Buffers for scanning string input stored in external arrays.
\begin{code}
-{-# OPTIONS -fno-prune-tydecls #-}
+
+{-# OPTIONS -optc-DNON_POSIX_SOURCE #-}
+
module StringBuffer
(
StringBuffer,
- -- creation
- hGetStringBuffer, -- :: FilePath -> IO StringBuffer
+ -- creation/destruction
+ hGetStringBuffer, -- :: FilePath -> IO StringBuffer
+ stringToStringBuffer, -- :: String -> IO StringBuffer
+ freeStringBuffer, -- :: StringBuffer -> IO ()
-- Lookup
currentChar, -- :: StringBuffer -> Char
stepOnBy#, -- :: StringBuffer -> Int# -> StringBuffer
stepOnTo#, -- :: StringBuffer -> Int# -> StringBuffer
stepOnUntil, -- :: (Char -> Bool) -> StringBuffer -> StringBuffer
+ stepOnUntilChar#, -- :: StringBuffer -> Char# -> StringBuffer
stepOverLexeme, -- :: StringBuffer -> StringBuffer
scanNumLit, -- :: Int -> StringBuffer -> (Int, StringBuffer)
squeezeLexeme, -- :: StringBuffer -> Int# -> StringBuffer
-- matching
prefixMatch, -- :: StringBuffer -> String -> Bool
untilEndOfString#, -- :: StringBuffer -> Int#
- untilChar#, -- :: StringBuffer -> Char# -> Int#
-- conversion
lexemeToString, -- :: StringBuffer -> String
#include "HsVersions.h"
-import GlaExts
-import Addr ( Addr(..) )
-import Foreign
-import ST
-import Char ( chr )
-
--- urk!
-#include "../lib/std/cbits/error.h"
-#if __GLASGOW_HASKELL__ >= 303
-import IO ( openFile, slurpFile )
-import PrelIOBase
-import PrelHandle
-import Addr
+#if __GLASGOW_HASKELL__ < 411
+import PrelAddr ( Addr(..) )
+import Panic ( panic )
#else
-import IO ( openFile, hFileSize, hClose, IOMode(..) )
-import Addr
+import Addr ( Addr(..) )
+import Ptr ( Ptr(..) )
#endif
-#if __GLASGOW_HASKELL__ < 301
-import IOBase ( IOError(..), IOErrorType(..) )
-import IOHandle ( readHandle, writeHandle, filePtr )
-import PackBase ( unpackCStringBA )
+#if __GLASGOW_HASKELL__ < 501
+import Char ( chr )
+#elif __GLASGOW_HASKELL__ < 503
+import PrelIO ( hGetcBuffered )
#else
-# if __GLASGOW_HASKELL__ <= 302
-import PrelIOBase ( Handle, IOError(..), IOErrorType(..),
- constructErrorAndFail )
-import PrelHandle ( readHandle, writeHandle, filePtr )
-# endif
-import PrelPack ( unpackCStringBA )
+import GHC.IO ( hGetcBuffered )
#endif
-#if __GLASGOW_HASKELL__ < 402
-import Util ( bracket )
-#else
+import PrimPacked
+import FastString
+
+import GlaExts
+import Foreign
+import IO ( openFile, isEOFError )
+import IOExts ( slurpFile )
+import Addr
import Exception ( bracket )
+
+import CString ( unpackCStringBA )
+
+#if __GLASGOW_HASKELL__ < 503
+import PrelIOBase
+import PrelHandle
+#else
+import GHC.IOBase
+import GHC.Handle
#endif
-import PrimPacked
-import FastString
-import Char (isDigit)
+import Char ( isDigit )
\end{code}
\begin{code}
\end{code}
\begin{code}
-instance Text StringBuffer where
+instance Show StringBuffer where
showsPrec _ s = showString ""
\end{code}
hGetStringBuffer expand_tabs fname = do
(a, read) <- if expand_tabs
then slurpFileExpandTabs fname
+#if __GLASGOW_HASKELL__ < 411
else slurpFile fname
+#else
+ else do
+ (Ptr a#, read) <- slurpFile fname
+ return (A# a#, read)
+#endif
- let (A# a#) = a; (I# read#) = read
+ -- urk! slurpFile gives us a buffer that doesn't have room for
+ -- the sentinel. Assume it has a final newline for now, and overwrite
+ -- that with the sentinel. slurpFileExpandTabs (below) leaves room
+ -- for the sentinel.
