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 46

-- See Note [Warnings in code generated by Alex] in compiler/parser/Lexer.x
47 48
{-# OPTIONS_GHC -fno-warn-unused-matches #-}
{-# OPTIONS_GHC -fno-warn-unused-binds #-}
49
{-# OPTIONS_GHC -fno-warn-unused-imports #-}
50
{-# OPTIONS_GHC -fno-warn-tabs #-}
51
{-# OPTIONS_GHC -fno-warn-missing-signatures #-}
52

53 54
{-# OPTIONS_GHC -funbox-strict-fields #-}

55
module Lexer (
56 57 58
   Token(..), lexer, pragState, mkPState, mkPStatePure, PState(..),
   P(..), ParseResult(..), mkParserFlags, ParserFlags(..), getSrcLoc,
   getPState, extopt, withThisPackage,
59
   failLocMsgP, failSpanMsgP, srcParseFail,
Ian Lynagh's avatar
Ian Lynagh committed
60
   getMessages,
61
   popContext, pushModuleContext, setLastToken, setSrcLoc,
vivian's avatar
vivian committed
62
   activeContext, nextIsEOF,
simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
63
   getLexState, popLexState, pushLexState,
64
   extension, bangPatEnabled, datatypeContextsEnabled,
65
   traditionalRecordSyntaxEnabled,
66 67
   explicitForallEnabled,
   inRulePrag,
cactus's avatar
cactus committed
68 69 70
   explicitNamespacesEnabled,
   patternSynonymsEnabled,
   sccProfilingOn, hpcEnabled,
Jedai's avatar
Jedai committed
71
   addWarning,
Alan Zimmerman's avatar
Alan Zimmerman committed
72
   lexTokenStream,
73
   addAnnotation,AddAnn,addAnnsAt,mkParensApiAnn,
74
   commentToAnnotation,
75
   moveAnnotations
76 77
  ) where

78 79
import GhcPrelude

rodlogic's avatar
rodlogic committed
80 81
-- base
import Control.Monad
82
import Control.Monad.Fail as MonadFail
rodlogic's avatar
rodlogic committed
83
import Data.Bits
84
import Data.Char
rodlogic's avatar
rodlogic committed
85 86 87 88
import Data.List
import Data.Maybe
import Data.Word

89 90
import EnumSet (EnumSet)
import qualified EnumSet
91

92 93 94
-- ghc-boot
import qualified GHC.LanguageExtensions as LangExt

rodlogic's avatar
rodlogic committed
95 96 97 98 99 100 101 102
-- bytestring
import Data.ByteString (ByteString)

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

-- compiler/utils
103
import Bag
104 105 106 107
import Outputable
import StringBuffer
import FastString
import UniqFM
108
import Util             ( readRational, readHexRational )
rodlogic's avatar
rodlogic committed
109 110 111

-- compiler/main
import ErrUtils
112
import DynFlags
rodlogic's avatar
rodlogic committed
113 114 115

-- compiler/basicTypes
import SrcLoc
116
import Module
117 118
import BasicTypes     ( InlineSpec(..), RuleMatchInfo(..),
                        IntegralLit(..), FractionalLit(..),
Alan Zimmerman's avatar
Alan Zimmerman committed
119
                        SourceText(..) )
120

rodlogic's avatar
rodlogic committed
121 122
-- compiler/parser
import Ctype
Alan Zimmerman's avatar
Alan Zimmerman committed
123 124

import ApiAnnotation
125 126
}

127 128 129
-- -----------------------------------------------------------------------------
-- Alex "Character set macros"

130 131
-- NB: The logic behind these definitions is also reflected in basicTypes/Lexeme.hs
-- Any changes here should likely be reflected there.
132
$unispace    = \x05 -- Trick Alex into handling Unicode. See [Unicode in Alex].
133 134 135
$nl          = [\n\r\f]
$whitechar   = [$nl\v\ $unispace]
$white_no_nl = $whitechar # \n -- TODO #8424
136
$tab         = \t
137 138

$ascdigit  = 0-9
139
$unidigit  = \x03 -- Trick Alex into handling Unicode. See [Unicode in Alex].
140
$decdigit  = $ascdigit -- for now, should really be $digit (ToDo)
141 142 143
$digit     = [$ascdigit $unidigit]

