#endif
}
-$unispace = \x05
+$unispace = \x05 -- Trick Alex into handling Unicode. See alexGetChar.
$whitechar = [\ \n\r\f\v\xa0 $unispace]
$white_no_nl = $whitechar # \n
$tab = \t
$ascdigit = 0-9
-$unidigit = \x03
+$unidigit = \x03 -- Trick Alex into handling Unicode. See alexGetChar.
$decdigit = $ascdigit -- for now, should really be $digit (ToDo)
$digit = [$ascdigit $unidigit]
$special = [\(\)\,\;\[\]\`\{\}]
$ascsymbol = [\!\#\$\%\&\*\+\.\/\<\=\>\?\@\\\^\|\-\~ \xa1-\xbf \xd7 \xf7]
-$unisymbol = \x04
+$unisymbol = \x04 -- Trick Alex into handling Unicode. See alexGetChar.
$symbol = [$ascsymbol $unisymbol] # [$special \_\:\"\']
-$unilarge = \x01
+$unilarge = \x01 -- Trick Alex into handling Unicode. See alexGetChar.
$asclarge = [A-Z \xc0-\xd6 \xd8-\xde]
$large = [$asclarge $unilarge]
-$unismall = \x02
+$unismall = \x02 -- Trick Alex into handling Unicode. See alexGetChar.
$ascsmall = [a-z \xdf-\xf6 \xf8-\xff]
$small = [$ascsmall $unismall \_]
-$unigraphic = \x06
+$unigraphic = \x06 -- Trick Alex into handling Unicode. See alexGetChar.
$graphic = [$small $large $symbol $digit $special $unigraphic \:\"\']
$octit = 0-7
@floating_point = @decimal \. @decimal @exponent? | @decimal @exponent
+-- normal signed numerical literals can only be explicitly negative,
+-- not explicitly positive (contrast @exponent)
+@negative = \-
+@signed = @negative ?
+
haskell :-
-- everywhere: skip whitespace and comments
-- Haddock comments
<0,glaexts> {
- "-- " / $docsym { multiline_doc_comment }
- "{-" \ ? / $docsym { nested_doc_comment }
+ "-- " $docsym / { ifExtension haddockEnabled } { multiline_doc_comment }
+ "{-" \ ? $docsym / { ifExtension haddockEnabled } { nested_doc_comment }
}
-- "special" symbols
@qual @conid { pop_and (idtoken qconid) }
}
-<glaexts> {
- @qual @varid "#"+ { idtoken qvarid }
- @qual @conid "#"+ { idtoken qconid }
- @varid "#"+ { varid }
- @conid "#"+ { idtoken conid }
+<0,glaexts> {
+ @qual @varid "#"+ / { ifExtension magicHashEnabled } { idtoken qvarid }
+ @qual @conid "#"+ / { ifExtension magicHashEnabled } { idtoken qconid }
+ @varid "#"+ / { ifExtension magicHashEnabled } { varid }
+ @conid "#"+ / { ifExtension magicHashEnabled } { idtoken conid }
}
-- ToDo: M.(,,,)
@consym { consym }
}
+-- For the normal boxed literals we need to be careful
+-- when trying to be close to Haskell98
<0,glaexts> {
- @decimal { tok_decimal }
- 0[oO] @octal { tok_octal }
- 0[xX] @hexadecimal { tok_hexadecimal }
+ -- Normal integral literals (:: Num a => a, from Integer)
+ @decimal { tok_num positive 0 0 decimal }
+ 0[oO] @octal { tok_num positive 2 2 octal }
+ 0[xX] @hexadecimal { tok_num positive 2 2 hexadecimal }
+
+ -- Normal rational literals (:: Fractional a => a, from Rational)
+ @floating_point { strtoken tok_float }
}
<glaexts> {
- @decimal \# { prim_decimal }
- 0[oO] @octal \# { prim_octal }
- 0[xX] @hexadecimal \# { prim_hexadecimal }
+ -- Unboxed ints (:: Int#)
+ -- It's simpler (and faster?) to give separate cases to the negatives,
+ -- especially considering octal/hexadecimal prefixes.
