X-Git-Url: https://git.martlubbers.net/?a=blobdiff_plain;f=lex.icl;h=0adb20c879a4b3492df770199df3ef302989af9b;hb=f4ee031491f6d753513b03f46a602321569e1114;hp=1475a0efacbe9bf2450bc130e76d2de24bfb8a51;hpb=4b8ac77a88528dd1eeac43fd7a6c2e6ac711d5a9;p=cc1516.git

diff --git a/lex.icl b/lex.icl
index 1475a0e..0adb20c 100644
--- a/lex.icl
+++ b/lex.icl
@@ -1,102 +1,72 @@
 implementation module lex
 
+import Control.Monad
 import Data.Either
-import Data.List
-import StdString
-import System.CommandLine
-import StdFile
-import StdMisc
-from StdFunc import id, const
 import Data.Maybe
-import Control.Applicative
-import Control.Monad
-import Control.Monad.State
-from Data.Func import $
-
-// Misschien moeten we hier continuation style van maken
-instance toString lexerOutput where
-	toString l = "dit is een lexer output, danwel error\n"
-
-lexer :: [Char] -> LexerOutput [Token]
-lexer _ = Left "Not Implemented"
-
-Start :: *World -> *World
-Start w
-# (args, w) = getCommandLine w // We lezen nu nog standaard van stdin
-# (out, w) = stdio w
-# (toparse, out) = readEntireFile out
-# out = out <<< toString (lexer toparse)
-# (b, w) = fclose out w
-| not b = setReturnCode 1 w
-= w
-	where
-		readEntireFile :: *File -> *([Char], *File)
-		readEntireFile f
-		# (b, c, f) = freadc f
-		| not b = ([], f)
-		# (cs, f) = readEntireFile f
-		= ([c:cs], f)
-
-
-
-// Clean adaption of Yard, a parsec like parser combinator
-:: Parser a = Parser ([Char] -> (LexerOutput a, [Char]))
-
-runParser :: (Parser a) [Char] -> (LexerOutput a, [Char])
-runParser (Parser p) s = p s
-
-instance Functor Parser where
-    fmap f s    = liftM f s
-
-instance Applicative Parser where
-    pure a      = Parser $ \s -> (Right a, s) 
-    (<*>) sf s  = ap sf s
-
-instance Monad Parser where
-    bind p f    = Parser $ \s -> let (out, rest) = runParser p s in case out of
-                    Left e  = (Left e, rest)
-                    Right t = runParser (f t) rest
-
-//gives us some, many and optional
-instance Alternative Parser where
-    empty       = zero
-    (<|>) p1 p2 = parserAlternative p1 p2
-parserAlternative p1 p2 = Parser $ \s -> let (out, rest) = runParser p1 s in case out of
-                            Left e  = runParser p2 s
-                            Right t = (Right t, rest)
-
-//parser that fails with error
-fail :: String -> Parser a
-fail e = Parser $ \s -> (Left e, s)
-
-//parser that always fails
-zero :: Parser a
-zero = fail "Zero parser"
-
-//matches exactly one Char
-item :: Parser Char
-item = Parser $ \s -> case s of 
-        []      = (Left "Unexpected empty input", s)
-        [x:xs]  = (Right x, xs)
-
-//matches any char which satisfies f
-satisfy :: (Char -> Bool) -> Parser Char
-satisfy f = item >>= (\r -> if (f r) (return r) zero)
-
-//tries a parser, if it fails returns a default value
-optionalDef :: a (Parser a) -> Parser a
-optionalDef def p = parserAlternative p (return def)
-
-//matched given char
-char :: Char -> Parser Char
-char c = satisfy (\i -> c==i) //I hate that we can't do: satisfy (==c)
