Lexer.x 36.3 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
-----------------------------------------------------------------------------
-- (c) The University of Glasgow, 2003
--
-- GHC's lexer.
--
-- This is a combination of an Alex-generated lexer from a regex
-- definition, with some hand-coded bits.
--
-- Completely accurate information about token-spans within the source
-- file is maintained.  Every token has a start and end SrcLoc attached to it.
--
-----------------------------------------------------------------------------

--   ToDo / known bugs:
--    - Unicode
--    - parsing integers is a bit slow
--    - readRational is a bit slow
--
--   Known bugs, that were also in the previous version:
--    - M... should be 3 tokens, not 1.
--    - pragma-end should be only valid in a pragma

{
module Lexer (
25
   Token(..), lexer, mkPState, PState(..),
26
   P(..), ParseResult(..), getSrcLoc, 
27 28 29
   failLocMsgP, failSpanMsgP, srcParseFail,
   popContext, pushCurrentContext, setLastToken, setSrcLoc,
   getLexState, popLexState, pushLexState
30 31 32 33 34 35 36 37 38 39 40
  ) where

#include "HsVersions.h"

import ErrUtils		( Message )
import Outputable
import StringBuffer
import FastString
import FastTypes
import SrcLoc
import UniqFM
41
import CmdLineOpts
42
import Ctype
43
import Util		( maybePrefixMatch, readRational )
44 45 46 47

import DATA_BITS
import Char
import Ratio
48
--import TRACE
49 50
}

51
$whitechar   = [\ \t\n\r\f\v\xa0]
52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
$white_no_nl = $whitechar # \n

$ascdigit  = 0-9
$unidigit  = \x01
$digit     = [$ascdigit $unidigit]

$special   = [\(\)\,\;\[\]\`\{\}]
$ascsymbol = [\!\#\$\%\&\*\+\.\/\<\=\>\?\@\\\^\|\-\~]
$unisymbol = \x02
$symbol    = [$ascsymbol $unisymbol] # [$special \_\:\"\']

$unilarge  = \x03
$asclarge  = [A-Z \xc0-\xd6 \xd8-\xde]
$large     = [$asclarge $unilarge]

$unismall  = \x04
$ascsmall  = [a-z \xdf-\xf6 \xf8-\xff]
$small     = [$ascsmall $unismall \_]

$graphic   = [$small $large $symbol $digit $special \:\"\']

$octit	   = 0-7
$hexit     = [$digit A-F a-f]
$symchar   = [$symbol \:]
$nl        = [\n\r]
$idchar    = [$small $large $digit \']

@varid     = $small $idchar*
@conid     = $large $idchar*

@varsym    = $symbol $symchar*
@consym    = \: $symchar*

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

-- we support the hierarchical module name extension:
@qual = (@conid \.)+

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

haskell :-

-- everywhere: skip whitespace and comments
$white_no_nl+ 				;

-- Everywhere: deal with nested comments.  We explicitly rule out
-- pragmas, "{-#", so that we don't accidentally treat them as comments.
-- (this can happen even though pragmas will normally take precedence due to
-- longest-match, because pragmas aren't valid in every state, but comments
-- are).
"{-" / { notFollowedBy '#' }		{ nested_comment }

-- Single-line comments are a bit tricky.  Haskell 98 says that two or
-- more dashes followed by a symbol should be parsed as a varsym, so we
-- have to exclude those.
-- The regex says: "munch all the characters after the dashes, as long as
-- the first one is not a symbol".
112 113
"--"\-* [^$symbol] .*			;
"--"\-* / { atEOL }			;
114 115 116 117 118 119 120 121 122 123 124 125 126

-- 'bol' state: beginning of a line.  Slurp up all the whitespace (including
-- blank lines) until we find a non-whitespace character, then do layout
-- processing.
--
-- One slight wibble here: what if the line begins with {-#? In
-- theory, we have to lex the pragma to see if it's one we recognise,
-- and if it is, then we backtrack and do_bol, otherwise we treat it
-- as a nested comment.  We don't bother with this: if the line begins
-- with {-#, then we'll assume it's a pragma we know about and go for do_bol.
<bol> {
  \n					;
  ^\# (line)?				{ begin line_prag1 }
127
  ^\# pragma .* \n			; -- GCC 3.3 CPP generated, apparently
128
  ^\# \! .* \n				; -- #!, for scripts
129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156
  ()					{ do_bol }
}

-- after a layout keyword (let, where, do, of), we begin a new layout
-- context if the curly brace is missing.
-- Careful! This stuff is quite delicate.
<layout, layout_do> {
  \{ / { notFollowedBy '-' }		{ pop_and open_brace }
	-- we might encounter {-# here, but {- has been handled already
  \n					;
  ^\# (line)?				{ begin line_prag1 }
}

-- do is treated in a subtly different way, see new_layout_context
<layout>    ()				{ new_layout_context True }
<layout_do> ()				{ new_layout_context False }

-- after a new layout context which was found to be to the left of the
-- previous context, we have generated a '{' token, and we now need to
-- generate a matching '}' token.
<layout_left>  ()			{ do_layout_left }

<0,glaexts> \n				{ begin bol }

"{-#" $whitechar* (line|LINE) 		{ begin line_prag2 }

-- single-line line pragmas, of the form
--    # <line> "<file>" <extra-stuff> \n
157 158
<line_prag1> $digit+			{ setLine line_prag1a }
<line_prag1a> \" [$graphic \ ]* \"	{ setFile line_prag1b }
159 160 161 162
<line_prag1b> .*			{ pop }

-- Haskell-style line pragmas, of the form
--    {-# LINE <line> "<file>" #-}
163 164
<line_prag2> $digit+			{ setLine line_prag2a }
<line_prag2a> \" [$graphic \ ]* \"	{ setFile line_prag2b }
165 166 167
<line_prag2b> "#-}"|"-}"		{ pop }
   -- NOTE: accept -} at the end of a LINE pragma, for compatibility
   -- with older versions of GHC which generated these.
168 169 170 171 172 173 174 175 176 177 178 179 180

