-\end{code}
-
------------------------------------------------------------------------------
-This very disturbing bit of code is used for expanding the tabs in a
-file before we start parsing it. Expanding the tabs early makes the
-lexer a lot simpler: we only have to record the beginning of the line
-in order to be able to calculate the column offset of the current
-token.
-
-We guess the size of the buffer required as 20% extra for
-expanded tabs, and enlarge it if necessary.
-
-\begin{code}
-#if __GLASGOW_HASKELL__ < 303
-mayBlock fo thing = thing
-
-writeCharOffAddr :: Addr -> Int -> Char -> IO ()
-writeCharOffAddr addr off c
- = _casm_ ``*((char *)%0+(int)%1)=(char)%2;'' addr off c
-#endif
-
-getErrType :: IO Int
-#if __GLASGOW_HASKELL__ < 303
-getErrType = _casm_ ``%r = ghc_errtype;''
-#else
-getErrType = _ccall_ getErrType__
-#endif
-
-slurpFileExpandTabs :: FilePath -> IO (Addr,Int)
-slurpFileExpandTabs fname = do
- bracket (openFile fname ReadMode) (hClose)
- (\ handle ->
- do sz <- hFileSize handle
- if sz > toInteger (maxBound::Int)
- then IOERROR (userError "slurpFile: file too big")
- else do
- let sz_i = fromInteger sz
- sz_i' = (sz_i * 12) `div` 10 -- add 20% for tabs
- chunk <- allocMem sz_i'
- trySlurp handle sz_i' chunk
- )
-
-trySlurp :: Handle -> Int -> Addr -> IO (Addr, Int)
-trySlurp handle sz_i chunk =
-#if __GLASGOW_HASKELL__ == 303
- wantReadableHandle "hGetChar" handle >>= \ handle_ ->
- let fo = haFO__ handle_ in
-#elif __GLASGOW_HASKELL__ > 303
- wantReadableHandle "hGetChar" handle $ \ handle_ ->
- let fo = haFO__ handle_ in
-#else
- readHandle handle >>= \ handle_ ->
- let fo = filePtr handle_ in
-#endif
- let
- (I# chunk_sz) = sz_i
-
- tAB_SIZE = 8#
-
- slurpFile :: Int# -> Int# -> Addr -> Int# -> Int# -> IO (Addr, Int)
- slurpFile c off chunk chunk_sz max_off = slurp c off
- where
-
- slurp :: Int# -> Int# -> IO (Addr, Int)
- slurp c off | off >=# max_off = do
- let new_sz = chunk_sz *# 2#
- chunk' <- reAllocMem chunk (I# new_sz)
- slurpFile c off chunk' new_sz (new_sz -# (tAB_SIZE +# 1#))
- slurp c off = do
- intc <- mayBlock fo (_ccall_ fileGetc fo)
- if intc == ((-1)::Int)
- then do errtype <- getErrType
- if errtype == (ERR_EOF :: Int)
- then return (chunk, I# off)
- else constructErrorAndFail "slurpFile"
- else case chr intc of
- '\t' -> tabIt c off
- ch -> do writeCharOffAddr chunk (I# off) ch
- let c' | ch == '\n' = 0#
- | otherwise = c +# 1#
- slurp c' (off +# 1#)
-
- tabIt :: Int# -> Int# -> IO (Addr, Int)
- -- can't run out of buffer in here, because we reserved an
- -- extra tAB_SIZE bytes at the end earlier.
- tabIt c off = do
- writeCharOffAddr chunk (I# off) ' '
- let c' = c +# 1#
- off' = off +# 1#
- if c' `remInt#` tAB_SIZE ==# 0#
- then slurp c' off'
- else tabIt c' off'
- in do
-
- -- allow space for a full tab at the end of the buffer
- -- (that's what the max_off thing is for),
- -- and add 1 to allow room for the final sentinel \NUL at
- -- the end of the file.
- (chunk', rc) <- slurpFile 0# 0# chunk chunk_sz (chunk_sz -# (tAB_SIZE +# 1#))
-#if __GLASGOW_HASKELL__ < 404
- writeHandle handle handle_
-#endif
- if rc < (0::Int)
- then constructErrorAndFail "slurpFile"
- else return (chunk', rc+1 {-room for sentinel-})
-
-
-reAllocMem :: Addr -> Int -> IO Addr
-reAllocMem ptr sz = do
- chunk <- _ccall_ realloc ptr sz
- if chunk == nullAddr
-#if __GLASGOW_HASKELL__ >= 400
- then fail "reAllocMem"
-#else
- then fail (userError "reAllocMem")
-#endif
- else return chunk
-
-allocMem :: Int -> IO Addr
-allocMem sz = do
- chunk <- _ccall_ malloc sz
-#if __GLASGOW_HASKELL__ < 303
- if chunk == nullAddr
- then fail (userError "allocMem")
- else return chunk
-#else
- if chunk == nullAddr
- then constructErrorAndFail "allocMem"
- else return chunk
-#endif
-\end{code}
-
-Lookup
-
-\begin{code}
-currentChar :: StringBuffer -> Char
-currentChar sb = case currentChar# sb of c -> C# c
-
-lookAhead :: StringBuffer -> Int -> Char
-lookAhead sb (I# i#) = case lookAhead# sb i# of c -> C# c
-
-indexSBuffer :: StringBuffer -> Int -> Char
-indexSBuffer sb (I# i#) = case indexSBuffer# sb i# of c -> C# c
-
-currentChar# :: StringBuffer -> Char#
-indexSBuffer# :: StringBuffer -> Int# -> Char#
-lookAhead# :: StringBuffer -> Int# -> Char#
-currentChar# (StringBuffer fo# _ _ current#) = indexCharOffAddr# fo# current#
-indexSBuffer# (StringBuffer fo# _ _ _) i# = indexCharOffAddr# fo# i#
-
- -- relative lookup, i.e, currentChar = lookAhead 0
-lookAhead# (StringBuffer fo# _ _ c#) i# = indexCharOffAddr# fo# (c# +# i#)
-
-currentIndex# :: StringBuffer -> Int#
-currentIndex# (StringBuffer fo# _ _ c#) = c#
-
-lexemeIndex :: StringBuffer -> Int#
-lexemeIndex (StringBuffer fo# _ c# _) = c#
-\end{code}
-
- moving the start point of the current lexeme.
-
-\begin{code}
- -- moving the end point of the current lexeme.
-setCurrentPos# :: StringBuffer -> Int# -> StringBuffer
-setCurrentPos# (StringBuffer fo l# s# c#) i# =
- StringBuffer fo l# s# (c# +# i#)
-
--- augmenting the current lexeme by one.
-incLexeme :: StringBuffer -> StringBuffer
-incLexeme (StringBuffer fo l# s# c#) = StringBuffer fo l# s# (c# +# 1#)
-
-decLexeme :: StringBuffer -> StringBuffer
-decLexeme (StringBuffer fo l# s# c#) = StringBuffer fo l# s# (c# -# 1#)
-
-\end{code}
-
--- move the start and end point of the buffer on by
--- x units.
-
-\begin{code}