move shortform to the linker

EBISPOT · Nov 8, 2024 · c58e873 · c58e873
1 parent f026c2b
commit c58e873
Show file tree

Hide file tree

Showing 4 changed files with 25 additions and 51 deletions.
diff --git a/dataload/linker/src/main/java/LinkerPass2.java b/dataload/linker/src/main/java/LinkerPass2.java
@@ -146,6 +146,14 @@ private static void writeEntityArray(JsonReader jsonReader, JsonWriter jsonWrite
 
             EntityDefinitionSet defOfThisEntity = pass1Result.iriToDefinitions.get(entityIri);
 
+            String curie = null;
+            if(defOfThisEntity.definingDefinitions.size() > 0) {
+                // always use the defining ontology's curie, as the defining
+                // ontology knows the base URI and we might not
+                //
+                curie = defOfThisEntity.definingDefinitions.iterator().next().curie.getAsString();
+            }
+
             while(jsonReader.peek() != JsonToken.END_OBJECT) {
 
                 String name = jsonReader.nextName();
@@ -159,22 +167,23 @@ private static void writeEntityArray(JsonReader jsonReader, JsonWriter jsonWrite
                 }
 
                 if(name == "curie") {
-                    if(defOfThisEntity.definingDefinitions.size() > 0) {
-                        // always use the defining ontology's curie, as the defining
-                        // ontology knows the base URI and we might not
-                        //
-                        com.google.gson.internal.Streams.write(
-                        defOfThisEntity.definingDefinitions.iterator().next().curie,
-                        jsonWriter);
-                        continue;
+                    if(curie != null) {
+                        // use the defining ontology curie
+                        jsonWriter.value(curie);
                     } else {
-                        // fallback to using the curie we already have
+                        // fallthrough to using the curie from rdf2json
+                        curie = jsonReader.nextString();
+                        jsonWriter.value(curie);
                     }
+                    continue;
                 }
 
                 CopyJsonGatheringStrings.copyJsonGatheringStrings(jsonReader, jsonWriter, stringsInEntity);
             }
 
+            jsonWriter.name("shortForm");
+            jsonWriter.value(curie.replaceFirst(":", "_"));
+
             if(defOfThisEntity != null) {
 
                 jsonWriter.name(IS_DEFINING_ONTOLOGY.getText());
@@ -452,39 +461,4 @@ private static class CurieMapResult {
         public String url;
         public String source;
     }
-
-    private static void processShortFormObject(JsonReader jsonReader, JsonWriter jsonWriter, LinkerPass1.LinkerPass1Result pass1Result, String entityIri) throws IOException {
-        jsonReader.beginObject();
-        JsonObject shortFormObject = new JsonObject();
-
-        while (jsonReader.peek() != JsonToken.END_OBJECT) {
-            String shortFormFieldName = jsonReader.nextName();
-            if (shortFormFieldName.equals("type")) {
-                JsonArray typeArray = new JsonArray();
-                jsonReader.beginArray();
-                while (jsonReader.peek() != JsonToken.END_ARRAY) {
-                    typeArray.add(jsonReader.nextString());
-                }
-                jsonReader.endArray();
-                shortFormObject.add("type", typeArray);
-            } else if (shortFormFieldName.equals("value")) {
-                String shortFormValue = jsonReader.nextString();
-                // Modify the value attribute
-                shortFormValue = getProcessedCurieValue(pass1Result, entityIri).replace(":", "_");
-                shortFormObject.addProperty("value", shortFormValue);
-            }
-        }
-        jsonReader.endObject();
-
-        // Write the modified short form object
-        jsonWriter.beginObject();
-        jsonWriter.name("type");
-        jsonWriter.beginArray();
-        for (JsonElement typeElement : shortFormObject.getAsJsonArray("type")) {
-            jsonWriter.value(typeElement.getAsString());
-        }
-        jsonWriter.endArray();
-        jsonWriter.name("value").value(shortFormObject.get("value").getAsString());
-        jsonWriter.endObject();
-    }
 }
diff --git a/dataload/rdf2json/src/main/java/uk/ac/ebi/rdf2json/OntologyGraph.java b/dataload/rdf2json/src/main/java/uk/ac/ebi/rdf2json/OntologyGraph.java
@@ -262,14 +262,14 @@ private String urlToFilename(String url) {
         HierarchicalParentsAnnotator.annotateHierarchicalParents(this); // must run after RelatedAnnotator
         AncestorsAnnotator.annotateAncestors(this);
         HierarchyMetricsAnnotator.annotateHierarchyMetrics(this); // must run after HierarchicalParentsAnnotator
-        ShortFormAnnotator.annotateShortForms(this);
+        CurieAnnotator.annotateCuries(this);
         DefinitionAnnotator.annotateDefinitions(this);
         SynonymAnnotator.annotateSynonyms(this);
         ReifiedPropertyAnnotator.annotateReifiedProperties(this);
         OntologyMetadataAnnotator.annotateOntologyMetadata(this);
         HierarchyFlagsAnnotator.annotateHierarchyFlags(this); // must run after DirectParentsAnnotator and HierarchicalParentsAnnotator
         IsObsoleteAnnotator.annotateIsObsolete(this);
-        LabelAnnotator.annotateLabels(this); // must run after ShortFormAnnotator
+        LabelAnnotator.annotateLabels(this); // must run after CurieAnnotator
         ConfigurablePropertyAnnotator.annotateConfigurableProperties(this);
         PreferredRootsAnnotator.annotatePreferredRoots(this);
         DisjointWithAnnotator.annotateDisjointWith(this);

diff --git a/...f2json/annotators/ShortFormAnnotator.java → ...i/rdf2json/annotators/CurieAnnotator.java b/...f2json/annotators/ShortFormAnnotator.java → ...i/rdf2json/annotators/CurieAnnotator.java
@@ -10,10 +10,10 @@
 import uk.ac.ebi.rdf2json.annotators.helpers.OntologyBaseUris;
 import uk.ac.ebi.rdf2json.properties.PropertyValueLiteral;
 
-public class ShortFormAnnotator {
-	private static final Logger logger = LoggerFactory.getLogger(ShortFormAnnotator.class);
+public class CurieAnnotator {
+	private static final Logger logger = LoggerFactory.getLogger(CurieAnnotator.class);
 
-	public static void annotateShortForms(OntologyGraph graph) {
+	public static void annotateCuries(OntologyGraph graph) {
 
 		long startTime3 = System.nanoTime();
 
@@ -41,7 +41,7 @@ public static void annotateShortForms(OntologyGraph graph) {
 		}
 	}
 		long endTime3 = System.nanoTime();
-		logger.info("annotate short forms: {}", ((endTime3 - startTime3) / 1000 / 1000 / 1000));
+		logger.info("annotate curies: {}", ((endTime3 - startTime3) / 1000 / 1000 / 1000));
 
 
 	}

diff --git a/dataload/rdf2json/src/main/java/uk/ac/ebi/rdf2json/annotators/LabelAnnotator.java b/dataload/rdf2json/src/main/java/uk/ac/ebi/rdf2json/annotators/LabelAnnotator.java
@@ -37,7 +37,7 @@ public static Set<String> getLabelProperties(OntologyGraph graph) {
     }
 
     public static void annotateLabels(OntologyGraph graph) {
-        collateProperties(graph, "label", getLabelProperties(graph), List.of("shortForm"));
+        collateProperties(graph, "label", getLabelProperties(graph), List.of("curie"));
     }
 
     private static void collateProperties(OntologyGraph graph, String destProp, Collection<String> sourceProps, Collection<String> fallbackProps) {