<0,glaexts> {
  "{-#" $whitechar* (SPECIALI[SZ]E|speciali[sz]e)
  					{ token ITspecialise_prag }
  "{-#" $whitechar* (SOURCE|source)	{ token ITsource_prag }
  "{-#" $whitechar* (INLINE|inline)	{ token ITinline_prag }
  "{-#" $whitechar* (NO(T?)INLINE|no(t?)inline)
  					{ token ITnoinline_prag }
  "{-#" $whitechar* (RULES|rules)	{ token ITrules_prag }
  "{-#" $whitechar* (DEPRECATED|deprecated)
  					{ token ITdeprecated_prag }
  "{-#" $whitechar* (SCC|scc)		{ token ITscc_prag }
  "{-#" $whitechar* (CORE|core)		{ token ITcore_prag }
181
  "{-#" $whitechar* (UNPACK|unpack)	{ token ITunpack_prag }
182 183 184 185 186 187 188 189 190 191 192 193 194
  
  "{-#" 				{ nested_comment }

  -- ToDo: should only be valid inside a pragma:
  "#-}" 				{ token ITclose_prag}
}


-- '0' state: ordinary lexemes
-- 'glaexts' state: glasgow extensions (postfix '#', etc.)

-- "special" symbols

195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211
<0,glaexts> {
  "[:" / { ifExtension parrEnabled }	{ token ITopabrack }
  ":]" / { ifExtension parrEnabled }	{ token ITcpabrack }
}
  
<0,glaexts> {
  "[|"	    / { ifExtension thEnabled }	{ token ITopenExpQuote }
  "[e|"	    / { ifExtension thEnabled }	{ token ITopenExpQuote }
  "[p|"	    / { ifExtension thEnabled }	{ token ITopenPatQuote }
  "[d|"	    / { ifExtension thEnabled }	{ layout_token ITopenDecQuote }
  "[t|"	    / { ifExtension thEnabled }	{ token ITopenTypQuote }
  "|]"	    / { ifExtension thEnabled }	{ token ITcloseQuote }
  \$ @varid / { ifExtension thEnabled }	{ skip_one_varid ITidEscape }
  "$("	    / { ifExtension thEnabled }	{ token ITparenEscape }
}

<0,glaexts> {
ross's avatar
ross committed
212 213
  "(|" / { ifExtension arrowsEnabled `alexAndPred` notFollowedBySymbol }
					{ special IToparenbar }
214 215 216 217 218 219 220 221
  "|)" / { ifExtension arrowsEnabled }  { special ITcparenbar }
}

<0,glaexts> {
  \? @varid / { ifExtension ipEnabled }	{ skip_one_varid ITdupipvarid }
  \% @varid / { ifExtension ipEnabled } { skip_one_varid ITsplitipvarid }
}

222
<glaexts> {
ross's avatar
ross committed
223
  "(#" / { notFollowedBySymbol }	{ token IToubxparen }
224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257
  "#)"					{ token ITcubxparen }
  "{|"					{ token ITocurlybar }
  "|}"					{ token ITccurlybar }
}

<0,glaexts> {
  \(					{ special IToparen }
  \)					{ special ITcparen }
  \[					{ special ITobrack }
  \]					{ special ITcbrack }
  \,					{ special ITcomma }
  \;					{ special ITsemi }
  \`					{ special ITbackquote }
 				
  \{					{ open_brace }
  \}					{ close_brace }
}

<0,glaexts> {
  @qual @varid			{ check_qvarid }
  @qual @conid			{ idtoken qconid }
  @varid			{ varid }
  @conid			{ idtoken conid }
}

-- after an illegal qvarid, such as 'M.let', 
-- we back up and try again in the bad_qvarid state:
<bad_qvarid> {
  @conid			{ pop_and (idtoken conid) }
  @qual @conid			{ pop_and (idtoken qconid) }
}

<glaexts> {
  @qual @varid "#"+		{ idtoken qvarid }
258
  @qual @conid "#"+		{ idtoken qconid }
259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305
  @varid "#"+			{ varid }
  @conid "#"+			{ idtoken conid }
}

-- ToDo: M.(,,,)

<0,glaexts> {
  @qual @varsym			{ idtoken qvarsym }
  @qual @consym			{ idtoken qconsym }
  @varsym			{ varsym }
  @consym			{ consym }
}

<0,glaexts> {
  @decimal			{ tok_decimal }
  0[oO] @octal			{ tok_octal }
  0[xX] @hexadecimal		{ tok_hexadecimal }
}

<glaexts> {
  @decimal \#			{ prim_decimal }
  0[oO] @octal \#		{ prim_octal }
  0[xX] @hexadecimal \#		{ prim_hexadecimal }
}

<0,glaexts> @floating_point		{ strtoken tok_float }
<glaexts>   @floating_point \#		{ init_strtoken 1 prim_float }
<glaexts>   @floating_point \# \#	{ init_strtoken 2 prim_double }

-- Strings and chars are lexed by hand-written code.  The reason is
-- that even if we recognise the string or char here in the regex
-- lexer, we would still have to parse the string afterward in order
-- to convert it to a String.
<0,glaexts> {
  \'				{ lex_char_tok }
  \" 				{ lex_string_tok }
}

{
-- work around bug in Alex 2.0
#if __GLASGOW_HASKELL__ < 503
unsafeAt arr i = arr ! i
#endif

-- -----------------------------------------------------------------------------
-- The token type

