| 1 | //! This module contains functionality for decompression. | 
| 2 |  | 
|---|
| 3 | #[ cfg(feature = "with-alloc")] | 
|---|
| 4 | use crate::alloc::{boxed::Box, vec, vec::Vec}; | 
|---|
| 5 | #[ cfg(all(feature = "std", feature = "with-alloc"))] | 
|---|
| 6 | use std::error::Error; | 
|---|
| 7 |  | 
|---|
| 8 | pub mod core; | 
|---|
| 9 | mod output_buffer; | 
|---|
| 10 | pub mod stream; | 
|---|
| 11 | use self::core::*; | 
|---|
| 12 |  | 
|---|
| 13 | const TINFL_STATUS_FAILED_CANNOT_MAKE_PROGRESS: i32 = -4; | 
|---|
| 14 | const TINFL_STATUS_BAD_PARAM: i32 = -3; | 
|---|
| 15 | const TINFL_STATUS_ADLER32_MISMATCH: i32 = -2; | 
|---|
| 16 | const TINFL_STATUS_FAILED: i32 = -1; | 
|---|
| 17 | const TINFL_STATUS_DONE: i32 = 0; | 
|---|
| 18 | const TINFL_STATUS_NEEDS_MORE_INPUT: i32 = 1; | 
|---|
| 19 | const TINFL_STATUS_HAS_MORE_OUTPUT: i32 = 2; | 
|---|
| 20 |  | 
|---|
| 21 | /// Return status codes. | 
|---|
| 22 | #[ repr(i8)] | 
|---|
| 23 | #[ derive(Debug, Copy, Clone, PartialEq, Eq, Hash)] | 
|---|
| 24 | pub enum TINFLStatus { | 
|---|
| 25 | /// More input data was expected, but the caller indicated that there was no more data, so the | 
|---|
| 26 | /// input stream is likely truncated. | 
|---|
| 27 | /// | 
|---|
| 28 | /// This can't happen if you have provided the | 
|---|
| 29 | /// [`TINFL_FLAG_HAS_MORE_INPUT`][core::inflate_flags::TINFL_FLAG_HAS_MORE_INPUT] flag to the | 
|---|
| 30 | /// decompression.  By setting that flag, you indicate more input exists but is not provided, | 
|---|
| 31 | /// and so reaching the end of the input data without finding the end of the compressed stream | 
|---|
| 32 | /// would instead return a [`NeedsMoreInput`][Self::NeedsMoreInput] status. | 
|---|
| 33 | FailedCannotMakeProgress = TINFL_STATUS_FAILED_CANNOT_MAKE_PROGRESS as i8, | 
|---|
| 34 |  | 
|---|
| 35 | /// The output buffer is an invalid size; consider the `flags` parameter. | 
|---|
| 36 | BadParam = TINFL_STATUS_BAD_PARAM as i8, | 
|---|
| 37 |  | 
|---|
| 38 | /// The decompression went fine, but the adler32 checksum did not match the one | 
|---|
| 39 | /// provided in the header. | 
|---|
| 40 | Adler32Mismatch = TINFL_STATUS_ADLER32_MISMATCH as i8, | 
|---|
| 41 |  | 
|---|
| 42 | /// Failed to decompress due to invalid data. | 
|---|
| 43 | Failed = TINFL_STATUS_FAILED as i8, | 
|---|
| 44 |  | 
|---|
| 45 | /// Finished decompression without issues. | 
|---|
| 46 | /// | 
|---|
| 47 | /// This indicates the end of the compressed stream has been reached. | 
|---|
| 48 | Done = TINFL_STATUS_DONE as i8, | 
|---|
| 49 |  | 
|---|
| 50 | /// The decompressor needs more input data to continue decompressing. | 
|---|
| 51 | /// | 
|---|
| 52 | /// This occurs when there's no more consumable input, but the end of the stream hasn't been | 
|---|
| 53 | /// reached, and you have supplied the | 
|---|
| 54 | /// [`TINFL_FLAG_HAS_MORE_INPUT`][core::inflate_flags::TINFL_FLAG_HAS_MORE_INPUT] flag to the | 
|---|
| 55 | /// decompressor.  Had you not supplied that flag (which would mean you were asserting that you | 
|---|
