Lexer.x 119 KB
Newer Older
1
-----------------------------------------------------------------------------
2
-- (c) The University of Glasgow, 2006
3
--
4
-- GHC's lexer for Haskell 2010 [1].
5
--
6 7
-- This is a combination of an Alex-generated lexer [2] from a regex
-- definition, with some hand-coded bits. [3]
8 9
--
-- 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
-- References:
-- [1] https://www.haskell.org/onlinereport/haskell2010/haskellch2.html
-- [2] http://www.haskell.org/alex/
-- [3] https://ghc.haskell.org/trac/ghc/wiki/Commentary/Compiler/Parser
--
18 19 20 21 22 23 24 25 26 27
-----------------------------------------------------------------------------

--   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

28
--   qualified operator NOTES.
Ian Lynagh's avatar
Ian Lynagh committed
29
--
30 31 32 33 34 35 36 37 38
--   - 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.

39 40 41 42

-- -----------------------------------------------------------------------------
-- Alex "Haskell code fragment top"

43
{
44
{-# LANGUAGE BangPatterns #-}
45
{-# LANGUAGE LambdaCase #-}
46

47 48
{-# OPTIONS_GHC -funbox-strict-fields #-}

49
module Lexer (
50
   Token(..), lexer, pragState, mkPState, mkPStatePure, PState(..),
51 52
   P(..), ParseResult(..), mkParserFlags, mkParserFlags', ParserFlags,
   getRealSrcLoc, getPState, withThisPackage,
53
   failLocMsgP, failSpanMsgP, srcParseFail,
Ian Lynagh's avatar
Ian Lynagh committed
54
   getMessages,
55
   popContext, pushModuleContext, setLastToken, setSrcLoc,
vivian's avatar
vivian committed
56
   activeContext, nextIsEOF,
simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
57
   getLexState, popLexState, pushLexState,
58
   ExtBits(..), getBit,
Jedai's avatar
Jedai committed
59
   addWarning,
Alan Zimmerman's avatar
Alan Zimmerman committed
60
   lexTokenStream,
61
   addAnnotation,AddAnn,addAnnsAt,mkParensApiAnn,
62
   commentToAnnotation
63 64
  ) where

65 66
import GhcPrelude

rodlogic's avatar
rodlogic committed
67 68
-- base
import Control.Monad
69
import Control.Monad.Fail as MonadFail
rodlogic's avatar
rodlogic committed
70
import Data.Bits
71
import Data.Char
rodlogic's avatar
rodlogic committed
72 73 74 75
import Data.List
import Data.Maybe
import Data.Word

76 77
import EnumSet (EnumSet)
import qualified EnumSet
78

79 80 81
-- ghc-boot
import qualified GHC.LanguageExtensions as LangExt

rodlogic's avatar
rodlogic committed
82 83 84 85 86 87 88 89
-- bytestring
import Data.ByteString (ByteString)

-- containers
import Data.Map (Map)
import qualified Data.Map as Map

-- compiler/utils
90
import Bag
91 92 93 94
import Outputable
import StringBuffer
import FastString
import UniqFM
95
import Util             ( readRational, readHexRational )
rodlogic's avatar
rodlogic committed
96 97 98

-- compiler/main
import ErrUtils
99
import DynFlags
rodlogic's avatar
rodlogic committed
100 101 102

-- compiler/basicTypes
import SrcLoc
103
import Module
104 105
import BasicTypes     ( InlineSpec(..), RuleMatchInfo(..),
                        IntegralLit(..), FractionalLit(..),
Alan Zimmerman's avatar
Alan Zimmerman committed
106
                        SourceText(..) )
107

rodlogic's avatar
rodlogic committed
108 109
-- compiler/parser
import Ctype
Alan Zimmerman's avatar
Alan Zimmerman committed
110 111

import ApiAnnotation
112 113
}

114 115 116
-- -----------------------------------------------------------------------------
-- Alex "Character set macros"

117 118
-- NB: The logic behind these definitions is also reflected in basicTypes/Lexeme.hs
-- Any changes here should likely be reflected there.
119
$unispace    = \x05 -- Trick Alex into handling Unicode. See [Unicode in Alex].
120 121 122
$nl          = [\n\r\f]
$whitechar   = [$nl\v\ $unispace]
$white_no_nl = $whitechar # \n -- TODO #8424
123
$tab         = \t
124 125

$ascdigit  = 0-9
126
$unidigit  = \x03 -- Trick Alex into handling Unicode. See [Unicode in Alex].
127
$decdigit  = $ascdigit -- for now, should really be $digit (ToDo)
128 129 130
$digit     = [$ascdigit $unidigit]

$special   = [\(\)\,\;\[\]\`\{\}]
131
$ascsymbol = [\!\#\$\%\&\*\+\.\/\<\=\>\?\@\\\^\|\-\~\:]
132
$unisymbol = \x04 -- Trick Alex into handling Unicode. See [Unicode in Alex].
133
$symbol    = [$ascsymbol $unisymbol] # [$special \_\"\']
134

135
$unilarge  = \x01 -- Trick Alex into handling Unicode. See [Unicode in Alex].
136
$asclarge  = [A-Z]
137 138
$large     = [$asclarge $unilarge]

139
$unismall  = \x02 -- Trick Alex into handling Unicode. See [Unicode in Alex].
140
$ascsmall  = [a-z]
141 142
$small     = [$ascsmall $unismall \_]

143
$unigraphic = \x06 -- Trick Alex into handling Unicode. See [Unicode in Alex].
144
$graphic   = [$small $large $symbol $digit $special $unigraphic \"\']
145

146
$binit     = 0-1
Ian Lynagh's avatar
Ian Lynagh committed
147
$octit     = 0-7
148
$hexit     = [$decdigit A-F a-f]
149

150
$uniidchar = \x07 -- Trick Alex into handling Unicode. See [Unicode in Alex].
151
$idchar    = [$small $large $digit $uniidchar \']
152

153 154
$pragmachar = [$small $large $digit]

155 156
$docsym    = [\| \^ \* \$]

157

158 159 160 161 162 163 164 165
-- -----------------------------------------------------------------------------
-- Alex "Regular expression macros"

