Lexer.x 118 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,
71
   starIsTypeEnabled,
Jedai's avatar
Jedai committed
72
   addWarning,
Alan Zimmerman's avatar
Alan Zimmerman committed
73
   lexTokenStream,
74
   addAnnotation,AddAnn,addAnnsAt,mkParensApiAnn,
75
   commentToAnnotation,
76
   moveAnnotations
77 78
  ) where

79 80
import GhcPrelude

rodlogic's avatar
rodlogic committed
81 82
-- base
import Control.Monad
83
import Control.Monad.Fail as MonadFail
rodlogic's avatar
rodlogic committed
84
import Data.Bits
85
import Data.Char
rodlogic's avatar
rodlogic committed
86 87 88 89
import Data.List
import Data.Maybe
import Data.Word

90 91
import EnumSet (EnumSet)
import qualified EnumSet
92

93 94 95
-- ghc-boot
import qualified GHC.LanguageExtensions as LangExt

rodlogic's avatar
rodlogic committed
96 97 98 99 100 101 102 103
-- bytestring
import Data.ByteString (ByteString)

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

-- compiler/utils
104
import Bag
105 106 107 108
import Outputable
import StringBuffer
import FastString
import UniqFM
109
import Util             ( readRational, readHexRational )
rodlogic's avatar
rodlogic committed
110 111 112

-- compiler/main
import ErrUtils
113
import DynFlags
rodlogic's avatar
rodlogic committed
114 115 116

-- compiler/basicTypes
import SrcLoc
117
import Module
118 119
import BasicTypes     ( InlineSpec(..), RuleMatchInfo(..),
                        IntegralLit(..), FractionalLit(..),
Alan Zimmerman's avatar
Alan Zimmerman committed
120
                        SourceText(..) )
121

rodlogic's avatar
rodlogic committed
122 123
-- compiler/parser
import Ctype
Alan Zimmerman's avatar
Alan Zimmerman committed
124 125

import ApiAnnotation
126 127
}

128 129 130
-- -----------------------------------------------------------------------------
-- Alex "Character set macros"

131 132
-- NB: The logic behind these definitions is also reflected in basicTypes/Lexeme.hs
-- Any changes here should likely be reflected there.
133
$unispace    = \x05 -- Trick Alex into handling Unicode. See [Unicode in Alex].
134 135 136
$nl          = [\n\r\f]
$whitechar   = [$nl\v\ $unispace]
$white_no_nl = $whitechar # \n -- TODO #8424
137
$tab         = \t
138 139

$ascdigit  = 0-9
140
$unidigit  = \x03 -- Trick Alex into handling Unicode. See [Unicode in Alex].
141
$decdigit  = $ascdigit -- for now, should really be $digit (ToDo)
142 143 144
$digit     = [$ascdigit $unidigit]

$special   = [\(\)\,\;\[\]\`\{\}]
145
$ascsymbol = [\!\#\$\%\&\*\+\.\/\<\=\>\?\@\\\^\|\-\~\:]
146
$unisymbol = \x04 -- Trick Alex into handling Unicode. See [Unicode in Alex].
147
$symbol    = [$ascsymbol $unisymbol] # [$special \_\"\']
148

149
$unilarge  = \x01 -- Trick Alex into handling Unicode. See [Unicode in Alex].
150
$asclarge  = [A-Z]
151 152
$large     = [$asclarge $unilarge]

153
$unismall  = \x02 -- Trick Alex into handling Unicode. See [Unicode in Alex].
154
$ascsmall  = [a-z]
155 156
$small     = [$ascsmall $unismall \_]

157
$unigraphic = \x06 -- Trick Alex into handling Unicode. See [Unicode in Alex].
158
$graphic   = [$small $large $symbol $digit $special $unigraphic \"\']
159

160
$binit     = 0-1
Ian Lynagh's avatar
Ian Lynagh committed
161
$octit     = 0-7
162
$hexit     = [$decdigit A-F a-f]
163

164
$uniidchar = \x07 -- Trick Alex into handling Unicode. See [Unicode in Alex].
165
$idchar    = [$small $large $digit $uniidchar \']
166

