# Adapted from the C# implementation const init_upper_mask::UInt8 = 0xFE function get_transmit_sequence(message) transmit_seq = Vector{UInt8}() carry::UInt8 = 0 upper_mask::UInt8 = init_upper_mask for message_byte in message if upper_mask == UInt8(0) # The carry now contains 7 bits. Flush it out. push!(transmit_seq, add_parity(carry)) carry = UInt8(0) upper_mask = init_upper_mask end shift_places = trailing_zeros(upper_mask) current = (carry << (8 - shift_places)) | (message_byte >>> shift_places) push!(transmit_seq, add_parity(current)) # Update parameters for next round carry = UInt8(message_byte & (~upper_mask)) # Shorten the upper mask upper_mask <<= 1 end if upper_mask != init_upper_mask last_group::UInt8 = UInt8(carry << count_ones(upper_mask)) # We have left over carry data push!(transmit_seq, add_parity(last_group)) end transmit_seq end add_parity(source::UInt8) = iseven(count_ones(source & 0x7F)) ? source << 1 : (source << 1) | 1 function decode_sequence(received_seq) length(received_seq) > 0 || return [] decoded_message = Vector{UInt8}() byte_to_add::UInt8 = 0x00 upper_mask::UInt8 = 0xFF for received_byte in received_seq if upper_mask == UInt8(0xFF) # This round is complete # Current byte too short byte_to_add = get_byte_data(received_byte) upper_mask = 0x80 continue end current_byte_data::UInt8 = get_byte_data(received_byte) contribution::UInt8 = current_byte_data >>> trailing_zeros(upper_mask) push!(decoded_message, byte_to_add | contribution) # Update parameters for next round byte_to_add = (current_byte_data & ~(upper_mask | 0x01)) << count_ones(upper_mask) upper_mask = (upper_mask >> 1) | 0x80 end decoded_message end function get_byte_data(data) iseven(count_ones(data::UInt8)) || error("Byte has incorrect parity") UInt8(data & 0xFE) end