fix (unified-storage): search returning empty results when query includes small term (#111140)

* fix search not returning anything when a word in the query has less than 3 characters

* fix test name

* fmt

* remove trimspace

pkg/storage/unified/search/bleve.go

@@ -1092,7 +1092,7 @@ func (b *bleveIndex) toBleveSearchRequest(ctx context.Context, req *resourcepb.R
 		queryPhrase.Analyzer = standard.Name
 
 		// Query 3: Match query with standard analyzer
-		queryAnalyzed := bleve.NewMatchQuery(req.Query)
+		queryAnalyzed := bleve.NewMatchQuery(removeSmallTerms(req.Query))
 		queryAnalyzed.Analyzer = standard.Name
 		queryAnalyzed.Operator = query.MatchQueryOperatorAnd // Make sure all terms from the query are matched
 
@@ -1178,6 +1178,23 @@ func (b *bleveIndex) toBleveSearchRequest(ctx context.Context, req *resourcepb.R
 	return searchrequest, nil
 }
 
+func removeSmallTerms(query string) string {
+	words := strings.Fields(query)
+	validWords := make([]string, 0, len(words))
+
+	for _, word := range words {
+		if len(word) >= EDGE_NGRAM_MIN_TOKEN {
+			validWords = append(validWords, word)
+		}
+	}
+
+	if len(validWords) == 0 {
+		return query
+	}
+
+	return strings.Join(validWords, " ")
+}
+
 func (b *bleveIndex) stopUpdaterAndCloseIndex() error {
 	// Signal updater to stop. We do this by 1) setting updaterShuttingDown + sending signal, and by 2) calling cancel.
 	b.updaterMu.Lock()

pkg/storage/unified/search/bleve_search_test.go

@@ -191,6 +191,22 @@ func TestCanSearchByTitle(t *testing.T) {
 			checkSearchQuery(t, index, newQueryByTitle(fmt.Sprintf(`foo%d%s`, i, v)), []string{name})
 		}
 	})
+
+	t.Run("title search will ignore terms < 3 characters", func(t *testing.T) {
+		index := newTestDashboardsIndex(t, threshold, 2, 2, noop)
+		indexDocumentsWithTitles(t, index, key, map[string]string{
+			"name1": "new dashboard",
+			"name2": "new dash",
+			"name3": "new",
+		})
+
+		// matches everything
+		checkSearchQuery(t, index, newTestQuery("new"), []string{"name3", "name2", "name1"})
+		// ignore terms shorter than 3 chars
+		checkSearchQuery(t, index, newTestQuery("new d"), []string{"name3", "name2", "name1"})
+		// include terms shorter that are exactly 3 chars
+		checkSearchQuery(t, index, newTestQuery("new das"), []string{"name2", "name1"})
+	})
 }
 
 func newTestQuery(query string) *resourcepb.ResourceSearchRequest {

pkg/storage/unified/search/custom_analyzers.go

@@ -10,6 +10,7 @@ import (
 )
 
 const TITLE_ANALYZER = "title_analyzer"
+const EDGE_NGRAM_MIN_TOKEN = 3.0
 
 func RegisterCustomAnalyzers(mapper *mapping.IndexMappingImpl) error {
 	return registerTitleAnalyzer(mapper)
@@ -22,7 +23,7 @@ func registerTitleAnalyzer(mapper *mapping.IndexMappingImpl) error {
 	// Define an N-Gram tokenizer (for substring search)
 	edgeNgramTokenFilter := map[string]interface{}{
 		"type": edgengram.Name,
-		"min":  3.0,
+		"min":  EDGE_NGRAM_MIN_TOKEN,
 		"max":  10.0,
 		"back": edgengram.FRONT,
 	}