change into topBoostOnly

clientlib/src/main/proto/yelp/nrtsearch/search.proto

@@ -1132,8 +1132,8 @@ message Highlight {
         google.protobuf.UInt32Value  boundary_max_scan = 15;
         // Locale used in boundary scanner when using "word" or "sentence" boundary_scanner. Examples: "en-US", "ch-ZH".
         google.protobuf.StringValue  boundary_scanner_locale = 16;
-        // Only highlight the top matched phrase (with the highest boost value) once per segment. By default, it is false.
-        google.protobuf.BoolValue top_phrase_once = 17;
+        // Only highlight the top matched phrases (with the highest boost value) per fragment. By default, it is false.
+        google.protobuf.BoolValue top_boost_only = 17;
     }
 
     // Highlight settings

grpc-gateway/luceneserver.swagger.json

@@ -1766,9 +1766,9 @@
           "type": "string",
           "description": "Locale used in boundary scanner when using \"word\" or \"sentence\" boundary_scanner. Examples: \"en-US\", \"ch-ZH\"."
         },
-        "topPhraseOnce": {
+        "topBoostOnly": {
           "type": "boolean",
-          "description": "Only highlight the top matched phrase (with the highest boost value) once per segment. By default, it is false."
+          "description": "Only highlight the top matched phrases (with the highest boost value) per fragment. By default, it is false."
         }
       }
     },

src/main/java/com/yelp/nrtsearch/server/luceneserver/highlights/HighlightSettings.java

@@ -38,7 +38,7 @@ public class HighlightSettings {
   private final Character[] boundaryChars;
   private final int boundaryMaxScan;
   private final Locale boundaryScannerLocale;
-  private final boolean topPhraseOnce;
+  private final boolean topBoostOnly;
 
   public HighlightSettings(
       Highlighter highlighter,
@@ -55,7 +55,7 @@ public HighlightSettings(
       Character[] boundaryChars,
       int boundaryMaxScan,
       Locale boundaryScannerLocale,
-      boolean topPhraseOnce,
+      boolean topBoostOnly,
       Map<String, Object> customHighlighterParams) {
     this.highlighter = highlighter;
     this.preTags = preTags;
@@ -71,7 +71,7 @@ public HighlightSettings(
     this.boundaryChars = boundaryChars;
     this.boundaryMaxScan = boundaryMaxScan;
     this.boundaryScannerLocale = boundaryScannerLocale;
-    this.topPhraseOnce = topPhraseOnce;
+    this.topBoostOnly = topBoostOnly;
     this.customHighlighterParams = customHighlighterParams;
   }
 
@@ -91,7 +91,7 @@ public Builder toBuilder() {
         .withBoundaryChars(this.boundaryChars)
         .withBoundaryMaxScan(this.boundaryMaxScan)
         .withBoundaryScannerLocale(this.boundaryScannerLocale)
-        .withTopPhraseOnce(this.topPhraseOnce)
+        .withTopBoostOnly(this.topBoostOnly)
         .withCustomHighlighterParams(this.customHighlighterParams);
   }
 
@@ -151,8 +151,8 @@ public Locale getBoundaryScannerLocale() {
     return boundaryScannerLocale;
   }
 
-  public boolean getTopPhraseOnce() {
-    return topPhraseOnce;
+  public boolean getTopBoostOnly() {
+    return topBoostOnly;
   }
 
   public Map<String, Object> getCustomHighlighterParams() {
@@ -194,8 +194,8 @@ public String toString() {
         + boundaryMaxScan
         + ", boundaryScannerLocale="
         + boundaryScannerLocale
-        + ", topPhraseOnce="
-        + topPhraseOnce
+        + ", topBoostOnly="
+        + topBoostOnly
         + '}';
   }
 
@@ -215,7 +215,7 @@ public static final class Builder {
     private Character[] boundaryChars;
     private int boundaryMaxScan;
     private Locale boundaryScannerLocale;
-    private boolean topPhraseOnce;
+    private boolean topBoostOnly;
     private Map<String, Object> customHighlighterParams;
 
     public Builder() {}
@@ -290,8 +290,8 @@ public Builder withBoundaryScannerLocale(Locale boundaryScannerLocale) {
       return this;
     }
 
-    public Builder withTopPhraseOnce(boolean topPhraseOnce) {
-      this.topPhraseOnce = topPhraseOnce;
+    public Builder withTopBoostOnly(boolean topBoostOnly) {
+      this.topBoostOnly = topBoostOnly;
       return this;
     }
 
