@@ -4,7 +4,6 @@
44-- |
55-- Module : Data.Binary.Bits.Get
66-- Copyright : (c) Lennart Kolmodin 2010-2011
7−-- (c) Sylvain Henry 2015
87-- License : BSD3-style (see LICENSE)
98--
109-- Maintainer : kolmodin@gmail.com
@@ -56,11 +55,12 @@ module Data.Binary.Bits.Get
5655 -- $bitget
5756
5857 BitGet
58+ , BitOrder(..)
5959 , runBitGet
60+ , selectBitOrder
6061
6162 -- ** Get bytes
6263 , getBool
63− , getBool#
6464 , getWord8
6565 , getWord16
6666 , getWord32
@@ -97,9 +97,6 @@ module Data.Binary.Bits.Get
9797
9898import Data.Binary.Get as B ( Get, getLazyByteString, isEmpty )
9999import Data.Binary.Get.Internal as B ( get, put, ensureN )
100−import Data.Binary.Bits.BitOrder
101−import Data.Binary.Bits.Internal
102−import Data.Binary.Bits.Alignment
103100
104101import Data.ByteString as B
105102import qualified Data.ByteString.Lazy as L
@@ -115,8 +112,12 @@ import Control.Applicative
115112import Control.Monad (when,foldM_)
116113
117114import Prelude as P
115+
116+#if defined(__GLASGOW_HASKELL__) && !defined(__HADDOCK__)
117+import GHC.Base
118118import GHC.Word
119−import GHC.Base (uncheckedShiftLWord8#, wordToWord8#, Int (..), andWord8#, Word8#, (-#), Int#, xorI#, eqWord8#, TYPE)
119+#endif
120+
120121
121122-- $bitget
122123-- Parse bits using a monad.
@@ -169,6 +170,16 @@ data S = S {-# UNPACK #-} !ByteString -- Input
169170 !BitOrder -- Bit order
170171 deriving (Show)
171172
173+-- | Compute bit offset (equivalent to x `mod` 8 but faster)
174+bit_offset :: Int -> Int
175+bit_offset n = make_mask 3 .&. n
176+{-# INLINE bit_offset #-}
177+
178+-- | Compute byte offset (equivalent to x `div` 8 but faster)
179+byte_offset :: Int -> Int
180+byte_offset n = n `shiftR` 3
181+{-# INLINE byte_offset #-}
182+
172183-- | Increment the current bit offset
173184incS :: Int -> S -> S
174185incS o (S bs n bo) = S (unsafeDrop d bs) n' bo
@@ -177,70 +188,83 @@ incS o (S bs n bo) = S (unsafeDrop d bs) n' bo
177188 !d = byte_offset o'
178189 !n' = bit_offset o'
179190
191+-- | Select the bit ordering
192+selectBitOrderS :: BitOrder -> S -> S
193+selectBitOrderS bo (S bs o _) = S bs o bo
194+
195+-- | make_mask 3 = 00000111
196+make_mask :: (Bits a, Num a) => Int -> a
197+make_mask n = (1 `shiftL` fromIntegral n) - 1
198+{-# SPECIALIZE make_mask :: Int -> Int #-}
199+{-# SPECIALIZE make_mask :: Int -> Word #-}
200+{-# SPECIALIZE make_mask :: Int -> Word8 #-}
201+{-# SPECIALIZE make_mask :: Int -> Word16 #-}
202+{-# SPECIALIZE make_mask :: Int -> Word32 #-}
203+{-# SPECIALIZE make_mask :: Int -> Word64 #-}
204+
205+-- | Keep only the n least-significant bits of the given value
206+mask :: (Bits a, Num a) => Int -> a -> a
207+mask n v = v .&. make_mask n
208+{-# INLINE mask #-}
209+
210+-- | Bit order
211+--
212+-- E.g. two words of 5 bits: ABCDE, VWXYZ
213+-- - BB: ABCDEVWX YZxxxxxx
214+-- - LB: XYZABCDE xxxxxxVW
215+-- - BL: EDCBAZYX WVxxxxxx -- not implemented
216+-- - LL: XWVEDCBA xxxxxxZY -- not implemented
217+data BitOrder = BB | LB deriving (Show)
218+
219+
180220-- | Read a single bit
181221readBool :: S -> Bool
182222readBool (S bs o bo) = case bo of
183223 BB -> testBit (unsafeHead bs) (7-o)
184− BL -> testBit (unsafeHead bs) (7-o)
185− LL -> testBit (unsafeHead bs) o
186224 LB -> testBit (unsafeHead bs) o
187225
188−-- | Read a single bit
189−readBool# :: S -> Int
190−readBool# (S bs (I# o) bo) = case bo of
191− BB -> testBit# (7# -# o)
192− BL -> testBit# (7# -# o)
