X-Git-Url: http://git.megacz.com/?p=ghc-hetmet.git;a=blobdiff_plain;f=compiler%2Fparser%2FLexer.x;h=84ee57eb1c96293ed65820f140524de3584af3d2;hp=be388d4fe17fa10daffb300b37e3d80c90008106;hb=f3399c446c7507d46d6cc550aa2fe7027dbc1b5b;hpb=bcaf1ab28c6525659f40da00d0bd014973646629 diff --git a/compiler/parser/Lexer.x b/compiler/parser/Lexer.x index be388d4..84ee57e 100644 --- a/compiler/parser/Lexer.x +++ b/compiler/parser/Lexer.x @@ -21,6 +21,13 @@ -- - pragma-end should be only valid in a pragma { +{-# OPTIONS -w #-} +-- The above warning supression flag is a temporary kludge. +-- While working on this module you are encouraged to remove it and fix +-- any warnings in the module. See +-- http://hackage.haskell.org/trac/ghc/wiki/Commentary/CodingStyle#Warnings +-- for details + module Lexer ( Token(..), lexer, pragState, mkPState, PState(..), P(..), ParseResult(..), getSrcLoc, @@ -28,7 +35,8 @@ module Lexer ( getMessages, popContext, pushCurrentContext, setLastToken, setSrcLoc, getLexState, popLexState, pushLexState, - extension, glaExtsEnabled, bangPatEnabled + extension, standaloneDerivingEnabled, bangPatEnabled, + addWarning ) where #include "HsVersions.h" @@ -47,7 +55,7 @@ import Util ( maybePrefixMatch, readRational ) import Control.Monad import Data.Bits -import Data.Char ( chr, isSpace ) +import Data.Char ( chr, ord, isSpace ) import Data.Ratio import Debug.Trace @@ -141,7 +149,7 @@ $tab+ { warn Opt_WarnTabs (text "Tab character") } -- space followed by a Haddock comment symbol (docsym) (in which case we'd -- have a Haddock comment). The rules then munch the rest of the line. -"-- " ~$docsym .* ; +"-- " ~[$docsym \#] .* ; "--" [^$symbol : \ ] .* ; -- Next, match Haddock comments if no -haddock flag @@ -202,7 +210,7 @@ $tab+ { warn Opt_WarnTabs (text "Tab character") } -- generate a matching '}' token. () { do_layout_left } -<0,option_prags,glaexts> \n { begin bol } +<0,option_prags> \n { begin bol } "{-#" $whitechar* (line|LINE) { begin line_prag2 } @@ -226,10 +234,10 @@ $tab+ { warn Opt_WarnTabs (text "Tab character") } -- (ToDo: we should really emit a warning when ignoring pragmas) -- XXX Now that we can enable this without the -fglasgow-exts hammer, -- is it better just to let the parse error happen? -<0,glaexts> +<0> "{-#" $whitechar* (RULES|rules) / { ifExtension explicitForallEnabled } { token ITrules_prag } -<0,option_prags,glaexts> { +<0,option_prags> { "{-#" $whitechar* (INLINE|inline) { token (ITinline_prag True) } "{-#" $whitechar* (NO(T?)INLINE|no(t?)inline) { token (ITinline_prag False) } @@ -249,9 +257,6 @@ $tab+ { warn Opt_WarnTabs (text "Tab character") } "{-#" $whitechar* (CORE|core) { token ITcore_prag } "{-#" $whitechar* (UNPACK|unpack) { token ITunpack_prag } - "{-#" $whitechar* (DOCOPTIONS|docoptions) - / { ifExtension haddockEnabled } { lex_string_prag ITdocOptions } - "{-#" { nested_comment lexToken } -- ToDo: should only be valid inside a pragma: @@ -259,36 +264,42 @@ $tab+ { warn Opt_WarnTabs (text "Tab character") } } { - "{-#" $whitechar* (OPTIONS|options) { lex_string_prag IToptions_prag } - "{-#" $whitechar* (OPTIONS_GHC|options_ghc) + "{-#" $whitechar* (OPTIONS|options) { lex_string_prag IToptions_prag } + "{-#" $whitechar* (OPTIONS_GHC|options_ghc) { lex_string_prag IToptions_prag } - "{-#" $whitechar* (LANGUAGE|language) { token ITlanguage_prag } - "{-#" $whitechar* (INCLUDE|include) { lex_string_prag ITinclude_prag } + "{-#" $whitechar* (OPTIONS_HADDOCK|options_haddock) + { lex_string_prag ITdocOptions } + "-- #" { multiline_doc_comment } + "{-#" $whitechar* (LANGUAGE|language) { token ITlanguage_prag } + "{-#" $whitechar* (INCLUDE|include) { lex_string_prag ITinclude_prag } +} + +<0> { + "-- #" .