$special   = [\(\)\,\;\[\]\`\{\}]
144
$ascsymbol = [\!\#\$\%\&\*\+\.\/\<\=\>\?\@\\\^\|\-\~\:]
145
$unisymbol = \x04 -- Trick Alex into handling Unicode. See [Unicode in Alex].
146
$symbol    = [$ascsymbol $unisymbol] # [$special \_\"\']
147

148
$unilarge  = \x01 -- Trick Alex into handling Unicode. See [Unicode in Alex].
149
$asclarge  = [A-Z]
150 151
$large     = [$asclarge $unilarge]

152
$unismall  = \x02 -- Trick Alex into handling Unicode. See [Unicode in Alex].
153
$ascsmall  = [a-z]
154 155
$small     = [$ascsmall $unismall \_]

156
$unigraphic = \x06 -- Trick Alex into handling Unicode. See [Unicode in Alex].
157
$graphic   = [$small $large $symbol $digit $special $unigraphic \"\']
158

159
$binit     = 0-1
Ian Lynagh's avatar
Ian Lynagh committed
160
$octit     = 0-7
161
$hexit     = [$decdigit A-F a-f]
162

163
$uniidchar = \x07 -- Trick Alex into handling Unicode. See [Unicode in Alex].
164
$idchar    = [$small $large $digit $uniidchar \']
165

166 167
$pragmachar = [$small $large $digit]

168 169
$docsym    = [\| \^ \* \$]

170

171 172 173 174 175 176 177 178
-- -----------------------------------------------------------------------------
-- 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
179

180 181 182 183 184 185 186 187
-- 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
188 189

@qual = (@conid \.)+
190 191 192 193
@qvarid = @qual @varid
@qconid = @qual @conid
@qvarsym = @qual @varsym
@qconsym = @qual @consym
194

195 196
@floating_point = @numspc @decimal \. @decimal @exponent? | @numspc @decimal @exponent
@hex_floating_point = @numspc @hexadecimal \. @hexadecimal @bin_exponent? | @numspc @hexadecimal @bin_exponent
197

198 199 200 201 202
-- normal signed numerical literals can only be explicitly negative,
-- not explicitly positive (contrast @exponent)
@negative = \-
@signed = @negative ?

203 204 205 206

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

207 208
haskell :-

209 210 211 212 213

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

-- everywhere: skip whitespace
Ian Lynagh's avatar
Ian Lynagh committed
214
$white_no_nl+ ;
215
$tab          { warnTab }
216 217 218 219 220

-- 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
221 222 223 224
-- are). We also rule out nested Haddock comments, if the -haddock flag is
-- set.

"{-" / { isNormalComment } { nested_comment lexToken }
225 226 227 228

-- 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.
229 230

-- Since Haddock comments aren't valid in every state, we need to rule them
Ian Lynagh's avatar
Ian Lynagh committed
231
-- out here.
232 233 234 235 236 237 238

-- 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.

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

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

244
"-- " $docsym .* / { ifExtension (not . haddockEnabled) } { lineCommentToken }
245 246 247 248 249 250 251

-- 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.

252
"---"\-* ~$symbol .* { lineCommentToken }
253 254 255 256

-- 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
257
"--"\-* / { atEOL } { lineCommentToken }
258 259 260 261

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

Jedai's avatar
Jedai committed
262
"-- " / { atEOL } { lineCommentToken }
263 264 265 266 267 268 269 270 271 272 273

-- '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
274
  \n                                    ;
Adam Gundry's avatar
Adam Gundry committed
275 276
  ^\# line                              { begin line_prag1 }
  ^\# / { followedByDigit }             { begin line_prag1 }
Ian Lynagh's avatar
Ian Lynagh committed
277 278 279
  ^\# pragma .* \n                      ; -- GCC 3.3 CPP generated, apparently
  ^\# \! .* \n                          ; -- #!, for scripts
  ()                                    { do_bol }
280 281 282 283 284
}

-- 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.
285
<layout, layout_do, layout_if> {
Ian Lynagh's avatar
Ian Lynagh committed
286 287 288 289
  \{ / { notFollowedBy '-' }            { hopefully_open_brace }
        -- we might encounter {-# here, but {- has been handled already
  \n                                    ;
  ^\# (line)?                           { begin line_prag1 }
290 291
}

292 293
-- after an 'if', a vertical bar starts a layout context for MultiWayIf
<layout_if> {
294
  \| / { notFollowedBySymbol }          { new_layout_context True dontGenerateSemic ITvbar }
295 296 297
  ()                                    { pop }
}

