anagrams.erl

%% @author OJ Reeves <oj@buffered.io>
%%
%% My first crappy attempt at Dave Thomas' 6th Code Kata
%% (located at http://codekata.pragprog.com/2007/01/kata_six_anagra.html)
%%
%% Extra bits are now done.
%%
%% Refactored some small bits, and made changes based on suggestions from
%% the guys on Google Wave.

-module(anagrams).
-author('OJ Reeves <oj@buffered.io>').
-export([list/1]).

list(FileName) ->
  Words = file_to_list(FileName),
  Grouped = group(Words),
  Anagrams = lists:filter(fun({_, W}) -> length(W) > 1 end, Grouped),
  {_, Longest} = find_by(fun longest/2, Anagrams),
  {_, Biggest} = find_by(fun biggest/2, Anagrams),
  io:format("Anagrams List:~n", []),
  lists:map(fun({_, W}) -> io:format("~s~n", [to_string(W)]) end, Anagrams),
  io:format("Biggest Anagram Set: ~s (~p items)~n", [to_string(Biggest), length(Biggest)]),
  io:format("Longest Anagram: ~s (~p chars)~n", [to_string(Longest), length(hd(Longest))]).

to_string(List) ->
  string:join(List, " ").

longest_list({L1,R1}, {L2,R2}) ->
  Diff = length(L1) - length(L2),
  if
    Diff < 0 -> R2;
    true -> R1
  end.

longest(Elem={E, _}, Acc={A, _}) ->
  longest_list({E, Elem}, {A, Acc}).

biggest(Elem={_, E}, Acc={_, A}) ->
  longest_list({E, Elem}, {A, Acc}). 

find_by(Fun, [H|Anagrams]) ->
  lists:foldl(Fun, H, Anagrams).

group(Words) ->
  KeyVals = lists:map(fun(Word) -> {lists:sort(Word), Word} end, Words),
  Sorted = lists:keysort(1, KeyVals),
  Grouped = group_by(fun key_equals/2, Sorted),
  flatten(Grouped).

flatten(List) ->
  Fun = fun(SubList=[{H,_}|_]) -> {H, lists:map(fun({_, W}) -> W end, SubList)} end,
  lists:map(Fun, List).

key_equals({X, _}, {Y, _}) ->
  X =:= Y.

group_by(_Fun, []) ->
  [];
group_by(Fun, [H|T]) ->
  {Ys, Zs} = span(fun(X) -> Fun(X, H) end, T),
  Front = [H|Ys],
  Back = group_by(Fun, Zs),
  [Front|Back].

span(_Fun, []) ->
  {[], []};
span(Fun, List=[H|T]) ->
  case Fun(H) of
    true ->
      {Ys, Zs} = span(Fun, T),
      {[H|Ys], Zs};
    _ ->
      {[], List}
  end.

file_to_list(FileName) ->
  {ok, Bin} = file:read_file(FileName),
  string:tokens(binary_to_list(Bin), "\n").