cleaned up lex

[cc1516.git] / lex.icl

diff --git a/lex.icl b/lex.icl
index f050062..6966cde 100644 (file)
--- 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 StdString
+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 ['/':'/':x:rest] = if (x == '\n') (lex rest) (lex ['/':'/':rest])
+lex ['/':'*':x1:x2:rest] = if (x1 == '*' && x2 == '/') 
+       (lex rest) (lex ['/':'*':rest])
+lex ['v':'a':'r':rest] = (Right VarToken, rest)
+lex ['V':'o':'i':'d':rest] = (Right VoidToken, rest)
+lex ['r':'e':'t':'u':'r':'n':rest] = (Right ReturnToken, rest)
+lex ['i':'f':rest] = (Right IfToken, rest)
+lex ['e':'l':'s':'e':rest] = (Right ElseToken, rest)
+lex ['w':'h':'i':'l':'e':rest] = (Right WhileToken, rest)
+lex ['T':'r':'u':'e':rest] = (Right TrueToken, rest)
+lex ['F':'a':'l':'s':'e':rest] = (Right FalseToken, rest)
+lex ['I':'n':'t':rest] = (Right IntTypeToken, rest)
+lex ['C':'h':'a':'r':rest] = (Right CharTypeToken, rest)
+lex ['B':'o':'o':'l':rest] = (Right BoolTypeToken, rest)
+lex [':':':':rest] = (Right DoubleColonToken, rest)
+lex ['!':'=':rest] = (Right NotEqualToken, rest)
+lex ['<':'=':rest] = (Right LesserEqToken, rest)
+lex ['>':'=':rest] = (Right GreaterEqToken, rest)
+lex ['=':'=':rest] = (Right EqualsToken, rest)
+lex ['&':'&':rest] = (Right AmpersandsToken, rest)
+lex ['|':'|':rest] = (Right PipesToken, rest)
+lex ['-':'>':rest] = (Right ArrowToken, rest)
+lex ['\'':x:'\'':rest] = (Right (CharToken x), rest)
+lex ['\'':'\\':x:'\'':rest] = case get x EscapeMap of
+       Just t = (Right (CharToken t), rest)
+       _ = (Left ("Unknown escape: \\" +++ toString x), [])
+lex [x:rest]
+# t = get x SingleCharTokens
+| isJust t = (Right (fromJust t), rest)
+| isSpace x = lex rest
+| isDigit x = let (is, rest`) = span isDigit rest in
+       (Right (NumberToken [x:is]), rest`)
+| isAlpha x = let (is, rest`) = span (\c.isAlphanum c || c == '_') rest in
+       (Right (IdentToken [x:is]), rest`)
+| otherwise = (Left ("Unexpected character: " +++ toString x), [])