Lexer.x 119 KB
Newer Older
1
-----------------------------------------------------------------------------
2
-- (c) The University of Glasgow, 2006
3
--
4
-- GHC's lexer for Haskell 2010 [1].
5
--
6 7
-- This is a combination of an Alex-generated lexer [2] from a regex
-- definition, with some hand-coded bits. [3]
8 9
--
-- Completely accurate information about token-spans within the source
Ian Lynagh's avatar
Ian Lynagh committed
10 11
-- file is maintained.  Every token has a start and end RealSrcLoc
-- attached to it.
12
--
13 14 15 16 17
-- References:
-- [1] https://www.haskell.org/onlinereport/haskell2010/haskellch2.html
-- [2] http://www.haskell.org/alex/
-- [3] https://ghc.haskell.org/trac/ghc/wiki/Commentary/Compiler/Parser
--
18 19 20 21 22 23 24 25 26 27
-----------------------------------------------------------------------------

--   ToDo / known bugs:
--    - parsing integers is a bit slow
--    - readRational is a bit slow
--
--   Known bugs, that were also in the previous version:
--    - M... should be 3 tokens, not 1.
--    - pragma-end should be only valid in a pragma

28
--   qualified operator NOTES.
Ian Lynagh's avatar
Ian Lynagh committed
29
--
30 31 32 33 34 35 36 37 38
--   - If M.(+) is a single lexeme, then..
--     - Probably (+) should be a single lexeme too, for consistency.
--       Otherwise ( + ) would be a prefix operator, but M.( + ) would not be.
--     - But we have to rule out reserved operators, otherwise (..) becomes
--       a different lexeme.
--     - Should we therefore also rule out reserved operators in the qualified
--       form?  This is quite difficult to achieve.  We don't do it for
--       qualified varids.

39 40 41 42

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

43
{
44
{-# LANGUAGE CPP #-}
45
{-# LANGUAGE BangPatterns #-}
46
{-# LANGUAGE LambdaCase #-}
47

48 49
{-# OPTIONS_GHC -funbox-strict-fields #-}

50
module Lexer (
51
   Token(..), lexer, pragState, mkPState, mkPStatePure, PState(..),
52 53
   P(..), ParseResult(..), mkParserFlags, mkParserFlags', ParserFlags,
   getRealSrcLoc, getPState, withThisPackage,
54
   failLocMsgP, failSpanMsgP, srcParseFail,
Ian Lynagh's avatar
Ian Lynagh committed
55
   getMessages,
56
   popContext, pushModuleContext, setLastToken, setSrcLoc,
vivian's avatar
vivian committed
57
   activeContext, nextIsEOF,
simonpj@microsoft.com's avatar
simonpj@microsoft.com committed
58
   getLexState, popLexState, pushLexState,
59
   ExtBits(..), getBit,
Vladislav Zavialov's avatar
Vladislav Zavialov committed
60
   addWarning, addError,
Alan Zimmerman's avatar
Alan Zimmerman committed
61
   lexTokenStream,
62
   addAnnotation,AddAnn,addAnnsAt,mkParensApiAnn,
63
   commentToAnnotation
64 65
  ) where

66 67
import GhcPrelude

rodlogic's avatar
rodlogic committed
68 69
-- base
import Control.Monad
70
import Control.Monad.Fail as MonadFail
rodlogic's avatar
rodlogic committed
71
import Data.Bits
72
import Data.Char
rodlogic's avatar
rodlogic committed
73 74 75 76
import Data.List
import Data.Maybe
import Data.Word

77 78
import EnumSet (EnumSet)
import qualified EnumSet
79

80 81 82
-- ghc-boot
import qualified GHC.LanguageExtensions as LangExt

rodlogic's avatar
rodlogic committed
83 84 85 86 87 88 89 90
-- bytestring
import Data.ByteString (ByteString)

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

-- compiler/utils
91
import Bag
92 93 94 95
import Outputable
import StringBuffer
import FastString
import UniqFM
96
import Util             ( readRational, readHexRational )
rodlogic's avatar
rodlogic committed
97 98 99

-- compiler/main
import ErrUtils
100
import DynFlags
rodlogic's avatar
rodlogic committed
101 102 103

