Gettext .mo format parser for Erlang

mo_parser.erl

%%% @author Sergey Prokhorov <[email protected]>
%%% @copyright (C) 2013, Sergey Prokhorov
%%% @doc
%%% Simple gettext .mo file format parser for Erlang.
%%% 
%%% Produce [{KeyPlurals::[binary()], TransPlurals::[binary()]}] orddict as
%%% output.
%%% Eg, for .po file (converted to .mo)
%%% <pre>
%%% msgid "Download"
%%% msgstr "Скачать"
%%% 
%%% msgid "Stone"
%%% msgid_plural "Stones"
%%% msgstr[0] "Камень"
%%% msgstr[1] "Камня"
%%% msgstr[2] "Камней"
%%% </pre>
%%% it will produce
%%% <pre><code>
%%% [{[<<"Download">>], [<<"Скачать">>]},
%%%  {[<<"Stone">>, <<"Stones">>], [<<"Камень">>, <<"Камня">>, <<"Камней">>]}]
%%% </code></pre>
%%% TODO: simple MIME parser (for key "" - translation metadata)
%%% TODO: plural form expression interpreter
%%% @end
%%% Created :  3 Sep 2013 by Sergey Prokhorov <[email protected]>

-module(mo_parser).
-export([parse/1, to_dict/1]).

-record(st,
        {bin :: binary(),
         obin :: binary(),
         catalog=[] :: [{[binary()], [binary()]}],
         bo :: little | big,
         version :: integer(),
         msg_cnt :: integer(),
         orig_tab_offset :: integer(),
         trans_tab_offset :: integer()}).


parse(Name) when is_list(Name) ->
    {ok, Bin} = file:read_file(Name),
    parse(Bin);
parse(Bin) when is_binary(Bin) ->
    State = #st{bin=Bin, obin=Bin},
    State2 = parse_magick(State),
    State3 = parse_meta(State2),
    parse_catalog(State3, 0).

to_dict(#st{catalog=Catalog}) ->
    Catalog.

parse_magick(#st{bin = <<16#950412de:32/little, Ver:32/little, Rest/binary>>} = S) ->
    S#st{bo=little, version=Ver, bin=Rest};
parse_magick(#st{bin = <<16#950412de:32/big, Ver:32/big, Rest/binary>>} = S) ->
    S#st{bo=big, version=Ver, bin=Rest}.


parse_meta(#st{bo=little, bin = <<MsgCnt:32/little, OrigTabOffset:32/little,
                                  TransTabOffset:32/little, Rest/binary>>} = S) ->
    S#st{msg_cnt = MsgCnt, orig_tab_offset = OrigTabOffset,
         trans_tab_offset = TransTabOffset, bin=Rest};
parse_meta(#st{bo=big, bin = <<MsgCnt:32/big, OrigTabOffset:32/big,
                               TransTabOffset:32/big, Rest/binary>>} = S) ->
    S#st{msg_cnt = MsgCnt, orig_tab_offset = OrigTabOffset,
         trans_tab_offset = TransTabOffset, bin=Rest}.


parse_catalog(#st{msg_cnt=N, catalog=Cat} = S, N) ->
    S#st{catalog=lists:reverse(Cat)};
parse_catalog(#st{orig_tab_offset=OrigO, trans_tab_offset=TransO,
                  obin=Bin, bo=Bo, catalog=Catalog} = S, N) ->
    Orig = get_string(N, OrigO, Bin, Bo),
    Trans = get_string(N, TransO, Bin, Bo),
    NewCatalog = [{Orig, Trans} | Catalog],
    parse_catalog(S#st{catalog=NewCatalog}, N + 1).

get_string(N, O, Bin, little) ->
    O1 = O + 8 * N,
    <<_:O1/binary, Len:32/little, StringO:32/little, _/binary>> = Bin,
    get_strings(StringO, Len, Bin);
get_string(N, O, Bin, big) ->
    O1 = O + 8 * N,
    <<_:O1/binary, Len:32/big, StringO:32/big, _/binary>> = Bin,
    get_strings(StringO, Len, Bin).

get_strings(StringO, Len, Bin) ->
    %% split by \0 to plural forms
    <<_:StringO/binary, String:Len/binary, _/binary>> = Bin,
    binary:split(String, [<<0>>], [global]).

This module was later included in https://github.com/seriyps/gettexter