167 168
$pragmachar = [$small $large $digit]

169 170
$docsym    = [\| \^ \* \$]

171

172 173 174 175 176 177 178 179
-- -----------------------------------------------------------------------------
-- 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
180

181 182 183 184 185 186 187 188
-- 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
189 190

@qual = (@conid \.)+
191 192 193 194
@qvarid = @qual @varid
@qconid = @qual @conid
@qvarsym = @qual @varsym
@qconsym = @qual @consym
195

196 197
@floating_point = @numspc @decimal \. @decimal @exponent? | @numspc @decimal @exponent
@hex_floating_point = @numspc @hexadecimal \. @hexadecimal @bin_exponent? | @numspc @hexadecimal @bin_exponent
198

199 200 201 202 203
-- normal signed numerical literals can only be explicitly negative,
-- not explicitly positive (contrast @exponent)
@negative = \-
@signed = @negative ?

204 205 206 207

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

208 209
haskell :-

210 211 212 213 214

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

-- everywhere: skip whitespace
Ian Lynagh's avatar
Ian Lynagh committed
215
$white_no_nl+ ;
216
$tab          { warnTab }
217 218 219 220 221

-- 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
222 223 224 225
-- are). We also rule out nested Haddock comments, if the -haddock flag is
-- set.

"{-" / { isNormalComment } { nested_comment lexToken }
226 227 228 229

-- 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.
230 231

-- Since Haddock comments aren't valid in every state, we need to rule them
Ian Lynagh's avatar
Ian Lynagh committed
232
-- out here.
233 234 235 236 237 238 239

-- 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.

240
"-- " ~$docsym .* { lineCommentToken }
241
"--" [^$symbol \ ] .* { lineCommentToken }
242 243 244

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

245
"-- " $docsym .* / { ifExtension (not . haddockEnabled) } { lineCommentToken }
246 247 248 249 250 251 252

-- 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.

253
"---"\-* ~$symbol .* { lineCommentToken }
254 255 256 257

-- 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
258
"--"\-* / { atEOL } { lineCommentToken }
259 260 261 262

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

Jedai's avatar
Jedai committed
263
"-- " / { atEOL } { lineCommentToken }
264 265 266 267 268 269 270 271 272 273 274

-- '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
275
  \n                                    ;
Adam Gundry's avatar
Adam Gundry committed
276 277
  ^\# line                              { begin line_prag1 }
  ^\# / { followedByDigit }             { begin line_prag1 }
Ian Lynagh's avatar
Ian Lynagh committed
278 279 280
  ^\# pragma .* \n                      ; -- GCC 3.3 CPP generated, apparently
  ^\# \! .* \n                          ; -- #!, for scripts
  ()                                    { do_bol }
281 282 283 284 285
}

-- 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.
286
<layout, layout_do, layout_if> {
Ian Lynagh's avatar
Ian Lynagh committed
287 288 289 290
  \{ / { notFollowedBy '-' }            { hopefully_open_brace }
        -- we might encounter {-# here, but {- has been handled already
  \n                                    ;
  ^\# (line)?                           { begin line_prag1 }
291 292
}

293 294
-- after an 'if', a vertical bar starts a layout context for MultiWayIf
<layout_if> {
295
  \| / { notFollowedBySymbol }          { new_layout_context True dontGenerateSemic ITvbar }
296 297 298
  ()                                    { pop }
}

299
-- do is treated in a subtly different way, see new_layout_context
300 301
<layout>    ()                          { new_layout_context True  generateSemic ITvocurly }
<layout_do> ()                          { new_layout_context False generateSemic ITvocurly }
302 303 304 305

-- 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
306
<layout_left>  ()                       { do_layout_left }
307

Ian Lynagh's avatar
Ian Lynagh committed
308
<0,option_prags> \n                     { begin bol }
309

310 311
"{-#" $whitechar* $pragmachar+ / { known_pragma linePrags }
                                { dispatch_pragmas linePrags }
312 313 314

