Lexer.x 120 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
   P(..), ParseResult(..), mkParserFlags, ParserFlags(..), getRealSrcLoc,
52
   getPState, extopt, 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 59 60 61 62 63
   extension, bangPatEnabled, datatypeContextsEnabled,
   traditionalRecordSyntaxEnabled,
   explicitForallEnabled,
   inRulePrag,
   explicitNamespacesEnabled,
   patternSynonymsEnabled,
64 65
   sccProfilingOn, hpcEnabled,
   starIsTypeEnabled,
Jedai's avatar
Jedai committed
66
   addWarning,
Alan Zimmerman's avatar
Alan Zimmerman committed
67
   lexTokenStream,
68
   addAnnotation,AddAnn,addAnnsAt,mkParensApiAnn,
69
   commentToAnnotation
70 71
  ) where

72 73
import GhcPrelude

rodlogic's avatar
rodlogic committed
74 75
-- base
import Control.Monad
76
import Control.Monad.Fail as MonadFail
rodlogic's avatar
rodlogic committed
77
import Data.Bits
78
import Data.Char
rodlogic's avatar
rodlogic committed
79 80 81 82
import Data.List
import Data.Maybe
import Data.Word

83 84
import EnumSet (EnumSet)
import qualified EnumSet
85

86 87 88
-- ghc-boot
import qualified GHC.LanguageExtensions as LangExt

rodlogic's avatar
rodlogic committed
89 90 91 92 93 94 95 96
-- bytestring
import Data.ByteString (ByteString)

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

-- compiler/utils
97
import Bag
98 99 100 101
import Outputable
import StringBuffer
import FastString
import UniqFM
102
import Util             ( readRational, readHexRational )
rodlogic's avatar
rodlogic committed
103 104 105

-- compiler/main
import ErrUtils
106
import DynFlags
rodlogic's avatar
rodlogic committed
107 108 109

-- compiler/basicTypes
import SrcLoc
110
import Module
111 112
import BasicTypes     ( InlineSpec(..), RuleMatchInfo(..),
                        IntegralLit(..), FractionalLit(..),
Alan Zimmerman's avatar
Alan Zimmerman committed
113
                        SourceText(..) )
114

rodlogic's avatar
rodlogic committed
115 116
-- compiler/parser
import Ctype
Alan Zimmerman's avatar
Alan Zimmerman committed
117 118

import ApiAnnotation
119 120
}

121 122 123
-- -----------------------------------------------------------------------------
-- Alex "Character set macros"

124 125
-- NB: The logic behind these definitions is also reflected in basicTypes/Lexeme.hs
-- Any changes here should likely be reflected there.
126
$unispace    = \x05 -- Trick Alex into handling Unicode. See [Unicode in Alex].
127 128 129
$nl          = [\n\r\f]
$whitechar   = [$nl\v\ $unispace]
$white_no_nl = $whitechar # \n -- TODO #8424
130
$tab         = \t
131 132

$ascdigit  = 0-9
133
$unidigit  = \x03 -- Trick Alex into handling Unicode. See [Unicode in Alex].
134
$decdigit  = $ascdigit -- for now, should really be $digit (ToDo)
135 136 137
$digit     = [$ascdigit $unidigit]

$special   = [\(\)\,\;\[\]\`\{\}]
138
$ascsymbol = [\!\#\$\%\&\*\+\.\/\<\=\>\?\@\\\^\|\-\~\:]
139
$unisymbol = \x04 -- Trick Alex into handling Unicode. See [Unicode in Alex].
140
$symbol    = [$ascsymbol $unisymbol] # [$special \_\"\']
141

142
$unilarge  = \x01 -- Trick Alex into handling Unicode. See [Unicode in Alex].
143
$asclarge  = [A-Z]
144 145
$large     = [$asclarge $unilarge]

146
$unismall  = \x02 -- Trick Alex into handling Unicode. See [Unicode in Alex].
147
$ascsmall  = [a-z]
148 149
$small     = [$ascsmall $unismall \_]

150
$unigraphic = \x06 -- Trick Alex into handling Unicode. See [Unicode in Alex].
151
$graphic   = [$small $large $symbol $digit $special $unigraphic \"\']
152

