Lexer.x 89.1 KB
Newer Older
1
-----------------------------------------------------------------------------
2
-- (c) The University of Glasgow, 2006
3 4 5 6 7 8 9
--
-- 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
Ian Lynagh's avatar
Ian Lynagh committed
10 11
-- file is maintained.  Every token has a start and end RealSrcLoc
-- attached to it.
12 13 14 15 16 17 18 19 20 21 22
--
-----------------------------------------------------------------------------

--   ToDo / known bugs:
--    - 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

23
--   qualified operator NOTES.
Ian Lynagh's avatar
Ian Lynagh committed
24
--
25 26 27 28 29 30 31 32 33
--   - If M.(+) is a single lexeme, then..
--     - Probably (+) should be a single lexeme too, for consistency.
--       Otherwise ( + ) would be a prefix operator, but M.( + ) would not be.
--     - But we have to rule out reserved operators, otherwise (..) becomes
--       a different lexeme.
--     - Should we therefore also rule out reserved operators in the qualified
--       form?  This is quite difficult to achieve.  We don't do it for
--       qualified varids.

34
{
35
-- XXX The above flags turn off warnings in the generated code:
36
{-# LANGUAGE BangPatterns #-}
37 38 39 40 41 42 43 44
{-# OPTIONS_GHC -fno-warn-unused-matches #-}
{-# OPTIONS_GHC -fno-warn-unused-binds #-}
{-# OPTIONS_GHC -fno-warn-unused-imports #-}
{-# OPTIONS_GHC -fno-warn-missing-signatures #-}
-- But alex still generates some code that causes the "lazy unlifted bindings"
-- warning, and old compilers don't know about it so we can't easily turn
-- it off, so for now we use the sledge hammer:
{-# OPTIONS_GHC -w #-}
45

46 47
{-# OPTIONS_GHC -funbox-strict-fields #-}

48
module Lexer (
49
   Token(..), lexer, pragState, mkPState, PState(..),
Ian Lynagh's avatar
Ian Lynagh committed
50
   P(..), ParseResult(..), getSrcLoc,
51
   getPState, getDynFlags, withThisPackage,
52
   failLocMsgP, failSpanMsgP, srcParseFail,
Ian Lynagh's avatar
Ian Lynagh committed
53
   getMessages,
54
   popContext, pushCurrentContext, setLastToken, setSrcLoc,
vivian's avatar
vivian committed
55
   activeContext, nextIsEOF,
simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
56
   getLexState, popLexState, pushLexState,
57
   extension, bangPatEnabled, datatypeContextsEnabled,
58
   traditionalRecordSyntaxEnabled,
59
   typeLiteralsEnabled,
60
   explicitNamespacesEnabled,
Jedai's avatar
Jedai committed
61 62
   addWarning,
   lexTokenStream
63 64
  ) where

65 66
import Bag
import ErrUtils
67 68 69 70 71
import Outputable
import StringBuffer
import FastString
import SrcLoc
import UniqFM
72
import DynFlags
73
import Module
74
import Ctype
Ian Lynagh's avatar
Ian Lynagh committed
75 76
import BasicTypes       ( InlineSpec(..), RuleMatchInfo(..), FractionalLit(..) )
import Util             ( readRational )
77

78
import Control.Monad
Simon Marlow's avatar
Simon Marlow committed
79
import Data.Bits
80
import Data.Char
81
import Data.List
Ian Lynagh's avatar
Ian Lynagh committed
82
import Data.Maybe
83 84
import Data.Map (Map)
import qualified Data.Map as Map
Simon Marlow's avatar
Simon Marlow committed
85
import Data.Ratio
86
import Data.Word
87 88
}

89
$unispace    = \x05 -- Trick Alex into handling Unicode. See alexGetChar.
90
$whitechar   = [\ \n\r\f\v $unispace]
91
$white_no_nl = $whitechar # \n
92
$tab         = \t
93 94

$ascdigit  = 0-9
95
$unidigit  = \x03 -- Trick Alex into handling Unicode. See alexGetChar.
96
$decdigit  = $ascdigit -- for now, should really be $digit (ToDo)
97 98 99
$digit     = [$ascdigit $unidigit]

$special   = [\(\)\,\;\[\]\`\{\}]
100
$ascsymbol = [\!\#\$\%\&\*\+\.\/\<\=\>\?\@\\\^\|\-\~]
101
$unisymbol = \x04 -- Trick Alex into handling Unicode. See alexGetChar.
102 103
$symbol    = [$ascsymbol $unisymbol] # [$special \_\:\"\']

104
$unilarge  = \x01 -- Trick Alex into handling Unicode. See alexGetChar.
105
$asclarge  = [A-Z]
106 107
$large     = [$asclarge $unilarge]

108
$unismall  = \x02 -- Trick Alex into handling Unicode. See alexGetChar.
109
$ascsmall  = [a-z]
110 111
$small     = [$ascsmall $unismall \_]

112
$unigraphic = \x06 -- Trick Alex into handling Unicode. See alexGetChar.
113
$graphic   = [$small $large $symbol $digit $special $unigraphic \:\"\']
114

Ian Lynagh's avatar
Ian Lynagh committed
115
$octit     = 0-7
116
$hexit     = [$decdigit A-F a-f]
117 118 119 120
$symchar   = [$symbol \:]
$nl        = [\n\r]
$idchar    = [$small $large $digit \']

121 122
$pragmachar = [$small $large $digit]

123 124
$docsym    = [\| \^ \* \$]

125 126 127 128 129 130
@varid     = $small $idchar*
@conid     = $large $idchar*

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

131
@decimal     = $decdigit+
132 133 134 135 136 137 138 139 140
@octal       = $octit+
@hexadecimal = $hexit+
@exponent    = [eE] [\-\+]? @decimal

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

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

141 142 143 144 145
-- normal signed numerical literals can only be explicitly negative,
-- not explicitly positive (contrast @exponent)
@negative = \-
@signed = @negative ?