+ @decimal \# { tok_primint positive 0 1 decimal }
+ 0[oO] @octal \# { tok_primint positive 2 3 octal }
+ 0[xX] @hexadecimal \# { tok_primint positive 2 3 hexadecimal }
+ @negative @decimal \# { tok_primint negative 1 2 decimal }
+ @negative 0[oO] @octal \# { tok_primint negative 3 4 octal }
+ @negative 0[xX] @hexadecimal \# { tok_primint negative 3 4 hexadecimal }
+
+ -- Unboxed floats and doubles (:: Float#, :: Double#)
+ -- prim_{float,double} work with signed literals
+ @signed @floating_point \# { init_strtoken 1 tok_primfloat }
+ @signed @floating_point \# \# { init_strtoken 2 tok_primdouble }
}
-<0,glaexts> @floating_point { strtoken tok_float }
-<glaexts> @floating_point \# { init_strtoken 1 prim_float }
-<glaexts> @floating_point \# \# { init_strtoken 2 prim_double }
-
-- 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
}
{
--- work around bug in Alex 2.0
-#if __GLASGOW_HASKELL__ < 503
-unsafeAt arr i = arr ! i
-#endif
-
-- -----------------------------------------------------------------------------
-- The token type
( "where", ITwhere, 0 ),
( "_scc_", ITscc, 0 ), -- ToDo: remove
- ( "forall", ITforall, bit tvBit),
- ( "mdo", ITmdo, bit glaExtsBit),
- ( "family", ITfamily, bit idxTysBit),
+ ( "forall", ITforall, bit explicitForallBit),
+ ( "mdo", ITmdo, bit recursiveDoBit),
+ ( "family", ITfamily, bit tyFamBit),
( "foreign", ITforeign, bit ffiBit),
( "export", ITexport, bit ffiBit),
( "proc", ITproc, bit arrowsBit)
]
+reservedSymsFM :: UniqFM (Token, Int -> Bool)
reservedSymsFM = listToUFM $
- map (\ (x,y,z) -> (mkFastString x,(y,z)))
- [ ("..", ITdotdot, 0)
- ,(":", ITcolon, 0) -- (:) is a reserved op,
- -- meaning only list cons
- ,("::", ITdcolon, 0)
- ,("=", ITequal, 0)
- ,("\\", ITlam, 0)
- ,("|", ITvbar, 0)
- ,("<-", ITlarrow, 0)
- ,("->", ITrarrow, 0)
- ,("@", ITat, 0)
- ,("~", ITtilde, 0)
- ,("=>", ITdarrow, 0)
- ,("-", ITminus, 0)
- ,("!", ITbang, 0)
-
- ,("*", ITstar, bit glaExtsBit .|.
- bit idxTysBit) -- For data T (a::*) = MkT
- ,(".", ITdot, bit tvBit) -- For 'forall a . t'
-
- ,("-<", ITlarrowtail, bit arrowsBit)
- ,(">-", ITrarrowtail, bit arrowsBit)
- ,("-<<", ITLarrowtail, bit arrowsBit)
- ,(">>-", ITRarrowtail, bit arrowsBit)
+ map (\ (x,y,z) -> (mkFastString x,(y,z)))
+ [ ("..", ITdotdot, always)
+ -- (:) is a reserved op, meaning only list cons
+ ,(":", ITcolon, always)
+ ,("::", ITdcolon, always)
+ ,("=", ITequal, always)
+ ,("\\", ITlam, always)
+ ,("|", ITvbar, always)
+ ,("<-", ITlarrow, always)
+ ,("->", ITrarrow, always)
+ ,("@", ITat, always)
+ ,("~", ITtilde, always)
+ ,("=>", ITdarrow, always)
+ ,("-", ITminus, always)
+ ,("!", ITbang, always)
+
+ -- For data T (a::*) = MkT
+ ,("*", ITstar, \i -> glaExtsEnabled i ||
+ kindSigsEnabled i ||
+ tyFamEnabled i)
+ -- For 'forall a . t'
+ ,(".", ITdot, explicitForallEnabled)
+
+ ,("-<", ITlarrowtail, arrowsEnabled)
+ ,(">-", ITrarrowtail, arrowsEnabled)
+ ,("-<<", ITLarrowtail, arrowsEnabled)
+ ,(">>-", ITRarrowtail, arrowsEnabled)
#if __GLASGOW_HASKELL__ >= 605
- ,("∷", ITdcolon, bit glaExtsBit)
- ,("⇒", ITdarrow, bit glaExtsBit)
- ,("∀", ITforall, bit glaExtsBit)
- ,("→", ITrarrow, bit glaExtsBit)
- ,("←", ITlarrow, bit glaExtsBit)
- ,("?", ITdotdot, bit glaExtsBit)
+ ,("∷", ITdcolon, unicodeSyntaxEnabled)
+ ,("⇒", ITdarrow, unicodeSyntaxEnabled)
+ ,("∀", ITforall, \i -> unicodeSyntaxEnabled i &&
+ explicitForallEnabled i)
+ ,("→", ITrarrow, unicodeSyntaxEnabled)
+ ,("←", ITlarrow, unicodeSyntaxEnabled)
+ ,("⋯", ITdotdot, unicodeSyntaxEnabled)
-- 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).
