X-Git-Url: https://git.martlubbers.net/?a=blobdiff_plain;f=lex.icl;h=876e3020563af235fb2604aac839e35132111f55;hb=1ee1a90bd336cd6c44ab4b51dd7f4946447ce8a8;hp=f050062d7f172760b708a867a469d06cba3dc75c;hpb=19c5f39c61f1692641570dc3cf13360e7dff5cdf;p=cc1516.git

diff --git a/lex.icl b/lex.icl
index f050062..876e302 100644
--- a/lex.icl
+++ b/lex.icl
@@ -1,3 +1,69 @@
-module lex
+implementation module lex
 
-Start = "Hello World!\n"
+import Control.Monad
+import Data.Either
+import Data.Maybe
+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), ('-', DashToken), ('*', 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 [x:xs]
+# t = get x SingleCharTokens
+| isJust t = (Right (fromJust t), xs)
+| isSpace x = lex xs
+| isDigit x = let (is, xs`) = span isDigit xs in
+	(Right (NumberToken [x:is]), xs`)
+| isAlpha x = let (is, xs`) = span (\c.isAlphanum c || c == '_') xs in
+	(Right (IdentToken [x:is]), xs`)
+| otherwise = (Left ("Unexpected character: " +++ toString x), [])