* ; } -<0,option_prags,glaexts> { +<0,option_prags> { -- This is to catch things like {-# OPTIONS OPTIONS_HUGS ... "{-#" $whitechar* $idchar+ { nested_comment lexToken } } -- '0' state: ordinary lexemes --- 'glaexts' state: glasgow extensions (postfix '#', etc.) -- Haddock comments -<0,glaexts> { - "-- " $docsym / { ifExtension haddockEnabled } { multiline_doc_comment } - "{-" \ ? $docsym / { ifExtension haddockEnabled } { nested_doc_comment } +<0> { + "-- " $docsym / { ifExtension haddockEnabled } { multiline_doc_comment } + "{-" \ ? $docsym / { ifExtension haddockEnabled } { nested_doc_comment } } -- "special" symbols -<0,glaexts> { +<0> { "[:" / { ifExtension parrEnabled } { token ITopabrack } ":]" / { ifExtension parrEnabled } { token ITcpabrack } } -<0,glaexts> { +<0> { "[|" / { ifExtension thEnabled } { token ITopenExpQuote } "[e|" / { ifExtension thEnabled } { token ITopenExpQuote } "[p|" / { ifExtension thEnabled } { token ITopenPatQuote } @@ -297,31 +308,34 @@ $tab+ { warn Opt_WarnTabs (text "Tab character") } "|]" / { ifExtension thEnabled } { token ITcloseQuote } \$ @varid / { ifExtension thEnabled } { skip_one_varid ITidEscape } "$(" / { ifExtension thEnabled } { token ITparenEscape } + + "[$" @varid "|" / { ifExtension qqEnabled } + { lex_quasiquote_tok } } -<0,glaexts> { +<0> { "(|" / { ifExtension arrowsEnabled `alexAndPred` notFollowedBySymbol } { special IToparenbar } "|)" / { ifExtension arrowsEnabled } { special ITcparenbar } } -<0,glaexts> { +<0> { \? @varid / { ifExtension ipEnabled } { skip_one_varid ITdupipvarid } } -<0,glaexts> { +<0> { "(#" / { ifExtension unboxedTuplesEnabled `alexAndPred` notFollowedBySymbol } { token IToubxparen } "#)" / { ifExtension unboxedTuplesEnabled } { token ITcubxparen } } -<0,glaexts> { +<0> { "{|" / { ifExtension genericsEnabled } { token ITocurlybar } "|}" / { ifExtension genericsEnabled } { token ITccurlybar } } -<0,option_prags,glaexts> { +<0,option_prags> { \( { special IToparen } \) { special ITcparen } \[ { special ITobrack } @@ -334,21 +348,14 @@ $tab+ { warn Opt_WarnTabs (text "Tab character") } \} { close_brace } } -<0,option_prags,glaexts> { - @qual @varid { check_qvarid } +<0,option_prags> { + @qual @varid { idtoken qvarid } @qual @conid { idtoken qconid } @varid { varid } @conid { idtoken conid } } --- after an illegal qvarid, such as 'M.let', --- we back up and try again in the bad_qvarid state: - { - @conid { pop_and (idtoken conid) } - @qual @conid { pop_and (idtoken qconid) } -} - -<0,glaexts> { +<0> { @qual @varid "#"+ / { ifExtension magicHashEnabled } { idtoken qvarid } @qual @conid "#"+ / { ifExtension magicHashEnabled } { idtoken qconid } @varid "#"+ / { ifExtension magicHashEnabled } { varid } @@ -357,7 +364,7 @@ $tab+ { warn Opt_WarnTabs (text "Tab character") } -- ToDo: M.