Crate fingertrees

Expand description

Finger Trees

Finger trees is a functional representation of persistent sequences supporting access to the ends in amortized constant time, and concatenation and splitting in time logarithmic in the size of the smaller piece. It also has split operation defined in general form, which can be used to implement sequence, priority queue, search tree, priority search queue and more datastructures.

Links:

Original paper: Finger Trees: A Simple General-purpose Data Structure
Wikipedia article: FingerTree

Notes:

This implementation does not use non-regular recursive types as implementation described in the paper. As rust’s monomorphization does not play well with such types.
Implmentation abstracts over reference counted types Rc/Arc. Using type family trick.
Uses strict spine in implementation.
Iterator returns cloned value, and in general this implementation assumes that value stored in a tree is cheaply clonable, if it is not you can always put it in a Rc/Arc or anything else.

Examples:

use fingertrees::measure::Size;
use fingertrees::monoid::Sum;
use fingertrees::{FingerTree, Measured, RcRefs};

// construct `Rc` based finger tree with `Size` measure
let ft: FingerTree<RcRefs, _> = vec!["one", "two", "three", "four", "five"]
    .into_iter()
    .map(Size)
    .collect();
assert_eq!(ft.measure(), Sum(5));

// split with predicate
let (left, right) = ft.split(|measure| *measure > Sum(2));
assert_eq!(left.measure(), Sum(2));
assert_eq!(Vec::from_iter(&left), vec![Size("one"), Size("two")]);
assert_eq!(right.measure(), Sum(3));
assert_eq!(Vec::from_iter(&right), vec![Size("three"), Size("four"), Size("five")]);

// concatinate
assert_eq!(ft, left + right);

// push values
assert_eq!(
    ft.push_left(Size("left")).push_right(Size("right")),
    vec!["left", "one", "two", "three", "four", "five", "right"]
         .into_iter()
         .map(Size)
         .collect(),
);