153
$binit     = 0-1
Ian Lynagh's avatar
Ian Lynagh committed
154
$octit     = 0-7
155
$hexit     = [$decdigit A-F a-f]
156

157
$uniidchar = \x07 -- Trick Alex into handling Unicode. See [Unicode in Alex].
158
$idchar    = [$small $large $digit $uniidchar \']
159

160 161
$pragmachar = [$small $large $digit]

162 163
$docsym    = [\| \^ \* \$]

164

165 166 167 168 169 170 171 172
-- -----------------------------------------------------------------------------
-- 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
173

174 175 176 177 178 179 180 181
-- 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
182 183

@qual = (@conid \.)+
184 185 186 187
@qvarid = @qual @varid
@qconid = @qual @conid
@qvarsym = @qual @varsym
@qconsym = @qual @consym
188

189 190
@floating_point = @numspc @decimal \. @decimal @exponent? | @numspc @decimal @exponent
@hex_floating_point = @numspc @hexadecimal \. @hexadecimal @bin_exponent? | @numspc @hexadecimal @bin_exponent
191

192 193 194 195 196
-- normal signed numerical literals can only be explicitly negative,
-- not explicitly positive (contrast @exponent)
@negative = \-
@signed = @negative ?

197 198 199 200

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

201 202
haskell :-

203 204 205 206 207

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

-- everywhere: skip whitespace
Ian Lynagh's avatar
Ian Lynagh committed
208
$white_no_nl+ ;
209
$tab          { warnTab }
210 211 212 213 214

-- 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
215 216 217 218
-- are). We also rule out nested Haddock comments, if the -haddock flag is
-- set.

"{-" / { isNormalComment } { nested_comment lexToken }
219 220 221 222

-- 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.
223 224

-- Since Haddock comments aren't valid in every state, we need to rule them
Ian Lynagh's avatar
Ian Lynagh committed
225
-- out here.
226 227 228 229 230 231 232

-- 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.

233
"-- " ~$docsym .* { lineCommentToken }
234
"--" [^$symbol \ ] .* { lineCommentToken }
235 236 237

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

238
"-- " $docsym .* / { ifExtension (not . haddockEnabled) } { lineCommentToken }
239 240 241 242 243 244 245

-- 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.

246
"---"\-* ~$symbol .* { lineCommentToken }
247 248 249 250

-- 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
251
"--"\-* / { atEOL } { lineCommentToken }
252 253 254 255

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

Jedai's avatar
Jedai committed
256
"-- " / { atEOL } { lineCommentToken }
257 258 259 260 261 262 263 264 265 266 267

-- '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
268
  \n                                    ;
Adam Gundry's avatar
Adam Gundry committed
269 270
  ^\# line                              { begin line_prag1 }
  ^\# / { followedByDigit }             { begin line_prag1 }
Ian Lynagh's avatar
Ian Lynagh committed
271 272 273
  ^\# pragma .* \n                      ; -- GCC 3.3 CPP generated, apparently
  ^\# \! .* \n                          ; -- #!, for scripts
  ()                                    { do_bol }
274 275 276 277 278
}

-- 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.
279
<layout, layout_do, layout_if> {
Ian Lynagh's avatar
Ian Lynagh committed
280 281 282 283
  \{ / { notFollowedBy '-' }            { hopefully_open_brace }
        -- we might encounter {-# here, but {- has been handled already
  \n                                    ;
  ^\# (line)?                           { begin line_prag1 }
284 285
}

286 287
-- after an 'if', a vertical bar starts a layout context for MultiWayIf
<layout_if> {
288
  \| / { notFollowedBySymbol }          { new_layout_context True dontGenerateSemic ITvbar }
289 290 291
  ()                                    { pop }
}

292
-- do is treated in a subtly different way, see new_layout_context
293 294
<layout>    ()                          { new_layout_context True  generateSemic ITvocurly }
<layout_do> ()                          { new_layout_context False generateSemic ITvocurly }
295 296 297 298

-- 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
299
<layout_left>  ()                       { do_layout_left }
300

Ian Lynagh's avatar
Ian Lynagh committed
301
<0,option_prags> \n                     { begin bol }
302