298
-- do is treated in a subtly different way, see new_layout_context
299 300
<layout>    ()                          { new_layout_context True  generateSemic ITvocurly }
<layout_do> ()                          { new_layout_context False generateSemic ITvocurly }
301 302 303 304

-- 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
305
<layout_left>  ()                       { do_layout_left }
306

Ian Lynagh's avatar
Ian Lynagh committed
307
<0,option_prags> \n                     { begin bol }
308

309 310
"{-#" $whitechar* $pragmachar+ / { known_pragma linePrags }
                                { dispatch_pragmas linePrags }
311 312 313

-- single-line line pragmas, of the form
--    # <line> "<file>" <extra-stuff> \n
314
<line_prag1> @decimal                   { setLine line_prag1a }
Ian Lynagh's avatar
Ian Lynagh committed
315 316
<line_prag1a> \" [$graphic \ ]* \"      { setFile line_prag1b }
<line_prag1b> .*                        { pop }
317 318 319

-- Haskell-style line pragmas, of the form
--    {-# LINE <line> "<file>" #-}
320
<line_prag2> @decimal                   { setLine line_prag2a }
Ian Lynagh's avatar
Ian Lynagh committed
321 322
<line_prag2a> \" [$graphic \ ]* \"      { setFile line_prag2b }
<line_prag2b> "#-}"|"-}"                { pop }
323 324
   -- NOTE: accept -} at the end of a LINE pragma, for compatibility
   -- with older versions of GHC which generated these.
325

Rufflewind's avatar
Rufflewind committed
326 327 328 329
-- Haskell-style column pragmas, of the form
--    {-# COLUMN <column> #-}
<column_prag> @decimal $whitechar* "#-}" { setColumn }

330
<0,option_prags> {
Ian Lynagh's avatar
Ian Lynagh committed
331
  "{-#" $whitechar* $pragmachar+
332 333 334 335 336
        $whitechar+ $pragmachar+ / { known_pragma twoWordPrags }
                                 { dispatch_pragmas twoWordPrags }

  "{-#" $whitechar* $pragmachar+ / { known_pragma oneWordPrags }
                                 { dispatch_pragmas oneWordPrags }
337

338
  -- We ignore all these pragmas, but don't generate a warning for them
339 340
  "{-#" $whitechar* $pragmachar+ / { known_pragma ignoredPrags }
                                 { dispatch_pragmas ignoredPrags }
341 342

  -- ToDo: should only be valid inside a pragma:
Ian Lynagh's avatar
Ian Lynagh committed
343
  "#-}"                          { endPrag }
344 345
}

346
<option_prags> {
347 348
  "{-#"  $whitechar* $pragmachar+ / { known_pragma fileHeaderPrags }
                                   { dispatch_pragmas fileHeaderPrags }
David Waern's avatar
David Waern committed
349 350
}

351 352
<0> {
  -- In the "0" mode we ignore these pragmas
353
  "{-#"  $whitechar* $pragmachar+ / { known_pragma fileHeaderPrags }
354 355 356
                     { nested_comment lexToken }
}

357
<0,option_prags> {
358 359
  "{-#"  { warnThen Opt_WarnUnrecognisedPragmas (text "Unrecognised pragma")
                    (nested_comment lexToken) }
360 361
}

362 363
-- '0' state: ordinary lexemes

364 365
-- Haddock comments

366
<0,option_prags> {
David Waern's avatar
David Waern committed
367 368
  "-- " $docsym      / { ifExtension haddockEnabled } { multiline_doc_comment }
  "{-" \ ? $docsym   / { ifExtension haddockEnabled } { nested_doc_comment }
369 370
}

371 372
-- "special" symbols

373
<0> {
374 375
  "[|"        / { ifExtension thQuotesEnabled } { token (ITopenExpQuote NoE
                                                                NormalSyntax) }
376
  "[||"       / { ifExtension thQuotesEnabled } { token (ITopenTExpQuote NoE) }
377 378
  "[e|"       / { ifExtension thQuotesEnabled } { token (ITopenExpQuote HasE
                                                                NormalSyntax) }
379 380 381 382
  "[e||"      / { ifExtension thQuotesEnabled } { token (ITopenTExpQuote HasE) }
  "[p|"       / { ifExtension thQuotesEnabled } { token ITopenPatQuote }
  "[d|"       / { ifExtension thQuotesEnabled } { layout_token ITopenDecQuote }
  "[t|"       / { ifExtension thQuotesEnabled } { token ITopenTypQuote }
383 384
  "|]"        / { ifExtension thQuotesEnabled } { token (ITcloseQuote
                                                                NormalSyntax) }
385
  "||]"       / { ifExtension thQuotesEnabled } { token ITcloseTExpQuote }
386 387 388 389
  \$ @varid   / { ifExtension thEnabled } { skip_one_varid ITidEscape }
  "$$" @varid / { ifExtension thEnabled } { skip_two_varid ITidTyEscape }
  "$("        / { ifExtension thEnabled } { token ITparenEscape }
  "$$("       / { ifExtension thEnabled } { token ITparenTyEscape }
390

