@@ -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
@@ -177,30 +189,58 @@ 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
217+-- - LL: XWVEDCBA xxxxxxZY
218+data BitOrder = BB | LB | BL | LL deriving (Show)
220+-- | Reverse the @n@ least important bits of the given value
221+reverseBits :: (Num a, FastBits a, Bits a) => Int -> a -> a
222+reverseBits n value = rec value n 0
225+ -- v is orginal value shifted
226+ -- i is the remaining number of bits
227+ -- r is current value
229+ rec 0 i r = r `fastShiftL` i
230+ rec v i r = rec (v `fastShiftR` 1) (i-1) ((r `fastShiftL` 1) .|. (v .&. 0x1))
180-- | Read a single bit
232-- | Read a single bit
181readBool :: S -> Bool
233readBool :: S -> Bool
182readBool (S bs o bo) = case bo of
234readBool (S bs o bo) = case bo of
183 BB -> testBit (unsafeHead bs) (7-o)
235 BB -> testBit (unsafeHead bs) (7-o)
184 BL -> testBit (unsafeHead bs) (7-o)
236 BL -> testBit (unsafeHead bs) (7-o)
185− LL -> testBit (unsafeHead bs) o
186 LB -> testBit (unsafeHead bs) o
237 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
238+ LL -> testBit (unsafeHead bs) o
197− testBit# n = I# (((w `andWord8#` (wordToWord8# 1## `uncheckedShiftLWord8#` n)) `eqWord8#` wordToWord8# 0##) `xorI#` 1#)
200-- | Extract a range of bits from (ws :: ByteString)
240-- | Extract a range of bits from (ws :: ByteString)
202-- Constraint: 8 * (length ws -1 ) < o+n <= 8 * length ws
242-- Constraint: 8 * (length ws -1 ) < o+n <= 8 * length ws
203−extract :: (Num a, FastBits a) => BitOrder -> ByteString -> Int -> Int -> a
243+extract :: (Num a, Bits a, FastBits a) => BitOrder -> ByteString -> Int -> Int -> a
204extract bo bs o n
244extract bo bs o n
205 | n == 0 = 0
245 | n == 0 = 0
206 | B.length bs == 0 = error "Empty ByteString"
246 | B.length bs == 0 = error "Empty ByteString"
@@ -218,29 +258,36 @@ extract bo bs o n
219 -- shift offset depending on the byte position (0..B.length-1)
259 -- shift offset depending on the byte position (0..B.length-1)
220 off i = case bo of
260 off i = case bo of
222 LB -> 8*i - o
261 LB -> 8*i - o
223 BB -> (B.length bs -1 - i) * 8 - r
263 BB -> (B.length bs -1 - i) * 8 - r
224 BL -> (B.length bs -1 - i) * 8 - r
264 BL -> (B.length bs -1 - i) * 8 - r
226 -- reverse bits if necessary
266 -- reverse bits if necessary
227 rev = case bo of
267 rev = case bo of
228− LB -> reverseBits n
268+ LL -> reverseBits n
229 BL -> reverseBits n
269 BL -> reverseBits n
230− BB -> id
270+ BB -> id
231− LL -> id
271+ LB -> id
234-- | Generic readWord
274-- | Generic readWord
235−readWord :: (Num a, FastBits a) => Int -> S -> a
275+readWord :: (Num a, Bits a, FastBits a) => Int -> S -> a
236readWord n (S bs o bo)
276readWord n (S bs o bo)
237 | n == 0 = 0
277 | n == 0 = 0
278+ | o+n <= 8 = extract bo (unsafeTake 1 bs) o n
279+ | o+n <= 16 = extract bo (unsafeTake 2 bs) o n
280+ | o+n <= 24 = extract bo (unsafeTake 3 bs) o n
281+ | o+n <= 32 = extract bo (unsafeTake 4 bs) o n
282+ | o+n <= 40 = extract bo (unsafeTake 5 bs) o n
283+ | o+n <= 48 = extract bo (unsafeTake 6 bs) o n
284+ | o+n <= 56 = extract bo (unsafeTake 7 bs) o n
285+ | o+n <= 64 = extract bo (unsafeTake 8 bs) o n
238− | otherwise = extract bo (unsafeTake nbytes bs) o n
286+ | 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
288-- | Check that the number of bits to read is not greater than the first parameter
242{-# INLINE readWordChecked #-}
289{-# INLINE readWordChecked #-}
