Handle negative values in lexer differently

lamagraph-compiler/src/Lamagraph/Compiler/Parser.y

@@ -239,11 +239,16 @@ atomic_type :: { LLmlType LmlcPs }
 ---------------
 -- Constants --
 ---------------
+
 constant :: { XLocated LmlcPs (LmlLit LmlcPs) }
   : integer_literal { sL1 $1 $ LmlInt noExtField (getInt $1) }
   | char_literal { sL1 $1 $ LmlChar noExtField (getChar $1) }
   | string_literal { sL1 $1 $ LmlString noExtField (getString $1)}
 
+signed_constant :: { XLocated LmlcPs (LmlLit LmlcPs) }
+  : constant { $1 }
+  | '-' integer_literal { sLL $1 $2 $ LmlInt noExtField (getMinusInt $2) }
+
 --------------
 -- Patterns --
 --------------
@@ -266,7 +271,7 @@ simple_pattern :: { LLmlPat LmlcPs }
   : delimited_pattern { $1 }
   | value_name { sL1 $1 $ LmlPatVar noExtField $1 }
   | '_' { sL1 $1 $ LmlPatAny noExtField }
-  | constant { sL1 $1 $ LmlPatConstant noExtField (unLoc $1) }
+  | signed_constant { sL1 $1 $ LmlPatConstant noExtField (unLoc $1) }
   | constr %prec below_DOT { sL1 $1 $ LmlPatConstruct noExtField $1 Nothing }
   | tuple_pattern '|' tuple_pattern { sLL $1 $3 $ LmlPatOr noExtField $1 $3 }
   | constr tuple_pattern %prec constr_appl { sLL $1 $2 $ LmlPatConstruct noExtField $1 (Just $2)}
@@ -315,8 +320,12 @@ compound_expr :: { LLmlExpr LmlcPs }
       sLL $1 $2 $ LmlExprApply noExtField prefixIdent (pure $2)
     }
   | '-' compound_expr %prec prec_unary_minus
-    { let prefixIdent = sL1 $1 $ LmlExprIdent noExtField (mkLongident $ pure "~-") in
-      sLL $1 $2 $ LmlExprApply noExtField prefixIdent (pure $2)
+    { case unLoc $2 of
+        LmlExprConstant _ (LmlInt _ n) ->
+          sLL $1 $2 $ LmlExprConstant noExtField (LmlInt noExtField (-n))
+        _ ->
+          let prefixIdent = sL1 $1 $ LmlExprIdent noExtField (mkLongident $ pure "~-")
+          in sLL $1 $2 $ LmlExprApply noExtField prefixIdent (pure $2)
     }
   | simple_expr infix_op compound_expr
     { let infixIdent = sL1 $2 $ LmlExprIdent noExtField ((mkLongident . pure . unLoc) $2) in
@@ -530,7 +539,10 @@ getLongident :: NonEmpty LToken -> Longident
 getLongident = mkLongident . fmap getIdent
 
 getInt :: LToken -> Int
-getInt (L _ (TokInt val)) = val
+getInt (L _ (TokInt val)) = Prelude.read $ toString val
+
+getMinusInt :: LToken -> Int
+getMinusInt (L _ (TokInt val)) = Prelude.read $ ("-" ++ (toString val))
 
 getChar :: LToken -> Char
 getChar (L _ (TokChar val)) = val

lamagraph-compiler/src/Lamagraph/Compiler/Parser/Lexer.x

@@ -49,6 +49,7 @@ import Lamagraph.Compiler.Parser.SrcLoc
 
 $digit = [0-9]
 $letter = [a-zA-Z]
+
 $capital_letter = [A-Z]
 $lowercase_letter = [a-z]
 
@@ -71,7 +72,7 @@ $operator_char = [\! \$ \% & \* \+ \. \/ \: \< \= \> \? \@ \^ \| \~]
 )
 
 -- Integer literals
-@integer_literal = \-? $digit ( $digit | \_ )*
+@integer_literal = $digit ( $digit | \_ )*
 
 -- Operators
 @infix_symbol = ( \= | \< | \> | \@ | \^ | \| | & | \+ | \- | \* | \/ | \$ | \% ) ( $operator_char )*
@@ -226,7 +227,7 @@ tokAnyIdent ctor input@(startPosn, _, _, str) len = do
 
 tokInt :: AlexAction LToken
 tokInt input@(startPosn, _, _, str) len = do
-  let num = read $ toString $ Text.take len str
+  let num = Text.take len str
   return $ L
     (alexPosnToSrcSpan startPosn $ getEndPos input len)
     (TokInt num)

lamagraph-compiler/src/Lamagraph/Compiler/Parser/LexerTypes.hs

@@ -41,7 +41,7 @@ data IdentType = Capitalized | Lowercase
 data Token
   = TokIdent IdentType Text
   | {- Integer literals -}
-    TokInt Int
+    TokInt Text
   | {- Character literals -}
     TokChar Char
   | {- String literals -}

lamagraph-compiler/test/Lamagraph/Compiler/Parser/LexerTest.hs

@@ -55,7 +55,7 @@ lowercaseIdent =
 integerLiteral :: TestTree
 integerLiteral =
   testCase "Lex Int literal" $ do
-    getTokenTypesFromText "42" @?= Right [TokInt 42, TokEOF]
+    getTokenTypesFromText "42" @?= Right [TokInt "42", TokEOF]
 
 allLetterKeywords :: TestTree -- Yep, strange name
 allLetterKeywords =

lamagraph-compiler/test/golden/eval/output/Minus64.lml.out

@@ -0,0 +1 @@
+-9223372036854775807-9223372036854775807-9223372036854775806-9223372036854775806-9223372036854775806

lamagraph-compiler/test/golden/eval/source/Minus64.lml

@@ -0,0 +1,18 @@
+module Minus64
+
+let bad_number_spaced = -9223372036854775808
+let bad_number=-9223372036854775808
+let bad_number_plus=9223372036854775807
+
+(* val x: int *)
+let f0 x = x +  -9223372036854775808
+let f1 x = x + - 9223372036854775808
+let f2 x = x- 9223372036854775807
+let f3 x = x-9223372036854775807
+let f4 x = x -9223372036854775807
+
+let s = print_int (f0 1)
+let s = print_int (f1 1)
+let s = print_int (f2 1)
+let s = print_int (f3 1)
+let s = print_int (f4 1)