| 1 |
use std::error::Error; |
| 2 |
use std::fmt; |
| 3 |
use std::io; |
| 4 |
use std::mem::MaybeUninit; |
| 5 |
|
| 6 |
use crate::ffi::{self, Backend, Deflate, DeflateBackend, ErrorMessage, Inflate, InflateBackend}; |
| 7 |
use crate::Compression; |
| 8 |
|
| 9 |
/// Raw in-memory compression stream for blocks of data. |
| 10 |
/// |
| 11 |
/// This type is the building block for the I/O streams in the rest of this |
| 12 |
/// crate. It requires more management than the [`Read`]/[`Write`] API but is |
| 13 |
/// maximally flexible in terms of accepting input from any source and being |
| 14 |
/// able to produce output to any memory location. |
| 15 |
/// |
| 16 |
/// It is recommended to use the I/O stream adaptors over this type as they're |
| 17 |
/// easier to use. |
| 18 |
/// |
| 19 |
/// [`Read`]: https://doc.rust-lang.org/std/io/trait.Read.html |
| 20 |
/// [`Write`]: https://doc.rust-lang.org/std/io/trait.Write.html |
| 21 |
#[derive(Debug)] |
| 22 |
pub struct Compress { |
| 23 |
inner: Deflate, |
| 24 |
} |
| 25 |
|
| 26 |
/// Raw in-memory decompression stream for blocks of data. |
| 27 |
/// |
| 28 |
/// This type is the building block for the I/O streams in the rest of this |
| 29 |
/// crate. It requires more management than the [`Read`]/[`Write`] API but is |
| 30 |
/// maximally flexible in terms of accepting input from any source and being |
| 31 |
/// able to produce output to any memory location. |
| 32 |
/// |
| 33 |
/// It is recommended to use the I/O stream adaptors over this type as they're |
| 34 |
/// easier to use. |
| 35 |
/// |
| 36 |
/// [`Read`]: https://doc.rust-lang.org/std/io/trait.Read.html |
| 37 |
/// [`Write`]: https://doc.rust-lang.org/std/io/trait.Write.html |
| 38 |
#[derive(Debug)] |
| 39 |
pub struct Decompress { |
| 40 |
inner: Inflate, |
| 41 |
} |
| 42 |
|
| 43 |
/// Values which indicate the form of flushing to be used when compressing |
| 44 |
/// in-memory data. |
| 45 |
#[derive(Copy, Clone, PartialEq, Eq, Debug)] |
| 46 |
#[non_exhaustive] |
| 47 |
#[allow(clippy::unnecessary_cast)] |
| 48 |
pub enum FlushCompress { |
| 49 |
/// A typical parameter for passing to compression/decompression functions, |
| 50 |
/// this indicates that the underlying stream to decide how much data to |
| 51 |
/// accumulate before producing output in order to maximize compression. |
| 52 |
None = ffi::MZ_NO_FLUSH as isize, |
| 53 |
|
| 54 |
/// All pending output is flushed to the output buffer, but the output is |
| 55 |
/// not aligned to a byte boundary. |
| 56 |
/// |
| 57 |
/// All input data so far will be available to the decompressor (as with |
| 58 |
/// `Flush::Sync`). This completes the current deflate block and follows it |
| 59 |
/// with an empty fixed codes block that is 10 bits long, and it assures |
| 60 |
/// that enough bytes are output in order for the decompressor to finish the |
| 61 |
/// block before the empty fixed code block. |
| 62 |
Partial = ffi::MZ_PARTIAL_FLUSH as isize, |
| 63 |
|
| 64 |
/// All pending output is flushed to the output buffer and the output is |
| 65 |
/// aligned on a byte boundary so that the decompressor can get all input |
| 66 |
/// data available so far. |
| 67 |
/// |
| 68 |
/// Flushing may degrade compression for some compression algorithms and so |
| 69 |
/// it should only be used when necessary. This will complete the current |
| 70 |
/// deflate block and follow it with an empty stored block. |
| 71 |
Sync = ffi::MZ_SYNC_FLUSH as isize, |
| 72 |
|
| 73 |
/// All output is flushed as with `Flush::Sync` and the compression state is |
| 74 |
/// reset so decompression can restart from this point if previous |
| 75 |
/// compressed data has been damaged or if random access is desired. |
| 76 |
/// |
| 77 |
/// Using this option too often can seriously degrade compression. |
| 78 |
Full = ffi::MZ_FULL_FLUSH as isize, |
| 79 |
|
| 80 |
/// Pending input is processed and pending output is flushed. |
| 81 |
/// |
| 82 |
/// The return value may indicate that the stream is not yet done and more |
| 83 |
/// data has yet to be processed. |
| 84 |
Finish = ffi::MZ_FINISH as isize, |
| 85 |
} |
| 86 |
|
| 87 |
/// Values which indicate the form of flushing to be used when |
| 88 |
/// decompressing in-memory data. |
| 89 |
#[derive(Copy, Clone, PartialEq, Eq, Debug)] |
| 90 |
#[non_exhaustive] |