-- single-line line pragmas, of the form
--    # <line> "<file>" <extra-stuff> \n
315
<line_prag1> @decimal                   { setLine line_prag1a }
Ian Lynagh's avatar
Ian Lynagh committed
316 317
<line_prag1a> \" [$graphic \ ]* \"      { setFile line_prag1b }
<line_prag1b> .*                        { pop }
318 319 320

-- Haskell-style line pragmas, of the form
--    {-# LINE <line> "<file>" #-}
321
<line_prag2> @decimal                   { setLine line_prag2a }
Ian Lynagh's avatar
Ian Lynagh committed
322 323
<line_prag2a> \" [$graphic \ ]* \"      { setFile line_prag2b }
<line_prag2b> "#-}"|"-}"                { pop }
324 325
   -- NOTE: accept -} at the end of a LINE pragma, for compatibility
   -- with older versions of GHC which generated these.
326

Rufflewind's avatar
Rufflewind committed
327 328 329 330
-- Haskell-style column pragmas, of the form
--    {-# COLUMN <column> #-}
<column_prag> @decimal $whitechar* "#-}" { setColumn }

331
<0,option_prags> {
Ian Lynagh's avatar
Ian Lynagh committed
332
  "{-#" $whitechar* $pragmachar+
333 334 335 336 337
        $whitechar+ $pragmachar+ / { known_pragma twoWordPrags }
                                 { dispatch_pragmas twoWordPrags }

  "{-#" $whitechar* $pragmachar+ / { known_pragma oneWordPrags }
                                 { dispatch_pragmas oneWordPrags }
338

339
  -- We ignore all these pragmas, but don't generate a warning for them
340 341
  "{-#" $whitechar* $pragmachar+ / { known_pragma ignoredPrags }
                                 { dispatch_pragmas ignoredPrags }
342 343

  -- ToDo: should only be valid inside a pragma:
Ian Lynagh's avatar
Ian Lynagh committed
344
  "#-}"                          { endPrag }
345 346
}

347
<option_prags> {
348 349
  "{-#"  $whitechar* $pragmachar+ / { known_pragma fileHeaderPrags }
                                   { dispatch_pragmas fileHeaderPrags }
David Waern's avatar
David Waern committed
350 351
}

352 353
<0> {
  -- In the "0" mode we ignore these pragmas
354
  "{-#"  $whitechar* $pragmachar+ / { known_pragma fileHeaderPrags }
355 356 357
                     { nested_comment lexToken }
}

358
<0,option_prags> {
359 360
  "{-#"  { warnThen Opt_WarnUnrecognisedPragmas (text "Unrecognised pragma")
                    (nested_comment lexToken) }
361 362
}

363 364
-- '0' state: ordinary lexemes

365 366
-- Haddock comments

367
<0,option_prags> {
David Waern's avatar
David Waern committed
368 369
  "-- " $docsym      / { ifExtension haddockEnabled } { multiline_doc_comment }
  "{-" \ ? $docsym   / { ifExtension haddockEnabled } { nested_doc_comment }
370 371
}

372 373
-- "special" symbols

374
<0> {
375 376
  "[|"        / { ifExtension thQuotesEnabled } { token (ITopenExpQuote NoE
                                                                NormalSyntax) }
377
  "[||"       / { ifExtension thQuotesEnabled } { token (ITopenTExpQuote NoE) }
378 379
  "[e|"       / { ifExtension thQuotesEnabled } { token (ITopenExpQuote HasE
                                                                NormalSyntax) }
380 381 382 383
  "[e||"      / { ifExtension thQuotesEnabled } { token (ITopenTExpQuote HasE) }
  "[p|"       / { ifExtension thQuotesEnabled } { token ITopenPatQuote }
  "[d|"       / { ifExtension thQuotesEnabled } { layout_token ITopenDecQuote }
  "[t|"       / { ifExtension thQuotesEnabled } { token ITopenTypQuote }
384 385
  "|]"        / { ifExtension thQuotesEnabled } { token (ITcloseQuote
                                                                NormalSyntax) }
386
  "||]"       / { ifExtension thQuotesEnabled } { token ITcloseTExpQuote }
387 388 389 390
  \$ @varid   / { ifExtension thEnabled } { skip_one_varid ITidEscape }
  "$$" @varid / { ifExtension thEnabled } { skip_two_varid ITidTyEscape }
  "$("        / { ifExtension thEnabled } { token ITparenEscape }
  "$$("       / { ifExtension thEnabled } { token ITparenTyEscape }
391

