lex.icl

   1 implementation module lex
   2
   3 import Data.Either
   4 import Data.List
   5 import StdString
   6 import System.CommandLine
   7 import StdFile
   8 import StdMisc
   9 import StdBool
  10 from StdFunc import id, const
  11 import Data.Maybe
  12 import Control.Applicative
  13 import Control.Monad
  14 import Control.Monad.State
  15 import Data.Functor
  16 from Data.Func import $
  17 from Text import class Text, instance Text String
  18 import qualified Text
  19
  20 Start :: *World -> *World
  21 Start w
  22 # (args, w) = getCommandLine w // We lezen nu nog standaard van stdin
  23 # (out, w) = stdio w
  24 # (toparse, out) = readEntireFile out
  25 # out = out <<< print (lexer toparse)
  26 # (b, w) = fclose out w
  27 | not b = setReturnCode 1 w
  28 = w
  29         where
  30                 readEntireFile :: *File -> *([Char], *File)
  31                 readEntireFile f
  32                 # (b, c, f) = freadc f
  33                 | not b = ([], f)
  34                 # (cs, f) = readEntireFile f
  35                 = ([c:cs], f)
  36
  37 lexer :: [Char] -> LexerOutput
  38 lexer [] = Right []
  39 lexer x = case lex x of
  40         (Right t, rest) = lexer rest >>= \ts.Right [t:ts]
  41         (Left e, _) = Left e
  42
  43 print :: LexerOutput -> String
  44 print (Left l) = "Error: " +++ l
  45 print (Right x) = 'Text'.concat (print` 0 x)
  46         where
  47                 print` :: Int [Token] -> [String]
  48                 print` _ [] = []
  49                 print` i [(IdentToken l):rest] = [toString l:print` i rest]
  50                 print` i [(NumberToken j):rest] = [toString j:print` i rest]
  51                 print` i [(CharToken c):rest] = ["'":toString c:"'":print` i rest]
  52                 print` i [VarToken:rest] = ["var ":print` i rest]
  53                 print` i [VoidToken:rest] = ["Void":print` i rest]
  54                 print` i [ReturnToken:rest] = ["return":print` i rest]
  55                 print` i [IfToken:rest] = ["if":print` i rest]
  56                 print` i [ElseToken:rest] = ["else":print` i rest]
  57                 print` i [WhileToken:rest] = ["while":print` i rest]
  58                 print` i [TrueToken:rest] = ["True":print` i rest]
  59                 print` i [FalseToken:rest] = ["False":print` i rest]
  60                 print` i [BraceOpenToken:rest] = ["(":print` i rest]
  61                 print` i [BraceCloseToken:rest] = [")":print` i rest]
  62                 print` i [CBraceOpenToken:rest] = ["{\n":tab (i+1):print` (i+1) rest]
  63                 print` i [CBraceCloseToken:rest] = ["}\n":tab i:print` i rest]
  64                 print` i [SquareOpenToken:rest] = ["[":print` i rest]
  65                 print` i [SquareCloseToken:rest] = ["]":print` i rest]
  66                 print` i [CommaToken:rest] = [", ":print` i rest]
  67                 print` i [ColonToken:rest] = [":":print` i rest]
  68                 print` i [SColonToken:CBraceCloseToken:rest] =
  69                         [";\n":tab (i-1):"}\n":print` (i-1) rest]
  70                 print` i [SColonToken:rest] = [";\n":tab i:print` i rest]
  71                 print` i [DotToken:rest] = [".":print` i rest]
  72                 print` i [PlusToken:rest] = [" + ":print` i rest]
  73                 print` i [DashToken:rest] = [" - ":print` i rest]
  74                 print` i [StarToken:rest] = [" * ":print` i rest]
  75                 print` i [SlashToken:rest] = [" / ":print` i rest]
  76                 print` i [PercentToken:rest] = [" % ":print` i rest]
  77                 print` i [AssignmentToken:rest] = [" = ":print` i rest]
  78                 print` i [LesserToken:rest] = [" < ":print` i rest]