193− LL -> testBit# o
194− LB -> testBit# o
195− where
196− !(W8# w) = unsafeHead bs
197− testBit# n = I# (((w `andWord8#` (wordToWord8# 1## `uncheckedShiftLWord8#` n)) `eqWord8#` wordToWord8# 0##) `xorI#` 1#)
198−
199−
200226-- | Extract a range of bits from (ws :: ByteString)
201227--
202228-- Constraint: 8 * (length ws -1 ) < o+n <= 8 * length ws
203−extract :: (Num a, FastBits a) => BitOrder -> ByteString -> Int -> Int -> a
229+extract :: (Num a, Bits a, FastBits a) => BitOrder -> ByteString -> Int -> Int -> a
204230extract bo bs o n
205231 | n == 0 = 0
206232 | B.length bs == 0 = error "Empty ByteString"
207− | otherwise = rev . mask n . foldlWithIndex' f 0 $ bs
233+ | otherwise = mask n (foldlWithIndex' f 0 bs)
208234 where
209235 -- B.foldl' with index
210236 foldlWithIndex' op b = fst . B.foldl' g (b,0)
211237 where g (b',i) w = (op b' w i, (i+1))
212238
213239 -- 'or' correctly shifted words
214− f b w i = b .|. (fromIntegral w `fastShift` off i)
240+ f b w i = b .|. (fromIntegral w `fastShift` off i)
215241
216242 -- co-offset
217− r = B.length bs * 8 - (o + n)
243+ r = B.length bs * 8 - (o + n)
218244
219245 -- shift offset depending on the byte position (0..B.length-1)
220− off i = case bo of
221− LL -> 8*i - o
246+ off i = case bo of
222247 LB -> 8*i - o
223248 BB -> (B.length bs -1 - i) * 8 - r
224− BL -> (B.length bs -1 - i) * 8 - r
225−
226− -- reverse bits if necessary
227− rev = case bo of
228− LB -> reverseBits n
229− BL -> reverseBits n
230− BB -> id
231− LL -> id
232249
233250
234251-- | Generic readWord
235−readWord :: (Num a, FastBits a) => Int -> S -> a
252+readWord :: (Num a, Bits a, FastBits a) => Int -> S -> a
236253readWord n (S bs o bo)
237254 | n == 0 = 0
238− | otherwise = extract bo (unsafeTake nbytes bs) o n
239− where nbytes = byte_offset (o+n+7)
255+ | o+n <= 8 = extract bo (unsafeTake 1 bs) o n
256+ | o+n <= 16 = extract bo (unsafeTake 2 bs) o n
257+ | o+n <= 24 = extract bo (unsafeTake 3 bs) o n
258+ | o+n <= 32 = extract bo (unsafeTake 4 bs) o n
259+ | o+n <= 40 = extract bo (unsafeTake 5 bs) o n
260+ | o+n <= 48 = extract bo (unsafeTake 6 bs) o n
261+ | o+n <= 56 = extract bo (unsafeTake 7 bs) o n
262+ | o+n <= 64 = extract bo (unsafeTake 8 bs) o n
263+ | otherwise = extract bo (unsafeTake ((o+n+7) `div` 8) bs) o n
240264
241265-- | Check that the number of bits to read is not greater than the first parameter
242266{-# INLINE readWordChecked #-}
243−readWordChecked :: (Num a, FastBits a) => Int -> Int -> S -> a
267+readWordChecked :: (Num a, Bits a, FastBits a) => Int -> Int -> S -> a
244268readWordChecked m n s
245269 | n > m = error $ "Tried to read more than " ++ show m ++ " bits (" ++ show n ++")"
246270 | otherwise = readWord n s
@@ -248,37 +272,39 @@ readWordChecked m n s
248272-- | Read the given number of bytes and return them in Big-Endian order
249273--
250274-- Examples:
251−-- BB: xxxABCDE FGHIJKLM NOPxxxxx -> ABCDEFGH IJKLMNOP
252−-- LL: LMNOPxxx DEFGHIJK xxxxxABC -> ABCDEFGH IJKLMNOP
253−-- BL: xxxPONML KJIHGFED CBAxxxxx -> ABCDEFGH IJKLMNOP
254−-- LB: EDCBAxxx MLKJIHGF xxxxxPON -> ABCDEFGH IJKLMNOP
275+-- BB: xxxABCDE FGHIJKLM NOPQRxxx -> ABCDEFGH IJKLMNOP QRxxxxxx
276+-- LB: NOPQRxxx FGHIJKLM xxxABCDE -> ABCDEFGH IJKLMNOP QRxxxxxx
255277readByteString :: Int -> S -> ByteString
256−readByteString n (S bs o bo) =
257− let
258− bs' = unsafeTake (n+1) bs
259− bs'' = unsafeTake n bs
260− rev = B.map (reverseBits 8)