-- compiler/basicTypes
import SrcLoc
104
import Module
105 106
import BasicTypes     ( InlineSpec(..), RuleMatchInfo(..),
                        IntegralLit(..), FractionalLit(..),
Alan Zimmerman's avatar
Alan Zimmerman committed
107
                        SourceText(..) )
108

rodlogic's avatar
rodlogic committed
109 110
-- compiler/parser
import Ctype
Alan Zimmerman's avatar
Alan Zimmerman committed
111 112

import ApiAnnotation
113 114
}

115 116 117
-- -----------------------------------------------------------------------------
-- Alex "Character set macros"

118 119
-- NB: The logic behind these definitions is also reflected in basicTypes/Lexeme.hs
-- Any changes here should likely be reflected there.
120
$unispace    = \x05 -- Trick Alex into handling Unicode. See [Unicode in Alex].
121 122 123
$nl          = [\n\r\f]
$whitechar   = [$nl\v\ $unispace]
$white_no_nl = $whitechar # \n -- TODO #8424
124
$tab         = \t
125 126

$ascdigit  = 0-9
127
$unidigit  = \x03 -- Trick Alex into handling Unicode. See [Unicode in Alex].
128
$decdigit  = $ascdigit -- for now, should really be $digit (ToDo)
129 130 131
$digit     = [$ascdigit $unidigit]

$special   = [\(\)\,\;\[\]\`\{\}]
132
$ascsymbol = [\!\#\$\%\&\*\+\.\/\<\=\>\?\@\\\^\|\-\~\:]
133
$unisymbol = \x04 -- Trick Alex into handling Unicode. See [Unicode in Alex].
134
$symbol    = [$ascsymbol $unisymbol] # [$special \_\"\']
135

136
$unilarge  = \x01 -- Trick Alex into handling Unicode. See [Unicode in Alex].
137
$asclarge  = [A-Z]
138 139
$large     = [$asclarge $unilarge]

140
$unismall  = \x02 -- Trick Alex into handling Unicode. See [Unicode in Alex].
141
$ascsmall  = [a-z]
142 143
$small     = [$ascsmall $unismall \_]

144
$unigraphic = \x06 -- Trick Alex into handling Unicode. See [Unicode in Alex].
145
$graphic   = [$small $large $symbol $digit $special $unigraphic \"\']
146

147
$binit     = 0-1
Ian Lynagh's avatar
Ian Lynagh committed
148
$octit     = 0-7
149
$hexit     = [$decdigit A-F a-f]
150

151
$uniidchar = \x07 -- Trick Alex into handling Unicode. See [Unicode in Alex].
152
$idchar    = [$small $large $digit $uniidchar \']
153

154 155
$pragmachar = [$small $large $digit]

156 157
$docsym    = [\| \^ \* \$]

158

159 160 161 162 163 164 165 166
-- -----------------------------------------------------------------------------
-- 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
167

168 169 170 171 172 173 174 175
-- 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
176 177

@qual = (@conid \.)+
178 179 180 181
@qvarid = @qual @varid
@qconid = @qual @conid
@qvarsym = @qual @varsym
@qconsym = @qual @consym
182

183 184
@floating_point = @numspc @decimal \. @decimal @exponent? | @numspc @decimal @exponent
@hex_floating_point = @numspc @hexadecimal \. @hexadecimal @bin_exponent? | @numspc @hexadecimal @bin_exponent
185

186 187 188 189 190
-- normal signed numerical literals can only be explicitly negative,
-- not explicitly positive (contrast @exponent)
@negative = \-
@signed = @negative ?

191 192 193 194

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

195 196
haskell :-

197 198 199 200 201

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

-- everywhere: skip whitespace
Ian Lynagh's avatar
Ian Lynagh committed
202
$white_no_nl+ ;
203
$tab          { warnTab }
204 205 206 207 208

-- 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
209 210 211 212
-- are). We also rule out nested Haddock comments, if the -haddock flag is
-- set.

"{-" / { isNormalComment } { nested_comment lexToken }
213 214 215 216

-- 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.
217 218

-- Since Haddock comments aren't valid in every state, we need to rule them
Ian Lynagh's avatar
Ian Lynagh committed
219
-- out here.
220 221 222 223 224 225 226

-- 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.

