From 5e559b540bce807177fa44d75a55423cbea3cc48 Mon Sep 17 00:00:00 2001
From: Peter Harpending <peter@harpending.org>
Date: Fri, 19 Sep 2025 22:24:38 -0700
Subject: [PATCH] wfc stuff

---
 src/wfc_utils.erl  | 10 ++++++
 src/wfc_wfchar.erl | 83 ++++++++++++++++++++++++++++++++++++++++++++++
 src/wfc_word.erl   | 82 +++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 175 insertions(+)
 create mode 100644 src/wfc_utils.erl
 create mode 100644 src/wfc_wfchar.erl
 create mode 100644 src/wfc_word.erl

diff --git a/src/wfc_utils.erl b/src/wfc_utils.erl
new file mode 100644
index 0000000..454494e
--- /dev/null
+++ b/src/wfc_utils.erl
@@ -0,0 +1,10 @@
+% @doc misc utility functions
+-module(wfc_utils).
+
+-export([err/2, emsg/2]).
+
+err(Fmt, Pat) ->
+    {error, emsg(Fmt, Pat)}.
+
+emsg(Fmt, Pat) ->
+    unicode:characters_to_list(io_lib:format(Fmt, Pat)).
diff --git a/src/wfc_wfchar.erl b/src/wfc_wfchar.erl
new file mode 100644
index 0000000..78b39e7
--- /dev/null
+++ b/src/wfc_wfchar.erl
@@ -0,0 +1,83 @@
+% @doc called wfchar to disambiguate from 
+%
+% mathematically, this is a variable like "a", "b", "c", etc
+-module(wfc_wfchar).
+
+-export_type([
+    wfchar/0
+]).
+
+-export([
+    validate/1,
+    from_binary/1, to_binary/1
+]).
+
+% @doc
+% a character is binary data that would be a valid atom
+%
+% [a-z][A-Za-z0-9_]+
+%
+% - leads with lowercase letter
+% - underscores ok
+% - numbers ok
+% - no non-ascii chars
+
+-opaque wfchar() :: {c, binary()}.
+
+%%-----------------------
+%% constructors/destructors
+%%-----------------------
+
+-spec validate(WfChar) -> Result
+    when WfChar :: wfchar(),
+         Result :: ok
+                 | {error, Reason :: string()}.
+% @doc validate if a candidate wfchar is well-formed and the inner binary is
+% legal
+
+validate({c, Binary}) ->
+    case from_binary(Binary) of
+        {ok, {c, X}} when X =:= Binary ->
+            ok;
+        Error ->
+            Error
+    end;
+validate(Term) ->
+    Error = wfc_utils:emsg("wfc_wfchar:validate failed; malformed wfchar: ~p", [Term]),
+    {error, Error}.
+
+
+-spec from_binary(Binary) -> Result
+    when Binary :: binary(),
+         Result :: {ok, wfchar()}
+                 | {error, Reason :: string()}.
+% @doc
+% make sure the binary is a legal wfchar and wrap it in a {c, Binary} tuple
+% @end
+
+%% initial char must be [a-z]
+from_binary(<<L:8, Rest/binary>>) when $a =< L, L =< $z ->
+    new2(<<L>>, Rest);
+from_binary(Char = <<L:8, _/binary>>) ->
+    {error, wfc_utils:emsg("wfc_wfchar:from_binary(~p): illegal lead character: ~p", [Char, L])};
+from_binary(Char) ->
+    wfc_utils:err("wfc_wfchar:from_binary(~p): malformed argument", [Char]).
+
+%% rest must be [A-Za-z0-9_]
+new2(Acc, <<L:8, Rest/binary>>) when       ($A =< L andalso L =< $Z)
+                                    orelse ($a =< L andalso L =< $z)
+                                    orelse ($0 =< L andalso L =< $9)
+                                    orelse (L =:= $_) ->
+    new2(<<Acc/binary, L>>, Rest);
+new2(Acc, <<>>) ->
+    {ok, {c, Acc}};
+new2(Acc, Rest = <<BadChar:8, _/binary>>) ->
+    WholeChar = <<Acc/binary, Rest/binary>>,
+    Error     = wfc_utils:emsg("wfc_wfchar:new2(~p, ~p): illegal character in wfchar: ~p; WholeChar: ~p", [Acc, Rest, BadChar, WholeChar]),
+    {error, Error}.
+
+
+
+-spec to_binary(wfchar()) -> binary().
+
+to_binary({c, X}) -> X.
diff --git a/src/wfc_word.erl b/src/wfc_word.erl
new file mode 100644
index 0000000..53bbb47
--- /dev/null
+++ b/src/wfc_word.erl
@@ -0,0 +1,82 @@
+% @doc a word is an ordset of wfchars
+%
+% multiplication is implied in a word
+%
+% empty word is 1
+%
+% anything times itself equals itself, so duplication is ignored
+-module(wfc_word).
+
+-export_type([
+    word/0
+]).
+
+-export([
+    %% constructors
+    one/0,
+    validate/1,
+    from_wfchars/1, to_wfchars/1,
+    %% ops
+    mul/2, mul/1
+]).
+
+-opaque word() :: {w, ordsets:ordset(wfc_wfchar:wfchar())}.
+
+%%----------------------------
+%% constructor
+%%----------------------------
+
+-spec one() -> word().
+
+one() ->
+    {w, []}.
+
+
+validate(_)   -> error(nyi).
+to_wfchars(_) -> error(nyi).
+
+-spec from_wfchars(WfChars) -> Result
+    when WfChars :: list(wfc_char:wfchar()),
+         Result  :: {ok, word()}
+                  | {error, Reason :: string()}.
+
+from_wfchars(Chars) ->
+    from_wfchars(ordsets:from_list(Chars), []).
+
+
+%% validate each char
+from_wfchars([WfChar | Rest], Acc) ->
+    case wfc_wfchar:validate(WfChar) of
+        ok    -> from_wfchars(Rest, [WfChar | Acc]);
+        Error -> Error
+    end;
+% done, all good
+from_wfchars([], Acc) ->
+    {ok, {w, lists:reverse(Acc)}}.
+
+
+%%----------------------------
+%% ops
+%%----------------------------
+
+-spec mul(word(), word()) -> Result
+    when Result :: {ok, word()}
+                 | {error, Reason :: string()}.
+% @doc product of two words
+
+mul({w, X}, {w, Y}) ->
+    case from_wfchars(ordsets:union(X, Y)) of
+        Result = {ok, _} -> Result;
+        Error            -> Error
+    end.
+
+
+-spec mul(Words) -> Result
+    when Words  :: [word()],
+         Result :: {ok, word()}
+                 | {error, Reason},
+         Reason :: string().
+% @doc multiply a list of words together
+
+mul([Word | Rest]) -> mul(Word, mul(Rest));
+mul([])            -> {ok, one()}.