| 91 |
#[allow(clippy::unnecessary_cast)] |
| 92 |
pub enum FlushDecompress { |
| 93 |
/// A typical parameter for passing to compression/decompression functions, |
| 94 |
/// this indicates that the underlying stream to decide how much data to |
| 95 |
/// accumulate before producing output in order to maximize compression. |
| 96 |
None = ffi::MZ_NO_FLUSH as isize, |
| 97 |
|
| 98 |
/// All pending output is flushed to the output buffer and the output is |
| 99 |
/// aligned on a byte boundary so that the decompressor can get all input |
| 100 |
/// data available so far. |
| 101 |
/// |
| 102 |
/// Flushing may degrade compression for some compression algorithms and so |
| 103 |
/// it should only be used when necessary. This will complete the current |
| 104 |
/// deflate block and follow it with an empty stored block. |
| 105 |
Sync = ffi::MZ_SYNC_FLUSH as isize, |
| 106 |
|
| 107 |
/// Pending input is processed and pending output is flushed. |
| 108 |
/// |
| 109 |
/// The return value may indicate that the stream is not yet done and more |
| 110 |
/// data has yet to be processed. |
| 111 |
Finish = ffi::MZ_FINISH as isize, |
| 112 |
} |
| 113 |
|
| 114 |
/// The inner state for an error when decompressing |
| 115 |
#[derive(Clone, Debug)] |
| 116 |
pub(crate) enum DecompressErrorInner { |
| 117 |
General { msg: ErrorMessage }, |
| 118 |
NeedsDictionary(u32), |
| 119 |
} |
| 120 |
|
| 121 |
/// Error returned when a decompression object finds that the input stream of |
| 122 |
/// bytes was not a valid input stream of bytes. |
| 123 |
#[derive(Clone, Debug)] |
| 124 |
pub struct DecompressError(pub(crate) DecompressErrorInner); |
| 125 |
|
| 126 |
impl DecompressError { |
| 127 |
/// Indicates whether decompression failed due to requiring a dictionary. |
| 128 |
/// |
| 129 |
/// The resulting integer is the Adler-32 checksum of the dictionary |
| 130 |
/// required. |
| 131 |
pub fn needs_dictionary(&self) -> Option<u32> { |
| 132 |
match self.0 { |
| 133 |
DecompressErrorInner::NeedsDictionary(adler) => Some(adler), |
| 134 |
_ => None, |
| 135 |
} |
| 136 |
} |
| 137 |
} |
| 138 |
|
| 139 |
#[inline] |
| 140 |
pub(crate) fn decompress_failed<T>(msg: ErrorMessage) -> Result<T, DecompressError> { |
| 141 |
Err(DecompressError(DecompressErrorInner::General { msg })) |
| 142 |
} |
| 143 |
|
| 144 |
#[inline] |
| 145 |
pub(crate) fn decompress_need_dict<T>(adler: u32) -> Result<T, DecompressError> { |
| 146 |
Err(DecompressError(DecompressErrorInner::NeedsDictionary( |
| 147 |
adler, |
| 148 |
))) |
| 149 |
} |
| 150 |
|
| 151 |
/// Error returned when a compression object is used incorrectly or otherwise |
| 152 |
/// generates an error. |
| 153 |
#[derive(Clone, Debug)] |
| 154 |
pub struct CompressError { |
| 155 |
pub(crate) msg: ErrorMessage, |
| 156 |
} |
| 157 |
|
| 158 |
#[inline] |
| 159 |
pub(crate) fn compress_failed<T>(msg: ErrorMessage) -> Result<T, CompressError> { |
| 160 |
Err(CompressError { msg }) |
| 161 |
} |
| 162 |
|
| 163 |
/// Possible status results of compressing some data or successfully |
| 164 |
/// decompressing a block of data. |
| 165 |
#[derive(Copy, Clone, PartialEq, Eq, Debug)] |
| 166 |
pub enum Status { |
| 167 |
/// Indicates success. |
| 168 |
/// |
| 169 |
/// Means that more input may be needed but isn't available |
| 170 |
/// and/or there's more output to be written but the output buffer is full. |
| 171 |
Ok, |
| 172 |
|
| 173 |
/// Indicates that forward progress is not possible due to input or output |
| 174 |
/// buffers being empty. |
| 175 |
/// |
| 176 |
/// For compression it means the input buffer needs some more data or the |
| 177 |
/// output buffer needs to be freed up before trying again. |
| 178 |
/// |
| 179 |
/// For decompression this means that more input is needed to continue or |
| 180 |
/// the output buffer isn't large enough to contain the result. The function |
| 181 |
/// can be called again after fixing both. |
| 182 |
BufError, |
| 183 |
|
| 184 |
/// Indicates that all input has been consumed and all output bytes have |
| 185 |
/// been written. Decompression/compression should not be called again. |
| 186 |
/// |
| 187 |
/// For decompression with zlib streams the adler-32 of the decompressed |
| 188 |
/// data has also been verified. |
| 189 |
StreamEnd, |
| 190 |
} |
| 191 |
|
| 192 |
impl Compress { |
