-module(term_codec).
-export([encode/1, decode/1]).

%% Erlang-side term <-> binary codec, built on the substrate fixes from
%% commits 24e3bf53 (binary_to_list / list_to_binary), 3d80bd8c ($X char
%% literals), 4852cca9 (atom_to_list / integer_to_list charlists).
%%
%% Wire format (netstring-ish; all length headers ASCII decimal):
%%
%%   atom        $a Len $: NameBytes
%%   integer     $i Len $: DecimalBytes   (negative ints carry leading $-)
%%   binary      $b Len $: RawBytes
%%   tuple       $t Count $: Enc1 Enc2 ... Encn
%%   list        $l Count $: Enc1 Enc2 ... Encn      (proper list)
%%   nil         $l $0 $:                            (empty list)
%%
%% Each Enc is itself one of these forms — recursive. The format is
%% byte-clean: binary bodies may contain any byte (newlines, NULs, etc.),
%% so callers can frame entries with a 4-byte big-endian length prefix
%% (Step 3b on-disk segment writer's job).

%% encode/1: term -> binary
encode(T) when is_atom(T) ->
    Cs = atom_to_list(T),
    list_to_binary([$a, integer_to_list(length(Cs)), $:, Cs]);
encode(T) when is_integer(T) ->
    Cs = integer_to_list(T),
    list_to_binary([$i, integer_to_list(length(Cs)), $:, Cs]);
encode(T) when is_binary(T) ->
    list_to_binary([$b, integer_to_list(byte_size(T)), $:, T]);
encode(T) when is_tuple(T) ->
    L = tuple_to_list(T),
    list_to_binary([$t, integer_to_list(length(L)), $:,
                    [encode(E) || E <- L]]);
encode([]) ->
    list_to_binary([$l, $0, $:]);
encode(T) when is_list(T) ->
    list_to_binary([$l, integer_to_list(length(T)), $:,
                    [encode(E) || E <- T]]).

%% decode/1: binary -> {ok, Term, RestBinary} | {error, badform}
%% On success returns the remaining unconsumed bytes so callers can
%% stream-decode multiple frames from one buffer.
decode(B) when is_binary(B) ->
    decode_chars(binary_to_list(B)).

decode_chars([$a | Rest]) ->
    {Len, Rest1} = read_len(Rest, 0),
    Rest2 = strip_colon(Rest1),
    {NameChars, Rest3} = split_at(Len, Rest2),
    {ok, list_to_atom(NameChars), list_to_binary(Rest3)};
decode_chars([$i | Rest]) ->
    {Len, Rest1} = read_len(Rest, 0),
    Rest2 = strip_colon(Rest1),
    {NumChars, Rest3} = split_at(Len, Rest2),
    {ok, list_to_integer(NumChars), list_to_binary(Rest3)};
decode_chars([$b | Rest]) ->
    {Len, Rest1} = read_len(Rest, 0),
    Rest2 = strip_colon(Rest1),
    {Bytes, Rest3} = split_at(Len, Rest2),
    {ok, list_to_binary(Bytes), list_to_binary(Rest3)};
decode_chars([$t | Rest]) ->
    {N, Rest1} = read_len(Rest, 0),
    Rest2 = strip_colon(Rest1),
    {Elems, Rest3} = decode_n(N, Rest2, []),
    {ok, list_to_tuple(Elems), list_to_binary(Rest3)};
decode_chars([$l | Rest]) ->
    {N, Rest1} = read_len(Rest, 0),
    Rest2 = strip_colon(Rest1),
    {Elems, Rest3} = decode_n(N, Rest2, []),
    {ok, Elems, list_to_binary(Rest3)};
decode_chars(_) ->
    {error, badform}.

read_len([C | Rest], Acc) when C >= $0, C =< $9 ->
    read_len(Rest, Acc * 10 + C - $0);
read_len([$- | Rest], 0) ->
    %% Leading minus for negative integer-body lengths is invalid for
    %% lengths, but appears inside integer-body bytes (handled in
    %% the body, not here — read_len only consumes digits before $:).
    {0, [$- | Rest]};
read_len(Rest, Acc) ->
    {Acc, Rest}.

strip_colon([$: | Rest]) -> Rest;
strip_colon(Other) -> erlang:error({badform, Other}).

split_at(0, Rest) -> {[], Rest};
split_at(N, [H | T]) ->
    {Hs, Tl} = split_at(N - 1, T),
    {[H | Hs], Tl};
split_at(_, []) ->
    erlang:error({badform, short}).

decode_n(0, Rest, Acc) ->
    {lists:reverse(Acc), Rest};
decode_n(N, Bytes, Acc) ->
    {Term, Rest} = decode_one(Bytes),
    decode_n(N - 1, Rest, [Term | Acc]).

decode_one(Bytes) ->
    case decode_chars(Bytes) of
        {ok, Term, RestBin} -> {Term, binary_to_list(RestBin)};
        {error, R} -> erlang:error({badform, R})
    end.