+ let (A# a#) = a;
+ (I# read#) = read;
+ end# = read# -# 1#
-- add sentinel '\NUL'
- _casm_ `` ((char *)%0)[(int)%1]=(char)0; '' (A# a#) (I# (read# -# 1#))
- return (StringBuffer a# read# 0# 0#)
-
-#if __GLASGOW_HASKELL__ < 303
-slurpFile fname =
- openFile fname ReadMode >>= \ hndl ->
- hFileSize hndl >>= \ len ->
- let len_i = fromInteger len in
- -- Allocate an array for system call to store its bytes into.
- -- ToDo: make it robust
--- trace (show ((len_i::Int)+1)) $
- _casm_ `` %r=(char *)malloc(sizeof(char)*(int)%0); '' (len_i::Int) >>= \ arr@(A# a#) ->
- if addr2Int# a# ==# 0# then
- fail (userError ("hGetStringBuffer: Could not allocate "++show len_i ++ " bytes"))
- else
- readHandle hndl >>= \ hndl_ ->
- writeHandle hndl hndl_ >>
- let ptr = filePtr hndl_ in
-#if __GLASGOW_HASKELL__ <= 302
- _ccall_ fread arr (1::Int) len_i (ptr::ForeignObj) >>= \ (I# read#) ->
-#else
- _ccall_ fread arr (1::Int) len_i (ptr::Addr) >>= \ (I# read#) ->
-#endif
- hClose hndl >>
- if read# ==# 0# then -- EOF or some other error
- fail (userError ("hGetStringBuffer: failed to slurp in interface file "++fname))
- else
- return (arr, I# read#)
-#endif
+ _casm_ `` ((char *)%0)[(int)%1]=(char)0; '' (A# a#) (I# end#)
+ return (StringBuffer a# end# 0# 0#)
unsafeWriteBuffer :: StringBuffer -> Int# -> Char# -> StringBuffer
unsafeWriteBuffer s@(StringBuffer a _ _ _) i# ch# =
\end{code}
-----------------------------------------------------------------------------
+-- Turn a String into a StringBuffer
+
+\begin{code}
+stringToStringBuffer :: String -> IO StringBuffer
+freeStringBuffer :: StringBuffer -> IO ()
+
+#if __GLASGOW_HASKELL__ >= 411
+stringToStringBuffer str =
+ do let sz@(I# sz#) = length str
+ (Ptr a#) <- mallocBytes (sz+1)
+ fill_in str (A# a#)
+ writeCharOffAddr (A# a#) sz '\0' -- sentinel
+ return (StringBuffer a# sz# 0# 0#)
+ where
+ fill_in [] _ = return ()
+ fill_in (c:cs) a = do
+ writeCharOffAddr a 0 c
+ fill_in cs (a `plusAddr` 1)
+
+freeStringBuffer (StringBuffer a# _ _ _) = Foreign.free (Ptr a#)
+#else
+stringToStringBuffer = panic "stringToStringBuffer: not implemented"
+freeStringBuffer sb = return ()
+#endif
+
+\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
expanded tabs, and enlarge it if necessary.