| 193 |
/// Creates a new object ready for compressing data that it's given. |
| 194 |
/// |
| 195 |
/// The `level` argument here indicates what level of compression is going |
| 196 |
/// to be performed, and the `zlib_header` argument indicates whether the |
| 197 |
/// output data should have a zlib header or not. |
| 198 |
pub fn new(level: Compression, zlib_header: bool) -> Compress { |
| 199 |
Compress { |
| 200 |
inner: Deflate::make(level, zlib_header, ffi::MZ_DEFAULT_WINDOW_BITS as u8), |
| 201 |
} |
| 202 |
} |
| 203 |
|
| 204 |
/// Creates a new object ready for compressing data that it's given. |
| 205 |
/// |
| 206 |
/// The `level` argument here indicates what level of compression is going |
| 207 |
/// to be performed, and the `zlib_header` argument indicates whether the |
| 208 |
/// output data should have a zlib header or not. The `window_bits` parameter |
| 209 |
/// indicates the base-2 logarithm of the sliding window size and must be |
| 210 |
/// between 9 and 15. |
| 211 |
/// |
| 212 |
/// # Panics |
| 213 |
/// |
| 214 |
/// If `window_bits` does not fall into the range 9 ..= 15, |
| 215 |
/// `new_with_window_bits` will panic. |
| 216 |
#[cfg(feature = "any_zlib")] |
| 217 |
pub fn new_with_window_bits( |
| 218 |
level: Compression, |
| 219 |
zlib_header: bool, |
| 220 |
window_bits: u8, |
| 221 |
) -> Compress { |
| 222 |
assert!( |
| 223 |
window_bits > 8 && window_bits < 16, |
| 224 |
"window_bits must be within 9 ..= 15" |
| 225 |
); |
| 226 |
Compress { |
| 227 |
inner: Deflate::make(level, zlib_header, window_bits), |
| 228 |
} |
| 229 |
} |
| 230 |
|
| 231 |
/// Creates a new object ready for compressing data that it's given. |
| 232 |
/// |
| 233 |
/// The `level` argument here indicates what level of compression is going |
| 234 |
/// to be performed. |
| 235 |
/// |
| 236 |
/// The Compress object produced by this constructor outputs gzip headers |
| 237 |
/// for the compressed data. |
| 238 |
/// |
| 239 |
/// # Panics |
| 240 |
/// |
| 241 |
/// If `window_bits` does not fall into the range 9 ..= 15, |
| 242 |
/// `new_with_window_bits` will panic. |
| 243 |
#[cfg(feature = "any_zlib")] |
| 244 |
pub fn new_gzip(level: Compression, window_bits: u8) -> Compress { |
| 245 |
assert!( |
| 246 |
window_bits > 8 && window_bits < 16, |
| 247 |
"window_bits must be within 9 ..= 15" |
| 248 |
); |
| 249 |
Compress { |
| 250 |
inner: Deflate::make(level, true, window_bits + 16), |
| 251 |
} |
| 252 |
} |
| 253 |
|
| 254 |
/// Returns the total number of input bytes which have been processed by |
| 255 |
/// this compression object. |
| 256 |
pub fn total_in(&self) -> u64 { |
| 257 |
self.inner.total_in() |
| 258 |
} |
| 259 |
|
| 260 |
/// Returns the total number of output bytes which have been produced by |
| 261 |
/// this compression object. |
| 262 |
pub fn total_out(&self) -> u64 { |
| 263 |
self.inner.total_out() |
| 264 |
} |
| 265 |
|
| 266 |
/// Specifies the compression dictionary to use. |
| 267 |
/// |
| 268 |
/// Returns the Adler-32 checksum of the dictionary. |
| 269 |
#[cfg(feature = "any_zlib")] |
| 270 |
pub fn set_dictionary(&mut self, dictionary: &[u8]) -> Result<u32, CompressError> { |
| 271 |
// SAFETY: The field `inner` must always be accessed as a raw pointer, |
| 272 |
// since it points to a cyclic structure. No copies of `inner` can be |
| 273 |
// retained for longer than the lifetime of `self.inner.inner.stream_wrapper`. |
| 274 |
let stream = self.inner.inner.stream_wrapper.inner; |
| 275 |
let rc = unsafe { |
| 276 |
(*stream).msg = std::ptr::null_mut(); |
| 277 |
assert!(dictionary.len() < ffi::uInt::MAX as usize); |
| 278 |
ffi::deflateSetDictionary(stream, dictionary.as_ptr(), dictionary.len() as ffi::uInt) |
| 279 |
}; |
| 280 |
|
| 281 |
match rc { |
| 282 |
ffi::MZ_STREAM_ERROR => compress_failed(self.inner.inner.msg()), |
| 283 |
#[allow(clippy::unnecessary_cast)] |
| 284 |
ffi::MZ_OK => Ok(unsafe { (*stream).adler } as u32), |
| 285 |
c => panic!("unknown return code: {}", c), |
| 286 |
} |
| 287 |
} |
| 288 |
|
| 289 |
/// Quickly resets this compressor without having to reallocate anything. |
| 290 |
/// |
| 291 |
/// This is equivalent to dropping this object and then creating a new one. |
| 292 |
pub fn reset(&mut self) { |
| 293 |
self.inner.reset(); |
| 294 |
} |
| 295 |
|
| 296 |
