compress

#!/usr/bin/env escript

%% compress -- compress standard input
%%
%% BUGS
%% The escape symbol should be passed down instead of written
%% directly on Pattern Matching.
%% The problem is that we can't call a function inside Pattern,
%% and I didn't wanted to increase the function arity only for
%% a constant value.
%%
%% Runs longer than 26 character aren't being broken into sev.
%% shorter ones, but this should be easy to correct.
compress(Input) ->
    compress(Input, [], 0, []).

compress([], Char, Count, Acc) ->
    StateList = putrep(Char, Count),
    Result = lists:concat([Acc, StateList]),
    ststd:putf("~s", [Result]);
compress([126|T], Char, Count, Acc) ->
                                                % ~ it's a run of itself of lenght 1
    StateList = putrep(Char, Count),
    compress(T, [], 0, lists:concat([Acc, StateList, [126, 65, 126]]));
compress([H|T], Char, Count, Acc) when H =/= Char ->
    compress(T, H, 1, lists:concat([Acc, putrep(Char, Count)]));
compress([_|T], Char, Count, Acc) ->
    compress(T, Char, Count+1, Acc).

%% putrep -- put out representations of run of n 'c's
%%
%% BUGS
%% This one isn't a exactly copy of the Pascal version.
%% This should use a constant instead of the escaping value 126.
putrep(Char, Count) ->
    case Count >= thresh() of
        true -> [126, init_symbol() + Count, Char];
        false -> lists:map(fun(_) -> Char end, lists:seq(1, Count))
    end.

%% init_symbol -- get the firs symbol for count repetitions
%%
%% This is a abstraction instead of using letter 'A' directly.
init_symbol() ->
    Symbol = 64,  % ASCII 64 = @, next one is 'A'
    Symbol.

%% thresh -- return the threshold for repetitions to compress
thresh() ->
    Threshold = 4,
    Threshold.

main([]) ->
    Input = ststd:getf(),
    compress(Input);
main(Arg) ->
    {ok, Bin} = file:read_file(Arg),
    File = erlang:binary_to_list(Bin),
    compress(File).