391
  "[" @varid "|"  / { ifExtension qqEnabled }
392
                     { lex_quasiquote_tok }
393 394

  -- qualified quasi-quote (#5555)
395 396
  "[" @qvarid "|"  / { ifExtension qqEnabled }
                     { lex_qquasiquote_tok }
397 398 399 400 401 402 403 404 405

  $unigraphic -- ⟦
    / { ifCurrentChar '⟦' `alexAndPred`
        ifExtension (\i -> unicodeSyntaxEnabled i && thQuotesEnabled i) }
    { token (ITopenExpQuote NoE UnicodeSyntax) }
  $unigraphic -- ⟧
    / { ifCurrentChar '⟧' `alexAndPred`
        ifExtension (\i -> unicodeSyntaxEnabled i && thQuotesEnabled i) }
    { token (ITcloseQuote UnicodeSyntax) }
406 407
}

eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
408 409 410 411 412 413 414 415
  -- See Note [Lexing type applications]
<0> {
    [^ $idchar \) ] ^
  "@"
    / { ifExtension typeApplicationEnabled `alexAndPred` notFollowedBySymbol }
    { token ITtypeApp }
}

416
<0> {
ross's avatar
ross committed
417
  "(|" / { ifExtension arrowsEnabled `alexAndPred` notFollowedBySymbol }
418 419 420 421 422 423 424 425 426 427 428
                                        { special (IToparenbar NormalSyntax) }
  "|)" / { ifExtension arrowsEnabled }  { special (ITcparenbar NormalSyntax) }

  $unigraphic -- ⦇
    / { ifCurrentChar '⦇' `alexAndPred`
        ifExtension (\i -> unicodeSyntaxEnabled i && arrowsEnabled i) }
    { special (IToparenbar UnicodeSyntax) }
  $unigraphic -- ⦈
    / { ifCurrentChar '⦈' `alexAndPred`
        ifExtension (\i -> unicodeSyntaxEnabled i && arrowsEnabled i) }
    { special (ITcparenbar UnicodeSyntax) }
429 430
}

431
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
432
  \? @varid / { ifExtension ipEnabled } { skip_one_varid ITdupipvarid }
433 434
}

Adam Gundry's avatar
Adam Gundry committed
435 436 437 438 439
<0> {
  "#" @varid / { ifExtension overloadedLabelsEnabled }
               { skip_one_varid ITlabelvarid }
}

440
<0> {
441
  "(#" / { orExtensions unboxedTuplesEnabled unboxedSumsEnabled }
Ian Lynagh's avatar
Ian Lynagh committed
442
         { token IToubxparen }
443
  "#)" / { orExtensions unboxedTuplesEnabled unboxedSumsEnabled }
Ian Lynagh's avatar
Ian Lynagh committed
444 445 446
         { token ITcubxparen }
}

447
<0,option_prags> {
Ian Lynagh's avatar
Ian Lynagh committed
448 449 450 451 452 453 454 455 456 457
  \(                                    { special IToparen }
  \)                                    { special ITcparen }
  \[                                    { special ITobrack }
  \]                                    { special ITcbrack }
  \,                                    { special ITcomma }
  \;                                    { special ITsemi }
  \`                                    { special ITbackquote }

  \{                                    { open_brace }
  \}                                    { close_brace }
458 459
}

460
<0,option_prags> {
461 462
  @qvarid                       { idtoken qvarid }
  @qconid                       { idtoken qconid }
Ian Lynagh's avatar
Ian Lynagh committed
463
  @varid                        { varid }
464
  @conid                        { idtoken conid }
465 466
}

467
<0> {
468 469
  @qvarid "#"+      / { ifExtension magicHashEnabled } { idtoken qvarid }
  @qconid "#"+      / { ifExtension magicHashEnabled } { idtoken qconid }
470
  @varid "#"+       / { ifExtension magicHashEnabled } { varid }
471
  @conid "#"+       / { ifExtension magicHashEnabled } { idtoken conid }
472 473
}

