ebb/src/word_split.erl

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

% 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) ->
    io:format("~p: process_char start~n", [node()]),
    {_, _, Start} = erlang:now(),
    Result = case lists:member(C, " \t\n\f") of
		 true ->
		     {segment, "", nil(), ""};
		 false ->
		     {chunk, [C]}
	     end,
    {_, _, Finish} = erlang:now(),
    io:format("~p: process_char finish, time ~p~n", [node(), Finish-Start]),
    Result.

combine_timed(Arg1, Arg2) ->
    io:format("~p: combine start~n", [node()]),
    {_, _, Start} = erlang:now(),
    Result = combine(Arg1, Arg2),
    {_, _, Finish} = erlang:now(),
    io:format("~p: combine finish, time ~p~n", [node(), Finish-Start]),
    Result.
    
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}.

words(Str) ->
    T = ebb_flow:map_reduce(fun(X) -> process_char(X) end,
			    fun(X, Y) -> combine_timed(X, Y) end,
			    Str),
    case ebb_run:run_distributed(T) of
	{ok, [{chunk, S}]} ->
	    to_list(maybe_word(S));
	{ok, [{segment, L, C, R}]} ->
	    to_list(conc(maybe_word(L), conc(C, maybe_word(R))));
	_ ->
	    error
    end.

words_from_file(Str) ->
    case file:read_file(Str) of
	{ok, Data} ->
	    io:format("Number of words: ~w\n", [length(words(binary_to_list(Data)))]);
	_ ->
	    io:format("File not found!")
    end.