146 147 148
haskell :-

-- everywhere: skip whitespace and comments
Ian Lynagh's avatar
Ian Lynagh committed
149
$white_no_nl+ ;
150
$tab+         { warn Opt_WarnTabs (text "Tab character") }
151 152 153 154 155

-- 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
156 157 158 159
-- are). We also rule out nested Haddock comments, if the -haddock flag is
-- set.

"{-" / { isNormalComment } { nested_comment lexToken }
160 161 162 163

-- 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.
164 165

-- Since Haddock comments aren't valid in every state, we need to rule them
Ian Lynagh's avatar
Ian Lynagh committed
166
-- out here.
167 168 169 170 171 172 173

-- The following two rules match comments that begin with two dashes, but
-- continue with a different character. The rules test that this character
-- is not a symbol (in which case we'd have a varsym), and that it's not a
-- space followed by a Haddock comment symbol (docsym) (in which case we'd
-- have a Haddock comment). The rules then munch the rest of the line.

Jedai's avatar
Jedai committed
174 175
"-- " ~[$docsym \#] .* { lineCommentToken }
"--" [^$symbol : \ ] .* { lineCommentToken }
176 177 178

-- Next, match Haddock comments if no -haddock flag

Jedai's avatar
Jedai committed
179
"-- " [$docsym \#] .* / { ifExtension (not . haddockEnabled) } { lineCommentToken }
180 181 182 183 184 185 186

-- Now, when we've matched comments that begin with 2 dashes and continue
-- with a different character, we need to match comments that begin with three
-- or more dashes (which clearly can't be Haddock comments). We only need to
-- make sure that the first non-dash character isn't a symbol, and munch the
-- rest of the line.

Jedai's avatar
Jedai committed
187
"---"\-* [^$symbol :] .* { lineCommentToken }
188 189 190 191

-- Since the previous rules all match dashes followed by at least one
-- character, we also need to match a whole line filled with just dashes.

Jedai's avatar
Jedai committed
192
"--"\-* / { atEOL } { lineCommentToken }
193 194 195 196

-- We need this rule since none of the other single line comment rules
-- actually match this case.

Jedai's avatar
Jedai committed
197
"-- " / { atEOL } { lineCommentToken }
198 199 200 201 202 203 204 205 206 207 208

-- '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> {
Ian Lynagh's avatar
Ian Lynagh committed
209 210 211 212 213
  \n                                    ;
  ^\# (line)?                           { begin line_prag1 }
  ^\# pragma .* \n                      ; -- GCC 3.3 CPP generated, apparently
  ^\# \! .* \n                          ; -- #!, for scripts
  ()                                    { do_bol }
214 215 216 217 218 219
}

-- 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> {
Ian Lynagh's avatar
Ian Lynagh committed
220 221 222 223
  \{ / { notFollowedBy '-' }            { hopefully_open_brace }
        -- we might encounter {-# here, but {- has been handled already
  \n                                    ;
  ^\# (line)?                           { begin line_prag1 }
224 225 226
}

-- do is treated in a subtly different way, see new_layout_context
Ian Lynagh's avatar
Ian Lynagh committed
227 228
<layout>    ()                          { new_layout_context True }
<layout_do> ()                          { new_layout_context False }
229 230 231 232

-- 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.
Ian Lynagh's avatar
Ian Lynagh committed
233
<layout_left>  ()                       { do_layout_left }
234

Ian Lynagh's avatar
Ian Lynagh committed
235
<0,option_prags> \n                     { begin bol }
236

237 238
"{-#" $whitechar* $pragmachar+ / { known_pragma linePrags }
                                { dispatch_pragmas linePrags }
239 240 241

-- single-line line pragmas, of the form
--    # <line> "<file>" <extra-stuff> \n
Ian Lynagh's avatar
Ian Lynagh committed
242 243 244
<line_prag1> $decdigit+                 { setLine line_prag1a }
<line_prag1a> \" [$graphic \ ]* \"      { setFile line_prag1b }
<line_prag1b> .*                        { pop }
245 246 247

-- Haskell-style line pragmas, of the form
--    {-# LINE <line> "<file>" #-}
Ian Lynagh's avatar
Ian Lynagh committed
248 249 250
<line_prag2> $decdigit+                 { setLine line_prag2a }
<line_prag2a> \" [$graphic \ ]* \"      { setFile line_prag2b }
<line_prag2b> "#-}"|"-}"                { pop }
251 252
   -- NOTE: accept -} at the end of a LINE pragma, for compatibility
   -- with older versions of GHC which generated these.
253

254
<0,option_prags> {
Ian Lynagh's avatar
Ian Lynagh committed
255
  "{-#" $whitechar* $pragmachar+
256 257 258 259 260
        $whitechar+ $pragmachar+ / { known_pragma twoWordPrags }
                                 { dispatch_pragmas twoWordPrags }

  "{-#" $whitechar* $pragmachar+ / { known_pragma oneWordPrags }
                                 { dispatch_pragmas oneWordPrags }
261

262
  -- We ignore all these pragmas, but don't generate a warning for them
263 264
  "{-#" $whitechar* $pragmachar+ / { known_pragma ignoredPrags }
                                 { dispatch_pragmas ignoredPrags }
265 266

  -- ToDo: should only be valid inside a pragma:
Ian Lynagh's avatar
Ian Lynagh committed
267
  "#-}"                          { endPrag }
268 269
}

270
<option_prags> {
271 272 273
  "{-#"  $whitechar* $pragmachar+ / { known_pragma fileHeaderPrags }
                                   { dispatch_pragmas fileHeaderPrags }

Ian Lynagh's avatar
Ian Lynagh committed
274
  "-- #"                           { multiline_doc_comment }
David Waern's avatar
David Waern committed
275 276
}

