strconv: detect invalid UTF-8 in the Unquote fast path

Fixes #23685

Change-Id: I3625bd01f860077ee0976df9e3dfb66754804bcd
Reviewed-on: https://go-review.googlesource.com/92535
Reviewed-by: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

src/strconv/quote.go

@@ -385,7 +385,9 @@ func Unquote(s string) (string, error) {
 	if !contains(s, '\\') && !contains(s, quote) {
 		switch quote {
 		case '"':
-			return s, nil
+			if utf8.ValidString(s) {
+				return s, nil
+			}
 		case '\'':
 			r, size := utf8.DecodeRuneInString(s)
 			if size == len(s) && (r != utf8.RuneError || size != 1) {

src/strconv/quote_test.go

@@ -326,6 +326,36 @@ func TestUnquote(t *testing.T) {
 	}
 }
 
+// Issue 23685: invalid UTF-8 should not go through the fast path.
+func TestUnquoteInvalidUTF8(t *testing.T) {
+	tests := []struct {
+		in string
+
+		// one of:
+		want    string
+		wantErr string
+	}{
+		{in: `"foo"`, want: "foo"},
+		{in: `"foo`, wantErr: "invalid syntax"},
+		{in: `"` + "\xc0" + `"`, want: "\xef\xbf\xbd"},
+		{in: `"a` + "\xc0" + `"`, want: "a\xef\xbf\xbd"},
+		{in: `"\t` + "\xc0" + `"`, want: "\t\xef\xbf\xbd"},
+	}
+	for i, tt := range tests {
+		got, err := Unquote(tt.in)
+		var gotErr string
+		if err != nil {
+			gotErr = err.Error()
+		}
+		if gotErr != tt.wantErr {
+			t.Errorf("%d. Unquote(%q) = err %v; want %q", i, tt.in, err, tt.wantErr)
+		}
+		if tt.wantErr == "" && err == nil && got != tt.want {
+			t.Errorf("%d. Unquote(%q) = %02x; want %02x", i, tt.in, []byte(got), []byte(tt.want))
+		}
+	}
+}
+
 func BenchmarkUnquoteEasy(b *testing.B) {
 	for i := 0; i < b.N; i++ {
 		Unquote(`"Give me a rock, paper and scissors and I will move the world."`)