[v11/forgejo] fix: skip empty tokens in SearchOptions.Tokens() (#8412)

backport of #8261 to v11

Co-authored-by: Danko Aleksejevs <danko@very.lv>
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/8412
Reviewed-by: Earl Warren <earl-warren@noreply.codeberg.org>
Co-authored-by: Shiny Nematoda <snematoda@noreply.codeberg.org>
Co-committed-by: Shiny Nematoda <snematoda@noreply.codeberg.org>

modules/indexer/issues/bleve/bleve.go

@@ -155,11 +155,12 @@ func (b *Indexer) Delete(_ context.Context, ids ...int64) error {
 func (b *Indexer) Search(ctx context.Context, options *internal.SearchOptions) (*internal.SearchResult, error) {
 	var queries []query.Query
 
-	if options.Keyword != "" {
 	tokens, err := options.Tokens()
 	if err != nil {
 		return nil, err
 	}
+
+	if len(tokens) > 0 {
 		q := bleve.NewBooleanQuery()
 		for _, token := range tokens {
 			innerQ := bleve.NewDisjunctionQuery(

modules/indexer/issues/elasticsearch/elasticsearch.go

@@ -148,12 +148,13 @@ func (b *Indexer) Delete(ctx context.Context, ids ...int64) error {
 func (b *Indexer) Search(ctx context.Context, options *internal.SearchOptions) (*internal.SearchResult, error) {
 	query := elastic.NewBoolQuery()
 
-	if options.Keyword != "" {
-		q := elastic.NewBoolQuery()
 	tokens, err := options.Tokens()
 	if err != nil {
 		return nil, err
 	}
+
+	if len(tokens) > 0 {
+		q := elastic.NewBoolQuery()
 		for _, token := range tokens {
 			innerQ := elastic.NewMultiMatchQuery(token.Term, "title", "content", "comments")
 			if token.Fuzzy {

modules/indexer/issues/internal/qstring.go

@@ -36,13 +36,10 @@ func (t *Tokenizer) next() (tk Token, err error) {
 
 	// skip all leading white space
 	for {
-		if r, _, err = t.in.ReadRune(); err == nil && r == ' ' {
-			//nolint:staticcheck,wastedassign // SA4006 the variable is used after the loop
-			r, _, err = t.in.ReadRune()
-			continue
-		}
+		if r, _, err = t.in.ReadRune(); err != nil || r != ' ' {
 			break
 		}
+	}
 	if err != nil {
 		return tk, err
 	}
@@ -98,12 +95,18 @@ nextEnd:
 
 // Tokenize the keyword
 func (o *SearchOptions) Tokens() (tokens []Token, err error) {
+	if o.Keyword == "" {
+		return nil, nil
+	}
+
 	in := strings.NewReader(o.Keyword)
 	it := Tokenizer{in: in}
 
 	for token, err := it.next(); err == nil; token, err = it.next() {
+		if token.Term != "" {
 			tokens = append(tokens, token)
 		}
+	}
 	if err != nil && err != io.EOF {
 		return nil, err
 	}

modules/indexer/issues/internal/qstring_test.go

@@ -41,6 +41,36 @@ var testOpts = []testIssueQueryStringOpt{
 			},
 		},
 	},
+	{
+		Keyword: "Hello  World",
+		Results: []Token{
+			{
+				Term:  "Hello",
+				Fuzzy: true,
+				Kind:  BoolOptShould,
+			},
+			{
+				Term:  "World",
+				Fuzzy: true,
+				Kind:  BoolOptShould,
+			},
+		},
+	},
+	{
+		Keyword: " Hello World ",
+		Results: []Token{
+			{
+				Term:  "Hello",
+				Fuzzy: true,
+				Kind:  BoolOptShould,
+			},
+			{
+				Term:  "World",
+				Fuzzy: true,
+				Kind:  BoolOptShould,
+			},
+		},
+	},
 	{
 		Keyword: "+Hello +World",
 		Results: []Token{
@@ -156,6 +186,68 @@ var testOpts = []testIssueQueryStringOpt{
 			},
 		},
 	},
+	{
+		Keyword: "\\",
+		Results: nil,
+	},
+	{
+		Keyword: "\"",
+		Results: nil,
+	},
+	{
+		Keyword: "Hello \\",
+		Results: []Token{
+			{
+				Term:  "Hello",
+				Fuzzy: true,
+				Kind:  BoolOptShould,
+			},
+		},
+	},
+	{
+		Keyword: "\"\"",
+		Results: nil,
+	},
+	{
+		Keyword: "\" World \"",
+		Results: []Token{
+			{
+				Term:  " World ",
+				Fuzzy: false,
+				Kind:  BoolOptShould,
+			},
+		},
+	},
+	{
+		Keyword: "\"\" World \"\"",
+		Results: []Token{
+			{
+				Term:  "World",
+				Fuzzy: true,
+				Kind:  BoolOptShould,
+			},
+		},
+	},
+	{
+		Keyword: "Best \"Hello World\" Ever",
+		Results: []Token{
+			{
+				Term:  "Best",
+				Fuzzy: true,
+				Kind:  BoolOptShould,
+			},
+			{
+				Term:  "Hello World",
+				Fuzzy: false,
+				Kind:  BoolOptShould,
+			},
+			{
+				Term:  "Ever",
+				Fuzzy: true,
+				Kind:  BoolOptShould,
+			},
+		},
+	},
 }
 
 func TestIssueQueryString(t *testing.T) {

modules/indexer/issues/internal/tests/tests.go

@@ -88,6 +88,11 @@ func TestIndexer(t *testing.T, indexer internal.Indexer) {
 	}
 }
 
+func allResults(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+	assert.Len(t, result.Hits, len(data))
+	assert.Equal(t, len(data), int(result.Total))
+}
+
 var cases = []*testIndexerCase{
 	{
 		Name:          "default",
@@ -97,6 +102,34 @@ var cases = []*testIndexerCase{
 			assert.Equal(t, len(data), int(result.Total))
 		},
 	},
+	{
+		Name: "empty keyword",
+		SearchOptions: &internal.SearchOptions{
+			Keyword: "",
+		},
+		Expected: allResults,
+	},
+	{
+		Name: "whitespace keyword",
+		SearchOptions: &internal.SearchOptions{
+			Keyword: "    ",
+		},
+		Expected: allResults,
+	},
+	{
+		Name: "dangling slash in keyword",
+		SearchOptions: &internal.SearchOptions{
+			Keyword: "\\",
+		},
+		Expected: allResults,
+	},
+	{
+		Name: "dangling quote in keyword",
+		SearchOptions: &internal.SearchOptions{
+			Keyword: "\"",
+		},
+		Expected: allResults,
+	},
 	{
 		Name: "empty",
 		SearchOptions: &internal.SearchOptions{