CmmLex.x 10.9 KB
Newer Older
1
-----------------------------------------------------------------------------
Simon Marlow's avatar
Simon Marlow committed
2 3
--
-- (c) The University of Glasgow, 2004-2006
4 5 6 7 8 9 10 11 12 13
--
-- Lexer for concrete Cmm.  We try to stay close to the C-- spec, but there
-- are a few minor differences:
--
--   * extra keywords for our macros, and float32/float64 types
--   * global registers (Sp,Hp, etc.)
--
-----------------------------------------------------------------------------

{
14
{-# LANGUAGE BangPatterns #-}
15
{-# OPTIONS -Wwarn -w #-}
16
-- The above -Wwarn suppression flag is a temporary kludge.
17 18
-- While working on this module you are encouraged to remove it and fix
-- any warnings in the module. See
19
--     http://ghc.haskell.org/trac/ghc/wiki/Commentary/CodingStyle#Warnings
20 21
-- for details

22 23 24 25
module CmmLex (
   CmmToken(..), cmmlex,
  ) where

26
import CmmExpr
27

28
import Lexer
29 30 31 32 33
import SrcLoc
import UniqFM
import StringBuffer
import FastString
import Ctype
Simon Marlow's avatar
Simon Marlow committed
34
import Util
35
--import TRACE
36 37 38

import Data.Word
import Data.Char
39 40
}

41
$whitechar   = [\ \t\n\r\f\v\xa0] -- \xa0 is Unicode no-break space
42 43 44
$white_no_nl = $whitechar # \n

$ascdigit  = 0-9
45
$unidigit  = \x01 -- Trick Alex into handling Unicode. See alexGetChar.
46
$digit     = [$ascdigit $unidigit]
47
$octit     = 0-7
48 49
$hexit     = [$digit A-F a-f]

50
$unilarge  = \x03 -- Trick Alex into handling Unicode. See alexGetChar.
51 52 53
$asclarge  = [A-Z \xc0-\xd6 \xd8-\xde]
$large     = [$asclarge $unilarge]

54
$unismall  = \x04 -- Trick Alex into handling Unicode. See alexGetChar.
55 56 57
$ascsmall  = [a-z \xdf-\xf6 \xf8-\xff]
$small     = [$ascsmall $unismall \_]

58
$namebegin = [$large $small \. \$ \@]
59 60 61 62 63 64 65 66 67
$namechar  = [$namebegin $digit]

@decimal     = $digit+
@octal       = $octit+
@hexadecimal = $hexit+
@exponent    = [eE] [\-\+]? @decimal

@floating_point = @decimal \. @decimal @exponent? | @decimal @exponent

68
@escape      = \\ ([abfnrt\\\'\"\?] | x $hexit{1,2} | $octit{1,3})
69 70 71 72
@strchar     = ($printable # [\"\\]) | @escape

cmm :-

73
$white_no_nl+           ;
74
^\# pragma .* \n        ; -- Apple GCC 3.3 CPP generates pragmas in its output
75

76
^\# (line)?             { begin line_prag }
77 78 79

-- single-line line pragmas, of the form
--    # <line> "<file>" <extra-stuff> \n
80 81 82
<line_prag> $digit+                     { setLine line_prag1 }
<line_prag1> \" [^\"]* \"       { setFile line_prag2 }
<line_prag2> .*                         { pop }
83 84

<0> {
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108
  \n                    ;

  [\:\;\{\}\[\]\(\)\=\`\~\/\*\%\-\+\&\^\|\>\<\,\!]      { special_char }

  ".."                  { kw CmmT_DotDot }
  "::"                  { kw CmmT_DoubleColon }
  ">>"                  { kw CmmT_Shr }
  "<<"                  { kw CmmT_Shl }
  ">="                  { kw CmmT_Ge }
  "<="                  { kw CmmT_Le }
  "=="                  { kw CmmT_Eq }
  "!="                  { kw CmmT_Ne }
  "&&"                  { kw CmmT_BoolAnd }
  "||"                  { kw CmmT_BoolOr }