/// Dynamically updates the compression level. |
| 297 |
/// |
| 298 |
/// This can be used to switch between compression levels for different |
| 299 |
/// kinds of data, or it can be used in conjunction with a call to reset |
| 300 |
/// to reuse the compressor. |
| 301 |
/// |
| 302 |
/// This may return an error if there wasn't enough output space to complete |
| 303 |
/// the compression of the available input data before changing the |
| 304 |
/// compression level. Flushing the stream before calling this method |
| 305 |
/// ensures that the function will succeed on the first call. |
| 306 |
#[cfg(feature = "any_zlib")] |
| 307 |
pub fn set_level(&mut self, level: Compression) -> Result<(), CompressError> { |
| 308 |
use std::os::raw::c_int; |
| 309 |
// SAFETY: The field `inner` must always be accessed as a raw pointer, |
| 310 |
// since it points to a cyclic structure. No copies of `inner` can be |
| 311 |
// retained for longer than the lifetime of `self.inner.inner.stream_wrapper`. |
| 312 |
let stream = self.inner.inner.stream_wrapper.inner; |
| 313 |
unsafe { |
| 314 |
(*stream).msg = std::ptr::null_mut(); |
| 315 |
} |
| 316 |
let rc = unsafe { ffi::deflateParams(stream, level.0 as c_int, ffi::MZ_DEFAULT_STRATEGY) }; |
| 317 |
|
| 318 |
match rc { |
| 319 |
ffi::MZ_OK => Ok(()), |
| 320 |
ffi::MZ_BUF_ERROR => compress_failed(self.inner.inner.msg()), |
| 321 |
c => panic!("unknown return code: {}", c), |
| 322 |
} |
| 323 |
} |
| 324 |
|
| 325 |
/// Compresses the input data into the output, consuming only as much |
| 326 |
/// input as needed and writing as much output as possible. |
| 327 |
/// |
| 328 |
/// The flush option can be any of the available `FlushCompress` parameters. |
| 329 |
/// |
| 330 |
/// To learn how much data was consumed or how much output was produced, use |
| 331 |
/// the `total_in` and `total_out` functions before/after this is called. |
| 332 |
pub fn compress( |
| 333 |
&mut self, |
| 334 |
input: &[u8], |
| 335 |
output: &mut [u8], |
| 336 |
flush: FlushCompress, |
| 337 |
) -> Result<Status, CompressError> { |
| 338 |
self.inner.compress(input, output, flush) |
| 339 |
} |
| 340 |
|
| 341 |
/// Similar to [`Self::compress`] but accepts uninitialized buffer. |
| 342 |
/// |
| 343 |
/// If you want to avoid the overhead of zero initializing the |
| 344 |
/// buffer and you don't want to use a [`Vec`], then please use |
| 345 |
/// this API. |
| 346 |
pub fn compress_uninit( |
| 347 |
&mut self, |
| 348 |
input: &[u8], |
| 349 |
output: &mut [MaybeUninit<u8>], |
| 350 |
flush: FlushCompress, |
| 351 |
) -> Result<Status, CompressError> { |
| 352 |
self.inner.compress_uninit(input, output, flush) |
| 353 |
} |
| 354 |
|
| 355 |
/// Compresses the input data into the extra space of the output, consuming |
| 356 |
/// only as much input as needed and writing as much output as possible. |
| 357 |
/// |
| 358 |
/// This function has the same semantics as `compress`, except that the |
| 359 |
/// length of `vec` is managed by this function. This will not reallocate |
| 360 |
/// the vector provided or attempt to grow it, so space for the output must |
| 361 |
/// be reserved in the output vector by the caller before calling this |
| 362 |
/// function. |
| 363 |
pub fn compress_vec( |
| 364 |
&mut self, |
| 365 |
input: &[u8], |
| 366 |
output: &mut Vec<u8>, |
| 367 |
flush: FlushCompress, |
| 368 |
) -> Result<Status, CompressError> { |
| 369 |
// SAFETY: bytes_written is the number of bytes written into `out` |
| 370 |
unsafe { |
| 371 |
write_to_spare_capacity_of_vec(output, |out| { |
| 372 |
let before = self.total_out(); |
| 373 |
let ret = self.compress_uninit(input, out, flush); |
| 374 |
let bytes_written = self.total_out() - before; |
| 375 |
(bytes_written as usize, ret) |
| 376 |
}) |
| 377 |
} |
| 378 |
} |
| 379 |
} |
| 380 |
|
| 381 |
impl Decompress { |
| 382 |
/// Creates a new object ready for decompressing data that it's given. |
| 383 |
/// |
| 384 |
/// The `zlib_header` argument indicates whether the input data is expected |
| 385 |
/// to have a zlib header or not. |
| 386 |
pub fn new(zlib_header: bool) -> Decompress { |
| 387 |
Decompress { |
| 388 |
inner: Inflate::make(zlib_header, ffi::MZ_DEFAULT_WINDOW_BITS as u8), |
| 389 |
} |
| 390 |
} |
| 391 |
|
| 392 |
/// Creates a new object ready for decompressing data that it's given. |
| 393 |