392
  "[" @varid "|"  / { ifExtension qqEnabled }
393
                     { lex_quasiquote_tok }
394 395

  -- qualified quasi-quote (#5555)
396 397
  "[" @qvarid "|"  / { ifExtension qqEnabled }
                     { lex_qquasiquote_tok }
398 399 400 401 402 403 404 405 406

  $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) }
407 408
}

eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
409 410 411 412 413 414 415 416
  -- See Note [Lexing type applications]
<0> {
    [^ $idchar \) ] ^
  "@"
    / { ifExtension typeApplicationEnabled `alexAndPred` notFollowedBySymbol }
    { token ITtypeApp }
}

417
<0> {
ross's avatar
ross committed
418
  "(|" / { ifExtension arrowsEnabled `alexAndPred` notFollowedBySymbol }
419 420 421 422 423 424 425 426 427 428 429
                                        { 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) }
430 431
}

432
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
433
  \? @varid / { ifExtension ipEnabled } { skip_one_varid ITdupipvarid }
434 435
}

Adam Gundry's avatar
Adam Gundry committed
436 437 438 439 440
<0> {
  "#" @varid / { ifExtension overloadedLabelsEnabled }
               { skip_one_varid ITlabelvarid }
}

441
<0> {
442
  "(#" / { orExtensions unboxedTuplesEnabled unboxedSumsEnabled }
Ian Lynagh's avatar
Ian Lynagh committed
443
         { token IToubxparen }
444
  "#)" / { orExtensions unboxedTuplesEnabled unboxedSumsEnabled }
Ian Lynagh's avatar
Ian Lynagh committed
445 446 447
         { token ITcubxparen }
}

448
<0,option_prags> {
Ian Lynagh's avatar
Ian Lynagh committed
449 450 451 452 453 454 455 456 457 458
  \(                                    { special IToparen }
  \)                                    { special ITcparen }
  \[                                    { special ITobrack }
  \]                                    { special ITcbrack }
  \,                                    { special ITcomma }
  \;                                    { special ITsemi }
  \`                                    { special ITbackquote }

  \{                                    { open_brace }
  \}                                    { close_brace }
459 460
}

461
<0,option_prags> {
462 463
  @qvarid                       { idtoken qvarid }
  @qconid                       { idtoken qconid }
Ian Lynagh's avatar
Ian Lynagh committed
464
  @varid                        { varid }
465
  @conid                        { idtoken conid }
466 467
}

468
<0> {
469 470
  @qvarid "#"+      / { ifExtension magicHashEnabled } { idtoken qvarid }
  @qconid "#"+      / { ifExtension magicHashEnabled } { idtoken qconid }
471
  @varid "#"+       / { ifExtension magicHashEnabled } { varid }
472
  @conid "#"+       / { ifExtension magicHashEnabled } { idtoken conid }
473 474
}

475 476
-- ToDo: - move `var` and (sym) into lexical syntax?
--       - remove backquote from $special?
477
<0> {
478 479
  @qvarsym                                         { idtoken qvarsym }
  @qconsym                                         { idtoken qconsym }
480 481
  @varsym                                          { varsym }
  @consym                                          { consym }
482 483
}

484 485
-- For the normal boxed literals we need to be careful
-- when trying to be close to Haskell98
486 487 488 489 490 491 492 493 494 495

-- 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.
--
496
<0> {
497
  -- Normal integral literals (:: Num a => a, from Integer)
498
  @decimal                                                               { tok_num positive 0 0 decimal }
499 500 501
  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 }
502
  @negative @decimal           / { ifExtension negativeLiteralsEnabled } { tok_num negative 1 1 decimal }
503 504 505 506
  @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 }
507 508