\begin{code}
-#if __GLASGOW_HASKELL__ < 303
-ioError = fail
-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
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
+ if sz_i == 0
+ -- empty file: just allocate a buffer containing '\0'
+ then do chunk <- allocMem 1
+ writeCharOffAddr chunk 0 '\0'
+ return (chunk, 0)
+ else do let 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
+#if __GLASGOW_HASKELL__ < 501
wantReadableHandle "hGetChar" handle $ \ handle_ ->
let fo = haFO__ handle_ in
#else
- readHandle handle >>= \ handle_ ->
- let fo = filePtr handle_ in
+ wantReadableHandle "hGetChar" handle $
+ \ handle_@Handle__{ haFD=fd, haBuffer=ref, haBufferMode=mode } ->
#endif
let
(I# chunk_sz) = sz_i
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)
+ slurpFile c off chunk' new_sz (new_sz -# (tAB_SIZE +# 1#))
slurp c off = do
+#if __GLASGOW_HASKELL__ < 501
intc <- mayBlock fo (_ccall_ fileGetc fo)
if intc == ((-1)::Int)
then do errtype <- getErrType
- if errtype == (ERR_EOF :: Int)
+ if errtype == (19{-ERR_EOF-} :: Int)
then return (chunk, I# off)
else constructErrorAndFail "slurpFile"
else case chr intc of
+#else
+ buf <- readIORef ref
+ ch <- (if not (bufferEmpty buf)
+ then hGetcBuffered fd ref buf
+ else do
+#if __GLASGOW_HASKELL__ >= 503
+ new_buf <- fillReadBuffer fd True False buf
+#else
+ new_buf <- fillReadBuffer fd True buf
+#endif
+ hGetcBuffered fd ref new_buf)
+ `catch` \e -> if isEOFError e
+ then return '\xFFFF'
+ else ioError e
+ case ch of
+ '\xFFFF' -> return (chunk, I# off)
+#endif
'\t' -> tabIt c off
ch -> do writeCharOffAddr chunk (I# off) ch
let c' | ch == '\n' = 0#
in do
-- allow space for a full tab at the end of the buffer
- -- (that's what the max_off thing is for)
- (chunk', rc) <- slurpFile 0# 0# chunk chunk_sz (chunk_sz -# tAB_SIZE)
+ -- (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_
- if rc < (0::Int)
- then constructErrorAndFail "slurpFile"
- else return (chunk', rc)
+#endif
+ 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__ < 303
- then fail (userError "reAllocMem")
-#else
then fail "reAllocMem"
-#endif
else return chunk
allocMem :: Int -> IO Addr
allocMem sz = do
-#if __GLASGOW_HASKELL__ < 303
chunk <- _ccall_ malloc sz
if chunk == nullAddr
- then fail (userError "allocMem")
- else return chunk
-#else
- chunk <- _ccall_ allocMemory__ sz
- if chunk == nullAddr
+#if __GLASGOW_HASKELL__ < 501
then constructErrorAndFail "allocMem"
- else return chunk
+#else
+ then ioException (IOError Nothing ResourceExhausted "malloc"
+ "out of memory" Nothing)
#endif
+ else return chunk
\end{code}
Lookup
| otherwise -> (acc,StringBuffer fo l# s# c#)
-expandUntilMatch :: StringBuffer -> String -> StringBuffer
+expandUntilMatch :: StringBuffer -> String -> Maybe StringBuffer
expandUntilMatch (StringBuffer fo l# s# c#) str =
loop c# str
where
- loop c# [] = StringBuffer fo l# s# c#
- loop c# ((C# x#):xs)
- | indexCharOffAddr# fo c# `eqChar#` x#
- = loop (c# +# 1#) xs
- | otherwise
- = loop (c# +# 1#) str
+ loop c# [] = Just (StringBuffer fo l# s# c#)
+ loop c# ((C# x#):xs) =
+ case indexCharOffAddr# fo c# of
+ ch# | ch# `eqChar#` '\NUL'# && c# >=# l# -> Nothing
+ | ch# `eqChar#` x# -> loop (c# +# 1#) xs
+ | otherwise -> loop (c# +# 1#) str
\end{code}
_ -> loop (c# +# 1#)
-untilChar# :: StringBuffer -> Char# -> StringBuffer
-untilChar# (StringBuffer fo l# s# c#) x# =
+stepOnUntilChar# :: StringBuffer -> Char# -> StringBuffer
+stepOnUntilChar# (StringBuffer fo l# s# c#) x# =
loop c#
where
loop c#
| c# >=# l# || indexCharOffAddr# fo c# `eqChar#` x#
- = StringBuffer fo l# s# c#
+ = StringBuffer fo l# c# c#
| otherwise
= loop (c# +# 1#)
else
unpackCStringBA (copySubStr (A# fo) (I# start_pos#) (I# (current# -# start_pos#)))
-lexemeToByteArray :: StringBuffer -> _ByteArray Int
+lexemeToByteArray :: StringBuffer -> ByteArray Int
lexemeToByteArray (StringBuffer fo _ start_pos# current#) =
if start_pos# ==# current# then
error "lexemeToByteArray"
projects / ghc-hetmet.git / blobdiff