/// |
| 394 |
/// The `zlib_header` argument indicates whether the input data is expected |
| 395 |
/// to have a zlib header or not. The `window_bits` parameter indicates the |
| 396 |
/// base-2 logarithm of the sliding window size and must be between 9 and 15. |
| 397 |
/// |
| 398 |
/// # Panics |
| 399 |
/// |
| 400 |
/// If `window_bits` does not fall into the range 9 ..= 15, |
| 401 |
/// `new_with_window_bits` will panic. |
| 402 |
#[cfg(feature = "any_zlib")] |
| 403 |
pub fn new_with_window_bits(zlib_header: bool, window_bits: u8) -> Decompress { |
| 404 |
assert!( |
| 405 |
window_bits > 8 && window_bits < 16, |
| 406 |
"window_bits must be within 9 ..= 15" |
| 407 |
); |
| 408 |
Decompress { |
| 409 |
inner: Inflate::make(zlib_header, window_bits), |
| 410 |
} |
| 411 |
} |
| 412 |
|
| 413 |
/// Creates a new object ready for decompressing data that it's given. |
| 414 |
/// |
| 415 |
/// The Decompress object produced by this constructor expects gzip headers |
| 416 |
/// for the compressed data. |
| 417 |
/// |
| 418 |
/// # Panics |
| 419 |
/// |
| 420 |
/// If `window_bits` does not fall into the range 9 ..= 15, |
| 421 |
/// `new_with_window_bits` will panic. |
| 422 |
#[cfg(feature = "any_zlib")] |
| 423 |
pub fn new_gzip(window_bits: u8) -> Decompress { |
| 424 |
assert!( |
| 425 |
window_bits > 8 && window_bits < 16, |
| 426 |
"window_bits must be within 9 ..= 15" |
| 427 |
); |
| 428 |
Decompress { |
| 429 |
inner: Inflate::make(true, window_bits + 16), |
| 430 |
} |
| 431 |
} |
| 432 |
|
| 433 |
/// Returns the total number of input bytes which have been processed by |
| 434 |
/// this decompression object. |
| 435 |
pub fn total_in(&self) -> u64 { |
| 436 |
self.inner.total_in() |
| 437 |
} |
| 438 |
|
| 439 |
/// Returns the total number of output bytes which have been produced by |
| 440 |
/// this decompression object. |
| 441 |
pub fn total_out(&self) -> u64 { |
| 442 |
self.inner.total_out() |
| 443 |
} |
| 444 |
|
| 445 |
/// Decompresses the input data into the output, consuming only as much |
| 446 |
/// input as needed and writing as much output as possible. |
| 447 |
/// |
| 448 |
/// The flush option can be any of the available `FlushDecompress` parameters. |
| 449 |
/// |
| 450 |
/// If the first call passes `FlushDecompress::Finish` it is assumed that |
| 451 |
/// the input and output buffers are both sized large enough to decompress |
| 452 |
/// the entire stream in a single call. |
| 453 |
/// |
| 454 |
/// A flush value of `FlushDecompress::Finish` indicates that there are no |
| 455 |
/// more source bytes available beside what's already in the input buffer, |
| 456 |
/// and the output buffer is large enough to hold the rest of the |
| 457 |
/// decompressed data. |
| 458 |
/// |
| 459 |
/// To learn how much data was consumed or how much output was produced, use |
| 460 |
/// the `total_in` and `total_out` functions before/after this is called. |
| 461 |
/// |
| 462 |
/// # Errors |
| 463 |
/// |
| 464 |
/// If the input data to this instance of `Decompress` is not a valid |
| 465 |
/// zlib/deflate stream then this function may return an instance of |
| 466 |
/// `DecompressError` to indicate that the stream of input bytes is corrupted. |
| 467 |
pub fn decompress( |
| 468 |
&mut self, |
| 469 |
input: &[u8], |
| 470 |
output: &mut [u8], |
| 471 |
flush: FlushDecompress, |
| 472 |
) -> Result<Status, DecompressError> { |
| 473 |
self.inner.decompress(input, output, flush) |
| 474 |
} |
| 475 |
|
| 476 |
/// Similar to [`Self::decompress`] but accepts uninitialized buffer |
| 477 |
/// |
| 478 |
/// If you want to avoid the overhead of zero initializing the |
| 479 |
/// buffer and you don't want to use a [`Vec`], then please use |
| 480 |
/// this API. |
| 481 |
pub fn decompress_uninit( |
| 482 |
&mut self, |
| 483 |
input: &[u8], |
| 484 |
output: &mut [MaybeUninit<u8>], |
| 485 |
flush: FlushDecompress, |
| 486 |
) -> Result<Status, DecompressError> { |
| 487 |
self.inner.decompress_uninit(input, output, flush) |
| 488 |
} |
| 489 |
|
| 490 |
/// Decompresses the input data into the extra space in the output vector |
| 491 |
/// specified by `output`. |
| 492 |
/// |
| 493 |
/// This function has the same semantics as `decompress`, except that the |
| 494 |
/// length of `vec` is managed by this function. This will not reallocate |
