snapshot of https://github.com/blevesearch/bleve.git (master, 2026-09-15; blobs > 131072B omitted)

fusion/rrf_test.go

@@ -0,0 +1,212 @@
+//  Copyright (c) 2025 Couchbase, Inc.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+// 		http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package fusion
+
+import (
+	"math"
+	"testing"
+
+	"github.com/blevesearch/bleve/v2/search"
+)
+
+const epsilon float64 = 1e-3
+
+func nearlyEqual(a float64, b float64, epsilon float64) bool {
+	return math.Abs(a-b) < epsilon
+}
+
+func compareFusionResults(a, b FusionResult) bool {
+	if a.Total != b.Total || !nearlyEqual(a.MaxScore, b.MaxScore, epsilon) || len(a.Hits) != len(b.Hits) {
+		return false
+	}
+	for i := range a.Hits {
+		if a.Hits[i].ID != b.Hits[i].ID || !nearlyEqual(a.Hits[i].Score, b.Hits[i].Score, epsilon) {
+			return false
+		}
+
+		if a.Hits[i].ScoreBreakdown != nil || b.Hits[i].ScoreBreakdown != nil {
+			return false
+		}
+	}
+	return true
+}
+
+func TestReciprocalRankFusion(t *testing.T) {
+	tests := []struct {
+		name          string
+		hits          search.DocumentMatchCollection
+		weights       []float64
+		rank_constant int
+		window_size   int
+		numKNNQueries int
+		want          FusionResult
+	}{
+		{
+			name:          "empty hits",
+			hits:          search.DocumentMatchCollection{},
+			weights:       []float64{0.5, 0.5},
+			rank_constant: 60,
+			window_size:   10,
+			numKNNQueries: 1,
+			want: FusionResult{
+				Hits:     search.DocumentMatchCollection{},
+				Total:    0,
+				MaxScore: 0.0,
+			},
+		},
+		{
+			name: "single knn query",
+			hits: search.DocumentMatchCollection{
+				{ID: "a", Score: 0.9, ScoreBreakdown: map[int]float64{0: 0.8}},
+				{ID: "b", Score: 0.8, ScoreBreakdown: map[int]float64{0: 0.9}},
+				{ID: "c", Score: 0.7, ScoreBreakdown: map[int]float64{0: 0.7}},
+			},
+			weights:       []float64{0.4, 0.6},
+			rank_constant: 1,
+			window_size:   3,
+			numKNNQueries: 1,
+			want: FusionResult{
+				Hits: search.DocumentMatchCollection{
+					{ID: "b", Score: 0.433},
+					{ID: "a", Score: 0.4},
+					{ID: "c", Score: 0.25},
+				},
+				Total:    3,
+				MaxScore: 0.433,
+			},
+		},
+		{
+			name: "multiple knn queries",
+			hits: search.DocumentMatchCollection{
+				{ID: "a", Score: 0.9, ScoreBreakdown: map[int]float64{0: 0.8, 1: 0.6}},
+				{ID: "b", Score: 0.8, ScoreBreakdown: map[int]float64{0: 0.9, 1: 0.5}},
+				{ID: "c", Score: 0.7, ScoreBreakdown: map[int]float64{0: 0.7, 1: 0.7}},
+			},
+			weights:       []float64{0.3, 0.4, 0.3},
+			rank_constant: 1,
+			window_size:   3,
+			numKNNQueries: 2,
+			want: FusionResult{
+				Hits: search.DocumentMatchCollection{
+					{ID: "a", Score: 0.383},
+					{ID: "b", Score: 0.375},
+					{ID: "c", Score: 0.325},
+				},
+				Total:    3,
+				MaxScore: 0.383,
+			},
+		},
+		{
+			name: "window size smaller than hits",
+			hits: search.DocumentMatchCollection{
+				{ID: "a", Score: 0.9, ScoreBreakdown: map[int]float64{0: 0.7}},
+				{ID: "b", Score: 0.8, ScoreBreakdown: map[int]float64{0: 0.9}},
+				{ID: "c", Score: 0.7, ScoreBreakdown: map[int]float64{0: 0.8}},
+			},
+			weights:       []float64{0.4, 0.6},
+			rank_constant: 1,
+			window_size:   2,
+			numKNNQueries: 1,
+			want: FusionResult{
+				Hits: search.DocumentMatchCollection{
+					{ID: "b", Score: 0.433},
+					{ID: "a", Score: 0.2},
+				},
+				Total:    2,
+				MaxScore: 0.433,
+			},
+		},
+		{
+			name: "documents with partial scores missing KNN scores",
+			hits: search.DocumentMatchCollection{
+				{ID: "a", Score: 0.9, ScoreBreakdown: map[int]float64{0: 0.8}},         // has FTS and KNN query 0, missing KNN query 1
