@@ -1,413 +1,496 @@
1−{-# LANGUAGE RankNTypes, MagicHash, CPP #-}
1+{-# LANGUAGE RankNTypes, MagicHash, BangPatterns, CPP #-}
2+
3+-----------------------------------------------------------------------------
4+-- |
5+-- Module : Data.Binary.Bits.Get
6+-- Copyright : (c) Lennart Kolmodin 2010-2011
7+-- (c) Sylvain Henry 2015
8+-- License : BSD3-style (see LICENSE)
9+--
10+-- Maintainer : kolmodin@gmail.com
11+-- Stability : experimental
12+-- Portability : portable (should run where the package binary runs)
13+--
14+-- Parse bits easily. Parsing can be done either in a monadic style, or more
15+-- efficiently, using the 'Applicative' style.
16+--
17+-- For the monadic style, write your parser as a 'BitGet' monad using the
18+--
19+-- * 'getBool'
20+--
21+-- * 'getWord8'
22+--
23+-- * 'getWord16be'
24+--
25+-- * 'getWord32be'
26+--
27+-- * 'getWord64be'
28+--
29+-- * 'getByteString'
30+--
31+-- functions and run it with 'runBitGet'.
32+--
33+-- For the applicative style, compose the fuctions
34+--
35+-- * 'bool'
36+--
37+-- * 'word8'
38+--
39+-- * 'word16be'
40+--
41+-- * 'word32be'
42+--
43+-- * 'word64be'
44+--
45+-- * 'byteString'
46+--
47+-- to make a 'Block'.
48+-- Use 'block' to turn it into the 'BitGet' monad to be able to run it with
49+-- 'runBitGet'.
50+-----------------------------------------------------------------------------
251
352module Data.Binary.Bits.Get
4− ( T(..)
53+ (
54+ -- * BitGet monad
55+
56+ -- $bitget
57+
58+ BitGet
559 , runBitGet
6− , runBitGetSimple
7− , BitGet
860
61+ -- ** Get bytes
962 , getBool
63+ , getBool#
1064 , getWord8
65+ , getWord16
66+ , getWord32
67+ , getWord64
1168 , getWord16be
1269 , getWord32be
1370 , getWord64be
1471
72+ -- ** Skip bits
73+ , skipBits
74+ , alignByte
75+
76+ -- * Blocks
77+
78+ -- $blocks
79+ , Block
80+ , block
81+
82+ -- ** Read in Blocks
83+ , bool
84+ , word8
85+ , word16
86+ , word32
87+ , word64
88+ , word16be
89+ , word32be
90+ , word64be
91+ , byteString
1592 , Data.Binary.Bits.Get.getByteString
16−
17− , readBool
18− , readWord8
19− , readWord16be
93+ , Data.Binary.Bits.Get.getLazyByteString
94+ , Data.Binary.Bits.Get.isEmpty
2095
2196 ) where
2297
23−import Data.Binary.Get as B ( runGet, Get, getByteString )
98+import Data.Binary.Get as B ( Get, getLazyByteString, isEmpty )
99+import 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
24103
25104import Data.ByteString as B
26−import Data.ByteString.Internal
27−import Data.ByteString.Unsafe
28−
29105import qualified Data.ByteString.Lazy as L
106+import Data.ByteString.Unsafe
107+import Foreign.Marshal.Alloc (mallocBytes)
108+import Foreign.Ptr
109+import Foreign.Storable (poke)
110+import System.IO.Unsafe (unsafePerformIO)
30111
31112import Data.Bits
32113import Data.Word
33−import Data.List as List ( reverse )
34−
35114import Control.Applicative
115+import Control.Monad (when,foldM_)
36116
37−#if defined(__GLASGOW_HASKELL__) && !defined(__HADDOCK__)
38−import GHC.Base
117+import Prelude as P
39118import GHC.Word
40−import GHC.Int
41−#endif
42−
43−data T a b = !a :*: !b deriving (Show)
44−
45−data S = S !ByteString -- ^ Input
46− !Int -- ^ Bit offset (0-7)
119+import GHC.Base (uncheckedShiftLWord8#, wordToWord8#, Int (..), andWord8#, Word8#, (-#), Int#, xorI#, eqWord8#, TYPE)
120+
121+-- $bitget
122+-- Parse bits using a monad.
