💾 Archived View for gemini.ucant.org › notes › tmp-misfin-parser.txt captured on 2024-02-05 at 09:36:50.
-=-=-=-=-=-=-
-module(miselfin_parse).
-include_lib("eunit/include/eunit.hrl").
-include("types.hrl").
-export([parse_request/1]).
-type header_tag() :: env_recipient | recipient | sender | timestamp.
-type header() :: {header_tag(), binary(), blurb()}.
-type misfin_b_result() :: {misfin_b, [header()], binary()}.
-type result() ::
misfin_b_result() |
empty |
no_mailbox |
wrong_protocol |
missing_footer |
missing_body |
incomplete.
-spec parse_request(Req :: binary()) ->
{ok, misfin_request()} | incomplete | {error, atom()}.
parse_request(Req) ->
case request(Req) of
{misfin_b, Headers, Body} ->
Result = full_request(Headers, Body),
{ok, Result};
incomplete ->
incomplete;
E ->
{error, E}
end.
-spec request(Req :: binary()) -> result().
request(<<"">>) ->
empty;
request(<<"misfin://", Rest/binary>>) ->
mailbox(Rest, <<"">>);
request(Req) when is_binary(Req) ->
wrong_protocol.
-spec mailbox(Frag :: binary(),
Acc :: binary()) -> result().
mailbox(<<"">>, <<"">>) ->
incomplete;
mailbox(<<"@", _Rest/binary>>, <<"">>) ->
no_mailbox;
mailbox(<<"@", Rest/binary>>, Mailbox) ->
host(Rest, <<"">>, {mailbox, Mailbox});
mailbox(<<Byte:8, Rest/binary>>, Acc) ->
mailbox(Rest, <<Acc/binary, Byte>>).
-spec host(Frag :: binary(),
Acc :: binary(),
Mailbox :: {mailbox, binary()})
-> result().
host(<<"">>, <<"">>, _) ->
incomplete;
host(<<"">>, _Acc, _) ->
missing_body;
host(<<" ", _Rest/binary>>, <<"">>, _) ->
incomplete;
host(<<" ", Rest/binary>>, Host, {mailbox, Mailbox}) ->
Rec = <<Mailbox/bytes, "@", Host/bytes>>,
Hdr = {env_recipient, Rec, no_blurb},
headers(Rest, no_header, [Hdr]);
host(<<Byte:8, Rest/binary>>, Acc, Mailbox) ->
host(Rest, <<Acc/binary, Byte>>, Mailbox).
% we'll assume that all the headers come at the top of the message, for now,
% since in the Misfin C proposal, this is mandatory
% headers can start with:
% "<" - sender
% ":" - recipient
% "@" - timestamp
-spec headers(Frag :: binary(),
Acc :: no_header | header(),
Headers :: [header()]) ->
result().
headers(<<"<", Rest/binary>>, no_header, Headers) ->
headers(Rest, {sender, <<"">>, no_blurb}, Headers);
headers(<<":", Rest/binary>>, no_header, Headers) ->
headers(Rest, {recipient, <<"">>, no_blurb}, Headers);
headers(<<"@", Rest/binary>>, no_header, Headers) ->
headers(Rest, {timestamp, <<"">>, no_blurb}, Headers);
headers(<<" ", Rest/binary>>, {Tag, <<"">>, Blurb}, Headers) ->
headers(Rest, {Tag, <<"">>, Blurb}, Headers);
headers(<<"\t", Rest/binary>>, {Tag, <<"">>, Blurb}, Headers) ->
headers(Rest, {Tag, <<"">>, Blurb}, Headers);
headers(<<"\r\n", Rest/binary>>, {Tag, Data, Blurb}, Headers) ->
headers(Rest, no_header, [{Tag, Data, Blurb} | Headers]);
headers(<<" ", Rest/binary>>, {Tag, Data, no_blurb}, Headers) ->
headers(Rest, {Tag, Data, <<"">>}, Headers);
headers(<<"\t", Rest/binary>>, {Tag, Data, no_blurb}, Headers) ->
headers(Rest, {Tag, Data, <<"">>}, Headers);
headers(<<" ", Rest/binary>>, {Tag, Data, <<"">>}, Headers) ->
headers(Rest, {Tag, Data, <<"">>}, Headers);
headers(<<"\t", Rest/binary>>, {Tag, Data, <<"">>}, Headers) ->
headers(Rest, {Tag, Data, <<"">>}, Headers);
headers(<<Byte:8, Rest/binary>>, {Tag, Data, no_blurb}, Headers) ->
headers(Rest, {Tag, <<Data/binary, Byte>>, no_blurb}, Headers);
headers(<<Byte:8, Rest/binary>>, {Tag, Data, Blurb}, Headers) ->
headers(Rest, {Tag, Data, <<Blurb/binary, Byte>>}, Headers);
headers(<<"">>, {_, _, _}, _Headers) ->
incomplete; % header does not end in CRLF?
headers(Rest, no_header, Headers) ->
body(Rest, <<"">>, Headers).
-spec body(Rest :: binary(),
Acc :: binary(),
Headers :: [header()]) ->
result().
body(<<"\r\n">>, Acc, Headers) ->
{misfin_b, lists:reverse(Headers), Acc};
body(<<"">>, _, _) ->
missing_footer;
body(<<Byte:8, Rest/binary>>, Acc, Headers) ->
body(Rest, <<Acc/binary, Byte>>, Headers).
%% TESTS %%
heterogeneous_test_() ->
Tests = [
{empty, <<"">>},
{wrong_protocol, <<"foo://">>},
{incomplete, <<"misfin://">>},
{incomplete, <<"misfin://username@">>},
{incomplete, <<"misfin://username@ ">>},
{no_mailbox, <<"misfin://@foo.com s y zar\r\n">>},
{missing_footer, <<"misfin://foo@bar.com ">>},
{missing_body, <<"misfin://martin@misfin.ucant.org">>},
{incomplete, <<"misfin://u@h.io < billg@ms.com">>},
{missing_footer, <<"misfin://username@host.com rest">>},
{
{misfin_b,[{env_recipient,<<"martin@misfin.ucant.org">>,
no_blurb},
{sender,<<"foo@bar.com">>,<<"Foo man choo">>},
{recipient,<<"martin@no.ucant.org">>,<<"Mk">>}],
<<"the message">>},
<<"misfin://martin@misfin.ucant.org < foo@bar.com\t Foo man choo\r\n:martin@no.ucant.org Mk\r\nthe message\r\n">>}
],
[?_assertEqual(Expected, request(Input)) ||
{Expected, Input} <- Tests].
% Given a list of 2-tuples `List`, get the cdr of those whose car matches `K`.
%
% this fn probably could have been done with list comprehensions
% or proplists:get_value?
matching(K, List) ->
lists:filtermap(fun(Elt) -> case Elt of
{K, V1, V2} -> {true, {V1, V2}};
_ -> false
end
end, List).
-spec full_request(Headers :: [header()],
Body :: binary()) ->
misfin_request().
full_request(Headers, Body) ->
ERs = matching(env_recipient, Headers),
Recipients = matching(recipient, Headers),
Senders = matching(sender, Headers),
RawTimestamps = matching(timestamp, Headers),
[{ER, _}] = ERs,
Timestamps = lists:filtermap(
fun({Timestamp, _}) ->
try iso8601:parse(Timestamp) of
{Date, Time} -> {true, {Date, Time}}
catch
error:badarg -> false
end
end, RawTimestamps),
#misfin_request{
envelope_recipient = ER,
recipients = Recipients,
senders = Senders,
timestamps = Timestamps,
body = Body
}.