@@ -4,7 +4,6 @@
5-- Module : Data.Binary.Bits.Get
5-- Module : Data.Binary.Bits.Get
6-- Copyright : (c) Lennart Kolmodin 2010-2011
6-- Copyright : (c) Lennart Kolmodin 2010-2011
7−-- (c) Sylvain Henry 2015
8-- License : BSD3-style (see LICENSE)
7-- License : BSD3-style (see LICENSE)
10-- Maintainer : kolmodin@gmail.com
9-- Maintainer : kolmodin@gmail.com
@@ -56,11 +55,13 @@ module Data.Binary.Bits.Get
56 -- $bitget
55 -- $bitget
59 , runBitGet
59 , runBitGet
61 -- ** Get bytes
63 -- ** Get bytes
64 , getWord8
65 , getWord8
65 , getWord16
66 , getWord16
66 , getWord32
67 , getWord32
@@ -97,9 +98,6 @@ module Data.Binary.Bits.Get
98import Data.Binary.Get as B ( Get, getLazyByteString, isEmpty )
99import Data.Binary.Get as B ( Get, getLazyByteString, isEmpty )
99import Data.Binary.Get.Internal as B ( get, put, ensureN )
100import 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
104import Data.ByteString as B
102import Data.ByteString as B
105import qualified Data.ByteString.Lazy as L
103import qualified Data.ByteString.Lazy as L
@@ -115,8 +113,12 @@ import Control.Applicative
115import Control.Monad (when,foldM_)
113import Control.Monad (when,foldM_)
117import Prelude as P
115import Prelude as P
117+#if defined(__GLASGOW_HASKELL__) && !defined(__HADDOCK__)
118import GHC.Word
119import GHC.Word
119−import GHC.Base (uncheckedShiftLWord8#, wordToWord8#, Int (..), andWord8#, Word8#, (-#), Int#, xorI#, eqWord8#, TYPE)
120+#endif
121-- $bitget
123-- $bitget
122-- Parse bits using a monad.
124-- Parse bits using a monad.
@@ -169,6 +171,16 @@ data S = S {-# UNPACK #-} !ByteString -- Input
169 !BitOrder -- Bit order
171 !BitOrder -- Bit order
170 deriving (Show)
172 deriving (Show)
174+-- | Compute bit offset (equivalent to x `mod` 8 but faster)
175+bit_offset :: Int -> Int
176+bit_offset n = make_mask 3 .&. n
177+{-# INLINE bit_offset #-}
179+-- | Compute byte offset (equivalent to x `div` 8 but faster)
180+byte_offset :: Int -> Int
181+byte_offset n = n `shiftR` 3
182+{-# INLINE byte_offset #-}
172-- | Increment the current bit offset
184-- | Increment the current bit offset
173incS :: Int -> S -> S
185incS :: Int -> S -> S
174incS o (S bs n bo) = S (unsafeDrop d bs) n' bo
186incS o (S bs n bo) = S (unsafeDrop d bs) n' bo
@@ -177,70 +189,83 @@ incS o (S bs n bo) = S (unsafeDrop d bs) n' bo
177 !d = byte_offset o'
189 !d = byte_offset o'
178 !n' = bit_offset o'
190 !n' = bit_offset o'
192+-- | Select the bit ordering
193+selectBitOrderS :: BitOrder -> S -> S
194+selectBitOrderS bo (S bs o _) = S bs o bo
196+-- | make_mask 3 = 00000111
197+make_mask :: (Bits a, Num a) => Int -> a
198+make_mask n = (1 `shiftL` fromIntegral n) - 1
199+{-# SPECIALIZE make_mask :: Int -> Int #-}
200+{-# SPECIALIZE make_mask :: Int -> Word #-}
201+{-# SPECIALIZE make_mask :: Int -> Word8 #-}
202+{-# SPECIALIZE make_mask :: Int -> Word16 #-}
203+{-# SPECIALIZE make_mask :: Int -> Word32 #-}
204+{-# SPECIALIZE make_mask :: Int -> Word64 #-}
206+-- | Keep only the n least-significant bits of the given value