  P@decimal             { global_regN (\n -> VanillaReg n VGcPtr) }
  R@decimal             { global_regN (\n -> VanillaReg n VNonGcPtr) }
  F@decimal             { global_regN FloatReg }
  D@decimal             { global_regN DoubleReg }
  L@decimal             { global_regN LongReg }
  Sp                    { global_reg Sp }
  SpLim                 { global_reg SpLim }
  Hp                    { global_reg Hp }
  HpLim                 { global_reg HpLim }
109 110 111
  CCCS                  { global_reg CCCS }
  CurrentTSO            { global_reg CurrentTSO }
  CurrentNursery        { global_reg CurrentNursery }
112 113 114 115 116 117 118 119 120 121 122
  HpAlloc               { global_reg HpAlloc }
  BaseReg               { global_reg BaseReg }

  $namebegin $namechar* { name }

  0 @octal              { tok_octal }
  @decimal              { tok_decimal }
  0[xX] @hexadecimal    { tok_hexadecimal }
  @floating_point       { strtoken tok_float }

  \" @strchar* \"       { strtoken tok_string }
123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148
}

{
data CmmToken
  = CmmT_SpecChar  Char
  | CmmT_DotDot
  | CmmT_DoubleColon
  | CmmT_Shr
  | CmmT_Shl
  | CmmT_Ge
  | CmmT_Le
  | CmmT_Eq
  | CmmT_Ne
  | CmmT_BoolAnd
  | CmmT_BoolOr
  | CmmT_CLOSURE
  | CmmT_INFO_TABLE
  | CmmT_INFO_TABLE_RET
  | CmmT_INFO_TABLE_FUN
  | CmmT_INFO_TABLE_CONSTR
  | CmmT_INFO_TABLE_SELECTOR
  | CmmT_else
  | CmmT_export
  | CmmT_section
  | CmmT_goto
  | CmmT_if
149
  | CmmT_call
150 151
  | CmmT_jump
  | CmmT_foreign
152
  | CmmT_never
153
  | CmmT_prim
154
  | CmmT_reserve
155
  | CmmT_return
156
  | CmmT_returns
157 158 159 160
  | CmmT_import
  | CmmT_switch
  | CmmT_case
  | CmmT_default
161
  | CmmT_push
Peter Wortmann's avatar
Peter Wortmann committed
162
  | CmmT_unwind
163 164 165 166
  | CmmT_bits8
  | CmmT_bits16
  | CmmT_bits32
  | CmmT_bits64
167
  | CmmT_bits128
168
  | CmmT_bits256
169
  | CmmT_bits512
170 171
  | CmmT_float32
  | CmmT_float64
172
  | CmmT_gcptr
173
  | CmmT_GlobalReg GlobalReg
174 175 176
  | CmmT_Name      FastString
  | CmmT_String    String
  | CmmT_Int       Integer
177 178 179 180 181 182 183
  | CmmT_Float     Rational
  | CmmT_EOF
  deriving (Show)

-- -----------------------------------------------------------------------------
-- Lexer actions

Ian Lynagh's avatar
Ian Lynagh committed
184
type Action = RealSrcSpan -> StringBuffer -> Int -> P (RealLocated CmmToken)
185 186 187 188 189 190 191 192 193 194 195 196 197 198

begin :: Int -> Action
begin code _span _str _len = do pushLexState code; lexToken

pop :: Action
pop _span _buf _len = do popLexState; lexToken

special_char :: Action
special_char span buf len = return (L span (CmmT_SpecChar (currentChar buf)))

kw :: CmmToken -> Action
kw tok span buf len = return (L span tok)

global_regN :: (Int -> GlobalReg) -> Action
199
global_regN con span buf len
200 201
  = return (L span (CmmT_GlobalReg (con (fromIntegral n))))
  where buf' = stepOn buf
202
        n = parseUnsignedInteger buf' (len-1) 10 octDecDigit
203 204 205 206 207