227
"-- " ~$docsym .* { lineCommentToken }
228
"--" [^$symbol \ ] .* { lineCommentToken }
229 230 231

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

232
"-- " $docsym .* / { alexNotPred (ifExtension HaddockBit) } { lineCommentToken }
233 234 235 236 237 238 239

-- 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.

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

-- 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
245
"--"\-* / { atEOL } { lineCommentToken }
246 247 248 249

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

Jedai's avatar
Jedai committed
250
"-- " / { atEOL } { lineCommentToken }
251 252 253 254 255 256 257 258 259 260 261

-- '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
262
  \n                                    ;
Adam Gundry's avatar
Adam Gundry committed
263 264
  ^\# line                              { begin line_prag1 }
  ^\# / { followedByDigit }             { begin line_prag1 }
Ian Lynagh's avatar
Ian Lynagh committed
265 266 267
  ^\# pragma .* \n                      ; -- GCC 3.3 CPP generated, apparently
  ^\# \! .* \n                          ; -- #!, for scripts
  ()                                    { do_bol }
268 269 270 271 272
}

-- 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.
273
<layout, layout_do, layout_if> {
Ian Lynagh's avatar
Ian Lynagh committed
274 275 276 277
  \{ / { notFollowedBy '-' }            { hopefully_open_brace }
        -- we might encounter {-# here, but {- has been handled already
  \n                                    ;
  ^\# (line)?                           { begin line_prag1 }
278 279
}

280 281
-- after an 'if', a vertical bar starts a layout context for MultiWayIf
<layout_if> {
282
  \| / { notFollowedBySymbol }          { new_layout_context True dontGenerateSemic ITvbar }
283 284 285
  ()                                    { pop }
}

286
-- do is treated in a subtly different way, see new_layout_context
287 288
<layout>    ()                          { new_layout_context True  generateSemic ITvocurly }
<layout_do> ()                          { new_layout_context False generateSemic ITvocurly }
289 290 291 292

-- 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
293
<layout_left>  ()                       { do_layout_left }
294

Ian Lynagh's avatar
Ian Lynagh committed
295
<0,option_prags> \n                     { begin bol }
296

297 298
"{-#" $whitechar* $pragmachar+ / { known_pragma linePrags }
                                { dispatch_pragmas linePrags }
299 300 301

-- single-line line pragmas, of the form
--    # <line> "<file>" <extra-stuff> \n
302 303 304 305 306
<line_prag1> {
  @decimal $white_no_nl+ \" [$graphic \ ]* \"  { setLineAndFile line_prag1a }
  ()                                           { failLinePrag1 }
}
<line_prag1a> .*                               { popLinePrag1 }
307 308 309

-- Haskell-style line pragmas, of the form
--    {-# LINE <line> "<file>" #-}
310 311 312 313
<line_prag2> {
  @decimal $white_no_nl+ \" [$graphic \ ]* \"  { setLineAndFile line_prag2a }
}
<line_prag2a> "#-}"|"-}"                       { pop }
314 315
   -- NOTE: accept -} at the end of a LINE pragma, for compatibility
   -- with older versions of GHC which generated these.
316

Rufflewind's avatar
Rufflewind committed
317 318 319 320
-- Haskell-style column pragmas, of the form
--    {-# COLUMN <column> #-}
<column_prag> @decimal $whitechar* "#-}" { setColumn }

321
<0,option_prags> {
Ian Lynagh's avatar
Ian Lynagh committed
322
  "{-#" $whitechar* $pragmachar+
323 324 325 326 327
        $whitechar+ $pragmachar+ / { known_pragma twoWordPrags }
                                 { dispatch_pragmas twoWordPrags }

  "{-#" $whitechar* $pragmachar+ / { known_pragma oneWordPrags }
                                 { dispatch_pragmas oneWordPrags }
328

329
  -- We ignore all these pragmas, but don't generate a warning for them
330 331
  "{-#" $whitechar* $pragmachar+ / { known_pragma ignoredPrags }
                                 { dispatch_pragmas ignoredPrags }
332 333

  -- ToDo: should only be valid inside a pragma:
Ian Lynagh's avatar
Ian Lynagh committed
334
  "#-}"                          { endPrag }
335 336
}

