-module(plagiarism_check).
-export([plagiarism_check/2]).

%% Re-derived as a token-isomorphism check instead of porting PHP's
%% placeholder/regex substitution dance: tokenize both snippets, build a
%% one-way rename map from the mismatched, non-numeric tokens of Code1 onto
%% Code2, then verify applying that map to every token of Code1 reproduces
%% Code2 exactly.
plagiarism_check(Code1, Code2) ->
    C1 = string:join(Code1, " "),
    C2 = string:join(Code2, " "),
    case C1 =:= C2 of
        true ->
            false;
        false ->
            D1 = tokenize(C1),
            D2 = tokenize(C2),
            Map = lists:foldl(fun({T1, T2}, Acc) ->
                case T1 =/= T2 andalso not is_numeric_token(T1) of
                    true -> maps:put(T1, T2, Acc);
                    false -> Acc
                end
            end, #{}, lists:zip(D1, D2)),
            [maps:get(T, Map, T) || T <- D1] =:= D2
    end.

tokenize(S) ->
    case re:run(S, "[A-Za-z0-9_]+", [global, {capture, all, list}]) of
        {match, Matches} -> [Tok || [Tok] <- Matches];
        nomatch -> []
    end.

is_numeric_token(T) ->
    case string:to_integer(T) of
        {_, ""} -> true;
        _ ->
            case string:to_float(T) of
                {_, ""} -> true;
                _ -> false
            end
    end.