123+--
124+-- @
125+--myBitParser :: 'Get' ('Word8', 'Word8')
126+--myBitParser = 'runGetBit' parse4by4
127+--
128+--parse4by4 :: 'BitGet' ('Word8', 'Word8')
129+--parse4by4 = do
130+-- bits <- 'getWord8' 4
131+-- more <- 'getWord8' 4
132+-- return (bits,more)
133+-- @
134+
135+-- $blocks
136+-- Parse more efficiently in blocks. Each block is read with only one boundry
137+-- check (checking that there is enough input) as the size of the block can be
138+-- calculated statically. This is somewhat limiting as you cannot make the
139+-- parsing depend on the input being parsed.
140+--
141+-- @
142+--data IPV6Header = IPV6Header {
143+-- ipv6Version :: 'Word8'
144+-- , ipv6TrafficClass :: 'Word8'
145+-- , ipv6FlowLabel :: 'Word32
146+-- , ipv6PayloadLength :: 'Word16'
147+-- , ipv6NextHeader :: 'Word8'
148+-- , ipv6HopLimit :: 'Word8'
149+-- , ipv6SourceAddress :: 'ByteString'
150+-- , ipv6DestinationAddress :: 'ByteString'
151+-- }
152+--
153+-- ipv6headerblock =
154+-- IPV6Header '<$>' 'word8' 4
155+-- '<*>' 'word8' 8
156+-- '<*>' 'word32be' 24
157+-- '<*>' 'word16be' 16
158+-- '<*>' 'word8' 8
159+-- '<*>' 'word8' 8
160+-- '<*>' 'byteString' 16
161+-- '<*>' 'byteString' 16
162+--
163+--ipv6Header :: 'Get' IPV6Header
164+--ipv6Header = 'runBitGet' ('block' ipv6headerblock)
165+-- @
166+
167+data S = S {-# UNPACK #-} !ByteString -- Input
168+ {-# UNPACK #-} !Int -- Bit offset (0-7)
169+ !BitOrder -- Bit order
47170 deriving (Show)
48171
49−-- | make_mask 3 = 00000111
50−make_mask :: Bits a => Int -> a
51−make_mask n = (1 `shiftL` fromIntegral n) - 1
52−{-# SPECIALIZE make_mask :: Int -> Word #-}
53−{-# SPECIALIZE make_mask :: Int -> Word8 #-}
54−{-# SPECIALIZE make_mask :: Int -> Word16 #-}
55−{-# SPECIALIZE make_mask :: Int -> Word32 #-}
56−{-# SPECIALIZE make_mask :: Int -> Word64 #-}
57−
58−readBool :: S -> T Bool S
59−readBool s@(S bs n) = testBit (unsafeHead bs) (7-n) :*: incS s 1
60−
61−incS :: S -> Int -> S
62−incS (S bs n) 1 =
63− let
64− n' = n + 1
65− s' | n' == 8 = S (unsafeTail bs) 0
66− | otherwise = S bs n'
67− in s'
68−incS (S bs n) o =
69− let (d,n') = divMod (n+o) 8
70− in S (unsafeDrop d bs) n'
71−
72−readByteString :: S -> Int -> T ByteString S
73−readByteString s@(S bs o) n
74− | o == 0 = unsafeTake n bs :*: (S (unsafeDrop n bs) 0)
75− | otherwise = unsafeTake n (unsafeTail bs) :*: (S (unsafeDrop (n+1) bs) 0)
76−
77−readWord8 :: S -> Int -> T Word8 S
78−readWord8 s@(S bs o) n
79− -- no bits at all, return 0
80− | n == 0 = 0 :*: s
81−
82− -- all bits are in the same byte
83− -- we just need to shift and mask them right
84− | n <= 8 - o = let w = unsafeHead bs
85− m = make_mask n
86− w' = (w `shiftr_w8` (8 - o - n)) .&. m
87− in w' :*: incS s n
88−
89− -- the bits are in two different bytes
90− -- make a word16 using both bytes, and then shift and mask
91− | n <= 8 = let w = (fromIntegral (unsafeHead bs) `shiftl_w16` 8) .|.
