue);
+      }
+    }
+  }
+
   private <T extends Enum<T>> T retrieveEnum(Term term, JsonNode data, boolean dwc,
       java.lang.Class<T> enumClass) {
     var value = termMapper.retrieveTerm(term, data, dwc);
@@ -691,14 +725,7 @@ private Integer parseToInteger(Term term, JsonNode data, boolean dwc) {
 
   private Double parseToDouble(Term term, JsonNode data, boolean dwc) {
     var value = termMapper.retrieveTerm(term, data, dwc);
-    try {
-      if (value != null) {
-        return Double.valueOf(value);
-      }
-    } catch (NumberFormatException ex) {
-      log.warn("Unable to parse value: {} to a double for term: {}", value, term.getTerm());
-    }
-    return null;
+    return stringToDouble(term, value);
   }
 
   private Boolean parseToBoolean(Term term, JsonNode data, boolean dwc) {

diff --git a/src/main/java/eu/dissco/core/translator/terms/specimen/location/AbstractMeterTerm.java b/src/main/java/eu/dissco/core/translator/terms/specimen/location/AbstractMeterTerm.java
@@ -0,0 +1,28 @@
+package eu.dissco.core.translator.terms.specimen.location;
+
+import eu.dissco.core.translator.terms.Term;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+import lombok.extern.slf4j.Slf4j;
+
+@Slf4j
+public abstract class AbstractMeterTerm extends Term {
+
+  private static final Pattern M_PATTERN = Pattern.compile("((-\\s?)?\\d+([.,]\\d+)?)\\s*m\\.?(eter)?(tr?.?)?(\\sm.?)?");
+
+  protected String sanitizeInput(String input) {
+    if (input == null) {
+      return null;
+    }
+    input = input.trim().toLowerCase();
+    Matcher matcher = M_PATTERN.matcher(input);
+    if (matcher.matches()) {
+      log.debug("Parsed Number: {}", matcher.group(1));
+      return matcher.group(1);
+    } else {
+      log.debug("Input string does not match the expected format: {}", input);
+      return input;
+    }
+  }
+
+}
diff --git a/src/main/java/eu/dissco/core/translator/terms/specimen/location/MaximumDepthInMeters.java b/src/main/java/eu/dissco/core/translator/terms/specimen/location/MaximumDepthInMeters.java
@@ -1,10 +1,9 @@
 package eu.dissco.core.translator.terms.specimen.location;
 
 import com.fasterxml.jackson.databind.JsonNode;
-import eu.dissco.core.translator.terms.Term;
 import java.util.List;
 
-public class MaximumDepthInMeters extends Term {
+public class MaximumDepthInMeters extends AbstractMeterTerm {
 
   public static final String TERM = DWC_PREFIX + "maximumDepthInMeters";
 
@@ -15,12 +14,14 @@ public class MaximumDepthInMeters extends Term {
 
   @Override
   public String retrieveFromDWCA(JsonNode unit) {
-    return super.searchJsonForTerm(unit, dwcaTerms);
+    var rawResult = super.searchJsonForTerm(unit, dwcaTerms);
+    return sanitizeInput(rawResult);
   }
 
   @Override
   public String retrieveFromABCD(JsonNode unit) {
-    return super.searchJsonForTerm(unit, abcdTerms);
+    var rawResult = super.searchJsonForTerm(unit, abcdTerms);
+    return sanitizeInput(rawResult);
   }
 
   @Override

diff --git a/...u/dissco/core/translator/terms/specimen/location/MaximumDistanceAboveSurfaceInMeters.java b/...u/dissco/core/translator/terms/specimen/location/MaximumDistanceAboveSurfaceInMeters.java
@@ -1,10 +1,9 @@
 package eu.dissco.core.translator.terms.specimen.location;
 
 import com.fasterxml.jackson.databind.JsonNode;
-import eu.dissco.core.translator.terms.Term;
 import java.util.List;
 