| 56 | /// believed all the data was available) you would have gotten a | 
|---|
| 57 | /// [`FailedCannotMakeProcess`][Self::FailedCannotMakeProgress] instead. | 
|---|
| 58 | NeedsMoreInput = TINFL_STATUS_NEEDS_MORE_INPUT as i8, | 
|---|
| 59 |  | 
|---|
| 60 | /// There is still pending data that didn't fit in the output buffer. | 
|---|
| 61 | HasMoreOutput = TINFL_STATUS_HAS_MORE_OUTPUT as i8, | 
|---|
| 62 | } | 
|---|
| 63 |  | 
|---|
| 64 | impl TINFLStatus { | 
|---|
| 65 | pub fn from_i32(value: i32) -> Option<TINFLStatus> { | 
|---|
| 66 | use self::TINFLStatus::*; | 
|---|
| 67 | match value { | 
|---|
| 68 | TINFL_STATUS_FAILED_CANNOT_MAKE_PROGRESS => Some(FailedCannotMakeProgress), | 
|---|
| 69 | TINFL_STATUS_BAD_PARAM => Some(BadParam), | 
|---|
| 70 | TINFL_STATUS_ADLER32_MISMATCH => Some(Adler32Mismatch), | 
|---|
| 71 | TINFL_STATUS_FAILED => Some(Failed), | 
|---|
| 72 | TINFL_STATUS_DONE => Some(Done), | 
|---|
| 73 | TINFL_STATUS_NEEDS_MORE_INPUT => Some(NeedsMoreInput), | 
|---|
| 74 | TINFL_STATUS_HAS_MORE_OUTPUT => Some(HasMoreOutput), | 
|---|
| 75 | _ => None, | 
|---|
| 76 | } | 
|---|
| 77 | } | 
|---|
| 78 | } | 
|---|
| 79 |  | 
|---|
| 80 | /// Struct return when decompress_to_vec functions fail. | 
|---|
| 81 | #[ cfg(feature = "with-alloc")] | 
|---|
| 82 | #[ derive(Debug)] | 
|---|
| 83 | pub struct DecompressError { | 
|---|
| 84 | /// Decompressor status on failure. See [TINFLStatus] for details. | 
|---|
| 85 | pub status: TINFLStatus, | 
|---|
| 86 | /// The currently decompressed data if any. | 
|---|
| 87 | pub output: Vec<u8>, | 
|---|
| 88 | } | 
|---|
| 89 |  | 
|---|
| 90 | #[ cfg(feature = "with-alloc")] | 
|---|
| 91 | impl alloc::fmt::Display for DecompressError { | 
|---|
| 92 | #[ cold] | 
|---|
| 93 | fn fmt(&self, f: &mut ::core::fmt::Formatter<'_>) -> ::core::fmt::Result { | 
|---|
| 94 | f.write_str(data:match self.status { | 
|---|
| 95 | TINFLStatus::FailedCannotMakeProgress => "Truncated input stream", | 
|---|
| 96 | TINFLStatus::BadParam => "Invalid output buffer size", | 
|---|
| 97 | TINFLStatus::Adler32Mismatch => "Adler32 checksum mismatch", | 
|---|
| 98 | TINFLStatus::Failed => "Invalid input data", | 
|---|
| 99 | TINFLStatus::Done => "", // Unreachable | 
|---|
| 100 | TINFLStatus::NeedsMoreInput => "Truncated input stream", | 
|---|
| 101 | TINFLStatus::HasMoreOutput => "Output size exceeded the specified limit", | 
|---|
| 102 | }) | 
|---|
| 103 | } | 
|---|
| 104 | } | 
|---|
| 105 |  | 
|---|
| 106 | /// Implement Error trait only if std feature is requested as it requires std. | 
|---|
| 107 | #[ cfg(all(feature = "std", feature = "with-alloc"))] | 
|---|
| 108 | impl Error for DecompressError {} | 
|---|
| 109 |  | 
|---|
| 110 | #[ cfg(feature = "with-alloc")] | 
|---|
| 111 | fn decompress_error(status: TINFLStatus, output: Vec<u8>) -> Result<Vec<u8>, DecompressError> { | 
|---|
| 112 | Err(DecompressError { status, output }) | 
|---|
| 113 | } | 
|---|
| 114 |  | 
|---|
| 115 | /// Decompress the deflate-encoded data in `input` to a vector. | 
|---|
| 116 | /// | 