337
<option_prags> {
338 339
  "{-#"  $whitechar* $pragmachar+ / { known_pragma fileHeaderPrags }
                                   { dispatch_pragmas fileHeaderPrags }
David Waern's avatar
David Waern committed
340 341
}

342 343
<0> {
  -- In the "0" mode we ignore these pragmas
344
  "{-#"  $whitechar* $pragmachar+ / { known_pragma fileHeaderPrags }
345 346 347
                     { nested_comment lexToken }
}

348
<0,option_prags> {
349 350
  "{-#"  { warnThen Opt_WarnUnrecognisedPragmas (text "Unrecognised pragma")
                    (nested_comment lexToken) }
351 352
}

353 354
-- '0' state: ordinary lexemes

355 356
-- Haddock comments

357
<0,option_prags> {
358 359
  "-- " $docsym      / { ifExtension HaddockBit } { multiline_doc_comment }
  "{-" \ ? $docsym   / { ifExtension HaddockBit } { nested_doc_comment }
360 361
}

362 363
-- "special" symbols

364
<0> {
365 366 367 368 369 370 371 372 373 374 375 376 377 378 379
  "[|"        / { ifExtension ThQuotesBit } { token (ITopenExpQuote NoE NormalSyntax) }
  "[||"       / { ifExtension ThQuotesBit } { token (ITopenTExpQuote NoE) }
  "[e|"       / { ifExtension ThQuotesBit } { token (ITopenExpQuote HasE NormalSyntax) }
  "[e||"      / { ifExtension ThQuotesBit } { token (ITopenTExpQuote HasE) }
  "[p|"       / { ifExtension ThQuotesBit } { token ITopenPatQuote }
  "[d|"       / { ifExtension ThQuotesBit } { layout_token ITopenDecQuote }
  "[t|"       / { ifExtension ThQuotesBit } { token ITopenTypQuote }
  "|]"        / { ifExtension ThQuotesBit } { token (ITcloseQuote NormalSyntax) }
  "||]"       / { ifExtension ThQuotesBit } { token ITcloseTExpQuote }
  \$ @varid   / { ifExtension ThBit }       { skip_one_varid ITidEscape }
  "$$" @varid / { ifExtension ThBit }       { skip_two_varid ITidTyEscape }
  "$("        / { ifExtension ThBit }       { token ITparenEscape }
  "$$("       / { ifExtension ThBit }       { token ITparenTyEscape }

  "[" @varid "|"  / { ifExtension QqBit }   { lex_quasiquote_tok }
380 381

  -- qualified quasi-quote (#5555)
382
  "[" @qvarid "|"  / { ifExtension QqBit }  { lex_qquasiquote_tok }
383 384 385

  $unigraphic -- ⟦
    / { ifCurrentChar '⟦' `alexAndPred`
386 387
        ifExtension UnicodeSyntaxBit `alexAndPred`
        ifExtension ThQuotesBit }
388 389 390
    { token (ITopenExpQuote NoE UnicodeSyntax) }
  $unigraphic -- ⟧
    / { ifCurrentChar '⟧' `alexAndPred`
391 392
        ifExtension UnicodeSyntaxBit `alexAndPred`
        ifExtension ThQuotesBit }
393
    { token (ITcloseQuote UnicodeSyntax) }
394 395
}

eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
396 397 398 399
  -- See Note [Lexing type applications]
<0> {
    [^ $idchar \) ] ^
  "@"
400
    / { ifExtension TypeApplicationsBit `alexAndPred` notFollowedBySymbol }
eir@cis.upenn.edu's avatar
eir@cis.upenn.edu committed
401 402 403
    { token ITtypeApp }
}

404
<0> {
405 406 407 408 409 410 411
  "(|"
    / { ifExtension ArrowsBit `alexAndPred`
        notFollowedBySymbol }
    { special (IToparenbar NormalSyntax) }
  "|)"
    / { ifExtension ArrowsBit }
    { special (ITcparenbar NormalSyntax) }
412 413 414

  $unigraphic -- ⦇
    / { ifCurrentChar '⦇' `alexAndPred`
415 416
        ifExtension UnicodeSyntaxBit `alexAndPred`
        ifExtension ArrowsBit }
417 418 419
    { special (IToparenbar UnicodeSyntax) }
  $unigraphic -- ⦈
    / { ifCurrentChar '⦈' `alexAndPred`
420 421
        ifExtension UnicodeSyntaxBit `alexAndPred`
        ifExtension ArrowsBit }
422
    { special (ITcparenbar UnicodeSyntax) }
423 424
}