474 475
-- ToDo: - move `var` and (sym) into lexical syntax?
--       - remove backquote from $special?
476
<0> {
477 478
  @qvarsym                                         { idtoken qvarsym }
  @qconsym                                         { idtoken qconsym }
479 480
  @varsym                                          { varsym }
  @consym                                          { consym }
481 482
}

483 484
-- For the normal boxed literals we need to be careful
-- when trying to be close to Haskell98
485 486 487 488 489 490 491 492 493 494

-- 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.
--
495
<0> {
496
  -- Normal integral literals (:: Num a => a, from Integer)
497
  @decimal                                                               { tok_num positive 0 0 decimal }
498 499 500
  0[bB] @numspc @binary        / { ifExtension binaryLiteralsEnabled }   { 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 }
501
  @negative @decimal           / { ifExtension negativeLiteralsEnabled } { tok_num negative 1 1 decimal }
502 503 504 505
  @negative 0[bB] @numspc @binary  / { ifExtension negativeLiteralsEnabled `alexAndPred`
                                       ifExtension binaryLiteralsEnabled }   { tok_num negative 3 3 binary }
  @negative 0[oO] @numspc @octal   / { ifExtension negativeLiteralsEnabled } { tok_num negative 3 3 octal }
  @negative 0[xX] @numspc @hexadecimal / { ifExtension negativeLiteralsEnabled } { tok_num negative 3 3 hexadecimal }
506 507

  -- Normal rational literals (:: Fractional a => a, from Rational)
508 509 510 511 512
  @floating_point                                                        { tok_frac 0 tok_float }
  @negative @floating_point    / { ifExtension negativeLiteralsEnabled } { tok_frac 0 tok_float }
  0[xX] @numspc @hex_floating_point     / { ifExtension hexFloatLiteralsEnabled } { tok_frac 0 tok_hex_float }
  @negative 0[xX] @numspc @hex_floating_point / { ifExtension hexFloatLiteralsEnabled `alexAndPred`
                                                  ifExtension negativeLiteralsEnabled } { tok_frac 0 tok_hex_float }
513 514
}

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

Ian Lynagh's avatar
Ian Lynagh committed
530
  @decimal                     \# \# / { ifExtension magicHashEnabled } { tok_primword 0 2 decimal }
531
  0[bB] @numspc @binary        \# \# / { ifExtension magicHashEnabled `alexAndPred`
532
                                         ifExtension binaryLiteralsEnabled } { tok_primword 2 4 binary }
533 534
  0[oO] @numspc @octal         \# \# / { ifExtension magicHashEnabled } { tok_primword 2 4 octal }
  0[xX] @numspc @hexadecimal   \# \# / { ifExtension magicHashEnabled } { tok_primword 2 4 hexadecimal }
Ian Lynagh's avatar
Ian Lynagh committed
535

536 537
  -- Unboxed floats and doubles (:: Float#, :: Double#)
  -- prim_{float,double} work with signed literals
538 539
  @signed @floating_point \# / { ifExtension magicHashEnabled } { tok_frac 1 tok_primfloat }
  @signed @floating_point \# \# / { ifExtension magicHashEnabled } { tok_frac 2 tok_primdouble }
540 541 542 543 544 545
}

-- 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.
546
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
547 548
  \'                            { lex_char_tok }
  \"                            { lex_string_tok }
549 550
}

eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
551 552 553 554 555 556 557 558 559 560 561
-- 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
562
-- Disambiguation is accomplished by requiring *something* to appear between
eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
563 564 565 566 567 568 569 570 571 572 573 574 575 576
-- 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.
577 578 579 580

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

581
{
582

583 584 585
-- -----------------------------------------------------------------------------
-- The token type

586
data Token
Ian Lynagh's avatar
Ian Lynagh committed
587
  = ITas                        -- Haskell keywords
588 589 590 591 592 593 594 595
  | ITcase
  | ITclass
  | ITdata
  | ITdefault
  | ITderiving
  | ITdo
  | ITelse
  | IThiding
596
  | ITforeign
597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612
  | ITif
  | ITimport
  | ITin
  | ITinfix
  | ITinfixl
  | ITinfixr
  | ITinstance
  | ITlet
  | ITmodule
  | ITnewtype
  | ITof
  | ITqualified
  | ITthen
  | ITtype
  | ITwhere