277 278
<0> {
  -- In the "0" mode we ignore these pragmas
279
  "{-#"  $whitechar* $pragmachar+ / { known_pragma fileHeaderPrags }
280 281 282
                     { nested_comment lexToken }
}

David Waern's avatar
David Waern committed
283
<0> {
Jedai's avatar
Jedai committed
284
  "-- #" .* { lineCommentToken }
285
}
286

287
<0,option_prags> {
288 289
  "{-#"  { warnThen Opt_WarnUnrecognisedPragmas (text "Unrecognised pragma")
                    (nested_comment lexToken) }
290 291
}

292 293
-- '0' state: ordinary lexemes

294 295
-- Haddock comments

296
<0,option_prags> {
David Waern's avatar
David Waern committed
297 298
  "-- " $docsym      / { ifExtension haddockEnabled } { multiline_doc_comment }
  "{-" \ ? $docsym   / { ifExtension haddockEnabled } { nested_doc_comment }
299 300
}

301 302
-- "special" symbols

303
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
304 305
  "[:" / { ifExtension parrEnabled }    { token ITopabrack }
  ":]" / { ifExtension parrEnabled }    { token ITcpabrack }
306
}
Ian Lynagh's avatar
Ian Lynagh committed
307

308
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
309 310 311 312 313 314 315 316
  "[|"      / { 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 }
317

318 319 320 321
-- For backward compatibility, accept the old dollar syntax
  "[$" @varid "|"  / { ifExtension qqEnabled }
                     { lex_quasiquote_tok }

322
  "[" @varid "|"  / { ifExtension qqEnabled }
323
                     { lex_quasiquote_tok }
324 325 326 327

  -- qualified quasi-quote (#5555)
  "[" @qual @varid "|"  / { ifExtension qqEnabled }
                          { lex_qquasiquote_tok }
328 329
}

330
<0> {
ross's avatar
ross committed
331
  "(|" / { ifExtension arrowsEnabled `alexAndPred` notFollowedBySymbol }
Ian Lynagh's avatar
Ian Lynagh committed
332
                                        { special IToparenbar }
333 334 335
  "|)" / { ifExtension arrowsEnabled }  { special ITcparenbar }
}

336
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
337
  \? @varid / { ifExtension ipEnabled } { skip_one_varid ITdupipvarid }
338 339
}

340
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
341 342 343 344 345 346
  "(#" / { ifExtension unboxedTuplesEnabled `alexAndPred` notFollowedBySymbol }
         { token IToubxparen }
  "#)" / { ifExtension unboxedTuplesEnabled }
         { token ITcubxparen }
}

347
<0,option_prags> {
Ian Lynagh's avatar
Ian Lynagh committed
348 349 350 351 352 353 354 355 356 357
  \(                                    { special IToparen }
  \)                                    { special ITcparen }
  \[                                    { special ITobrack }
  \]                                    { special ITcbrack }
  \,                                    { special ITcomma }
  \;                                    { special ITsemi }
  \`                                    { special ITbackquote }

  \{                                    { open_brace }
  \}                                    { close_brace }
358 359
}

360
<0,option_prags> {
Ian Lynagh's avatar
Ian Lynagh committed
361 362 363 364
  @qual @varid                  { idtoken qvarid }
  @qual @conid                  { idtoken qconid }
  @varid                        { varid }
  @conid                        { idtoken conid }
365 366
}

367
<0> {
368 369 370 371
  @qual @varid "#"+ / { ifExtension magicHashEnabled } { idtoken qvarid }
  @qual @conid "#"+ / { ifExtension magicHashEnabled } { idtoken qconid }
  @varid "#"+       / { ifExtension magicHashEnabled } { varid }
  @conid "#"+       / { ifExtension magicHashEnabled } { idtoken conid }
372 373
}

374 375
-- ToDo: - move `var` and (sym) into lexical syntax?
--       - remove backquote from $special?
376
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
377 378
  @qual @varsym                                    { idtoken qvarsym }
  @qual @consym                                    { idtoken qconsym }
379 380
  @varsym                                          { varsym }
  @consym                                          { consym }
381 382
}

383 384
-- For the normal boxed literals we need to be careful
-- when trying to be close to Haskell98
385
<0> {
386
  -- Normal integral literals (:: Num a => a, from Integer)
Ian Lynagh's avatar
Ian Lynagh committed
387 388 389
  @decimal           { tok_num positive 0 0 decimal }
  0[oO] @octal       { tok_num positive 2 2 octal }
  0[xX] @hexadecimal { tok_num positive 2 2 hexadecimal }
390 391

  -- Normal rational literals (:: Fractional a => a, from Rational)
Ian Lynagh's avatar
Ian Lynagh committed
392
  @floating_point    { strtoken tok_float }
393 394
}

395
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
396
  -- Unboxed ints (:: Int#) and words (:: Word#)
397 398
  -- It's simpler (and faster?) to give separate cases to the negatives,
  -- especially considering octal/hexadecimal prefixes.
Ian Lynagh's avatar
Ian Lynagh committed
399 400 401 402 403
  @decimal                     \# / { ifExtension magicHashEnabled } { tok_primint positive 0 1 decimal }
  0[oO] @octal                 \# / { ifExtension magicHashEnabled } { tok_primint positive 2 3 octal }
  0[xX] @hexadecimal           \# / { ifExtension magicHashEnabled } { tok_primint positive 2 3 hexadecimal }
  @negative @decimal           \# / { ifExtension magicHashEnabled } { tok_primint negative 1 2 decimal }
  @negative 0[oO] @octal       \# / { ifExtension magicHashEnabled } { tok_primint negative 3 4 octal }
404
  @negative 0[xX] @hexadecimal \# / { ifExtension magicHashEnabled } { tok_primint negative 3 4 hexadecimal }
405

Ian Lynagh's avatar
Ian Lynagh committed
406 407 408 409
  @decimal                     \# \# / { ifExtension magicHashEnabled } { tok_primword 0 2 decimal }
  0[oO] @octal                 \# \# / { ifExtension magicHashEnabled } { tok_primword 2 4 octal }
  0[xX] @hexadecimal           \# \# / { ifExtension magicHashEnabled } { tok_primword 2 4 hexadecimal }