303 304
"{-#" $whitechar* $pragmachar+ / { known_pragma linePrags }
                                { dispatch_pragmas linePrags }
305 306 307

-- single-line line pragmas, of the form
--    # <line> "<file>" <extra-stuff> \n
308 309 310 311 312
<line_prag1> {
  @decimal $white_no_nl+ \" [$graphic \ ]* \"  { setLineAndFile line_prag1a }
  ()                                           { failLinePrag1 }
}
<line_prag1a> .*                               { popLinePrag1 }
313 314 315

-- Haskell-style line pragmas, of the form
--    {-# LINE <line> "<file>" #-}
316 317 318 319
<line_prag2> {
  @decimal $white_no_nl+ \" [$graphic \ ]* \"  { setLineAndFile line_prag2a }
}
<line_prag2a> "#-}"|"-}"                       { pop }
320 321
   -- NOTE: accept -} at the end of a LINE pragma, for compatibility
   -- with older versions of GHC which generated these.
322

Rufflewind's avatar
Rufflewind committed
323 324 325 326
-- Haskell-style column pragmas, of the form
--    {-# COLUMN <column> #-}
<column_prag> @decimal $whitechar* "#-}" { setColumn }

327
<0,option_prags> {
Ian Lynagh's avatar
Ian Lynagh committed
328
  "{-#" $whitechar* $pragmachar+
329 330 331 332 333
        $whitechar+ $pragmachar+ / { known_pragma twoWordPrags }
                                 { dispatch_pragmas twoWordPrags }

  "{-#" $whitechar* $pragmachar+ / { known_pragma oneWordPrags }
                                 { dispatch_pragmas oneWordPrags }
334

335
  -- We ignore all these pragmas, but don't generate a warning for them
336 337
  "{-#" $whitechar* $pragmachar+ / { known_pragma ignoredPrags }
                                 { dispatch_pragmas ignoredPrags }
338 339

  -- ToDo: should only be valid inside a pragma:
Ian Lynagh's avatar
Ian Lynagh committed
340
  "#-}"                          { endPrag }
341 342
}

343
<option_prags> {
344 345
  "{-#"  $whitechar* $pragmachar+ / { known_pragma fileHeaderPrags }
                                   { dispatch_pragmas fileHeaderPrags }
David Waern's avatar
David Waern committed
346 347
}

348 349
<0> {
  -- In the "0" mode we ignore these pragmas
350
  "{-#"  $whitechar* $pragmachar+ / { known_pragma fileHeaderPrags }
351 352 353
                     { nested_comment lexToken }
}

354
<0,option_prags> {
355 356
  "{-#"  { warnThen Opt_WarnUnrecognisedPragmas (text "Unrecognised pragma")
                    (nested_comment lexToken) }
357 358
}

359 360
-- '0' state: ordinary lexemes

361 362
-- Haddock comments

363
<0,option_prags> {
364 365
  "-- " $docsym      / { ifExtension haddockEnabled } { multiline_doc_comment }
  "{-" \ ? $docsym   / { ifExtension haddockEnabled } { nested_doc_comment }
366 367
}

368 369
-- "special" symbols

370
<0> {
371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389
  "[|"        / { ifExtension thQuotesEnabled } { token (ITopenExpQuote NoE
                                                                NormalSyntax) }
  "[||"       / { ifExtension thQuotesEnabled } { token (ITopenTExpQuote NoE) }
  "[e|"       / { ifExtension thQuotesEnabled } { token (ITopenExpQuote HasE
                                                                NormalSyntax) }
  "[e||"      / { ifExtension thQuotesEnabled } { token (ITopenTExpQuote HasE) }
  "[p|"       / { ifExtension thQuotesEnabled } { token ITopenPatQuote }
  "[d|"       / { ifExtension thQuotesEnabled } { layout_token ITopenDecQuote }
  "[t|"       / { ifExtension thQuotesEnabled } { token ITopenTypQuote }
  "|]"        / { ifExtension thQuotesEnabled } { token (ITcloseQuote
                                                                NormalSyntax) }
  "||]"       / { ifExtension thQuotesEnabled } { token ITcloseTExpQuote }
  \$ @varid   / { ifExtension thEnabled } { skip_one_varid ITidEscape }
  "$$" @varid / { ifExtension thEnabled } { skip_two_varid ITidTyEscape }
  "$("        / { ifExtension thEnabled } { token ITparenEscape }
  "$$("       / { ifExtension thEnabled } { token ITparenTyEscape }