notFollowedBySymbol _ _ _ (AI _ _ buf)
= nextCharIs buf (`notElem` "!#$%&*+./<=>?@\\^|-~")
+-- 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.
isNormalComment bits _ _ (AI _ _ buf)
| haddockEnabled bits = notFollowedByDocOrPragma
| otherwise = nextCharIs buf (/='#')
- where
- notFollowedByDocOrPragma
- = not $ spaceAndP buf (`nextCharIs` (`elem` "|^*$#"))
+ where
+ notFollowedByDocOrPragma
+ = not $ spaceAndP buf (`nextCharIs` (`elem` "|^*$#"))
spaceAndP buf p = p buf || nextCharIs buf (==' ') && p (snd (nextChar buf))
Just (c,input) -> go (c:commentAcc) input docType False
withLexedDocType lexDocComment = do
- input <- getInput
- case alexGetChar input of
- Nothing -> error "Can't happen"
- Just ('|', input) -> lexDocComment input ITdocCommentNext False
- Just ('^', input) -> lexDocComment input ITdocCommentPrev False
- Just ('$', input) -> lexDocComment input ITdocCommentNamed False
- Just ('*', input) -> lexDocSection 1 input
+ input@(AI _ _ buf) <- getInput
+ case prevChar buf ' ' of
+ '|' -> lexDocComment input ITdocCommentNext False
+ '^' -> lexDocComment input ITdocCommentPrev False
+ '$' -> lexDocComment input ITdocCommentNamed False
+ '*' -> lexDocSection 1 input
where
lexDocSection n input = case alexGetChar input of
Just ('*', input) -> lexDocSection (n+1) input
sym con span buf len =
case lookupUFM reservedSymsFM fs of
- Just (keyword,0) -> return (L span keyword)
Just (keyword,exts) -> do
- b <- extension (\i -> exts .&. i /= 0)
+ b <- extension exts
if b then return (L span keyword)
else return (L span $! con fs)
_other -> return (L span $! con fs)
where
fs = lexemeToFastString buf len
-tok_decimal span buf len
- = return (L span (ITinteger $! parseInteger buf len 10 octDecDigit))
-
-tok_octal span buf len
- = return (L span (ITinteger $! parseInteger (offsetBytes 2 buf) (len-2) 8 octDecDigit))
-
-tok_hexadecimal span buf len
- = return (L span (ITinteger $! parseInteger (offsetBytes 2 buf) (len-2) 16 hexDigit))
-
-prim_decimal span buf len
- = return (L span (ITprimint $! parseInteger buf (len-1) 10 octDecDigit))
-
-prim_octal span buf len
- = return (L span (ITprimint $! parseInteger (offsetBytes 2 buf) (len-3) 8 octDecDigit))
-
-prim_hexadecimal span buf len
- = return (L span (ITprimint $! parseInteger (offsetBytes 2 buf) (len-3) 16 hexDigit))
-
+-- Variations on the integral numeric literal.
+tok_integral :: (Integer -> Token)
+ -> (Integer -> Integer)
+ -- -> (StringBuffer -> StringBuffer) -> (Int -> Int)
+ -> Int -> Int
+ -> (Integer, (Char->Int)) -> Action
+tok_integral itint transint transbuf translen (radix,char_to_int) span buf len =
+ return $ L span $ itint $! transint $ parseUnsignedInteger
+ (offsetBytes transbuf buf) (subtract translen len) radix char_to_int
+
+-- some conveniences for use with tok_integral
+tok_num = tok_integral ITinteger
+tok_primint = tok_integral ITprimint
+positive = id
+negative = negate
+decimal = (10,octDecDigit)
+octal = (8,octDecDigit)
+hexadecimal = (16,hexDigit)
+
+-- readRational can understand negative rationals, exponents, everything.
tok_float str = ITrational $! readRational str
-prim_float str = ITprimfloat $! readRational str
-prim_double str = ITprimdouble $! readRational str
+tok_primfloat str = ITprimfloat $! readRational str
+tok_primdouble str = ITprimdouble $! readRational str
-- -----------------------------------------------------------------------------
-- Layout processing
setLine :: Int -> Action
setLine code span buf len = do
- let line = parseInteger buf len 10 octDecDigit
+ let line = parseUnsignedInteger buf len 10 octDecDigit
setSrcLoc (mkSrcLoc (srcSpanFile span) (fromIntegral line - 1) 0)
-- subtract one: the line number refers to the *following* line
popLexState
adj_c
| c <= '\x06' = non_graphic
| c <= '\xff' = c
+ -- Alex doesn't handle Unicode, so when Unicode
+ -- character is encoutered we output these values
+ -- with the actual character value hidden in the state.