243−readWordChecked :: (Num a, FastBits a) => Int -> Int -> S -> a
290+readWordChecked :: (Num a, Bits a, FastBits a) => Int -> Int -> S -> a
244readWordChecked m n s
291readWordChecked m n s
245 | n > m = error $ "Tried to read more than " ++ show m ++ " bits (" ++ show n ++")"
292 | n > m = error $ "Tried to read more than " ++ show m ++ " bits (" ++ show n ++")"
246 | otherwise = readWord n s
293 | otherwise = readWord n s
@@ -248,24 +295,23 @@ readWordChecked m n s
248-- | Read the given number of bytes and return them in Big-Endian order
295-- | Read the given number of bytes and return them in Big-Endian order
250-- Examples:
297-- Examples:
251−-- BB: xxxABCDE FGHIJKLM NOPxxxxx -> ABCDEFGH IJKLMNOP
298+-- BB: xxxABCDE FGHIJKLM NOPQRxxx -> ABCDEFGH IJKLMNOP QRxxxxxx
252−-- LL: LMNOPxxx DEFGHIJK xxxxxABC -> ABCDEFGH IJKLMNOP
299+-- LB: NOPQRxxx FGHIJKLM xxxABCDE -> ABCDEFGH IJKLMNOP QRxxxxxx
253−-- BL: xxxPONML KJIHGFED CBAxxxxx -> ABCDEFGH IJKLMNOP
300+-- BL: xxxRQPON MLKJIHGF EDCBAxxx -> ABCDEFGH IJKLMNOP QRxxxxxx
254−-- LB: EDCBAxxx MLKJIHGF xxxxxPON -> ABCDEFGH IJKLMNOP
301+-- LL: EDCBAxxx MLKJIHGF xxxRQPON -> ABCDEFGH IJKLMNOP QRxxxxxx
255readByteString :: Int -> S -> ByteString
302readByteString :: Int -> S -> ByteString
256readByteString n (S bs o bo) =
303readByteString n (S bs o bo) =
258− bs' = unsafeTake (n+1) bs
305+ bs' = unsafeTake (n+1) bs
259− bs'' = unsafeTake n bs
260− rev = B.map (reverseBits 8)
306+ rev = B.map (reverseBits 9)
261 in case (o,bo) of
307 in case (o,bo) of
262− (0,BB) -> bs''
308+ (0,BB) -> unsafeTake n bs
263− (0,LL) -> B.reverse bs''
309+ (0,LB) -> B.reverse (unsafeTake n bs)
264− (0,LB) -> rev bs''
310+ (0,BL) -> rev (unsafeTake n bs)
265− (0,BL) -> rev . B.reverse $ bs''
311+ (0,LL) -> rev (B.reverse (unsafeTake n bs))
266− (_,LL) -> readByteString n (S (B.reverse bs') (8-o) BB)
312+ (_,BL) -> rev (readByteString n (S (B.reverse bs') o LB))
267− (_,BL) -> rev . B.reverse $ readByteString n (S bs' o BB)
313+ (_,LL) -> rev (readByteString n (S (B.reverse bs') o BB))
268− (_,LB) -> rev . B.reverse $ readByteString n (S bs' o LL)
314+ (_,LB) -> readByteString n (S (B.reverse bs') o BB)
269 (_,BB) -> unsafePerformIO $ do
315 (_,BB) -> unsafePerformIO $ do
270 let len = n+1
316 let len = n+1
271 ptr <- mallocBytes len
317 ptr <- mallocBytes len
@@ -287,6 +333,7 @@ newtype BitGet a = B { runState :: S -> Get (S,a) }
288instance Monad BitGet where
334instance Monad BitGet where
289 return x = B $ \s -> return (s,x)
335 return x = B $ \s -> return (s,x)
336+ fail str = B $ \s -> putBackState s >> fail str
290 (B f) >>= g = B $ \s -> do (s',a) <- f s
337 (B f) >>= g = B $ \s -> do (s',a) <- f s
291 runState (g a) s'
338 runState (g a) s'
@@ -297,29 +344,6 @@ instance Applicative BitGet where
297 pure x = return x
344 pure x = return x
298 fm <*> m = fm >>= \f -> m >>= \v -> return (f v)
345 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
347-- | 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.
348-- been partially consumed it will be discarded once 'runBitGet' is finished.
325runBitGet :: BitGet a -> Get a
349runBitGet :: BitGet a -> Get a
@@ -354,6 +378,13 @@ withState f = do
354 s <- getState
378 s <- getState
355 putState $! f s
379 putState $! f s
381+-- | Get the bit ordering
382+getBitOrder :: BitGet BitOrder
384+ (S _ _ bo) <- getState
357-- | Make sure there are at least @n@ bits.
388-- | Make sure there are at least @n@ bits.
