1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
|
{-# LANGUAGE CPP #-}
-----------------------------------------------------------------------------
-- |
-- Copyright : (c) 2006-2008 Duncan Coutts
-- License : BSD-style
--
-- Maintainer : duncan@haskell.org
-- Stability : provisional
-- Portability : portable (H98 + FFI)
--
-- Pure stream based interface to lower level bzlib wrapper
--
-----------------------------------------------------------------------------
module Codec.Compression.BZip.Internal (
-- * Compression
compress,
CompressParams(..),
defaultCompressParams,
-- * Decompression
decompress,
DecompressParams(..),
defaultDecompressParams,
-- * The compression parameter types
Stream.BlockSize(..),
Stream.WorkFactor(..),
Stream.MemoryLevel(..),
) where
import Prelude hiding (length)
import Control.Monad (when)
import Control.Exception (assert)
import qualified Data.ByteString.Lazy as L
#ifdef BYTESTRING_IN_BASE
import qualified Data.ByteString.Base as S
#else
import qualified Data.ByteString.Lazy.Internal as L
import qualified Data.ByteString.Internal as S
#endif
import qualified Codec.Compression.BZip.Stream as Stream
import Codec.Compression.BZip.Stream (Stream)
-- | The full set of parameters for compression. The defaults are
-- 'defaultCompressParams'.
--
-- The 'compressBufferSize' is the size of the first output buffer containing
-- the compressed data. If you know an approximate upper bound on the size of
-- the compressed data then setting this parameter can save memory. The default
-- compression output buffer size is @16k@. If your extimate is wrong it does
-- not matter too much, the default buffer size will be used for the remaining
-- chunks.
--
data CompressParams = CompressParams {
compressBlockSize :: Stream.BlockSize,
compressWorkFactor :: Stream.WorkFactor,
compressBufferSize :: Int
}
-- | The full set of parameters for decompression. The defaults are
-- 'defaultDecompressParams'.
--
-- The 'decompressBufferSize' is the size of the first output buffer,
-- containing the uncompressed data. If you know an exact or approximate upper
-- bound on the size of the decompressed data then setting this parameter can
-- save memory. The default decompression output buffer size is @32k@. If your
-- extimate is wrong it does not matter too much, the default buffer size will
-- be used for the remaining chunks.
--
-- One particular use case for setting the 'decompressBufferSize' is if you
-- know the exact size of the decompressed data and want to produce a strict
-- 'Data.ByteString.ByteString'. The compression and deccompression functions
-- use lazy 'Data.ByteString.Lazy.ByteString's but if you set the
-- 'decompressBufferSize' correctly then you can generate a lazy
-- 'Data.ByteString.Lazy.ByteString' with exactly one chunk, which can be
-- converted to a strict 'Data.ByteString.ByteString' in @O(1)@ time using
-- @'Data.ByteString.concat' . 'Data.ByteString.Lazy.toChunks'@.
--
data DecompressParams = DecompressParams {
decompressMemoryLevel :: Stream.MemoryLevel,
decompressBufferSize :: Int
}
-- | The default set of parameters for compression. This is typically used with
-- the @compressWith@ function with specific paramaters overridden.
--
defaultCompressParams :: CompressParams
defaultCompressParams = CompressParams {
compressBlockSize = Stream.DefaultBlockSize,
compressWorkFactor = Stream.DefaultWorkFactor,
compressBufferSize = defaultCompressBufferSize
}
-- | The default set of parameters for decompression. This is typically used with
-- the @compressWith@ function with specific paramaters overridden.
--
defaultDecompressParams :: DecompressParams
defaultDecompressParams = DecompressParams {
decompressMemoryLevel = Stream.DefaultMemoryLevel,
decompressBufferSize = defaultDecompressBufferSize
}
-- | The default chunk sizes for the output of compression and decompression
-- are 16k and 32k respectively (less a small accounting overhead).
