--------------------------------------------------------
\begin{code}
-{-# OPTIONS -#include "ctypes.h" #-}
module Lex (
- ifaceParseErr,
+ ifaceParseErr, srcParseErr,
-- Monad for parser
Token(..), lexer, ParseResult(..), PState(..),
#include "HsVersions.h"
-import Char ( ord, isSpace, toUpper )
+import Char ( isSpace, toUpper )
import List ( isSuffixOf )
-import IdInfo ( InlinePragInfo(..), CprInfo(..) )
-import Name ( isLowerISO, isUpperISO )
-import PrelMods ( mkTupNameStr, mkUbxTupNameStr )
-import CmdLineOpts ( opt_IgnoreIfacePragmas, opt_HiVersion, opt_NoHiCheck )
+import IdInfo ( InlinePragInfo(..) )
+import PrelNames ( mkTupNameStr )
+import CmdLineOpts ( opt_HiVersion, opt_NoHiCheck )
import Demand ( Demand(..) {- instance Read -} )
-import UniqFM ( UniqFM, listToUFM, lookupUFM)
-import BasicTypes ( NewOrData(..) )
+import UniqFM ( listToUFM, lookupUFM )
+import BasicTypes ( NewOrData(..), Boxity(..) )
import SrcLoc ( SrcLoc, incSrcLine, srcLocFile, srcLocLine,
replaceSrcLine, mkSrcLoc )
-import Maybes ( MaybeErr(..) )
import ErrUtils ( Message )
import Outputable
import FastString
import StringBuffer
import GlaExts
-import ST ( runST )
-
-#if __GLASGOW_HASKELL__ >= 303
-import Bits
-import Word
-#endif
-
-import Char ( chr )
-import Addr
+import Ctype
+import Char ( chr, ord )
import PrelRead ( readRational__ ) -- Glasgow non-std
\end{code}
| ITbottom
| ITinteger_lit
| ITfloat_lit
+ | ITword_lit
+ | ITword64_lit
+ | ITint64_lit
| ITrational_lit
| ITaddr_lit
+ | ITlabel_lit
| ITlit_lit
| ITstring_lit
| ITtypeapp
| ITunfold InlinePragInfo
| ITstrict ([Demand], Bool)
| ITrules
- | ITcprinfo (CprInfo)
+ | ITcprinfo
+ | ITdeprecated
| IT__scc
| ITsccAllCafs
| ITinline_prag
| ITnoinline_prag
| ITrules_prag
+ | ITdeprecated_prag
| ITline_prag
| ITclose_prag
| ITpragma StringBuffer
- | ITchar Char
+ | ITchar Int
| ITstring FAST_STRING
- | ITinteger Integer
+ | ITinteger Integer
| ITrational Rational
- | ITprimchar Char
+ | ITprimchar Int
| ITprimstring FAST_STRING
| ITprimint Integer
| ITprimfloat Rational
| ITunknown String -- Used when the lexer can't make sense of it
| ITeof -- end of file token
- deriving Text -- debugging
+ deriving Show -- debugging
\end{code}
-----------------------------------------------------------------------------
( "NOTINLINE", ITnoinline_prag ),
( "LINE", ITline_prag ),
( "RULES", ITrules_prag ),
- ( "RULEZ", ITrules_prag ) -- american spelling :-)
+ ( "RULEZ", ITrules_prag ), -- american spelling :-)
+ ( "DEPRECATED", ITdeprecated_prag )
]
haskellKeywordsFM = listToUFM $
( "_scc_", ITscc )
]
+-- IMPORTANT: Keep this in synch with ParseIface.y's var_fs production! (SUP)
ghcExtensionKeywordsFM = listToUFM $
map (\ (x,y) -> (_PK_ x,y))
[ ( "forall", ITforall ),
("__bot", ITbottom),
("__integer", ITinteger_lit),
("__float", ITfloat_lit),
+ ("__int64", ITint64_lit),
+ ("__word", ITword_lit),
+ ("__word64", ITword64_lit),
("__rational", ITrational_lit),
("__addr", ITaddr_lit),
+ ("__label", ITlabel_lit),
