X-Git-Url: https://git.martlubbers.net/?a=blobdiff_plain;f=src%2Flex.icl;h=ac5168f4d36ffcebd808061c60e8019996c07bb7;hb=65f6bf827fa2f4d2f79f9e3bda990b7e4c4df5f1;hp=e3c8484e7f888bd57995cd0db92e8426e52a96f8;hpb=2c3c54fbde3f89ed2b87cbf6d4be5d4e7811d694;p=cc1516.git

diff --git a/src/lex.icl b/src/lex.icl
index e3c8484..ac5168f 100644
--- a/src/lex.icl
+++ b/src/lex.icl
@@ -1,7 +1,7 @@
 implementation module lex
 
 import Control.Monad, Control.Applicative
-import Data.Either, Data.Func
+import Data.Either, Data.Func, Data.Maybe, Data.Functor
 from StdFunc import o
 import StdBool
 import StdList
@@ -15,54 +15,61 @@ lexer r = case runParser lexProgram r of
 	(Left e, _) = Left $ toString e
 
 lexProgram :: Parser Char [Token]
-lexProgram = some lexToken <* many (satisfy isSpace) <* eof
-	>>= \ts->pure (map (\t->(0, 0, t)) ts)
+lexProgram = catMaybes <$> some lexToken <* eof
+	>>= \ts->pure $ (map (\t->(0, 0, t)) ts)
 
-lexToken :: Parser Char TokenValue
+lexToken :: Parser Char (Maybe TokenValue)
 lexToken = 
+    //Comments
+    (list (fromString "//") >>| lexUntilNL >>| pure Nothing) <|>
+    (list (fromString "/*") >>| lexUntilCommentClose >>| pure Nothing) <|>
 	//Keyword tokens
-	(lexKw "var" VarToken) <|>
-	(lexKw "Void" VoidToken) <|>
-	(lexKw "return" ReturnToken) <|>
-	(lexKw "if" IfToken) <|>
-	(lexKw "else" ElseToken) <|>
-	(lexKw "while" WhileToken) <|>
-	(lexKw "True" TrueToken) <|>
-	(lexKw "False" FalseToken) <|>
-	(lexKw "Int" IntTypeToken) <|>
-	(lexKw "Bool" BoolTypeToken) <|>
-	(lexKw "Char" CharTypeToken) <|>
+	lexKw "var" VarToken <|>
+	lexKw "Void" VoidToken <|>
+	lexKw "return" ReturnToken <|>
+	lexKw "if" IfToken <|>
+	lexKw "else" ElseToken <|>
+	lexKw "while" WhileToken <|>
+	lexKw "True" TrueToken <|>
+	lexKw "False" FalseToken <|>
+	lexKw "Int" IntTypeToken <|>
+	lexKw "Bool" BoolTypeToken <|>
+	lexKw "Char" CharTypeToken <|>
 	//Escape chars tokens
-	(liftM CharToken $ item '\'' *> item '\\' *> lexEscape <* item '\'') <|>
+	liftM (Just o CharToken)
+		(list (fromString "'\\") *> lexEscape <* item '\'') <|>
 	//Normal chars tokens
-	(liftM CharToken $ item '\'' *> satisfy ((<>) '\'') <* item '\'') <|>
+	liftM (Just o CharToken)
+		(item '\'' *> satisfy ((<>) '\'') <* item '\'') <|>
 	//Two char ops tokens
-	(lexOp "::" DoubleColonToken) <|> (lexOp "!=" NotEqualToken) <|>
-	(lexOp "<=" LesserEqToken) <|> (lexOp ">=" GreaterEqToken) <|>
-	(lexOp "==" EqualsToken) <|> (lexOp "&&" AmpersandsToken) <|>
-	(lexOp "||" PipesToken) <|> (lexOp "[]" EmptyListToken) <|>
-	(lexOp "->" ArrowToken) <|> (lexOp "(" BraceOpenToken) <|>
+	lexOp "::" DoubleColonToken <|> lexOp "!=" NotEqualToken <|>
+	lexOp "<=" LesserEqToken <|> lexOp ">=" GreaterEqToken <|>
+	lexOp "==" EqualsToken <|> lexOp "&&" AmpersandsToken <|>
+	lexOp "||" PipesToken <|> lexOp "[]" EmptyListToken <|>
+	lexOp "->" ArrowToken <|> 
 	//One char ops tokens
-	(lexOp ")" BraceCloseToken) <|> (lexOp "{" CBraceOpenToken) <|>
-	(lexOp "}" CBraceCloseToken) <|> (lexOp "[" SquareOpenToken) <|>
-	(lexOp "]" SquareCloseToken) <|> (lexOp "," CommaToken) <|>
-	(lexOp ":" ColonToken) <|> (lexOp ";" SColonToken) <|>
-	(lexOp "." DotToken) <|> (lexOp "+" PlusToken) <|>
-	(lexOp "*" StarToken) <|> (lexOp "/" SlashToken) <|>
-	(lexOp "%" PercentToken) <|> (lexOp "=" AssignmentToken) <|>
-	(lexOp "<" LesserToken) <|> (lexOp ">" BiggerToken) <|>
-	(lexOp "!" ExclamationToken) <|> (lexOp "-" DashToken) <|>
+    lexOp "(" BraceOpenToken <|>
+	lexOp ")" BraceCloseToken <|> lexOp "{" CBraceOpenToken <|>
+	lexOp "}" CBraceCloseToken <|> lexOp "[" SquareOpenToken <|>
+	lexOp "]" SquareCloseToken <|> lexOp "," CommaToken <|>
+	lexOp ":" ColonToken <|> lexOp ";" SColonToken <|>
+	lexOp "." DotToken <|> lexOp "+" PlusToken <|>
+	lexOp "*" StarToken <|> lexOp "/" SlashToken <|>
+	lexOp "%" PercentToken <|> lexOp "=" AssignmentToken <|>
+	lexOp "<" LesserToken <|> lexOp ">" BiggerToken <|>
+	lexOp "!" ExclamationToken <|> lexOp "-" DashToken <|>
 	//Number tokens
-	(liftM (NumberToken o toInt o toString) $ some $ satisfy isDigit) <|>
+	liftM (Just o NumberToken o toInt o toString) (some $ satisfy isDigit) <|>
 	//Ident tokens
-	(liftM IdentToken $ some $ satisfy isIdentChar) <|>
-	(satisfy isSpace >>| lexToken) //<|>
-//	(eof >>| pure EndOfFileToken) 
+	liftM (Just o IdentToken o toString) (some $ satisfy isIdentChar) <|>
+	(satisfy isSpace >>| pure Nothing)
 	where
+		lexUntilNL = top until (eof <|> (item '\n' >>| pure Void))
+		lexUntilCommentClose = top until (list (fromString "*/"))
 		isIdentChar c = isAlphanum c || c == '_'
-		lexOp s tv = list (fromString s) >>| pure tv
-		lexKw kw tv = lexOp kw tv <* check (not o isIdentChar) >>| pure tv
-		lexEscape = (
+		lexOp s tv = list (fromString s) >>| pure (Just tv)
+		lexKw kw tv = lexOp kw tv <* check (not o isIdentChar)
+		lexEscape = fromJust <$> ((
 			lexOp "a" (toChar 7) <|> lexOp "b" '\b' <|> lexOp "f" '\f' <|>
 			lexOp "n" '\n' <|> lexOp "r" '\t' <|> lexOp "v" '\v' <|>
-			lexOp "'" '\'') <?> ("Unknown escape", 0)
+			lexOp "'" '\'') <?> ("Unknown escape", 0))