MB-61640: Fuzzy Dynamic Scoring

 - Removed levenshtein distance calculation from bleve
 - Propagated edit distances from vellum
 - Removed all changes for wildcard

index_test.go

@@ -2953,3 +2953,62 @@ func TestCopyIndex(t *testing.T) {
 		}
 	}
 }
+
+func TestFuzzyScoring(t *testing.T) {
+	tmpIndexPath := createTmpIndexPath(t)
+	defer cleanupTmpIndexPath(t, tmpIndexPath)
+
+	mp := NewIndexMapping()
+	mp.DefaultAnalyzer = "simple"
+	idx, err := New(tmpIndexPath, mp)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	batch := idx.NewBatch()
+
+	docs := []map[string]interface{}{
+		{
+			"textField": "ab",
+		},
+		{
+			"textField": "abc",
+		},
+		{
+			"textField": "abcd",
+		},
+	}
+
+	for _, doc := range docs {
+		err := batch.Index(fmt.Sprintf("%v", doc["textField"]), doc)
+		if err != nil {
+			t.Fatal(err)
+		}
+	}
+
+	err = idx.Batch(batch)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	query := NewFuzzyQuery("ab")
+	query.Fuzziness = 2
+	searchRequest := NewSearchRequestOptions(query, 10, 0, true)
+	res, err := idx.Search(searchRequest)
+	if err != nil {
+		t.Error(err)
+	}
+
+	maxScore := res.Hits[0].Score
+
+	for i, hit := range res.Hits {
+		if maxScore/float64(i+1) != hit.Score {
+			t.Errorf("expected score - %f, got score - %f", maxScore/float64(i+1), hit.Score)
+		}
+	}
+
+	err = idx.Close()
+	if err != nil {
+		t.Fatal(err)
+	}
+}

search/searcher/search_fuzzy.go

@@ -52,9 +52,11 @@ func NewFuzzySearcher(ctx context.Context, indexReader index.IndexReader, term s
 	}
 
 	var candidates []string
+	var editDistances []uint8
 	var dictBytesRead uint64
 	if fuzzyCandidates != nil {
 		candidates = fuzzyCandidates.candidates
+		editDistances = fuzzyCandidates.editDistances
 		dictBytesRead = fuzzyCandidates.bytesRead
 	}
 
@@ -68,7 +70,7 @@ func NewFuzzySearcher(ctx context.Context, indexReader index.IndexReader, term s
 	}
 
 	return NewMultiTermSearcherBoosted(ctx, indexReader, candidates, field,
-		boost, fuzzyCandidates.editDistances, options, true)
+		boost, editDistances, options, true)
 }
 
 type fuzzyCandidates struct {
@@ -95,7 +97,7 @@ func findFuzzyCandidateTerms(indexReader index.IndexReader, term string,
 		candidates:    make([]string, 0),
 		editDistances: make([]uint8, 0),
 	}
-	var reuse []int
+
 	// in case of advanced reader implementations directly call
 	// the levenshtein automaton based iterator to collect the
 	// candidate terms
@@ -112,7 +114,7 @@ func findFuzzyCandidateTerms(indexReader index.IndexReader, term string,
 		tfd, err := fieldDict.Next()
 		for err == nil && tfd != nil {
 			rv.candidates = append(rv.candidates, tfd.Term)
-			rv.editDistances = append(rv.editDistances, tfd.Distance)
+			rv.editDistances = append(rv.editDistances, tfd.EditDistance)
 			if tooManyClauses(len(rv.candidates)) {
 				return nil, tooManyClausesErr(field, len(rv.candidates))
 			}
@@ -139,6 +141,7 @@ func findFuzzyCandidateTerms(indexReader index.IndexReader, term string,
 	}()
 
 	// enumerate terms and check levenshtein distance
+	var reuse []int
 	tfd, err := fieldDict.Next()
 	for err == nil && tfd != nil {
 		var ld int

search/searcher/search_multi_term.go

@@ -45,6 +45,8 @@ func NewMultiTermSearcher(ctx context.Context, indexReader index.IndexReader, te
 		options, limit)
 }
 
+// Works similarly to the multi term searcher but additionally boosts individual terms based on
+// their edit distance from the query terms
 func NewMultiTermSearcherBoosted(ctx context.Context, indexReader index.IndexReader, terms []string,
 	field string, boost float64, editDistances []uint8, options search.SearcherOptions, limit bool) (
 	search.Searcher, error) {