425
<0> {
426
  \? @varid / { ifExtension IpBit } { skip_one_varid ITdupipvarid }
427 428
}

Adam Gundry's avatar
Adam Gundry committed
429
<0> {
430
  "#" @varid / { ifExtension OverloadedLabelsBit } { skip_one_varid ITlabelvarid }
Adam Gundry's avatar
Adam Gundry committed
431 432
}

433
<0> {
434 435
  "(#" / { ifExtension UnboxedTuplesBit `alexOrPred`
           ifExtension UnboxedSumsBit }
Ian Lynagh's avatar
Ian Lynagh committed
436
         { token IToubxparen }
437 438
  "#)" / { ifExtension UnboxedTuplesBit `alexOrPred`
           ifExtension UnboxedSumsBit }
Ian Lynagh's avatar
Ian Lynagh committed
439 440 441
         { token ITcubxparen }
}

442
<0,option_prags> {
Ian Lynagh's avatar
Ian Lynagh committed
443 444 445 446 447 448 449 450 451 452
  \(                                    { special IToparen }
  \)                                    { special ITcparen }
  \[                                    { special ITobrack }
  \]                                    { special ITcbrack }
  \,                                    { special ITcomma }
  \;                                    { special ITsemi }
  \`                                    { special ITbackquote }

  \{                                    { open_brace }
  \}                                    { close_brace }
453 454
}

455
<0,option_prags> {
456 457
  @qvarid                       { idtoken qvarid }
  @qconid                       { idtoken qconid }
Ian Lynagh's avatar
Ian Lynagh committed
458
  @varid                        { varid }
459
  @conid                        { idtoken conid }
460 461
}

462
<0> {
463 464 465 466
  @qvarid "#"+      / { ifExtension MagicHashBit } { idtoken qvarid }
  @qconid "#"+      / { ifExtension MagicHashBit } { idtoken qconid }
  @varid "#"+       / { ifExtension MagicHashBit } { varid }
  @conid "#"+       / { ifExtension MagicHashBit } { idtoken conid }
467 468
}

469 470
-- ToDo: - move `var` and (sym) into lexical syntax?
--       - remove backquote from $special?
471
<0> {
472 473
  @qvarsym                                         { idtoken qvarsym }
  @qconsym                                         { idtoken qconsym }
474 475
  @varsym                                          { varsym }
  @consym                                          { consym }
476 477
}

478 479
-- For the normal boxed literals we need to be careful
-- when trying to be close to Haskell98
480 481 482 483 484 485 486 487 488 489

-- 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.
--
490
<0> {
491
  -- Normal integral literals (:: Num a => a, from Integer)
492 493 494 495 496 497 498 499 500
  @decimal                                                                   { tok_num positive 0 0 decimal }
  0[bB] @numspc @binary                / { ifExtension BinaryLiteralsBit }   { tok_num positive 2 2 binary }
  0[oO] @numspc @octal                                                       { tok_num positive 2 2 octal }
  0[xX] @numspc @hexadecimal                                                 { tok_num positive 2 2 hexadecimal }
  @negative @decimal                   / { ifExtension NegativeLiteralsBit } { tok_num negative 1 1 decimal }
  @negative 0[bB] @numspc @binary      / { ifExtension NegativeLiteralsBit `alexAndPred`
                                           ifExtension BinaryLiteralsBit }   { tok_num negative 3 3 binary }
  @negative 0[oO] @numspc @octal       / { ifExtension NegativeLiteralsBit } { tok_num negative 3 3 octal }
  @negative 0[xX] @numspc @hexadecimal / { ifExtension NegativeLiteralsBit } { tok_num negative 3 3 hexadecimal }
501 502