306
data Token
307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354
  = ITas  			-- Haskell keywords
  | ITcase
  | ITclass
  | ITdata
  | ITdefault
  | ITderiving
  | ITdo
  | ITelse
  | IThiding
  | ITif
  | ITimport
  | ITin
  | ITinfix
  | ITinfixl
  | ITinfixr
  | ITinstance
  | ITlet
  | ITmodule
  | ITnewtype
  | ITof
  | ITqualified
  | ITthen
  | ITtype
  | ITwhere
  | ITscc			-- ToDo: remove (we use {-# SCC "..." #-} now)

  | ITforall			-- GHC extension keywords
  | ITforeign
  | ITexport
  | ITlabel
  | ITdynamic
  | ITsafe
  | ITthreadsafe
  | ITunsafe
  | ITstdcallconv
  | ITccallconv
  | ITdotnet
  | ITmdo

  | ITspecialise_prag		-- Pragmas
  | ITsource_prag
  | ITinline_prag
  | ITnoinline_prag
  | ITrules_prag
  | ITdeprecated_prag
  | ITline_prag
  | ITscc_prag
  | ITcore_prag                 -- hdaume: core annotations
355
  | ITunpack_prag
356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427
  | ITclose_prag

  | ITdotdot  			-- reserved symbols
  | ITcolon
  | ITdcolon
  | ITequal
  | ITlam
  | ITvbar
  | ITlarrow
  | ITrarrow
  | ITat
  | ITtilde
  | ITdarrow
  | ITminus
  | ITbang
  | ITstar
  | ITdot

  | ITbiglam			-- GHC-extension symbols

  | ITocurly  			-- special symbols
  | ITccurly
  | ITocurlybar                 -- {|, for type applications
  | ITccurlybar                 -- |}, for type applications
  | ITvocurly
  | ITvccurly
  | ITobrack
  | ITopabrack			-- [:, for parallel arrays with -fparr
  | ITcpabrack			-- :], for parallel arrays with -fparr
  | ITcbrack
  | IToparen
  | ITcparen
  | IToubxparen
  | ITcubxparen
  | ITsemi
  | ITcomma
  | ITunderscore
  | ITbackquote

  | ITvarid   FastString	-- identifiers
  | ITconid   FastString
  | ITvarsym  FastString
  | ITconsym  FastString
  | ITqvarid  (FastString,FastString)
  | ITqconid  (FastString,FastString)
  | ITqvarsym (FastString,FastString)
  | ITqconsym (FastString,FastString)

  | ITdupipvarid   FastString	-- GHC extension: implicit param: ?x
  | ITsplitipvarid FastString	-- GHC extension: implicit param: %x

  | ITpragma StringBuffer

  | ITchar       Char
  | ITstring     FastString
  | ITinteger    Integer
  | ITrational   Rational

  | ITprimchar   Char
  | ITprimstring FastString
  | ITprimint    Integer
  | ITprimfloat  Rational
  | ITprimdouble Rational

  -- MetaHaskell extension tokens
  | ITopenExpQuote  		-- [| or [e|
  | ITopenPatQuote		-- [p|
  | ITopenDecQuote		-- [d|
  | ITopenTypQuote		-- [t|         
  | ITcloseQuote		-- |]
  | ITidEscape   FastString	-- $x
  | ITparenEscape		-- $( 
428 429
  | ITvarQuote			-- '
  | ITtyQuote			-- ''
430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446

  -- Arrow notation extension
  | ITproc
  | ITrec
  | IToparenbar			-- (|
  | ITcparenbar			-- |)
  | ITlarrowtail		-- -<
  | ITrarrowtail		-- >-
  | ITLarrowtail		-- -<<
  | ITRarrowtail		-- >>-

  | ITunknown String		-- Used when the lexer can't make sense of it
  | ITeof			-- end of file token
#ifdef DEBUG
  deriving Show -- debugging
#endif

447
isSpecial :: Token -> Bool
448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516
-- If we see M.x, where x is a keyword, but
-- is special, we treat is as just plain M.x, 
-- not as a keyword.
isSpecial ITas        	= True
isSpecial IThiding    	= True
isSpecial ITqualified 	= True
isSpecial ITforall    	= True
isSpecial ITexport    	= True
isSpecial ITlabel     	= True
isSpecial ITdynamic   	= True
isSpecial ITsafe    	= True
isSpecial ITthreadsafe 	= True
isSpecial ITunsafe    	= True
isSpecial ITccallconv   = True
isSpecial ITstdcallconv = True
isSpecial ITmdo		= True
isSpecial _             = False

-- the bitmap provided as the third component indicates whether the
-- corresponding extension keyword is valid under the extension options
-- provided to the compiler; if the extension corresponding to *any* of the
-- bits set in the bitmap is enabled, the keyword is valid (this setup
-- facilitates using a keyword in two different extensions that can be
-- activated independently)
--
reservedWordsFM = listToUFM $
	map (\(x, y, z) -> (mkFastString x, (y, z)))
       [( "_",		ITunderscore, 	0 ),
	( "as",		ITas, 		0 ),
	( "case",	ITcase, 	0 ),     
	( "class",	ITclass, 	0 ),    
	( "data",	ITdata, 	0 ),     
	( "default",	ITdefault, 	0 ),  
	( "deriving",	ITderiving, 	0 ), 
	( "do",		ITdo, 		0 ),       
	( "else",	ITelse, 	0 ),     
	( "hiding",	IThiding, 	0 ),
	( "if",		ITif, 		0 ),       
	( "import",	ITimport, 	0 ),   
	( "in",		ITin, 		0 ),       
	( "infix",	ITinfix, 	0 ),    
	( "infixl",	ITinfixl, 	0 ),   
	( "infixr",	ITinfixr, 	0 ),   
	( "instance",	ITinstance, 	0 ), 
	( "let",	ITlet, 		0 ),      
	( "module",	ITmodule, 	0 ),   
	( "newtype",	ITnewtype, 	0 ),  
	( "of",		ITof, 		0 ),       
	( "qualified",	ITqualified, 	0 ),
	( "then",	ITthen, 	0 ),     
	( "type",	ITtype, 	0 ),     
	( "where",	ITwhere, 	0 ),
	( "_scc_",	ITscc, 		0 ),		-- ToDo: remove

