Lexer.x 37.9 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(..), Token__(..), lexer, mkPState, showPFailed,
26 27 28 29 30 31 32
   P(..), ParseResult(..), setSrcLocFor, getSrcLoc, 
   failMsgP, failLocMsgP, srcParseFail,
   popContext, pushCurrentContext,
  ) where

#include "HsVersions.h"

33
import ForeignCall	( Safety(..) )
34 35 36 37 38 39 40
import ErrUtils		( Message )
import Outputable
import StringBuffer
import FastString
import FastTypes
import SrcLoc
import UniqFM
41
import CmdLineOpts
42 43 44 45 46 47
import Ctype
import Util		( maybePrefixMatch )

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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125
$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".
"--"\-* ([^$symbol] .*)?		;

-- '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 }
126
  ^\# pragma .* \n			; -- GCC 3.3 CPP generated, apparently
127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155
  ()					{ 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
<line_prag1> $digit+			{ set_line line_prag1a }
156
<line_prag1a> \" [$graphic \ ]* \"	{ set_file line_prag1b }
157 158 159 160 161
<line_prag1b> .*			{ pop }

-- Haskell-style line pragmas, of the form
--    {-# LINE <line> "<file>" #-}
<line_prag2> $digit+			{ set_line line_prag2a }
162
<line_prag2a> \" [$graphic \ ]* \"	{ set_file line_prag2b }
163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189
<line_prag2b> "#-}"			{ pop }

<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 }
  
  "{-#" 				{ 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

190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206
<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
207 208
  "(|" / { ifExtension arrowsEnabled `alexAndPred` notFollowedBySymbol }
					{ special IToparenbar }
209 210 211 212 213 214 215 216
  "|)" / { ifExtension arrowsEnabled }  { special ITcparenbar }
}

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

217
<glaexts> {
ross's avatar
ross committed
218
  "(#" / { notFollowedBySymbol }	{ token IToubxparen }
219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252
  "#)"					{ 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 }
253
  @qual @conid "#"+		{ idtoken qconid }
254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300
  @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

301
data Token = T SrcLoc{-start-} SrcLoc{-end-} Token__
302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423

data Token__
  = 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
  | 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		-- $( 
424 425
  | ITvarQuote			-- '
  | ITtyQuote			-- ''
426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512

  -- 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

isSpecial :: Token__ -> Bool
-- 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),
513
	( "proc",	ITproc,		 bit arrowsBit)
514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544
     ]

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

545
type Action = SrcLoc -> SrcLoc -> StringBuffer -> Int -> P Token
546 547

special :: Token__ -> Action
548
special tok loc end _buf len = return (T loc end tok)
549

550
token, layout_token :: Token__ -> Action
551 552
token t loc end buf len = return (T loc end t)
layout_token t loc end buf len = pushLexState layout >> return (T loc end t)
553 554

idtoken :: (StringBuffer -> Int -> Token__) -> Action
555
idtoken f loc end buf len = return (T loc end $! (f buf len))
556 557

skip_one_varid :: (FastString -> Token__) -> Action
558 559
skip_one_varid f loc end buf len 
  = return (T loc end $! f (lexemeToFastString (stepOn buf) (len-1)))
560 561

strtoken :: (String -> Token__) -> Action
562 563
strtoken f loc end buf len = 
  return (T loc end $! (f $! lexemeToString buf len))
564 565 566

init_strtoken :: Int -> (String -> Token__) -> Action
-- like strtoken, but drops the last N character(s)
567 568
init_strtoken drop f loc end buf len = 
  return (T loc end $! (f $! lexemeToString buf (len-drop)))
569 570

begin :: Int -> Action
571
begin code _loc _end _str _len = do pushLexState code; lexToken
572 573

pop :: Action
574
pop _loc _end _buf _len = do popLexState; lexToken
575 576

pop_and :: Action -> Action
577
pop_and act loc end buf len = do popLexState; act loc end buf len
578 579 580

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

ross's avatar
ross committed
581 582 583
notFollowedBySymbol _ _ _ (_,buf)
  = atEnd buf || currentChar buf `notElem` "!#$%&*+./<=>?@\\^|-~"

584 585
ifExtension pred bits _ _ _ = pred bits

586 587 588 589 590
{-
  nested comments require traversing by hand, they can't be parsed
  using regular expressions.
-}
nested_comment :: Action
591
nested_comment loc _end _str _len = do
592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611
  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