207+mask :: (Bits a, Num a) => Int -> a -> a
208+mask n v = v .&. make_mask n
213+-- E.g. two words of 5 bits: ABCDE, VWXYZ
214+-- - BB: ABCDEVWX YZxxxxxx
215+-- - LB: XYZABCDE xxxxxxVW
216+-- - BL: EDCBAZYX WVxxxxxx -- not implemented
217+-- - LL: XWVEDCBA xxxxxxZY -- not implemented
218+data BitOrder = BB | LB deriving (Show)
180-- | Read a single bit
221-- | Read a single bit
181readBool :: S -> Bool
222readBool :: S -> Bool
182readBool (S bs o bo) = case bo of
223readBool (S bs o bo) = case bo of
183 BB -> testBit (unsafeHead bs) (7-o)
224 BB -> testBit (unsafeHead bs) (7-o)
184− BL -> testBit (unsafeHead bs) (7-o)
185− LL -> testBit (unsafeHead bs) o
186 LB -> testBit (unsafeHead bs) o
225 LB -> testBit (unsafeHead bs) o
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)
196− !(W8# w) = unsafeHead bs
197− testBit# n = I# (((w `andWord8#` (wordToWord8# 1## `uncheckedShiftLWord8#` n)) `eqWord8#` wordToWord8# 0##) `xorI#` 1#)
200-- | Extract a range of bits from (ws :: ByteString)
227-- | Extract a range of bits from (ws :: ByteString)
202-- Constraint: 8 * (length ws -1 ) < o+n <= 8 * length ws
229-- Constraint: 8 * (length ws -1 ) < o+n <= 8 * length ws
203−extract :: (Num a, FastBits a) => BitOrder -> ByteString -> Int -> Int -> a
230+extract :: (Num a, Bits a, FastBits a) => BitOrder -> ByteString -> Int -> Int -> a
204extract bo bs o n
231extract bo bs o n
205 | n == 0 = 0
232 | n == 0 = 0
206 | B.length bs == 0 = error "Empty ByteString"
233 | B.length bs == 0 = error "Empty ByteString"
207− | otherwise = rev . mask n . foldlWithIndex' f 0 $ bs
234+ | otherwise = mask n (foldlWithIndex' f 0 bs)
209 -- B.foldl' with index
236 -- B.foldl' with index
210 foldlWithIndex' op b = fst . B.foldl' g (b,0)
237 foldlWithIndex' op b = fst . B.foldl' g (b,0)
211 where g (b',i) w = (op b' w i, (i+1))
238 where g (b',i) w = (op b' w i, (i+1))
213 -- 'or' correctly shifted words
240 -- 'or' correctly shifted words
214− f b w i = b .|. (fromIntegral w `fastShift` off i)
241+ f b w i = b .|. (fromIntegral w `fastShift` off i)
216 -- co-offset
243 -- co-offset
217− r = B.length bs * 8 - (o + n)
244+ r = B.length bs * 8 - (o + n)
219 -- shift offset depending on the byte position (0..B.length-1)
246 -- shift offset depending on the byte position (0..B.length-1)
220− off i = case bo of
247+ off i = case bo of
222 LB -> 8*i - o
248 LB -> 8*i - o
223 BB -> (B.length bs -1 - i) * 8 - r
249 BB -> (B.length bs -1 - i) * 8 - r
224− BL -> (B.length bs -1 - i) * 8 - r
226− -- reverse bits if necessary
234-- | Generic readWord
252-- | Generic readWord
235−readWord :: (Num a, FastBits a) => Int -> S -> a
253+readWord :: (Num a, Bits a, FastBits a) => Int -> S -> a
236readWord n (S bs o bo)
254readWord n (S bs o bo)
237 | n == 0 = 0
255 | n == 0 = 0
256+ | o+n <= 8 = extract bo (unsafeTake 1 bs) o n
257+ | o+n <= 16 = extract bo (unsafeTake 2 bs) o n
258+ | o+n <= 24 = extract bo (unsafeTake 3 bs) o n
259+ | o+n <= 32 = extract bo (unsafeTake 4 bs) o n
260+ | o+n <= 40 = extract bo (unsafeTake 5 bs) o n
261+ | o+n <= 48 = extract bo (unsafeTake 6 bs) o n