|---|
| 117 | /// NOTE: This function will not bound the output, so if the output is large enough it can result in an out of memory error. | 
|---|
| 118 | /// It is therefore suggested to not use this for anything other than test programs, use the functions with a specified limit, or | 
|---|
| 119 | /// ideally streaming decompression via the [flate2](https://github.com/alexcrichton/flate2-rs) library instead. | 
|---|
| 120 | /// | 
|---|
| 121 | /// Returns a [`Result`] containing the [`Vec`] of decompressed data on success, and a [struct][DecompressError] containing the status and so far decompressed data if any on failure. | 
|---|
| 122 | #[ inline] | 
|---|
| 123 | #[ cfg(feature = "with-alloc")] | 
|---|
| 124 | pub fn decompress_to_vec(input: &[u8]) -> Result<Vec<u8>, DecompressError> { | 
|---|
| 125 | decompress_to_vec_inner(input, flags:0, max_output_size:usize::MAX) | 
|---|
| 126 | } | 
|---|
| 127 |  | 
|---|
| 128 | /// Decompress the deflate-encoded data (with a zlib wrapper) in `input` to a vector. | 
|---|
| 129 | /// | 
|---|
| 130 | /// NOTE: This function will not bound the output, so if the output is large enough it can result in an out of memory error. | 
|---|
| 131 | /// It is therefore suggested to not use this for anything other than test programs, use the functions with a specified limit, or | 
|---|
| 132 | /// ideally streaming decompression via the [flate2](https://github.com/alexcrichton/flate2-rs) library instead. | 
|---|
| 133 | /// | 
|---|
| 134 | /// Returns a [`Result`] containing the [`Vec`] of decompressed data on success, and a [struct][DecompressError] containing the status and so far decompressed data if any on failure. | 
|---|
| 135 | #[ inline] | 
|---|
| 136 | #[ cfg(feature = "with-alloc")] | 
|---|
| 137 | pub fn decompress_to_vec_zlib(input: &[u8]) -> Result<Vec<u8>, DecompressError> { | 
|---|
| 138 | decompress_to_vec_inner( | 
|---|
| 139 | input, | 
|---|
| 140 | flags:inflate_flags::TINFL_FLAG_PARSE_ZLIB_HEADER, | 
|---|
| 141 | max_output_size:usize::MAX, | 
|---|
| 142 | ) | 
|---|
| 143 | } | 
|---|
| 144 |  | 
|---|
| 145 | /// Decompress the deflate-encoded data in `input` to a vector. | 
|---|
| 146 | /// | 
|---|
| 147 | /// The vector is grown to at most `max_size` bytes; if the data does not fit in that size, | 
|---|
| 148 | /// the error [struct][DecompressError] will contain the status [`TINFLStatus::HasMoreOutput`] and the data that was decompressed on failure. | 
|---|
| 149 | /// | 
|---|
| 150 | /// As this function tries to decompress everything in one go, it's not ideal for general use outside of tests or where the output size is expected to be small. | 
|---|
| 151 | /// It is suggested to use streaming decompression via the [flate2](https://github.com/alexcrichton/flate2-rs) library instead. | 
|---|
| 152 | /// | 
|---|
| 153 | /// Returns a [`Result`] containing the [`Vec`] of decompressed data on success, and a [struct][DecompressError] on failure. | 
|---|
| 154 | #[ inline] | 
|---|
| 155 | #[ cfg(feature = "with-alloc")] | 
|---|
| 156 | pub fn decompress_to_vec_with_limit( | 