612
        err input = do failLocMsgP loc (fst input) "unterminated `{-'"
613 614

open_brace, close_brace :: Action
615
open_brace  loc end _str _len = do 
616 617
  ctx <- getContext
  setContext (NoLayout:ctx)
618 619
  return (T loc end ITocurly)
close_brace loc end _str _len = do 
620
  popContext
621
  return (T loc end ITccurly)
622 623 624 625

-- 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.
626
check_qvarid loc end buf len = do
627 628 629 630 631 632 633 634 635 636 637 638
  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
639
	token     = T loc end (ITqvarid (mod,var))
640 641

	try_again = do
642
		setInput (loc,buf)
643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670
		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))

671
varid loc end buf len = 
672 673 674
  case lookupUFM reservedWordsFM fs of
	Just (keyword,0)    -> do
		maybe_layout keyword
675
		return (T loc end keyword)
676 677 678
	Just (keyword,exts) -> do
		b <- extension (\i -> exts .&. i /= 0)
		if b then do maybe_layout keyword
679 680 681
			     return (T loc end keyword)
		     else return (T loc end (ITvarid fs))
	_other -> return (T loc end (ITvarid fs))
682 683 684 685 686 687 688 689 690 691 692 693
  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

694
sym con loc end buf len = 
695
  case lookupUFM reservedSymsFM fs of
696
	Just (keyword,0)    -> return (T loc end keyword)
697 698
	Just (keyword,exts) -> do
		b <- extension (\i -> exts .&. i /= 0)
699 700 701
		if b then return (T loc end keyword)
		     else return (T loc end $! con fs)
	_other -> return (T loc end $! con fs)
702 703 704
  where
	fs = lexemeToFastString buf len

705 706
tok_decimal loc end buf len 
  = return (T loc end (ITinteger  $! parseInteger buf len 10 oct_or_dec))
707

708 709
tok_octal loc end buf len 
  = return (T loc end (ITinteger  $! parseInteger (stepOnBy 2 buf) (len-2) 8 oct_or_dec))
710

711 712
tok_hexadecimal loc end buf len 
  = return (T loc end (ITinteger  $! parseInteger (stepOnBy 2 buf) (len-2) 16 hex))
713

714 715
prim_decimal loc end buf len 
  = return (T loc end (ITprimint  $! parseInteger buf (len-1) 10 oct_or_dec))
716

717 718
prim_octal loc end buf len 
  = return (T loc end (ITprimint  $! parseInteger (stepOnBy 2 buf) (len-3) 8 oct_or_dec))
719

720 721
prim_hexadecimal loc end buf len 
  = return (T loc end (ITprimint  $! parseInteger (stepOnBy 2 buf) (len-3) 16 hex))
722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737

tok_float        str = ITrational $! readRational__ str
prim_float       str = ITprimfloat  $! readRational__ str
prim_double      str = ITprimdouble $! readRational__ str

parseInteger :: StringBuffer -> Int -> Integer -> (Char->Int) -> Integer
parseInteger buf len radix to_int 
  = go 0 0
  where go i x | i == len  = x
	       | otherwise = go (i+1) (x * radix + toInteger (to_int (lookAhead buf i)))

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

-- we're at the first token on a line, insert layout tokens if necessary
do_bol :: Action
738 739
do_bol loc end _str _len = do
	pos <- getOffside end
740 741 742 743 744
	case pos of
	    LT -> do
                --trace "layout: inserting '}'" $ do
		popContext
		-- do NOT pop the lex state, we might have a ';' to insert
745
		return (T loc end ITvccurly)
746 747 748
	    EQ -> do
                --trace "layout: inserting ';'" $ do
		popLexState
749
		return (T loc end ITsemi)
750 751 752 753 754 755 756
	    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
757
maybe_layout ITmdo	= pushLexState layout_do
758 759 760
maybe_layout ITof	= pushLexState layout
maybe_layout ITlet	= pushLexState layout
maybe_layout ITwhere	= pushLexState layout
761
maybe_layout ITrec	= pushLexState layout
762 763 764 765 766 767 768 769 770 771 772
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.
--
773
new_layout_context strict loc end _buf _len = do
774
    popLexState
775
    let offset = srcLocCol loc
776 777 778 779 780 781 782 783
    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
784
		return (T loc end ITvocurly)
785 786
	other -> do
		setContext (Layout offset : ctx)
787
		return (T loc end ITvocurly)
788

789
do_layout_left loc end _buf _len = do
790 791
    popLexState
    pushLexState bol  -- we must be at the start of a line
792
    return (T loc end ITvccurly)
793 794 795 796 797

