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 | #[cfg (not(feature = "rustc-dep-of-std" ))] |
11 | pub mod stream; |
12 | #[cfg (not(feature = "rustc-dep-of-std" ))] |
13 | use self::core::*; |
14 | |
15 | const TINFL_STATUS_FAILED_CANNOT_MAKE_PROGRESS: i32 = -4; |
16 | const TINFL_STATUS_BAD_PARAM: i32 = -3; |
17 | const TINFL_STATUS_ADLER32_MISMATCH: i32 = -2; |
18 | const TINFL_STATUS_FAILED: i32 = -1; |
19 | const TINFL_STATUS_DONE: i32 = 0; |
20 | const TINFL_STATUS_NEEDS_MORE_INPUT: i32 = 1; |
21 | const TINFL_STATUS_HAS_MORE_OUTPUT: i32 = 2; |
22 | #[cfg (feature = "block-boundary" )] |
23 | const TINFL_STATUS_BLOCK_BOUNDARY: i32 = 3; |
24 | |
25 | /// Return status codes. |
26 | #[repr (i8)] |
27 | #[cfg_attr (not(feature = "rustc-dep-of-std" ), derive(Hash, Debug))] |
28 | #[derive (Copy, Clone, PartialEq, Eq)] |
29 | pub enum TINFLStatus { |
30 | /// More input data was expected, but the caller indicated that there was no more data, so the |
31 | /// input stream is likely truncated. |
32 | /// |
33 | /// This can't happen if you have provided the |
34 | /// [`TINFL_FLAG_HAS_MORE_INPUT`][core::inflate_flags::TINFL_FLAG_HAS_MORE_INPUT] flag to the |
35 | /// decompression. By setting that flag, you indicate more input exists but is not provided, |
36 | /// and so reaching the end of the input data without finding the end of the compressed stream |
37 | /// would instead return a [`NeedsMoreInput`][Self::NeedsMoreInput] status. |
38 | FailedCannotMakeProgress = TINFL_STATUS_FAILED_CANNOT_MAKE_PROGRESS as i8, |
39 | |
40 | /// The output buffer is an invalid size; consider the `flags` parameter. |
41 | BadParam = TINFL_STATUS_BAD_PARAM as i8, |
42 | |
43 | /// The decompression went fine, but the adler32 checksum did not match the one |
44 | /// provided in the header. |
45 | Adler32Mismatch = TINFL_STATUS_ADLER32_MISMATCH as i8, |
46 | |
47 | /// Failed to decompress due to invalid data. |
48 | Failed = TINFL_STATUS_FAILED as i8, |
49 | |
50 | /// Finished decompression without issues. |
51 | /// |
52 | /// This indicates the end of the compressed stream has been reached. |
53 | Done = TINFL_STATUS_DONE as i8, |
54 | |
55 | /// The decompressor needs more input data to continue decompressing. |
56 | /// |
57 | /// This occurs when there's no more consumable input, but the end of the stream hasn't been |
58 | /// reached, and you have supplied the |
59 | /// [`TINFL_FLAG_HAS_MORE_INPUT`][core::inflate_flags::TINFL_FLAG_HAS_MORE_INPUT] flag to the |
60 | /// decompressor. Had you not supplied that flag (which would mean you were asserting that you |
61 | /// believed all the data was available) you would have gotten a |
62 | /// [`FailedCannotMakeProcess`][Self::FailedCannotMakeProgress] instead. |
63 | NeedsMoreInput = TINFL_STATUS_NEEDS_MORE_INPUT as i8, |
64 | |
65 | /// There is still pending data that didn't fit in the output buffer. |
66 | HasMoreOutput = TINFL_STATUS_HAS_MORE_OUTPUT as i8, |
67 | |
68 | /// Reached the end of a deflate block, and the start of the next block. |
69 | /// |
70 | /// At this point, you can suspend decompression and later resume with a new `DecompressorOxide`. |
71 | /// The only state that must be preserved is [`DecompressorOxide::block_boundary_state()`], |
72 | /// plus the last 32KiB of the output buffer (or less if you know the stream was compressed with |
73 | /// a smaller window size). |
74 | /// |
75 | /// This is only returned if you use the |
76 | /// [`TINFL_FLAG_STOP_ON_BLOCK_BOUNDARY`][core::inflate_flags::TINFL_FLAG_STOP_ON_BLOCK_BOUNDARY] flag. |
77 | #[cfg (feature = "block-boundary" )] |
78 | BlockBoundary = TINFL_STATUS_BLOCK_BOUNDARY as i8, |
79 | } |
80 | |
81 | impl TINFLStatus { |
82 | pub fn from_i32(value: i32) -> Option<TINFLStatus> { |
83 | use self::TINFLStatus::*; |
84 | match value { |
85 | TINFL_STATUS_FAILED_CANNOT_MAKE_PROGRESS => Some(FailedCannotMakeProgress), |
86 | TINFL_STATUS_BAD_PARAM => Some(BadParam), |
87 | TINFL_STATUS_ADLER32_MISMATCH => Some(Adler32Mismatch), |