  -- Normal rational literals (:: Fractional a => a, from Rational)
509 510 511 512 513
  @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 }
514 515
}

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

Ian Lynagh's avatar
Ian Lynagh committed
531
  @decimal                     \# \# / { ifExtension magicHashEnabled } { tok_primword 0 2 decimal }
532
  0[bB] @numspc @binary        \# \# / { ifExtension magicHashEnabled `alexAndPred`
533
                                         ifExtension binaryLiteralsEnabled } { tok_primword 2 4 binary }
534 535
  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
536

537 538
  -- Unboxed floats and doubles (:: Float#, :: Double#)
  -- prim_{float,double} work with signed literals
539 540
  @signed @floating_point \# / { ifExtension magicHashEnabled } { tok_frac 1 tok_primfloat }
  @signed @floating_point \# \# / { ifExtension magicHashEnabled } { tok_frac 2 tok_primdouble }
541 542 543 544 545 546
}

-- 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.
547
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
548 549
  \'                            { lex_char_tok }
  \"                            { lex_string_tok }
550 551
}

eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
552 553 554 555 556 557 558 559 560 561 562
-- 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
563
-- Disambiguation is accomplished by requiring *something* to appear between
eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
564 565 566 567 568 569 570 571 572 573 574 575 576 577
-- 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.
578 579 580 581

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

582
{
583

584 585 586
-- -----------------------------------------------------------------------------
-- The token type

587
data Token
Ian Lynagh's avatar
Ian Lynagh committed
588
  = ITas                        -- Haskell keywords
589 590 591 592 593 594 595 596
  | ITcase
  | ITclass
  | ITdata
  | ITdefault
  | ITderiving
  | ITdo
  | ITelse
  | IThiding
597
  | ITforeign
598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613
  | ITif
  | ITimport
  | ITin
  | ITinfix
  | ITinfixl
  | ITinfixr
  | ITinstance
  | ITlet
  | ITmodule
  | ITnewtype
  | ITof
  | ITqualified
  | ITthen
  | ITtype
  | ITwhere

614
  | ITforall            IsUnicodeSyntax -- GHC extension keywords
615 616 617 618
  | ITexport
  | ITlabel
  | ITdynamic
  | ITsafe
619
  | ITinterruptible
620 621 622
  | ITunsafe
  | ITstdcallconv
  | ITccallconv
623
  | ITcapiconv
624
  | ITprimcallconv
thoughtpolice's avatar
thoughtpolice committed
625
  | ITjavascriptcallconv
626
  | ITmdo
627
  | ITfamily
628
  | ITrole
629 630 631
  | ITgroup
  | ITby
  | ITusing
cactus's avatar
cactus committed
632
  | ITpattern
Facundo Domínguez's avatar
Facundo Domínguez committed
633
  | ITstatic
Ryan Scott's avatar
Ryan Scott committed
634 635
  | ITstock
  | ITanyclass
Ryan Scott's avatar
Ryan Scott committed
636
  | ITvia
637

Edward Z. Yang's avatar
Edward Z. Yang committed
638 639 640 641 642 643
  -- Backpack tokens
  | ITunit
  | ITsignature
  | ITdependency
  | ITrequires

Alan Zimmerman's avatar
Alan Zimmerman committed
644 645 646 647 648 649 650 651
  -- 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
652 653
  | 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
654 655 656 657 658 659
  | ITscc_prag          SourceText
  | ITgenerated_prag    SourceText
  | ITcore_prag         SourceText         -- hdaume: core annotations
  | ITunpack_prag       SourceText
  | ITnounpack_prag     SourceText
  | ITann_prag          SourceText
660
  | ITcomplete_prag     SourceText
661
  | ITclose_prag
662 663 664
  | IToptions_prag String
  | ITinclude_prag String
  | ITlanguage_prag
Alan Zimmerman's avatar
Alan Zimmerman committed
665 666 667 668 669 670
  | 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
671

