cl-structures: Use hashbrown's hash table implementation for deduplication.

This commit is contained in:
John 2024-04-24 17:43:02 -05:00
parent e70ffd1895
commit be604b7b45
3 changed files with 26 additions and 37 deletions

View File

@ -8,3 +8,4 @@ license.workspace = true
publish.workspace = true
[dependencies]
hashbrown = { version = "0.14.3", default-features = false }

View File

@ -1,55 +1,49 @@
//! A string interner with deduplication
use super::{symbol::Symbol, StringArena};
use std::{
collections::{hash_map::RawEntryMut, HashMap},
hash::{BuildHasher, RandomState},
};
use hashbrown::hash_table::HashTable;
use std::hash::{BuildHasher, RandomState};
#[derive(Debug)]
pub struct Interner<T: Symbol, H: BuildHasher = RandomState> {
map: HashMap<T, ()>,
arena: StringArena<T>,
pub struct Interner<Sym: Symbol, H: BuildHasher = RandomState> {
set: HashTable<Sym>,
arena: StringArena<Sym>,
hasher: H,
}
impl<T: Symbol, H: BuildHasher + Default> Default for Interner<T, H> {
impl<Sym: Symbol, H: BuildHasher + Default> Default for Interner<Sym, H> {
fn default() -> Self {
Self { map: Default::default(), arena: Default::default(), hasher: Default::default() }
Self { set: Default::default(), arena: Default::default(), hasher: Default::default() }
}
}
impl<T: Symbol, H: BuildHasher> Interner<T, H> {
pub fn get_or_insert(&mut self, s: &str) -> T {
let Self { map, arena, hasher } = self;
impl<Sym: Symbol, H: BuildHasher> Interner<Sym, H> {
pub fn get_or_insert(&mut self, s: &str) -> Sym {
let Self { set: map, arena, hasher } = self;
let hash = hasher.hash_one(s);
match map.raw_entry_mut().from_hash(hash, is_match(s, arena)) {
RawEntryMut::Occupied(entry) => *entry.into_key(),
RawEntryMut::Vacant(entry) => {
let tok = arena.push_string(s);
*(entry.insert_hashed_nocheck(hash, tok, ()).0)
}
}
*map.entry(hash, is_match(s, arena), |t| {
hasher.hash_one(arena.get(*t).unwrap())
})
.or_insert_with(|| arena.push_string(s))
.get()
}
pub fn get(&self, s: &str) -> Option<T> {
let Self { map, arena, hasher } = self;
map.raw_entry()
.from_hash(hasher.hash_one(s), is_match(s, arena))
.map(|entry| *entry.0)
pub fn get(&self, s: &str) -> Option<Sym> {
let Self { set: map, arena, hasher } = self;
map.find(hasher.hash_one(s), is_match(s, arena)).copied()
}
pub fn get_str(&self, sym: T) -> Option<&str> {
pub fn get_str(&self, sym: Sym) -> Option<&str> {
self.arena.get(sym)
}
}
fn is_match<'a, T: Symbol>(
target: &'a str,
arena: &'a StringArena<T>,
) -> impl Fn(&T) -> bool + 'a {
fn is_match<'a, Sym: Symbol>(
s: &'a str,
arena: &'a StringArena<Sym>,
) -> impl Fn(&Sym) -> bool + 'a {
move |sym| match arena.get(*sym) {
Some(sym) => sym == target,
Some(sym) => sym == s,
None => false,
}
}

View File

@ -2,13 +2,7 @@
//! - [Span](struct@span::Span): Stores a start and end [Loc](struct@span::Loc)
//! - [Loc](struct@span::Loc): Stores the index in a stream
#![warn(clippy::all)]
#![feature(
inline_const,
dropck_eyepatch,
decl_macro,
get_many_mut,
hash_raw_entry
)]
#![feature(inline_const, dropck_eyepatch, decl_macro, get_many_mut)]
#![deny(unsafe_op_in_unsafe_fn)]
pub mod arena;