410 411
  -- Unboxed floats and doubles (:: Float#, :: Double#)
  -- prim_{float,double} work with signed literals
412 413
  @signed @floating_point \# / { ifExtension magicHashEnabled } { init_strtoken 1 tok_primfloat }
  @signed @floating_point \# \# / { ifExtension magicHashEnabled } { init_strtoken 2 tok_primdouble }
414 415 416 417 418 419
}

-- 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.
420
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
421 422
  \'                            { lex_char_tok }
  \"                            { lex_string_tok }
423 424 425 426 427 428
}

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

429
data Token
Ian Lynagh's avatar
Ian Lynagh committed
430
  = ITas                        -- Haskell keywords
431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453
  | ITcase
  | ITclass
  | ITdata
  | ITdefault
  | ITderiving
  | ITdo
  | ITelse
  | IThiding
  | ITif
  | ITimport
  | ITin
  | ITinfix
  | ITinfixl
  | ITinfixr
  | ITinstance
  | ITlet
  | ITmodule
  | ITnewtype
  | ITof
  | ITqualified
  | ITthen
  | ITtype
  | ITwhere
Ian Lynagh's avatar
Ian Lynagh committed
454
  | ITscc                       -- ToDo: remove (we use {-# SCC "..." #-} now)
455

Ian Lynagh's avatar
Ian Lynagh committed
456
  | ITforall                    -- GHC extension keywords
457 458 459 460 461
  | ITforeign
  | ITexport
  | ITlabel
  | ITdynamic
  | ITsafe
462
  | ITinterruptible
463 464 465
  | ITunsafe
  | ITstdcallconv
  | ITccallconv
466
  | ITcapiconv
467
  | ITprimcallconv
468
  | ITmdo
469
  | ITfamily
470 471 472
  | ITgroup
  | ITby
  | ITusing
473

Ian Lynagh's avatar
Ian Lynagh committed
474
  -- Pragmas
475
  | ITinline_prag InlineSpec RuleMatchInfo
Ian Lynagh's avatar
Ian Lynagh committed
476 477
  | ITspec_prag                 -- SPECIALISE
  | ITspec_inline_prag Bool     -- SPECIALISE INLINE (or NOINLINE)
478 479
  | ITsource_prag
  | ITrules_prag
Ian Lynagh's avatar
Ian Lynagh committed
480
  | ITwarning_prag
481 482 483
  | ITdeprecated_prag
  | ITline_prag
  | ITscc_prag
andy@galois.com's avatar
andy@galois.com committed
484
  | ITgenerated_prag
485
  | ITcore_prag                 -- hdaume: core annotations
486
  | ITunpack_prag
487
  | ITnounpack_prag
488
  | ITann_prag
489
  | ITclose_prag
490 491 492
  | IToptions_prag String
  | ITinclude_prag String
  | ITlanguage_prag
493 494
  | ITvect_prag
  | ITvect_scalar_prag
495
  | ITnovect_prag
496
  | ITctype
497

Ian Lynagh's avatar
Ian Lynagh committed
498
  | ITdotdot                    -- reserved symbols
499 500 501 502 503 504 505 506 507
  | ITcolon
  | ITdcolon
  | ITequal
  | ITlam
  | ITvbar
  | ITlarrow
  | ITrarrow
  | ITat
  | ITtilde
batterseapower's avatar
batterseapower committed
508
  | ITtildehsh
509 510 511 512 513 514
  | ITdarrow
  | ITminus
  | ITbang
  | ITstar
  | ITdot

Ian Lynagh's avatar
Ian Lynagh committed
515
  | ITbiglam                    -- GHC-extension symbols
516

Ian Lynagh's avatar
Ian Lynagh committed
517
  | ITocurly                    -- special symbols
518 519 520 521
  | ITccurly
  | ITvocurly
  | ITvccurly
  | ITobrack
Ian Lynagh's avatar
Ian Lynagh committed
522 523
  | ITopabrack                  -- [:, for parallel arrays with -XParallelArrays
  | ITcpabrack                  -- :], for parallel arrays with -XParallelArrays
524 525 526 527 528 529 530 531 532
  | ITcbrack
  | IToparen
  | ITcparen
  | IToubxparen
  | ITcubxparen
  | ITsemi
  | ITcomma
  | ITunderscore
  | ITbackquote
dreixel's avatar
dreixel committed
533
  | ITsimpleQuote               --  '
534

Ian Lynagh's avatar
Ian Lynagh committed
535
  | ITvarid   FastString        -- identifiers
536 537 538 539 540 541 542
  | ITconid   FastString
  | ITvarsym  FastString
  | ITconsym  FastString
  | ITqvarid  (FastString,FastString)
  | ITqconid  (FastString,FastString)
  | ITqvarsym (FastString,FastString)
  | ITqconsym (FastString,FastString)
543 544
  | ITprefixqvarsym (FastString,FastString)
  | ITprefixqconsym (FastString,FastString)
545

Ian Lynagh's avatar
Ian Lynagh committed
546
  | ITdupipvarid   FastString   -- GHC extension: implicit param: ?x
547 548 549 550

  | ITchar       Char
  | ITstring     FastString
  | ITinteger    Integer
551
  | ITrational   FractionalLit
552 553 554 555

  | ITprimchar   Char
  | ITprimstring FastString
  | ITprimint    Integer
Ian Lynagh's avatar
Ian Lynagh committed
556
  | ITprimword   Integer
557 558
  | ITprimfloat  FractionalLit
  | ITprimdouble FractionalLit
559