("__litlit", ITlit_lit),
("__string", ITstring_lit),
("__a", ITtypeapp),
("__P", ITspecialise),
("__C", ITnocaf),
("__R", ITrules),
+ ("__D", ITdeprecated),
("__U", ITunfold NoInlinePragInfo),
("__ccall", ITccall (False, False, False)),
if next `eqChar#` '-'# then trundle (n +# 1#)
else if is_symbol next || n <# 2#
then is_a_token
- else case untilChar# (stepOnBy# buf n) '\n'# of
- { buf' -> tab y bol atbol (stepOverLexeme buf')
- }
+ else tab y bol atbol
+ (stepOnUntilChar# (stepOnBy# buf n) '\n'#)
in trundle 1#
-- comments and pragmas. We deal with LINE pragmas here,
let lexeme = mkFastString -- ToDo: too slow
(map toUpper (lexemeToString buf2)) in
case lookupUFM pragmaKeywordsFM lexeme of
- Just ITline_prag -> line_prag (lexer cont) buf2 s'
+ Just ITline_prag ->
+ line_prag skip_to_end buf2 s'
Just other -> is_a_token
- Nothing -> skip_to_end (stepOnBy# buf 2#)
+ Nothing -> skip_to_end (stepOnBy# buf 2#) s'
}}
-
- else skip_to_end (stepOnBy# buf 2#)
+
+ else skip_to_end (stepOnBy# buf 2#) s'
+ where
+ skip_to_end = nested_comment (lexer cont)
+
+ -- special GHC extension: we grok cpp-style #line pragmas
+ '#'# | lexemeIndex buf ==# bol -> -- the '#' must be in column 0
+ line_prag next_line (stepOn buf) s'
where
- skip_to_end buf = nested_comment (lexer cont) buf s'
+ next_line buf = lexer cont (stepOnUntilChar# buf '\n'#)
-- tabs have been expanded beforehand
c | is_space c -> tab y bol atbol (stepOn buf)
| otherwise = lexToken cont glaexts buf s'
-- {-# LINE .. #-} pragmas. yeuch.
-line_prag cont buf =
+line_prag cont buf s@PState{loc=loc} =
case expandWhile# is_space buf of { buf1 ->
case scanNumLit 0 (stepOverLexeme buf1) of { (line,buf2) ->
-- subtract one: the line number refers to the *following* line.
let real_line = line - 1 in
case fromInteger real_line of { i@(I# l) ->
+ -- ToDo, if no filename then we skip the newline.... d'oh
case expandWhile# is_space buf2 of { buf3 ->
case currentChar# buf3 of
'\"'#{-"-} ->
case untilEndOfString# (stepOn (stepOverLexeme buf3)) of { buf4 ->
- let file = lexemeToFastString buf4 in
- \s@PState{loc=loc} -> skipToEnd buf4 s{loc = mkSrcLoc file i}
+ let
+ file = lexemeToFastString buf4
+ new_buf = stepOn (stepOverLexeme buf4)
+ in
+ if nullFastString file
+ then cont new_buf s{loc = replaceSrcLine loc l}
+ else cont new_buf s{loc = mkSrcLoc file i}
}
- other -> \s@PState{loc=loc} -> skipToEnd buf3 s{loc = replaceSrcLine loc l}
+ _other -> cont (stepOverLexeme buf3) s{loc = replaceSrcLine loc l}
}}}}
- where
- skipToEnd buf = nested_comment cont buf
nested_comment :: P a -> P a
nested_comment cont buf = loop buf
_ -> (layoutOff `thenP_` cont ITocurly) (incLexeme buf)
-- strings/characters -------------------------------------------------
- '\"'#{-"-} -> lex_string cont glaexts "" (incLexeme buf)
+ '\"'#{-"-} -> lex_string cont glaexts [] (incLexeme buf)
'\''# -> lex_char (char_end cont) glaexts (incLexeme buf)
-- strictness and cpr pragmas and __scc treated specially.
