Fix Unicode literals in textual bytes and fixed values

bytes.go

@@ -118,10 +118,13 @@ func bytesNativeFromTextual(buf []byte) (interface{}, []byte, error) {
 				// necessarily represent text. Therefore, Avro bytes are not
 				// encoded in UTF-16. Each \u is followed by 4 hexadecimal
 				// digits, the first and second of which must be 0.
-				v, err := parseUint64FromHexSlice(buf[i+3 : i+5])
+				v, err := parseUint64FromHexSlice(buf[i+1 : i+5])
 				if err != nil {
 					return nil, nil, fmt.Errorf("cannot decode textual bytes: %s", err)
 				}
+				if v > 255 {
+					return nil, nil, fmt.Errorf("cannot decode textual bytes: code point out of range: U+%04X", v)
+				}
 				i += 4 // absorb 4 characters: one 'u' and three of the digits
 				newBytes = append(newBytes, byte(v))
 				continue
@@ -136,6 +139,15 @@ func bytesNativeFromTextual(buf []byte) (interface{}, []byte, error) {
 		if b == '"' {
 			return newBytes, buf[i+1:], nil
 		}
+		if b >= utf8.RuneSelf {
+			r, width := utf8.DecodeRune(buf[i:])
+			if r == utf8.RuneError || r > 255 {
+				return nil, nil, fmt.Errorf("cannot decode textual bytes: invalid byte code point: %U", r)
+			}
+			newBytes = append(newBytes, byte(r))
+			i += width - 1
+			continue
+		}
 		newBytes = append(newBytes, b)
 	}
 	return nil, nil, fmt.Errorf("cannot decode textual bytes: expected final \"; found: %#U", buf[buflen-1])

bytes_test.go

@@ -199,3 +199,50 @@ func TestStringCodecAcceptsBytes(t *testing.T) {
 		testTextEncodePass(t, schema, []byte("abcd"), []byte(`"abcd"`))
 	})
 }
+
+func TestBytesTextUnicodeCodePoints(t *testing.T) {
+	codec, err := NewCodec(`"bytes"`)
+	if err != nil {
+		t.Fatal(err)
+	}
+	for value := 0; value <= 255; value++ {
+		text, err := json.Marshal(string(rune(value)))
+		if err != nil {
+			t.Fatal(err)
+		}
+		decoded, remaining, err := codec.NativeFromTextual(append(text, ' '))
+		if err != nil {
+			t.Fatalf("code point %d: %v", value, err)
+		}
+		got := decoded.([]byte)
+		if len(got) != 1 || got[0] != byte(value) || string(remaining) != " " {
+			t.Errorf("code point %d: got %v with remaining %q", value, got, remaining)
+		}
+	}
+}
+
+func TestBytesTextRejectsOutOfRangeCodePoints(t *testing.T) {
+	for _, text := range []string{`"\u0100"`, `"\uffff"`, `"\uGGff"`, `"Ā"`, `"😂"`, "\"\xff\""} {
+		_, _, err := bytesNativeFromTextual([]byte(text))
+		if err == nil {
+			t.Errorf("expected error for %q", text)
+		}
+	}
+}
+
+func TestFixedTextUnicodeCodePoint(t *testing.T) {
+	codec, err := NewCodec(`{"type":"fixed","name":"single","size":1}`)
+	if err != nil {
+		t.Fatal(err)
+	}
+	for _, text := range []string{`"ÿ"`, `"\u00ff"`} {
+		decoded, rest, err := codec.NativeFromTextual([]byte(text))
+		if err != nil {
+			t.Fatal(err)
+		}
+		value := decoded.([]byte)
+		if len(value) != 1 || value[0] != 255 || len(rest) != 0 {
+			t.Fatalf("%s decoded to %v, remaining %q", text, value, rest)
+		}
+	}
+}