92− (fromIntegral (unsafeIndex bs 1))
93− m = make_mask n
94− w' = (w `shiftr_w16` (16 - o - n)) .&. m
95− in fromIntegral w' :*: incS s n
96−
97−bit_offset :: Int -> Int
98−bit_offset n = make_mask 3 .&. n
99−
100−byte_offset :: Int -> Int
101−byte_offset n = n `shiftR` 3
102−
103−readWord16be :: S -> Int -> T Word16 S
104−readWord16be s@(S bs o) n
105−
106− -- 8 or fewer bits, use readWord8
107− | n <= 8 = let w :*: s' = readWord8 s n
108− in fromIntegral w :*: s'
109−
110− -- handle 9 or more bits, stored in two bytes
111−
112− -- no offset, plain and simple 16 bytes
113− | o == 0 && n == 16 = let msb = fromIntegral (unsafeHead bs)
114− lsb = fromIntegral (unsafeIndex bs 1)
115− w = (msb `shiftl_w16` 8) .|. lsb
116− in w :*: incS s n
117−
118− -- no offset, but not full 16 bytes
119− | o == 0 = let msb = fromIntegral (unsafeHead bs)
120− lsb = fromIntegral (unsafeIndex bs 1)
121− w = (msb `shiftl_w16` (n-8)) .|. (lsb `shiftr_w16` (16-n))
122− in w :*: incS s n
123−
124− -- with offset, and n=9-16
125− | n <= 16 = readWithOffset s shiftl_w16 shiftr_w16 n
126−
127− | otherwise = error "readWord16be: tried to read more than 16 bits"
128−
129−readWithoutOffset :: (Bits a, Num a)
130− => S -> (a -> Int -> a) -> (a -> Int -> a) -> Int -> T a S
131−readWithoutOffset s@(S bs o) shifterL shifterR n
132− | o /= 0 = error "readWithoutOffset: there is an offset"
133−
134− | bit_offset n == 0 && byte_offset n <= 4 =
135− let segs = byte_offset n
136− bn 0 = fromIntegral (unsafeHead bs)
137− bn n = (bn (n-1) `shifterL` 8) .|. fromIntegral (unsafeIndex bs n)
138−
139− in (bn (segs-1) :*: incS s n)
140−
141− | n <= 64 = let segs = byte_offset n
142− o' = bit_offset (n - 8 + o)
143−
144− bn 0 = fromIntegral (unsafeHead bs)
145− bn n = (bn (n-1) `shifterL` 8) .|. fromIntegral (unsafeIndex bs n)
146−
147− msegs = bn (segs-1) `shifterL` o'
148−
149− last = (fromIntegral (unsafeIndex bs segs)) `shifterR` (8 - o')
150−
151− w = msegs .|. last
152− in w :*: incS s n
153−
154−readWithOffset :: (Bits a, Num a)
155− => S -> (a -> Int -> a) -> (a -> Int -> a) -> Int -> T a S
156−readWithOffset s@(S bs o) shifterL shifterR n
157− | n <= 64 = let bits_in_msb = 8 - o
158− (n',top) = (n - bits_in_msb
159− , (fromIntegral (unsafeHead bs) .&. make_mask bits_in_msb) `shifterL` n')
160−
161− segs = byte_offset n'
162−
163− bn 0 = 0
164− bn 1 = fromIntegral (unsafeIndex bs 1)
165− bn n = (bn (n-1) `shifterL` 8) .|. fromIntegral (unsafeIndex bs n)
166−
167− o' = bit_offset n'
168−
169− mseg = bn segs `shifterL` o'
170−
171− last | o' > 0 = (fromIntegral (unsafeIndex bs (segs + 1))) `shifterR` (8 - o')
172− | otherwise = 0
173−
