X-Git-Url: http://git.megacz.com/?a=blobdiff_plain;f=ghc%2Fcompiler%2Fparser%2FLexer.x;h=c7ffc592e01cf8b0842b1c5711498c3fe5156010;hb=22df1e2a699d6eda6d5ada5073bc97c9f35e2947;hp=e83bf944b6e54ed7cb833fd955bb3608e36e7cc6;hpb=81d2f8a9de4b6f17ca5ee00d984c07fecc98f67f;p=ghc-hetmet.git diff --git a/ghc/compiler/parser/Lexer.x b/ghc/compiler/parser/Lexer.x index e83bf94..c7ffc59 100644 --- a/ghc/compiler/parser/Lexer.x +++ b/ghc/compiler/parser/Lexer.x @@ -22,10 +22,11 @@ { module Lexer ( - Token(..), lexer, mkPState, + Token(..), lexer, mkPState, PState(..), P(..), ParseResult(..), getSrcLoc, - failMsgP, failLocMsgP, failSpanMsgP, srcParseFail, - popContext, pushCurrentContext, + failLocMsgP, failSpanMsgP, srcParseFail, + popContext, pushCurrentContext, setLastToken, setSrcLoc, + getLexState, popLexState, pushLexState ) where #include "HsVersions.h" @@ -37,9 +38,9 @@ import FastString import FastTypes import SrcLoc import UniqFM -import CmdLineOpts +import DynFlags import Ctype -import Util ( maybePrefixMatch ) +import Util ( maybePrefixMatch, readRational ) import DATA_BITS import Char @@ -108,7 +109,8 @@ $white_no_nl+ ; -- have to exclude those. -- The regex says: "munch all the characters after the dashes, as long as -- the first one is not a symbol". -"--"\-* ([^$symbol] .*)? ; +"--"\-* [^$symbol] .* ; +"--"\-* / { atEOL } ; -- 'bol' state: beginning of a line. Slurp up all the whitespace (including -- blank lines) until we find a non-whitespace character, then do layout @@ -123,6 +125,7 @@ $white_no_nl+ ; \n ; ^\# (line)? { begin line_prag1 } ^\# pragma .* \n ; -- GCC 3.3 CPP generated, apparently + ^\# \! .* \n ; -- #!, for scripts () { do_bol } } @@ -151,18 +154,26 @@ $white_no_nl+ ; -- single-line line pragmas, of the form -- # "" \n - $digit+ { set_line line_prag1a } - \" [$graphic \ ]* \" { set_file line_prag1b } + $digit+ { setLine line_prag1a } + \" [$graphic \ ]* \" { setFile line_prag1b } .* { pop } -- Haskell-style line pragmas, of the form -- {-# LINE "" #-} - $digit+ { set_line line_prag2a } - \" [$graphic \ ]* \" { set_file line_prag2b } + $digit+ { setLine line_prag2a } + \" [$graphic \ ]* \" { setFile line_prag2b } "#-}"|"-}" { pop } -- NOTE: accept -} at the end of a LINE pragma, for compatibility -- with older versions of GHC which generated these. +-- We only want RULES pragmas to be picked up when -fglasgow-exts +-- is on, because the contents of the pragma is always written using +-- glasgow-exts syntax (using forall etc.), so if glasgow exts are not +-- enabled, we're sure to get a parse error. +-- (ToDo: we should really emit a warning when ignoring pragmas) + + "{-#" $whitechar* (RULES|rules) { token ITrules_prag } + <0,glaexts> { "{-#" $whitechar* (SPECIALI[SZ]E|speciali[sz]e) { token ITspecialise_prag } @@ -170,7 +181,6 @@ $white_no_nl+ ; "{-#" $whitechar* (INLINE|inline) { token ITinline_prag } "{-#" $whitechar* (NO(T?)INLINE|no(t?)inline) { token ITnoinline_prag } - "{-#" $whitechar* (RULES|rules) { token ITrules_prag } "{-#" $whitechar* (DEPRECATED|deprecated) { token ITdeprecated_prag } "{-#" $whitechar* (SCC|scc) { token ITscc_prag } @@ -415,25 +425,25 @@ data Token | ITprimdouble Rational -- MetaHaskell extension tokens - | ITopenExpQuote -- [| or [e| - | ITopenPatQuote -- [p| - | ITopenDecQuote -- [d| - | ITopenTypQuote -- [t| - | ITcloseQuote -- |] - | ITidEscape FastString -- $x - | ITparenEscape -- $( - | ITvarQuote -- ' - | ITtyQuote -- '' + | ITopenExpQuote -- [| or [e| + | ITopenPatQuote -- [p| + | ITopenDecQuote -- [d| + | ITopenTypQuote -- [t| + | ITcloseQuote -- |] + | ITidEscape FastString -- $x + | ITparenEscape -- $( + | ITvarQuote -- ' + | ITtyQuote -- '' -- Arrow notation extension | ITproc | ITrec - | IToparenbar -- (| - | ITcparenbar -- |) - | ITlarrowtail -- -< - | ITrarrowtail -- >- - | ITLarrowtail -- -<< - | ITRarrowtail -- >>- + | IToparenbar -- (| + | ITcparenbar -- |) + | ITlarrowtail -- -< + | ITrarrowtail -- >- + | ITLarrowtail -- -<< + | ITRarrowtail -- >>- | ITunknown String -- Used when the lexer can't make sense of it | ITeof -- end of file token @@ -496,7 +506,7 @@ reservedWordsFM = listToUFM $ ( "where", ITwhere, 0 ), ( "_scc_", ITscc, 0 ), -- ToDo: remove - ( "forall", ITforall, bit glaExtsBit), + ( "forall", ITforall, bit tvBit), ( "mdo", ITmdo, bit glaExtsBit), ( "foreign", ITforeign, bit ffiBit), @@ -532,7 +542,7 @@ reservedSymsFM = listToUFM $ ,("!", ITbang, 0) ,("*", ITstar, bit glaExtsBit) -- For data T (a::*) = MkT - ,(".", ITdot, bit glaExtsBit) -- For 'forall a . t' + ,(".", ITdot, bit tvBit) -- For 'forall a . t' ,("-<", ITlarrowtail, bit arrowsBit) ,(">-", ITrarrowtail, bit arrowsBit) @@ -582,6 +592,8 @@ notFollowedBy char _ _ _ (_,buf) = atEnd buf || currentChar buf /= char notFollowedBySymbol _ _ _ (_,buf) = atEnd buf || currentChar buf `notElem` "!#$%&*+./<=>?@\\^|-~" +atEOL _ _ _ (_,buf) = atEnd buf || currentChar buf == '\n' + ifExtension pred bits _ _ _ = pred bits {- @@ -705,32 +717,26 @@ sym con span buf len = fs = lexemeToFastString buf len tok_decimal span buf len - = return (L span (ITinteger $! parseInteger buf len 10 oct_or_dec)) + = return (L span (ITinteger $! parseInteger buf len 10 octDecDigit)) tok_octal span buf len - = return (L span (ITinteger $! parseInteger (stepOnBy 2 buf) (len-2) 8 oct_or_dec)) + = return (L span (ITinteger $! parseInteger (stepOnBy 2 buf) (len-2) 8 octDecDigit)) tok_hexadecimal span buf len - = return (L span (ITinteger $! parseInteger (stepOnBy 2 buf) (len-2) 16 hex)) + = return (L span (ITinteger $! parseInteger (stepOnBy 2 buf) (len-2) 16 hexDigit)) prim_decimal span buf len - = return (L span (ITprimint $! parseInteger buf (len-1) 10 oct_or_dec)) + = return (L span (ITprimint $! parseInteger buf (len-1) 10 octDecDigit)) prim_octal span buf len - = return (L span (ITprimint $! parseInteger (stepOnBy 2 buf) (len-3) 8 oct_or_dec)) + = return (L span (ITprimint $! parseInteger (stepOnBy 2 buf) (len-3) 8 octDecDigit)) prim_hexadecimal span buf len - = return (L span (ITprimint $! parseInteger (stepOnBy 2 buf) (len-3) 16 hex)) + = return (L span (ITprimint $! parseInteger (stepOnBy 2 buf) (len-3) 16 hexDigit)) -tok_float str = ITrational $! readRational__ str -prim_float str = ITprimfloat $! readRational__ str -prim_double str = ITprimdouble $! readRational__ str - -parseInteger :: StringBuffer -> Int -> Integer -> (Char->Int) -> Integer -parseInteger buf len radix to_int - = go 0 0 - where go i x | i == len = x - | otherwise = go (i+1) (x * radix + toInteger (to_int (lookAhead buf i))) +tok_float str = ITrational $! readRational str +prim_float str = ITprimfloat $! readRational str +prim_double str = ITprimdouble $! readRational str -- ----------------------------------------------------------------------------- -- Layout processing @@ -796,17 +802,17 @@ do_layout_left span _buf _len = do -- ----------------------------------------------------------------------------- -- LINE pragmas -set_line :: Int -> Action -set_line code span buf len = do - let line = parseInteger buf len 10 oct_or_dec +setLine :: Int -> Action +setLine code span buf len = do + let line = parseInteger buf len 10 octDecDigit setSrcLoc (mkSrcLoc (srcSpanFile span) (fromIntegral line - 1) 0) -- subtract one: the line number refers to the *following* line popLexState pushLexState code lexToken -set_file :: Int -> Action -set_file code span buf len = do +setFile :: Int -> Action +setFile code span buf len = do let file = lexemeToFastString (stepOn buf) (len-2) setSrcLoc (mkSrcLoc file (srcSpanEndLine span) (srcSpanEndCol span)) popLexState @@ -953,9 +959,9 @@ lex_escape = do then return (chr (ord c - ord '@')) else lit_error - 'x' -> readNum is_hexdigit 16 hex - 'o' -> readNum is_octdigit 8 oct_or_dec - x | is_digit x -> readNum2 is_digit 10 oct_or_dec (oct_or_dec x) + 'x' -> readNum is_hexdigit 16 hexDigit + 'o' -> readNum is_octdigit 8 octDecDigit + x | is_digit x -> readNum2 is_digit 10 octDecDigit (octDecDigit x) c1 -> do i <- getInput @@ -996,22 +1002,6 @@ readNum2 is_digit base conv i = do then return (chr i) else lit_error -is_hexdigit c - = is_digit c - || (c >= 'a' && c <= 'f') - || (c >= 'A' && c <= 'F') - -hex c | is_digit c = ord c - ord '0' - | otherwise = ord (to_lower c) - ord 'a' + 10 - -oct_or_dec c = ord c - ord '0' - -is_octdigit c = c >= '0' && c <= '7' - -to_lower c - | c >= 'A' && c <= 'Z' = chr (ord c - (ord 'A' - ord 'a')) - | otherwise = c - silly_escape_chars = [ ("NUL", '\NUL'), ("SOH", '\SOH'), @@ -1059,54 +1049,6 @@ getCharOrFail = do Just (c,i) -> do setInput i; return c -- ----------------------------------------------------------------------------- --- Floats - -readRational :: ReadS Rational -- NB: doesn't handle leading "-" -readRational r = do - (n,d,s) <- readFix r - (k,t) <- readExp s - return ((n%1)*10^^(k-d), t) - where - readFix r = do - (ds,s) <- lexDecDigits r - (ds',t) <- lexDotDigits s - return (read (ds++ds'), length ds', t) - - readExp (e:s) | e `elem` "eE" = readExp' s - readExp s = return (0,s) - - readExp' ('+':s) = readDec s - readExp' ('-':s) = do - (k,t) <- readDec s - return (-k,t) - readExp' s = readDec s - - readDec s = do - (ds,r) <- nonnull isDigit s - return (foldl1 (\n d -> n * 10 + d) [ ord d - ord '0' | d <- ds ], - r) - - lexDecDigits = nonnull isDigit - - lexDotDigits ('.':s) = return (span isDigit s) - lexDotDigits s = return ("",s) - - nonnull p s = do (cs@(_:_),t) <- return (span p s) - return (cs,t) - -readRational__ :: String -> Rational -- NB: *does* handle a leading "-" -readRational__ top_s - = case top_s of - '-' : xs -> - (read_me xs) - xs -> read_me xs - where - read_me s - = case (do { (x,"") <- readRational s ; return x }) of - [x] -> x - [] -> error ("readRational__: no parse:" ++ top_s) - _ -> error ("readRational__: ambiguous parse:" ++ top_s) - --- ----------------------------------------------------------------------------- -- The Parse Monad data LayoutContext @@ -1131,7 +1073,10 @@ data PState = PState { lex_state :: [Int] } -- last_loc and last_len are used when generating error messages, - -- and in pushCurrentContext only. + -- and in pushCurrentContext only. Sigh, if only Happy passed the + -- current token to happyError, we could at least get rid of last_len. + -- Getting rid of last_loc would require finding another way to + -- implement pushCurrentContext (which is only called from one place). newtype P a = P { unP :: PState -> ParseResult a } @@ -1215,6 +1160,7 @@ parrBit = 2 arrowsBit = 4 thBit = 5 ipBit = 6 +tvBit = 7 -- Scoped type variables enables 'forall' keyword glaExtsEnabled, ffiEnabled, parrEnabled :: Int -> Bool glaExtsEnabled flags = testBit flags glaExtsBit @@ -1223,6 +1169,7 @@ 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 -- create a parse state -- @@ -1245,6 +1192,7 @@ mkPState buf loc 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 -- setBitIf :: Int -> Bool -> Int b `setBitIf` cond | cond = bit b @@ -1297,7 +1245,7 @@ srcParseErr buf len -- detected during parsing. srcParseFail :: P a srcParseFail = P $ \PState{ buffer = buf, last_len = len, - last_loc = last_loc, loc = loc } -> + last_loc = last_loc } -> PFailed last_loc (srcParseErr buf len) -- A lexical error is reported at a particular position in the source file,