  "[" @varid "|"  / { ifExtension qqEnabled }
                     { lex_quasiquote_tok }
390 391

  -- qualified quasi-quote (#5555)
392 393
  "[" @qvarid "|"  / { ifExtension qqEnabled }
                     { lex_qquasiquote_tok }
394 395 396

  $unigraphic -- ⟦
    / { ifCurrentChar '⟦' `alexAndPred`
397
        ifExtension (\i -> unicodeSyntaxEnabled i && thQuotesEnabled i) }
398 399 400
    { token (ITopenExpQuote NoE UnicodeSyntax) }
  $unigraphic -- ⟧
    / { ifCurrentChar '⟧' `alexAndPred`
401
        ifExtension (\i -> unicodeSyntaxEnabled i && thQuotesEnabled i) }
402
    { token (ITcloseQuote UnicodeSyntax) }
403 404
}

eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
405 406 407 408
  -- See Note [Lexing type applications]
<0> {
    [^ $idchar \) ] ^
  "@"
409
    / { ifExtension typeApplicationEnabled `alexAndPred` notFollowedBySymbol }
eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
410 411 412
    { token ITtypeApp }
}

413
<0> {
414 415 416
  "(|" / { ifExtension arrowsEnabled `alexAndPred` notFollowedBySymbol }
                                        { special (IToparenbar NormalSyntax) }
  "|)" / { ifExtension arrowsEnabled }  { special (ITcparenbar NormalSyntax) }
417 418 419

  $unigraphic -- ⦇
    / { ifCurrentChar '⦇' `alexAndPred`
420
        ifExtension (\i -> unicodeSyntaxEnabled i && arrowsEnabled i) }
421 422 423
    { special (IToparenbar UnicodeSyntax) }
  $unigraphic -- ⦈
    / { ifCurrentChar '⦈' `alexAndPred`
424
        ifExtension (\i -> unicodeSyntaxEnabled i && arrowsEnabled i) }
425
    { special (ITcparenbar UnicodeSyntax) }
426 427
}

428
<0> {
429
  \? @varid / { ifExtension ipEnabled } { skip_one_varid ITdupipvarid }
430 431
}

Adam Gundry's avatar
Adam Gundry committed
432
<0> {
433 434
  "#" @varid / { ifExtension overloadedLabelsEnabled }
               { skip_one_varid ITlabelvarid }
Adam Gundry's avatar
Adam Gundry committed
435 436
}

437
<0> {
438
  "(#" / { orExtensions unboxedTuplesEnabled unboxedSumsEnabled }
Ian Lynagh's avatar
Ian Lynagh committed
439
         { token IToubxparen }
440
  "#)" / { orExtensions unboxedTuplesEnabled unboxedSumsEnabled }
Ian Lynagh's avatar
Ian Lynagh committed
441 442 443
         { token ITcubxparen }
}

444
<0,option_prags> {
Ian Lynagh's avatar
Ian Lynagh committed
445 446 447 448 449 450 451 452 453 454
  \(                                    { special IToparen }
  \)                                    { special ITcparen }
  \[                                    { special ITobrack }
  \]                                    { special ITcbrack }
  \,                                    { special ITcomma }
  \;                                    { special ITsemi }
  \`                                    { special ITbackquote }

  \{                                    { open_brace }
  \}                                    { close_brace }
455 456
}

457
<0,option_prags> {
458 459
  @qvarid                       { idtoken qvarid }
  @qconid                       { idtoken qconid }
Ian Lynagh's avatar
Ian Lynagh committed
460
  @varid                        { varid }
461
  @conid                        { idtoken conid }
462 463
}

464
<0> {
465 466 467 468
  @qvarid "#"+      / { ifExtension magicHashEnabled } { idtoken qvarid }
  @qconid "#"+      / { ifExtension magicHashEnabled } { idtoken qconid }
  @varid "#"+       / { ifExtension magicHashEnabled } { varid }
  @conid "#"+       / { ifExtension magicHashEnabled } { idtoken conid }
469 470
}