(,,,) -<0,glaexts> { +<0> { @qual @varsym { idtoken qvarsym } @qual @consym { idtoken qconsym } @varsym { varsym } @@ -366,7 +373,7 @@ $tab+ { warn Opt_WarnTabs (text "Tab character") } -- For the normal boxed literals we need to be careful -- when trying to be close to Haskell98 -<0,glaexts> { +<0> { -- Normal integral literals (:: Num a => a, from Integer) @decimal { tok_num positive 0 0 decimal } 0[oO] @octal { tok_num positive 2 2 octal } @@ -376,7 +383,7 @@ $tab+ { warn Opt_WarnTabs (text "Tab character") } @floating_point { strtoken tok_float } } -<0,glaexts> { +<0> { -- Unboxed ints (:: Int#) -- It's simpler (and faster?) to give separate cases to the negatives, -- especially considering octal/hexadecimal prefixes. @@ -397,7 +404,7 @@ $tab+ { warn Opt_WarnTabs (text "Tab character") } -- 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. -<0,glaexts> { +<0> { \' { lex_char_tok } \" { lex_string_tok } } @@ -413,7 +420,6 @@ data Token | ITdata | ITdefault | ITderiving - | ITderive | ITdo | ITelse | IThiding @@ -447,6 +453,9 @@ data Token | ITdotnet | ITmdo | ITfamily + | ITgroup + | ITby + | ITusing -- Pragmas | ITinline_prag Bool -- True <=> INLINE, False <=> NOINLINE @@ -536,6 +545,7 @@ data Token | ITparenEscape -- $( | ITvarQuote -- ' | ITtyQuote -- '' + | ITquasiQuote (FastString,FastString,SrcSpan) -- [:...|...|] -- Arrow notation extension | ITproc @@ -556,6 +566,7 @@ data Token | ITdocCommentNamed String -- something beginning '-- $' | ITdocSection Int String -- a section heading | ITdocOptions String -- doc options (prune, ignore-exports, etc) + | ITdocOptionsOld String -- doc options declared "-- # ..."-style #ifdef DEBUG deriving Show -- debugging @@ -567,7 +578,6 @@ isSpecial :: Token -> Bool -- not as a keyword. isSpecial ITas = True isSpecial IThiding = True -isSpecial ITderive = True isSpecial ITqualified = True isSpecial ITforall = True isSpecial ITexport = True @@ -580,6 +590,9 @@ isSpecial ITccallconv = True isSpecial ITstdcallconv = True isSpecial ITmdo = True isSpecial ITfamily = True +isSpecial ITgroup = True +isSpecial ITby = True +isSpecial ITusing = True isSpecial _ = False -- the bitmap provided as the third component indicates whether the @@ -598,7 +611,6 @@ reservedWordsFM = listToUFM $ ( "data", ITdata, 0 ), ( "default", ITdefault, 0 ), ( "deriving", ITderiving, 0 ), - ( "derive", ITderive, 0 ), ( "do", ITdo, 0 ), ( "else", ITelse, 0 ), ( "hiding", IThiding, 0 ), @@ -619,9 +631,12 @@ reservedWordsFM = listToUFM $ ( "where", ITwhere, 0 ), ( "_scc_", ITscc, 0 ), -- ToDo: remove - ( "forall", ITforall, bit explicitForallBit), + ( "forall", ITforall, bit explicitForallBit), ( "mdo", ITmdo, bit recursiveDoBit), ( "family", ITfamily, bit tyFamBit), + ( "group", ITgroup, bit transformComprehensionsBit), + ( "by", ITby, bit transformComprehensionsBit), + ( "using", ITusing, bit transformComprehensionsBit), ( "foreign", ITforeign, bit ffiBit), ( "export", ITexport, bit ffiBit), @@ -657,9 +672,7 @@ reservedSymsFM = listToUFM $ ,("!", ITbang, always) -- For data T (a::*) = MkT - ,("*", ITstar, \i -> glaExtsEnabled i || - kindSigsEnabled i || - tyFamEnabled i) + ,("*", ITstar, \i -> kindSigsEnabled i || tyFamEnabled i) -- For 'forall a . t' ,(".", ITdot, explicitForallEnabled) @@ -784,7 +797,7 @@ multiline_doc_comment span buf _len = withLexedDocType (worker "") nested_comment :: P (Located Token) -> Action nested_comment cont span _str _len = do input <- getInput - go 1 input + go (1::Int) input where go 0 input = do setInput input; cont go n input = case alexGetChar input of @@ -824,7 +837,8 @@ withLexedDocType lexDocComment = do '|' -> lexDocComment input ITdocCommentNext False '^' -> lexDocComment input ITdocCommentPrev False '$' -> lexDocComment input ITdocCommentNamed False - '*' -> lexDocSection 1 input + '*' -> lexDocSection 1 input + '#' -> lexDocComment input ITdocOptionsOld False where lexDocSection n input = case alexGetChar input of Just ('*', input) -> lexDocSection (n+1) input @@ -874,30 +888,6 @@ close_brace span _str _len = do popContext return (L span ITccurly) --- We have to be careful not to count M. as a qualified name --- when is a keyword. We hack around this by catching --- the offending tokens afterward, and re-lexing in a different state. -check_qvarid span buf len = do - case lookupUFM reservedWordsFM var of - Just (keyword,exts) - | not (isSpecial keyword) -> - if exts == 0 - then try_again - else do - b <- extension (\i -> exts .&. i /= 0) - if b then try_again - else return token - _other -> return token - where - (mod,var) = splitQualName buf len - token = L span (ITqvarid (mod,var)) - - try_again = do - (AI _ offs _) <- getInput - setInput (AI (srcSpanStart span) (offs-len) buf) - pushLexState bad_qvarid - lexToken - qvarid buf len = ITqvarid $! splitQualName buf len qconid buf len = ITqconid $! splitQualName buf len @@ -1112,8 +1102,8 @@ lex_string s = do Just ('"',i) -> do setInput i - glaexts <- extension glaExtsEnabled - if glaexts + magicHash <- extension magicHashEnabled + if magicHash then do i <- getInput case alexGetChar' i of @@ -1197,9 +1187,9 @@ lex_char_tok span buf len = do -- We've seen ' finish_char_tok :: SrcLoc -> Char -> P (Located Token) finish_char_tok loc ch -- We've already seen the closing quote -- Just need to check for trailing # - = do glaexts <- extension glaExtsEnabled + = do magicHash <- extension magicHashEnabled i@(AI end _ _) <- getInput - if glaexts then do + if magicHash then do case alexGetChar' i of Just ('#',i@(AI end _ _)) -> do setInput i @@ -1332,11 +1322,47 @@ getCharOrFail = do Just (c,i) -> do setInput i; return c -- ----------------------------------------------------------------------------- +-- QuasiQuote + +lex_quasiquote_tok :: Action +lex_quasiquote_tok span buf len = do + let quoter = reverse $ takeWhile (/= '$') + $ reverse $ lexemeToString buf (len - 1) + quoteStart <- getSrcLoc + quote <- lex_quasiquote "" + end <- getSrcLoc + return (L (mkSrcSpan (srcSpanStart span) end) + (ITquasiQuote (mkFastString quoter, + mkFastString (reverse quote), + mkSrcSpan quoteStart end))) + +lex_quasiquote :: String -> P String +lex_quasiquote s = do + i <- getInput + case alexGetChar' i of + Nothing -> lit_error + + Just ('\\',i) + | Just ('|',i) <- next -> do + setInput i; lex_quasiquote ('|' : s) + | Just (']',i) <- next -> do + setInput i; lex_quasiquote (']' : s) + where next = alexGetChar' i + + Just ('|',i) + | Just (']',i) <- next -> do + setInput i; return s + where next = alexGetChar' i + + Just (c, i) -> do + setInput i; lex_quasiquote (c : s) + +-- ----------------------------------------------------------------------------- -- Warnings warn :: DynFlag -> SDoc -> Action -warn option warning span _buf _len = do - addWarning option (mkWarnMsg span alwaysQualify warning) +warn option warning srcspan _buf _len = do + addWarning option srcspan warning lexToken -- ----------------------------------------------------------------------------- @@ -1515,8 +1541,8 @@ getLexState = P $ \s@PState{ lex_state=ls:l } -> POk s ls -- -fglasgow-exts or -fparr) are represented by a bitmap stored in an unboxed -- integer -glaExtsBit, ffiBit, parrBit :: Int -glaExtsBit = 0 +genericsBit, ffiBit, parrBit :: Int +genericsBit = 0 -- {| and |} ffiBit = 1 parrBit = 2 arrowsBit = 4 @@ -1532,11 +1558,13 @@ kindSigsBit = 12 -- Kind signatures on type variables recursiveDoBit = 13 -- mdo unicodeSyntaxBit = 14 -- the forall symbol, arrow symbols, etc unboxedTuplesBit = 15 -- (# and #) -genericsBit = 16 -- {| and |} +standaloneDerivingBit = 16 -- standalone instance deriving declarations +transformComprehensionsBit = 17 +qqBit = 18 -- enable quasiquoting -glaExtsEnabled, ffiEnabled, parrEnabled :: Int -> Bool +genericsEnabled, ffiEnabled, parrEnabled :: Int -> Bool always _ = True -glaExtsEnabled flags = testBit flags glaExtsBit +genericsEnabled flags = testBit flags genericsBit ffiEnabled flags = testBit flags ffiBit parrEnabled flags = testBit flags parrBit arrowsEnabled flags = testBit flags arrowsBit @@ -1551,7 +1579,9 @@ kindSigsEnabled flags = testBit flags kindSigsBit recursiveDoEnabled flags = testBit flags recursiveDoBit unicodeSyntaxEnabled flags = testBit flags unicodeSyntaxBit unboxedTuplesEnabled flags = testBit flags unboxedTuplesBit -genericsEnabled flags = testBit flags genericsBit +standaloneDerivingEnabled flags = testBit flags standaloneDerivingBit +transformComprehensionsEnabled flags = testBit flags transformComprehensionsBit +qqEnabled flags = testBit flags qqBit -- PState for parsing options pragmas -- @@ -1589,15 +1619,16 @@ mkPState buf loc flags = loc = loc, extsBitmap = fromIntegral bitmap, context = [], - lex_state = [bol, if glaExtsEnabled bitmap then glaexts else 0] + lex_state = [bol, 0] -- 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 + bitmap = genericsBit `setBitIf` dopt Opt_Generics flags + .|. ffiBit `setBitIf` dopt Opt_ForeignFunctionInterface flags .|. parrBit `setBitIf` dopt Opt_PArr flags .|. arrowsBit `setBitIf` dopt Opt_Arrows flags - .|. thBit `setBitIf` dopt Opt_TH flags + .|. thBit `setBitIf` dopt Opt_TemplateHaskell flags + .|. qqBit `setBitIf` dopt Opt_QuasiQuotes flags .|. ipBit `setBitIf` dopt Opt_ImplicitParams flags .|. explicitForallBit `setBitIf` dopt Opt_ScopedTypeVariables flags .|. explicitForallBit `setBitIf` dopt Opt_PolymorphicComponents flags @@ -1612,16 +1643,18 @@ mkPState buf loc flags = .|. recursiveDoBit `setBitIf` dopt Opt_RecursiveDo flags .|. unicodeSyntaxBit `setBitIf` dopt Opt_UnicodeSyntax flags .|. unboxedTuplesBit `setBitIf` dopt Opt_UnboxedTuples flags - .|. genericsBit `setBitIf` dopt Opt_Generics flags + .|. standaloneDerivingBit `setBitIf` dopt Opt_StandaloneDeriving flags + .|. transformComprehensionsBit `setBitIf` dopt Opt_TransformListComp flags -- setBitIf :: Int -> Bool -> Int b `setBitIf` cond | cond = bit b | otherwise = 0 -addWarning :: DynFlag -> WarnMsg -> P () -addWarning option w +addWarning :: DynFlag -> SrcSpan -> SDoc -> P () +addWarning option srcspan warning = P $ \s@PState{messages=(ws,es), dflags=d} -> - let ws' = if dopt option d then ws `snocBag` w else ws + let warning' = mkWarnMsg srcspan alwaysQualify warning + ws' = if dopt option d then ws `snocBag` warning' else ws in POk s{messages=(ws', es)} () getMessages :: PState -> Messages