LibTextCodec: Change UTF-8's decoder to replace invalid code points

The UTF-8 decoder will currently crash if it is provided invalid UTF-8
input. Instead, change its behavior to match that of all other decoders
to replace invalid code points with U+FFFD. This is required by the web.
This commit is contained in:
Timothy Flynn 2023-05-11 11:22:41 -04:00 committed by Andreas Kling
parent cb32bb3502
commit 00fa23237a
Notes: sideshowbarker 2024-07-16 21:51:02 +09:00

View file

@ -213,10 +213,12 @@ ErrorOr<String> convert_input_to_utf8_using_given_decoder_unless_there_is_a_byte
VERIFY(actual_decoder);
// FIXME: 3. Process a queue with an instance of encodings decoder, ioQueue, output, and "replacement".
// This isn't the exact same as the spec, especially the error mode of "replacement", which we don't have the concept of yet.
// 3. Process a queue with an instance of encodings decoder, ioQueue, output, and "replacement".
// FIXME: This isn't the exact same as the spec, which is written in terms of I/O queues.
auto output = TRY(actual_decoder->to_utf8(input));
// 4. Return output.
return actual_decoder->to_utf8(input);
return output;
}
ErrorOr<String> Decoder::to_utf8(StringView input)
@ -242,7 +244,7 @@ ErrorOr<String> UTF8Decoder::to_utf8(StringView input)
bomless_input = input.substring_view(3);
}
return String::from_utf8(bomless_input);
return Decoder::to_utf8(bomless_input);
}
ErrorOr<void> UTF16BEDecoder::process(StringView input, Function<ErrorOr<void>(u32)> on_code_point)