-
-alpha :: Parser Char
-alpha = satisfy isAlpha
-
-digit :: Parser Char
-digit = satisfy isDigit
-
-//matches a given String
-string :: [Char] -> Parser [Char]
-string s = mapM_ char s >>| return s
\ No newline at end of file
+import Data.Map
+import Data.Func
+import StdString
+import StdTuple
+import StdBool
+import StdList
+import StdChar
+import qualified Text
+
+SingleCharTokens :: Map Char Token
+SingleCharTokens = fromList [
+	('(', BraceOpenToken), (')', BraceCloseToken), ('{', CBraceOpenToken),
+	('}', CBraceCloseToken), ('[', SquareOpenToken), (']', SquareCloseToken),
+	(',', CommaToken), (':', ColonToken), (';', SColonToken),
+	('.', DotToken), ('+', PlusToken), ('*', StarToken), ('/', SlashToken),
+	('%', PercentToken), ('=', AssignmentToken), ('<', LesserToken),
+	('>', BiggerToken), ('!', ExclamationToken)]
+
+EscapeMap :: Map Char Char
+EscapeMap = fromList [('a', toChar 7), ('b', '\b'), ('f', '\f'), ('n', '\n'),
+	('r', '\r'), ('t', '\t'), ('v', '\v')]
+
+lexer :: [Char] -> LexerOutput
+lexer [] = Right []
+lexer x = case lex x of
+	(Right t, rest) = lexer rest >>= \ts.Right [t:ts]
+	(Left e, _) = Left e
+
+lex :: [Char] -> (Either String Token, [Char])
+lex [] = (Right EndOfFileToken, [])
+lex ['/':'/':xs] = lex $ dropWhile ((<>) '\n') xs
+lex ['/':'*':xs] =
+	lex $ tl $ map snd $ dropWhile ((<>) ('*', '/')) $ zip2 [zero:xs] xs
+lex ['v':'a':'r':xs] = (Right VarToken, xs)
+lex ['V':'o':'i':'d':xs] = (Right VoidToken, xs)
+lex ['r':'e':'t':'u':'r':'n':xs] = (Right ReturnToken, xs)
+lex ['i':'f':xs] = (Right IfToken, xs)
+lex ['e':'l':'s':'e':xs] = (Right ElseToken, xs)
+lex ['w':'h':'i':'l':'e':xs] = (Right WhileToken, xs)
+lex ['T':'r':'u':'e':xs] = (Right TrueToken, xs)
+lex ['F':'a':'l':'s':'e':xs] = (Right FalseToken, xs)
+lex ['I':'n':'t':xs] = (Right IntTypeToken, xs)
+lex ['C':'h':'a':'r':xs] = (Right CharTypeToken, xs)
+lex ['B':'o':'o':'l':xs] = (Right BoolTypeToken, xs)
+lex [':':':':xs] = (Right DoubleColonToken, xs)
+lex ['!':'=':xs] = (Right NotEqualToken, xs)
+lex ['<':'=':xs] = (Right LesserEqToken, xs)
+lex ['>':'=':xs] = (Right GreaterEqToken, xs)
+lex ['=':'=':xs] = (Right EqualsToken, xs)
+lex ['&':'&':xs] = (Right AmpersandsToken, xs)
+lex ['|':'|':xs] = (Right PipesToken, xs)
+lex ['-':'>':xs] = (Right ArrowToken, xs)
+lex ['\'':x:'\'':xs] = (Right (CharToken x), xs)
+lex ['\'':'\\':x:'\'':xs] = case get x EscapeMap of
+	Just t = (Right (CharToken t), xs)
+	_ = (Left ("Unknown escape: \\" +++ toString x), [])
+lex ['-':t=:[x:xs]]
+| isDigit x = let (i, r) = span isDigit t in (Right (NumberToken ['-':i]), r)
+| otherwise = (Right DashToken, xs)
+lex t=:[x:xs] = case get x SingleCharTokens of
+	(Just tok) = (Right tok, xs)
+	Nothing
+	| isSpace x = lex xs
+	| isDigit x = let (i, r) = span isDigit t in (Right (NumberToken i), r)
+	| isAlpha x = let (i, r) = span isIdent t in (Right (IdentToken i), r)
+		with isIdent c = isAlphanum c || c == '_'
+	| otherwise = (Left ("Unexpected character: " +++ toString x), [])