560
  -- Template Haskell extension tokens
Ian Lynagh's avatar
Ian Lynagh committed
561 562 563 564 565 566 567 568
  | ITopenExpQuote              --  [| or [e|
  | ITopenPatQuote              --  [p|
  | ITopenDecQuote              --  [d|
  | ITopenTypQuote              --  [t|
  | ITcloseQuote                --  |]
  | ITidEscape   FastString     --  $x
  | ITparenEscape               --  $(
  | ITtyQuote                   --  ''
569 570 571 572 573 574 575 576
  | ITquasiQuote (FastString,FastString,RealSrcSpan)
    -- ITquasiQuote(quoter, quote, loc)
    -- represents a quasi-quote of the form
    -- [quoter| quote |]
  | ITqQuasiQuote (FastString,FastString,FastString,RealSrcSpan)
    -- ITqQuasiQuote(Qual, quoter, quote, loc)
    -- represents a qualified quasi-quote of the form
    -- [Qual.quoter| quote |]
577 578 579 580

  -- Arrow notation extension
  | ITproc
  | ITrec
Ian Lynagh's avatar
Ian Lynagh committed
581 582 583 584 585 586
  | IToparenbar                 --  (|
  | ITcparenbar                 --  |)
  | ITlarrowtail                --  -<
  | ITrarrowtail                --  >-
  | ITLarrowtail                --  -<<
  | ITRarrowtail                --  >>-
587

Ian Lynagh's avatar
Ian Lynagh committed
588 589
  | ITunknown String            -- Used when the lexer can't make sense of it
  | ITeof                       -- end of file token
590 591 592 593 594 595 596

  -- Documentation annotations
  | ITdocCommentNext  String     -- something beginning '-- |'
  | ITdocCommentPrev  String     -- something beginning '-- ^'
  | ITdocCommentNamed String     -- something beginning '-- $'
  | ITdocSection      Int String -- a section heading
  | ITdocOptions      String     -- doc options (prune, ignore-exports, etc)
David Waern's avatar
David Waern committed
597
  | ITdocOptionsOld   String     -- doc options declared "-- # ..."-style
Jedai's avatar
Jedai committed
598 599
  | ITlineComment     String     -- comment starting by "--"
  | ITblockComment    String     -- comment in {- -}
600

601
  deriving Show
602 603 604 605 606 607 608 609

-- 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)
--
Ian Lynagh's avatar
Ian Lynagh committed
610
reservedWordsFM :: UniqFM (Token, Int)
611
reservedWordsFM = listToUFM $
Ian Lynagh's avatar
Ian Lynagh committed
612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657
    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 explicitForallBit .|.
                                              bit inRulePragBit),
         ( "mdo",            ITmdo,           bit recursiveDoBit),
         ( "family",         ITfamily,        bit tyFamBit),
         ( "group",          ITgroup,         bit transformComprehensionsBit),
         ( "by",             ITby,            bit transformComprehensionsBit),
         ( "using",          ITusing,         bit transformComprehensionsBit),

         ( "foreign",        ITforeign,       bit ffiBit),
         ( "export",         ITexport,        bit ffiBit),
         ( "label",          ITlabel,         bit ffiBit),
         ( "dynamic",        ITdynamic,       bit ffiBit),
         ( "safe",           ITsafe,          bit ffiBit .|.
                                              bit safeHaskellBit),
         ( "interruptible",  ITinterruptible, bit interruptibleFfiBit),
         ( "unsafe",         ITunsafe,        bit ffiBit),
         ( "stdcall",        ITstdcallconv,   bit ffiBit),
         ( "ccall",          ITccallconv,     bit ffiBit),
658
         ( "capi",           ITcapiconv,      bit cApiFfiBit),
Ian Lynagh's avatar
Ian Lynagh committed
659 660 661 662
         ( "prim",           ITprimcallconv,  bit ffiBit),

         ( "rec",            ITrec,           bit recBit),
         ( "proc",           ITproc,          bit arrowsBit)
663 664
     ]

Ian Lynagh's avatar
Ian Lynagh committed
665
reservedSymsFM :: UniqFM (Token, Int -> Bool)
666
reservedSymsFM = listToUFM $
Ian Lynagh's avatar
Ian Lynagh committed
667 668 669 670 671 672 673 674 675 676 677 678
    map (\ (x,y,z) -> (mkFastString x,(y,z)))
      [ ("..",  ITdotdot,   always)
        -- (:) is a reserved op, meaning only list cons
       ,(":",   ITcolon,    always)
       ,("::",  ITdcolon,   always)
       ,("=",   ITequal,    always)
       ,("\\",  ITlam,      always)
       ,("|",   ITvbar,     always)
       ,("<-",  ITlarrow,   always)
       ,("->",  ITrarrow,   always)
       ,("@",   ITat,       always)
       ,("~",   ITtilde,    always)
batterseapower's avatar
batterseapower committed
679
       ,("~#",  ITtildehsh, always)
Ian Lynagh's avatar
Ian Lynagh committed
680 681 682 683 684
       ,("=>",  ITdarrow,   always)
       ,("-",   ITminus,    always)
       ,("!",   ITbang,     always)

        -- For data T (a::*) = MkT
685
       ,("*", ITstar, always) -- \i -> kindSigsEnabled i || tyFamEnabled i)
Ian Lynagh's avatar
Ian Lynagh committed
686
        -- For 'forall a . t'
687
       ,(".", ITdot,  always) -- \i -> explicitForallEnabled i || inRulePrag i)
Ian Lynagh's avatar
Ian Lynagh committed
688 689 690 691 692

       ,("-<",  ITlarrowtail, arrowsEnabled)
       ,(">-",  ITrarrowtail, arrowsEnabled)
       ,("-<<", ITLarrowtail, arrowsEnabled)
       ,(">>-", ITRarrowtail, arrowsEnabled)
693