| otherwise =
case generalCategory c of
UppercaseLetter -> upper
arrowsBit = 4
thBit = 5
ipBit = 6
-tvBit = 7 -- Scoped type variables enables 'forall' keyword
+explicitForallBit = 7 -- the 'forall' keyword and '.' symbol
bangPatBit = 8 -- Tells the parser to understand bang-patterns
-- (doesn't affect the lexer)
-idxTysBit = 9 -- indexed type families: 'family' keyword and kind sigs
+tyFamBit = 9 -- indexed type families: 'family' keyword and kind sigs
haddockBit = 10 -- Lex and parse Haddock comments
+magicHashBit = 11 -- # in both functions and operators
+kindSigsBit = 12 -- Kind signatures on type variables
+recursiveDoBit = 13 -- mdo
+unicodeSyntaxBit = 14 -- the forall symbol, arrow symbols, etc
glaExtsEnabled, ffiEnabled, parrEnabled :: Int -> Bool
-glaExtsEnabled flags = testBit flags glaExtsBit
-ffiEnabled flags = testBit flags ffiBit
-parrEnabled flags = testBit flags parrBit
-arrowsEnabled flags = testBit flags arrowsBit
-thEnabled flags = testBit flags thBit
-ipEnabled flags = testBit flags ipBit
-tvEnabled flags = testBit flags tvBit
-bangPatEnabled flags = testBit flags bangPatBit
-idxTysEnabled flags = testBit flags idxTysBit
-haddockEnabled flags = testBit flags haddockBit
+always _ = True
+glaExtsEnabled flags = testBit flags glaExtsBit
+ffiEnabled flags = testBit flags ffiBit
+parrEnabled flags = testBit flags parrBit
+arrowsEnabled flags = testBit flags arrowsBit
+thEnabled flags = testBit flags thBit
+ipEnabled flags = testBit flags ipBit
+explicitForallEnabled flags = testBit flags explicitForallBit
+bangPatEnabled flags = testBit flags bangPatBit
+tyFamEnabled flags = testBit flags tyFamBit
+haddockEnabled flags = testBit flags haddockBit
+magicHashEnabled flags = testBit flags magicHashBit
+kindSigsEnabled flags = testBit flags kindSigsBit
+recursiveDoEnabled flags = testBit flags recursiveDoBit
+unicodeSyntaxEnabled flags = testBit flags unicodeSyntaxBit
-- PState for parsing options pragmas
--
-- we begin in the layout state if toplev_layout is set
}
where
- bitmap = glaExtsBit `setBitIf` dopt Opt_GlasgowExts flags
- .|. ffiBit `setBitIf` dopt Opt_FFI flags
- .|. parrBit `setBitIf` dopt Opt_PArr flags
- .|. arrowsBit `setBitIf` dopt Opt_Arrows flags
- .|. thBit `setBitIf` dopt Opt_TH flags
- .|. ipBit `setBitIf` dopt Opt_ImplicitParams flags
- .|. tvBit `setBitIf` dopt Opt_ScopedTypeVariables flags
- .|. bangPatBit `setBitIf` dopt Opt_BangPatterns flags
- .|. idxTysBit `setBitIf` dopt Opt_IndexedTypes flags
- .|. haddockBit `setBitIf` dopt Opt_Haddock flags
+ bitmap = glaExtsBit `setBitIf` dopt Opt_GlasgowExts flags
+ .|. ffiBit `setBitIf` dopt Opt_FFI flags
+ .|. parrBit `setBitIf` dopt Opt_PArr flags
+ .|. arrowsBit `setBitIf` dopt Opt_Arrows flags
+ .|. thBit `setBitIf` dopt Opt_TH flags
+ .|. ipBit `setBitIf` dopt Opt_ImplicitParams flags
+ .|. explicitForallBit `setBitIf` dopt Opt_ScopedTypeVariables flags
+ .|. explicitForallBit `setBitIf` dopt Opt_PolymorphicComponents flags
+ .|. bangPatBit `setBitIf` dopt Opt_BangPatterns flags
+ .|. tyFamBit `setBitIf` dopt Opt_TypeFamilies flags
+ .|. haddockBit `setBitIf` dopt Opt_Haddock flags
+ .|. magicHashBit `setBitIf` dopt Opt_MagicHash flags
+ .|. kindSigsBit `setBitIf` dopt Opt_KindSignatures flags
+ .|. recursiveDoBit `setBitIf` dopt Opt_RecursiveDo flags
+ .|. unicodeSyntaxBit `setBitIf` dopt Opt_UnicodeSyntax flags
--
setBitIf :: Int -> Bool -> Int
b `setBitIf` cond | cond = bit b
projects / ghc-hetmet.git / blobdiff