88 | TINFL_STATUS_FAILED => Some(Failed), |
89 | TINFL_STATUS_DONE => Some(Done), |
90 | TINFL_STATUS_NEEDS_MORE_INPUT => Some(NeedsMoreInput), |
91 | TINFL_STATUS_HAS_MORE_OUTPUT => Some(HasMoreOutput), |
92 | #[cfg (feature = "block-boundary" )] |
93 | TINFL_STATUS_BLOCK_BOUNDARY => Some(BlockBoundary), |
94 | _ => None, |
95 | } |
96 | } |
97 | } |
98 | |
99 | /// Struct return when decompress_to_vec functions fail. |
100 | #[cfg (feature = "with-alloc" )] |
101 | #[derive (Debug)] |
102 | pub struct DecompressError { |
103 | /// Decompressor status on failure. See [TINFLStatus] for details. |
104 | pub status: TINFLStatus, |
105 | /// The currently decompressed data if any. |
106 | pub output: Vec<u8>, |
107 | } |
108 | |
109 | #[cfg (feature = "with-alloc" )] |
110 | impl alloc::fmt::Display for DecompressError { |
111 | #[cold ] |
112 | fn fmt(&self, f: &mut ::core::fmt::Formatter<'_>) -> ::core::fmt::Result { |
113 | f.write_str(data:match self.status { |
114 | TINFLStatus::FailedCannotMakeProgress => "Truncated input stream" , |
115 | TINFLStatus::BadParam => "Invalid output buffer size" , |
116 | TINFLStatus::Adler32Mismatch => "Adler32 checksum mismatch" , |
117 | TINFLStatus::Failed => "Invalid input data" , |
118 | TINFLStatus::Done => "" , // Unreachable |
119 | TINFLStatus::NeedsMoreInput => "Truncated input stream" , |
120 | TINFLStatus::HasMoreOutput => "Output size exceeded the specified limit" , |
121 | #[cfg (feature = "block-boundary" )] |
122 | TINFLStatus::BlockBoundary => "Reached end of a deflate block" , |
123 | }) |
124 | } |
125 | } |
126 | |
127 | /// Implement Error trait only if std feature is requested as it requires std. |
128 | #[cfg (all(feature = "std" , feature = "with-alloc" ))] |
129 | impl Error for DecompressError {} |
130 | |
131 | #[cfg (feature = "with-alloc" )] |
132 | fn decompress_error(status: TINFLStatus, output: Vec<u8>) -> Result<Vec<u8>, DecompressError> { |
133 | Err(DecompressError { status, output }) |
134 | } |
135 | |
136 | /// Decompress the deflate-encoded data in `input` to a vector. |
137 | /// |
138 | /// NOTE: This function will not bound the output, so if the output is large enough it can result in an out of memory error. |
139 | /// It is therefore suggested to not use this for anything other than test programs, use the functions with a specified limit, or |
140 | /// ideally streaming decompression via the [flate2](https://github.com/alexcrichton/flate2-rs) library instead. |
141 | /// |
142 | /// 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. |
143 | #[inline ] |
144 | #[cfg (feature = "with-alloc" )] |
145 | pub fn decompress_to_vec(input: &[u8]) -> Result<Vec<u8>, DecompressError> { |
146 | decompress_to_vec_inner(input, flags:0, max_output_size:usize::MAX) |
147 | } |
148 | |
149 | /// Decompress the deflate-encoded data (with a zlib wrapper) in `input` to a vector. |
150 | /// |
151 | /// NOTE: This function will not bound the output, so if the output is large enough it can result in an out of memory error. |
152 | /// It is therefore suggested to not use this for anything other than test programs, use the functions with a specified limit, or |
153 | /// ideally streaming decompression via the [flate2](https://github.com/alexcrichton/flate2-rs) library instead. |
154 | /// |
155 | /// 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. |
156 | #[inline ] |
157 | #[cfg (feature = "with-alloc" )] |
158 | pub fn decompress_to_vec_zlib(input: &[u8]) -> Result<Vec<u8>, DecompressError> { |
159 | decompress_to_vec_inner( |
160 | input, |
161 | flags:inflate_flags::TINFL_FLAG_PARSE_ZLIB_HEADER, |
162 | max_output_size:usize::MAX, |
163 | ) |
164 | } |
165 | |
166 | /// Decompress the deflate-encoded data in `input` to a vector. |
167 | /// |
168 | /// The vector is grown to at most `max_size` bytes; if the data does not fit in that size, |
169 | /// the error [struct][DecompressError] will contain the status [`TINFLStatus::HasMoreOutput`] and the data that was decompressed on failure. |