613
  | ITforall            IsUnicodeSyntax -- GHC extension keywords
614 615 616 617
  | ITexport
  | ITlabel
  | ITdynamic
  | ITsafe
618
  | ITinterruptible
619 620 621
  | ITunsafe
  | ITstdcallconv
  | ITccallconv
622
  | ITcapiconv
623
  | ITprimcallconv
thoughtpolice's avatar
thoughtpolice committed
624
  | ITjavascriptcallconv
625
  | ITmdo
626
  | ITfamily
627
  | ITrole
628 629 630
  | ITgroup
  | ITby
  | ITusing
cactus's avatar
cactus committed
631
  | ITpattern
Facundo Domínguez's avatar
Facundo Domínguez committed
632
  | ITstatic
Ryan Scott's avatar
Ryan Scott committed
633 634
  | ITstock
  | ITanyclass
635

Edward Z. Yang's avatar
Edward Z. Yang committed
636 637 638 639 640 641
  -- Backpack tokens
  | ITunit
  | ITsignature
  | ITdependency
  | ITrequires

Alan Zimmerman's avatar
Alan Zimmerman committed
642 643 644 645 646 647 648 649
  -- 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
650 651
  | ITline_prag         SourceText  -- not usually produced, see 'use_pos_prags'
  | ITcolumn_prag       SourceText  -- not usually produced, see 'use_pos_prags'
Alan Zimmerman's avatar
Alan Zimmerman committed
652 653 654 655 656 657
  | ITscc_prag          SourceText
  | ITgenerated_prag    SourceText
  | ITcore_prag         SourceText         -- hdaume: core annotations
  | ITunpack_prag       SourceText
  | ITnounpack_prag     SourceText
  | ITann_prag          SourceText
658
  | ITcomplete_prag     SourceText
659
  | ITclose_prag
660 661 662
  | IToptions_prag String
  | ITinclude_prag String
  | ITlanguage_prag
Alan Zimmerman's avatar
Alan Zimmerman committed
663 664 665 666 667 668
  | 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
669

Ian Lynagh's avatar
Ian Lynagh committed
670
  | ITdotdot                    -- reserved symbols
671
  | ITcolon
672
  | ITdcolon            IsUnicodeSyntax
673 674
  | ITequal
  | ITlam
675
  | ITlcase
676
  | ITvbar
677 678
  | ITlarrow            IsUnicodeSyntax
  | ITrarrow            IsUnicodeSyntax
679 680
  | ITat
  | ITtilde
681
  | ITdarrow            IsUnicodeSyntax
682 683 684 685
  | ITminus
  | ITbang
  | ITdot

Ian Lynagh's avatar
Ian Lynagh committed
686
  | ITbiglam                    -- GHC-extension symbols
687

Ian Lynagh's avatar
Ian Lynagh committed
688
  | ITocurly                    -- special symbols
689 690 691 692
  | ITccurly
  | ITvocurly
  | ITvccurly
  | ITobrack
Ian Lynagh's avatar
Ian Lynagh committed
693 694
  | ITopabrack                  -- [:, for parallel arrays with -XParallelArrays
  | ITcpabrack                  -- :], for parallel arrays with -XParallelArrays
695 696 697 698 699 700 701 702 703
  | ITcbrack
  | IToparen
  | ITcparen
  | IToubxparen
  | ITcubxparen
  | ITsemi
  | ITcomma
  | ITunderscore
  | ITbackquote
dreixel's avatar
dreixel committed
704
  | ITsimpleQuote               --  '
705

Ian Lynagh's avatar
Ian Lynagh committed
706
  | ITvarid   FastString        -- identifiers
707 708 709 710 711 712 713 714
  | 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
715
  | ITdupipvarid   FastString   -- GHC extension: implicit param: ?x
Adam Gundry's avatar
Adam Gundry committed
716
  | ITlabelvarid   FastString   -- Overloaded label: #x
717

718 719
  | ITchar     SourceText Char       -- Note [Literal source text] in BasicTypes
  | ITstring   SourceText FastString -- Note [Literal source text] in BasicTypes
720
  | ITinteger  IntegralLit           -- Note [Literal source text] in BasicTypes
Alan Zimmerman's avatar
Alan Zimmerman committed
721
  | ITrational FractionalLit
722

723 724 725 726
  | 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
727 728
  | ITprimfloat  FractionalLit
  | ITprimdouble FractionalLit
729

730
  -- Template Haskell extension tokens
731 732 733 734 735 736 737 738 739 740 741 742
  | 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                           --  ''
743 744 745 746 747 748 749 750
  | 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 |]
751 752 753 754

