decoder2: add support for decoding utf-16 surrogates, produced by some JSON encoder implementations (Python, Java, C#) (#25193)

Larsimusrex · web-flow · commit 24f91280d9a5 · 2025-08-30T13:02:39.000+03:00
diff --git a/vlib/x/json2/decoder2/decode.v b/vlib/x/json2/decoder2/decode.v
@@ -412,11 +412,41 @@ fn (mut decoder Decoder) decode_value[T](mut val T) ! {
 							string_buffer << `\t`
 						}
 						`u` {
-							string_buffer << rune(strconv.parse_uint(decoder.json[
+							unicode_point := rune(strconv.parse_uint(decoder.json[
 								string_info.position + string_index..string_info.position +
-								string_index + 4], 16, 32)!).bytes()
+								string_index + 4], 16, 32)!)
 
 							string_index += 4
+
+							if unicode_point < 0xD800 { // normal utf-8
+								string_buffer << unicode_point.bytes()
+							} else if unicode_point >= 0xDC00 { // trail surrogate -> invalid
+								decoder.decode_error('Got trail surrogate: ${u32(unicode_point):04X} before head surrogate.')!
+							} else { // head surrogate -> treat as utf-16
+								if string_index > string_info.length - 6 {
+									decoder.decode_error('Expected a trail surrogate after a head surrogate, but got no valid escape sequence.')!
+								}
+								if decoder.json[string_info.position + string_index..
+									string_info.position + string_index + 2] != '\\u' {
+									decoder.decode_error('Expected a trail surrogate after a head surrogate, but got no valid escape sequence.')!
+								}
+
+								string_index += 2
+
+								unicode_point2 := rune(strconv.parse_uint(decoder.json[
+									string_info.position + string_index..string_info.position +
+									string_index + 4], 16, 32)!)
+
+								string_index += 4
+
+								if unicode_point2 < 0xDC00 {
+									decoder.decode_error('Expected a trail surrogate after a head surrogate, but got ${u32(unicode_point):04X}.')!
+								}
+
+								final_unicode_point := (unicode_point2 & 0x3FF) +
+									((unicode_point & 0x3FF) << 10) + 0x10000
+								string_buffer << final_unicode_point.bytes()
+							}
 						}
 						else {} // has already been checked
 					}
diff --git a/vlib/x/json2/decoder2/tests/decode_escaped_string_test.v b/vlib/x/json2/decoder2/tests/decode_escaped_string_test.v
@@ -10,3 +10,50 @@ fn test_decode_escaped_string() {
 
 	assert escaped_strings == decoded_strings
 }
+
+fn test_surrogate() {
+	assert decoder2.decode[string](r'"\ud83d\ude00"')! == '😀'
+	assert decoder2.decode[string](r'"\ud83d\ude00 text"')! == '😀 text'
+}
+
+fn test_invalid_surrogate() {
+	if x := decoder2.decode[string](r'"\ud83d"') {
+		assert false
+	} else {
+		if err is decoder2.JsonDecodeError {
+			assert err.line == 1
+			assert err.character == 1
+			assert err.message == 'Data: Expected a trail surrogate after a head surrogate, but got no valid escape sequence.'
+		}
+	}
+
+	if x := decoder2.decode[string](r'"\ud83d\n\n\n\n"') {
+		assert false
+	} else {
+		if err is decoder2.JsonDecodeError {
+			assert err.line == 1
+			assert err.character == 1
+			assert err.message == 'Data: Expected a trail surrogate after a head surrogate, but got no valid escape sequence.'
+		}
+	}
+
+	if x := decoder2.decode[string](r'"\ud83d\ud83d"') {
+		assert false
+	} else {
+		if err is decoder2.JsonDecodeError {
+			assert err.line == 1
+			assert err.character == 1
+			assert err.message == 'Data: Expected a trail surrogate after a head surrogate, but got D83D.'
+		}
+	}
+
+	if x := decoder2.decode[string](r'"\ude00\ud83d"') {
+		assert false
+	} else {
+		if err is decoder2.JsonDecodeError {
+			assert err.line == 1
+			assert err.character == 1
+			assert err.message == 'Data: Got trail surrogate: DE00 before head surrogate.'
+		}
+	}
+}