Ian Lynagh's avatar
Ian Lynagh committed
672
  | ITdotdot                    -- reserved symbols
673
  | ITcolon
674
  | ITdcolon            IsUnicodeSyntax
675 676
  | ITequal
  | ITlam
677
  | ITlcase
678
  | ITvbar
679 680
  | ITlarrow            IsUnicodeSyntax
  | ITrarrow            IsUnicodeSyntax
681 682
  | ITat
  | ITtilde
683
  | ITdarrow            IsUnicodeSyntax
684 685
  | ITminus
  | ITbang
686
  | ITstar              IsUnicodeSyntax
687 688
  | ITdot

Ian Lynagh's avatar
Ian Lynagh committed
689
  | ITbiglam                    -- GHC-extension symbols
690

Ian Lynagh's avatar
Ian Lynagh committed
691
  | ITocurly                    -- special symbols
692 693 694 695
  | ITccurly
  | ITvocurly
  | ITvccurly
  | ITobrack
Ian Lynagh's avatar
Ian Lynagh committed
696 697
  | ITopabrack                  -- [:, for parallel arrays with -XParallelArrays
  | ITcpabrack                  -- :], for parallel arrays with -XParallelArrays
698 699 700 701 702 703 704 705 706
  | ITcbrack
  | IToparen
  | ITcparen
  | IToubxparen
  | ITcubxparen
  | ITsemi
  | ITcomma
  | ITunderscore
  | ITbackquote
dreixel's avatar
dreixel committed
707
  | ITsimpleQuote               --  '
708

Ian Lynagh's avatar
Ian Lynagh committed
709
  | ITvarid   FastString        -- identifiers
710 711 712 713 714 715 716 717
  | 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
718
  | ITdupipvarid   FastString   -- GHC extension: implicit param: ?x
Adam Gundry's avatar
Adam Gundry committed
719
  | ITlabelvarid   FastString   -- Overloaded label: #x
720

721 722
  | ITchar     SourceText Char       -- Note [Literal source text] in BasicTypes
  | ITstring   SourceText FastString -- Note [Literal source text] in BasicTypes
723
  | ITinteger  IntegralLit           -- Note [Literal source text] in BasicTypes
Alan Zimmerman's avatar
Alan Zimmerman committed
724
  | ITrational FractionalLit
725

726 727 728 729
  | 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
730 731
  | ITprimfloat  FractionalLit
  | ITprimdouble FractionalLit
732

733
  -- Template Haskell extension tokens
734 735 736 737 738 739 740 741 742 743 744 745
  | 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                           --  ''
746 747 748 749 750 751 752 753
  | 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 |]
754 755 756 757

  -- Arrow notation extension
  | ITproc
  | ITrec
758 759
  | IToparenbar  IsUnicodeSyntax --  (|
  | ITcparenbar  IsUnicodeSyntax --  |)
760 761 762 763
  | ITlarrowtail IsUnicodeSyntax --  -<
  | ITrarrowtail IsUnicodeSyntax --  >-
  | ITLarrowtail IsUnicodeSyntax --  -<<
  | ITRarrowtail IsUnicodeSyntax --  >>-
764

eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
765 766 767 768 769 770 771
  -- 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
772 773 774 775 776 777 778

  -- 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
779 780
  | ITlineComment     String     -- comment starting by "--"
  | ITblockComment    String     -- comment in {- -}
781

782
  deriving Show
783

Alan Zimmerman's avatar
Alan Zimmerman committed
784 785 786
instance Outputable Token where
  ppr x = text (show x)

Alan Zimmerman's avatar
Alan Zimmerman committed
787

788 789 790 791 792 793 794
-- 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)
--
795
reservedWordsFM :: UniqFM (Token, ExtsBitmap)
796
reservedWordsFM = listToUFM $
Ian Lynagh's avatar
Ian Lynagh committed
797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823
    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 ),