|---|
| 157 | input: &[u8], | 
|---|
| 158 | max_size: usize, | 
|---|
| 159 | ) -> Result<Vec<u8>, DecompressError> { | 
|---|
| 160 | decompress_to_vec_inner(input, flags:0, max_output_size:max_size) | 
|---|
| 161 | } | 
|---|
| 162 |  | 
|---|
| 163 | /// Decompress the deflate-encoded data (with a zlib wrapper) in `input` to a vector. | 
|---|
| 164 | /// The vector is grown to at most `max_size` bytes; if the data does not fit in that size, | 
|---|
| 165 | /// the error [struct][DecompressError] will contain the status [`TINFLStatus::HasMoreOutput`] and the data that was decompressed on failure. | 
|---|
| 166 | /// | 
|---|
| 167 | /// As this function tries to decompress everything in one go, it's not ideal for general use outside of tests or where the output size is expected to be small. | 
|---|
| 168 | /// It is suggested to use streaming decompression via the [flate2](https://github.com/alexcrichton/flate2-rs) library instead. | 
|---|
| 169 | /// | 
|---|
| 170 | /// Returns a [`Result`] containing the [`Vec`] of decompressed data on success, and a [struct][DecompressError] on failure. | 
|---|
| 171 | #[ inline] | 
|---|
| 172 | #[ cfg(feature = "with-alloc")] | 
|---|
| 173 | pub fn decompress_to_vec_zlib_with_limit( | 
|---|
| 174 | input: &[u8], | 
|---|
| 175 | max_size: usize, | 
|---|
| 176 | ) -> Result<Vec<u8>, DecompressError> { | 
|---|
| 177 | decompress_to_vec_inner(input, flags:inflate_flags::TINFL_FLAG_PARSE_ZLIB_HEADER, max_output_size:max_size) | 
|---|
| 178 | } | 
|---|
| 179 |  | 
|---|
| 180 | /// Backend of various to-[`Vec`] decompressions. | 
|---|
| 181 | /// | 
|---|
| 182 | /// Returns [`Vec`] of decompressed data on success and the [error struct][DecompressError] with details on failure. | 
|---|
| 183 | #[ cfg(feature = "with-alloc")] | 
|---|
| 184 | fn decompress_to_vec_inner( | 
|---|
| 185 | mut input: &[u8], | 
|---|
| 186 | flags: u32, | 
|---|
| 187 | max_output_size: usize, | 
|---|
| 188 | ) -> Result<Vec<u8>, DecompressError> { | 
|---|
| 189 | let flags = flags | inflate_flags::TINFL_FLAG_USING_NON_WRAPPING_OUTPUT_BUF; | 
|---|
| 190 | let mut ret: Vec<u8> = vec![0; input.len().saturating_mul(2).min(max_output_size)]; | 
|---|
| 191 |  | 
|---|
| 192 | let mut decomp = Box::<DecompressorOxide>::default(); | 
|---|
| 193 |  | 
|---|
| 194 | let mut out_pos = 0; | 
|---|
| 195 | loop { | 
|---|
| 196 | // Wrap the whole output slice so we know we have enough of the | 
|---|
| 197 | // decompressed data for matches. | 
|---|
| 198 | let (status, in_consumed, out_consumed) = | 
|---|
| 199 | decompress(&mut decomp, input, &mut ret, out_pos, flags); | 
|---|
| 200 | out_pos += out_consumed; | 
|---|
| 201 |  | 
|---|
| 202 | match status { | 
|---|
| 203 | TINFLStatus::Done => { | 
|---|
| 204 | ret.truncate(out_pos); | 
|---|
| 205 | return Ok(ret); | 
|---|
| 206 | } | 
|---|
| 207 |  | 
|---|
| 208 | TINFLStatus::HasMoreOutput => { | 
|---|
| 209 | // in_consumed is not expected to be out of bounds, | 
|---|
| 210 | // but the check eliminates a panicking code path | 
|---|
