Open
Description
In CBORGenerator.java
, there is a limit to the length of the string that can be output before the generator will use the indefinite length encoding. The reason behind this appears to be the complexity of determining the length of the encoded byte stream when it does not fit into an internal buffer, given that the length prefix needs to be output before the data.
For now, we have changed the implementation of _writeChunkedString(...)
to the following:
protected final void writeLongString(char[] text, int offset, int len) throws IOException {
CharBuffer charBuffer = CharBuffer.wrap(text, offset, len);
ByteBuffer byteBuffer = StandardCharsets.UTF_8.encode(charBuffer);
int byteLen = byteBuffer.remaining();
_writeLengthMarker(PREFIX_TYPE_TEXT, byteLen);
_writeBytes(byteBuffer.array(), byteBuffer.position(), byteLen);
}
It should also be possible to use the existing UTF-8 encoder to write the encoding out in pieces with the addition of a int encodedLength(...)
method to precompute and output the length.
Also, thanks for the awesome library!