| 495 |
/// the vector provided or attempt to grow it, so space for the output must |
| 496 |
/// be reserved in the output vector by the caller before calling this |
| 497 |
/// function. |
| 498 |
/// |
| 499 |
/// # Errors |
| 500 |
/// |
| 501 |
/// If the input data to this instance of `Decompress` is not a valid |
| 502 |
/// zlib/deflate stream then this function may return an instance of |
| 503 |
/// `DecompressError` to indicate that the stream of input bytes is corrupted. |
| 504 |
pub fn decompress_vec( |
| 505 |
&mut self, |
| 506 |
input: &[u8], |
| 507 |
output: &mut Vec<u8>, |
| 508 |
flush: FlushDecompress, |
| 509 |
) -> Result<Status, DecompressError> { |
| 510 |
// SAFETY: bytes_written is the number of bytes written into `out` |
| 511 |
unsafe { |
| 512 |
write_to_spare_capacity_of_vec(output, |out| { |
| 513 |
let before = self.total_out(); |
| 514 |
let ret = self.decompress_uninit(input, out, flush); |
| 515 |
let bytes_written = self.total_out() - before; |
| 516 |
(bytes_written as usize, ret) |
| 517 |
}) |
| 518 |
} |
| 519 |
} |
| 520 |
|
| 521 |
/// Specifies the decompression dictionary to use. |
| 522 |
#[cfg(feature = "any_zlib")] |
| 523 |
pub fn set_dictionary(&mut self, dictionary: &[u8]) -> Result<u32, DecompressError> { |
| 524 |
// SAFETY: The field `inner` must always be accessed as a raw pointer, |
| 525 |
// since it points to a cyclic structure. No copies of `inner` can be |
| 526 |
// retained for longer than the lifetime of `self.inner.inner.stream_wrapper`. |
| 527 |
let stream = self.inner.inner.stream_wrapper.inner; |
| 528 |
let rc = unsafe { |
| 529 |
(*stream).msg = std::ptr::null_mut(); |
| 530 |
assert!(dictionary.len() < ffi::uInt::MAX as usize); |
| 531 |
ffi::inflateSetDictionary(stream, dictionary.as_ptr(), dictionary.len() as ffi::uInt) |
| 532 |
}; |
| 533 |
|
| 534 |
#[allow(clippy::unnecessary_cast)] |
| 535 |
match rc { |
| 536 |
ffi::MZ_STREAM_ERROR => decompress_failed(self.inner.inner.msg()), |
| 537 |
ffi::MZ_DATA_ERROR => decompress_need_dict(unsafe { (*stream).adler } as u32), |
| 538 |
ffi::MZ_OK => Ok(unsafe { (*stream).adler } as u32), |
| 539 |
c => panic!("unknown return code: {}", c), |
| 540 |
} |
| 541 |
} |
| 542 |
|
| 543 |
/// Performs the equivalent of replacing this decompression state with a |
| 544 |
/// freshly allocated copy. |
| 545 |
/// |
| 546 |
/// This function may not allocate memory, though, and attempts to reuse any |
| 547 |
/// previously existing resources. |
| 548 |
/// |
| 549 |
/// The argument provided here indicates whether the reset state will |
| 550 |
/// attempt to decode a zlib header first or not. |
| 551 |
pub fn reset(&mut self, zlib_header: bool) { |
| 552 |
self.inner.reset(zlib_header); |
| 553 |
} |
| 554 |
} |
| 555 |
|
| 556 |
impl Error for DecompressError {} |
| 557 |
|
| 558 |
impl DecompressError { |
| 559 |
/// Retrieve the implementation's message about why the operation failed, if one exists. |
| 560 |
pub fn message(&self) -> Option<&str> { |
| 561 |
match &self.0 { |
| 562 |
DecompressErrorInner::General { msg } => msg.get(), |
| 563 |
_ => None, |
| 564 |
} |
| 565 |
} |
| 566 |
} |
| 567 |
|
| 568 |
impl From<DecompressError> for io::Error { |
| 569 |
fn from(data: DecompressError) -> io::Error { |
| 570 |
io::Error::new(io::ErrorKind::Other, data) |
| 571 |
} |
| 572 |
} |
| 573 |
|
| 574 |
impl fmt::Display for DecompressError { |
| 575 |
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
| 576 |
let msg = match &self.0 { |
| 577 |
DecompressErrorInner::General { msg } => msg.get(), |
| 578 |
DecompressErrorInner::NeedsDictionary { .. } => Some("requires a dictionary"), |
| 579 |
}; |
| 580 |
match msg { |
| 581 |
Some(msg) => write!(f, "deflate decompression error: {msg}"), |
| 582 |
None => write!(f, "deflate decompression error"), |
| 583 |
} |
| 584 |
} |
| 585 |
} |
| 586 |
|
| 587 |
impl Error for CompressError {} |
| 588 |
|
| 589 |
impl CompressError { |
| 590 |
/// Retrieve the implementation's message about why the operation failed, if one exists. |
| 591 |
pub fn message(&self) -> Option<&str> { |
| 592 |
self.msg.get() |
| 593 |
} |
| 594 |
} |
| 595 |
|
| 596 |
impl From<CompressError> for io::Error { |
| 597 |
fn from(data: CompressError) -> io::Error { |
| 598 |
io::Error::new(io::ErrorKind::Other, data) |
| 599 |
} |
| 600 |
} |
