toodee

TooDee is a lightweight and high performance two-dimensional wrapper around a growable Vec.

TooDeeView and TooDeeViewMut allow you create two-dimensional wrappers around a slice.

Core features

• Raw access to the underlying vector's slice via data() and data_mut().
• Creation of performant two-dimensional subsets using view() and view_mut().
• get_unchecked(Coordinate) and get_unchecked_row(usize) for faster (unsafe) access to cells or rows.
• Most operations are implemented for both TooDee and TooDeeViewMut structs - see below for how this pattern can be extended.
• Get/set specific cells using indexing, e.g., let my_row = toodee[row]; my_row[col] = val;.
• Index with a Coordinate if you prefer, e.g., toodee[(col, row)] = val.
• Index by row index (i.e., row major) to access row slices, e.g., &toodee[row].
• Iteration, any which way - rows(), rows_mut(), col(), col_mut(), cells(), cells_mut().
• #[no_std] compliant.
• Can create a new TooDeeView from a &[T], or a TooDeeViewMut from a &mut [T].
• insert_col(), remove_col(), insert_row(), and remove_row() implementations with good performance.

Features

copy, included by default

The CopyOps trait provides various operations that copy data within the same 2D array, or copy data from one array to another. Many of these operations are named like their slice counterparts, e.g., copy_from_slice() or copy_from_toodee().

translate, included by default

The TranslateOps trait provides common translation algorithms, including:

• translate_with_wrap(), a way to shift data around vertically and horizontally.
• flip_rows(), i.e., a mirror translation of data about the center row.
• flip_cols(), i.e., a mirror translation of data about the center column.

sort, included by default

The SortOps trait provides efficient implementations of:

• sort_by_row() operations, with stable and unstable variants.
• sort_by_col() operations, with stable and unstable variants.

serde, included by default

Serialization and deserialization of TooDee objects. TooDeeView and TooDeeViewMut can be serialized, but must be deserialized into a TooDee structure.

Build Your Own 2D Algorithms

Traits such as SortOps contain additional algorithms. These traits are defined by extending the TooDeeOpsMut trait, which has been implemented for TooDee and TooDeeViewMut. I recommend taking the same approach because the algorithms you implement will then work on both structs. Additionally, you can override the trait's default implementation if necessary.

The implementation of a new trait could look something like:

pub trait FooOps<T> : TooDeeOpsMut<T> {

    fn foo(&mut self) -> Bar {
        ...
        return bar;
    }
}

The above code would provide a default foo() implementation that could be overridden if required. Then it's simply a matter of stating that both TooDee and TooDeeOpsMut implement FooOps:

impl<T> FooOps<T> for TooDeeViewMut<'_, T> {}

impl<T> FooOps<T> for TooDee<T> {}

Once the implementations are available, just call the methods, e.g.,

let bar = my_toodee.foo();
let bar_view = my_toodee_mut_view.foo();

Happy coding 😄

TODO

• Pathfinding algorithms?
• Image/bitmap algorithms?!

Motivation

Similar libraries do exist, but they lacked either performance, flexibility, or functionality.

Here's a small feature comparison chart:

	Storage order	Structs supported	Growable?	Mutable views?	Raw data access?	Iterate over row slices?	Notes
toodee::TooDee	Row-major	Anything (Sized)	Yes	Yes	Yes	Yes
image::ImageBuffer	Row-major	image::Pixel	No	No	Yes	No	Good for image processing - see the imageproc crate.
image::SubImage	Row-major	image::Pixel	No	Yes	No	No
grid::Grid	Row-major	Clone	Yes	No	Yes	No	Similar to TooDee, but not as functionally rich.
array2d::Array2D	Row-major	Clone	No	No	No	No
imgref::Img	Row-major	Anything (Sized)	No	Yes	Yes	Yes
nalgebra::Matrix	Column-major	Scalar	Yes	Yes	Yes	No	Use this for vector/matrix math.

Goals

• High performance and good flexibility, with the constraint of using a 1-D vector.
• Suitable for use in image processing, but not restricted to this problem domain.
• Provide solid implementations of non-trivial 2D operations.

Non-goals

• GPU integration

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.