Ian Lynagh's avatar
Ian Lynagh committed
694 695
       ,("∷",   ITdcolon, unicodeSyntaxEnabled)
       ,("⇒",   ITdarrow, unicodeSyntaxEnabled)
Ian Lynagh's avatar
Ian Lynagh committed
696 697
       ,("∀",   ITforall, \i -> unicodeSyntaxEnabled i &&
                                explicitForallEnabled i)
Ian Lynagh's avatar
Ian Lynagh committed
698 699
       ,("→",   ITrarrow, unicodeSyntaxEnabled)
       ,("←",   ITlarrow, unicodeSyntaxEnabled)
700 701 702 703 704 705 706 707

       ,("⤙",   ITlarrowtail, \i -> unicodeSyntaxEnabled i && arrowsEnabled i)
       ,("⤚",   ITrarrowtail, \i -> unicodeSyntaxEnabled i && arrowsEnabled i)
       ,("⤛",   ITLarrowtail, \i -> unicodeSyntaxEnabled i && arrowsEnabled i)
       ,("⤜",   ITRarrowtail, \i -> unicodeSyntaxEnabled i && arrowsEnabled i)

       ,("★", ITstar, unicodeSyntaxEnabled)

Simon Marlow's avatar
Simon Marlow committed
708 709 710
        -- ToDo: ideally, → and ∷ should be "specials", so that they cannot
        -- form part of a large operator.  This would let us have a better
        -- syntax for kinds: ɑ∷*→* would be a legal kind signature. (maybe).
711 712 713 714 715
       ]

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

Ian Lynagh's avatar
Ian Lynagh committed
716
type Action = RealSrcSpan -> StringBuffer -> Int -> P (RealLocated Token)
717

718
special :: Token -> Action
twanvl's avatar
twanvl committed
719
special tok span _buf _len = return (L span tok)
720

721
token, layout_token :: Token -> Action
twanvl's avatar
twanvl committed
722 723
token t span _buf _len = return (L span t)
layout_token t span _buf _len = pushLexState layout >> return (L span t)
724

725 726
idtoken :: (StringBuffer -> Int -> Token) -> Action
idtoken f span buf len = return (L span $! (f buf len))
727

728
skip_one_varid :: (FastString -> Token) -> Action
Ian Lynagh's avatar
Ian Lynagh committed
729
skip_one_varid f span buf len
730
  = return (L span $! f (lexemeToFastString (stepOn buf) (len-1)))
731

732
strtoken :: (String -> Token) -> Action
Ian Lynagh's avatar
Ian Lynagh committed
733
strtoken f span buf len =
734
  return (L span $! (f $! lexemeToString buf len))
735

736
init_strtoken :: Int -> (String -> Token) -> Action
737
-- like strtoken, but drops the last N character(s)
Ian Lynagh's avatar
Ian Lynagh committed
738
init_strtoken drop f span buf len =
739
  return (L span $! (f $! lexemeToString buf (len-drop)))
740 741

begin :: Int -> Action
742
begin code _span _str _len = do pushLexState code; lexToken
743 744

pop :: Action
Ian Lynagh's avatar
Ian Lynagh committed
745 746
pop _span _buf _len = do _ <- popLexState
                         lexToken
747

748 749 750 751 752 753 754 755 756 757 758
hopefully_open_brace :: Action
hopefully_open_brace span buf len
 = do relaxed <- extension relaxedLayout
      ctx <- getContext
      (AI l _) <- getInput
      let offset = srcLocCol l
          isOK = relaxed ||
                 case ctx of
                 Layout prev_off : _ -> prev_off < offset
                 _                   -> True
      if isOK then pop_and open_brace span buf len
Ian Lynagh's avatar
Ian Lynagh committed
759
              else failSpanMsgP (RealSrcSpan span) (text "Missing block")
760

761
pop_and :: Action -> Action
Ian Lynagh's avatar
Ian Lynagh committed
762 763
pop_and act span buf len = do _ <- popLexState
                              act span buf len
764

765
{-# INLINE nextCharIs #-}
Ian Lynagh's avatar
Ian Lynagh committed
766
nextCharIs :: StringBuffer -> (Char -> Bool) -> Bool
767 768
nextCharIs buf p = not (atEnd buf) && p (currentChar buf)

Ian Lynagh's avatar
Ian Lynagh committed
769
notFollowedBy :: Char -> AlexAccPred Int
Ian Lynagh's avatar
Ian Lynagh committed
770
notFollowedBy char _ _ _ (AI _ buf)
771
  = nextCharIs buf (/=char)
772

Ian Lynagh's avatar
Ian Lynagh committed
773
notFollowedBySymbol :: AlexAccPred Int
774
notFollowedBySymbol _ _ _ (AI _ buf)
775
  = nextCharIs buf (`notElem` "!#$%&*+./<=>?@\\^|-~")
ross's avatar
ross committed
776

777 778 779 780 781
-- We must reject doc comments as being ordinary comments everywhere.
-- In some cases the doc comment will be selected as the lexeme due to
-- maximal munch, but not always, because the nested comment rule is
-- valid in all states, but the doc-comment rules are only valid in
-- the non-layout states.
Ian Lynagh's avatar
Ian Lynagh committed
782
isNormalComment :: AlexAccPred Int
783
isNormalComment bits _ _ (AI _ buf)
784 785 786 787 788 789
  | haddockEnabled bits = notFollowedByDocOrPragma
  | otherwise           = nextCharIs buf (/='#')
  where
    notFollowedByDocOrPragma
       = not $ spaceAndP buf (`nextCharIs` (`elem` "|^*$#"))

Ian Lynagh's avatar
Ian Lynagh committed
790
spaceAndP :: StringBuffer -> (StringBuffer -> Bool) -> Bool
791
spaceAndP buf p = p buf || nextCharIs buf (==' ') && p (snd (nextChar buf))
792

Ian Lynagh's avatar
Ian Lynagh committed
793
atEOL :: AlexAccPred Int
794
atEOL _ _ _ (AI _ buf) = atEnd buf || currentChar buf == '\n'
795

Ian Lynagh's avatar
Ian Lynagh committed
796
ifExtension :: (Int -> Bool) -> AlexAccPred Int
797 798
ifExtension pred bits _ _ _ = pred bits

