1.0.0-alpha2 |
|
---|
#14 in #fastest
41KB
99 lines
rle-decode-helper
THE fastest way to implement any kind of decoding for Run Length Encoded data in Rust.
Writing a fast decoder that is also safe can be quite challenging, so this crate is here to save you the hassle of maintaining and testing your own implementation.
Usage
Of course, you need to depend on this crate:
rle-decode-helper = "1.0.0-alpha"
There is only a single function to use, rle_decode<T>(&mut Vec<T>, lookbehind_size: usize, fill_length: usize)
.
It takes :
- a vector to modify,
- the number of items to copy from said vector (basically
vector[(vector.len() - lookbehind)..])
- the number of items to append.
Afterwards the vector will contain an extra
fill_length
items.
use rle_decode_helper::rle_decode;
let mut decode_buffer = vec![0, 0, 1, 1, 0, 2, 3];
let lookbehind_length = 4;
let output_length = 10;
rle_decode(&mut decode_buffer, lookbehind_length, output_length);
assert_eq!(decode_buffer, [0, 0, 1, 1, 0, 2, 3, 1, 0, 2, 3, 1, 0, 2, 3, 1, 0]);
Panics
There are cases where the decode functions panics
- The lookbehind length is 0
- The lookbehind length is bigger than the Vec's length
- The output length + Vec's length would overflow
Background
The idea for this crate first originated from this pre-RFC. It brought to attention a weak-point in the standard library, which lead some crates writing their own unsafe by-pass.
During the exploration happening in the pre-RFC, some experimentation was conducted. For examples see here and here.
Some Stats
There is a benchmark comparing a naive (repeated push) implementation, a vulnerable implementation that might lead to uninitialized memory and this crate.
The results for a very small lookbehind length of 2 and for a bigger lookbehind length of 333
License
Licensed under either of
- Apache License, Version 2.0 (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
- MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)
at your option.
Contribution
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.