Fun with Elixir bit bashing and weird Microsoft serialization formats

parse.ex

defmodule OcrMap.Parse do
  use Bitwise, only_operators: true

  def parse_7bit_encoded_int(rest) do
    parse_7bit_encoded_int(rest, 0, 0)
  end

  def parse_7bit_encoded_int(<< 0::1, bits::7, rest::binary >>, shift, acc) do
    { (bits <<< shift) + acc, rest }
  end

  def parse_7bit_encoded_int(<< 1::1, bits::7, rest::binary >>, shift, acc) do
    parse_7bit_encoded_int(rest, shift+7, (bits <<< shift) + acc)
  end
end

defmodule OcrMap.ParseTest do
  use ExUnit.Case

  test "parse 7-bit encoded integers" do
    { i, rest } = OcrMap.Parse.parse_7bit_encoded_int(<< 0x1c, 0x62, 0x00 >>)
    assert i == 28
    assert rest == << 0x62, 0x00 >>

    { i, rest } = OcrMap.Parse.parse_7bit_encoded_int(<< 0xff, 0x01, 0x00 >>)
    assert i == 255
    assert rest == << 0x00 >>
  end
end