174− w = top .|. mseg .|. last
175− in w :*: incS s n
176−
177−
178−readWord32be :: S -> Int -> T Word32 S
179−readWord32be s@(S bs o) n
180− -- 8 or fewer bits, use readWord8
181− | n <= 8 = let w :*: s' = readWord8 s n
182− in fromIntegral w :*: s'
172+-- | Increment the current bit offset
173+incS :: Int -> S -> S
174+incS o (S bs n bo) = S (unsafeDrop d bs) n' bo
175+ where
176+ !o' = (n+o)
177+ !d = byte_offset o'
178+ !n' = bit_offset o'
179+
180+-- | Read a single bit
181+readBool :: S -> Bool
182+readBool (S bs o bo) = case bo of
183+ 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
187+
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+
200+-- | Extract a range of bits from (ws :: ByteString)
201+--
202+-- Constraint: 8 * (length ws -1 ) < o+n <= 8 * length ws
203+extract :: (Num a, FastBits a) => BitOrder -> ByteString -> Int -> Int -> a
204+extract bo bs o n
205+ | n == 0 = 0
206+ | B.length bs == 0 = error "Empty ByteString"
207+ | otherwise = rev . mask n . foldlWithIndex' f 0 $ bs
208+ where
209+ -- B.foldl' with index
210+ foldlWithIndex' op b = fst . B.foldl' g (b,0)
211+ where g (b',i) w = (op b' w i, (i+1))
212+
213+ -- 'or' correctly shifted words
214+ f b w i = b .|. (fromIntegral w `fastShift` off i)
215+
216+ -- co-offset
217+ r = B.length bs * 8 - (o + n)
218+
219+ -- shift offset depending on the byte position (0..B.length-1)
220+ off i = case bo of
221+ LL -> 8*i - o
222+ LB -> 8*i - o
223+ 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
232+
233+
234+-- | Generic readWord
235+readWord :: (Num a, FastBits a) => Int -> S -> a
236+readWord n (S bs o bo)
237+ | n == 0 = 0
238+ | otherwise = extract bo (unsafeTake nbytes bs) o n
239+ where nbytes = byte_offset (o+n+7)
240+
241+-- | Check that the number of bits to read is not greater than the first parameter
242+{-# INLINE readWordChecked #-}
243+readWordChecked :: (Num a, FastBits a) => Int -> Int -> S -> a
244+readWordChecked m n s
245+ | n > m = error $ "Tried to read more than " ++ show m ++ " bits (" ++ show n ++")"
246+ | otherwise = readWord n s
247+
248+-- | Read the given number of bytes and return them in Big-Endian order
249+--
250+-- 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
255+readByteString :: 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)
183281
184− -- 16 or fewer bits, use readWord16be
185− | n <= 16 = let w :*: s' = readWord16be s n
186− in fromIntegral w :*: s'
187282
188− | o == 0 = readWithoutOffset s shiftl_w32 shiftr_w32 n
283+------------------------------------------------------------------------
284+-- | 'BitGet' is a monad, applicative and a functor. See 'runBitGet'
285+-- for how to run it.