262+ | o+n <= 56 = extract bo (unsafeTake 7 bs) o n
263+ | o+n <= 64 = extract bo (unsafeTake 8 bs) o n
238− | otherwise = extract bo (unsafeTake nbytes bs) o n
264+ | otherwise = extract bo (unsafeTake ((o+n+7) `div` 8) bs) o n
239− where nbytes = byte_offset (o+n+7)
241-- | Check that the number of bits to read is not greater than the first parameter
266-- | Check that the number of bits to read is not greater than the first parameter
242{-# INLINE readWordChecked #-}
267{-# INLINE readWordChecked #-}
243−readWordChecked :: (Num a, FastBits a) => Int -> Int -> S -> a
268+readWordChecked :: (Num a, Bits a, FastBits a) => Int -> Int -> S -> a
244readWordChecked m n s
269readWordChecked m n s
245 | n > m = error $ "Tried to read more than " ++ show m ++ " bits (" ++ show n ++")"
270 | n > m = error $ "Tried to read more than " ++ show m ++ " bits (" ++ show n ++")"
246 | otherwise = readWord n s
271 | otherwise = readWord n s
@@ -248,37 +273,39 @@ readWordChecked m n s
248-- | Read the given number of bytes and return them in Big-Endian order
273-- | Read the given number of bytes and return them in Big-Endian order
250-- Examples:
275-- Examples:
251−-- BB: xxxABCDE FGHIJKLM NOPxxxxx -> ABCDEFGH IJKLMNOP
276+-- BB: xxxABCDE FGHIJKLM NOPQRxxx -> ABCDEFGH IJKLMNOP QRxxxxxx
252−-- LL: LMNOPxxx DEFGHIJK xxxxxABC -> ABCDEFGH IJKLMNOP
277+-- LB: NOPQRxxx FGHIJKLM xxxABCDE -> ABCDEFGH IJKLMNOP QRxxxxxx
253−-- BL: xxxPONML KJIHGFED CBAxxxxx -> ABCDEFGH IJKLMNOP
254−-- LB: EDCBAxxx MLKJIHGF xxxxxPON -> ABCDEFGH IJKLMNOP
255readByteString :: Int -> S -> ByteString
278readByteString :: Int -> S -> ByteString
256−readByteString n (S bs o bo) =
279+readByteString n (S bs o bo)
258− bs' = unsafeTake (n+1) bs
259− bs'' = unsafeTake n bs
260− rev = B.map (reverseBits 8)
261− in case (o,bo) of
280+ | o == 0 = case bo of
262− (0,BB) -> bs''
281+ BB -> unsafeTake n bs
263− (0,LL) -> B.reverse bs''
265− (0,BL) -> rev . B.reverse $ bs''
282+ LB -> B.reverse (unsafeTake n bs)
266− (_,LL) -> readByteString n (S (B.reverse bs') (8-o) BB)
283+
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
284+ | otherwise = unsafePerformIO $ do
270− let len = n+1
285+ let len = n+1
271− ptr <- mallocBytes len
286+ ptr <- mallocBytes len
272− let f r i = do
289+ let f r i = do
274− w = unsafeIndex bs (len-i)
291+ w = unsafeIndex bs (len-i)
275− w' = (w `fastShiftL` o) .|. r
292+ w' = (w `fastShiftL` o) .|. r
276− r' = w `fastShiftR` (8-o)
293+ r' = w `fastShiftR` (8-o)
277− poke (ptr `plusPtr` (len-i)) w'
294+ poke (ptr `plusPtr` (len-i)) w'
278− return r'
295+ return r'
279− foldM_ f 0 [1..len]
296+ foldM_ f 0 [1..len]
280− unsafeInit <$> unsafePackMallocCStringLen (ptr,len)
297+ unsafeInit <$> unsafePackMallocCStringLen (ptr,len)
302+ w = unsafeIndex bs (i-1)
303+ w' = (w `fastShiftL` (8-o)) .|. r
304+ r' = w `fastShiftR` o
305+ poke (ptr `plusPtr` (len-i)) w'
308+ unsafeTail <$> unsafePackMallocCStringLen (ptr,len)
283------------------------------------------------------------------------
