add operator parsing

maybe in the future but too ambitious for now

pages/syntax_reference.md

@@ -52,7 +52,6 @@ Operator | Associativity
 `+` `-`  | left
 `\|>`    | left
 `==`     | requires parens
-`..`     | requires parens
 `=>`     | right
 `->`     | right
 `=`      | right

src/akh_lex.erl

@@ -15,10 +15,38 @@ This module contains functions for tokenizing Akhamoth source code.
 -define(is_id_start(C), C >= $a, C =< $z; C >= $A, C =< $Z; C =:= $_).
 -define(is_space(C), C =:= $\s; C =:= $\t).
 
+-define(is_op2(T), T =:= <<"|>">>; T =:= <<"=>">>; T =:= <<"->">>; T =:= <<"==">>).
+-define(is_op1(T),
+    T =:= <<"+">>;
+    T =:= <<"-">>;
+    T =:= <<"*">>;
+    T =:= <<"/">>;
+    T =:= <<"=">>;
+    T =:= <<".">>
+).
+
 -doc """
 Tokens for which the category is the same as the content.
 """.
--type token_simple() :: '(' | ')' | '[' | ']' | '{' | '}'.
+-type token_simple() ::
+    '.'
+    | '+'
+    | '-'
+    | '*'
+    | '/'
+    | '='
+    | '|>'
+    | '=>'
+    | '->'
+    | '=='
+    | ','
+    | ':'
+    | '('
+    | ')'
+    | '['
+    | ']'
+    | '{'
+    | '}'.
 
 -doc """
 Tokens for which there is content beyond the category.
@@ -62,16 +90,20 @@ next(Lexer)
 Attempts to get the next token in the input.
 """.
 -spec next(lexer()) -> none | {ok, token(), lexer()}.
-next(#lexer{source = <<C, Rest/bytes>>} = Lx) when ?is_id_start(C) ->
+next(#lexer{source = <<C, Rest/binary>>} = Lx) when ?is_id_start(C) ->
     lex_id(Lx#lexer{source = Rest}, 1);
-next(#lexer{source = <<C, Rest/bytes>>} = Lx) when ?is_digit(C) ->
+next(#lexer{source = <<C, Rest/binary>>} = Lx) when ?is_digit(C) ->
     lex_number(Lx#lexer{source = Rest}, 1);
-next(#lexer{source = <<C, Rest/bytes>>, offset = Offset} = Lx) when ?is_space(C) ->
-    next(Lx#lexer{source = Rest, offset = Offset + 1});
-next(#lexer{source = <<$\n, Rest/bytes>>, offset = Offset} = Lx) ->
-    new_line(Lx#lexer{source = Rest, offset = Offset + 1});
-next(#lexer{source = <<$\r, $\n, Rest/bytes>>, offset = Offset} = Lx) ->
-    new_line(Lx#lexer{source = Rest, offset = Offset + 2});
+next(#lexer{source = <<C, Rest/binary>>} = Lx) when ?is_space(C) ->
+    next(Lx#lexer{source = Rest, offset = Lx#lexer.offset + 1});
+next(#lexer{source = <<T:2/binary, Rest/binary>>, offset = Offset} = Lx) when ?is_op2(T) ->
+    {ok, {binary_to_atom(T), Offset}, Lx#lexer{source = Rest, offset = Offset + 2}};
+next(#lexer{source = <<T:1/binary, Rest/binary>>, offset = Offset} = Lx) when ?is_op1(T) ->
+    {ok, {binary_to_atom(T), Offset}, Lx#lexer{source = Rest, offset = Offset + 1}};
+next(#lexer{source = <<$\n, Rest/binary>>} = Lx) ->
+    new_line(Lx#lexer{source = Rest, offset = Lx#lexer.offset + 1});
+next(#lexer{source = <<$\r, $\n, Rest/binary>>} = Lx) ->
+    new_line(Lx#lexer{source = Rest, offset = Lx#lexer.offset + 2});
 next(#lexer{source = <<>>}) ->
     none.
 
@@ -79,7 +111,7 @@ next(#lexer{source = <<>>}) ->
 
 -spec lex_id(#lexer{}, non_neg_integer()) -> return().
 lex_id(
-    #lexer{source = <<C, Rest/bytes>>} = Lx,
+    #lexer{source = <<C, Rest/binary>>} = Lx,
     Len
 ) when ?is_id_start(C); ?is_digit(C) ->
     lex_id(Lx#lexer{source = Rest}, Len + 1);
@@ -88,7 +120,7 @@ lex_id(#lexer{offset = Offset} = Lx, Len) ->
 
 -spec lex_number(#lexer{}, non_neg_integer()) -> return().
 lex_number(
-    #lexer{source = <<C, Rest/bytes>>} = Lx,
+    #lexer{source = <<C, Rest/binary>>} = Lx,
     Len
 ) when ?is_digit(C); C =:= $_ ->
     lex_number(Lx#lexer{source = Rest}, Len + 1);

src/akhamoth.erl

@@ -1,6 +1,9 @@
 -module(akhamoth).
 
--export([compile_file/1, compile_binary/1, collect/2]).
+-export([
+    compile_file/1,
+    compile_binary/1
+]).
 
 compile_file(Path) ->
     {ok, Source} = file:read_file(Path),
@@ -9,10 +12,15 @@ compile_file(Path) ->
 compile_binary(Source) ->
     Lx = akh_lex:new(Source),
     {Tokens, SourceMap} = collect(Lx, []),
-    lists:map(fun({T, O, _}) -> {T, akh_source_map:location(O, SourceMap)} end, Tokens).
+    lists:map(fun(T) -> location(T, SourceMap) end, Tokens).
 
 collect(Lx, Acc) ->
     case akh_lex:next(Lx) of
         none -> {Acc, akh_lex:source_map(Lx)};
         {ok, T, L} -> collect(L, [T | Acc])
     end.
+
+location({T, O, _}, SourceMap) ->
+    {T, akh_source_map:location(O, SourceMap)};
+location({T, O}, SourceMap) ->
+    {T, akh_source_map:location(O, SourceMap)}.