      	( "forall",	ITforall,	 bit glaExtsBit),
	( "mdo",	ITmdo,		 bit glaExtsBit),

	( "foreign",	ITforeign,	 bit ffiBit),
	( "export",	ITexport,	 bit ffiBit),
	( "label",	ITlabel,	 bit ffiBit),
	( "dynamic",	ITdynamic,	 bit ffiBit),
	( "safe",	ITsafe,		 bit ffiBit),
	( "threadsafe",	ITthreadsafe,	 bit ffiBit),
	( "unsafe",	ITunsafe,	 bit ffiBit),
	( "stdcall",    ITstdcallconv,	 bit ffiBit),
	( "ccall",      ITccallconv,	 bit ffiBit),
	( "dotnet",     ITdotnet,	 bit ffiBit),

	( "rec",	ITrec,		 bit arrowsBit),
517
	( "proc",	ITproc,		 bit arrowsBit)
518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548
     ]

reservedSymsFM = listToUFM $
	map (\ (x,y,z) -> (mkFastString x,(y,z)))
      [ ("..",	ITdotdot,	0)
       ,(":",	ITcolon,	0)	-- (:) is a reserved op, 
						-- meaning only list cons
       ,("::",	ITdcolon, 	0)
       ,("=",	ITequal, 	0)
       ,("\\",	ITlam, 		0)
       ,("|",	ITvbar, 	0)
       ,("<-",	ITlarrow, 	0)
       ,("->",	ITrarrow, 	0)
       ,("@",	ITat, 		0)
       ,("~",	ITtilde, 	0)
       ,("=>",	ITdarrow, 	0)
       ,("-",	ITminus, 	0)
       ,("!",	ITbang, 	0)

       ,("*",	ITstar,		bit glaExtsBit)	-- For data T (a::*) = MkT
       ,(".",	ITdot,		bit glaExtsBit)	-- For 'forall a . t'

       ,("-<",	ITlarrowtail,	bit arrowsBit)
       ,(">-",	ITrarrowtail,	bit arrowsBit)
       ,("-<<",	ITLarrowtail,	bit arrowsBit)
       ,(">>-",	ITRarrowtail,	bit arrowsBit)
       ]

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

549
type Action = SrcSpan -> StringBuffer -> Int -> P (Located Token)
550

551 552
special :: Token -> Action
special tok span _buf len = return (L span tok)
553

554 555 556
token, layout_token :: Token -> Action
token t span buf len = return (L span t)
layout_token t span buf len = pushLexState layout >> return (L span t)
557

558 559
idtoken :: (StringBuffer -> Int -> Token) -> Action
idtoken f span buf len = return (L span $! (f buf len))
560

561 562 563
skip_one_varid :: (FastString -> Token) -> Action
skip_one_varid f span buf len 
  = return (L span $! f (lexemeToFastString (stepOn buf) (len-1)))
564

565 566 567
strtoken :: (String -> Token) -> Action
strtoken f span buf len = 
  return (L span $! (f $! lexemeToString buf len))
568

569
init_strtoken :: Int -> (String -> Token) -> Action
570
-- like strtoken, but drops the last N character(s)
571 572
init_strtoken drop f span buf len = 
  return (L span $! (f $! lexemeToString buf (len-drop)))
573 574

begin :: Int -> Action
575
begin code _span _str _len = do pushLexState code; lexToken
576 577

pop :: Action
578
pop _span _buf _len = do popLexState; lexToken
579 580

pop_and :: Action -> Action
581
pop_and act span buf len = do popLexState; act span buf len
582 583 584

notFollowedBy char _ _ _ (_,buf) = atEnd buf || currentChar buf /= char

ross's avatar
ross committed
585 586 587
notFollowedBySymbol _ _ _ (_,buf)
  = atEnd buf || currentChar buf `notElem` "!#$%&*+./<=>?@\\^|-~"

588 589
atEOL _ _ _ (_,buf) = atEnd buf || currentChar buf == '\n'

590 591
ifExtension pred bits _ _ _ = pred bits

592 593 594 595 596
{-
  nested comments require traversing by hand, they can't be parsed
  using regular expressions.
-}
nested_comment :: Action
597
nested_comment span _str _len = do
598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617
  input <- getInput
  go 1 input
  where go 0 input = do setInput input; lexToken
	go n input = do
	  case alexGetChar input of
	    Nothing  -> err input
	    Just (c,input) -> do
	      case c of
	    	'-' -> do
		  case alexGetChar input of
		    Nothing  -> err input
		    Just ('\125',input) -> go (n-1) input
		    Just (c,_)          -> go n input
	     	'\123' -> do
		  case alexGetChar input of
		    Nothing  -> err input
		    Just ('-',input') -> go (n+1) input'
		    Just (c,input)    -> go n input
	    	c -> go n input

618 619
        err input = do failLocMsgP (srcSpanStart span) (fst input) 
			"unterminated `{-'"
620 621

open_brace, close_brace :: Action
622
open_brace span _str _len = do 
623 624
  ctx <- getContext
  setContext (NoLayout:ctx)
625 626
  return (L span ITocurly)
close_brace span _str _len = do 
627
  popContext
628
  return (L span ITccurly)
629 630 631 632

-- We have to be careful not to count M.<varid> as a qualified name
-- when <varid> is a keyword.  We hack around this by catching 
-- the offending tokens afterward, and re-lexing in a different state.
633
check_qvarid span buf len = do
634 635 636 637 638 639 640 641 642 643 644 645
  case lookupUFM reservedWordsFM var of
	Just (keyword,exts)
	  | not (isSpecial keyword) ->
	  if exts == 0 
	     then try_again
	     else do
		b <- extension (\i -> exts .&. i /= 0)
		if b then try_again
		     else return token
	_other -> return token
  where
	(mod,var) = splitQualName buf len
646
	token     = L span (ITqvarid (mod,var))
647 648

	try_again = do
649
		setInput (srcSpanStart span,buf)
650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677
		pushLexState bad_qvarid
		lexToken

qvarid buf len = ITqvarid $! splitQualName buf len
qconid buf len = ITqconid $! splitQualName buf len

splitQualName :: StringBuffer -> Int -> (FastString,FastString)
-- takes a StringBuffer and a length, and returns the module name
-- and identifier parts of a qualified name.  Splits at the *last* dot,
-- because of hierarchical module names.
splitQualName orig_buf len = split orig_buf 0 0
  where
    split buf dot_off n
	| n == len		  = done dot_off
	| lookAhead buf n == '.'  = split2 buf n (n+1)
	| otherwise 		  = split buf dot_off (n+1)	
  