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

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

set_file :: Int -> Action
807
set_file code loc end buf len = do
808
  let file = lexemeToFastString (stepOn buf) (len-2)
809
  setSrcLoc (mkSrcLoc file (srcLocLine end) (srcLocCol end))
810 811 812 813 814 815 816 817 818 819
  popLexState
  pushLexState code
  lexToken

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

-- This stuff is horrible.  I hates it.

lex_string_tok :: Action
820
lex_string_tok loc end buf len = do
821 822
  tok <- lex_string ""
  end <- getSrcLoc 
823
  return (T loc end tok)
824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860

lex_string :: String -> P Token__
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)

861 862 863 864 865 866 867 868
lex_char :: P Char
lex_char = do
  mc <- getCharOrFail
  case mc of
      '\\' -> lex_escape
      c | is_any c -> return c
      _other -> lit_error

869 870 871 872 873 874 875 876 877
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
878 879 880 881 882 883
-- 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
884
lex_char_tok loc _end buf len = do	-- We've seen '
885 886 887 888 889 890 891 892
   i1 <- getInput	-- Look ahead to first character
   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
893
			return (T loc end2 ITtyQuote)
894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915
		   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	
916
			if th_exts then return (T loc (fst i1) ITvarQuote)
917 918 919 920 921 922 923 924 925
				   else lit_error

finish_char_tok :: SrcLoc -> Char -> P Token
finish_char_tok loc ch	-- We've already seen the closing quote
			-- Just need to check for trailing #
  = do	glaexts <- extension glaExtsEnabled
	if glaexts then do
		i@(end,_) <- getInput
		case alexGetChar i of
926 927
			Just ('#',i@(end,_)) -> do
				setInput i
928
				return (T loc end (ITprimchar ch))
929
			_other ->
930 931 932
					return (T loc end (ITchar ch))
	  else do end <- getSrcLoc
		  return (T loc end (ITchar ch))
933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 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 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115

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

	'x'   -> readNum is_hexdigit 16 hex
	'o'   -> readNum is_octdigit  8 oct_or_dec
	x | is_digit x -> readNum2 is_digit 10 oct_or_dec (oct_or_dec x)

	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

is_hexdigit c
	=  is_digit c 
	|| (c >= 'a' && c <= 'f')
	|| (c >= 'A' && c <= 'F')

hex c | is_digit c = ord c - ord '0'
      | otherwise  = ord (to_lower c) - ord 'a' + 10

oct_or_dec c = ord c - ord '0'

is_octdigit c = c >= '0' && c <= '7'

to_lower c 
  | c >=  'A' && c <= 'Z' = chr (ord c - (ord 'A' - ord 'a'))
  | otherwise = c

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

-- -----------------------------------------------------------------------------
-- Floats

readRational :: ReadS Rational -- NB: doesn't handle leading "-"
readRational r = do 
     (n,d,s) <- readFix r
     (k,t)   <- readExp s
     return ((n%1)*10^^(k-d), t)
 where
     readFix r = do
	(ds,s)  <- lexDecDigits r
	(ds',t) <- lexDotDigits s
	return (read (ds++ds'), length ds', t)

     readExp (e:s) | e `elem` "eE" = readExp' s
     readExp s			   = return (0,s)

     readExp' ('+':s) = readDec s
     readExp' ('-':s) = do
			(k,t) <- readDec s
			return (-k,t)
     readExp' s	      = readDec s

     readDec s = do
        (ds,r) <- nonnull isDigit s
        return (foldl1 (\n d -> n * 10 + d) [ ord d - ord '0' | d <- ds ],
                r)

     lexDecDigits = nonnull isDigit

     lexDotDigits ('.':s) = return (span isDigit s)
     lexDotDigits s       = return ("",s)

     nonnull p s = do (cs@(_:_),t) <- return (span p s)
                      return (cs,t)

readRational__ :: String -> Rational -- NB: *does* handle a leading "-"
readRational__ top_s
  = case top_s of
      '-' : xs -> - (read_me xs)
      xs       -> read_me xs
  where
    read_me s
      = case (do { (x,"") <- readRational s ; return x }) of
	  [x] -> x
	  []  -> error ("readRational__: no parse:"        ++ top_s)
	  _   -> error ("readRational__: ambiguous parse:" ++ top_s)

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

data LayoutContext
  = NoLayout
  | Layout !Int

data ParseResult a
  = POk PState a
  | PFailed 
1116 1117 1118
	SrcLoc SrcLoc	-- The start and end of the text span related to
			-- the error.  Might be used in environments which can 
			-- show this span, e.g. by highlighting it.
1119 1120
	Message		-- The error message

1121
showPFailed loc1 loc2 err = hcat [ppr loc1, text ": ", err]
1122 1123 1124

data PState = PState { 
	buffer	   :: StringBuffer,
1125 1126
        last_loc   :: SrcLoc,		-- pos of previous token
	last_len   :: !Int,		-- len of previous token
1127 1128 1129 1130 1131
        loc        :: SrcLoc,   -- current loc (end of prev token + 1)
	extsBitmap :: !Int,	-- bitmap that determines permitted extensions
	context	   :: [LayoutContext],
	lex_state  :: [Int]
     }
1132
	-- last_loc and last_len are used when generating error messages,
1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147
	-- and in pushCurrentContext only.