799 800 801
multiline_doc_comment :: Action
multiline_doc_comment span buf _len = withLexedDocType (worker "")
  where
802
    worker commentAcc input docType oneLine = case alexGetChar' input of
Ian Lynagh's avatar
Ian Lynagh committed
803
      Just ('\n', input')
804 805 806 807 808 809
        | oneLine -> docCommentEnd input commentAcc docType buf span
        | otherwise -> case checkIfCommentLine input' of
          Just input -> worker ('\n':commentAcc) input docType False
          Nothing -> docCommentEnd input commentAcc docType buf span
      Just (c, input) -> worker (c:commentAcc) input docType oneLine
      Nothing -> docCommentEnd input commentAcc docType buf span
Ian Lynagh's avatar
Ian Lynagh committed
810

811 812
    checkIfCommentLine input = check (dropNonNewlineSpace input)
      where
813 814 815
        check input = case alexGetChar' input of
          Just ('-', input) -> case alexGetChar' input of
            Just ('-', input) -> case alexGetChar' input of
816 817 818 819 820
              Just (c, _) | c /= '-' -> Just input
              _ -> Nothing
            _ -> Nothing
          _ -> Nothing

821
        dropNonNewlineSpace input = case alexGetChar' input of
Ian Lynagh's avatar
Ian Lynagh committed
822
          Just (c, input')
823 824 825 826
            | isSpace c && c /= '\n' -> dropNonNewlineSpace input'
            | otherwise -> input
          Nothing -> input

Jedai's avatar
Jedai committed
827 828 829 830 831
lineCommentToken :: Action
lineCommentToken span buf len = do
  b <- extension rawTokenStreamEnabled
  if b then strtoken ITlineComment span buf len else lexToken

832 833 834 835
{-
  nested comments require traversing by hand, they can't be parsed
  using regular expressions.
-}
Ian Lynagh's avatar
Ian Lynagh committed
836
nested_comment :: P (RealLocated Token) -> Action
837
nested_comment cont span _str _len = do
838
  input <- getInput
Jedai's avatar
Jedai committed
839
  go "" (1::Int) input
840
  where
Jedai's avatar
Jedai committed
841 842 843 844 845
    go commentAcc 0 input = do setInput input
                               b <- extension rawTokenStreamEnabled
                               if b
                                 then docCommentEnd input commentAcc ITblockComment _str span
                                 else cont
846
    go commentAcc n input = case alexGetChar' input of
847
      Nothing -> errBrace input span
848
      Just ('-',input) -> case alexGetChar' input of
849
        Nothing  -> errBrace input span
Jedai's avatar
Jedai committed
850 851
        Just ('\125',input) -> go commentAcc (n-1) input
        Just (_,_)          -> go ('-':commentAcc) n input
852
      Just ('\123',input) -> case alexGetChar' input of
853
        Nothing  -> errBrace input span
Jedai's avatar
Jedai committed
854 855 856
        Just ('-',input) -> go ('-':'\123':commentAcc) (n+1) input
        Just (_,_)       -> go ('\123':commentAcc) n input
      Just (c,input) -> go (c:commentAcc) n input
857 858 859 860

nested_doc_comment :: Action
nested_doc_comment span buf _len = withLexedDocType (go "")
  where
861
    go commentAcc input docType _ = case alexGetChar' input of
862
      Nothing -> errBrace input span
863
      Just ('-',input) -> case alexGetChar' input of
864
        Nothing -> errBrace input span
twanvl's avatar
twanvl committed
865
        Just ('\125',input) ->
866
          docCommentEnd input commentAcc docType buf span
twanvl's avatar
twanvl committed
867
        Just (_,_) -> go ('-':commentAcc) input docType False
868
      Just ('\123', input) -> case alexGetChar' input of
869 870 871 872 873
        Nothing  -> errBrace input span
        Just ('-',input) -> do
          setInput input
          let cont = do input <- getInput; go commentAcc input docType False
          nested_comment cont span buf _len
twanvl's avatar
twanvl committed
874
        Just (_,_) -> go ('\123':commentAcc) input docType False
875 876
      Just (c,input) -> go (c:commentAcc) input docType False

Ian Lynagh's avatar
Ian Lynagh committed
877 878
withLexedDocType :: (AlexInput -> (String -> Token) -> Bool -> P (RealLocated Token))
                 -> P (RealLocated Token)
879
withLexedDocType lexDocComment = do
880
  input@(AI _ buf) <- getInput
881 882 883 884
  case prevChar buf ' ' of
    '|' -> lexDocComment input ITdocCommentNext False
    '^' -> lexDocComment input ITdocCommentPrev False
    '$' -> lexDocComment input ITdocCommentNamed False
David Waern's avatar
David Waern committed
885 886
    '*' -> lexDocSection 1 input
    '#' -> lexDocComment input ITdocOptionsOld False
Ian Lynagh's avatar
Ian Lynagh committed
887
    _ -> panic "withLexedDocType: Bad doc type"
Ian Lynagh's avatar
Ian Lynagh committed
888
 where
889
    lexDocSection n input = case alexGetChar' input of
890
      Just ('*', input) -> lexDocSection (n+1) input
twanvl's avatar
twanvl committed
891
      Just (_,   _)     -> lexDocComment input (ITdocSection n) True
892 893
      Nothing -> do setInput input; lexToken -- eof reached, lex it normally

894 895 896
-- RULES pragmas turn on the forall and '.' keywords, and we turn them
-- off again at the end of the pragma.
rulePrag :: Action
897
rulePrag span _buf _len = do
898
  setExts (.|. bit inRulePragBit)
899 900 901
  return (L span ITrules_prag)

endPrag :: Action
902
endPrag span _buf _len = do
903 904 905
  setExts (.&. complement (bit inRulePragBit))
  return (L span ITclose_prag)