    -- careful, we might get names like M....
    -- so, if the character after the dot is not upper-case, this is
    -- the end of the qualifier part.
    split2 buf dot_off n
	| isUpper (lookAhead buf n) = split buf dot_off (n+1)
	| otherwise 		    = done dot_off

    done dot_off =
	(lexemeToFastString orig_buf dot_off, 
	 lexemeToFastString (stepOnBy (dot_off+1) orig_buf) (len - dot_off -1))

678
varid span buf len = 
679 680 681
  case lookupUFM reservedWordsFM fs of
	Just (keyword,0)    -> do
		maybe_layout keyword
682
		return (L span keyword)
683 684 685
	Just (keyword,exts) -> do
		b <- extension (\i -> exts .&. i /= 0)
		if b then do maybe_layout keyword
686 687 688
			     return (L span keyword)
		     else return (L span (ITvarid fs))
	_other -> return (L span (ITvarid fs))
689 690 691 692 693 694 695 696 697 698 699 700
  where
	fs = lexemeToFastString buf len

conid buf len = ITconid fs
  where fs = lexemeToFastString buf len

qvarsym buf len = ITqvarsym $! splitQualName buf len
qconsym buf len = ITqconsym $! splitQualName buf len

varsym = sym ITvarsym
consym = sym ITconsym

701
sym con span buf len = 
702
  case lookupUFM reservedSymsFM fs of
703
	Just (keyword,0)    -> return (L span keyword)
704 705
	Just (keyword,exts) -> do
		b <- extension (\i -> exts .&. i /= 0)
706 707 708
		if b then return (L span keyword)
		     else return (L span $! con fs)
	_other -> return (L span $! con fs)
709 710 711
  where
	fs = lexemeToFastString buf len

712
tok_decimal span buf len 
713
  = return (L span (ITinteger  $! parseInteger buf len 10 octDecDigit))
714

715
tok_octal span buf len 
716
  = return (L span (ITinteger  $! parseInteger (stepOnBy 2 buf) (len-2) 8 octDecDigit))
717

718
tok_hexadecimal span buf len 
719
  = return (L span (ITinteger  $! parseInteger (stepOnBy 2 buf) (len-2) 16 hexDigit))
720

721
prim_decimal span buf len 
722
  = return (L span (ITprimint  $! parseInteger buf (len-1) 10 octDecDigit))
723

724
prim_octal span buf len 
725
  = return (L span (ITprimint  $! parseInteger (stepOnBy 2 buf) (len-3) 8 octDecDigit))
726

727
prim_hexadecimal span buf len 
728
  = return (L span (ITprimint  $! parseInteger (stepOnBy 2 buf) (len-3) 16 hexDigit))
729

730 731 732
tok_float        str = ITrational   $! readRational str
prim_float       str = ITprimfloat  $! readRational str
prim_double      str = ITprimdouble $! readRational str
733 734 735 736 737 738

-- -----------------------------------------------------------------------------
-- Layout processing

-- we're at the first token on a line, insert layout tokens if necessary
do_bol :: Action
739 740
do_bol span _str _len = do
	pos <- getOffside (srcSpanEnd span)
741 742 743 744 745
	case pos of
	    LT -> do
                --trace "layout: inserting '}'" $ do
		popContext
		-- do NOT pop the lex state, we might have a ';' to insert
746
		return (L span ITvccurly)
747 748 749
	    EQ -> do
                --trace "layout: inserting ';'" $ do
		popLexState
750
		return (L span ITsemi)
751 752 753 754 755 756 757
	    GT -> do
		popLexState
		lexToken

-- certain keywords put us in the "layout" state, where we might
-- add an opening curly brace.
maybe_layout ITdo	= pushLexState layout_do
758
maybe_layout ITmdo	= pushLexState layout_do
759 760 761
maybe_layout ITof	= pushLexState layout
maybe_layout ITlet	= pushLexState layout
maybe_layout ITwhere	= pushLexState layout
762
maybe_layout ITrec	= pushLexState layout
763 764 765 766 767 768 769 770 771 772 773
maybe_layout _	        = return ()

-- Pushing a new implicit layout context.  If the indentation of the
-- next token is not greater than the previous layout context, then
-- Haskell 98 says that the new layout context should be empty; that is
-- the lexer must generate {}.
--
-- We are slightly more lenient than this: when the new context is started
-- by a 'do', then we allow the new context to be at the same indentation as
-- the previous context.  This is what the 'strict' argument is for.
--
774
new_layout_context strict span _buf _len = do
775
    popLexState
776
    let offset = srcSpanStartCol span
777 778 779 780 781 782 783 784
    ctx <- getContext
    case ctx of
	Layout prev_off : _  | 
	   (strict     && prev_off >= offset  ||
	    not strict && prev_off > offset) -> do
		-- token is indented to the left of the previous context.
		-- we must generate a {} sequence now.
		pushLexState layout_left
785
		return (L span ITvocurly)
786 787
	other -> do
		setContext (Layout offset : ctx)
788
		return (L span ITvocurly)
789

790
do_layout_left span _buf _len = do
791 792
    popLexState
    pushLexState bol  -- we must be at the start of a line
793
    return (L span ITvccurly)
794 795 796 797

-- -----------------------------------------------------------------------------
-- LINE pragmas