lex_demand cont (stepOnUntil (not . isSpace)
(stepOnBy# buf 3#)) -- past __S
'M'# ->
- lex_cpr cont (stepOnUntil (not . isSpace)
- (stepOnBy# buf 3#)) -- past __M
+ cont ITcprinfo (stepOnBy# buf 3#) -- past __M
+
's'# ->
case prefixMatch (stepOnBy# buf 3#) "cc" of
Just buf' -> lex_scc cont (stepOverLexeme buf')
cont (ITunknown "\NUL") (stepOn buf)
'?'# | flag glaexts && is_lower (lookAhead# buf 1#) ->
- lex_ip cont (setCurrentPos# buf 1#)
+ lex_ip cont (incLexeme buf)
c | is_digit c -> lex_num cont glaexts 0 buf
| is_symbol c -> lex_sym cont buf
| is_upper c -> lex_con cont glaexts buf
lex_string cont glaexts s buf
= case currentChar# buf of
'"'#{-"-} ->
- let buf' = incLexeme buf; s' = mkFastString (reverse s) in
+ let buf' = incLexeme buf; s' = mkFastStringInt (reverse s) in
case currentChar# buf' of
- '#'# | flag glaexts -> cont (ITprimstring s') (incLexeme buf')
+ '#'# | flag glaexts -> if all (<= 0xFF) s
+ then cont (ITprimstring s') (incLexeme buf')
+ else lexError "primitive string literal must contain only characters <= '\xFF'" buf'
_ -> cont (ITstring s') buf'
-- ignore \& in a string, deal with string gaps
lex_next_string cont s glaexts c buf = lex_string cont glaexts (c:s) buf
-lex_char :: (Int# -> Char -> P a) -> Int# -> P a
+lex_char :: (Int# -> Int -> P a) -> Int# -> P a
lex_char cont glaexts buf
= case currentChar# buf of
'\\'# -> lex_escape (cont glaexts) (incLexeme buf)
- c | is_any c -> cont glaexts (C# c) (incLexeme buf)
+ c | is_any c -> cont glaexts (I# (ord# c)) (incLexeme buf)
other -> charError buf
char_end cont glaexts c buf
lex_escape cont buf
= let buf' = incLexeme buf in
case currentChar# buf of
- 'a'# -> cont '\a' buf'
- 'b'# -> cont '\b' buf'
- 'f'# -> cont '\f' buf'
- 'n'# -> cont '\n' buf'
- 'r'# -> cont '\r' buf'
- 't'# -> cont '\t' buf'
- 'v'# -> cont '\v' buf'
- '\\'# -> cont '\\' buf'
- '"'# -> cont '\"' buf'
- '\''# -> cont '\'' buf'
+ 'a'# -> cont (ord '\a') buf'
+ 'b'# -> cont (ord '\b') buf'
+ 'f'# -> cont (ord '\f') buf'
+ 'n'# -> cont (ord '\n') buf'
+ 'r'# -> cont (ord '\r') buf'
+ 't'# -> cont (ord '\t') buf'
+ 'v'# -> cont (ord '\v') buf'
+ '\\'# -> cont (ord '\\') buf'
+ '"'# -> cont (ord '\"') buf'
+ '\''# -> cont (ord '\'') buf'
'^'# -> let c = currentChar# buf' in
if c `geChar#` '@'# && c `leChar#` '_'#
- then cont (C# (chr# (ord# c -# ord# '@'#))) (incLexeme buf')
+ then cont (I# (ord# c -# ord# '@'#)) (incLexeme buf')
else charError buf'
'x'# -> readNum (after_charnum cont) buf' is_hexdigit 16 hex
_ -> case [ (c,buf2) | (p,c) <- silly_escape_chars,
Just buf2 <- [prefixMatch buf p] ] of
- (c,buf2):_ -> cont c buf2
+ (c,buf2):_ -> cont (ord c) buf2
[] -> charError buf'
-after_charnum cont i buf
- = let int = fromInteger i in
- if i >= 0 && i <= 255
- then cont (chr int) buf
+after_charnum cont i buf
+ = if i >= 0 && i <= 0x7FFFFFFF
+ then cont (fromInteger i) buf
else charError buf
readNum cont buf is_digit base conv = read buf 0
= case read_em [] buf of
(stuff, rest) -> read_em (WwUnpack new_or_data wrapper_unpacks stuff : acc) rest
-lex_cpr cont buf =
- case read_em [] buf of { (cpr_inf,buf') ->
- ASSERT ( null (tail cpr_inf) )
- cont (ITcprinfo $ head cpr_inf) buf'
- }
- where
- -- code snatched from lex_demand above
- read_em acc buf =
- case currentChar# buf of
- '-'# -> read_em (NoCPRInfo : acc) (stepOn buf)
- '('# -> do_unpack acc (stepOn buf)
- ')'# -> (reverse acc, stepOn buf)
- _ -> (reverse acc, buf)
-
- do_unpack acc buf
- = case read_em [] buf of
- (stuff, rest) -> read_em ((CPRInfo stuff) : acc) rest
------------------
lex_scc cont buf =
(mergeLexemes buf buf')
Nothing -> lexError "unterminated ``" buf
-------------------------------------------------------------------------------
--- Character Classes
-
-is_ident, is_symbol, is_any, is_upper, is_digit :: Char# -> Bool
-
-{-# INLINE is_ctype #-}
-#if __GLASGOW_HASKELL__ >= 303
-is_ctype :: Word8 -> Char# -> Bool
-is_ctype mask = \c ->
- (indexWord8OffAddr (``char_types'' :: Addr) (ord (C# c)) .&. mask) /= 0
-#else
-is_ctype :: Int -> Char# -> Bool
-is_ctype (I# mask) = \c ->
- let (A# ctype) = ``char_types'' :: Addr
- flag_word = int2Word# (ord# (indexCharOffAddr# ctype (ord# c)))
- in
- (flag_word `and#` (int2Word# mask)) `neWord#` (int2Word# 0#)
-#endif
-
-is_ident = is_ctype 1
-is_symbol = is_ctype 2
-is_any = is_ctype 4
-is_space = is_ctype 8
-is_lower = is_ctype 16
-is_upper = is_ctype 32
-is_digit = is_ctype 64
-
-----------------------------------------------------------------------------
-- identifiers, symbols etc.