  -- Normal rational literals (:: Fractional a => a, from Rational)
503 504 505 506 507 508
  @floating_point                                                            { tok_frac 0 tok_float }
  @negative @floating_point            / { ifExtension NegativeLiteralsBit } { tok_frac 0 tok_float }
  0[xX] @numspc @hex_floating_point    / { ifExtension HexFloatLiteralsBit } { tok_frac 0 tok_hex_float }
  @negative 0[xX] @numspc @hex_floating_point
                                       / { ifExtension HexFloatLiteralsBit `alexAndPred`
                                           ifExtension NegativeLiteralsBit } { tok_frac 0 tok_hex_float }
509 510
}

511
<0> {
Ian Lynagh's avatar
Ian Lynagh committed
512
  -- Unboxed ints (:: Int#) and words (:: Word#)
513 514
  -- It's simpler (and faster?) to give separate cases to the negatives,
  -- especially considering octal/hexadecimal prefixes.
515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531
  @decimal                          \# / { ifExtension MagicHashBit }        { tok_primint positive 0 1 decimal }
  0[bB] @numspc @binary             \# / { ifExtension MagicHashBit `alexAndPred`
                                           ifExtension BinaryLiteralsBit }   { tok_primint positive 2 3 binary }
  0[oO] @numspc @octal              \# / { ifExtension MagicHashBit }        { tok_primint positive 2 3 octal }
  0[xX] @numspc @hexadecimal        \# / { ifExtension MagicHashBit }        { tok_primint positive 2 3 hexadecimal }
  @negative @decimal                \# / { ifExtension MagicHashBit }        { tok_primint negative 1 2 decimal }
  @negative 0[bB] @numspc @binary   \# / { ifExtension MagicHashBit `alexAndPred`
                                           ifExtension BinaryLiteralsBit }   { tok_primint negative 3 4 binary }
  @negative 0[oO] @numspc @octal    \# / { ifExtension MagicHashBit }        { tok_primint negative 3 4 octal }
  @negative 0[xX] @numspc @hexadecimal \#
                                       / { ifExtension MagicHashBit }        { tok_primint negative 3 4 hexadecimal }

  @decimal                       \# \# / { ifExtension MagicHashBit }        { tok_primword 0 2 decimal }
  0[bB] @numspc @binary          \# \# / { ifExtension MagicHashBit `alexAndPred`
                                           ifExtension BinaryLiteralsBit }   { tok_primword 2 4 binary }
  0[oO] @numspc @octal           \# \# / { ifExtension MagicHashBit }        { tok_primword 2 4 octal }
  0[xX] @numspc @hexadecimal     \# \# / { ifExtension MagicHashBit }        { tok_primword 2 4 hexadecimal }
Ian Lynagh's avatar
Ian Lynagh committed
532

533 534
  -- Unboxed floats and doubles (:: Float#, :: Double#)
  -- prim_{float,double} work with signed literals
535 536
  @signed @floating_point           \# / { ifExtension MagicHashBit }        { tok_frac 1 tok_primfloat }
  @signed @floating_point        \# \# / { ifExtension MagicHashBit }        { 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 'UsePosPragsBit'
  | ITcolumn_prag       SourceText  -- not usually produced, see 'UsePosPragsBit'
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,        xbit StaticPointersBit ),
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, IsUnicodeSyntax, ExtsBitmap)
878
reservedSymsFM = listToUFM $
879 880
    map (\ (x,w,y,z) -> (mkFastString x,(w,y,z)))
      [ ("..",  ITdotdot,                   NormalSyntax,  0 )
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,                    NormalSyntax,  0 )
       ,("::",  ITdcolon NormalSyntax,      NormalSyntax,  0 )
       ,("=",   ITequal,                    NormalSyntax,  0 )
       ,("\\",  ITlam,                      NormalSyntax,  0 )
       ,("|",   ITvbar,                     NormalSyntax,  0 )
       ,("<-",  ITlarrow NormalSyntax,      NormalSyntax,  0 )
       ,("->",  ITrarrow NormalSyntax,      NormalSyntax,  0 )
       ,("@",   ITat,                       NormalSyntax,  0 )
       ,("~",   ITtilde,                    NormalSyntax,  0 )
       ,("=>",  ITdarrow NormalSyntax,      NormalSyntax,  0 )
       ,("-",   ITminus,                    NormalSyntax,  0 )
       ,("!",   ITbang,                     NormalSyntax,  0 )

       ,("*",   ITstar NormalSyntax,        NormalSyntax,  xbit StarIsTypeBit)
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
       ,(".",   ITdot,                      NormalSyntax,  0 )