@varid     = $small $idchar*          -- variable identifiers
@conid     = $large $idchar*          -- constructor identifiers

@varsym    = ($symbol # \:) $symbol*  -- variable (operator) symbol
@consym    = \: $symbol*              -- constructor (operator) symbol
166

167 168 169 170 171 172 173 174
-- See Note [Lexing NumericUnderscores extension] and #14473
@numspc       = _*                   -- numeric spacer (#14473)
@decimal      = $decdigit(@numspc $decdigit)*
@binary       = $binit(@numspc $binit)*
@octal        = $octit(@numspc $octit)*
@hexadecimal  = $hexit(@numspc $hexit)*
@exponent     = @numspc [eE] [\-\+]? @decimal
@bin_exponent = @numspc [pP] [\-\+]? @decimal
175 176

@qual = (@conid \.)+
177 178 179 180
@qvarid = @qual @varid
@qconid = @qual @conid
@qvarsym = @qual @varsym
@qconsym = @qual @consym
181

182 183
@floating_point = @numspc @decimal \. @decimal @exponent? | @numspc @decimal @exponent
@hex_floating_point = @numspc @hexadecimal \. @hexadecimal @bin_exponent? | @numspc @hexadecimal @bin_exponent
184

185 186 187 188 189
-- normal signed numerical literals can only be explicitly negative,
-- not explicitly positive (contrast @exponent)
@negative = \-
@signed = @negative ?

190 191 192 193

-- -----------------------------------------------------------------------------
-- Alex "Identifier"

194 195
haskell :-

196 197 198 199 200

-- -----------------------------------------------------------------------------
-- Alex "Rules"

-- everywhere: skip whitespace
Ian Lynagh's avatar
Ian Lynagh committed
201
$white_no_nl+ ;
202
$tab          { warnTab }
203 204 205 206 207

-- 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
208 209 210 211
-- are). We also rule out nested Haddock comments, if the -haddock flag is
-- set.

"{-" / { isNormalComment } { nested_comment lexToken }
212 213 214 215

-- 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.
216 217

-- Since Haddock comments aren't valid in every state, we need to rule them
Ian Lynagh's avatar
Ian Lynagh committed
218
-- out here.
219 220 221 222 223 224 225

-- 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.

226
"-- " ~$docsym .* { lineCommentToken }
227
"--" [^$symbol \ ] .* { lineCommentToken }
228 229 230

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

231
"-- " $docsym .* / { alexNotPred (ifExtension HaddockBit) } { lineCommentToken }
232 233 234 235 236 237 238

-- 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.

239
"---"\-* ~$symbol .* { lineCommentToken }
240 241 242 243

-- 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
244
"--"\-* / { atEOL } { lineCommentToken }
245 246 247 248

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

Jedai's avatar
Jedai committed
249
"-- " / { atEOL } { lineCommentToken }
250 251 252 253 254 255 256 257 258 259 260

-- '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
261
  \n                                    ;
Adam Gundry's avatar
Adam Gundry committed
262 263
  ^\# line                              { begin line_prag1 }
  ^\# / { followedByDigit }             { begin line_prag1 }
Ian Lynagh's avatar
Ian Lynagh committed
264 265 266
  ^\# pragma .* \n                      ; -- GCC 3.3 CPP generated, apparently
  ^\# \! .* \n                          ; -- #!, for scripts
  ()                                    { do_bol }
267 268 269 270 271
}

-- 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.
272
<layout, layout_do, layout_if> {
Ian Lynagh's avatar
Ian Lynagh committed
273 274 275 276
  \{ / { notFollowedBy '-' }            { hopefully_open_brace }
        -- we might encounter {-# here, but {- has been handled already
  \n                                    ;
  ^\# (line)?                           { begin line_prag1 }
277 278
}

279 280
-- after an 'if', a vertical bar starts a layout context for MultiWayIf
<layout_if> {
281
  \| / { notFollowedBySymbol }          { new_layout_context True dontGenerateSemic ITvbar }
282 283 284
  ()                                    { pop }
}

285
-- do is treated in a subtly different way, see new_layout_context
286 287
<layout>    ()                          { new_layout_context True  generateSemic ITvocurly }
<layout_do> ()                          { new_layout_context False generateSemic ITvocurly }
288 289 290 291

-- 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
292
<layout_left>  ()                       { do_layout_left }
293

Ian Lynagh's avatar
Ian Lynagh committed
294
<0,option_prags> \n                     { begin bol }
295

296 297
"{-#" $whitechar* $pragmachar+ / { known_pragma linePrags }
                                { dispatch_pragmas linePrags }
298 299 300

-- single-line line pragmas, of the form
--    # <line> "<file>" <extra-stuff> \n
301 302 303 304 305
<line_prag1> {
  @decimal $white_no_nl+ \" [$graphic \ ]* \"  { setLineAndFile line_prag1a }
  ()                                           { failLinePrag1 }
}
<line_prag1a> .*                               { popLinePrag1 }
306 307 308

-- Haskell-style line pragmas, of the form
--    {-# LINE <line> "<file>" #-}
309 310 311 312
<line_prag2> {
  @decimal $white_no_nl+ \" [$graphic \ ]* \"  { setLineAndFile line_prag2a }
}
<line_prag2a> "#-}"|"-}"                       { pop }
313 314
   -- NOTE: accept -} at the end of a LINE pragma, for compatibility
   -- with older versions of GHC which generated these.
315

Rufflewind's avatar
Rufflewind committed
316 317 318 319
-- Haskell-style column pragmas, of the form
--    {-# COLUMN <column> #-}
<column_prag> @decimal $whitechar* "#-}" { setColumn }

320
<0,option_prags> {
Ian Lynagh's avatar
Ian Lynagh committed
321
  "{-#" $whitechar* $pragmachar+
322 323 324 325 326
        $whitechar+ $pragmachar+ / { known_pragma twoWordPrags }
                                 { dispatch_pragmas twoWordPrags }

  "{-#" $whitechar* $pragmachar+ / { known_pragma oneWordPrags }
                                 { dispatch_pragmas oneWordPrags }
327

328
  -- We ignore all these pragmas, but don't generate a warning for them
329 330
  "{-#" $whitechar* $pragmachar+ / { known_pragma ignoredPrags }
                                 { dispatch_pragmas ignoredPrags }
331 332