261− in case (o,bo) of
262− (0,BB) -> bs''
263− (0,LL) -> B.reverse bs''
264− (0,LB) -> rev bs''
265− (0,BL) -> rev . B.reverse $ bs''
266− (_,LL) -> readByteString n (S (B.reverse bs') (8-o) BB)
267− (_,BL) -> rev . B.reverse $ readByteString n (S bs' o BB)
268− (_,LB) -> rev . B.reverse $ readByteString n (S bs' o LL)
269− (_,BB) -> unsafePerformIO $ do
270− let len = n+1
271− ptr <- mallocBytes len
272− let f r i = do
273− let
274− w = unsafeIndex bs (len-i)
275− w' = (w `fastShiftL` o) .|. r
276− r' = w `fastShiftR` (8-o)
277− poke (ptr `plusPtr` (len-i)) w'
278− return r'
279− foldM_ f 0 [1..len]
280− unsafeInit <$> unsafePackMallocCStringLen (ptr,len)
281−
278+readByteString n (S bs o bo)
279+ | o == 0 = case bo of
280+ BB -> unsafeTake n bs
281+ LB -> B.reverse (unsafeTake n bs)
282+
283+ | otherwise = unsafePerformIO $ do
284+ let len = n+1
285+ ptr <- mallocBytes len
286+ case bo of
287+ BB -> do
288+ let f r i = do
289+ let
290+ w = unsafeIndex bs (len-i)
291+ w' = (w `fastShiftL` o) .|. r
292+ r' = w `fastShiftR` (8-o)
293+ poke (ptr `plusPtr` (len-i)) w'
294+ return r'
295+ foldM_ f 0 [1..len]
296+ unsafeInit <$> unsafePackMallocCStringLen (ptr,len)
297+
298+ LB -> do
299+ let f r i = do
300+ let
301+ w = unsafeIndex bs (i-1)
302+ w' = (w `fastShiftL` (8-o)) .|. r
303+ r' = w `fastShiftR` o
304+ poke (ptr `plusPtr` (len-i)) w'
305+ return r'
306+ foldM_ f 0 [1..len]
307+ unsafeTail <$> unsafePackMallocCStringLen (ptr,len)
282308
283309------------------------------------------------------------------------
284310-- | 'BitGet' is a monad, applicative and a functor. See 'runBitGet'
@@ -287,6 +313,7 @@ newtype BitGet a = B { runState :: S -> Get (S,a) }
287313
288314instance Monad BitGet where
289315 return x = B $ \s -> return (s,x)
316+ fail str = B $ \s -> putBackState s >> fail str
290317 (B f) >>= g = B $ \s -> do (s',a) <- f s
291318 runState (g a) s'
292319
@@ -297,29 +324,6 @@ instance Applicative BitGet where
297324 pure x = return x
298325 fm <*> m = fm >>= \f -> m >>= \v -> return (f v)
299326
300−instance BitOrderable BitGet where
301− setBitOrder bo = do
302− (S bs o _) <- getState
303− putState (S bs o bo)
304−
305− getBitOrder = do
306− (S _ _ bo) <- getState
307− return bo
308−
309−instance Alignable BitGet where
310− -- | Skip the given number of bits
311− skipBits n = do
312− ensureBits n
313− withState (incS n)
314−
315− -- | Skip bits if necessary to align to the next byte
316− alignByte = do
317− (S _ o _) <- getState
318− when (o /= 0) $
319− skipBits (8-o)
320−
321−
322−
323327-- | Run a 'BitGet' within the Binary packages 'Get' monad. If a byte has
324328-- been partially consumed it will be discarded once 'runBitGet' is finished.
325329runBitGet :: BitGet a -> Get a
@@ -367,6 +371,23 @@ ensureBits n = do
367371 put B.empty
368372 return (S (bs`append`bs') o bo, ())
369373
374+-- | Skip the given number of bits
375+skipBits :: Int -> BitGet ()
376+skipBits n = do
377+ ensureBits n
378+ withState (incS n)
379+
380+-- | Skip bits if necessary to align to the next byte
381+alignByte :: BitGet ()
382+alignByte = do
383+ (S _ o _) <- getState
384+ when (o /= 0) $
385+ skipBits (8-o)
386+
387+-- | Select the bit ordering
388+selectBitOrder :: BitOrder -> BitGet ()
389+selectBitOrder = withState . selectBitOrderS
390+
370391-- | Test whether all input has been consumed, i.e. there are no remaining
371392-- undecoded bytes.