286+newtype BitGet a = B { runState :: S -> Get (S,a) }
189287
190− | n <= 32 = readWithOffset s shiftl_w32 shiftr_w32 n
288+instance Monad BitGet where
289+ return x = B $ \s -> return (s,x)
290+ (B f) >>= g = B $ \s -> do (s',a) <- f s
291+ runState (g a) s'
191292
192− | otherwise = error "readWord32be: tried to read more than 32 bits"
293+instance Functor BitGet where
294+ fmap f m = m >>= \a -> return (f a)
193295
296+instance Applicative BitGet where
297+ pure x = return x
298+ fm <*> m = fm >>= \f -> m >>= \v -> return (f v)
194299
195−readWord64be :: S -> Int -> T Word64 S
196−readWord64be s@(S bs o) n
197− -- 8 or fewer bits, use readWord8
198− | n <= 8 = let w :*: s' = readWord8 s n
199− in fromIntegral w :*: s'
300+instance BitOrderable BitGet where
301+ setBitOrder bo = do
302+ (S bs o _) <- getState
303+ putState (S bs o bo)
200304
201− -- 16 or fewer bits, use readWord16be
202− | n <= 16 = let w :*: s' = readWord16be s n
203− in fromIntegral w :*: s'
305+ getBitOrder = do
306+ (S _ _ bo) <- getState
307+ return bo
204308
205− | o == 0 = readWithoutOffset s shiftl_w64 shiftr_w64 n
309+instance Alignable BitGet where
310+ -- | Skip the given number of bits
311+ skipBits n = do
312+ ensureBits n
313+ withState (incS n)
206314
207− | n <= 64 = readWithOffset s shiftl_w64 shiftr_w64 n
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)
208320
209− | otherwise = error "readWord64be: tried to read more than 64 bits"
210321
211−------------------------------------------------------------------------
212−-- unrolled codensity/state monad
213−newtype BitGet a = C { runCont :: forall r.
214− S ->
215− Success a r ->
216− Get r }
217322
218−type Success a r = S -> a -> Get r
323+-- | 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.
325+runBitGet :: BitGet a -> Get a
326+runBitGet bg = do
327+ s <- mkInitState
328+ (s',a) <- runState bg s
329+ putBackState s'
330+ return a
219331
220−instance Monad BitGet where
221− return x = C $ \s ks -> ks s x
222− fail str = C $ \s ks -> fail str
223− (C c) >>= f = C $ \s ks -> c s (\s' a -> runCont (f a) s' ks)
332+mkInitState :: Get S
333+mkInitState = do
334+ bs <- get
335+ put B.empty
336+ return (S bs 0 BB)
224337
225−instance Functor BitGet where
226− fmap f m = m >>= \a -> return (f a)
338+putBackState :: S -> Get ()
339+putBackState (S bs o _) = do
340+ remaining <- get
341+ let bs' = case o of
342+ 0 -> bs
343+ _ -> unsafeDrop 1 bs
344+ put (bs' `B.append` remaining)
227345
228−instance Applicative BitGet where
229− pure x = return x
230− fm <*> m = fm >>= \f -> m >>= \v -> return (f v)
346+getState :: BitGet S
347+getState = B $ \s -> return (s,s)
231348
232−runBitGetSimple :: BitGet a -> Get a
233−runBitGetSimple bg = runBitGet bg
349+putState :: S -> BitGet ()
350+putState s = B $ \_ -> return (s,())
234351
235−runBitGet :: BitGet a -> Get a
236−runBitGet bg = runCont bg (S B.empty 0) (\_s a -> return a)
352+withState :: (S -> S) -> BitGet ()
353+withState f = do
354+ s <- getState
355+ putState $! f s
237356
238−-- | Make sure there are at lest @n@ bits.
357+-- | Make sure there are at least @n@ bits.
239358ensureBits :: Int -> BitGet ()
240−ensureBits n = C $ \s ks ->
241− let loop (S bs o) | n <= (B.length bs * 8 - o) = ks (S bs o) ()
242− | otherwise = do
243− let currentBits = B.length bs * 8 - o
244− let byteCount = (n - currentBits + 7) `div` 8
245− bs' <- B.getByteString byteCount
246− -- should always be enough going once throguh the loop
247− loop (S (bs`append`bs') o)
248− in loop s
249−
359+ensureBits n = do
360+ (S bs o bo) <- getState
361+ if n <= (B.length bs * 8 - o)
362+ then return ()
363+ else do let currentBits = B.length bs * 8 - o
364+ let byteCount = byte_offset (n - currentBits + 7)
365+ B $ \_ -> do B.ensureN byteCount
366+ bs' <- B.get
367+ put B.empty
368+ return (S (bs`append`bs') o bo, ())
369+
370+-- | Test whether all input has been consumed, i.e. there are no remaining
371+-- undecoded bytes.