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
1148 1149
		POk s1 a          -> (unP (k a)) s1
		PFailed l1 l2 err -> PFailed l1 l2 err
1150 1151

failP :: String -> P a
1152
failP msg = P $ \s -> PFailed (last_loc s) (loc s) (text msg)
1153 1154

failMsgP :: String -> P a
1155
failMsgP msg = P $ \s -> PFailed (last_loc s) (loc s) (text msg)
1156 1157

failLocMsgP :: SrcLoc -> SrcLoc -> String -> P a
1158
failLocMsgP loc1 loc2 str = P $ \s -> PFailed loc1 loc2 (text str)
1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173

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} ()

-- tmp, for supporting stuff in RdrHsSyn.  The scope better not include
-- any calls to the lexer, because it assumes things about the SrcLoc.
setSrcLocFor :: SrcLoc -> P a -> P a
setSrcLocFor new_loc scope = P $ \s@PState{ loc = old_loc } -> 
  case unP scope s{loc=new_loc} of
1174
	PFailed l1 l2 msg -> PFailed l1 l2 msg
1175 1176 1177 1178 1179
	POk _ r -> POk s r

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

1180 1181
setLastToken :: SrcLoc -> Int -> P ()
setLastToken loc len = P $ \s -> POk s{ last_loc=loc, last_len=len } ()
1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219

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
1220 1221
thBit	   = 5
ipBit      = 6
1222 1223 1224 1225 1226 1227

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
1228 1229
thEnabled      flags = testBit flags thBit
ipEnabled      flags = testBit flags ipBit
1230 1231 1232

-- create a parse state
--
1233 1234
mkPState :: StringBuffer -> SrcLoc -> DynFlags -> PState
mkPState buf loc flags  = 
1235 1236
  PState {
      buffer	 = buf,
1237
      last_loc   = loc,
1238 1239 1240 1241 1242 1243 1244 1245
      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
1246 1247 1248 1249 1250 1251
      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
1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264
      --
      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, 
1265
			   loc = loc, last_len = len, last_loc = last_loc }) ->
1266 1267
  case ctx of
	(_:tl) -> POk s{ context = tl } ()
1268
	[]     -> PFailed last_loc loc (srcParseErr buf len)
1269 1270 1271 1272 1273

-- 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 ()
1274 1275
pushCurrentContext = P $ \ s@PState{ last_loc=loc, context=ctx } ->
  POk s{ context = Layout (srcLocCol loc) : ctx} ()
1276

1277 1278
getOffside :: SrcLoc -> P Ordering
getOffside loc = P $ \s@PState{context=stk} ->
1279
		let ord = case stk of
1280
			(Layout n:_) -> compare (srcLocCol loc) n
1281
			_            -> GT
1282
		in POk s ord
1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303

-- ---------------------------------------------------------------------------
-- 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, 	
1304 1305
				last_loc = last_loc, loc = loc } ->
    PFailed last_loc loc (srcParseErr buf len)
1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321

-- 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
  failLocMsgP loc loc str

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

lexer :: (Token -> P a) -> P a
lexer cont = do
1322
  tok@(T _ _ tok__) <- lexToken
1323 1324 1325 1326 1327 1328 1329 1330 1331
  --trace ("token: " ++ show tok__) $ do
  cont tok

lexToken :: P Token
lexToken = do
  inp@(loc1,buf) <- getInput
  sc <- getLexState
  exts <- getExts
  case alexScanUser exts inp sc of
1332 1333
    AlexEOF -> do setLastToken loc1 0
		  return (T loc1 loc1 ITeof)
1334 1335 1336 1337 1338 1339
    AlexError (loc2,_) -> do failLocMsgP loc1 loc2 "lexical error"
    AlexSkip inp2 _ -> do
	setInput inp2
	lexToken
    AlexToken inp2@(end,buf2) len t -> do
	setInput inp2
1340 1341
	setLastToken loc1 len
	t loc1 end buf len
1342
}