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

eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
762 763 764 765 766 767 768
  -- type application '@' (lexed differently than as-pattern '@',
  -- due to checking for preceding whitespace)
  | ITtypeApp


  | ITunknown String            -- Used when the lexer can't make sense of it
  | ITeof                       -- end of file token
769 770 771 772 773 774 775

  -- 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)
Jedai's avatar
Jedai committed
776 777
  | ITlineComment     String     -- comment starting by "--"
  | ITblockComment    String     -- comment in {- -}
778

779
  deriving Show
780

Alan Zimmerman's avatar
Alan Zimmerman committed
781 782 783
instance Outputable Token where
  ppr x = text (show x)

Alan Zimmerman's avatar
Alan Zimmerman committed
784

785 786 787 788 789 790 791
-- 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)
--
792
reservedWordsFM :: UniqFM (Token, ExtsBitmap)
793
reservedWordsFM = listToUFM $
Ian Lynagh's avatar
Ian Lynagh committed
794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820
    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 ),

821 822
         ( "forall",         ITforall NormalSyntax,
                                              xbit ExplicitForallBit .|.
823 824
                                              xbit InRulePragBit),
         ( "mdo",            ITmdo,           xbit RecursiveDoBit),
825 826 827
             -- See Note [Lexing type pseudo-keywords]
         ( "family",         ITfamily,        0 ),
         ( "role",           ITrole,          0 ),
828
         ( "pattern",        ITpattern,       xbit PatternSynonymsBit),
Facundo Domínguez's avatar
Facundo Domínguez committed
829
         ( "static",         ITstatic,        0 ),
Ryan Scott's avatar
Ryan Scott committed
830 831
         ( "stock",          ITstock,         0 ),
         ( "anyclass",       ITanyclass,      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, ExtsBitmap -> Bool)
877
reservedSymsFM = listToUFM $
Ian Lynagh's avatar
Ian Lynagh committed
878
    map (\ (x,y,z) -> (mkFastString x,(y,z)))
879
      [ ("..",  ITdotdot,              always)
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
       ,(":",   ITcolon,               always)
       ,("::",  ITdcolon NormalSyntax, always)
       ,("=",   ITequal,               always)
       ,("\\",  ITlam,                 always)
       ,("|",   ITvbar,                always)
       ,("<-",  ITlarrow NormalSyntax, always)
       ,("->",  ITrarrow NormalSyntax, always)
       ,("@",   ITat,                  always)
       ,("~",   ITtilde,               always)
       ,("=>",  ITdarrow NormalSyntax, always)
       ,("-",   ITminus,               always)
       ,("!",   ITbang,                always)
Ian Lynagh's avatar
Ian Lynagh committed
893 894

        -- For 'forall a . t'
895
       ,(".", ITdot,  always) -- \i -> explicitForallEnabled i || inRulePrag i)
Ian Lynagh's avatar
Ian Lynagh committed
896

897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916
       ,("-<",  ITlarrowtail NormalSyntax, arrowsEnabled)
       ,(">-",  ITrarrowtail NormalSyntax, arrowsEnabled)
       ,("-<<", ITLarrowtail NormalSyntax, arrowsEnabled)
       ,(">>-", ITRarrowtail NormalSyntax, arrowsEnabled)

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

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

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

956 957 958 959 960 961 962 963
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
964 965
                 Layout prev_off _ : _ -> prev_off < offset
                 _                     -> True
966
      if isOK then pop_and open_brace span buf len
Ian Lynagh's avatar
Ian Lynagh committed
967
              else failSpanMsgP (RealSrcSpan span) (text "Missing block")
968

969
pop_and :: Action -> Action
Ian Lynagh's avatar
Ian Lynagh committed
970 971
pop_and act span buf len = do _ <- popLexState
                              act span buf len
972

973
{-# INLINE nextCharIs #-}
Ian Lynagh's avatar
Ian Lynagh committed
974
nextCharIs :: StringBuffer -> (Char -> Bool) -> Bool
975 976
nextCharIs buf p = not (atEnd buf) && p (currentChar buf)

977 978 979 980
{-# INLINE nextCharIsNot #-}
nextCharIsNot :: StringBuffer -> (Char -> Bool) -> Bool
nextCharIsNot buf p = not (nextCharIs buf p)

