11 releases (1 stable)
1.0.0 | Oct 30, 2024 |
---|---|
0.7.0 | Sep 19, 2022 |
0.6.0 | Jan 25, 2022 |
0.5.0 | Feb 3, 2021 |
0.2.0 | Dec 13, 2019 |
#347 in Text processing
23,432 downloads per month
Used in 10 crates
(4 directly)
71KB
1.5K
SLoC
unicode-joining-type
Fast lookup of the Unicode Joining Type and Joining Group properties for char
in Rust using Unicode 16.0 data. This crate is no-std compatible.
Usage
use unicode_joining_type::{get_joining_type, JoiningType};
use unicode_joining_type::{get_joining_group, JoiningGroup};
fn main() {
assert_eq!(get_joining_type('A'), JoiningType::NonJoining);
assert_eq!(get_joining_group('ھ'), JoiningGroup::KnottedHeh);
}
Performance & Implementation Notes
ucd-generate is used to generate joining_type_tables.rs
and
joining_group_tables.rs
. A build script (build.rs
) compiles each of these
into a two level look up tables. The look up time is constant as it is just
indexing into two arrays.
The two level approach maps a code point to a block, then to a position within a block. The allows the second level of block to be deduplicated, saving space. The code is parameterised over the block size, which must be a power of 2. The value in the build script is optimal for the data set.
This approach trades off some space for faster lookups. The joining type tables take up about 26KiB, the joining group tables take up about 6.75KiB. Benchmarks showed this approach to be ~5–10× faster than the typical binary search approach.
There is still room for further size reduction. For example, by eliminating repeated block mappings at the end of the first level block array.