358ensureBits :: Int -> BitGet ()
389ensureBits :: Int -> BitGet ()
359ensureBits n = do
390ensureBits n = do
@@ -367,6 +398,33 @@ ensureBits n = do
367 put B.empty
398 put B.empty
368 return (S (bs`append`bs') o bo, ())
399 return (S (bs`append`bs') o bo, ())
401+-- | Skip the given number of bits
402+skipBits :: Int -> BitGet ()
407+-- | Skip bits if necessary to align to the next byte
408+alignByte :: BitGet ()
410+ (S _ o _) <- getState
414+-- | Select the bit ordering
415+selectBitOrder :: BitOrder -> BitGet ()
416+selectBitOrder = withState . selectBitOrderS
418+withBitOrder :: BitOrder -> BitGet a -> BitGet a
419+withBitOrder bo f = do
370-- | Test whether all input has been consumed, i.e. there are no remaining
428-- | Test whether all input has been consumed, i.e. there are no remaining
371-- undecoded bytes.
429-- undecoded bytes.
372isEmpty :: BitGet Bool
430isEmpty :: BitGet Bool
@@ -378,10 +436,6 @@ isEmpty = B $ \ (S bs o bo) -> if B.null bs
378getBool :: BitGet Bool
436getBool :: BitGet Bool
379getBool = block bool
437getBool = 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]@.
439-- | Get @n@ bits as a 'Word8'. @n@ must be within @[0..8]@.
386getWord8 :: Int -> BitGet Word8
440getWord8 :: Int -> BitGet Word8
387getWord8 n = block (word8 n)
441getWord8 n = block (word8 n)
@@ -457,10 +511,6 @@ block (Block i p) = do
457bool :: Block Bool
511bool :: Block Bool
458bool = Block 1 readBool
512bool = 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]@.
514-- | Read @n@ bits as a 'Word8'. @n@ must be within @[0..8]@.
465word8 :: Int -> Block Word8
515word8 :: Int -> Block Word8
466word8 n = Block n (readWordChecked 8 n)
516word8 n = Block n (readWordChecked 8 n)
@@ -494,3 +544,84 @@ byteString :: Int -> Block ByteString
494byteString n | n > 0 = Block (n*8) (readByteString n)
544byteString n | n > 0 = Block (n*8) (readByteString n)
495 | otherwise = Block 0 (\_ -> B.empty)
545 | otherwise = Block 0 (\_ -> B.empty)
548+---------------------------------------------------------------------
549+-- Unchecked shifts, from the "binary" package
551+-- | Class for types supporting fast bit shifting
552+class FastBits a where
553+ fastShiftR :: a -> Int -> a
554+ fastShiftL :: a -> Int -> a
556+ {-# INLINE fastShift #-}
557+ fastShift :: a -> Int -> a
559+ | n > 0 = fastShiftL x n
560+ | n < 0 = fastShiftR x (negate n)
563+instance FastBits Word8 where
564+ fastShiftR = shiftr_w8
565+ fastShiftL = shiftl_w8
567+instance FastBits Word16 where
568+ fastShiftR = shiftr_w16
569+ fastShiftL = shiftl_w16
571+instance FastBits Word32 where
572+ fastShiftR = shiftr_w32
573+ fastShiftL = shiftl_w32
575+instance FastBits Word64 where
576+ fastShiftR = shiftr_w64
577+ fastShiftL = shiftl_w64
580+shiftl_w8 :: Word8 -> Int -> Word8
581+shiftl_w16 :: Word16 -> Int -> Word16
582+shiftl_w32 :: Word32 -> Int -> Word32
583+shiftl_w64 :: Word64 -> Int -> Word64
585+shiftr_w8 :: Word8 -> Int -> Word8
586+shiftr_w16 :: Word16 -> Int -> Word16
587+shiftr_w32 :: Word32 -> Int -> Word32
588+shiftr_w64 :: Word64 -> Int -> Word64
590+#if defined(__GLASGOW_HASKELL__) && !defined(__HADDOCK__)
591+shiftl_w8 (W8# w) (I# i) = W8# (w `uncheckedShiftL#` i)
592+shiftl_w16 (W16# w) (I# i) = W16# (w `uncheckedShiftL#` i)
593+shiftl_w32 (W32# w) (I# i) = W32# (w `uncheckedShiftL#` i)
595+shiftr_w8 (W8# w) (I# i) = W8# (w `uncheckedShiftRL#` i)
596+shiftr_w16 (W16# w) (I# i) = W16# (w `uncheckedShiftRL#` i)
597+shiftr_w32 (W32# w) (I# i) = W32# (w `uncheckedShiftRL#` i)
600+#if WORD_SIZE_IN_BITS < 64
601+shiftl_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftL64#` i)
602+shiftr_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftRL64#` i)
604+#if __GLASGOW_HASKELL__ <= 606
605+-- Exported by GHC.Word in GHC 6.8 and higher
606+foreign import ccall unsafe "stg_uncheckedShiftL64"
607+ uncheckedShiftL64# :: Word64# -> Int# -> Word64#
608+foreign import ccall unsafe "stg_uncheckedShiftRL64"
609+ uncheckedShiftRL64# :: Word64# -> Int# -> Word64#
613+shiftl_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftL#` i)
614+shiftr_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftRL#` i)