372+isEmpty :: BitGet Bool
373+isEmpty = B $ \ (S bs o bo) -> if B.null bs
374+ then B.isEmpty >>= \e -> return (S bs o bo, e)
375+ else return (S bs o bo, False)
376+
377+-- | Get 1 bit as a 'Bool'.
250378getBool :: BitGet Bool
251−getBool = ensureBits 1 >> modifyState readBool
379+getBool = block bool
252380
381+-- | Get 1 bit as a 'Bool'.
382+getBool# :: BitGet Int
383+getBool# = block bool#
384+
385+-- | Get @n@ bits as a 'Word8'. @n@ must be within @[0..8]@.
253386getWord8 :: Int -> BitGet Word8
254−getWord8 n = ensureBits n >> modifyState (flip readWord8 n)
387+getWord8 n = block (word8 n)
388+
389+-- | Get @n@ bits as a 'Word16'. @n@ must be within @[0..16]@.
390+getWord16 :: Int -> BitGet Word16
391+getWord16 n = block (word16 n)
255392
256393getWord16be :: Int -> BitGet Word16
257−getWord16be n = ensureBits n >> modifyState (flip readWord16be n)
394+getWord16be = getWord16
395+{-# DEPRECATED getWord16be "Use 'getWord16' instead" #-}
396+
397+-- | Get @n@ bits as a 'Word32'. @n@ must be within @[0..32]@.
398+getWord32 :: Int -> BitGet Word32
399+getWord32 n = block (word32 n)
258400
259401getWord32be :: Int -> BitGet Word32
260−getWord32be n = ensureBits n >> modifyState (flip readWord32be n)
402+getWord32be = getWord32
403+{-# DEPRECATED getWord32be "Use 'getWord32' instead" #-}
404+
405+-- | Get @n@ bits as a 'Word64'. @n@ must be within @[0..64]@.
406+getWord64 :: Int -> BitGet Word64
407+getWord64 n = block (word64 n)
261408
262409getWord64be :: Int -> BitGet Word64
263−getWord64be n = ensureBits n >> modifyState (flip readWord64be n)
410+getWord64be = getWord64
411+{-# DEPRECATED getWord64be "Use 'getWord64' instead" #-}
264412
265−-- | Get a @ByteString@. Drops the current byte if we've started to read
266−-- from it.
413+-- | Get @n@ bytes as a 'ByteString'.
267414getByteString :: Int -> BitGet ByteString
268−getByteString n = do
269− ensureBits (n*8)
270− -- drop the first byte if we've started to use it
271− modifyState $ \s@(S bs o) ->
272− let bs' | o == 0 = bs
273− | otherwise = B.tail bs
274− in B.take n bs' :*: S (B.drop n bs') 0
275−
276−getState :: BitGet S
277−getState = C $ \s ks -> ks s s
278−
279−putState :: S -> BitGet ()
280−putState s = C $ \_ ks -> ks s ()
281−
282−modifyState :: (S -> (T a S)) -> BitGet a
283−modifyState f = C $ \s ks -> case f s of
284− w :*: s' -> ks s' w
285−
286−------------------------------------------------------------------------
287−-- Unchecked shifts, from the package binary
288−
289−shiftl_w16 :: Word16 -> Int -> Word16
290−shiftl_w32 :: Word32 -> Int -> Word32
291−shiftl_w64 :: Word64 -> Int -> Word64
292−
293−#if defined(__GLASGOW_HASKELL__) && !defined(__HADDOCK__)
294−shiftl_w8 (W8# w) (I# i) = W8# (w `uncheckedShiftL#` i)
295−shiftl_w16 (W16# w) (I# i) = W16# (w `uncheckedShiftL#` i)
296−shiftl_w32 (W32# w) (I# i) = W32# (w `uncheckedShiftL#` i)
297−
298−shiftr_w8 (W8# w) (I# i) = W8# (w `uncheckedShiftRL#` i)
299−shiftr_w16 (W16# w) (I# i) = W16# (w `uncheckedShiftRL#` i)
300−shiftr_w32 (W32# w) (I# i) = W32# (w `uncheckedShiftRL#` i)
301−
302−
303−#if WORD_SIZE_IN_BITS < 64
304−shiftl_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftL64#` i)
