Skip to content

Commit 5defd4a

Browse files
mridula-s109mridula-s109
committed
Clean the mess
1 parent 7fb7346 commit 5defd4a

File tree

2 files changed

+9
-184
lines changed

2 files changed

+9
-184
lines changed

‎x-pack/plugin/rank-rrf/src/internalClusterTest/java/org/elasticsearch/xpack/rank/rrf/RRFRetrieverBuilderIT.java‎

Lines changed: 0 additions & 180 deletions
Original file line numberDiff line numberDiff line change
@@ -47,7 +47,6 @@
4747
import java.util.concurrent.atomic.AtomicInteger;
4848

4949
import static org.elasticsearch.cluster.metadata.IndexMetadata.SETTING_NUMBER_OF_SHARDS;
50-
import static org.elasticsearch.search.rank.RankBuilder.DEFAULT_RANK_WINDOW_SIZE;
5150
import static org.elasticsearch.test.hamcrest.ElasticsearchAssertions.assertResponse;
5251
import static org.hamcrest.CoreMatchers.is;
5352
import static org.hamcrest.Matchers.containsString;
@@ -840,183 +839,4 @@ public XContentBuilder toXContent(XContentBuilder builder, Params params) throws
840839
assertThat(numAsyncCalls.get(), equalTo(4));
841840
}
842841

