same_string.erl

-module(same_string).

-export([test/0,
         is_re_message_the_same/2,
         is_message_the_same/2]).
         
%% XXX - make sure this works with unicode
non_word() ->
    [$\n, $\r, $\t,$!,$",$#,$$,$%,$&,$',$(,
     $),$*, $+,$,, $\\, $-,
     $., $/, $:,$;,$<,$=,$>,$?,$@,$[,$],$^,
     $_,$`,${,$|,$},$~].

is_re_message_the_same(Msg, Msg2) ->
    A = re:replace(Msg,"([[:punct:]]+|\\s+)", "", [{return, binary}, global, unicode]),
    B = re:replace(Msg2,"([[:punct:]]+|\\s+)", "", [{return, binary}, global, unicode]),
    nomatch =/= re:run(A, <<"^", B/binary,"$">>, [caseless, unicode]).

is_message_the_same(Msg1, Msg2) ->
    NewMsg1 = replace(Msg1),
    NewMsg2 = replace(Msg2),
    string:equal(NewMsg1, NewMsg2).

replace(Str) ->
    string:to_lower(string:join(string:tokens(Str, non_word() ++" "), "")).

test() ->
    TestCount = 100000,
    Val1 = timer:tc(fun () ->
                     lists:foreach(fun(_Word) ->
                                           is_message_the_same("test", "Test")
                                   end,
                                   lists:seq(1, TestCount))
             end),
    io:format("First test ~p ~n", [Val1]),
    Val2 = timer:tc(fun () ->
                     lists:foreach(fun(_Word) ->
                                           is_re_message_the_same("test", "Test")
                                   end,
                                   lists:seq(1, TestCount))
             end),
    io:format("Second test ~p ~n", [Val2]),

    ok.