305−shiftr_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftRL64#` i)
306−
307−#if __GLASGOW_HASKELL__ <= 606
308−-- Exported by GHC.Word in GHC 6.8 and higher
309−foreign import ccall unsafe "stg_uncheckedShiftL64"
310− uncheckedShiftL64# :: Word64# -> Int# -> Word64#
311−foreign import ccall unsafe "stg_uncheckedShiftRL64"
312− uncheckedShiftRL64# :: Word64# -> Int# -> Word64#
313−#endif
314−
315−#else
316−shiftl_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftL#` i)
317−shiftr_w64 (W64# w) (I# i) = W64# (w `uncheckedShiftRL#` i)
318−#endif
319−
320−#else
321−shiftl_w8 = shiftL
322−shiftl_w16 = shiftL
323−shiftl_w32 = shiftL
324−shiftl_w64 = shiftL
325−
326−shiftr_w8 = shiftR
327−shiftr_w16 = shiftR
328−shiftr_w32 = shiftR
329−shiftr_w64 = shiftR
330−#endif
331−
332−{-
333−data R a where
334− RBool :: R Bool
335− RWord8 :: Int -> R Word8
336− RWord16be :: Int -> R Word16
337− RWord32be :: Int -> R Word32
338− RByteString :: Int -> R ByteString
339− RThis :: a -> R a
340− RNextTo :: R a -> R b -> R (T a b)
341− RMap :: R b -> (b -> R a) -> R a
342− RMapPure :: R a -> (a -> b) -> R b
343− RList :: Int -> R a -> R [a]
344− RCheck :: R a -> (a -> Bool) -> String -> R a
345−
346−instance Functor R where
347− fmap f m = RMapPure m f
348−
349−instance Show (R a) where
350− show r = case r of
351− RBool -> "Bool"
352−
353−size_in_bits :: forall a. R a -- ^ The record
354− -> Int -- ^ Number of bits
355−size_in_bits r = case r of
356− RBool -> 1
357− RWord8 n -> min n 8
358− RWord16be n -> min n 16
359− RWord32be n -> min n 16
360− RByteString n -> 8 * n
361− RThis _ -> 0
362− RList n r -> n * size_in_bits r
363− RMap r _ -> size_in_bits r
364− RMapPure r _ -> size_in_bits r
365− RCheck r _ _ -> size_in_bits r
366− RNextTo a b -> size_in_bits a + size_in_bits b
367−
368−size_in_bytes :: forall a. R a
369− -> Int
370−size_in_bytes r = byte_offset (size_in_bits r + 7)
371−
372−get :: R a -> Get a
373−get r = do
374− bs <- getByteString (size_in_bytes r)
375− a :*: s <- getR (S bs 0) r
376− return a
377−
378−getR :: S -> R a -> Get (T a S)
379−getR s0 r = do
380− case r of
381− RBool -> return (readBool s0)
382− RWord8 n -> return (readWord8 s0 n)
383− RWord16be n -> return (readWord16be s0 n)
384− RWord32be n -> return (readWord32be s0 n)
385− RByteString n -> return (readByteString s0 n)
386− RThis x -> return (x :*: s0)
387− RNextTo a b -> do
388− t :*: s <- getR s0 a
389− u :*: s' <- getR s b
390− return (t:*:u:*:s')
391− RMap r p -> do
392− a :*: s@(S bs o) <- getR s0 r
393− let codepath = p a
394− required_bytes = byte_offset (size_in_bits r - o)
395− bs' <- getByteString required_bytes
396− getR (S (bs `append` bs') o) codepath
397− RMapPure r f -> do
398− a :*: s <- getR s0 r
399− return (f a :*: s)
400− RList n r ->
401− let loop 0 s acc = return (List.reverse acc :*: s)
402− loop m s acc = do
403− a :*: s' <- getR s r
404− loop (m-1) s' (a:acc)
405− in loop n s0 []
406− RCheck r c m -> do
407− a :*: s <- getR s0 r
408− if c a
409− then return (a :*: s)
410− else fail m
411−-}
412−
415+getByteString n = block (byteString n)
416+
417+-- | Get @n@ bytes as a lazy ByteString.