  -- ToDo: should only be valid inside a pragma:
Ian Lynagh's avatar
Ian Lynagh committed
333
  "#-}"                          { endPrag }
334 335
}

336
<option_prags> {
337 338
  "{-#"  $whitechar* $pragmachar+ / { known_pragma fileHeaderPrags }
                                   { dispatch_pragmas fileHeaderPrags }
David Waern's avatar
David Waern committed
339 340
}

341 342
<0> {
  -- In the "0" mode we ignore these pragmas
343
  "{-#"  $whitechar* $pragmachar+ / { known_pragma fileHeaderPrags }
344 345 346
                     { nested_comment lexToken }
}

347
<0,option_prags> {
348 349
  "{-#"  { warnThen Opt_WarnUnrecognisedPragmas (text "Unrecognised pragma")
                    (nested_comment lexToken) }
350 351
}

352 353
-- '0' state: ordinary lexemes

354 355
-- Haddock comments

356
<0,option_prags> {
357 358
  "-- " $docsym      / { ifExtension HaddockBit } { multiline_doc_comment }
  "{-" \ ? $docsym   / { ifExtension HaddockBit } { nested_doc_comment }
359 360
}

361 362
-- "special" symbols

363
<0> {
364 365 366 367 368 369 370 371 372 373 374 375 376 377 378
  "[|"        / { ifExtension ThQuotesBit } { token (ITopenExpQuote NoE NormalSyntax) }
  "[||"       / { ifExtension ThQuotesBit } { token (ITopenTExpQuote NoE) }
  "[e|"       / { ifExtension ThQuotesBit } { token (ITopenExpQuote HasE NormalSyntax) }
  "[e||"      / { ifExtension ThQuotesBit } { token (ITopenTExpQuote HasE) }
  "[p|"       / { ifExtension ThQuotesBit } { token ITopenPatQuote }
  "[d|"       / { ifExtension ThQuotesBit } { layout_token ITopenDecQuote }
  "[t|"       / { ifExtension ThQuotesBit } { token ITopenTypQuote }
  "|]"        / { ifExtension ThQuotesBit } { token (ITcloseQuote NormalSyntax) }
  "||]"       / { ifExtension ThQuotesBit } { token ITcloseTExpQuote }
  \$ @varid   / { ifExtension ThBit }       { skip_one_varid ITidEscape }
  "$$" @varid / { ifExtension ThBit }       { skip_two_varid ITidTyEscape }
  "$("        / { ifExtension ThBit }       { token ITparenEscape }
  "$$("       / { ifExtension ThBit }       { token ITparenTyEscape }

  "[" @varid "|"  / { ifExtension QqBit }   { lex_quasiquote_tok }
379 380

  -- qualified quasi-quote (#5555)
381
  "[" @qvarid "|"  / { ifExtension QqBit }  { lex_qquasiquote_tok }
382 383 384

  $unigraphic -- ⟦
    / { ifCurrentChar '⟦' `alexAndPred`
385 386
        ifExtension UnicodeSyntaxBit `alexAndPred`
        ifExtension ThQuotesBit }
387 388 389
    { token (ITopenExpQuote NoE UnicodeSyntax) }
  $unigraphic -- ⟧
    / { ifCurrentChar '⟧' `alexAndPred`
390 391
        ifExtension UnicodeSyntaxBit `alexAndPred`
        ifExtension ThQuotesBit }
392
    { token (ITcloseQuote UnicodeSyntax) }
393 394
}

eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
395 396 397 398
  -- See Note [Lexing type applications]
<0> {
    [^ $idchar \) ] ^
  "@"
399
    / { ifExtension TypeApplicationsBit `alexAndPred` notFollowedBySymbol }
eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
400 401 402
    { token ITtypeApp }
}

403
<0> {
404 405 406 407 408 409 410
  "(|"
    / { ifExtension ArrowsBit `alexAndPred`
        notFollowedBySymbol }
    { special (IToparenbar NormalSyntax) }
  "|)"
    / { ifExtension ArrowsBit }
    { special (ITcparenbar NormalSyntax) }
411 412 413

  $unigraphic -- ⦇
    / { ifCurrentChar '⦇' `alexAndPred`
414 415
        ifExtension UnicodeSyntaxBit `alexAndPred`
        ifExtension ArrowsBit }
416 417 418
    { special (IToparenbar UnicodeSyntax) }
  $unigraphic -- ⦈
    / { ifCurrentChar '⦈' `alexAndPred`
419 420
        ifExtension UnicodeSyntaxBit `alexAndPred`
        ifExtension ArrowsBit }
421
    { special (ITcparenbar UnicodeSyntax) }
422 423
}

424
<0> {
425
  \? @varid / { ifExtension IpBit } { skip_one_varid ITdupipvarid }
426 427
}

Adam Gundry's avatar
Adam Gundry committed
428
<0> {
429
  "#" @varid / { ifExtension OverloadedLabelsBit } { skip_one_varid ITlabelvarid }
Adam Gundry's avatar
Adam Gundry committed
430 431
}

432
<0> {
433 434
  "(#" / { ifExtension UnboxedTuplesBit `alexOrPred`
           ifExtension UnboxedSumsBit }
Ian Lynagh's avatar
Ian Lynagh committed
435
         { token IToubxparen }
436 437
  "#)" / { ifExtension UnboxedTuplesBit `alexOrPred`
           ifExtension UnboxedSumsBit }
Ian Lynagh's avatar
Ian Lynagh committed
438 439 440
         { token ITcubxparen }
}

441
<0,option_prags> {
Ian Lynagh's avatar
Ian Lynagh committed
442 443 444 445 446 447 448 449 450 451
  \(                                    { special IToparen }
  \)                                    { special ITcparen }
  \[                                    { special ITobrack }
  \]                                    { special ITcbrack }
  \,                                    { special ITcomma }
  \;                                    { special ITsemi }
  \`                                    { special ITbackquote }

  \{                                    { open_brace }
  \}                                    { close_brace }
452 453
}

454
<0,option_prags> {
455 456
  @qvarid                       { idtoken qvarid }
  @qconid                       { idtoken qconid }
Ian Lynagh's avatar
Ian Lynagh committed
457
  @varid                        { varid }
458
  @conid                        { idtoken conid }
459 460
}