471 472
-- ToDo: - move `var` and (sym) into lexical syntax?
--       - remove backquote from $special?
473
<0> {
474 475
  @qvarsym                                         { idtoken qvarsym }
  @qconsym                                         { idtoken qconsym }
476 477
  @varsym                                          { varsym }
  @consym                                          { consym }
478 479
}

480 481
-- For the normal boxed literals we need to be careful
-- when trying to be close to Haskell98
482 483 484 485 486 487 488 489 490 491

-- 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.
--
492
<0> {
493
  -- Normal integral literals (:: Num a => a, from Integer)
494 495 496 497 498 499 500 501 502
  @decimal                                                               { tok_num positive 0 0 decimal }
  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 }
  @negative @decimal           / { ifExtension negativeLiteralsEnabled } { tok_num negative 1 1 decimal }
  @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 }
503 504

  -- Normal rational literals (:: Fractional a => a, from Rational)
505 506 507 508 509
  @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 }
510 511
}

512
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
513
  -- Unboxed ints (:: Int#) and words (:: Word#)
514 515
  -- It's simpler (and faster?) to give separate cases to the negatives,
  -- especially considering octal/hexadecimal prefixes.
516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531
  @decimal                     \# / { ifExtension magicHashEnabled } { tok_primint positive 0 1 decimal }
  0[bB] @numspc @binary        \# / { ifExtension magicHashEnabled `alexAndPred`
                                      ifExtension binaryLiteralsEnabled } { tok_primint positive 2 3 binary }
  0[oO] @numspc @octal         \# / { ifExtension magicHashEnabled } { tok_primint positive 2 3 octal }
  0[xX] @numspc @hexadecimal   \# / { ifExtension magicHashEnabled } { tok_primint positive 2 3 hexadecimal }
  @negative @decimal           \# / { ifExtension magicHashEnabled } { tok_primint negative 1 2 decimal }
  @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 }

  @decimal                     \# \# / { ifExtension magicHashEnabled } { tok_primword 0 2 decimal }
  0[bB] @numspc @binary        \# \# / { ifExtension magicHashEnabled `alexAndPred`
                                         ifExtension binaryLiteralsEnabled } { tok_primword 2 4 binary }
  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
532

533 534
  -- Unboxed floats and doubles (:: Float#, :: Double#)
  -- prim_{float,double} work with signed literals
535 536
  @signed @floating_point \# / { ifExtension magicHashEnabled } { tok_frac 1 tok_primfloat }
  @signed @floating_point \# \# / { ifExtension magicHashEnabled } { tok_frac 2 tok_primdouble }
537 538 539 540 541 542
}

-- 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.
543
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
544 545
  \'                            { lex_char_tok }
  \"                            { lex_string_tok }
546 547
}

eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
548 549 550 551 552 553 554 555 556 557 558
-- 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
559
-- Disambiguation is accomplished by requiring *something* to appear between
eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
560 561 562 563 564 565 566 567 568 569 570 571 572 573
-- 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.
574 575 576 577

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

578
{
579

580 581 582
-- -----------------------------------------------------------------------------
-- The token type

583
data Token
Ian Lynagh's avatar
Ian Lynagh committed
584
  = ITas                        -- Haskell keywords
585 586 587 588 589 590 591 592
  | ITcase
  | ITclass
  | ITdata
  | ITdefault
  | ITderiving
  | ITdo
  | ITelse
  | IThiding
593
  | ITforeign
594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609
  | ITif
  | ITimport
  | ITin
  | ITinfix
  | ITinfixl
  | ITinfixr
  | ITinstance
  | ITlet
  | ITmodule
  | ITnewtype
  | ITof
  | ITqualified
  | ITthen
  | ITtype
  | ITwhere