global_reg :: GlobalReg -> Action
global_reg r span buf len = return (L span (CmmT_GlobalReg r))

strtoken :: (String -> CmmToken) -> Action
208
strtoken f span buf len =
209 210 211
  return (L span $! (f $! lexemeToString buf len))

name :: Action
212
name span buf len =
213
  case lookupUFM reservedWordsFM fs of
214 215
        Just tok -> return (L span tok)
        Nothing  -> return (L span (CmmT_Name fs))
216
  where
217
        fs = lexemeToFastString buf len
218 219

reservedWordsFM = listToUFM $
220 221 222 223 224 225 226 227 228 229 230 231
        map (\(x, y) -> (mkFastString x, y)) [
        ( "CLOSURE",            CmmT_CLOSURE ),
        ( "INFO_TABLE",         CmmT_INFO_TABLE ),
        ( "INFO_TABLE_RET",     CmmT_INFO_TABLE_RET ),
        ( "INFO_TABLE_FUN",     CmmT_INFO_TABLE_FUN ),
        ( "INFO_TABLE_CONSTR",  CmmT_INFO_TABLE_CONSTR ),
        ( "INFO_TABLE_SELECTOR",CmmT_INFO_TABLE_SELECTOR ),
        ( "else",               CmmT_else ),
        ( "export",             CmmT_export ),
        ( "section",            CmmT_section ),
        ( "goto",               CmmT_goto ),
        ( "if",                 CmmT_if ),
232 233 234
        ( "call",               CmmT_call ),
        ( "jump",               CmmT_jump ),
        ( "foreign",            CmmT_foreign ),
235 236
        ( "never",              CmmT_never ),
        ( "prim",               CmmT_prim ),
237 238
        ( "reserve",            CmmT_reserve ),
        ( "return",             CmmT_return ),
239 240 241 242
        ( "returns",            CmmT_returns ),
        ( "import",             CmmT_import ),
        ( "switch",             CmmT_switch ),
        ( "case",               CmmT_case ),
243 244
        ( "default",            CmmT_default ),
        ( "push",               CmmT_push ),
Peter Wortmann's avatar
Peter Wortmann committed
245
        ( "unwind",             CmmT_unwind ),
246
        ( "bits8",              CmmT_bits8 ),
247 248 249 250 251 252 253 254
        ( "bits16",             CmmT_bits16 ),
        ( "bits32",             CmmT_bits32 ),
        ( "bits64",             CmmT_bits64 ),
        ( "bits128",            CmmT_bits128 ),
        ( "bits256",            CmmT_bits256 ),
        ( "bits512",            CmmT_bits512 ),
        ( "float32",            CmmT_float32 ),
        ( "float64",            CmmT_float64 ),
255
-- New forms
256 257 258 259 260 261 262 263 264 265 266 267 268
        ( "b8",                 CmmT_bits8 ),
        ( "b16",                CmmT_bits16 ),
        ( "b32",                CmmT_bits32 ),
        ( "b64",                CmmT_bits64 ),
        ( "b128",               CmmT_bits128 ),
        ( "b256",               CmmT_bits256 ),
        ( "b512",               CmmT_bits512 ),
        ( "f32",                CmmT_float32 ),
        ( "f64",                CmmT_float64 ),
        ( "gcptr",              CmmT_gcptr )
        ]

tok_decimal span buf len
269
  = return (L span (CmmT_Int  $! parseUnsignedInteger buf len 10 octDecDigit))
270

271
tok_octal span buf len
272
  = return (L span (CmmT_Int  $! parseUnsignedInteger (offsetBytes 1 buf) (len-1) 8 octDecDigit))
273

274
tok_hexadecimal span buf len
275
  = return (L span (CmmT_Int  $! parseUnsignedInteger (offsetBytes 2 buf) (len-2) 16 hexDigit))
276 277 278 279

tok_float str = CmmT_Float $! readRational str

tok_string str = CmmT_String (read str)
280
                 -- urk, not quite right, but it'll do for now
281 282 283 284 285 286