310------------------------------------------------------------------------
284-- | 'BitGet' is a monad, applicative and a functor. See 'runBitGet'
311-- | 'BitGet' is a monad, applicative and a functor. See 'runBitGet'
@@ -287,6 +314,7 @@ newtype BitGet a = B { runState :: S -> Get (S,a) }
288instance Monad BitGet where
315instance Monad BitGet where
289 return x = B $ \s -> return (s,x)
316 return x = B $ \s -> return (s,x)
317+ fail str = B $ \s -> putBackState s >> fail str
290 (B f) >>= g = B $ \s -> do (s',a) <- f s
318 (B f) >>= g = B $ \s -> do (s',a) <- f s
291 runState (g a) s'
319 runState (g a) s'
@@ -297,29 +325,6 @@ instance Applicative BitGet where
297 pure x = return x
325 pure x = return x
298 fm <*> m = fm >>= \f -> m >>= \v -> return (f v)
326 fm <*> m = fm >>= \f -> m >>= \v -> return (f v)
300−instance BitOrderable BitGet where
302− (S bs o _) <- getState
303− putState (S bs o bo)
306− (S _ _ bo) <- getState
309−instance Alignable BitGet where
310− -- | Skip the given number of bits
315− -- | Skip bits if necessary to align to the next byte
317− (S _ o _) <- getState
323-- | Run a 'BitGet' within the Binary packages 'Get' monad. If a byte has
328-- | Run a 'BitGet' within the Binary packages 'Get' monad. If a byte has
324-- been partially consumed it will be discarded once 'runBitGet' is finished.
329-- been partially consumed it will be discarded once 'runBitGet' is finished.
325runBitGet :: BitGet a -> Get a
330runBitGet :: BitGet a -> Get a
@@ -354,6 +359,13 @@ withState f = do
354 s <- getState
359 s <- getState
355 putState $! f s
360 putState $! f s
362+-- | Get the bit ordering
363+getBitOrder :: BitGet BitOrder
365+ (S _ _ bo) <- getState
357-- | Make sure there are at least @n@ bits.
369-- | Make sure there are at least @n@ bits.
358ensureBits :: Int -> BitGet ()
370ensureBits :: Int -> BitGet ()
359ensureBits n = do
371ensureBits n = do
@@ -367,6 +379,33 @@ ensureBits n = do
367 put B.empty
379 put B.empty
368 return (S (bs`append`bs') o bo, ())
380 return (S (bs`append`bs') o bo, ())
382+-- | Skip the given number of bits
383+skipBits :: Int -> BitGet ()
388+-- | Skip bits if necessary to align to the next byte
389+alignByte :: BitGet ()
391+ (S _ o _) <- getState
395+-- | Select the bit ordering
396+selectBitOrder :: BitOrder -> BitGet ()
397+selectBitOrder = withState . selectBitOrderS
399+withBitOrder :: BitOrder -> BitGet a -> BitGet a
400+withBitOrder bo f = do
370-- | Test whether all input has been consumed, i.e. there are no remaining
409-- | Test whether all input has been consumed, i.e. there are no remaining
371-- undecoded bytes.
410-- undecoded bytes.
372isEmpty :: BitGet Bool
411isEmpty :: BitGet Bool
@@ -378,10 +417,6 @@ isEmpty = B $ \ (S bs o bo) -> if B.null bs
378getBool :: BitGet Bool
417getBool :: BitGet Bool
379getBool = block bool
418getBool = block bool
381−-- | Get 1 bit as a 'Bool'.
382−getBool# :: BitGet Int
383−getBool# = block bool#
385-- | Get @n@ bits as a 'Word8'. @n@ must be within @[0..8]@.
420-- | Get @n@ bits as a 'Word8'. @n@ must be within @[0..8]@.