where lexeme = lexemeToFastString buf'
lex_id cont glaexts buf =
- case expandWhile# is_ident buf of { buf1 ->
+ let buf1 = expandWhile# is_ident buf in
+ seq buf1 $
case (if flag glaexts
then expandWhile# (eqChar# '#'#) buf1 -- slurp trailing hashes
cont kwd_token buf';
Nothing ->
- let var_token = cont (mk_var_token lexeme) buf' in
+ let var_token = cont (ITvarid lexeme) buf' in
if not (flag glaexts)
then var_token
Just kwd_token -> cont kwd_token buf';
Nothing -> var_token
- }}}}
+ }}}
lex_sym cont buf =
case expandWhile# is_symbol buf of
_ -> just_a_conid
'-'# -> case lookAhead# buf 1# of
- '>'# -> cont (ITqconid (mod,SLIT("->"))) (setCurrentPos# buf 2#)
+ '>'# -> cont (ITqconid (mod,SLIT("(->)"))) (setCurrentPos# buf 2#)
_ -> lex_id3 cont glaexts mod buf just_a_conid
_ -> lex_id3 cont glaexts mod buf just_a_conid
mk_var_token pk_str
| is_upper f = ITconid pk_str
- -- _[A-Z] is treated as a constructor in interface files.
- | f `eqChar#` '_'# && not (_NULL_ tl)
- && (case _HEAD_ tl of { C# g -> is_upper g }) = ITconid pk_str
| is_ident f = ITvarid pk_str
| f `eqChar#` ':'# = ITconsym pk_str
| otherwise = ITvarsym pk_str
where
(C# f) = _HEAD_ pk_str
- tl = _TAIL_ pk_str
mk_qvar_token m token =
case mk_var_token token of
go n buf =
case currentChar# buf of
','# -> go (n+1) (stepOn buf)
- ')'# -> cont (ITqconid (mod, snd (mkTupNameStr n))) (stepOn buf)
+ ')'# -> cont (ITqconid (mod, snd (mkTupNameStr Boxed n))) (stepOn buf)
_ -> back_off
lex_ubx_tuple cont mod buf back_off =
case currentChar# buf of
','# -> go (n+1) (stepOn buf)
'#'# -> case lookAhead# buf 1# of
- ')'# -> cont (ITqconid (mod, snd (mkUbxTupNameStr n)))
+ ')'# -> cont (ITqconid (mod, snd (mkTupNameStr Unboxed n)))
(stepOnBy# buf 2#)
_ -> back_off
_ -> back_off
POk s{ context = NoLayout:ctx } ()
popContext :: P ()
-popContext = \ buf s@(PState{ context = ctx }) ->
+popContext = \ buf s@(PState{ context = ctx, loc = loc }) ->
case ctx of
(_:tl) -> POk s{ context = tl } ()
- [] -> panic "Lex.popContext: empty context"
+ [] -> PFailed (srcParseErr buf loc)
{-
Note that if the name of the file we're processing ends
Nothing -> ptext SLIT("pre ghc-3.02 version")
Just v -> ptext SLIT("version") <+> integer v
+-----------------------------------------------------------------------------
+
+srcParseErr :: StringBuffer -> SrcLoc -> Message
+srcParseErr s l
+ = hcat [ppr l,
+ if null token
+ then ptext SLIT(": parse error (possibly incorrect indentation)")
+ else hcat [ptext SLIT(": parse error on input "),
+ char '`', text token, char '\'']
+ ]
+ where
+ token = lexemeToString s
+
\end{code}
projects / ghc-hetmet.git / blobdiff