981
notFollowedBy :: Char -> AlexAccPred ExtsBitmap
Ian Lynagh's avatar
Ian Lynagh committed
982
notFollowedBy char _ _ _ (AI _ buf)
983
  = nextCharIsNot buf (== char)
984

985
notFollowedBySymbol :: AlexAccPred ExtsBitmap
986
notFollowedBySymbol _ _ _ (AI _ buf)
987
  = nextCharIsNot buf (`elem` "!#$%&*+./<=>?@\\^|-~")
ross's avatar
ross committed
988

Adam Gundry's avatar
Adam Gundry committed
989 990 991 992
followedByDigit :: AlexAccPred ExtsBitmap
followedByDigit _ _ _ (AI _ buf)
  = afterOptionalSpace buf (\b -> nextCharIs b (`elem` ['0'..'9']))

993 994 995 996
ifCurrentChar :: Char -> AlexAccPred ExtsBitmap
ifCurrentChar char _ (AI _ buf) _ _
  = nextCharIs buf (== char)

997 998 999 1000 1001
-- 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.
1002
isNormalComment :: AlexAccPred ExtsBitmap
1003
isNormalComment bits _ _ (AI _ buf)
1004
  | haddockEnabled bits = notFollowedByDocOrPragma
1005
  | otherwise           = nextCharIsNot buf (== '#')
1006 1007
  where
    notFollowedByDocOrPragma
1008
       = afterOptionalSpace buf (\b -> nextCharIsNot b (`elem` "|^*$#"))
1009

1010 1011 1012 1013 1014
afterOptionalSpace :: StringBuffer -> (StringBuffer -> Bool) -> Bool
afterOptionalSpace buf p
    = if nextCharIs buf (== ' ')
      then p (snd (nextChar buf))
      else p buf
1015

1016
atEOL :: AlexAccPred ExtsBitmap
1017
atEOL _ _ _ (AI _ buf) = atEnd buf || currentChar buf == '\n'
1018

1019
ifExtension :: (ExtsBitmap -> Bool) -> AlexAccPred ExtsBitmap
1020 1021
ifExtension pred bits _ _ _ = pred bits

1022 1023 1024
orExtensions :: (ExtsBitmap -> Bool) -> (ExtsBitmap -> Bool) -> AlexAccPred ExtsBitmap
orExtensions pred1 pred2 bits _ _ _ = pred1 bits || pred2 bits

1025 1026 1027
multiline_doc_comment :: Action
multiline_doc_comment span buf _len = withLexedDocType (worker "")
  where
1028
    worker commentAcc input docType checkNextLine = case alexGetChar' input of
Ian Lynagh's avatar
Ian Lynagh committed
1029
      Just ('\n', input')
1030 1031
        | checkNextLine -> case checkIfCommentLine input' of
          Just input -> worker ('\n':commentAcc) input docType checkNextLine
1032
          Nothing -> docCommentEnd input commentAcc docType buf span
1033 1034
        | otherwise -> docCommentEnd input commentAcc docType buf span
      Just (c, input) -> worker (c:commentAcc) input docType checkNextLine
1035
      Nothing -> docCommentEnd input commentAcc docType buf span
Ian Lynagh's avatar
Ian Lynagh committed
1036

1037 1038 1039 1040 1041 1042 1043 1044
    -- 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 "-- $", because that would be the
    --     start of a /new/ named haddock chunk (#10398).
    checkIfCommentLine :: AlexInput -> Maybe AlexInput
1045 1046
    checkIfCommentLine input = check (dropNonNewlineSpace input)
      where
1047 1048 1049 1050 1051 1052 1053 1054 1055 1056
        check input = do
          ('-', input) <- alexGetChar' input
          ('-', input) <- alexGetChar' input
          (c, after_c) <- alexGetChar' input
          case c of
            '-' -> Nothing
            ' ' -> case alexGetChar' after_c of
                     Just ('$', _) -> Nothing
                     _ -> Just input
            _   -> Just input
1057

1058
        dropNonNewlineSpace input = case alexGetChar' input of
Ian Lynagh's avatar
Ian Lynagh committed
1059
          Just (c, input')
1060 1061 1062 1063
            | isSpace c && c /= '\n' -> dropNonNewlineSpace input'
            | otherwise -> input
          Nothing -> input

Jedai's avatar
Jedai committed
1064 1065 1066 1067 1068
lineCommentToken :: Action
lineCommentToken span buf len = do
  b <- extension rawTokenStreamEnabled
  if b then strtoken ITlineComment span buf len else lexToken