  79                 print` i [BiggerToken:rest] = [" > ":print` i rest]
  80                 print` i [ExclamationToken:rest] = ["!":print` i rest]
  81                 print` i [DoubleColonToken:rest] = [" :: ":print` i rest]
  82                 print` i [LesserEqToken:rest] = [" <= ":print` i rest]
  83                 print` i [GreaterEqToken:rest] = [" >= ":print` i rest]
  84                 print` i [EqualsToken:rest] = [" == ":print` i rest]
  85                 print` i [AmpersandsToken:rest] = [" && ":print` i rest]
  86                 print` i [PipesToken:rest] = [" || ":print` i rest]
  87                 print` i [ArrowToken:rest] = [" -> ":print` i rest]
  88                 print` i [EndOfFileToken:rest] = ["\n":print` i rest]
  89
  90 tab :: Int -> String
  91 tab 0 = ""
  92 tab i = "\t" +++ tab (i-1)
  93
  94
  95 lex :: [Char] -> (Either String Token, [Char])
  96 lex [] = (Right EndOfFileToken, [])
  97 lex ['v':'a':'r':rest] = (Right VarToken, rest)
  98 lex ['V':'o':'i':'d':rest] = (Right VoidToken, rest)
  99 lex ['r':'e':'t':'u':'r':'n':rest] = (Right ReturnToken, rest)
 100 lex ['i':'f':rest] = (Right IfToken, rest)
 101 lex ['e':'l':'s':'e':rest] = (Right ElseToken, rest)
 102 lex ['w':'h':'i':'l':'e':rest] = (Right WhileToken, rest)
 103 lex ['T':'r':'u':'e':rest] = (Right TrueToken, rest)
 104 lex ['F':'a':'l':'s':'e':rest] = (Right FalseToken, rest)
 105 lex ['\'':'\\':x:'\'':rest] = case x of
 106         'a' = (Right (CharToken (toChar 7)), rest) //Alarm
 107         'b' = (Right (CharToken '\b'), rest) //Backspace
 108         'f' = (Right (CharToken '\f'), rest) //Formfeed
 109         'n' = (Right (CharToken '\n'), rest) //Newline
 110         'r' = (Right (CharToken '\r'), rest) //Carriage Return
 111         't' = (Right (CharToken '\t'), rest) //Horizontal tab
 112         'v' = (Right (CharToken '\v'), rest) //Vertical tab
 113         _ = (Left ("Unknown escape: \\" +++ toString x), rest)
 114 lex ['\'':x:'\'':rest] = (Right (CharToken x), rest)
 115 lex [':':':':rest] = (Right DoubleColonToken, rest)
 116 lex ['<':'=':rest] = (Right LesserEqToken, rest)
 117 lex ['>':'=':rest] = (Right GreaterEqToken, rest)
 118 lex ['=':'=':rest] = (Right EqualsToken, rest)
 119 lex ['&':'&':rest] = (Right AmpersandsToken, rest)
 120 lex ['|':'|':rest] = (Right PipesToken, rest)
 121 lex ['-':'>':rest] = (Right ArrowToken, rest)
 122 lex ['(':rest] = (Right BraceOpenToken, rest)
 123 lex [')':rest] = (Right BraceCloseToken, rest)
 124 lex ['{':rest] = (Right CBraceOpenToken, rest)
 125 lex ['}':rest] = (Right CBraceCloseToken, rest)
 126 lex ['[':rest] = (Right SquareOpenToken, rest)
 127 lex [']':rest] = (Right SquareCloseToken, rest)
 128 lex [',':rest] = (Right CommaToken, rest)
 129 lex [':':rest] = (Right ColonToken, rest)
 130 lex [';':rest] = (Right SColonToken, rest)
 131 lex ['.':rest] = (Right DotToken, rest)
 132 lex ['+':rest] = (Right PlusToken, rest)
 133 lex ['-':rest] = (Right DashToken, rest)
 134 lex ['*':rest] = (Right StarToken, rest)
 135 lex ['/':rest] = (Right SlashToken, rest)
 136 lex ['%':rest] = (Right PercentToken, rest)
 137 lex ['=':rest] = (Right AssignmentToken, rest)
 138 lex ['<':rest] = (Right LesserToken, rest)
 139 lex ['>':rest] = (Right BiggerToken, rest)
 140 lex ['!':rest] = (Right ExclamationToken, rest)
 141 lex [x:xs]
 142 | isSpace x = lex xs
 143 | isDigit x
 144         # (is, rest) = span isDigit xs
 145         = (Right (NumberToken [x:is]), rest)
 146 | isAlpha x
 147         # (is, rest) = span (\c.isAlphanum c || c == '_') xs
 148         = (Right (IdentToken [x:is]), rest)
 149 | otherwise = (Left ("Unexpected character: " +++ toString x), [])