-public class MaximumDistanceAboveSurfaceInMeters extends Term {
+public class MaximumDistanceAboveSurfaceInMeters extends AbstractMeterTerm {
 
   public static final String TERM = DWC_PREFIX + "maximumDistanceAboveSurfaceInMeters";
 
@@ -14,12 +13,14 @@ public class MaximumDistanceAboveSurfaceInMeters extends Term {
 
   @Override
   public String retrieveFromDWCA(JsonNode unit) {
-    return super.searchJsonForTerm(unit, dwcaTerms);
+    var rawResult = super.searchJsonForTerm(unit, dwcaTerms);
+    return sanitizeInput(rawResult);
   }
 
   @Override
   public String retrieveFromABCD(JsonNode unit) {
-    return super.searchJsonForTerm(unit, abcdTerms);
+    var rawResult = super.searchJsonForTerm(unit, abcdTerms);
+    return sanitizeInput(rawResult);
   }
 
   @Override

diff --git a/...main/java/eu/dissco/core/translator/terms/specimen/location/MaximumElevationInMeters.java b/...main/java/eu/dissco/core/translator/terms/specimen/location/MaximumElevationInMeters.java
@@ -1,10 +1,9 @@
 package eu.dissco.core.translator.terms.specimen.location;
 
 import com.fasterxml.jackson.databind.JsonNode;
-import eu.dissco.core.translator.terms.Term;
 import java.util.List;
 
-public class MaximumElevationInMeters extends Term {
+public class MaximumElevationInMeters extends AbstractMeterTerm {
 
   public static final String TERM = DWC_PREFIX + "maximumElevationInMeters";
 
@@ -15,12 +14,14 @@ public class MaximumElevationInMeters extends Term {
 
   @Override
   public String retrieveFromDWCA(JsonNode unit) {
-    return super.searchJsonForTerm(unit, dwcaTerms);
+    var rawResult = super.searchJsonForTerm(unit, dwcaTerms);
+    return sanitizeInput(rawResult);
   }
 
   @Override
   public String retrieveFromABCD(JsonNode unit) {
-    return super.searchJsonForTerm(unit, abcdTerms);
+    var rawResult = super.searchJsonForTerm(unit, abcdTerms);
+    return sanitizeInput(rawResult);
   }
 
   @Override

diff --git a/src/main/java/eu/dissco/core/translator/terms/specimen/location/MinimumDepthInMeters.java b/src/main/java/eu/dissco/core/translator/terms/specimen/location/MinimumDepthInMeters.java
@@ -1,10 +1,9 @@
 package eu.dissco.core.translator.terms.specimen.location;
 
 import com.fasterxml.jackson.databind.JsonNode;
-import eu.dissco.core.translator.terms.Term;
 import java.util.List;
 
-public class MinimumDepthInMeters extends Term {
+public class MinimumDepthInMeters extends AbstractMeterTerm {
 
   public static final String TERM = DWC_PREFIX + "minimumDepthInMeters";
 
@@ -15,12 +14,14 @@ public class MinimumDepthInMeters extends Term {
 
   @Override
   public String retrieveFromDWCA(JsonNode unit) {
-    return super.searchJsonForTerm(unit, dwcaTerms);
+    var rawResult = super.searchJsonForTerm(unit, dwcaTerms);
+    return sanitizeInput(rawResult);
   }
 
   @Override
   public String retrieveFromABCD(JsonNode unit) {
-    return super.searchJsonForTerm(unit, abcdTerms);
+    var rawResult = super.searchJsonForTerm(unit, abcdTerms);
+    return sanitizeInput(rawResult);
   }
 
   @Override

diff --git a/...u/dissco/core/translator/terms/specimen/location/MinimumDistanceAboveSurfaceInMeters.java b/...u/dissco/core/translator/terms/specimen/location/MinimumDistanceAboveSurfaceInMeters.java
@@ -1,10 +1,9 @@
 package eu.dissco.core.translator.terms.specimen.location;
 
 import com.fasterxml.jackson.databind.JsonNode;
-import eu.dissco.core.translator.terms.Term;
 import java.util.List;
 
