Kaitai ETF (Erlang's External Term Fomat)

etf.ksy

meta:
  id: etf
  file-extension: etf
  title: Erlang's External Term Format
  endian: be
doc-ref: https://erlang.org/doc/apps/erts/erl_ext_dist.html
seq:
  - id: messages
    type: message
    repeat: eos
types:
  empty: {}
  message:
    seq:
      - id: magic
        contents: [131]
      - id: term
        type: term
  term:
    seq:
      - id: tag
        type: u1
        enum: term_type
      - id: data
        type:
          switch-on: tag
          cases:
            term_type::map: map
            term_type::small_atom_latin1: atom(1, false)
            term_type::atom_latin1: atom(2, false)
            term_type::small_atom_utf8: atom(1, true)
            term_type::atom_utf8: atom(2, true)
            term_type::list: list
            term_type::binary: binary
            term_type::nil: empty
            term_type::int: s4
            term_type::small_int: u1
            term_type::small_bignum: bignum(1)
            term_type::large_bignum: bignum(4)
            _: unimplemented
    -webide-representation: "{tag} {data:dec}"
  bignum:
    instances:
      sign_char:
        value: "sign == 0 ? \"+\" : \"-\""
    params:
      - id: length_len
        type: u1
    seq:
      - id: length
        type:
          switch-on: length_len
          cases:
            1: u1
            4: u4
        valid:
          max: 999
      - id: sign
        type: u1
        valid:
          max: 1
      - id: data
        size: length
    -webide-representation: "{sign_char}{data}"
  binary:
    instances:
      text:
        value: "data.to_s(\"latin1\")"
      repr:
        value: "length < 30 ? \" \\\"\"+text+\"\\\"\" : \"\""
    seq:
      - id: length
        type: u4
      - id: data
        size: length
    -webide-representation: "<{length:dec} bytes>{repr}"
  list:
    instances:
      tail:
        value: elements.last
    seq:
      - id: length
        type: u4
      - id: elements
        type: term
        repeat: expr
        repeat-expr: "length > 999 ? 0 : length+1"
    -webide-representation: "[{elements}]"
  map:
    seq:
      - id: arity
        type: u4
      - id: pairs
        type: pair
        repeat: expr
        repeat-expr: "arity > 999 ? 0 : arity"
    -webide-representation: "<{arity:dec} pairs>"
  pair:
    seq:
      - id: key
        type: term
      - id: value
        type: term
    -webide-representation: "{key}: {value}"
  atom:
    params:
      - id: length_len
        type: u8
      - id: is_utf8
        type: b1
    instances:
      text:
        value: "is_utf8 ? data.to_s(\"utf8\") : data.to_s(\"latin1\")"
    seq:
      - id: length
        type:
          switch-on: length_len
          cases:
            1: u1
            2: u2
      - id: data
        size: length
    -webide-representation: "{text}"
  unimplemented:
    -webide-representation: UNIMPLEMENTED!
enums:
  term_type:
    82: atom_cache_ref_idx
    97: small_int
    98: int
    99: float_str
    102: port
    89: new_port
    120: v4_port
    103: pid
    88: new_pid
    104: small_tuple
    105: large_tuple
    116: map
    106: nil
    107: string
    108: list
    109: binary
    110: small_bignum
    111: large_bignum
    101: reference
    114: new_reference
    90: newer_reference
    117: function
    112: new_function
    113: export
    77: bit_binary
    70: new_float
    118: atom_utf8
    119: small_atom_utf8
    100: atom_latin1
    115: small_atom_latin1