461
<0> {
462 463 464 465
  @qvarid "#"+      / { ifExtension MagicHashBit } { idtoken qvarid }
  @qconid "#"+      / { ifExtension MagicHashBit } { idtoken qconid }
  @varid "#"+       / { ifExtension MagicHashBit } { varid }
  @conid "#"+       / { ifExtension MagicHashBit } { idtoken conid }
466 467
}

468 469
-- ToDo: - move `var` and (sym) into lexical syntax?
--       - remove backquote from $special?
470
<0> {
471 472
  @qvarsym                                         { idtoken qvarsym }
  @qconsym                                         { idtoken qconsym }
473 474
  @varsym                                          { varsym }
  @consym                                          { consym }
475 476
}

477 478
-- For the normal boxed literals we need to be careful
-- when trying to be close to Haskell98
479 480 481 482 483 484 485 486 487 488

-- Note [Lexing NumericUnderscores extension] (#14473)
--
-- NumericUnderscores extension allows underscores in numeric literals.
-- Multiple underscores are represented with @numspc macro.
-- To be simpler, we have only the definitions with underscores.
-- And then we have a separate function (tok_integral and tok_frac)
-- that validates the literals.
-- If extensions are not enabled, check that there are no underscores.
--
489
<0> {
490
  -- Normal integral literals (:: Num a => a, from Integer)
491 492 493 494 495 496 497 498 499
  @decimal                                                                   { tok_num positive 0 0 decimal }
  0[bB] @numspc @binary                / { ifExtension BinaryLiteralsBit }   { tok_num positive 2 2 binary }
  0[oO] @numspc @octal                                                       { tok_num positive 2 2 octal }
  0[xX] @numspc @hexadecimal                                                 { tok_num positive 2 2 hexadecimal }
  @negative @decimal                   / { ifExtension NegativeLiteralsBit } { tok_num negative 1 1 decimal }
  @negative 0[bB] @numspc @binary      / { ifExtension NegativeLiteralsBit `alexAndPred`
                                           ifExtension BinaryLiteralsBit }   { tok_num negative 3 3 binary }
  @negative 0[oO] @numspc @octal       / { ifExtension NegativeLiteralsBit } { tok_num negative 3 3 octal }
  @negative 0[xX] @numspc @hexadecimal / { ifExtension NegativeLiteralsBit } { tok_num negative 3 3 hexadecimal }
500 501

  -- Normal rational literals (:: Fractional a => a, from Rational)
502 503 504 505 506 507
  @floating_point                                                            { tok_frac 0 tok_float }
  @negative @floating_point            / { ifExtension NegativeLiteralsBit } { tok_frac 0 tok_float }
  0[xX] @numspc @hex_floating_point    / { ifExtension HexFloatLiteralsBit } { tok_frac 0 tok_hex_float }
  @negative 0[xX] @numspc @hex_floating_point
                                       / { ifExtension HexFloatLiteralsBit `alexAndPred`
                                           ifExtension NegativeLiteralsBit } { tok_frac 0 tok_hex_float }
508 509
}

510
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
511
  -- Unboxed ints (:: Int#) and words (:: Word#)
512 513
  -- It's simpler (and faster?) to give separate cases to the negatives,
  -- especially considering octal/hexadecimal prefixes.
514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530
  @decimal                          \# / { ifExtension MagicHashBit }        { tok_primint positive 0 1 decimal }
  0[bB] @numspc @binary             \# / { ifExtension MagicHashBit `alexAndPred`
                                           ifExtension BinaryLiteralsBit }   { tok_primint positive 2 3 binary }
  0[oO] @numspc @octal              \# / { ifExtension MagicHashBit }        { tok_primint positive 2 3 octal }
  0[xX] @numspc @hexadecimal        \# / { ifExtension MagicHashBit }        { tok_primint positive 2 3 hexadecimal }
  @negative @decimal                \# / { ifExtension MagicHashBit }        { tok_primint negative 1 2 decimal }
  @negative 0[bB] @numspc @binary   \# / { ifExtension MagicHashBit `alexAndPred`
                                           ifExtension BinaryLiteralsBit }   { tok_primint negative 3 4 binary }
  @negative 0[oO] @numspc @octal    \# / { ifExtension MagicHashBit }        { tok_primint negative 3 4 octal }
  @negative 0[xX] @numspc @hexadecimal \#
                                       / { ifExtension MagicHashBit }        { tok_primint negative 3 4 hexadecimal }

  @decimal                       \# \# / { ifExtension MagicHashBit }        { tok_primword 0 2 decimal }
  0[bB] @numspc @binary          \# \# / { ifExtension MagicHashBit `alexAndPred`
                                           ifExtension BinaryLiteralsBit }   { tok_primword 2 4 binary }
  0[oO] @numspc @octal           \# \# / { ifExtension MagicHashBit }        { tok_primword 2 4 octal }
  0[xX] @numspc @hexadecimal     \# \# / { ifExtension MagicHashBit }        { tok_primword 2 4 hexadecimal }
Ian Lynagh's avatar
Ian Lynagh committed
531

532 533
  -- Unboxed floats and doubles (:: Float#, :: Double#)
  -- prim_{float,double} work with signed literals
534 535
  @signed @floating_point           \# / { ifExtension MagicHashBit }        { tok_frac 1 tok_primfloat }
  @signed @floating_point        \# \# / { ifExtension MagicHashBit }        { tok_frac 2 tok_primdouble }
536 537 538 539 540 541
}

-- 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.
542
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
543 544
  \'                            { lex_char_tok }
  \"                            { lex_string_tok }
545 546
}

eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
547 548 549 550 551 552 553 554 555 556 557
-- Note [Lexing type applications]
-- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-- The desired syntax for type applications is to prefix the type application
-- with '@', like this:
--
--   foo @Int @Bool baz bum
--
-- This, of course, conflicts with as-patterns. The conflict arises because
-- expressions and patterns use the same parser, and also because we want
-- to allow type patterns within expression patterns.
--
Gabor Greif's avatar
Gabor Greif committed
558
-- Disambiguation is accomplished by requiring *something* to appear between
eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
559 560 561 562 563 564 565 566 567 568 569 570 571 572
-- type application and the preceding token. This something must end with
-- a character that cannot be the end of the variable bound in an as-pattern.
-- Currently (June 2015), this means that the something cannot end with a
-- $idchar or a close-paren. (The close-paren is necessary if the as-bound
-- identifier is symbolic.)
--
-- Note that looking for whitespace before the '@' is insufficient, because
-- of this pathological case:
--
--   foo {- hi -}@Int
--
-- This design is predicated on the fact that as-patterns are generally
-- whitespace-free, and also that this whole thing is opt-in, with the
-- TypeApplications extension.
573 574 575 576