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

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

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

       ,("★",   ITstar UnicodeSyntax,       UnicodeSyntax, xbit StarIsTypeBit)
917

Simon Marlow's avatar
Simon Marlow committed
918 919 920
        -- 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).
921 922 923 924 925
       ]

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

Ian Lynagh's avatar
Ian Lynagh committed
926
type Action = RealSrcSpan -> StringBuffer -> Int -> P (RealLocated Token)
927

928
special :: Token -> Action
twanvl's avatar
twanvl committed
929
special tok span _buf _len = return (L span tok)
930

931
token, layout_token :: Token -> Action
twanvl's avatar
twanvl committed
932 933
token t span _buf _len = return (L span t)
layout_token t span _buf _len = pushLexState layout >> return (L span t)
934

935 936
idtoken :: (StringBuffer -> Int -> Token) -> Action
idtoken f span buf len = return (L span $! (f buf len))
937

938
skip_one_varid :: (FastString -> Token) -> Action
Ian Lynagh's avatar
Ian Lynagh committed
939
skip_one_varid f span buf len
940
  = return (L span $! f (lexemeToFastString (stepOn buf) (len-1)))
941

942 943 944 945
skip_two_varid :: (FastString -> Token) -> Action
skip_two_varid f span buf len
  = return (L span $! f (lexemeToFastString (stepOn (stepOn buf)) (len-2)))

946
strtoken :: (String -> Token) -> Action
Ian Lynagh's avatar
Ian Lynagh committed
947
strtoken f span buf len =
948
  return (L span $! (f $! lexemeToString buf len))
949 950

begin :: Int -> Action
951
begin code _span _str _len = do pushLexState code; lexToken
952 953

pop :: Action
Ian Lynagh's avatar
Ian Lynagh committed
954 955
pop _span _buf _len = do _ <- popLexState
                         lexToken
956 957 958
-- See Note [Nested comment line pragmas]
failLinePrag1 :: Action
failLinePrag1 span _buf _len = do
959
  b <- getBit InNestedCommentBit
960 961 962 963 964 965
  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
966
  b <- getBit InNestedCommentBit
967 968 969
  if b then return (L span ITcomment_line_prag) else do
    _ <- popLexState
    lexToken
970

971 972
hopefully_open_brace :: Action
hopefully_open_brace span buf len
973
 = do relaxed <- getBit RelaxedLayoutBit
974 975 976 977 978
      ctx <- getContext
      (AI l _) <- getInput
      let offset = srcLocCol l
          isOK = relaxed ||
                 case ctx of
979 980
                 Layout prev_off _ : _ -> prev_off < offset
                 _                     -> True
981
      if isOK then pop_and open_brace span buf len
Ian Lynagh's avatar
Ian Lynagh committed
982
              else failSpanMsgP (RealSrcSpan span) (text "Missing block")
983

984
pop_and :: Action -> Action
Ian Lynagh's avatar
Ian Lynagh committed
985 986
pop_and act span buf len = do _ <- popLexState
                              act span buf len
987

988
{-# INLINE nextCharIs #-}
Ian Lynagh's avatar
Ian Lynagh committed
989
nextCharIs :: StringBuffer -> (Char -> Bool) -> Bool
990 991
nextCharIs buf p = not (atEnd buf) && p (currentChar buf)

992 993 994 995
{-# INLINE nextCharIsNot #-}
nextCharIsNot :: StringBuffer -> (Char -> Bool) -> Bool
nextCharIsNot buf p = not (nextCharIs buf p)

996
notFollowedBy :: Char -> AlexAccPred ExtsBitmap
Ian Lynagh's avatar
Ian Lynagh committed
997
notFollowedBy char _ _ _ (AI _ buf)
998
  = nextCharIsNot buf (== char)
999

1000
notFollowedBySymbol :: AlexAccPred ExtsBitmap
1001
notFollowedBySymbol _ _ _ (AI _ buf)
1002
  = nextCharIsNot buf (`elem` "!#$%&*+./<=>?@\\^|-~")
ross's avatar
ross committed
1003