@@ -316,7 +316,7 @@ public HighlightSettings build() {
           boundaryChars,
           boundaryMaxScan,
           boundaryScannerLocale,
-          topPhraseOnce,
+          topBoostOnly,
           customHighlighterParams);
     }
   }

src/main/java/com/yelp/nrtsearch/server/luceneserver/highlights/HighlightUtils.java

@@ -133,10 +133,10 @@ static Map<String, HighlightSettings> createPerFieldSettings(
                     settings.hasBoundaryScannerLocale()
                         ? Locale.forLanguageTag(settings.getBoundaryScannerLocale().getValue())
                         : globalSettings.getBoundaryScannerLocale())
-                .withTopPhraseOnce(
-                    settings.hasTopPhraseOnce()
-                        ? settings.getTopPhraseOnce().getValue()
-                        : globalSettings.getTopPhraseOnce())
+                .withTopBoostOnly(
+                    settings.hasTopBoostOnly()
+                        ? settings.getTopBoostOnly().getValue()
+                        : globalSettings.getTopBoostOnly())
                 .withCustomHighlighterParams(
                     settings.hasCustomHighlighterParams()
                         ? StructValueTransformer.transformStruct(
@@ -217,9 +217,9 @@ private static HighlightSettings createGlobalFieldSettings(
             settings.hasBoundaryScannerLocale()
                 ? Locale.forLanguageTag(settings.getBoundaryScannerLocale().getValue())
                 : DEFAULT_BOUNDARY_SCANNER_LOCALE)
-        .withTopPhraseOnce(
-            settings.hasTopPhraseOnce()
-                ? settings.getTopPhraseOnce().getValue()
+        .withTopBoostOnly(
+            settings.hasTopBoostOnly()
+                ? settings.getTopBoostOnly().getValue()
                 : DEFAULT_TOP_PHRASE_ONCE)
         .withCustomHighlighterParams(
             settings.hasCustomHighlighterParams()

src/main/java/com/yelp/nrtsearch/server/luceneserver/highlights/NRTFastVectorHighlighter.java

@@ -120,12 +120,12 @@ public String[] getHighlights(
     }
 
     BaseFragmentsBuilder fragmentsBuilder =
-        new TopPhraseOnceFragmentsBuilderAdaptor(
+        new TopBoostOnlyFragmentsBuilderAdaptor(
             settings.isScoreOrdered()
                 ? new ScoreOrderFragmentsBuilder()
                 : new SimpleFragmentsBuilder(),
             boundaryScanner,
-            settings.getTopPhraseOnce());
+            settings.getTopBoostOnly());
     fragmentsBuilder.setDiscreteMultiValueHighlighting(settings.getDiscreteMultivalue());
 
     try {

src/main/java/com/yelp/nrtsearch/server/luceneserver/highlights/TopPhraseOnceFragmentsBuilderAdaptor.java → src/main/java/com/yelp/nrtsearch/server/luceneserver/highlights/TopBoostOnlyFragmentsBuilderAdaptor.java

@@ -15,7 +15,6 @@
  */
 package com.yelp.nrtsearch.server.luceneserver.highlights;
 
-import java.util.Comparator;
 import java.util.List;
 import org.apache.lucene.document.Field;
 import org.apache.lucene.search.highlight.Encoder;
@@ -25,18 +24,18 @@
 import org.apache.lucene.search.vectorhighlight.FieldFragList.WeightedFragInfo.SubInfo;
 import org.apache.lucene.search.vectorhighlight.FieldPhraseList.WeightedPhraseInfo.Toffs;
 
-public class TopPhraseOnceFragmentsBuilderAdaptor extends BaseFragmentsBuilder {
+public class TopBoostOnlyFragmentsBuilderAdaptor extends BaseFragmentsBuilder {
   private final BaseFragmentsBuilder innerBaseFragmentsBuilder;
-  private final boolean topPhraseOnce;
+  private final boolean topBoostOnly;
 
   /** a constructor. */
-  public TopPhraseOnceFragmentsBuilderAdaptor(
+  public TopBoostOnlyFragmentsBuilderAdaptor(
       BaseFragmentsBuilder baseFragmentsBuilder,
       BoundaryScanner boundaryScanner,
-      boolean topPhraseOnce) {
+      boolean topBoostOnly) {
     super(boundaryScanner);
     this.innerBaseFragmentsBuilder = baseFragmentsBuilder;
-    this.topPhraseOnce = topPhraseOnce;
+    this.topBoostOnly = topBoostOnly;
   }
 