906 907 908 909
-- docCommentEnd
-------------------------------------------------------------------------------
-- This function is quite tricky. We can't just return a new token, we also
-- need to update the state of the parser. Why? Because the token is longer
Ian Lynagh's avatar
Ian Lynagh committed
910
-- than what was lexed by Alex, and the lexToken function doesn't know this, so
911
-- it writes the wrong token length to the parser state. This function is
Ian Lynagh's avatar
Ian Lynagh committed
912
-- called afterwards, so it can just update the state.
913 914

docCommentEnd :: AlexInput -> String -> (String -> Token) -> StringBuffer ->
Ian Lynagh's avatar
Ian Lynagh committed
915
                 RealSrcSpan -> P (RealLocated Token)
916 917
docCommentEnd input commentAcc docType buf span = do
  setInput input
918
  let (AI loc nextBuf) = input
919
      comment = reverse commentAcc
Ian Lynagh's avatar
Ian Lynagh committed
920
      span' = mkRealSrcSpan (realSrcSpanStart span) loc
921
      last_len = byteDiff buf nextBuf
Ian Lynagh's avatar
Ian Lynagh committed
922

923
  span `seq` setLastToken span' last_len
924
  return (L span' (docType comment))
Ian Lynagh's avatar
Ian Lynagh committed
925

Ian Lynagh's avatar
Ian Lynagh committed
926 927
errBrace :: AlexInput -> RealSrcSpan -> P a
errBrace (AI end _) span = failLocMsgP (realSrcSpanStart span) end "unterminated `{-'"
Ian Lynagh's avatar
Ian Lynagh committed
928

929
open_brace, close_brace :: Action
Ian Lynagh's avatar
Ian Lynagh committed
930
open_brace span _str _len = do
931 932
  ctx <- getContext
  setContext (NoLayout:ctx)
933
  return (L span ITocurly)
Ian Lynagh's avatar
Ian Lynagh committed
934
close_brace span _str _len = do
935
  popContext
936
  return (L span ITccurly)
937

Ian Lynagh's avatar
Ian Lynagh committed
938
qvarid, qconid :: StringBuffer -> Int -> Token
939 940
qvarid buf len = ITqvarid $! splitQualName buf len False
qconid buf len = ITqconid $! splitQualName buf len False
941

942
splitQualName :: StringBuffer -> Int -> Bool -> (FastString,FastString)
943 944 945
-- 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.
946
splitQualName orig_buf len parens = split orig_buf orig_buf
947
  where
948
    split buf dot_buf
Ian Lynagh's avatar
Ian Lynagh committed
949 950 951
        | orig_buf `byteDiff` buf >= len  = done dot_buf
        | c == '.'                        = found_dot buf'
        | otherwise                       = split buf' dot_buf
952 953
      where
       (c,buf') = nextChar buf
Ian Lynagh's avatar
Ian Lynagh committed
954

955 956 957
    -- 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.
958
    found_dot buf -- buf points after the '.'
Ian Lynagh's avatar
Ian Lynagh committed
959 960
        | isUpper c    = split buf' buf
        | otherwise    = done buf
961 962 963 964
      where
       (c,buf') = nextChar buf

    done dot_buf =
Ian Lynagh's avatar
Ian Lynagh committed
965 966
        (lexemeToFastString orig_buf (qual_size - 1),
         if parens -- Prelude.(+)
967 968
            then lexemeToFastString (stepOn dot_buf) (len - qual_size - 2)
            else lexemeToFastString dot_buf (len - qual_size))
969
      where
Ian Lynagh's avatar
Ian Lynagh committed
970
        qual_size = orig_buf `byteDiff` dot_buf
971

Ian Lynagh's avatar
Ian Lynagh committed
972 973
varid :: Action
varid span buf len =
Ian Lynagh's avatar
Ian Lynagh committed
974
  fs `seq`
975
  case lookupUFM reservedWordsFM fs of
Ian Lynagh's avatar
Ian Lynagh committed
976 977 978 979 980 981 982 983 984
        Just (keyword,0)    -> do
                maybe_layout keyword
                return (L span keyword)
        Just (keyword,exts) -> do
                b <- extension (\i -> exts .&. i /= 0)
                if b then do maybe_layout keyword
                             return (L span keyword)
                     else return (L span (ITvarid fs))
        _other -> return (L span (ITvarid fs))
985
  where
Ian Lynagh's avatar
Ian Lynagh committed
986
        fs = lexemeToFastString buf len
987

Ian Lynagh's avatar
Ian Lynagh committed
988
conid :: StringBuffer -> Int -> Token
989 990 991
conid buf len = ITconid fs
  where fs = lexemeToFastString buf len

Ian Lynagh's avatar
Ian Lynagh committed
992
qvarsym, qconsym, prefixqvarsym, prefixqconsym :: StringBuffer -> Int -> Token
993 994 995 996
qvarsym buf len = ITqvarsym $! splitQualName buf len False
qconsym buf len = ITqconsym $! splitQualName buf len False
prefixqvarsym buf len = ITprefixqvarsym $! splitQualName buf len True
prefixqconsym buf len = ITprefixqconsym $! splitQualName buf len True
997

Ian Lynagh's avatar
Ian Lynagh committed
998
varsym, consym :: Action
999 1000 1001
varsym = sym ITvarsym
consym = sym ITconsym

Ian Lynagh's avatar
Ian Lynagh committed
1002 1003
sym :: (FastString -> Token) -> RealSrcSpan -> StringBuffer -> Int
    -> P (RealLocated Token)
Ian Lynagh's avatar
Ian Lynagh committed
1004
sym con span buf len =
1005
  case lookupUFM reservedSymsFM fs of
Ian Lynagh's avatar
Ian Lynagh committed
1006 1007 1008 1009 1010
        Just (keyword,exts) -> do
                b <- extension exts
                if b then return (L span keyword)
                     else return (L span $! con fs)
        _other -> return (L span $! con fs)