| 211 | if in_consumed > input.len() { | 
|---|
| 212 | return decompress_error(TINFLStatus::HasMoreOutput, ret); | 
|---|
| 213 | } | 
|---|
| 214 | input = &input[in_consumed..]; | 
|---|
| 215 |  | 
|---|
| 216 | // if the buffer has already reached the size limit, return an error | 
|---|
| 217 | if ret.len() >= max_output_size { | 
|---|
| 218 | return decompress_error(TINFLStatus::HasMoreOutput, ret); | 
|---|
| 219 | } | 
|---|
| 220 | // calculate the new length, capped at `max_output_size` | 
|---|
| 221 | let new_len = ret.len().saturating_mul(2).min(max_output_size); | 
|---|
| 222 | ret.resize(new_len, 0); | 
|---|
| 223 | } | 
|---|
| 224 |  | 
|---|
| 225 | _ => return decompress_error(status, ret), | 
|---|
| 226 | } | 
|---|
| 227 | } | 
|---|
| 228 | } | 
|---|
| 229 |  | 
|---|
| 230 | /// Decompress one or more source slices from an iterator into the output slice. | 
|---|
| 231 | /// | 
|---|
| 232 | /// * On success, returns the number of bytes that were written. | 
|---|
| 233 | /// * On failure, returns the failure status code. | 
|---|
| 234 | /// | 
|---|
| 235 | /// This will fail if the output buffer is not large enough, but in that case | 
|---|
| 236 | /// the output buffer will still contain the partial decompression. | 
|---|
| 237 | /// | 
|---|
| 238 | /// * `out` the output buffer. | 
|---|
| 239 | /// * `it` the iterator of input slices. | 
|---|
| 240 | /// * `zlib_header` if the first slice out of the iterator is expected to have a | 
|---|
| 241 | ///   Zlib header. Otherwise the slices are assumed to be the deflate data only. | 
|---|
| 242 | /// * `ignore_adler32` if the adler32 checksum should be calculated or not. | 
|---|
| 243 | pub fn decompress_slice_iter_to_slice<'out, 'inp>( | 
|---|
| 244 | out: &'out mut [u8], | 
|---|
| 245 | it: impl Iterator<Item = &'inp [u8]>, | 
|---|
| 246 | zlib_header: bool, | 
|---|
| 247 | ignore_adler32: bool, | 
|---|
| 248 | ) -> Result<usize, TINFLStatus> { | 
|---|
| 249 | use self::core::inflate_flags::*; | 
|---|
| 250 |  | 
|---|
| 251 | let mut it = it.peekable(); | 
|---|
| 252 | let r = &mut DecompressorOxide::new(); | 
|---|
| 253 | let mut out_pos = 0; | 
|---|
| 254 | while let Some(in_buf) = it.next() { | 
|---|
| 255 | let has_more = it.peek().is_some(); | 
|---|
| 256 | let flags = { | 
|---|
| 257 | let mut f = TINFL_FLAG_USING_NON_WRAPPING_OUTPUT_BUF; | 
|---|
| 258 | if zlib_header { | 
|---|
| 259 | f |= TINFL_FLAG_PARSE_ZLIB_HEADER; | 
|---|
| 260 | } | 
|---|
| 261 | if ignore_adler32 { | 
|---|
| 262 | f |= TINFL_FLAG_IGNORE_ADLER32; | 
|---|
| 263 | } | 
|---|
| 264 | if has_more { | 
|---|
| 265 | f |= TINFL_FLAG_HAS_MORE_INPUT; | 
|---|
| 266 | } | 
|---|
| 267 | f | 
|---|
| 268 | }; | 
|---|
| 269 | let (status, _input_read, bytes_written) = decompress(r, in_buf, out, out_pos, flags); | 
|---|
| 270 | out_pos += bytes_written; | 
|---|
| 271 | match status { | 
|---|
| 272 | TINFLStatus::NeedsMoreInput => continue, | 
|---|
| 273 | TINFLStatus::Done => return Ok(out_pos), | 
|---|
| 274 | e => return Err(e), | 
|---|
| 275 | } | 
|---|