+				{ID: "b", Score: 0.8, ScoreBreakdown: map[int]float64{1: 0.7}},         // has FTS and KNN query 1, missing KNN query 0
+				{ID: "c", Score: 0.7, ScoreBreakdown: map[int]float64{0: 0.6, 1: 0.9}}, // has all scores
+				{ID: "d", Score: 0.6, ScoreBreakdown: map[int]float64{}},               // has only FTS, missing all KNN scores
+			},
+			weights:       []float64{0.3, 0.4, 0.3}, // FTS, KNN query 0, KNN query 1
+			rank_constant: 1,
+			window_size:   4,
+			numKNNQueries: 2,
+			want: FusionResult{
+				Hits: search.DocumentMatchCollection{
+					{ID: "c", Score: 0.358}, // FTS rank 3, KNN0 rank 2, KNN1 rank 1: 0.3/4 + 0.4/3 + 0.3/2 = 0.075 + 0.133 + 0.15 = 0.358
+					{ID: "a", Score: 0.35},  // FTS rank 1, KNN0 rank 1, no KNN1: 0.3/2 + 0.4/2 + 0 = 0.15 + 0.2 + 0 = 0.35
+					{ID: "b", Score: 0.2},   // FTS rank 2, no KNN0, KNN1 rank 2: 0.3/3 + 0 + 0.3/3 = 0.1 + 0 + 0.1 = 0.2
+					{ID: "d", Score: 0.06},  // FTS rank 4, no KNN0, no KNN1: 0.3/5 + 0 + 0 = 0.06
+				},
+				Total:    4,
+				MaxScore: 0.358,
+			},
+		},
+		{
+			name: "documents with only KNN scores",
+			hits: search.DocumentMatchCollection{
+				{ID: "a", Score: 0.0, ScoreBreakdown: map[int]float64{0: 0.9}},         // no FTS rank (Score 0.0), only KNN query 0
+				{ID: "b", Score: 0.0, ScoreBreakdown: map[int]float64{1: 0.8}},         // no FTS rank (Score 0.0), only KNN query 1
+				{ID: "c", Score: 0.0, ScoreBreakdown: map[int]float64{0: 0.7, 1: 0.6}}, // no FTS rank (Score 0.0), both KNN queries
+			},
+			weights:       []float64{0.5, 0.3, 0.2}, // FTS, KNN query 0, KNN query 1
+			rank_constant: 1,
+			window_size:   3,
+			numKNNQueries: 2,
+			want: FusionResult{
+				Hits: search.DocumentMatchCollection{
+					{ID: "c", Score: 0.167}, // no FTS rank, KNN0 rank 2, KNN1 rank 2: 0 + 0.3/3 + 0.2/3 = 0 + 0.1 + 0.067 = 0.167
+					{ID: "a", Score: 0.15},  // no FTS rank, KNN0 rank 1, no KNN1: 0 + 0.3/2 + 0 = 0 + 0.15 + 0 = 0.15
+					{ID: "b", Score: 0.1},   // no FTS rank, no KNN0, KNN1 rank 1: 0 + 0 + 0.2/2 = 0 + 0 + 0.1 = 0.1
+				},
+				Total:    3,
+				MaxScore: 0.167,
+			},
+		},
+		{
+			name: "mixed scenario with gaps in KNN queries",
+			hits: search.DocumentMatchCollection{
+				{ID: "a", Score: 0.8, ScoreBreakdown: map[int]float64{1: 0.9}}, // has FTS and KNN query 1, missing KNN query 0
+				{ID: "b", Score: 0.6, ScoreBreakdown: map[int]float64{0: 0.8}}, // has FTS and KNN query 0, missing KNN query 1
+				{ID: "c", Score: 0.0, ScoreBreakdown: map[int]float64{0: 0.7}}, // no FTS rank (Score 0.0), only KNN query 0
+				{ID: "d", Score: 0.4, ScoreBreakdown: map[int]float64{}},       // only FTS, no KNN scores
+			},
+			weights:       []float64{0.4, 0.3, 0.3}, // FTS, KNN query 0, KNN query 1
+			rank_constant: 1,
+			window_size:   4,
+			numKNNQueries: 2,
+			want: FusionResult{
+				Hits: search.DocumentMatchCollection{
+					{ID: "a", Score: 0.35},  // FTS rank 1, no KNN0, KNN1 rank 1: 0.4/2 + 0 + 0.3/2 = 0.2 + 0 + 0.15 = 0.35
+					{ID: "b", Score: 0.283}, // FTS rank 2, KNN0 rank 1, no KNN1: 0.4/3 + 0.3/2 + 0 = 0.133 + 0.15 + 0 = 0.283
+					{ID: "d", Score: 0.1},   // FTS rank 3, no KNN0, no KNN1: 0.4/4 + 0 + 0 = 0.1
+					{ID: "c", Score: 0.1},   // no FTS rank, KNN0 rank 2, no KNN1: 0 + 0.3/3 + 0 = 0 + 0.1 + 0 = 0.1
+				},
+				Total:    4,
+				MaxScore: 0.35,
+			},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			for i, hit := range tt.hits {
+				hit.HitNumber = uint64(i)
+			}
+
+			if got := ReciprocalRankFusion(tt.hits, tt.weights, tt.rank_constant, tt.window_size, tt.numKNNQueries, false); !compareFusionResults(*got, tt.want) {
+				t.Errorf("ReciprocalRankFusion() = %v, want %v", got, tt.want)
+			}
+		})
+	}
+}