switch to binaries, parse identifiers

src/akh_lexer.erl

@@ -5,30 +5,8 @@
     next/1
 ]).
 
--define(open_delim(T),
-    T =:= $(;
-    T =:= $[;
-    T =:= ${
-).
-
--define(close_delim(T),
-    T =:= $);
-    T =:= $];
-    T =:= $}
-).
-
--define(op1(T),
-    T =:= $,;
-    T =:= $;;
-    T =:= $.;
-    T =:= $+
-).
-
 -type token_kind() ::
-    comment
-    | atom
-    | identifier
-    | integer
+    binary()
     | '{'
     | '}'
     | '['
@@ -42,31 +20,35 @@
     | '->'
     | '+'.
 
--define(LOC(Line, Column), {{line, Line}, {column, Column}}).
--type location() :: {{line, pos_integer()}, {column, pos_integer()}} | inserted.
+-type token() :: {token_kind(), non_neg_integer() | inserted}.
 
--type token() :: {token_kind(), location()}.
+-define(is_id(C), (C >= $A andalso C =< $Z); (C >= $a andalso C =< $z); C =:= $_).
 
 -record(lexer, {
-    source :: string(),
-    line = 1 :: pos_integer(),
-    column = 1 :: pos_integer(),
-    delim_stack = [] :: [token()]
+    source :: binary(),
+    offset = 0 :: non_neg_integer()
 }).
 
--spec new(string()) -> #lexer{}.
+-spec new(binary()) -> #lexer{}.
 new(Source) -> #lexer{source = Source}.
 
 -spec next(#lexer{}) -> none | {ok, token(), #lexer{}}.
-next(#lexer{source = [], delim_stack = []}) ->
-    none;
-next(#lexer{source = [$\n | Rest], line = Line} = Lexer) ->
-    next(Lexer#lexer{source = Rest, line = Line + 1, column = 1});
-next(#lexer{source = [T | Rest], line = Line, column = Column} = Lexer) when ?op1(T) ->
-    Token = {list_to_atom([T]), ?LOC(Line, Column)},
-    {ok, Token, Lexer#lexer{source = Rest, column = Column + 1}};
-next(#lexer{source = [$-, $> | Rest], line = Line, column = Column} = Lexer) ->
-    Token = {'->', ?LOC(Line, Column)},
-    {ok, Token, Lexer#lexer{source = Rest, column = Column + 2}};
-next(_) ->
+next(#lexer{source = <<$\n, Rest/bytes>>, offset = Offset} = Lexer) ->
+    next(Lexer#lexer{source = Rest, offset = Offset + 1});
+next(#lexer{source = <<C, _/bytes>>} = Lexer) when ?is_id(C) ->
+    lex_id(Lexer, 1);
+next(#lexer{source = <<>>}) ->
     none.
+
+-spec lex_id(#lexer{}, pos_integer()) -> {ok, token(), #lexer{}}.
+lex_id(#lexer{source = Source, offset = Offset} = Lexer, Len) when Len < byte_size(Source) ->
+    C = binary:at(Source, Len),
+    if
+        ?is_id(C) ->
+            lex_id(Lexer, Len + 1);
+        true ->
+            {Id, Rest} = split_binary(Source, Len),
+            {ok, {Id, Offset}, Lexer#lexer{source = Rest, offset = Offset + Len}}
+    end;
+lex_id(#lexer{source = Source, offset = Offset} = Lexer, Len) ->
+    {ok, {Source, Offset}, Lexer#lexer{source = <<>>, offset = Offset + Len}}.

src/akhamoth.erl

@@ -1,11 +1,9 @@
 -module(akhamoth).
 
--export([compile/1]).
+-export([compile/1, collect/2]).
 
--spec compile(Path :: file:name_all()) -> any().
 compile(Path) ->
-    {ok, SrcBin} = file:read_file(Path),
-    Src = unicode:characters_to_list(SrcBin),
+    {ok, Src} = file:read_file(Path),
     Lexer = akh_lexer:new(Src),
     collect(Lexer, []).