-- -----------------------------------------------------------------------------
-- Alex "Haskell code fragment bottom"

577
{
578

579 580 581
-- -----------------------------------------------------------------------------
-- The token type

582
data Token
Ian Lynagh's avatar
Ian Lynagh committed
583
  = ITas                        -- Haskell keywords
584 585 586 587 588 589 590 591
  | ITcase
  | ITclass
  | ITdata
  | ITdefault
  | ITderiving
  | ITdo
  | ITelse
  | IThiding
592
  | ITforeign
593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608
  | ITif
  | ITimport
  | ITin
  | ITinfix
  | ITinfixl
  | ITinfixr
  | ITinstance
  | ITlet
  | ITmodule
  | ITnewtype
  | ITof
  | ITqualified
  | ITthen
  | ITtype
  | ITwhere

609
  | ITforall            IsUnicodeSyntax -- GHC extension keywords
610 611 612 613
  | ITexport
  | ITlabel
  | ITdynamic
  | ITsafe
614
  | ITinterruptible
615 616 617
  | ITunsafe
  | ITstdcallconv
  | ITccallconv
618
  | ITcapiconv
619
  | ITprimcallconv
thoughtpolice's avatar
thoughtpolice committed
620
  | ITjavascriptcallconv
621
  | ITmdo
622
  | ITfamily
623
  | ITrole
624 625 626
  | ITgroup
  | ITby
  | ITusing
cactus's avatar
cactus committed
627
  | ITpattern
Facundo Domínguez's avatar
Facundo Domínguez committed
628
  | ITstatic
Ryan Scott's avatar
Ryan Scott committed
629 630
  | ITstock
  | ITanyclass
Ryan Scott's avatar
Ryan Scott committed
631
  | ITvia
632

Edward Z. Yang's avatar
Edward Z. Yang committed
633 634 635 636 637 638
  -- Backpack tokens
  | ITunit
  | ITsignature
  | ITdependency
  | ITrequires

Alan Zimmerman's avatar
Alan Zimmerman committed
639 640 641 642 643 644 645 646
  -- Pragmas, see  note [Pragma source text] in BasicTypes
  | ITinline_prag       SourceText InlineSpec RuleMatchInfo
  | ITspec_prag         SourceText                -- SPECIALISE
  | ITspec_inline_prag  SourceText Bool    -- SPECIALISE INLINE (or NOINLINE)
  | ITsource_prag       SourceText
  | ITrules_prag        SourceText
  | ITwarning_prag      SourceText
  | ITdeprecated_prag   SourceText
647 648
  | ITline_prag         SourceText  -- not usually produced, see 'UsePosPragsBit'
  | ITcolumn_prag       SourceText  -- not usually produced, see 'UsePosPragsBit'
Alan Zimmerman's avatar
Alan Zimmerman committed
649 650 651 652 653 654
  | ITscc_prag          SourceText
  | ITgenerated_prag    SourceText
  | ITcore_prag         SourceText         -- hdaume: core annotations
  | ITunpack_prag       SourceText
  | ITnounpack_prag     SourceText
  | ITann_prag          SourceText
655
  | ITcomplete_prag     SourceText
656
  | ITclose_prag
657 658 659
  | IToptions_prag String
  | ITinclude_prag String
  | ITlanguage_prag
Alan Zimmerman's avatar
Alan Zimmerman committed
660 661 662 663 664 665
  | ITminimal_prag      SourceText
  | IToverlappable_prag SourceText  -- instance overlap mode
  | IToverlapping_prag  SourceText  -- instance overlap mode
  | IToverlaps_prag     SourceText  -- instance overlap mode
  | ITincoherent_prag   SourceText  -- instance overlap mode
  | ITctype             SourceText
666
  | ITcomment_line_prag         -- See Note [Nested comment line pragmas]
667

Ian Lynagh's avatar
Ian Lynagh committed
668
  | ITdotdot                    -- reserved symbols
669
  | ITcolon
670
  | ITdcolon            IsUnicodeSyntax
671 672
  | ITequal
  | ITlam
673
  | ITlcase
674
  | ITvbar
675 676
  | ITlarrow            IsUnicodeSyntax
  | ITrarrow            IsUnicodeSyntax
677 678
  | ITat
  | ITtilde
679
  | ITdarrow            IsUnicodeSyntax
680 681
  | ITminus
  | ITbang
682
  | ITstar              IsUnicodeSyntax
683 684
  | ITdot

Ian Lynagh's avatar
Ian Lynagh committed
685
  | ITbiglam                    -- GHC-extension symbols
686

Ian Lynagh's avatar
Ian Lynagh committed
687
  | ITocurly                    -- special symbols
688 689 690 691
  | ITccurly
  | ITvocurly
  | ITvccurly
  | ITobrack
Ian Lynagh's avatar
Ian Lynagh committed
692 693
  | ITopabrack                  -- [:, for parallel arrays with -XParallelArrays
  | ITcpabrack                  -- :], for parallel arrays with -XParallelArrays
694 695 696 697 698 699 700 701 702
  | ITcbrack
  | IToparen
  | ITcparen
  | IToubxparen
  | ITcubxparen
  | ITsemi
  | ITcomma
  | ITunderscore
  | ITbackquote
dreixel's avatar
dreixel committed
703
  | ITsimpleQuote               --  '
704

Ian Lynagh's avatar
Ian Lynagh committed
705
  | ITvarid   FastString        -- identifiers
706 707 708 709 710 711 712 713
  | ITconid   FastString
  | ITvarsym  FastString
  | ITconsym  FastString
  | ITqvarid  (FastString,FastString)
  | ITqconid  (FastString,FastString)
  | ITqvarsym (FastString,FastString)
  | ITqconsym (FastString,FastString)