610
  | ITforall            IsUnicodeSyntax -- GHC extension keywords
611 612 613 614
  | ITexport
  | ITlabel
  | ITdynamic
  | ITsafe
615
  | ITinterruptible
616 617 618
  | ITunsafe
  | ITstdcallconv
  | ITccallconv
619
  | ITcapiconv
620
  | ITprimcallconv
thoughtpolice's avatar
thoughtpolice committed
621
  | ITjavascriptcallconv
622
  | ITmdo
623
  | ITfamily
624
  | ITrole
625 626 627
  | ITgroup
  | ITby
  | ITusing
cactus's avatar
cactus committed
628
  | ITpattern
Facundo Domínguez's avatar
Facundo Domínguez committed
629
  | ITstatic
Ryan Scott's avatar
Ryan Scott committed
630 631
  | ITstock
  | ITanyclass
Ryan Scott's avatar
Ryan Scott committed
632
  | ITvia
633

Edward Z. Yang's avatar
Edward Z. Yang committed
634 635 636 637 638 639
  -- Backpack tokens
  | ITunit
  | ITsignature
  | ITdependency
  | ITrequires

Alan Zimmerman's avatar
Alan Zimmerman committed
640 641 642 643 644 645 646 647
  -- 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
648 649
  | 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
650 651 652 653 654 655
  | ITscc_prag          SourceText
  | ITgenerated_prag    SourceText
  | ITcore_prag         SourceText         -- hdaume: core annotations
  | ITunpack_prag       SourceText
  | ITnounpack_prag     SourceText
  | ITann_prag          SourceText
656
  | ITcomplete_prag     SourceText
657
  | ITclose_prag
658 659 660
  | IToptions_prag String
  | ITinclude_prag String
  | ITlanguage_prag
Alan Zimmerman's avatar
Alan Zimmerman committed
661 662 663 664 665 666
  | 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
667
  | ITcomment_line_prag         -- See Note [Nested comment line pragmas]
668

Ian Lynagh's avatar
Ian Lynagh committed
669
  | ITdotdot                    -- reserved symbols
670
  | ITcolon
671
  | ITdcolon            IsUnicodeSyntax
672 673
  | ITequal
  | ITlam
674
  | ITlcase
675
  | ITvbar
676 677
  | ITlarrow            IsUnicodeSyntax
  | ITrarrow            IsUnicodeSyntax
678 679
  | ITat
  | ITtilde
680
  | ITdarrow            IsUnicodeSyntax
681 682
  | ITminus
  | ITbang
683
  | ITstar              IsUnicodeSyntax
684 685
  | 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
755 756 757 758 759 760 761 762
  | 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
763 764 765 766
  -- due to checking for preceding whitespace)
  | ITtypeApp


767 768
  | ITunknown String            -- Used when the lexer can't make sense of it
  | ITeof                       -- end of file token
769 770

  -- Documentation annotations
771 772 773 774 775 776 777
  | 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 {- -}
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),
829
         ( "static",         ITstatic,        0 ),
Ryan Scott's avatar
Ryan Scott committed
830 831
         ( "stock",          ITstock,         0 ),
         ( "anyclass",       ITanyclass,      0 ),
Ryan Scott's avatar
Ryan Scott committed
832
         ( "via",            ITvia,           0 ),
833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850
         ( "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
851 852 853 854
         ( "unit",           ITunit,          0 ),
         ( "dependency",     ITdependency,       0 ),
         ( "signature",      ITsignature,     0 ),

855 856 857
         ( "rec",            ITrec,           xbit ArrowsBit .|.
                                              xbit RecursiveDoBit),
         ( "proc",           ITproc,          xbit ArrowsBit)
858 859
     ]

860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875
{-----------------------------------
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.
-------------------------------------}
876

877
reservedSymsFM :: UniqFM (Token, ExtsBitmap -> Bool)
878
reservedSymsFM = listToUFM $
879 880
    map (\ (x,y,z) -> (mkFastString x,(y,z)))
      [ ("..",  ITdotdot,              always)
Ian Lynagh's avatar
Ian Lynagh committed
881
        -- (:) is a reserved op, meaning only list cons
882 883 884 885 886 887 888 889 890 891 892 893 894 895
       ,(":",   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)

       ,("*", ITstar NormalSyntax, starIsTypeEnabled)
896

Ian Lynagh's avatar
Ian Lynagh committed
897
        -- For 'forall a . t'
898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920
       ,(".", ITdot,  always) -- \i -> explicitForallEnabled i || inRulePrag i)