| 601 |
|
| 602 |
impl fmt::Display for CompressError { |
| 603 |
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
| 604 |
match self.msg.get() { |
| 605 |
Some(msg) => write!(f, "deflate compression error: {msg}"), |
| 606 |
None => write!(f, "deflate compression error"), |
| 607 |
} |
| 608 |
} |
| 609 |
} |
| 610 |
|
| 611 |
/// Allows `writer` to write data into the spare capacity of the `output` vector. |
| 612 |
/// This will not reallocate the vector provided or attempt to grow it, so space |
| 613 |
/// for the `output` must be reserved by the caller before calling this |
| 614 |
/// function. |
| 615 |
/// |
| 616 |
/// `writer` needs to return the number of bytes written (and can also return |
| 617 |
/// another arbitrary return value). |
| 618 |
/// |
| 619 |
/// # Safety: |
| 620 |
/// |
| 621 |
/// The length returned by the `writer` must be equal to actual number of bytes written |
| 622 |
/// to the uninitialized slice passed in and initialized. |
| 623 |
unsafe fn write_to_spare_capacity_of_vec<T>( |
| 624 |
output: &mut Vec<u8>, |
| 625 |
writer: impl FnOnce(&mut [MaybeUninit<u8>]) -> (usize, T), |
| 626 |
) -> T { |
| 627 |
let cap = output.capacity(); |
| 628 |
let len = output.len(); |
| 629 |
|
| 630 |
let (bytes_written, ret) = writer(output.spare_capacity_mut()); |
| 631 |
output.set_len(cap.min(len + bytes_written)); // Sanitizes `bytes_written`. |
| 632 |
|
| 633 |
ret |
| 634 |
} |
| 635 |
|
| 636 |
#[cfg(test)] |
| 637 |
mod tests { |
| 638 |
use std::io::Write; |
| 639 |
|
| 640 |
use crate::write; |
| 641 |
use crate::{Compression, Decompress, FlushDecompress}; |
| 642 |
|
| 643 |
#[cfg(feature = "any_zlib")] |
| 644 |
use crate::{Compress, FlushCompress}; |
| 645 |
|
| 646 |
#[test] |
| 647 |
fn issue51() { |
| 648 |
let data = [ |
| 649 |
0x1f, 0x8b, 0x08, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0xb3, 0xc9, 0x28, 0xc9, |
| 650 |
0xcd, 0xb1, 0xe3, 0xe5, 0xb2, 0xc9, 0x48, 0x4d, 0x4c, 0xb1, 0xb3, 0x29, 0xc9, 0x2c, |
| 651 |
0xc9, 0x49, 0xb5, 0x33, 0x31, 0x30, 0x51, 0xf0, 0xcb, 0x2f, 0x51, 0x70, 0xcb, 0x2f, |
| 652 |
0xcd, 0x4b, 0xb1, 0xd1, 0x87, 0x08, 0xda, 0xe8, 0x83, 0x95, 0x00, 0x95, 0x26, 0xe5, |
| 653 |
0xa7, 0x54, 0x2a, 0x24, 0xa5, 0x27, 0xe7, 0xe7, 0xe4, 0x17, 0xd9, 0x2a, 0x95, 0x67, |
| 654 |
0x64, 0x96, 0xa4, 0x2a, 0x81, 0x8c, 0x48, 0x4e, 0xcd, 0x2b, 0x49, 0x2d, 0xb2, 0xb3, |
| 655 |
0xc9, 0x30, 0x44, 0x37, 0x01, 0x28, 0x62, 0xa3, 0x0f, 0x95, 0x06, 0xd9, 0x05, 0x54, |
| 656 |
0x04, 0xe5, 0xe5, 0xa5, 0x67, 0xe6, 0x55, 0xe8, 0x1b, 0xea, 0x99, 0xe9, 0x19, 0x21, |
| 657 |
0xab, 0xd0, 0x07, 0xd9, 0x01, 0x32, 0x53, 0x1f, 0xea, 0x3e, 0x00, 0x94, 0x85, 0xeb, |
| 658 |
0xe4, 0xa8, 0x00, 0x00, 0x00, |
| 659 |
]; |
| 660 |
|
| 661 |
let mut decoded = Vec::with_capacity(data.len() * 2); |
| 662 |
|
| 663 |
let mut d = Decompress::new(false); |
| 664 |
// decompressed whole deflate stream |
| 665 |
assert!(d |
| 666 |
.decompress_vec(&data[10..], &mut decoded, FlushDecompress::Finish) |
| 667 |
.is_ok()); |
| 668 |
|
| 669 |
// decompress data that has nothing to do with the deflate stream (this |
| 670 |
// used to panic) |
| 671 |
drop(d.decompress_vec(&[0], &mut decoded, FlushDecompress::None)); |
| 672 |
} |
| 673 |
|
| 674 |
#[test] |
| 675 |
fn reset() { |
| 676 |
let string = "hello world".as_bytes(); |
| 677 |
let mut zlib = Vec::new(); |
| 678 |
let mut deflate = Vec::new(); |
| 679 |
|
| 680 |
let comp = Compression::default(); |
| 681 |
write::ZlibEncoder::new(&mut zlib, comp) |
| 682 |
.write_all(string) |
| 683 |
.unwrap(); |
| 684 |
write::DeflateEncoder::new(&mut deflate, comp) |
| 685 |
.write_all(string) |
| 686 |
.unwrap(); |
| 687 |
|
| 688 |
let mut dst = [0; 1024]; |
| 689 |
let mut decoder = Decompress::new(true); |
| 690 |
decoder |
| 691 |
.decompress(&zlib, &mut dst, FlushDecompress::Finish) |
| 692 |
.unwrap(); |
| 693 |
assert_eq!(decoder.total_out(), string.len() as u64); |
| 694 |
assert!(dst.starts_with(string)); |
| 695 |
|
| 696 |
decoder.reset(false); |
| 697 |
decoder |
| 698 |
.decompress(&deflate, &mut dst, FlushDecompress::Finish) |
| 699 |
.unwrap(); |
| 700 |
assert_eq!(decoder.total_out(), string.len() as u64); |
| 701 |
assert!(dst.starts_with(string)); |
| 702 |
} |
| 703 |
|
| 704 |
#[cfg(feature = "any_zlib")] |
| 705 |
#[test] |
| 706 |
fn set_dictionary_with_zlib_header() { |
| 707 |