386getWord8 :: Int -> BitGet Word8
421getWord8 :: Int -> BitGet Word8
387getWord8 n = block (word8 n)
422getWord8 n = block (word8 n)
@@ -457,10 +492,6 @@ block (Block i p) = do
457bool :: Block Bool
492bool :: Block Bool
458bool = Block 1 readBool
493bool = Block 1 readBool
460−-- | Read a 1 bit 'Bool'.
462−bool# = Block 1 readBool#
464-- | Read @n@ bits as a 'Word8'. @n@ must be within @[0..8]@.
495-- | Read @n@ bits as a 'Word8'. @n@ must be within @[0..8]@.
465word8 :: Int -> Block Word8
496word8 :: Int -> Block Word8
466word8 n = Block n (readWordChecked 8 n)
497word8 n = Block n (readWordChecked 8 n)
@@ -494,3 +525,84 @@ byteString :: Int -> Block ByteString
494byteString n | n > 0 = Block (n*8) (readByteString n)
525byteString n | n > 0 = Block (n*8) (readByteString n)
495 | otherwise = Block 0 (\_ -> B.empty)
526 | otherwise = Block 0 (\_ -> B.empty)
529+---------------------------------------------------------------------
530+-- Unchecked shifts, from the "binary" package
532+-- | Class for types supporting fast bit shifting
533+class FastBits a where
534+ fastShiftR :: a -> Int -> a
535+ fastShiftL :: a -> Int -> a
537+ {-# INLINE fastShift #-}
538+ fastShift :: a -> Int -> a
540+ | n > 0 = fastShiftL x n
541+ | n < 0 = fastShiftR x (negate n)
544+instance FastBits Word8 where
545+ fastShiftR = shiftr_w8
546+ fastShiftL = shiftl_w8
548+instance FastBits Word16 where
549+ fastShiftR = shiftr_w16
550+ fastShiftL = shiftl_w16
552+instance FastBits Word32 where
553+ fastShiftR = shiftr_w32
554+ fastShiftL = shiftl_w32
556+instance FastBits Word64 where
557+ fastShiftR = shiftr_w64
558+ fastShiftL = shiftl_w64
561+shiftl_w8 :: Word8 -> Int -> Word8
562+shiftl_w16 :: Word16 -> Int -> Word16
563+shiftl_w32 :: Word32 -> Int -> Word32
564+shiftl_w64 :: Word64 -> Int -> Word64
566+shiftr_w8 :: Word8 -> Int -> Word8
567+shiftr_w16 :: Word16 -> Int -> Word16
568+shiftr_w32 :: Word32 -> Int -> Word32
569+shiftr_w64 :: Word64 -> Int -> Word64
571+#if defined(__GLASGOW_HASKELL__) && !defined(__HADDOCK__)
572+shiftl_w8 (W8# w) (I# i) = W8# (w `uncheckedShiftL#` i)
573+shiftl_w16 (W16# w) (I# i) = W16# (w `uncheckedShiftL#` i)
574+shiftl_w32 (W32# w) (I# i) = W32# (w `uncheckedShiftL#` i)
576+shiftr_w8 (W8# w) (I# i) = W8# (w `uncheckedShiftRL#` i)
577+shiftr_w16 (W16# w) (I# i) = W16# (w `uncheckedShiftRL#` i)
578+shiftr_w32 (W32# w) (I# i) = W32# (w `uncheckedShiftRL#` i)
581+#if WORD_SIZE_IN_BITS < 64
582+shiftl_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftL64#` i)
583+shiftr_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftRL64#` i)
585+#if __GLASGOW_HASKELL__ <= 606
586+-- Exported by GHC.Word in GHC 6.8 and higher
587+foreign import ccall unsafe "stg_uncheckedShiftL64"
588+ uncheckedShiftL64# :: Word64# -> Int# -> Word64#
589+foreign import ccall unsafe "stg_uncheckedShiftRL64"
590+ uncheckedShiftRL64# :: Word64# -> Int# -> Word64#
594+shiftl_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftL#` i)
595+shiftr_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftRL#` i)