798 799 800
setLine :: Int -> Action
setLine code span buf len = do
  let line = parseInteger buf len 10 octDecDigit
801
  setSrcLoc (mkSrcLoc (srcSpanFile span) (fromIntegral line - 1) 0)
802 803 804 805 806
	-- subtract one: the line number refers to the *following* line
  popLexState
  pushLexState code
  lexToken

807 808
setFile :: Int -> Action
setFile code span buf len = do
809
  let file = lexemeToFastString (stepOn buf) (len-2)
810
  setSrcLoc (mkSrcLoc file (srcSpanEndLine span) (srcSpanEndCol span))
811 812 813 814 815 816 817 818 819 820
  popLexState
  pushLexState code
  lexToken

-- -----------------------------------------------------------------------------
-- Strings & Chars

-- This stuff is horrible.  I hates it.

lex_string_tok :: Action
821
lex_string_tok span buf len = do
822 823
  tok <- lex_string ""
  end <- getSrcLoc 
824
  return (L (mkSrcSpan (srcSpanStart span) end) tok)
825

826
lex_string :: String -> P Token
827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870
lex_string s = do
  i <- getInput
  case alexGetChar i of
    Nothing -> lit_error

    Just ('"',i)  -> do
	setInput i
	glaexts <- extension glaExtsEnabled
	if glaexts
	  then do
	    i <- getInput
	    case alexGetChar i of
	      Just ('#',i) -> do
		   setInput i
		   if any (> '\xFF') s
                    then failMsgP "primitive string literal must contain only characters <= \'\\xFF\'"
                    else let s' = mkFastStringNarrow (reverse s) in
			 -- always a narrow string/byte array
			 return (ITprimstring s')
	      _other ->
		return (ITstring (mkFastString (reverse s)))
	  else
		return (ITstring (mkFastString (reverse s)))

    Just ('\\',i)
	| Just ('&',i) <- next -> do 
		setInput i; lex_string s
	| Just (c,i) <- next, is_space c -> do 
		setInput i; lex_stringgap s
	where next = alexGetChar i

    Just _ -> do
	c <- lex_char
	lex_string (c:s)

lex_stringgap s = do
  c <- getCharOrFail
  case c of
    '\\' -> lex_string s
    c | is_space c -> lex_stringgap s
    _other -> lit_error


lex_char_tok :: Action
871 872 873 874 875 876
-- Here we are basically parsing character literals, such as 'x' or '\n'
-- but, when Template Haskell is on, we additionally spot
-- 'x and ''T, returning ITvarQuote and ITtyQuote respectively, 
-- but WIHTOUT CONSUMING the x or T part  (the parser does that).
-- So we have to do two characters of lookahead: when we see 'x we need to
-- see if there's a trailing quote
877
lex_char_tok span buf len = do	-- We've seen '
878
   i1 <- getInput	-- Look ahead to first character
879
   let loc = srcSpanStart span
880 881 882 883 884 885 886
   case alexGetChar i1 of
	Nothing -> lit_error 

	Just ('\'', i2@(end2,_)) -> do 	-- We've seen ''
		  th_exts <- extension thEnabled
		  if th_exts then do
			setInput i2
887
			return (L (mkSrcSpan loc end2)  ITtyQuote)
888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909
		   else lit_error

	Just ('\\', i2@(end2,_)) -> do 	-- We've seen 'backslash 
		  setInput i2
		  lit_ch <- lex_escape
		  mc <- getCharOrFail	-- Trailing quote
		  if mc == '\'' then finish_char_tok loc lit_ch
			        else lit_error 

        Just (c, i2@(end2,_)) | not (is_any c) -> lit_error
			      | otherwise      ->

		-- We've seen 'x, where x is a valid character
		--  (i.e. not newline etc) but not a quote or backslash
	   case alexGetChar i2 of	-- Look ahead one more character
		Nothing -> lit_error
		Just ('\'', i3) -> do 	-- We've seen 'x'
			setInput i3 
			finish_char_tok loc c
		_other -> do 		-- We've seen 'x not followed by quote
					-- If TH is on, just parse the quote only
			th_exts <- extension thEnabled	
910
			if th_exts then return (L (mkSrcSpan loc (fst i1)) ITvarQuote)
911 912
				   else lit_error

913
finish_char_tok :: SrcLoc -> Char -> P (Located Token)
914 915 916
finish_char_tok loc ch	-- We've already seen the closing quote
			-- Just need to check for trailing #
  = do	glaexts <- extension glaExtsEnabled
917
	i@(end,_) <- getInput
918 919
	if glaexts then do
		case alexGetChar i of
920 921
			Just ('#',i@(end,_)) -> do
				setInput i
922
				return (L (mkSrcSpan loc end) (ITprimchar ch))
923
			_other ->
924 925 926 927 928 929 930 931 932 933 934
				return (L (mkSrcSpan loc end) (ITchar ch))
	        else do
		   return (L (mkSrcSpan loc end) (ITchar ch))

lex_char :: P Char
lex_char = do
  mc <- getCharOrFail
  case mc of
      '\\' -> lex_escape
      c | is_any c -> return c
      _other -> lit_error
935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954

lex_escape :: P Char
lex_escape = do
  c <- getCharOrFail
  case c of
	'a'   -> return '\a'
	'b'   -> return '\b'
	'f'   -> return '\f'
	'n'   -> return '\n'
	'r'   -> return '\r'
	't'   -> return '\t'
	'v'   -> return '\v'
	'\\'  -> return '\\'
	'"'   -> return '\"'
	'\''  -> return '\''
	'^'   -> do c <- getCharOrFail
		    if c >= '@' && c <= '_'
			then return (chr (ord c - ord '@'))
			else lit_error