       ,("-<",  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)
       ,("★",   ITstar UnicodeSyntax,
                  \i -> unicodeSyntaxEnabled i && starIsTypeEnabled i)
921

Simon Marlow's avatar
Simon Marlow committed
922 923 924
        -- 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).
925 926 927 928 929
       ]

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

Ian Lynagh's avatar
Ian Lynagh committed
930
type Action = RealSrcSpan -> StringBuffer -> Int -> P (RealLocated Token)
931

932
special :: Token -> Action
twanvl's avatar
twanvl committed
933
special tok span _buf _len = return (L span tok)
934

935
token, layout_token :: Token -> Action
twanvl's avatar
twanvl committed
936 937
token t span _buf _len = return (L span t)
layout_token t span _buf _len = pushLexState layout >> return (L span t)
938

939 940
idtoken :: (StringBuffer -> Int -> Token) -> Action
idtoken f span buf len = return (L span $! (f buf len))
941

942
skip_one_varid :: (FastString -> Token) -> Action
Ian Lynagh's avatar
Ian Lynagh committed
943
skip_one_varid f span buf len
944
  = return (L span $! f (lexemeToFastString (stepOn buf) (len-1)))
945

946 947 948 949
skip_two_varid :: (FastString -> Token) -> Action
skip_two_varid f span buf len
  = return (L span $! f (lexemeToFastString (stepOn (stepOn buf)) (len-2)))

950
strtoken :: (String -> Token) -> Action
Ian Lynagh's avatar
Ian Lynagh committed
951
strtoken f span buf len =
952
  return (L span $! (f $! lexemeToString buf len))
953 954

begin :: Int -> Action
955
begin code _span _str _len = do pushLexState code; lexToken
956 957

pop :: Action
Ian Lynagh's avatar
Ian Lynagh committed
958 959
pop _span _buf _len = do _ <- popLexState
                         lexToken
960 961 962
-- See Note [Nested comment line pragmas]
failLinePrag1 :: Action
failLinePrag1 span _buf _len = do
963
  b <- extension inNestedComment
964 965 966 967 968 969
  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
970
  b <- extension inNestedComment
971 972 973
  if b then return (L span ITcomment_line_prag) else do
    _ <- popLexState
    lexToken
974

975 976
hopefully_open_brace :: Action
hopefully_open_brace span buf len
977
 = do relaxed <- extension relaxedLayout
978 979 980 981 982
      ctx <- getContext
      (AI l _) <- getInput
      let offset = srcLocCol l
          isOK = relaxed ||
                 case ctx of
983 984
                 Layout prev_off _ : _ -> prev_off < offset
                 _                     -> True
985
      if isOK then pop_and open_brace span buf len
Ian Lynagh's avatar
Ian Lynagh committed
986
              else failSpanMsgP (RealSrcSpan span) (text "Missing block")
987

988
pop_and :: Action -> Action
Ian Lynagh's avatar
Ian Lynagh committed
989 990
pop_and act span buf len = do _ <- popLexState
                              act span buf len
991

992
{-# INLINE nextCharIs #-}
Ian Lynagh's avatar
Ian Lynagh committed
993
nextCharIs :: StringBuffer -> (Char -> Bool) -> Bool
994 995
nextCharIs buf p = not (atEnd buf) && p (currentChar buf)

996 997 998 999
{-# INLINE nextCharIsNot #-}
nextCharIsNot :: StringBuffer -> (Char -> Bool) -> Bool
nextCharIsNot buf p = not (nextCharIs buf p)

1000
notFollowedBy :: Char -> AlexAccPred ExtsBitmap
Ian Lynagh's avatar
Ian Lynagh committed
1001
notFollowedBy char _ _ _ (AI _ buf)
1002
  = nextCharIsNot buf (== char)
1003

1004
notFollowedBySymbol :: AlexAccPred ExtsBitmap
1005
notFollowedBySymbol _ _ _ (AI _ buf)
1006
  = nextCharIsNot buf (`elem` "!#$%&*+./<=>?@\\^|-~")
ross's avatar
ross committed
1007

Adam Gundry's avatar
Adam Gundry committed
1008 1009 1010 1011
followedByDigit :: AlexAccPred ExtsBitmap
followedByDigit _ _ _ (AI _ buf)
  = afterOptionalSpace buf (\b -> nextCharIs b (`elem` ['0'..'9']))

