Add type_s field to nested documents as their prefix

wwtamu · Jan 8, 2025 · 4b8f188 · 4b8f188
1 parent fca75a7
commit 4b8f188
Show file tree

Hide file tree

Showing 6 changed files with 35 additions and 19 deletions.
diff --git a/src/main/java/edu/tamu/scholars/discovery/AppConstants.java b/src/main/java/edu/tamu/scholars/discovery/AppConstants.java
@@ -2,6 +2,10 @@
 
 public class AppConstants {
 
+    public static final char DOT = '.';
+
+    public static final char FORWARD_SLASH = '/';
+
     public static final String ID = "id";
 
     public static final String CLASS = "class";
@@ -18,6 +22,8 @@ public class AppConstants {
 
     public static final String COLLECTIONS = "_collections_";
 
+    public static final String TYPE_S = "type_s";
+
     public static final String WILDCARD = "*";
 
     public static final String QUERY_DELIMETER = ":";

diff --git a/src/main/java/edu/tamu/scholars/discovery/etl/EtlUtility.java b/src/main/java/edu/tamu/scholars/discovery/etl/EtlUtility.java
@@ -1,5 +1,6 @@
 package edu.tamu.scholars.discovery.etl;
 
+import static edu.tamu.scholars.discovery.AppConstants.DOT;
 import static org.apache.commons.lang3.StringUtils.isNotEmpty;
 
 import edu.tamu.scholars.discovery.etl.model.DataFieldDescriptor;
@@ -10,6 +11,13 @@ private EtlUtility() {
 
     }
 
+    public static String getFieldPrefix(DataFieldDescriptor descriptor) {
+        String name = getFieldName(descriptor);
+        int index = name.indexOf(DOT);
+
+        return index < 0 ? name : name.substring(0, index);
+    }
+
     public static String getFieldName(DataFieldDescriptor descriptor) {
         return isNestedReference(descriptor)
             ? descriptor.getNestedReference().getKey()

diff --git a/src/main/java/edu/tamu/scholars/discovery/etl/load/solr/SolrIndexLoader.java b/src/main/java/edu/tamu/scholars/discovery/etl/load/solr/SolrIndexLoader.java
@@ -99,9 +99,9 @@ private void preprocessFields() {
         List<Field> fields = new ArrayList<>();
         List<CopyField> copyFields = new ArrayList<>();
 
-        processField(getDescriptor(LABEL, false), fields, copyFields);
-        processField(getDescriptor(CLASS, false), fields, copyFields);
-        processField(getDescriptor("_collections_", true), fields, copyFields);
+        processField(getDescriptor(LABEL, true, false), fields, copyFields);
+        processField(getDescriptor(CLASS, false, false), fields, copyFields);
+        processField(getDescriptor("_collections_", false, true), fields, copyFields);
 
         this.data.getFields()
             .stream()
@@ -165,9 +165,10 @@ private void processField(DataFieldDescriptor descriptor, List<Field> fields, Li
         }
     }
 
-    private DataFieldDescriptor getDescriptor(String name, boolean multiValued) {
+    private DataFieldDescriptor getDescriptor(String name, boolean docValues, boolean multiValued) {
         DataFieldDescriptor descriptor = new DataFieldDescriptor();
         FieldDestination destination = new FieldDestination();
+        destination.setDocValues(docValues);
         destination.setMultiValued(multiValued);
 
         descriptor.setName(name);

diff --git a/...edu/tamu/scholars/discovery/etl/transform/solr/FlatMapToSolrInputDocumentTransformer.java b/...edu/tamu/scholars/discovery/etl/transform/solr/FlatMapToSolrInputDocumentTransformer.java
@@ -6,7 +6,9 @@
 import static edu.tamu.scholars.discovery.AppConstants.ID_PATH_DELIMITER;
 import static edu.tamu.scholars.discovery.AppConstants.LABEL;
 import static edu.tamu.scholars.discovery.AppConstants.SYNC_IDS;
+import static edu.tamu.scholars.discovery.AppConstants.TYPE_S;
 import static edu.tamu.scholars.discovery.etl.EtlConstants.NESTED_DELIMITER_PATTERN;
+import static edu.tamu.scholars.discovery.etl.EtlUtility.getFieldPrefix;
 import static org.apache.commons.lang3.StringUtils.isNotEmpty;
 
 import java.text.ParseException;
@@ -148,6 +150,8 @@ private SolrInputDocument processNestedValue(
         nestedDocument.setField(ID, nestedDocumentId);
         nestedDocument.setField(LABEL, nestedDocumentLabel);
 
+        nestedDocument.setField(TYPE_S, getFieldPrefix(descriptor));
+
         rootContext.syncIds.add(id);
 
         processNestedReferences(

diff --git a/src/main/java/edu/tamu/scholars/discovery/factory/index/solr/builder/SolrQueryBuilder.java b/src/main/java/edu/tamu/scholars/discovery/factory/index/solr/builder/SolrQueryBuilder.java
@@ -137,18 +137,16 @@ public SolrQueryBuilder withFacets(List<FacetArg> facets) {
 
     public SolrQueryBuilder withBoosts(List<BoostArg> boosts) {
         StringBuilder boostedQuery = new StringBuilder(this.query.getQuery());
-        boosts.forEach(boost -> {
-            boostedQuery.append(" OR ")
-                .append("(")
-                .append(boost.getField())
-                .append(":")
-                .append("(")
-                .append(query)
-                .append(")")
-                .append("^")
-                .append(boost.getValue())
-                .append(")");
-        });
+        boosts.forEach(boost -> boostedQuery.append(" OR ")
+            .append("(")
+            .append(boost.getField())
+            .append(":")
+            .append("(")
+            .append(query)
+            .append(")")
+            .append("^")
+            .append(boost.getValue())
+            .append(")"));
 
         this.query.setQuery(boostedQuery.toString());
 

diff --git a/...ain/java/edu/tamu/scholars/discovery/model/serializer/UnwrappingIndividualSerializer.java b/...ain/java/edu/tamu/scholars/discovery/model/serializer/UnwrappingIndividualSerializer.java
@@ -1,6 +1,8 @@
 package edu.tamu.scholars.discovery.model.serializer;
 
 import static edu.tamu.scholars.discovery.AppConstants.COLLECTIONS;
+import static edu.tamu.scholars.discovery.AppConstants.DOT;
+import static edu.tamu.scholars.discovery.AppConstants.FORWARD_SLASH;
 import static edu.tamu.scholars.discovery.AppConstants.ID;
 
 import java.io.IOException;
@@ -17,9 +19,6 @@
 
 public class UnwrappingIndividualSerializer extends JsonSerializer<Individual> {
 
-    private static final char DOT = '.';
-    private static final char FORWARD_SLASH = '/';
-
     private static final Set<String> EXCLUDED_FIELDS = Set.of(
         "_root_",
         "_version_",