170 | /// |
171 | /// 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. |
172 | /// It is suggested to use streaming decompression via the [flate2](https://github.com/alexcrichton/flate2-rs) library instead. |
173 | /// |
174 | /// Returns a [`Result`] containing the [`Vec`] of decompressed data on success, and a [struct][DecompressError] on failure. |
175 | #[inline ] |
176 | #[cfg (feature = "with-alloc" )] |
177 | pub fn decompress_to_vec_with_limit( |
178 | input: &[u8], |
179 | max_size: usize, |
180 | ) -> Result<Vec<u8>, DecompressError> { |
181 | decompress_to_vec_inner(input, flags:0, max_output_size:max_size) |
182 | } |
183 | |
184 | /// Decompress the deflate-encoded data (with a zlib wrapper) in `input` to a vector. |
185 | /// The vector is grown to at most `max_size` bytes; if the data does not fit in that size, |
186 | /// the error [struct][DecompressError] will contain the status [`TINFLStatus::HasMoreOutput`] and the data that was decompressed on failure. |
187 | /// |
188 | /// 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. |
189 | /// It is suggested to use streaming decompression via the [flate2](https://github.com/alexcrichton/flate2-rs) library instead. |
190 | /// |
191 | /// Returns a [`Result`] containing the [`Vec`] of decompressed data on success, and a [struct][DecompressError] on failure. |
192 | #[inline ] |
193 | #[cfg (feature = "with-alloc" )] |
194 | pub fn decompress_to_vec_zlib_with_limit( |
195 | input: &[u8], |
196 | max_size: usize, |
197 | ) -> Result<Vec<u8>, DecompressError> { |
198 | decompress_to_vec_inner(input, flags:inflate_flags::TINFL_FLAG_PARSE_ZLIB_HEADER, max_output_size:max_size) |
199 | } |
200 | |
201 | /// Backend of various to-[`Vec`] decompressions. |
202 | /// |
203 | /// Returns [`Vec`] of decompressed data on success and the [error struct][DecompressError] with details on failure. |
204 | #[cfg (feature = "with-alloc" )] |
205 | fn decompress_to_vec_inner( |
206 | mut input: &[u8], |
207 | flags: u32, |
208 | max_output_size: usize, |
209 | ) -> Result<Vec<u8>, DecompressError> { |
210 | let flags = flags | inflate_flags::TINFL_FLAG_USING_NON_WRAPPING_OUTPUT_BUF; |
211 | let mut ret: Vec<u8> = vec![0; input.len().saturating_mul(2).min(max_output_size)]; |
212 | |
213 | let mut decomp = Box::<DecompressorOxide>::default(); |
214 | |
215 | let mut out_pos = 0; |
216 | loop { |
217 | // Wrap the whole output slice so we know we have enough of the |
218 | // decompressed data for matches. |
219 | let (status, in_consumed, out_consumed) = |
220 | decompress(&mut decomp, input, &mut ret, out_pos, flags); |
221 | out_pos += out_consumed; |
222 | |
223 | match status { |
224 | TINFLStatus::Done => { |
225 | ret.truncate(out_pos); |
226 | return Ok(ret); |
227 | } |
228 | |
229 | TINFLStatus::HasMoreOutput => { |
230 | // in_consumed is not expected to be out of bounds, |
231 | // but the check eliminates a panicking code path |
232 | if in_consumed > input.len() { |
233 | return decompress_error(TINFLStatus::HasMoreOutput, ret); |
234 | } |
235 | input = &input[in_consumed..]; |
236 | |
237 | // if the buffer has already reached the size limit, return an error |
238 | if ret.len() >= max_output_size { |
239 | return decompress_error(TINFLStatus::HasMoreOutput, ret); |
240 | } |
241 | // calculate the new length, capped at `max_output_size` |
242 | let new_len = ret.len().saturating_mul(2).min(max_output_size); |
243 | ret.resize(new_len, 0); |
244 | } |
245 | |
246 | _ => return decompress_error(status, ret), |
247 | } |
248 | } |
249 | } |
250 | |
251 | /// Decompress one or more source slices from an iterator into the output slice. |
252 | /// |
253 | /// * On success, returns the number of bytes that were written. |
254 | /// * On failure, returns the failure status code. |
255 | /// |
256 | /// This will fail if the output buffer is not large enough, but in that case |
257 | /// the output buffer will still contain the partial decompression. |