372393isEmpty :: BitGet Bool
@@ -378,10 +399,6 @@ isEmpty = B $ \ (S bs o bo) -> if B.null bs
378399getBool :: BitGet Bool
379400getBool = block bool
380401
381−-- | Get 1 bit as a 'Bool'.
382−getBool# :: BitGet Int
383−getBool# = block bool#
384−
385402-- | Get @n@ bits as a 'Word8'. @n@ must be within @[0..8]@.
386403getWord8 :: Int -> BitGet Word8
387404getWord8 n = block (word8 n)
@@ -457,10 +474,6 @@ block (Block i p) = do
457474bool :: Block Bool
458475bool = Block 1 readBool
459476
460−-- | Read a 1 bit 'Bool'.
461−bool# :: Block Int
462−bool# = Block 1 readBool#
463−
464477-- | Read @n@ bits as a 'Word8'. @n@ must be within @[0..8]@.
465478word8 :: Int -> Block Word8
466479word8 n = Block n (readWordChecked 8 n)
@@ -494,3 +507,84 @@ byteString :: Int -> Block ByteString
494507byteString n | n > 0 = Block (n*8) (readByteString n)
495508 | otherwise = Block 0 (\_ -> B.empty)
496509
510+
511+---------------------------------------------------------------------
512+-- Unchecked shifts, from the "binary" package
513+
514+-- | Class for types supporting fast bit shifting
515+class FastBits a where
516+ fastShiftR :: a -> Int -> a
517+ fastShiftL :: a -> Int -> a
518+
519+ {-# INLINE fastShift #-}
520+ fastShift :: a -> Int -> a
521+ fastShift x n
522+ | n > 0 = fastShiftL x n
523+ | n < 0 = fastShiftR x (negate n)
524+ | otherwise = x
525+
526+instance FastBits Word8 where
527+ fastShiftR = shiftr_w8
528+ fastShiftL = shiftl_w8
529+
530+instance FastBits Word16 where
531+ fastShiftR = shiftr_w16
532+ fastShiftL = shiftl_w16
533+
534+instance FastBits Word32 where
535+ fastShiftR = shiftr_w32
536+ fastShiftL = shiftl_w32
537+
538+instance FastBits Word64 where
539+ fastShiftR = shiftr_w64
540+ fastShiftL = shiftl_w64
541+
542+
543+shiftl_w8 :: Word8 -> Int -> Word8
544+shiftl_w16 :: Word16 -> Int -> Word16
545+shiftl_w32 :: Word32 -> Int -> Word32
546+shiftl_w64 :: Word64 -> Int -> Word64
547+
548+shiftr_w8 :: Word8 -> Int -> Word8
549+shiftr_w16 :: Word16 -> Int -> Word16
550+shiftr_w32 :: Word32 -> Int -> Word32
551+shiftr_w64 :: Word64 -> Int -> Word64
552+
553+#if defined(__GLASGOW_HASKELL__) && !defined(__HADDOCK__)
554+shiftl_w8 (W8# w) (I# i) = W8# (w `uncheckedShiftL#` i)
555+shiftl_w16 (W16# w) (I# i) = W16# (w `uncheckedShiftL#` i)
556+shiftl_w32 (W32# w) (I# i) = W32# (w `uncheckedShiftL#` i)
557+
558+shiftr_w8 (W8# w) (I# i) = W8# (w `uncheckedShiftRL#` i)
559+shiftr_w16 (W16# w) (I# i) = W16# (w `uncheckedShiftRL#` i)
560+shiftr_w32 (W32# w) (I# i) = W32# (w `uncheckedShiftRL#` i)
561+
562+
563+#if WORD_SIZE_IN_BITS < 64
564+shiftl_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftL64#` i)
565+shiftr_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftRL64#` i)
566+
567+#if __GLASGOW_HASKELL__ <= 606
568+-- Exported by GHC.Word in GHC 6.8 and higher
569+foreign import ccall unsafe "stg_uncheckedShiftL64"
570+ uncheckedShiftL64# :: Word64# -> Int# -> Word64#
571+foreign import ccall unsafe "stg_uncheckedShiftRL64"
572+ uncheckedShiftRL64# :: Word64# -> Int# -> Word64#
573+#endif
574+
575+#else
576+shiftl_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftL#` i)
577+shiftr_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftRL#` i)
578+#endif
579+
580+#else
581+shiftl_w8 = shiftL
582+shiftl_w16 = shiftL
583+shiftl_w32 = shiftL
584+shiftl_w64 = shiftL
585+
586+shiftr_w8 = shiftR
587+shiftr_w16 = shiftR
588+shiftr_w32 = shiftR
589+shiftr_w64 = shiftR
590+#endif