Help to make (and understand) better and beautiful code

Questions / Help
,
1

Years ago i worte keccak_ex with only support for hash 256

Now I completed the support for hash 512 and 128 bytes

The “problem”
Both implementation are the same ( :upside_down_face: copy and paste) and just change

for hash 256

  @digest_length 32

  @block_length 200 - 2 * @digest_length

for hash 512

  @digest_length 64

  @block_length 200 - 2 * @digest_length

i thik something like an “interface/inheritance” can help with all the duplicated code making an isolation with the concret type but i cant find anything like that

the other problematic line is

defp process_block_decode_loop(%__MODULE__{} = value, index) when index < @block_length do

because you need the block_length to make the loop work…

i appretiate any idea/insight/suggestion to make the code looks better and to learn a better way

2

if i understood it correctly the struct that you’re using seems to be an internal abstraction. so why not internalize the digest_length and the block_length in the struct, this way you can always have it available on the function head by just destructing it and keep the same guard validations.

3

Are you sure you got all the references? This take(134) seems extremely suspicious - it’s two less than @block_length for @digest_length == 32 in 256-bit, but not changed for 512-bit where @digest_length is different.

4

Your right!!! also that pointed me to an edge case…

The correct version is:

  defp padding(%__MODULE__{} = value) do
    fix = if (value.input_length + 1) == byte_size(value.buffer) do
      value.buffer
      |> binary_part(0, value.input_length)
      |> set_value(<<0x81>>)
5

Nice one, already applied, thanks!!!