let string = "hello, hello!".as_bytes(); |
| 708 |
let dictionary = "hello".as_bytes(); |
| 709 |
|
| 710 |
let mut encoded = Vec::with_capacity(1024); |
| 711 |
|
| 712 |
let mut encoder = Compress::new(Compression::default(), true); |
| 713 |
|
| 714 |
let dictionary_adler = encoder.set_dictionary(&dictionary).unwrap(); |
| 715 |
|
| 716 |
encoder |
| 717 |
.compress_vec(string, &mut encoded, FlushCompress::Finish) |
| 718 |
.unwrap(); |
| 719 |
|
| 720 |
assert_eq!(encoder.total_in(), string.len() as u64); |
| 721 |
assert_eq!(encoder.total_out(), encoded.len() as u64); |
| 722 |
|
| 723 |
let mut decoder = Decompress::new(true); |
| 724 |
let mut decoded = [0; 1024]; |
| 725 |
let decompress_error = decoder |
| 726 |
.decompress(&encoded, &mut decoded, FlushDecompress::Finish) |
| 727 |
.expect_err("decompression should fail due to requiring a dictionary"); |
| 728 |
|
| 729 |
let required_adler = decompress_error.needs_dictionary() |
| 730 |
.expect("the first call to decompress should indicate a dictionary is required along with the required Adler-32 checksum"); |
| 731 |
|
| 732 |
assert_eq!(required_adler, dictionary_adler, |
| 733 |
"the Adler-32 checksum should match the value when the dictionary was set on the compressor"); |
| 734 |
|
| 735 |
let actual_adler = decoder.set_dictionary(&dictionary).unwrap(); |
| 736 |
|
| 737 |
assert_eq!(required_adler, actual_adler); |
| 738 |
|
| 739 |
// Decompress the rest of the input to the remainder of the output buffer |
| 740 |
let total_in = decoder.total_in(); |
| 741 |
let total_out = decoder.total_out(); |
| 742 |
|
| 743 |
let decompress_result = decoder.decompress( |
| 744 |
&encoded[total_in as usize..], |
| 745 |
&mut decoded[total_out as usize..], |
| 746 |
FlushDecompress::Finish, |
| 747 |
); |
| 748 |
assert!(decompress_result.is_ok()); |
| 749 |
|
| 750 |
assert_eq!(&decoded[..decoder.total_out() as usize], string); |
| 751 |
} |
| 752 |
|
| 753 |
#[cfg(feature = "any_zlib")] |
| 754 |
#[test] |
| 755 |
fn set_dictionary_raw() { |
| 756 |
let string = "hello, hello!".as_bytes(); |
| 757 |
let dictionary = "hello".as_bytes(); |
| 758 |
|
| 759 |
let mut encoded = Vec::with_capacity(1024); |
| 760 |
|
| 761 |
let mut encoder = Compress::new(Compression::default(), false); |
| 762 |
|
| 763 |
encoder.set_dictionary(&dictionary).unwrap(); |
| 764 |
|
| 765 |
encoder |
| 766 |
.compress_vec(string, &mut encoded, FlushCompress::Finish) |
| 767 |
.unwrap(); |
| 768 |
|
| 769 |
assert_eq!(encoder.total_in(), string.len() as u64); |
| 770 |
assert_eq!(encoder.total_out(), encoded.len() as u64); |
| 771 |
|
| 772 |
let mut decoder = Decompress::new(false); |
| 773 |
|
| 774 |
decoder.set_dictionary(&dictionary).unwrap(); |
| 775 |
|
| 776 |
let mut decoded = [0; 1024]; |
| 777 |
let decompress_result = decoder.decompress(&encoded, &mut decoded, FlushDecompress::Finish); |
| 778 |
|
| 779 |
assert!(decompress_result.is_ok()); |
| 780 |
|
| 781 |
assert_eq!(&decoded[..decoder.total_out() as usize], string); |
| 782 |
} |
| 783 |
|
| 784 |
#[cfg(feature = "any_zlib")] |
| 785 |
#[test] |
| 786 |
fn test_gzip_flate() { |
| 787 |
let string = "hello, hello!".as_bytes(); |
| 788 |
|
| 789 |
let mut encoded = Vec::with_capacity(1024); |
| 790 |
|
| 791 |
let mut encoder = Compress::new_gzip(Compression::default(), 9); |
| 792 |
|
| 793 |
encoder |
| 794 |
.compress_vec(string, &mut encoded, FlushCompress::Finish) |
| 795 |
.unwrap(); |
| 796 |
|
| 797 |
assert_eq!(encoder.total_in(), string.len() as u64); |
| 798 |
assert_eq!(encoder.total_out(), encoded.len() as u64); |
| 799 |
|
| 800 |
let mut decoder = Decompress::new_gzip(9); |
| 801 |
|
| 802 |
let mut decoded = [0; 1024]; |
| 803 |
decoder |
| 804 |
.decompress(&encoded, &mut decoded, FlushDecompress::Finish) |
| 805 |
.unwrap(); |
| 806 |
|
| 807 |
assert_eq!(&decoded[..decoder.total_out() as usize], string); |
| 808 |
} |
| 809 |
|
| 810 |
#[cfg(feature = "any_zlib")] |
| 811 |
#[test] |
| 812 |
fn test_error_message() { |
| 813 |
let mut decoder = Decompress::new(false); |
| 814 |
let mut decoded = [0; 128]; |
| 815 |
let garbage = b"xbvxzi"; |
| 816 |
|
| 817 |
let err = decoder |
| 818 |
.decompress(&*garbage, &mut decoded, FlushDecompress::Finish) |
| 819 |
.unwrap_err(); |
| 820 |
|
| 821 |
assert_eq!(err.message(), Some("invalid stored block lengths")); |
| 822 |
} |
| 823 |
} |