#length #decoding #run #encoded #kind #data #fastest

deprecated rle-decode-fast

Deprecated: this is available in stable Rust since 1.53 as Vec::extend_from_within(). Previsouly, the fastest way to implement any kind of decoding for Run Length Encoded data in Rust

4 stable releases

Uses old Rust 2015

1.0.3 Jan 30, 2022
1.0.1 Jun 30, 2019

#9 in #fastest

Download history 123816/week @ 2024-07-28 125700/week @ 2024-08-04 141246/week @ 2024-08-11 130529/week @ 2024-08-18 150323/week @ 2024-08-25 146707/week @ 2024-09-01 132179/week @ 2024-09-08 132135/week @ 2024-09-15 148863/week @ 2024-09-22 146810/week @ 2024-09-29 137887/week @ 2024-10-06 157369/week @ 2024-10-13 160660/week @ 2024-10-20 143327/week @ 2024-10-27 139760/week @ 2024-11-03 148034/week @ 2024-11-10

598,217 downloads per month
Used in 885 crates (2 directly)

MIT/Apache

63KB
92 lines

rle-decode-fast

The same functionallity is available in stable Rust since 1.53, use that instead

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-fast = "1.0"

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_fast::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 fill length lookbehind=333 and for a bigger fill lengths lookbehind=2

License

Licensed under either of

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.

Dependencies

~0–8MB
~55K SLoC