--
defaultCompressBufferSize, defaultDecompressBufferSize :: Int
#ifdef BYTESTRING_IN_BASE
defaultCompressBufferSize = 16 * 1024 - 16
defaultDecompressBufferSize = 32 * 1024 - 16
#else
defaultCompressBufferSize = 16 * 1024 - L.chunkOverhead
defaultDecompressBufferSize = 32 * 1024 - L.chunkOverhead
#endif
{-# NOINLINE compress #-}
compress
:: CompressParams
-> L.ByteString
-> L.ByteString
compress (CompressParams blockSize workFactor initChunkSize) input =
L.fromChunks $ Stream.run $ do
Stream.compressInit blockSize Stream.Silent workFactor
case L.toChunks input of
[] -> fillBuffers 14 [] --bzip2 header is 14 bytes
S.PS inFPtr offset length : chunks -> do
Stream.pushInputBuffer inFPtr offset length
fillBuffers initChunkSize chunks
where
-- we flick between two states:
-- * where one or other buffer is empty
-- - in which case we refill one or both
-- * where both buffers are non-empty
-- - in which case we compress until a buffer is empty
fillBuffers :: Int
-> [S.ByteString]
-> Stream [S.ByteString]
fillBuffers outChunkSize inChunks = do
Stream.consistencyCheck
-- in this state there are two possabilities:
-- * no outbut buffer space is available
-- - in which case we must make more available
-- * no input buffer is available
-- - in which case we must supply more
inputBufferEmpty <- Stream.inputBufferEmpty
outputBufferFull <- Stream.outputBufferFull
assert (inputBufferEmpty || outputBufferFull) $ return ()
when outputBufferFull $ do
outFPtr <- Stream.unsafeLiftIO (S.mallocByteString outChunkSize)
Stream.pushOutputBuffer outFPtr 0 outChunkSize
if inputBufferEmpty
then case inChunks of
[] -> drainBuffers []
S.PS inFPtr offset length : inChunks' -> do
Stream.pushInputBuffer inFPtr offset length
drainBuffers inChunks'
else drainBuffers inChunks
drainBuffers ::
[S.ByteString]
-> Stream [S.ByteString]
drainBuffers inChunks = do
inputBufferEmpty' <- Stream.inputBufferEmpty
outputBufferFull' <- Stream.outputBufferFull
assert(not outputBufferFull'
&& (null inChunks || not inputBufferEmpty')) $ return ()
-- this invariant guarantees we can always make forward progress
let action = if null inChunks then Stream.Finish else Stream.Run
status <- Stream.compress action
case status of
Stream.Ok -> do
outputBufferFull <- Stream.outputBufferFull
if outputBufferFull
then do (outFPtr, offset, length) <- Stream.popOutputBuffer
outChunks <- Stream.unsafeInterleave
(fillBuffers defaultCompressBufferSize inChunks)
return (S.PS outFPtr offset length : outChunks)
else do fillBuffers defaultCompressBufferSize inChunks
Stream.StreamEnd -> do
inputBufferEmpty <- Stream.inputBufferEmpty
assert inputBufferEmpty $ return ()
outputBufferBytesAvailable <- Stream.outputBufferBytesAvailable
if outputBufferBytesAvailable > 0
then do (outFPtr, offset, length) <- Stream.popOutputBuffer
Stream.finalise
return [S.PS outFPtr offset length]
else do Stream.finalise
return []
{-# NOINLINE decompress #-}
decompress
:: DecompressParams
-> L.ByteString
-> L.ByteString
decompress (DecompressParams memLevel initChunkSize) input =
L.fromChunks $ Stream.run $ do
Stream.decompressInit Stream.Silent memLevel
case L.toChunks input of
[] -> fillBuffers 4 [] --always an error anyway
S.PS inFPtr offset length : chunks -> do
Stream.pushInputBuffer inFPtr offset length
fillBuffers initChunkSize chunks
where
-- we flick between two states:
-- * where one or other buffer is empty
-- - in which case we refill one or both
-- * where both buffers are non-empty
-- - in which case we compress until a buffer is empty
fillBuffers :: Int
-> [S.ByteString]
-> Stream [S.ByteString]
fillBuffers outChunkSize inChunks = do
-- in this state there are two possabilities:
-- * no outbut buffer space is available
-- - in which case we must make more available
-- * no input buffer is available
-- - in which case we must supply more
inputBufferEmpty <- Stream.inputBufferEmpty
outputBufferFull <- Stream.outputBufferFull
assert (inputBufferEmpty || outputBufferFull) $ return ()
when outputBufferFull $ do
outFPtr <- Stream.unsafeLiftIO (S.mallocByteString outChunkSize)
Stream.pushOutputBuffer outFPtr 0 outChunkSize
if inputBufferEmpty
then case inChunks of
[] -> drainBuffers []
S.PS inFPtr offset length : inChunks' -> do
Stream.pushInputBuffer inFPtr offset length
drainBuffers inChunks'
else drainBuffers inChunks
drainBuffers ::
[S.ByteString]
-> Stream [S.ByteString]
drainBuffers inChunks = do
inputBufferEmpty' <- Stream.inputBufferEmpty
outputBufferFull' <- Stream.outputBufferFull
assert(not outputBufferFull'
&& (null inChunks || not inputBufferEmpty')) $ return ()
-- this invariant guarantees we can always make forward progress or at
-- least detect premature EOF
status <- Stream.decompress
case status of
Stream.Ok -> do
outputBufferFull <- Stream.outputBufferFull
if outputBufferFull
then do (outFPtr, offset, length) <- Stream.popOutputBuffer
outChunks <- Stream.unsafeInterleave
(fillBuffers defaultDecompressBufferSize inChunks)
return (S.PS outFPtr offset length : outChunks)
else do -- We need to detect if we ran out of input:
inputBufferEmpty <- Stream.inputBufferEmpty
if inputBufferEmpty && null inChunks
then fail "premature end of compressed stream"
else fillBuffers defaultDecompressBufferSize inChunks
Stream.StreamEnd -> do
-- Note that there may be input bytes still available if the stream
-- is embeded in some other data stream. Here we just silently discard
-- any trailing data.
outputBufferBytesAvailable <- Stream.outputBufferBytesAvailable
if outputBufferBytesAvailable > 0
then do (outFPtr, offset, length) <- Stream.popOutputBuffer
Stream.finalise
return [S.PS outFPtr offset length]
else do Stream.finalise
return []
|