Adam Gundry's avatar
Adam Gundry committed
1004 1005 1006 1007
followedByDigit :: AlexAccPred ExtsBitmap
followedByDigit _ _ _ (AI _ buf)
  = afterOptionalSpace buf (\b -> nextCharIs b (`elem` ['0'..'9']))

1008 1009 1010 1011
ifCurrentChar :: Char -> AlexAccPred ExtsBitmap
ifCurrentChar char _ (AI _ buf) _ _
  = nextCharIs buf (== char)

1012 1013 1014 1015 1016
-- 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.
1017
isNormalComment :: AlexAccPred ExtsBitmap
1018
isNormalComment bits _ _ (AI _ buf)
1019 1020
  | HaddockBit `xtest` bits = notFollowedByDocOrPragma
  | otherwise               = nextCharIsNot buf (== '#')
1021 1022
  where
    notFollowedByDocOrPragma
1023
       = afterOptionalSpace buf (\b -> nextCharIsNot b (`elem` "|^*$#"))
1024

1025 1026 1027 1028 1029
afterOptionalSpace :: StringBuffer -> (StringBuffer -> Bool) -> Bool
afterOptionalSpace buf p
    = if nextCharIs buf (== ' ')
      then p (snd (nextChar buf))
      else p buf
1030

1031
atEOL :: AlexAccPred ExtsBitmap
1032
atEOL _ _ _ (AI _ buf) = atEnd buf || currentChar buf == '\n'
1033

1034 1035 1036 1037 1038
ifExtension :: ExtBits -> AlexAccPred ExtsBitmap
ifExtension extBits bits _ _ _ = extBits `xtest` bits

alexNotPred p userState in1 len in2
  = not (p userState in1 len in2)
1039

1040 1041
alexOrPred p1 p2 userState in1 len in2
  = p1 userState in1 len in2 || p2 userState in1 len in2
1042

1043 1044 1045
multiline_doc_comment :: Action
multiline_doc_comment span buf _len = withLexedDocType (worker "")
  where
1046
    worker commentAcc input docType checkNextLine = case alexGetChar' input of
Ian Lynagh's avatar
Ian Lynagh committed
1047
      Just ('\n', input')
1048 1049
        | checkNextLine -> case checkIfCommentLine input' of
          Just input -> worker ('\n':commentAcc) input docType checkNextLine
1050
          Nothing -> docCommentEnd input commentAcc docType buf span
1051 1052
        | otherwise -> docCommentEnd input commentAcc docType buf span
      Just (c, input) -> worker (c:commentAcc) input docType checkNextLine
1053
      Nothing -> docCommentEnd input commentAcc docType buf span
Ian Lynagh's avatar
Ian Lynagh committed
1054

1055 1056 1057 1058 1059 1060 1061 1062
    -- 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 "-- $", because that would be the
    --     start of a /new/ named haddock chunk (#10398).
    checkIfCommentLine :: AlexInput -> Maybe AlexInput
1063 1064
    checkIfCommentLine input = check (dropNonNewlineSpace input)
      where
1065 1066 1067 1068 1069 1070 1071 1072 1073 1074
        check input = do
          ('-', input) <- alexGetChar' input
          ('-', input) <- alexGetChar' input
          (c, after_c) <- alexGetChar' input
          case c of
            '-' -> Nothing
            ' ' -> case alexGetChar' after_c of
                     Just ('$', _) -> Nothing
                     _ -> Just input
            _   -> Just input
1075

1076
        dropNonNewlineSpace input = case alexGetChar' input of
Ian Lynagh's avatar
Ian Lynagh committed
1077
          Just (c, input')
1078 1079 1080 1081
            | isSpace c && c /= '\n' -> dropNonNewlineSpace input'
            | otherwise -> input
          Nothing -> input

Jedai's avatar
Jedai committed
1082 1083
lineCommentToken :: Action
lineCommentToken span buf len = do
1084
  b <- getBit RawTokenStreamBit
Jedai's avatar
Jedai committed
1085 1086
  if b then strtoken ITlineComment span buf len else lexToken

1087 1088 1089 1090
{-
  nested comments require traversing by hand, they can't be parsed
  using regular expressions.
-}
Ian Lynagh's avatar
Ian Lynagh committed
1091
nested_comment :: P (RealLocated Token) -> Action
Alan Zimmerman's avatar
Alan Zimmerman committed
1092