824 825
         ( "forall",         ITforall NormalSyntax,
                                              xbit ExplicitForallBit .|.
826 827
                                              xbit InRulePragBit),
         ( "mdo",            ITmdo,           xbit RecursiveDoBit),
828 829 830
             -- See Note [Lexing type pseudo-keywords]
         ( "family",         ITfamily,        0 ),
         ( "role",           ITrole,          0 ),
831
         ( "pattern",        ITpattern,       xbit PatternSynonymsBit),
Facundo Domínguez's avatar
Facundo Domínguez committed
832
         ( "static",         ITstatic,        0 ),
Ryan Scott's avatar
Ryan Scott committed
833 834
         ( "stock",          ITstock,         0 ),
         ( "anyclass",       ITanyclass,      0 ),
Ryan Scott's avatar
Ryan Scott committed
835
         ( "via",            ITvia,           0 ),
836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853
         ( "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
854 855 856 857
         ( "unit",           ITunit,          0 ),
         ( "dependency",     ITdependency,       0 ),
         ( "signature",      ITsignature,     0 ),

858 859 860
         ( "rec",            ITrec,           xbit ArrowsBit .|.
                                              xbit RecursiveDoBit),
         ( "proc",           ITproc,          xbit ArrowsBit)
861 862
     ]

863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878
{-----------------------------------
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.
-------------------------------------}
879

880
reservedSymsFM :: UniqFM (Token, ExtsBitmap -> Bool)
881
reservedSymsFM = listToUFM $
Ian Lynagh's avatar
Ian Lynagh committed
882
    map (\ (x,y,z) -> (mkFastString x,(y,z)))
883
      [ ("..",  ITdotdot,              always)
Ian Lynagh's avatar
Ian Lynagh committed
884
        -- (:) is a reserved op, meaning only list cons
885 886 887 888 889 890 891 892 893 894 895 896
       ,(":",   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
897

898 899
       ,("*", ITstar NormalSyntax, starIsTypeEnabled)

Ian Lynagh's avatar
Ian Lynagh committed
900
        -- For 'forall a . t'
901
       ,(".", ITdot,  always) -- \i -> explicitForallEnabled i || inRulePrag i)
Ian Lynagh's avatar
Ian Lynagh committed
902

903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921
       ,("-<",  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)
922 923
       ,("★",   ITstar UnicodeSyntax,
                  \i -> unicodeSyntaxEnabled i && starIsTypeEnabled i)
924

Simon Marlow's avatar
Simon Marlow committed
925 926 927
        -- 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).
928 929 930 931 932
       ]

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

Ian Lynagh's avatar
Ian Lynagh committed
933
type Action = RealSrcSpan -> StringBuffer -> Int -> P (RealLocated Token)
934

935
special :: Token -> Action
twanvl's avatar
twanvl committed
936
special tok span _buf _len = return (L span tok)
937

938
token, layout_token :: Token -> Action
twanvl's avatar
twanvl committed
939 940
token t span _buf _len = return (L span t)
layout_token t span _buf _len = pushLexState layout >> return (L span t)
941

942 943
idtoken :: (StringBuffer -> Int -> Token) -> Action
idtoken f span buf len = return (L span $! (f buf len))
944

945
skip_one_varid :: (FastString -> Token) -> Action
Ian Lynagh's avatar
Ian Lynagh committed
946
skip_one_varid f span buf len
947
  = return (L span $! f (lexemeToFastString (stepOn buf) (len-1)))
948

949 950 951 952
skip_two_varid :: (FastString -> Token) -> Action
skip_two_varid f span buf len
  = return (L span $! f (lexemeToFastString (stepOn (stepOn buf)) (len-2)))

953
strtoken :: (String -> Token) -> Action
Ian Lynagh's avatar
Ian Lynagh committed
954
strtoken f span buf len =
955
  return (L span $! (f $! lexemeToString buf len))
956 957

begin :: Int -> Action
958
begin code _span _str _len = do pushLexState code; lexToken
959 960

pop :: Action
Ian Lynagh's avatar
Ian Lynagh committed
961 962
pop _span _buf _len = do _ <- popLexState
                         lexToken
963

964 965 966 967 968 969 970 971
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
972 973
                 Layout prev_off _ : _ -> prev_off < offset
                 _                     -> True
974
      if isOK then pop_and open_brace span buf len
Ian Lynagh's avatar
Ian Lynagh committed
975
              else failSpanMsgP (RealSrcSpan span) (text "Missing block")
976