843-
public void testRRFWithWeightedFields() {
844-
// Test that weighted fields affect ranking as expected
845-
client().prepareIndex(INDEX).setId("1").setSource("title", "elasticsearch guide", "content", "comprehensive tutorial").get();
846-
client().prepareIndex(INDEX).setId("2").setSource("title", "advanced elasticsearch", "content", "expert guide").get();
847-
client().prepareIndex(INDEX).setId("3").setSource("title", "tutorial", "content", "elasticsearch basics").get();
848-
refresh();
849-
850-
// First search without weights - baseline
851-
var retrieverNoWeights = new RRFRetrieverBuilder(
852-
null,
853-
List.of("title", "content"),
854-
"elasticsearch",
855-
DEFAULT_RANK_WINDOW_SIZE,
856-
RRFRetrieverBuilder.DEFAULT_RANK_CONSTANT,
857-
new float[0]
858-
);
859-
860-
var responseNoWeights = client().prepareSearch(INDEX).setSource(new SearchSourceBuilder().retriever(retrieverNoWeights)).get();
861-
862-
// Second search with title field heavily weighted
863-
var retrieverWithWeights = new RRFRetrieverBuilder(
864-
null,
865-
List.of("title^5.0", "content^1.0"),
866-
"elasticsearch",
867-
DEFAULT_RANK_WINDOW_SIZE,
868-
RRFRetrieverBuilder.DEFAULT_RANK_CONSTANT,
869-
new float[0]
870-
);
871-
872-
var responseWithWeights = client().prepareSearch(INDEX).setSource(new SearchSourceBuilder().retriever(retrieverWithWeights)).get();
873-
874-
// Both searches should return the same documents
875-
assertEquals(responseNoWeights.getHits().getTotalHits().value(), responseWithWeights.getHits().getTotalHits().value());
876-
assertEquals(3L, responseWithWeights.getHits().getTotalHits().value());
877-
878-
// Verify that weighting title field more heavily affects the ranking
879-
// Document 2 has "elasticsearch" in title, so should rank higher with title weighted
880-
var hitsWithWeights = responseWithWeights.getHits().getHits();
881-
assertTrue("Should have results", hitsWithWeights.length > 0);
882-
883-
// The exact ranking may vary, but we should get consistent results
884-
for (var hit : hitsWithWeights) {
885-
assertTrue("All results should have positive scores", hit.getScore() > 0);
886-
}
887-
}
888-
889-
public void testRRFWeightValidation() {
890-
// Test that negative weights are properly rejected
891-
var retrieverWithNegativeWeight = new RRFRetrieverBuilder(
892-
null,
893-
List.of("title^-1.0", "content"),
894-
"test",
895-
DEFAULT_RANK_WINDOW_SIZE,
896-
RRFRetrieverBuilder.DEFAULT_RANK_CONSTANT,
897-
new float[0]
898-
);
899-
900-
var exception = expectThrows(
901-
IllegalArgumentException.class,
902-
() -> client().prepareSearch(INDEX).setSource(new SearchSourceBuilder().retriever(retrieverWithNegativeWeight)).get()
903-
);
904-
905-
assertThat(exception.getMessage(), containsString("per-field weights must be non-negative"));
906-
}
907-
908-
public void testRRFZeroWeights() {
909-
// Test that zero weights are accepted but effectively disable the field
910-
client().prepareIndex(INDEX).setId("1").setSource("title", "test document", "content", "content text").get();
911-
refresh();
912-
913-
var retrieverWithZeroWeight = new RRFRetrieverBuilder(
914-
null,
915-
List.of("title^0.0", "content^1.0"),
916-
"test",
917-
DEFAULT_RANK_WINDOW_SIZE,
918-
RRFRetrieverBuilder.DEFAULT_RANK_CONSTANT,
919-
new float[0]
920-
);
921-
922-
var response = client().prepareSearch(INDEX).setSource(new SearchSourceBuilder().retriever(retrieverWithZeroWeight)).get();
923-
924-
assertEquals(1L, response.getHits().getTotalHits().value());
925-
assertTrue("Should find the document via content field", response.getHits().getHits()[0].getScore() > 0);
926-
}
927-
928-
public void testRRFLargeWeightValues() {
929-
// Test that very large weight values are handled gracefully
930-
client().prepareIndex(INDEX).setId("1").setSource("title", "elasticsearch", "content", "search engine").get();
931-
client().prepareIndex(INDEX).setId("2").setSource("title", "search", "content", "elasticsearch engine").get();
932-
refresh();
933-
934-
var retrieverWithLargeWeights = new RRFRetrieverBuilder(
935-
null,
936-
List.of("title^1000000.0", "content^1.0"),
937-
"elasticsearch",
938-
DEFAULT_RANK_WINDOW_SIZE,
939-
RRFRetrieverBuilder.DEFAULT_RANK_CONSTANT,
940-
new float[0]
941-
);
942-
943-
var response = client().prepareSearch(INDEX).setSource(new SearchSourceBuilder().retriever(retrieverWithLargeWeights)).get();
944-
945-
assertTrue("Should find documents", response.getHits().getTotalHits().value() > 0);
946-
// Verify that large weights don't cause overflow issues
947-
for (var hit : response.getHits().getHits()) {
948-
assertTrue("Scores should be finite", Float.isFinite(hit.getScore()));
949-
assertTrue("Scores should be positive", hit.getScore() > 0);
950-
}
951-
}
952-
953-
public void testRRFMixedWeightedAndUnweightedFields() {
954-
// Test scenario with both weighted and unweighted fields
955-
client().prepareIndex(INDEX).setId("1").setSource("title", "elasticsearch", "content", "search", "description", "engine").get();
956-
refresh();
957-
958-
var retrieverMixed = new RRFRetrieverBuilder(
959-
null,
960-
List.of("title^3.0", "content", "description^0.5"), // Mixed weights
961-
"elasticsearch",
962-
DEFAULT_RANK_WINDOW_SIZE,
963-
RRFRetrieverBuilder.DEFAULT_RANK_CONSTANT,
964-
new float[0]
965-
);
966-
967-
var response = client().prepareSearch(INDEX).setSource(new SearchSourceBuilder().retriever(retrieverMixed)).get();
968-
969-
assertEquals(1L, response.getHits().getTotalHits().value());
970-
assertTrue("Should find the document", response.getHits().getHits()[0].getScore() > 0);
971-
}
972-
973-
public void testRRFWeightedFieldsRankingImpact() {
974-
// Test that different weight configurations produce different ranking results
975-
client().prepareIndex(INDEX).setId("1").setSource("title", "elasticsearch search", "content", "powerful engine").get();
976-
client().prepareIndex(INDEX).setId("2").setSource("title", "search engine", "content", "elasticsearch technology").get();
977-
refresh();
978-
979-
// Title-weighted query
980-
var titleWeightedRetriever = new RRFRetrieverBuilder(
981-
null,
982-
List.of("title^10.0", "content^1.0"),
983-
"elasticsearch",
984-
DEFAULT_RANK_WINDOW_SIZE,
985-
RRFRetrieverBuilder.DEFAULT_RANK_CONSTANT,
986-
new float[0]
987-
);
988-
989-
var titleWeightedResponse = client().prepareSearch(INDEX)
990-
.setSource(new SearchSourceBuilder().retriever(titleWeightedRetriever))
991-
.get();
992-
993-
// Content-weighted query
994-
var contentWeightedRetriever = new RRFRetrieverBuilder(
995-
null,
996-
List.of("title^1.0", "content^10.0"),
997-
"elasticsearch",
998-
DEFAULT_RANK_WINDOW_SIZE,
999-
RRFRetrieverBuilder.DEFAULT_RANK_CONSTANT,
1000-
new float[0]
1001-
);
1002-
1003-
var contentWeightedResponse = client().prepareSearch(INDEX)
1004-
.setSource(new SearchSourceBuilder().retriever(contentWeightedRetriever))
1005-
.get();
1006-
1007-
// Both should return results
1008-
assertEquals(2L, titleWeightedResponse.getHits().getTotalHits().value());
1009-
assertEquals(2L, contentWeightedResponse.getHits().getTotalHits().value());
1010-
1011-
// Verify that both return finite, positive scores
1012-
for (var hit : titleWeightedResponse.getHits().getHits()) {
1013-
assertTrue("Title-weighted scores should be finite", Float.isFinite(hit.getScore()));
1014-
assertTrue("Title-weighted scores should be positive", hit.getScore() > 0);
1015-
}
1016-
1017-
for (var hit : contentWeightedResponse.getHits().getHits()) {
1018-
assertTrue("Content-weighted scores should be finite", Float.isFinite(hit.getScore()));
1019-
assertTrue("Content-weighted scores should be positive", hit.getScore() > 0);
1020-
}
1021-
}
1022842
}

‎x-pack/plugin/rank-rrf/src/main/java/org/elasticsearch/xpack/rank/rrf/RRFRetrieverBuilder.java‎

Lines changed: 9 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -190,10 +190,15 @@ public RRFRetrieverBuilder(
190190
this.query = query;
191191
this.rankConstant = rankConstant;
192192
Objects.requireNonNull(weights, "weights must not be null");
193-
if (weights.length != innerRetrievers.size()) {
194-
throw new IllegalArgumentException(
195-
"weights array length [" + weights.length + "] must match retrievers count [" + innerRetrievers.size() + "]"
196-
);
193+
// For simplified syntax (fields + query), allow empty weights array since weights are handled during rewrite
194+
// For explicit retriever syntax, weights must match the number of retrievers
195+
if (fields == null || query == null) {
196+
// Explicit retriever syntax - validate weights match retrievers
197+
if (weights.length != innerRetrievers.size()) {
198+
throw new IllegalArgumentException(
199+
"weights array length [" + weights.length + "] must match retrievers count [" + innerRetrievers.size() + "]"
200+
);
201+
}
197202
}
198203
this.weights = weights;
199204
}

0 commit comments

Comments
 (0)