   @Override
@@ -53,7 +52,7 @@ protected String makeFragment(
       String[] preTags,
       String[] postTags,
       Encoder encoder) {
-    if (!topPhraseOnce) {
+    if (!topBoostOnly) {
       return super.makeFragment(buffer, index, values, fragInfo, preTags, postTags, encoder);
     }
     StringBuilder fragment = new StringBuilder();
@@ -63,11 +62,12 @@ protected String makeFragment(
         getFragmentSourceMSO(
             buffer, index, values, s, fragInfo.getEndOffset(), modifiedStartOffset);
     int srcIndex = 0;
-    // get a list of the first highest boosted phrase only
-    List<SubInfo> subInfos =
-        fragInfo.getSubInfos().stream().max(Comparator.comparingDouble(SubInfo::getBoost)).stream()
-            .toList();
-    for (SubInfo subInfo : subInfos) {
+    double topBoostValue =
+        fragInfo.getSubInfos().stream().map(SubInfo::getBoost).max(Float::compare).orElse(0f);
+    for (SubInfo subInfo : fragInfo.getSubInfos()) {
+      if (subInfo.getBoost() < topBoostValue) {
+        continue;
+      }
       for (Toffs to : subInfo.getTermsOffsets()) {
         fragment
             .append(

src/test/java/com/yelp/nrtsearch/server/luceneserver/highlights/NRTFastVectorHighlighterTest.java

@@ -82,7 +82,7 @@ protected void initIndex(String name) throws Exception {
                             "The food is good there, but the service is terrible.",
                             "I personally don't like the staff at this place.",
                             "Not all food are good.",
-                            "This pizza place was one of my favorites in downtown, and as promised by other users, the pepperoni pizza is the best pepperoni pizza in town."))
+                            "The margarita pizza and the marinara pizza in this pizzeria are yummy and inexpensive."))
                     .build())
             .putFields(
                 "boundary_scanner_field",
@@ -194,7 +194,7 @@ public void testHighlightMultivalueField() {
   }
 
   @Test
-  public void testHighlightMultivalueFieldWithTopPhraseOnly() {
+  public void testHighlightMultivalueFieldWithTopBoostOnly() {
     Highlight highlight =
         Highlight.newBuilder()
             .addFields("comment_multivalue")
@@ -212,21 +212,41 @@ public void testHighlightMultivalueFieldWithTopPhraseOnly() {
                                                         PhraseQuery.newBuilder()
                                                             .setField("comment_multivalue")
                                                             .addAllTerms(
-                                                                List.of("pepperoni", "pizza")))
+                                                                List.of("margarita", "pizza")))
                                                     .setBoost(3))
                                             .setOccurValue(BooleanClause.Occur.SHOULD_VALUE))
+                                    .addClauses(
+                                        BooleanClause.newBuilder()
+                                            .setQuery(
+                                                Query.newBuilder()
+                                                    .setPhraseQuery(
+                                                        PhraseQuery.newBuilder()
+                                                            .setField("comment_multivalue")
+                                                            .addAllTerms(
+                                                                List.of("marinara", "pizza")))
+                                                    .setBoost(3))
+                                            .setOccurValue(BooleanClause.Occur.SHOULD_VALUE))
+                                    .addClauses(
+                                        BooleanClause.newBuilder()
+                                            .setQuery(
+                                                Query.newBuilder()
+                                                    .setTermQuery(
+                                                        TermQuery.newBuilder()
+                                                            .setField("comment_multivalue")
+                                                            .setTextValue("delicious"))
+                                                    .setBoost(4)))
                                     .addClauses(
                                         BooleanClause.newBuilder()
                                             .setQuery(
                                                 Query.newBuilder()
                                                     .setTermQuery(
                                                         TermQuery.newBuilder()
                                                             .setField("comment_multivalue")
-                                                            .setTextValue("pizza"))
+                                                            .setTextValue("yummy"))
                                                     .setBoost(2)))))
                     .setMaxNumberOfFragments(UInt32Value.of(1))
                     .setFragmentSize(UInt32Value.of(250))
-                    .setTopPhraseOnce(BoolValue.of(true))
+                    .setTopBoostOnly(BoolValue.of(true))
                     .setScoreOrdered(BoolValue.of(true))
                     .setDiscreteMultivalue(BoolValue.of(true)))
             .build();
@@ -236,7 +256,7 @@ public void testHighlightMultivalueFieldWithTopPhraseOnly() {
 
     assertThat(response.getHits(0).getHighlightsMap().get("comment_multivalue").getFragmentsList())
         .containsExactly(
-            "This pizza place was one of my favorites in downtown, and as promised by other users, the <em>pepperoni pizza</em> is the best pepperoni pizza in town.");
+            "The <em>margarita pizza</em> and the <em>marinara pizza</em> in this pizzeria are yummy and inexpensive.");
     assertThat(response.getDiagnostics().getHighlightTimeMs()).isGreaterThan(0);
   }