955 956 957
	'x'   -> readNum is_hexdigit 16 hexDigit
	'o'   -> readNum is_octdigit  8 octDecDigit
	x | is_digit x -> readNum2 is_digit 10 octDecDigit (octDecDigit x)
958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053

	c1 ->  do
	   i <- getInput
	   case alexGetChar i of
	    Nothing -> lit_error
	    Just (c2,i2) -> 
              case alexGetChar i2 of
		Nothing	-> lit_error
		Just (c3,i3) -> 
		   let str = [c1,c2,c3] in
		   case [ (c,rest) | (p,c) <- silly_escape_chars,
			      	     Just rest <- [maybePrefixMatch p str] ] of
			  (escape_char,[]):_ -> do
				setInput i3
				return escape_char
			  (escape_char,_:_):_ -> do
				setInput i2
				return escape_char
			  [] -> lit_error

readNum :: (Char -> Bool) -> Int -> (Char -> Int) -> P Char
readNum is_digit base conv = do
  c <- getCharOrFail
  if is_digit c 
	then readNum2 is_digit base conv (conv c)
	else lit_error

readNum2 is_digit base conv i = do
  input <- getInput
  read i input
  where read i input = do
	  case alexGetChar input of
	    Just (c,input') | is_digit c -> do
		read (i*base + conv c) input'
	    _other -> do
		setInput input
		if i >= 0 && i <= 0x10FFFF
		   then return (chr i)
		   else lit_error

silly_escape_chars = [
	("NUL", '\NUL'),
	("SOH", '\SOH'),
	("STX", '\STX'),
	("ETX", '\ETX'),
	("EOT", '\EOT'),
	("ENQ", '\ENQ'),
	("ACK", '\ACK'),
	("BEL", '\BEL'),
	("BS", '\BS'),
	("HT", '\HT'),
	("LF", '\LF'),
	("VT", '\VT'),
	("FF", '\FF'),
	("CR", '\CR'),
	("SO", '\SO'),
	("SI", '\SI'),
	("DLE", '\DLE'),
	("DC1", '\DC1'),
	("DC2", '\DC2'),
	("DC3", '\DC3'),
	("DC4", '\DC4'),
	("NAK", '\NAK'),
	("SYN", '\SYN'),
	("ETB", '\ETB'),
	("CAN", '\CAN'),
	("EM", '\EM'),
	("SUB", '\SUB'),
	("ESC", '\ESC'),
	("FS", '\FS'),
	("GS", '\GS'),
	("RS", '\RS'),
	("US", '\US'),
	("SP", '\SP'),
	("DEL", '\DEL')
	]

lit_error = lexError "lexical error in string/character literal"

getCharOrFail :: P Char
getCharOrFail =  do
  i <- getInput
  case alexGetChar i of
	Nothing -> lexError "unexpected end-of-file in string/character literal"
	Just (c,i)  -> do setInput i; return c

-- -----------------------------------------------------------------------------
-- The Parse Monad

data LayoutContext
  = NoLayout
  | Layout !Int

data ParseResult a
  = POk PState a
  | PFailed 
1054
	SrcSpan		-- The start and end of the text span related to
1055 1056
			-- the error.  Might be used in environments which can 
			-- show this span, e.g. by highlighting it.
1057 1058 1059 1060
	Message		-- The error message

data PState = PState { 
	buffer	   :: StringBuffer,
1061 1062
        last_loc   :: SrcSpan,	-- pos of previous token
	last_len   :: !Int,	-- len of previous token
1063 1064 1065 1066 1067
        loc        :: SrcLoc,   -- current loc (end of prev token + 1)
	extsBitmap :: !Int,	-- bitmap that determines permitted extensions
	context	   :: [LayoutContext],
	lex_state  :: [Int]
     }
1068
	-- last_loc and last_len are used when generating error messages,
1069 1070 1071 1072
	-- and in pushCurrentContext only.  Sigh, if only Happy passed the
	-- current token to happyError, we could at least get rid of last_len.
	-- Getting rid of last_loc would require finding another way to 
	-- implement pushCurrentContext (which is only called from one place).
1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086

newtype P a = P { unP :: PState -> ParseResult a }

instance Monad P where
  return = returnP
  (>>=) = thenP
  fail = failP

returnP :: a -> P a
returnP a = P $ \s -> POk s a

thenP :: P a -> (a -> P b) -> P b
(P m) `thenP` k = P $ \ s ->
	case m s of
1087 1088
		POk s1 a         -> (unP (k a)) s1
		PFailed span err -> PFailed span err
1089 1090

failP :: String -> P a
1091
failP msg = P $ \s -> PFailed (last_loc s) (text msg)
1092 1093

failMsgP :: String -> P a
1094
failMsgP msg = P $ \s -> PFailed (last_loc s) (text msg)
1095 1096

failLocMsgP :: SrcLoc -> SrcLoc -> String -> P a
1097 1098 1099 1100
failLocMsgP loc1 loc2 str = P $ \s -> PFailed (mkSrcSpan loc1 loc2) (text str)

failSpanMsgP :: SrcSpan -> String -> P a
failSpanMsgP span msg = P $ \s -> PFailed span (text msg)
1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113

extension :: (Int -> Bool) -> P Bool
extension p = P $ \s -> POk s (p $! extsBitmap s)

getExts :: P Int
getExts = P $ \s -> POk s (extsBitmap s)

setSrcLoc :: SrcLoc -> P ()
setSrcLoc new_loc = P $ \s -> POk s{loc=new_loc} ()

getSrcLoc :: P SrcLoc
getSrcLoc = P $ \s@(PState{ loc=loc }) -> POk s loc

1114
setLastToken :: SrcSpan -> Int -> P ()
1115
setLastToken loc len = P $ \s -> POk s{ last_loc=loc, last_len=len } ()
1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153

type AlexInput = (SrcLoc,StringBuffer)

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

alexGetChar :: AlexInput -> Maybe (Char,AlexInput)
alexGetChar (loc,s) 
  | atEnd s   = Nothing
  | otherwise = c `seq` loc' `seq` s' `seq` Just (c, (loc', s'))
  where c = currentChar s
        loc' = advanceSrcLoc loc c
	s'   = stepOn s