-- -----------------------------------------------------------------------------
-- Line pragmas

setLine :: Int -> Action
setLine code span buf len = do
287
  let line = parseUnsignedInteger buf len 10 octDecDigit
Ian Lynagh's avatar
Ian Lynagh committed
288
  setSrcLoc (mkRealSrcLoc (srcSpanFile span) (fromIntegral line - 1) 1)
289
        -- subtract one: the line number refers to the *following* line
290 291 292 293 294 295 296 297
  -- trace ("setLine "  ++ show line) $ do
  popLexState
  pushLexState code
  lexToken

setFile :: Int -> Action
setFile code span buf len = do
  let file = lexemeToFastString (stepOn buf) (len-2)
Ian Lynagh's avatar
Ian Lynagh committed
298
  setSrcLoc (mkRealSrcLoc file (srcSpanEndLine span) (srcSpanEndCol span))
299 300 301 302 303 304 305 306 307 308
  popLexState
  pushLexState code
  lexToken

-- -----------------------------------------------------------------------------
-- This is the top-level function: called from the parser each time a
-- new token is to be read from the input.

cmmlex :: (Located CmmToken -> P a) -> P a
cmmlex cont = do
Ian Lynagh's avatar
Ian Lynagh committed
309 310 311
  (L span tok) <- lexToken
  --trace ("token: " ++ show tok) $ do
  cont (L (RealSrcSpan span) tok)
312

Ian Lynagh's avatar
Ian Lynagh committed
313
lexToken :: P (RealLocated CmmToken)
314 315 316 317
lexToken = do
  inp@(loc1,buf) <- getInput
  sc <- getLexState
  case alexScan inp sc of
Ian Lynagh's avatar
Ian Lynagh committed
318
    AlexEOF -> do let span = mkRealSrcSpan loc1 loc1
319 320
                  setLastToken span 0
                  return (L span CmmT_EOF)
321 322
    AlexError (loc2,_) -> do failLocMsgP loc1 loc2 "lexical error"
    AlexSkip inp2 _ -> do
323 324
        setInput inp2
        lexToken
325
    AlexToken inp2@(end,buf2) len t -> do
326 327 328 329
        setInput inp2
        let span = mkRealSrcSpan loc1 end
        span `seq` setLastToken span len
        t span buf len
330 331 332 333 334

-- -----------------------------------------------------------------------------
-- Monad stuff

-- Stuff that Alex needs to know about our input type:
Ian Lynagh's avatar
Ian Lynagh committed
335
type AlexInput = (RealSrcLoc,StringBuffer)
336 337 338 339

alexInputPrevChar :: AlexInput -> Char
alexInputPrevChar (_,s) = prevChar s '\n'

340
-- backwards compatibility for Alex 2.x
341
alexGetChar :: AlexInput -> Maybe (Char,AlexInput)
342 343 344 345 346 347 348
alexGetChar inp = case alexGetByte inp of
                    Nothing    -> Nothing
                    Just (b,i) -> c `seq` Just (c,i)
                       where c = chr $ fromIntegral b

alexGetByte :: AlexInput -> Maybe (Word8,AlexInput)
alexGetByte (loc,s)
349
  | atEnd s   = Nothing
350 351 352
  | otherwise = b `seq` loc' `seq` s' `seq` Just (b, (loc', s'))
  where c    = currentChar s
        b    = fromIntegral $ ord $ c
353
        loc' = advanceSrcLoc loc c
354
        s'   = stepOn s
355 356 357 358 359 360 361

getInput :: P AlexInput
getInput = P $ \s@PState{ loc=l, buffer=b } -> POk s (l,b)

setInput :: AlexInput -> P ()
setInput (l,b) = P $ \s -> POk s{ loc=l, buffer=b } ()
}