| 276 | } | 
|---|
| 277 | // If we ran out of source slices without getting a `Done` from the | 
|---|
| 278 | // decompression we can call it a failure. | 
|---|
| 279 | Err(TINFLStatus::FailedCannotMakeProgress) | 
|---|
| 280 | } | 
|---|
| 281 |  | 
|---|
| 282 | #[ cfg(all(test, feature = "with-alloc"))] | 
|---|
| 283 | mod test { | 
|---|
| 284 | use super::{ | 
|---|
| 285 | decompress_slice_iter_to_slice, decompress_to_vec_zlib, decompress_to_vec_zlib_with_limit, | 
|---|
| 286 | DecompressError, TINFLStatus, | 
|---|
| 287 | }; | 
|---|
| 288 | const ENCODED: [u8; 20] = [ | 
|---|
| 289 | 120, 156, 243, 72, 205, 201, 201, 215, 81, 168, 202, 201, 76, 82, 4, 0, 27, 101, 4, 19, | 
|---|
| 290 | ]; | 
|---|
| 291 |  | 
|---|
| 292 | #[ test] | 
|---|
| 293 | fn decompress_vec() { | 
|---|
| 294 | let res = decompress_to_vec_zlib(&ENCODED[..]).unwrap(); | 
|---|
| 295 | assert_eq!(res.as_slice(), & b"Hello, zlib!"[..]); | 
|---|
| 296 | } | 
|---|
| 297 |  | 
|---|
| 298 | #[ test] | 
|---|
| 299 | fn decompress_vec_with_high_limit() { | 
|---|
| 300 | let res = decompress_to_vec_zlib_with_limit(&ENCODED[..], 100_000).unwrap(); | 
|---|
| 301 | assert_eq!(res.as_slice(), & b"Hello, zlib!"[..]); | 
|---|
| 302 | } | 
|---|
| 303 |  | 
|---|
| 304 | #[ test] | 
|---|
| 305 | fn fail_to_decompress_with_limit() { | 
|---|
| 306 | let res = decompress_to_vec_zlib_with_limit(&ENCODED[..], 8); | 
|---|
| 307 | match res { | 
|---|
| 308 | Err(DecompressError { | 
|---|
| 309 | status: TINFLStatus::HasMoreOutput, | 
|---|
| 310 | .. | 
|---|
| 311 | }) => (), // expected result | 
|---|
| 312 | _ => panic!( "Decompression output size limit was not enforced"), | 
|---|
| 313 | } | 
|---|
| 314 | } | 
|---|
| 315 |  | 
|---|
| 316 | #[ test] | 
|---|
| 317 | fn test_decompress_slice_iter_to_slice() { | 
|---|
| 318 | // one slice | 
|---|
| 319 | let mut out = [0_u8; 12_usize]; | 
|---|
| 320 | let r = | 
|---|
| 321 | decompress_slice_iter_to_slice(&mut out, Some(&ENCODED[..]).into_iter(), true, false); | 
|---|
| 322 | assert_eq!(r, Ok(12)); | 
|---|
| 323 | assert_eq!(&out[..12], & b"Hello, zlib!"[..]); | 
|---|
| 324 |  | 
|---|
| 325 | // some chunks at a time | 
|---|
| 326 | for chunk_size in 1..13 { | 
|---|
| 327 | // Note: because of https://github.com/Frommi/miniz_oxide/issues/110 our | 
|---|
| 328 | // out buffer needs to have +1 byte available when the chunk size cuts | 
|---|
| 329 | // the adler32 data off from the last actual data. | 
|---|
| 330 | let mut out = [0_u8; 12_usize + 1]; | 
|---|
| 331 | let r = | 
|---|
| 332 | decompress_slice_iter_to_slice(&mut out, ENCODED.chunks(chunk_size), true, false); | 
|---|
| 333 | assert_eq!(r, Ok(12)); | 
|---|
| 334 | assert_eq!(&out[..12], & b"Hello, zlib!"[..]); | 
|---|
| 335 | } | 
|---|
| 336 |  | 
|---|
| 337 | // output buffer too small | 
|---|
| 338 | let mut out = [0_u8; 3_usize]; | 
|---|
| 339 | let r = decompress_slice_iter_to_slice(&mut out, ENCODED.chunks(7), true, false); | 
|---|
| 340 | assert!(r.is_err()); | 
|---|
| 341 | } | 
|---|
| 342 | } | 
|---|
| 343 |  | 
|---|