more rewrite of lexer, add simple ints

src/akh_lexer.erl

@@ -1,54 +1,64 @@
 -module(akh_lexer).
--include("akhamoth.hrl").
+-moduledoc """
--export([
+This module contains functions for tokenizing Akhamoth source code.
-    new/1,
+""".
-    next/1
-]).
 
--type token_kind() ::
+-export([new/1, next/1]).
-    binary()
-    | '{'
-    | '}'
-    | '['
-    | ']'
-    | '('
-    | ')'
-    | ','
-    | ';'
-    | ':'
-    | '.'
-    | '->'
-    | '+'.
 
--type token() :: {token_kind(), non_neg_integer() | inserted}.
+-define(is_digit(C), C >= $0, C =< $9).
+-define(is_id_start(C), C >= $a, C =< $z; C >= $A, C =< $Z; C =:= $_).
 
--define(is_id(C), (C >= $A andalso C =< $Z); (C >= $a andalso C =< $z); C =:= $_).
+-doc """
+A token in the input stream
+""".
+-type token() :: any().
 
 -record(lexer, {
     source :: binary(),
     offset = 0 :: non_neg_integer()
 }).
 
+%%% exports
+
+-doc """
+Initializes a lexer to tokenize the given binary.
+""".
 -spec new(binary()) -> #lexer{}.
 new(Source) -> #lexer{source = Source}.
 
+-doc """
+Attempts to get the next token in the input.
+""".
 -spec next(#lexer{}) -> none | {ok, token(), #lexer{}}.
-next(#lexer{source = <<$\n, Rest/bytes>>, offset = Offset} = Lexer) ->
+next(#lexer{source = <<C, _/bytes>>} = Lx) when ?is_id_start(C) ->
-    next(Lexer#lexer{source = Rest, offset = Offset + 1});
+    lex_id(Lx, 1);
-next(#lexer{source = <<C, _/bytes>>} = Lexer) when ?is_id(C) ->
+next(#lexer{source = <<C, _/bytes>>} = Lx) when ?is_digit(C) ->
-    lex_id(Lexer, 1);
+    lex_number(Lx, 1, C - $0);
 next(#lexer{source = <<>>}) ->
     none.
 
--spec lex_id(#lexer{}, pos_integer()) -> {ok, token(), #lexer{}}.
+%%% local functions
-lex_id(#lexer{source = Source, offset = Offset} = Lexer, Len) when Len < byte_size(Source) ->
+
+lex_id(#lexer{source = Source, offset = Offset} = Lx, Len) when Len < byte_size(Source) ->
     C = binary:at(Source, Len),
     if
-        ?is_id(C) ->
+        ?is_id_start(C); ?is_digit(C) ->
-            lex_id(Lexer, Len + 1);
+            lex_id(Lx, Len + 1);
         true ->
             {Id, Rest} = split_binary(Source, Len),
-            {ok, {Id, Offset}, Lexer#lexer{source = Rest, offset = Offset + Len}}
+            {ok, {id, Offset, Id}, Lx#lexer{source = Rest, offset = Offset + Len}}
     end;
-lex_id(#lexer{source = Source, offset = Offset} = Lexer, Len) ->
+lex_id(#lexer{source = Source, offset = Offset} = Lx, Len) ->
-    {ok, {Source, Offset}, Lexer#lexer{source = <<>>, offset = Offset + Len}}.
+    {ok, {id, Offset, Source}, Lx#lexer{source = <<>>, offset = Offset + Len}}.
+
+lex_number(#lexer{source = Source, offset = Offset} = Lx, Len, Acc) when Len < byte_size(Source) ->
+    C = binary:at(Source, Len),
+    if
+        ?is_digit(C) ->
+            lex_number(Lx, Len + 1, Acc * 10 + C - $0);
+        true ->
+            {_, Rest} = split_binary(Source, Len),
+            {ok, {number, Offset, Acc}, Lx#lexer{source = Rest, offset = Offset + Len}}
+    end;
+lex_number(#lexer{offset = Offset} = Lx, Len, Acc) ->
+    {ok, {number, Offset, Acc}, Lx#lexer{source = <<>>, offset = Offset + Len}}.