1012 1013 1014 1015
ifCurrentChar :: Char -> AlexAccPred ExtsBitmap
ifCurrentChar char _ (AI _ buf) _ _
  = nextCharIs buf (== char)

1016 1017 1018 1019 1020
-- 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.
1021
isNormalComment :: AlexAccPred ExtsBitmap
1022
isNormalComment bits _ _ (AI _ buf)
1023 1024
  | haddockEnabled bits = notFollowedByDocOrPragma
  | otherwise           = nextCharIsNot buf (== '#')
1025 1026
  where
    notFollowedByDocOrPragma
1027
       = afterOptionalSpace buf (\b -> nextCharIsNot b (`elem` "|^*$#"))
1028

1029 1030 1031 1032 1033
afterOptionalSpace :: StringBuffer -> (StringBuffer -> Bool) -> Bool
afterOptionalSpace buf p
    = if nextCharIs buf (== ' ')
      then p (snd (nextChar buf))
      else p buf
1034

1035
atEOL :: AlexAccPred ExtsBitmap
1036
atEOL _ _ _ (AI _ buf) = atEnd buf || currentChar buf == '\n'
1037

1038 1039
ifExtension :: (ExtsBitmap -> Bool) -> AlexAccPred ExtsBitmap
ifExtension pred bits _ _ _ = pred bits
1040

1041 1042
orExtensions :: (ExtsBitmap -> Bool) -> (ExtsBitmap -> Bool) -> AlexAccPred ExtsBitmap
orExtensions pred1 pred2 bits _ _ _ = pred1 bits || pred2 bits
1043

1044 1045 1046
multiline_doc_comment :: Action
multiline_doc_comment span buf _len = withLexedDocType (worker "")
  where
1047
    worker commentAcc input docType checkNextLine = case alexGetChar' input of
Ian Lynagh's avatar
Ian Lynagh committed
1048
      Just ('\n', input')
1049 1050
        | checkNextLine -> case checkIfCommentLine input' of
          Just input -> worker ('\n':commentAcc) input docType checkNextLine
1051
          Nothing -> docCommentEnd input commentAcc docType buf span
1052 1053
        | otherwise -> docCommentEnd input commentAcc docType buf span
      Just (c, input) -> worker (c:commentAcc) input docType checkNextLine
1054
      Nothing -> docCommentEnd input commentAcc docType buf span
Ian Lynagh's avatar
Ian Lynagh committed
1055

1056 1057 1058 1059 1060 1061 1062 1063
    -- 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 "-- $", because that would be the
    --     start of a /new/ named haddock chunk (#10398).
    checkIfCommentLine :: AlexInput -> Maybe AlexInput
1064 1065
    checkIfCommentLine input = check (dropNonNewlineSpace input)
      where
1066 1067 1068 1069 1070 1071 1072 1073 1074 1075
        check input = do
          ('-', input) <- alexGetChar' input
          ('-', input) <- alexGetChar' input
          (c, after_c) <- alexGetChar' input
          case c of
            '-' -> Nothing
            ' ' -> case alexGetChar' after_c of
                     Just ('$', _) -> Nothing
                     _ -> Just input
            _   -> Just input
1076

1077
        dropNonNewlineSpace input = case alexGetChar' input of
Ian Lynagh's avatar
Ian Lynagh committed
1078
          Just (c, input')
1079 1080 1081 1082
            | isSpace c && c /= '\n' -> dropNonNewlineSpace input'
            | otherwise -> input
          Nothing -> input

Jedai's avatar
Jedai committed
1083 1084
lineCommentToken :: Action
lineCommentToken span buf len = do
1085
  b <- extension rawTokenStreamEnabled
Jedai's avatar
Jedai committed
1086 1087
  if b then strtoken ITlineComment span buf len else lexToken

1088 1089 1090 1091
{-
  nested comments require traversing by hand, they can't be parsed
  using regular expressions.
-}
Ian Lynagh's avatar
Ian Lynagh committed
1092
nested_comment :: P (RealLocated Token) -> Action
Alan Zimmerman's avatar