ebb/sample/word_split.erl

-module(word_split).
-export([words/1, to_list/1]).

%%-include_lib("src/ebb_flow.erl").

% data WordState = {chunk, S} | {segment, L, C, R}

nil() ->
    nil.

singleton(S) ->
    {single, S}.

conc(A, B) ->
    {conc, A, B}.

to_list(nil) ->
    [];
to_list({single, S}) ->
    [S];
to_list({conc, A, B}) ->
    to_list(A) ++ to_list(B).

maybe_word("") ->
    nil();
maybe_word(S) ->
    singleton(S).

process_char(C) ->
    case lists:member(C, " \t\n\f\l") of
	true ->
	    {segment, "", nil(), ""};
	false ->
	    {chunk, [C]}
    end.

combine({chunk, S1}, {chunk, S2}) ->
    {chunk, S1 ++ S2};
combine({chunk, S}, {segment, L, C, R}) ->
    {segment, S ++ L, C, R};
combine({segment, L, C, R}, {chunk, S}) ->
    {segment, L, C, R ++ S};
combine({segment, L1, C1, R1}, {segment, L2, C2, R2}) ->
    {segment, L1, conc(C1, conc(maybe_word(R1 ++ L2), C2)), R2}.

head([X|_]) ->
    X.

words(Str) ->
    T = ebb_flow:map_reduce(fun process_char/1,
			    fun combine/2,
			    Str),
    case head(ebb_run:run_linear(T)) of
	{chunk, S} ->
	    maybe_word(S);
	{segment, L, C, R} ->
	    conc(maybe_word(L), conc(C, maybe_word(R)));
	_ ->
	    error
    end.