Ian Lynagh's avatar
Ian Lynagh committed
714
  | ITdupipvarid   FastString   -- GHC extension: implicit param: ?x
Adam Gundry's avatar
Adam Gundry committed
715
  | ITlabelvarid   FastString   -- Overloaded label: #x
716

717 718
  | ITchar     SourceText Char       -- Note [Literal source text] in BasicTypes
  | ITstring   SourceText FastString -- Note [Literal source text] in BasicTypes
719
  | ITinteger  IntegralLit           -- Note [Literal source text] in BasicTypes
Alan Zimmerman's avatar
Alan Zimmerman committed
720
  | ITrational FractionalLit
721

722 723 724 725
  | ITprimchar   SourceText Char     -- Note [Literal source text] in BasicTypes
  | ITprimstring SourceText ByteString -- Note [Literal source text] @BasicTypes
  | ITprimint    SourceText Integer  -- Note [Literal source text] in BasicTypes
  | ITprimword   SourceText Integer  -- Note [Literal source text] in BasicTypes
726 727
  | ITprimfloat  FractionalLit
  | ITprimdouble FractionalLit
728

729
  -- Template Haskell extension tokens
730 731 732 733 734 735 736 737 738 739 740 741
  | ITopenExpQuote HasE IsUnicodeSyntax --  [| or [e|
  | ITopenPatQuote                      --  [p|
  | ITopenDecQuote                      --  [d|
  | ITopenTypQuote                      --  [t|
  | ITcloseQuote IsUnicodeSyntax        --  |]
  | ITopenTExpQuote HasE                --  [|| or [e||
  | ITcloseTExpQuote                    --  ||]
  | ITidEscape   FastString             --  $x
  | ITparenEscape                       --  $(
  | ITidTyEscape   FastString           --  $$x
  | ITparenTyEscape                     --  $$(
  | ITtyQuote                           --  ''
742 743 744 745 746 747 748 749
  | 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 |]
750 751 752 753

  -- Arrow notation extension
  | ITproc
  | ITrec
754 755 756 757 758 759 760 761
  | IToparenbar  IsUnicodeSyntax -- ^ @(|@
  | ITcparenbar  IsUnicodeSyntax -- ^ @|)@
  | ITlarrowtail IsUnicodeSyntax -- ^ @-<@
  | ITrarrowtail IsUnicodeSyntax -- ^ @>-@
  | ITLarrowtail IsUnicodeSyntax -- ^ @-<<@
  | ITRarrowtail IsUnicodeSyntax -- ^ @>>-@

  -- | Type application '@' (lexed differently than as-pattern '@',
eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
762 763 764 765
  -- due to checking for preceding whitespace)
  | ITtypeApp


766 767
  | ITunknown String             -- ^ Used when the lexer can't make sense of it
  | ITeof                        -- ^ end of file token
768 769

  -- Documentation annotations
770 771 772 773 774 775 776
  | 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)
  | ITlineComment     String     -- ^ comment starting by "--"
  | ITblockComment    String     -- ^ comment in {- -}
777

778
  deriving Show
779

Alan Zimmerman's avatar
Alan Zimmerman committed
780 781 782
instance Outputable Token where
  ppr x = text (show x)

Alan Zimmerman's avatar
Alan Zimmerman committed
783

784 785 786 787 788 789 790
-- 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)
--
791
reservedWordsFM :: UniqFM (Token, ExtsBitmap)
792
reservedWordsFM = listToUFM $
Ian Lynagh's avatar
Ian Lynagh committed
793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819
    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 ),

820 821
         ( "forall",         ITforall NormalSyntax,
                                              xbit ExplicitForallBit .|.
822 823
                                              xbit InRulePragBit),
         ( "mdo",            ITmdo,           xbit RecursiveDoBit),
824 825 826
             -- See Note [Lexing type pseudo-keywords]
         ( "family",         ITfamily,        0 ),
         ( "role",           ITrole,          0 ),
827
         ( "pattern",        ITpattern,       xbit PatternSynonymsBit),
828
         ( "static",         ITstatic,        xbit StaticPointersBit ),
Ryan Scott's avatar
Ryan Scott committed
829 830
         ( "stock",          ITstock,         0 ),
         ( "anyclass",       ITanyclass,      0 ),
Ryan Scott's avatar
Ryan Scott committed
831
         ( "via",            ITvia,           0 ),
832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849
         ( "group",          ITgroup,         xbit TransformComprehensionsBit),
         ( "by",             ITby,            xbit TransformComprehensionsBit),
         ( "using",          ITusing,         xbit TransformComprehensionsBit),

         ( "foreign",        ITforeign,       xbit FfiBit),
         ( "export",         ITexport,        xbit FfiBit),
         ( "label",          ITlabel,         xbit FfiBit),
         ( "dynamic",        ITdynamic,       xbit FfiBit),
         ( "safe",           ITsafe,          xbit FfiBit .|.
                                              xbit SafeHaskellBit),
         ( "interruptible",  ITinterruptible, xbit InterruptibleFfiBit),
         ( "unsafe",         ITunsafe,        xbit FfiBit),
         ( "stdcall",        ITstdcallconv,   xbit FfiBit),
         ( "ccall",          ITccallconv,     xbit FfiBit),
         ( "capi",           ITcapiconv,      xbit CApiFfiBit),
         ( "prim",           ITprimcallconv,  xbit FfiBit),
         ( "javascript",     ITjavascriptcallconv, xbit FfiBit),

Edward Z. Yang's avatar
Edward Z. Yang committed
850 851 852 853
         ( "unit",           ITunit,          0 ),
         ( "dependency",     ITdependency,       0 ),
         ( "signature",      ITsignature,     0 ),

854 855 856
         ( "rec",            ITrec,           xbit ArrowsBit .|.
                                              xbit RecursiveDoBit),
         ( "proc",           ITproc,          xbit ArrowsBit)
857 858
     ]

859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874
{-----------------------------------
Note [Lexing type pseudo-keywords]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

One might think that we wish to treat 'family' and 'role' as regular old
varids whenever -XTypeFamilies and -XRoleAnnotations are off, respectively.
But, there is no need to do so. These pseudo-keywords are not stolen syntax:
they are only used after the keyword 'type' at the top-level, where varids are
not allowed. Furthermore, checks further downstream (TcTyClsDecls) ensure that
type families and role annotations are never declared without their extensions
on. In fact, by unconditionally lexing these pseudo-keywords as special, we
can get better error messages.

