walrus 0.8.0

//! Walrus-specific hash maps and hash sets which typically hash much more
//! quickly than libstd's hash map which isn't optimized for speed.

use id_arena::Id;
use std::collections::{HashMap, HashSet};
use std::hash::{BuildHasher, Hasher};

/// Type parameter for the hasher of a `HashMap` or `HashSet`
#[derive(Default, Copy, Clone, Debug)]
pub struct BuildIdHasher;

/// Hasher constructed by `BuildIdHasher`, only ever used to hash an `Id`
#[derive(Default, Copy, Clone, Debug)]
pub struct IdHasher {
    hash: u64,
}

pub type IdHashMap<K, V> = HashMap<Id<K>, V, BuildIdHasher>;
pub type IdHashSet<T> = HashSet<Id<T>, BuildIdHasher>;

impl BuildHasher for BuildIdHasher {
    type Hasher = IdHasher;

    fn build_hasher(&self) -> IdHasher {
        IdHasher { hash: 0 }
    }
}

// This is the "speed" of this hasher. We're only ever going to hash `Id<T>`,
// and `Id<T>` is composed of two parts: one part arena ID and one part index
// in the arena. The arena ID is a 32-bit integer and the index is a `usize`, so
// we're only going to handle those two types here.
//
// The goal is to produce a 64-bit result, and 32 of those comes from the arena
// ID. We can assume that none of the arena indexes are larger than 32-bit
// because wasm can't encode more than 2^32 items anyway, so we can basically
// just shift each item into the lower 32-bits of the hash. Ideally this'll end
// up producing a very optimized version of hashing!
//
// To explore this a bit, see https://godbolt.org/z/QxkXgF
impl Hasher for IdHasher {
    fn write_u32(&mut self, amt: u32) {
        self.hash <<= 32;
        self.hash |= amt as u64;
    }

    fn write_usize(&mut self, amt: usize) {
        self.hash <<= 32;
        self.hash |= amt as u64;
    }

    fn write(&mut self, _other: &[u8]) {
        panic!("hashing an `Id` should only be usize/u32")
    }

    fn finish(&self) -> u64 {
        self.hash
    }
}