Fix for CatalogueOfLife/data#722

Fixes for ITIS id matching and removal of wildcard prefix Datasets config Fix for parameter ordering Removing the call to matchByUsageKey from matchByExternalKey - which was bypassing prefix matching index metadata generation fix for docker Include breakdown for all ranks IUCN #1350 Include atomised name to avoid pipelines and other services reparsing the name #1350
CatalogueOfLife · Nov 4, 2024 · 6db3805 · 6db3805
1 parent 5aa3e90
commit 6db3805
Show file tree

Hide file tree

Showing 13 changed files with 501 additions and 158 deletions.
diff --git a/matching-ws/src/main/java/life/catalogue/matching/MatchingApplication.java b/matching-ws/src/main/java/life/catalogue/matching/MatchingApplication.java
@@ -87,7 +87,7 @@ public void run(ApplicationArguments args) {
   }
 
   private void initialiseWebapp() {
-    Optional<APIMetadata> metadata = matchingService.getAPIMetadata(false);
+    Optional<APIMetadata> metadata = matchingService.getAPIMetadata(true);
     if (metadata.isEmpty()) {
       log.error("No main index found. Cannot start web services");
       return;
@@ -136,7 +136,8 @@ private void runIndexingIfRequired(ApplicationArguments args) throws Exception {
         indexingService.indexIdentifiers(id);
     }
 
-    log.info("Indexing completed");
+    matchingService.getAPIMetadata(true);
+    log.info("Indexing ready");
   }
 
   private ExecutionMode getMode(ApplicationArguments args) {

diff --git a/matching-ws/src/main/java/life/catalogue/matching/controller/MatchController.java b/matching-ws/src/main/java/life/catalogue/matching/controller/MatchController.java
@@ -15,6 +15,7 @@
 import java.util.stream.Collectors;
 import life.catalogue.matching.model.*;
 import life.catalogue.matching.service.MatchingService;
+import life.catalogue.matching.util.IUCNUtils;
 import lombok.AllArgsConstructor;
 import lombok.Builder;
 import lombok.Data;
@@ -112,9 +113,13 @@ public NameUsageMatch matchOldPaths(
     HttpServletRequest response) {
     return matchV2(
       usageKey,
-      taxonID,taxonConceptID,scientificNameID,
-      scientificName2, scientificName,
-      authorship, authorship2,
+      taxonID,
+      taxonConceptID,
+      scientificNameID,
+      scientificName2,
+      scientificName,
+      authorship,
+      authorship2,
       genericName,
       specificEpithet,
       infraspecificEpithet,
@@ -677,15 +682,15 @@ public Map<String, Object> iucnRedListV1(@PathVariable(value = "usageKey", requi
       return Map.of();
     }
     NameUsageMatch.Status status = statusList.get(0);
-    String formatted = formatIucn(status.getStatus());
+    String formatted = IUCNUtils.formatIucn(status.getStatus());
     if (formatted == null || formatted.isEmpty()) {
       return Map.of();
     }
 
     String scientificName = match.getAcceptedUsage() != null ? match.getAcceptedUsage().getCanonicalName() : match.getUsage().getCanonicalName();
 
     try {
-      IUCN iucn = IUCN.valueOf(formatted); // throws IllegalArgumentException if not found
+      IUCNUtils.IUCN iucn =  IUCNUtils.IUCN.valueOf(formatted); // throws IllegalArgumentException if not found
       watch.stop();
       log("v1/species/iucnRedListCategory", usageKey, watch);
       return Map.of(
@@ -695,7 +700,7 @@ public Map<String, Object> iucnRedListV1(@PathVariable(value = "usageKey", requi
         "taxonomicStatus", NameUsageMatchV1.TaxonomicStatusV1.convert(
           match.getDiagnostics().getStatus()),
         "iucnTaxonID", status.getSourceId(),
-        "code", iucn.code
+        "code", iucn.getCode()
       );
     } catch (IllegalArgumentException e) {
       log.error("IUCN category not found: {}", formatted, e);
@@ -757,37 +762,6 @@ private static void addIfNotNull(StringJoiner joiner, Object value) {
     }
   }
 
-  String formatIucn(String original){
-    if (original == null) {
-      return null;
-    }
-    // Trim the string
-    String trimmed = original.trim();
-    // Convert to uppercase
-    String uppercased = trimmed.toUpperCase();
-    // Replace any whitespace with a single underscore
-    return uppercased.replaceAll("\\s+", "_");
-  }
-
-   enum IUCN {
-    EXTINCT("EX"),
-    EXTINCT_IN_THE_WILD("EW"),
-    CRITICALLY_ENDANGERED ("CR"),
-    ENDANGERED ("EN"),
-    VULNERABLE ("VU"),
-    NEAR_THREATENED ("NT"),
-    CONSERVATION_DEPENDENT ("CD"),
-    LEAST_CONCERN ("LC"),
-    DATA_DEFICIENT ("DD"),
-    NOT_EVALUATED ("NE");
-
-    private final String code;
-
-    IUCN(String code) {
-      this.code = code;
-    }
-  }
-
   @Data
   @Builder
   @NoArgsConstructor