977
pop_and :: Action -> Action
Ian Lynagh's avatar
Ian Lynagh committed
978 979
pop_and act span buf len = do _ <- popLexState
                              act span buf len
980

981
{-# INLINE nextCharIs #-}
Ian Lynagh's avatar
Ian Lynagh committed
982
nextCharIs :: StringBuffer -> (Char -> Bool) -> Bool
983 984
nextCharIs buf p = not (atEnd buf) && p (currentChar buf)

985 986 987 988
{-# INLINE nextCharIsNot #-}
nextCharIsNot :: StringBuffer -> (Char -> Bool) -> Bool
nextCharIsNot buf p = not (nextCharIs buf p)

989
notFollowedBy :: Char -> AlexAccPred ExtsBitmap
Ian Lynagh's avatar
Ian Lynagh committed
990
notFollowedBy char _ _ _ (AI _ buf)
991
  = nextCharIsNot buf (== char)
992

993
notFollowedBySymbol :: AlexAccPred ExtsBitmap
994
notFollowedBySymbol _ _ _ (AI _ buf)
995
  = nextCharIsNot buf (`elem` "!#$%&*+./<=>?@\\^|-~")
ross's avatar
ross committed
996

Adam Gundry's avatar
Adam Gundry committed
997 998 999 1000
followedByDigit :: AlexAccPred ExtsBitmap
followedByDigit _ _ _ (AI _ buf)
  = afterOptionalSpace buf (\b -> nextCharIs b (`elem` ['0'..'9']))

1001 1002 1003 1004
ifCurrentChar :: Char -> AlexAccPred ExtsBitmap
ifCurrentChar char _ (AI _ buf) _ _
  = nextCharIs buf (== char)

1005 1006 1007 1008 1009
-- 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.
1010
isNormalComment :: AlexAccPred ExtsBitmap
1011
isNormalComment bits _ _ (AI _ buf)
1012
  | haddockEnabled bits = notFollowedByDocOrPragma
1013
  | otherwise           = nextCharIsNot buf (== '#')
1014 1015
  where
    notFollowedByDocOrPragma
1016
       = afterOptionalSpace buf (\b -> nextCharIsNot b (`elem` "|^*$#"))
1017

1018 1019 1020 1021 1022
afterOptionalSpace :: StringBuffer -> (StringBuffer -> Bool) -> Bool
afterOptionalSpace buf p
    = if nextCharIs buf (== ' ')
      then p (snd (nextChar buf))
      else p buf
1023

1024
atEOL :: AlexAccPred ExtsBitmap
1025
atEOL _ _ _ (AI _ buf) = atEnd buf || currentChar buf == '\n'
1026

1027
ifExtension :: (ExtsBitmap -> Bool) -> AlexAccPred ExtsBitmap
1028 1029
ifExtension pred bits _ _ _ = pred bits

1030 1031 1032
orExtensions :: (ExtsBitmap -> Bool) -> (ExtsBitmap -> Bool) -> AlexAccPred ExtsBitmap
orExtensions pred1 pred2 bits _ _ _ = pred1 bits || pred2 bits

1033 1034 1035
multiline_doc_comment :: Action
multiline_doc_comment span buf _len = withLexedDocType (worker "")
  where
1036
    worker commentAcc input docType checkNextLine = case alexGetChar' input of
Ian Lynagh's avatar
Ian Lynagh committed
1037
      Just ('\n', input')
1038 1039
        | checkNextLine -> case checkIfCommentLine input' of
          Just input -> worker ('\n':commentAcc) input docType checkNextLine
1040
          Nothing -> docCommentEnd input commentAcc docType buf span
1041 1042
        | otherwise -> docCommentEnd input commentAcc docType buf span
      Just (c, input) -> worker (c:commentAcc) input docType checkNextLine
1043
      Nothing -> docCommentEnd input commentAcc docType buf span
Ian Lynagh's avatar
Ian Lynagh committed
1044

1045 1046 1047 1048 1049 1050 1051 1052
    -- 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
1053 1054
    checkIfCommentLine input = check (dropNonNewlineSpace input)
      where