Also, note that these are included in the `varid` production in the parser --
a key detail to make all this work.
-------------------------------------}
875

876
reservedSymsFM :: UniqFM (Token, IsUnicodeSyntax, ExtsBitmap)
877
reservedSymsFM = listToUFM $
878 879
    map (\ (x,w,y,z) -> (mkFastString x,(w,y,z)))
      [ ("..",  ITdotdot,                   NormalSyntax,  0 )
Ian Lynagh's avatar
Ian Lynagh committed
880
        -- (:) is a reserved op, meaning only list cons
881 882 883 884 885 886 887 888 889 890 891 892 893 894
       ,(":",   ITcolon,                    NormalSyntax,  0 )
       ,("::",  ITdcolon NormalSyntax,      NormalSyntax,  0 )
       ,("=",   ITequal,                    NormalSyntax,  0 )
       ,("\\",  ITlam,                      NormalSyntax,  0 )
       ,("|",   ITvbar,                     NormalSyntax,  0 )
       ,("<-",  ITlarrow NormalSyntax,      NormalSyntax,  0 )
       ,("->",  ITrarrow NormalSyntax,      NormalSyntax,  0 )
       ,("@",   ITat,                       NormalSyntax,  0 )
       ,("~",   ITtilde,                    NormalSyntax,  0 )
       ,("=>",  ITdarrow NormalSyntax,      NormalSyntax,  0 )
       ,("-",   ITminus,                    NormalSyntax,  0 )
       ,("!",   ITbang,                     NormalSyntax,  0 )

       ,("*",   ITstar NormalSyntax,        NormalSyntax,  xbit StarIsTypeBit)
895

Ian Lynagh's avatar
Ian Lynagh committed
896
        -- For 'forall a . t'
897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915
       ,(".",   ITdot,                      NormalSyntax,  0 )

       ,("-<",  ITlarrowtail NormalSyntax,  NormalSyntax,  xbit ArrowsBit)
       ,(">-",  ITrarrowtail NormalSyntax,  NormalSyntax,  xbit ArrowsBit)
       ,("-<<", ITLarrowtail NormalSyntax,  NormalSyntax,  xbit ArrowsBit)
       ,(">>-", ITRarrowtail NormalSyntax,  NormalSyntax,  xbit ArrowsBit)

       ,("∷",   ITdcolon UnicodeSyntax,     UnicodeSyntax, 0 )
       ,("⇒",   ITdarrow UnicodeSyntax,     UnicodeSyntax, 0 )
       ,("∀",   ITforall UnicodeSyntax,     UnicodeSyntax, 0 )
       ,("→",   ITrarrow UnicodeSyntax,     UnicodeSyntax, 0 )
       ,("←",   ITlarrow UnicodeSyntax,     UnicodeSyntax, 0 )

       ,("⤙",   ITlarrowtail UnicodeSyntax, UnicodeSyntax, xbit ArrowsBit)
       ,("⤚",   ITrarrowtail UnicodeSyntax, UnicodeSyntax, xbit ArrowsBit)
       ,("⤛",   ITLarrowtail UnicodeSyntax, UnicodeSyntax, xbit ArrowsBit)
       ,("⤜",   ITRarrowtail UnicodeSyntax, UnicodeSyntax, xbit ArrowsBit)

       ,("★",   ITstar UnicodeSyntax,       UnicodeSyntax, xbit StarIsTypeBit)
916

Simon Marlow's avatar
Simon Marlow committed
917 918 919
        -- 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).
920 921 922 923 924
       ]

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

Ian Lynagh's avatar
Ian Lynagh committed
925
type Action = RealSrcSpan -> StringBuffer -> Int -> P (RealLocated Token)
926

927
special :: Token -> Action
twanvl's avatar
twanvl committed
928
special tok span _buf _len = return (L span tok)
929

930
token, layout_token :: Token -> Action
twanvl's avatar
twanvl committed
931 932
token t span _buf _len = return (L span t)
layout_token t span _buf _len = pushLexState layout >> return (L span t)
933

934 935
idtoken :: (StringBuffer -> Int -> Token) -> Action
idtoken f span buf len = return (L span $! (f buf len))
936

937
skip_one_varid :: (FastString -> Token) -> Action
Ian Lynagh's avatar
Ian Lynagh committed
938
skip_one_varid f span buf len
939
  = return (L span $! f (lexemeToFastString (stepOn buf) (len-1)))
940

941 942 943 944
skip_two_varid :: (FastString -> Token) -> Action
skip_two_varid f span buf len
  = return (L span $! f (lexemeToFastString (stepOn (stepOn buf)) (len-2)))

945
strtoken :: (String -> Token) -> Action
Ian Lynagh's avatar
Ian Lynagh committed
946
strtoken f span buf len =
947
  return (L span $! (f $! lexemeToString buf len))
948 949

begin :: Int -> Action
950
begin code _span _str _len = do pushLexState code; lexToken
951 952

pop :: Action
Ian Lynagh's avatar
Ian Lynagh committed
953 954
pop _span _buf _len = do _ <- popLexState
                         lexToken
955 956 957
-- See Note [Nested comment line pragmas]
failLinePrag1 :: Action
failLinePrag1 span _buf _len = do
958
  b <- getBit InNestedCommentBit
959 960 961 962 963 964
  if b then return (L span ITcomment_line_prag)
       else lexError "lexical error in pragma"

-- See Note [Nested comment line pragmas]
popLinePrag1 :: Action
popLinePrag1 span _buf _len = do
965
  b <- getBit InNestedCommentBit
966 967 968
  if b then return (L span ITcomment_line_prag) else do
    _ <- popLexState
    lexToken
969