418+getLazyByteString :: Int -> BitGet L.ByteString
419+getLazyByteString n = do
420+ (S _ o bo) <- getState
421+ case o of
422+ 0 -> B $ \s -> do
423+ putBackState s
424+ lbs <- B.getLazyByteString (fromIntegral n)
425+ return (S B.empty 0 bo, lbs)
426+ _ -> L.fromChunks . (:[]) <$> Data.Binary.Bits.Get.getByteString n
427+
428+
429+
430+
431+
432+-- | A block that will be read with only one boundary check. Needs to know the
433+-- number of bits in advance.
434+data Block a = Block Int (S -> a)
435+
436+instance Functor Block where
437+ fmap f (Block i p) = Block i (\s -> f (p s))
438+
439+instance Applicative Block where
440+ pure a = Block 0 (\_ -> a)
441+ (Block i p) <*> (Block j q) = Block (i+j) (\s -> p s $ q (incS i s))
442+ (Block i _) *> (Block j q) = Block (i+j) (q . incS i)
443+ (Block i p) <* (Block j _) = Block (i+j) p
444+
445+-- | Get a block. Will be read with one single boundry check, and
446+-- therefore requires a statically known number of bits.
447+-- Build blocks using 'bool', 'word8', 'word16be', 'word32be', 'word64be',
448+-- 'byteString' and 'Applicative'.
449+block :: Block a -> BitGet a
450+block (Block i p) = do
451+ ensureBits i
452+ s <- getState
453+ putState $! (incS i s)
454+ return $! p s
455+
456+-- | Read a 1 bit 'Bool'.
457+bool :: Block Bool
458+bool = Block 1 readBool
459+
460+-- | Read a 1 bit 'Bool'.
461+bool# :: Block Int
462+bool# = Block 1 readBool#
463+
464+-- | Read @n@ bits as a 'Word8'. @n@ must be within @[0..8]@.
465+word8 :: Int -> Block Word8
466+word8 n = Block n (readWordChecked 8 n)
467+
468+-- | Read @n@ bits as a 'Word16'. @n@ must be within @[0..16]@.
469+word16 :: Int -> Block Word16
470+word16 n = Block n (readWordChecked 16 n)
471+
472+word16be :: Int -> Block Word16
473+word16be = word16
474+{-# DEPRECATED word16be "Use 'word16' instead" #-}
475+
476+-- | Read @n@ bits as a 'Word32'. @n@ must be within @[0..32]@.
477+word32 :: Int -> Block Word32
478+word32 n = Block n (readWordChecked 32 n)
479+
480+word32be :: Int -> Block Word32
481+word32be = word32
482+{-# DEPRECATED word32be "Use 'word32' instead" #-}
483+
484+-- | Read @n@ bits as a 'Word64'. @n@ must be within @[0..64]@.
485+word64 :: Int -> Block Word64
486+word64 n = Block n (readWordChecked 64 n)
487+
488+word64be :: Int -> Block Word64
489+word64be = word64
490+{-# DEPRECATED word64be "Use 'word64' instead" #-}
491+
492+-- | Read @n@ bytes as a 'ByteString'.
493+byteString :: Int -> Block ByteString
494+byteString n | n > 0 = Block (n*8) (readByteString n)
495+ | otherwise = Block 0 (\_ -> B.empty)
413496