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 } ()

pushLexState :: Int -> P ()
pushLexState ls = P $ \s@PState{ lex_state=l } -> POk s{lex_state=ls:l} ()

popLexState :: P Int
popLexState = P $ \s@PState{ lex_state=ls:l } -> POk s{ lex_state=l } ls

getLexState :: P Int
getLexState = P $ \s@PState{ lex_state=ls:l } -> POk s ls

-- for reasons of efficiency, flags indicating language extensions (eg,
-- -fglasgow-exts or -fparr) are represented by a bitmap stored in an unboxed
-- integer

glaExtsBit, ffiBit, parrBit :: Int
glaExtsBit = 0
ffiBit	   = 1
parrBit	   = 2
arrowsBit  = 4
1154 1155
thBit	   = 5
ipBit      = 6
1156 1157 1158 1159 1160 1161

glaExtsEnabled, ffiEnabled, parrEnabled :: Int -> Bool
glaExtsEnabled flags = testBit flags glaExtsBit
ffiEnabled     flags = testBit flags ffiBit
parrEnabled    flags = testBit flags parrBit
arrowsEnabled  flags = testBit flags arrowsBit
1162 1163
thEnabled      flags = testBit flags thBit
ipEnabled      flags = testBit flags ipBit
1164 1165 1166

-- create a parse state
--
1167 1168
mkPState :: StringBuffer -> SrcLoc -> DynFlags -> PState
mkPState buf loc flags  = 
1169 1170
  PState {
      buffer	 = buf,
1171
      last_loc   = mkSrcSpan loc loc,
1172 1173 1174 1175 1176 1177 1178 1179
      last_len   = 0,
      loc        = loc,
      extsBitmap = fromIntegral bitmap,
      context    = [],
      lex_state  = [bol, if glaExtsEnabled bitmap then glaexts else 0]
	-- we begin in the layout state if toplev_layout is set
    }
    where
1180 1181 1182 1183 1184 1185
      bitmap =     glaExtsBit `setBitIf` dopt Opt_GlasgowExts flags
	       .|. ffiBit     `setBitIf` dopt Opt_FFI         flags
	       .|. parrBit    `setBitIf` dopt Opt_PArr        flags
	       .|. arrowsBit  `setBitIf` dopt Opt_Arrows      flags
	       .|. thBit      `setBitIf` dopt Opt_TH          flags
	       .|. ipBit      `setBitIf` dopt Opt_ImplicitParams flags
1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198
      --
      setBitIf :: Int -> Bool -> Int
      b `setBitIf` cond | cond      = bit b
			| otherwise = 0

getContext :: P [LayoutContext]
getContext = P $ \s@PState{context=ctx} -> POk s ctx

setContext :: [LayoutContext] -> P ()
setContext ctx = P $ \s -> POk s{context=ctx} ()

popContext :: P ()
popContext = P $ \ s@(PState{ buffer = buf, context = ctx, 
1199
			   loc = loc, last_len = len, last_loc = last_loc }) ->
1200 1201
  case ctx of
	(_:tl) -> POk s{ context = tl } ()
1202
	[]     -> PFailed last_loc (srcParseErr buf len)
1203 1204 1205 1206 1207

-- Push a new layout context at the indentation of the last token read.
-- This is only used at the outer level of a module when the 'module'
-- keyword is missing.
pushCurrentContext :: P ()
1208
pushCurrentContext = P $ \ s@PState{ last_loc=loc, context=ctx } ->
1209
  POk s{ context = Layout (srcSpanStartCol loc) : ctx} ()
1210

1211 1212
getOffside :: SrcLoc -> P Ordering
getOffside loc = P $ \s@PState{context=stk} ->
1213
		let ord = case stk of
1214
			(Layout n:_) -> compare (srcLocCol loc) n
1215
			_            -> GT
1216
		in POk s ord
1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237

-- ---------------------------------------------------------------------------
-- Construct a parse error

srcParseErr
  :: StringBuffer	-- current buffer (placed just after the last token)
  -> Int		-- length of the previous token
  -> Message
srcParseErr buf len
  = hcat [ if null token 
	     then ptext SLIT("parse error (possibly incorrect indentation)")
	     else hcat [ptext SLIT("parse error on input "),
          	  	char '`', text token, char '\'']
    ]
  where token = lexemeToString (stepOnBy (-len) buf) len

-- Report a parse failure, giving the span of the previous token as
-- the location of the error.  This is the entry point for errors
-- detected during parsing.
srcParseFail :: P a
srcParseFail = P $ \PState{ buffer = buf, last_len = len, 	
1238
			    last_loc = last_loc } ->
1239
    PFailed last_loc (srcParseErr buf len)
1240 1241 1242 1243 1244 1245 1246 1247

-- A lexical error is reported at a particular position in the source file,
-- not over a token range.  TODO: this is slightly wrong, because we record
-- the error at the character position following the one which caused the
-- error.  We should somehow back up by one character.
lexError :: String -> P a
lexError str = do
  loc <- getSrcLoc
1248 1249
  i@(end,_) <- getInput
  failLocMsgP loc end str
1250 1251 1252 1253 1254

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

1255
lexer :: (Located Token -> P a) -> P a
1256
lexer cont = do
1257
  tok@(L _ tok__) <- lexToken
1258 1259 1260
  --trace ("token: " ++ show tok__) $ do
  cont tok

1261
lexToken :: P (Located Token)
1262 1263 1264 1265 1266
lexToken = do
  inp@(loc1,buf) <- getInput
  sc <- getLexState
  exts <- getExts
  case alexScanUser exts inp sc of
1267 1268 1269
    AlexEOF -> do let span = mkSrcSpan loc1 loc1
		  setLastToken span 0
		  return (L span ITeof)
1270 1271 1272 1273 1274 1275
    AlexError (loc2,_) -> do failLocMsgP loc1 loc2 "lexical error"
    AlexSkip inp2 _ -> do
	setInput inp2
	lexToken
    AlexToken inp2@(end,buf2) len t -> do
	setInput inp2
1276 1277 1278
	let span = mkSrcSpan loc1 end
	span `seq` setLastToken span len
	t span buf len
1279
}