970 971
hopefully_open_brace :: Action
hopefully_open_brace span buf len
972
 = do relaxed <- getBit RelaxedLayoutBit
973 974 975 976 977
      ctx <- getContext
      (AI l _) <- getInput
      let offset = srcLocCol l
          isOK = relaxed ||
                 case ctx of
978 979
                 Layout prev_off _ : _ -> prev_off < offset
                 _                     -> True
980
      if isOK then pop_and open_brace span buf len
Ian Lynagh's avatar
Ian Lynagh committed
981
              else failSpanMsgP (RealSrcSpan span) (text "Missing block")
982

983
pop_and :: Action -> Action
Ian Lynagh's avatar
Ian Lynagh committed
984 985
pop_and act span buf len = do _ <- popLexState
                              act span buf len
986

987
{-# INLINE nextCharIs #-}
Ian Lynagh's avatar
Ian Lynagh committed
988
nextCharIs :: StringBuffer -> (Char -> Bool) -> Bool
989 990
nextCharIs buf p = not (atEnd buf) && p (currentChar buf)

991 992 993 994
{-# INLINE nextCharIsNot #-}
nextCharIsNot :: StringBuffer -> (Char -> Bool) -> Bool
nextCharIsNot buf p = not (nextCharIs buf p)

995
notFollowedBy :: Char -> AlexAccPred ExtsBitmap
Ian Lynagh's avatar
Ian Lynagh committed
996
notFollowedBy char _ _ _ (AI _ buf)
997
  = nextCharIsNot buf (== char)
998

999
notFollowedBySymbol :: AlexAccPred ExtsBitmap
1000
notFollowedBySymbol _ _ _ (AI _ buf)
1001
  = nextCharIsNot buf (`elem` "!#$%&*+./<=>?@\\^|-~")
ross's avatar
ross committed
1002

Adam Gundry's avatar
Adam Gundry committed
1003 1004 1005 1006
followedByDigit :: AlexAccPred ExtsBitmap
followedByDigit _ _ _ (AI _ buf)
  = afterOptionalSpace buf (\b -> nextCharIs b (`elem` ['0'..'9']))

1007 1008 1009 1010
ifCurrentChar :: Char -> AlexAccPred ExtsBitmap
ifCurrentChar char _ (AI _ buf) _ _
  = nextCharIs buf (== char)

1011 1012 1013 1014 1015
-- 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.
1016
isNormalComment :: AlexAccPred ExtsBitmap
1017
isNormalComment bits _ _ (AI _ buf)
1018 1019
  | HaddockBit `xtest` bits = notFollowedByDocOrPragma
  | otherwise               = nextCharIsNot buf (== '#')
1020 1021
  where
    notFollowedByDocOrPragma
1022
       = afterOptionalSpace buf (\b -> nextCharIsNot b (`elem` "|^*$#"))
1023

1024 1025 1026 1027 1028
afterOptionalSpace :: StringBuffer -> (StringBuffer -> Bool) -> Bool
afterOptionalSpace buf p
    = if nextCharIs buf (== ' ')
      then p (snd (nextChar buf))
      else p buf
1029

1030
atEOL :: AlexAccPred ExtsBitmap
1031
atEOL _ _ _ (AI _ buf) = atEnd buf || currentChar buf == '\n'
1032

1033 1034 1035 1036 1037
ifExtension :: ExtBits -> AlexAccPred ExtsBitmap
ifExtension extBits bits _ _ _ = extBits `xtest` bits

alexNotPred p userState in1 len in2
  = not (p userState in1 len in2)
1038

1039 1040
alexOrPred p1 p2 userState in1 len in2
  = p1 userState in1 len in2 || p2 userState in1 len in2
1041

1042 1043 1044
multiline_doc_comment :: Action
multiline_doc_comment span buf _len = withLexedDocType (worker "")
  where
1045
    worker commentAcc input docType checkNextLine = case alexGetChar' input of
Ian Lynagh's avatar
Ian Lynagh committed
1046
      Just ('\n', input')
1047 1048
        | checkNextLine -> case checkIfCommentLine input' of
          Just input -> worker ('\n':commentAcc) input docType checkNextLine
1049
          Nothing -> docCommentEnd input commentAcc docType buf span
1050 1051
        | otherwise -> docCommentEnd input commentAcc docType buf span
      Just (c, input) -> worker (c:commentAcc) input docType checkNextLine
1052
      Nothing -> docCommentEnd input commentAcc docType buf span
Ian Lynagh's avatar
Ian Lynagh committed
1053

1054 1055 1056 1057 1058 1059 1060 1061
    -- Check if the next line of input belongs to this doc comment as well.
    -- A doc comment continues onto the next line when the following
    -- conditions are met:
    --   * The line starts with "--"
    --   * The line doesn't start with "---".
    --   * The line doesn't start with "-- $", because that would be the
    --     start of a /new/ named haddock chunk (#10398).
    checkIfCommentLine :: AlexInput -> Maybe AlexInput
1062 1063
    checkIfCommentLine input = check (dropNonNewlineSpace input)
      where
1064 1065 1066 1067 1068 1069 1070 1071 1072 1073
        check input = do
          ('-', input) <- alexGetChar' input
          ('-', input) <- alexGetChar' input
          (c, after_c) <- alexGetChar' input
          case c of
            '-' -> Nothing
            ' ' -> case alexGetChar' after_c of
                     Just ('$', _) -> Nothing
                     _ -> Just input
            _   -> Just input
1074

1075
        dropNonNewlineSpace input = case alexGetChar' input of
Ian Lynagh's avatar
Ian Lynagh committed
1076
          Just (c, input')
1077 1078 1079 1080
            | isSpace c && c /= '\n' -> dropNonNewlineSpace input'
            | otherwise -> input
          Nothing -> input

Jedai's avatar
Jedai committed
1081 1082
lineCommentToken :: Action
lineCommentToken span buf len = do
1083
  b <- getBit RawTokenStreamBit
Jedai's avatar
Jedai committed
1084 1085
  if b then strtoken ITlineComment span buf len else lexToken

1086 1087 1088 1089
{-
  nested comments require traversing by hand, they can't be parsed
  using regular expressions.
-}
Ian Lynagh's avatar
Ian Lynagh committed
1090
nested_comment :: P (RealLocated Token) -> Action
Alan Zimmerman's avatar
Alan Zimmerman committed
1091
nested_comment cont span buf len = do