-public class MinimumDistanceAboveSurfaceInMeters extends Term {
+public class MinimumDistanceAboveSurfaceInMeters extends AbstractMeterTerm {
 
   public static final String TERM = DWC_PREFIX + "minimumDistanceAboveSurfaceInMeters";
 
@@ -14,12 +13,14 @@ public class MinimumDistanceAboveSurfaceInMeters extends Term {
 
   @Override
   public String retrieveFromDWCA(JsonNode unit) {
-    return super.searchJsonForTerm(unit, dwcaTerms);
+    var rawResult = super.searchJsonForTerm(unit, dwcaTerms);
+    return sanitizeInput(rawResult);
   }
 
   @Override
   public String retrieveFromABCD(JsonNode unit) {
-    return super.searchJsonForTerm(unit, abcdTerms);
+    var rawResult = super.searchJsonForTerm(unit, abcdTerms);
+    return sanitizeInput(rawResult);
   }
 
   @Override

diff --git a/...main/java/eu/dissco/core/translator/terms/specimen/location/MinimumElevationInMeters.java b/...main/java/eu/dissco/core/translator/terms/specimen/location/MinimumElevationInMeters.java
@@ -1,10 +1,9 @@
 package eu.dissco.core.translator.terms.specimen.location;
 
 import com.fasterxml.jackson.databind.JsonNode;
-import eu.dissco.core.translator.terms.Term;
 import java.util.List;
 
-public class MinimumElevationInMeters extends Term {
+public class MinimumElevationInMeters extends AbstractMeterTerm {
 
   public static final String TERM = DWC_PREFIX + "minimumElevationInMeters";
 
@@ -15,12 +14,14 @@ public class MinimumElevationInMeters extends Term {
 
   @Override
   public String retrieveFromDWCA(JsonNode unit) {
-    return super.searchJsonForTerm(unit, dwcaTerms);
+    var rawResult = super.searchJsonForTerm(unit, dwcaTerms);
+    return sanitizeInput(rawResult);
   }
 
   @Override
   public String retrieveFromABCD(JsonNode unit) {
-    return super.searchJsonForTerm(unit, abcdTerms);
+    var rawResult = super.searchJsonForTerm(unit, abcdTerms);
+    return sanitizeInput(rawResult);
   }
 
   @Override

diff --git a/src/main/java/eu/dissco/core/translator/terms/utils/IdentifierUtils.java b/src/main/java/eu/dissco/core/translator/terms/utils/IdentifierUtils.java
@@ -1,6 +1,11 @@
 package eu.dissco.core.translator.terms.utils;
 
-import static eu.dissco.core.translator.schema.Identifier.DctermsType.*;
+import static eu.dissco.core.translator.schema.Identifier.DctermsType.ARK;
+import static eu.dissco.core.translator.schema.Identifier.DctermsType.DOI;
+import static eu.dissco.core.translator.schema.Identifier.DctermsType.HANDLE;
+import static eu.dissco.core.translator.schema.Identifier.DctermsType.PURL;
+import static eu.dissco.core.translator.schema.Identifier.DctermsType.URL;
+import static eu.dissco.core.translator.schema.Identifier.DctermsType.UUID;
 import static eu.dissco.core.translator.schema.Identifier.OdsGupriLevel.GLOBALLY_UNIQUE_STABLE;
 import static eu.dissco.core.translator.schema.Identifier.OdsGupriLevel.GLOBALLY_UNIQUE_STABLE_PERSISTENT_RESOLVABLE;
 import static eu.dissco.core.translator.schema.Identifier.OdsGupriLevel.GLOBALLY_UNIQUE_STABLE_PERSISTENT_RESOLVABLE_FDO_COMPLIANT;