258 | /// |
259 | /// * `out` the output buffer. |
260 | /// * `it` the iterator of input slices. |
261 | /// * `zlib_header` if the first slice out of the iterator is expected to have a |
262 | /// Zlib header. Otherwise the slices are assumed to be the deflate data only. |
263 | /// * `ignore_adler32` if the adler32 checksum should be calculated or not. |
264 | #[cfg (not(feature = "rustc-dep-of-std" ))] |
265 | pub fn decompress_slice_iter_to_slice<'out, 'inp>( |
266 | out: &'out mut [u8], |
267 | it: impl Iterator<Item = &'inp [u8]>, |
268 | zlib_header: bool, |
269 | ignore_adler32: bool, |
270 | ) -> Result<usize, TINFLStatus> { |
271 | use self::core::inflate_flags::*; |
272 | |
273 | let mut it = it.peekable(); |
274 | let r = &mut DecompressorOxide::new(); |
275 | let mut out_pos = 0; |
276 | while let Some(in_buf) = it.next() { |
277 | let has_more = it.peek().is_some(); |
278 | let flags = { |
279 | let mut f = TINFL_FLAG_USING_NON_WRAPPING_OUTPUT_BUF; |
280 | if zlib_header { |
281 | f |= TINFL_FLAG_PARSE_ZLIB_HEADER; |
282 | } |
283 | if ignore_adler32 { |
284 | f |= TINFL_FLAG_IGNORE_ADLER32; |
285 | } |
286 | if has_more { |
287 | f |= TINFL_FLAG_HAS_MORE_INPUT; |
288 | } |
289 | f |
290 | }; |
291 | let (status, _input_read, bytes_written) = decompress(r, in_buf, out, out_pos, flags); |
292 | out_pos += bytes_written; |
293 | match status { |
294 | TINFLStatus::NeedsMoreInput => continue, |
295 | TINFLStatus::Done => return Ok(out_pos), |
296 | e => return Err(e), |
297 | } |
298 | } |
299 | // If we ran out of source slices without getting a `Done` from the |
300 | // decompression we can call it a failure. |
301 | Err(TINFLStatus::FailedCannotMakeProgress) |
302 | } |
303 | |
304 | #[cfg (all(test, feature = "with-alloc" ))] |
305 | mod test { |
306 | use super::{ |
307 | decompress_slice_iter_to_slice, decompress_to_vec_zlib, decompress_to_vec_zlib_with_limit, |
308 | DecompressError, TINFLStatus, |
309 | }; |
310 | const ENCODED: [u8; 20] = [ |
311 | 120, 156, 243, 72, 205, 201, 201, 215, 81, 168, 202, 201, 76, 82, 4, 0, 27, 101, 4, 19, |
312 | ]; |
313 | |
314 | #[test ] |
315 | fn decompress_vec() { |
316 | let res = decompress_to_vec_zlib(&ENCODED[..]).unwrap(); |
317 | assert_eq!(res.as_slice(), &b"Hello, zlib!" [..]); |
318 | } |
319 | |
320 | #[test ] |
321 | fn decompress_vec_with_high_limit() { |
322 | let res = decompress_to_vec_zlib_with_limit(&ENCODED[..], 100_000).unwrap(); |
323 | assert_eq!(res.as_slice(), &b"Hello, zlib!" [..]); |
324 | } |
325 | |
326 | #[test ] |
327 | fn fail_to_decompress_with_limit() { |
328 | let res = decompress_to_vec_zlib_with_limit(&ENCODED[..], 8); |
329 | match res { |
330 | Err(DecompressError { |
331 | status: TINFLStatus::HasMoreOutput, |
332 | .. |
333 | }) => (), // expected result |
334 | _ => panic!("Decompression output size limit was not enforced" ), |
335 | } |
336 | } |
337 | |
338 | #[test ] |
339 | fn test_decompress_slice_iter_to_slice() { |
340 | // one slice |
341 | let mut out = [0_u8; 12_usize]; |
342 | let r = |
343 | decompress_slice_iter_to_slice(&mut out, Some(&ENCODED[..]).into_iter(), true, false); |
344 | assert_eq!(r, Ok(12)); |
345 | assert_eq!(&out[..12], &b"Hello, zlib!" [..]); |
346 | |
347 | // some chunks at a time |
348 | for chunk_size in 1..13 { |
349 | // Note: because of https://github.com/Frommi/miniz_oxide/issues/110 our |
350 | // out buffer needs to have +1 byte available when the chunk size cuts |
351 | // the adler32 data off from the last actual data. |
352 | let mut out = [0_u8; 12_usize + 1]; |
353 | let r = |
354 | decompress_slice_iter_to_slice(&mut out, ENCODED.chunks(chunk_size), true, false); |
355 | assert_eq!(r, Ok(12)); |
356 | assert_eq!(&out[..12], &b"Hello, zlib!" [..]); |
357 | } |
358 | |
359 | // output buffer too small |
360 | let mut out = [0_u8; 3_usize]; |
361 | let r = decompress_slice_iter_to_slice(&mut out, ENCODED.chunks(7), true, false); |
362 | assert!(r.is_err()); |
363 | } |
364 | } |
365 | |