Skip to content

Commit

Permalink
chore: add example and correct README
Browse files Browse the repository at this point in the history
Signed-off-by: feathercyc <[email protected]>
  • Loading branch information
GG2002 committed Aug 20, 2024
1 parent 2caee42 commit a5d944e
Show file tree
Hide file tree
Showing 3 changed files with 106 additions and 10 deletions.
21 changes: 11 additions & 10 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@ The implementation of the interval tree in interval_map references "Introduction
To safely and efficiently handle insertion and deletion operations in Rust, `interval_map` innovatively **uses arrays to simulate pointers** for managing the parent-child references in the red-black tree. This approach also ensures that interval_map has the `Send` and `Unpin` traits, allowing it to be safely transferred between threads and to maintain a fixed memory location during asynchronous operations.

`interval_map` implements an `IntervalMap` struct:
- It accepts `Interval<T>` as the key, where `T` can be any type that implements `Ord+Clone` trait. Therefore, intervals such as $[1, 2)$ and $["aaa", "bbb")$ are allowed
- It accepts `Interval<T>` as the key, where `T` can be any type that implements `Ord` trait. Therefore, intervals such as $[1, 2)$ and $["aaa", "bbb")$ are allowed
- The value can be of any type

`interval_map` supports `insert`, `delete`, and `iter` fns. Traversal is performed in the order of `Interval<T>` . For instance, with intervals of type `Interval<u32>`:
Expand All @@ -22,15 +22,16 @@ Currently, `interval_map` only supports half-open intervals, i.e., $[...,...)$.

The benchmark was conducted on a platform with `AMD R7 7840H + DDR5 5600MHz`. The result are as follows:
1. Only insert
| insert | 100 | 1000 | 10, 000 | 100, 000 |
| --------------- | --------- | --------- | --------- | --------- |
| Time per insert | 5.4168 µs | 80.518 µs | 2.2823 ms | 36.528 ms |
| insert | 100 | 1000 | 10, 000 | 100, 000 |
| ---------- | --------- | --------- | --------- | --------- |
| Total time | 5.4168 µs | 80.518 µs | 2.2823 ms | 36.528 ms |
2. Insert N and remove N
| insert_and_remove | 100 | 1000 | 10, 000 | 100, 000 |
| ------------------ | --------- | --------- | --------- | --------- |
| Time per operation | 10.333 µs | 223.43 µs | 4.9358 ms | 81.634 ms |
| insert_and_remove | 100 | 1000 | 10, 000 | 100, 000 |
| ----------------- | --------- | --------- | --------- | --------- |
| Total time | 10.333 µs | 223.43 µs | 4.9358 ms | 81.634 ms |

## TODO
- [] Support for $(...,...)$, $[...,...]$ and $(...,...]$ interval types.
- [] Add more tests like [etcd](https://github.com/etcd-io/etcd/blob/main/pkg/adt/interval_tree_test.go)
- [] Add Point type for Interval
- [ ] ~~Support for $(...,...)$, $[...,...]$ and $(...,...]$ interval types.~~ There's no way to support these interval type without performance loss now.
- [ ] ~~Add Point type for Interval~~ To support Point type, it should also support $[...,...]$, so it couldn't be supported now, either. But you could write code like [examples/new_point](examples/new_point.rs).
- [x] Add more tests like [etcd](https://github.com/etcd-io/etcd/blob/main/pkg/adt/interval_tree_test.go).
- [x] Refine iter mod.
27 changes: 27 additions & 0 deletions examples/new_point.rs
Original file line number Diff line number Diff line change
@@ -0,0 +1,27 @@
use interval_map::{Interval, IntervalMap};

trait Point<T> {
fn new_point(x: T) -> Interval<T>;
}

impl Point<u32> for Interval<u32> {
fn new_point(x: u32) -> Self {
Interval::new(x, x + 1)
}
}

fn main() {
let mut interval_map = IntervalMap::<u32, i32>::new();
interval_map.insert(Interval::new(3, 7), 20);
interval_map.insert(Interval::new(2, 6), 15);

let tmp_point = Interval::new_point(5);
assert_eq!(tmp_point, Interval::new(5, 6));

interval_map.insert(tmp_point.clone(), 10);
assert_eq!(interval_map.get(&tmp_point).unwrap(), &10);
assert_eq!(
interval_map.find_all_overlap(&Interval::new_point(5)).len(),
3
);
}
68 changes: 68 additions & 0 deletions examples/string_affine.rs
Original file line number Diff line number Diff line change
@@ -0,0 +1,68 @@
use std::cmp;

use interval_map::{Interval, IntervalMap};

#[derive(Debug, Clone, PartialEq, Eq)]
pub enum StringAffine {
/// String
String(String),
/// Unbounded
Unbounded,
}

impl StringAffine {
pub fn new_key(s: &str) -> Self {
Self::String(s.to_string())
}

pub fn new_unbounded() -> Self {
Self::Unbounded
}
}

impl PartialOrd for StringAffine {
fn partial_cmp(&self, other: &Self) -> Option<cmp::Ordering> {
Some(self.cmp(other))
}
}

impl Ord for StringAffine {
fn cmp(&self, other: &Self) -> cmp::Ordering {
match (self, other) {
(StringAffine::String(x), StringAffine::String(y)) => x.cmp(y),
(StringAffine::String(_), StringAffine::Unbounded) => cmp::Ordering::Less,
(StringAffine::Unbounded, StringAffine::String(_)) => cmp::Ordering::Greater,
(StringAffine::Unbounded, StringAffine::Unbounded) => cmp::Ordering::Equal,
}
}
}

trait Point<T> {
fn new_point(x: T) -> Interval<T>;
}

impl Point<StringAffine> for Interval<StringAffine> {
fn new_point(x: StringAffine) -> Interval<StringAffine> {
match x {
StringAffine::String(mut x_string) => {
let low = x_string.clone();
x_string.push('\0');
Interval::new(
StringAffine::new_key(&low),
StringAffine::new_key(&x_string),
)
}
_ => panic!("new_point only receive StringAffine::String!"),
}
}
}

fn main() {
let mut interval_map = IntervalMap::<StringAffine, u32>::new();
interval_map.insert(
Interval::new(StringAffine::new_key("8"), StringAffine::Unbounded),
123,
);
assert!(interval_map.overlaps(&Interval::new_point(StringAffine::new_key("9"))));
assert!(!interval_map.overlaps(&Interval::new_point(StringAffine::new_key("7"))));
}

0 comments on commit a5d944e

Please sign in to comment.