Break up library into macros and map nested libraries for shared dependency with constant map macro.
This commit is contained in:
paladin
parent
805874a4c1
commit
d037e22164
8 changed files with 88 additions and 39 deletions
337
map/src/lib.rs
Normal file
337
map/src/lib.rs
Normal file
|
|
@ -0,0 +1,337 @@
|
|||
// I've already spent 2 months on this interpreter (now GC!?), and I'm tired, so I've cut a few corners...
|
||||
/* See some of Lua's notes on this table: https://www.lua.org/source/5.5/ltable.c.html
|
||||
This table uses an 'open-addressed' hashmap https://en.wikipedia.org/wiki/Open_addressing
|
||||
where collisions are handled by 'linear-probing' and something I call 'displacement'.
|
||||
|
||||
Motivation:
|
||||
Afaik, without having some mechanism that records or ensures known proximity between a displaced element
|
||||
that collided with other(s) in the hashmap and it's original home (at element's hash's index),
|
||||
the entire array may have to be checked just to see if an element is present. How can we fix this?
|
||||
|
||||
Solution:
|
||||
Lua uses Brent's method, which I couldn't find a concrete explanation of, here I'm just recording
|
||||
the maximum displacement that will need to be linearly-probed from the home entry to be certain of
|
||||
any 'home' element's presence in the greater array.
|
||||
|
||||
Notes:
|
||||
- If this value becomes large, the load factor is probably high and there would be a resize.
|
||||
- Values would only need to be shuffled around closer to their 'home' during a resize.
|
||||
- It is unlikely the displacement would large value without a resize amending that problem.
|
||||
*/
|
||||
use std::fmt::Debug;
|
||||
use std::hash::{DefaultHasher, Hash, Hasher};
|
||||
|
||||
type Displacement = u8;
|
||||
|
||||
#[cfg(feature = "bighash")]
|
||||
pub type Hashed = u64;
|
||||
#[cfg(not(feature = "bighash"))]
|
||||
pub type Hashed = u32;
|
||||
|
||||
pub trait Equivalent<T>: Sized {
|
||||
fn matches(&self, other: &T) -> bool;
|
||||
}
|
||||
|
||||
impl<T: PartialEq> Equivalent<T> for T {
|
||||
fn matches(&self, other: &T) -> bool {
|
||||
self == other
|
||||
}
|
||||
}
|
||||
|
||||
pub trait Hashes {
|
||||
fn hashed(&self) -> Hashed;
|
||||
}
|
||||
|
||||
impl Hashes for &str {
|
||||
fn hashed(&self) -> Hashed {
|
||||
let mut hasher = DefaultHasher::new();
|
||||
self[..self.len().min(6)].hash(&mut hasher);
|
||||
hasher.finish() as Hashed
|
||||
}
|
||||
}
|
||||
|
||||
pub trait TableEntry: Equivalent<Self> + Hashes {
|
||||
const VACANT: Self;
|
||||
fn is_vacant(&self) -> bool;
|
||||
}
|
||||
|
||||
#[derive(Debug)]
|
||||
pub struct Entry<E: TableEntry> {
|
||||
pub entry: E,
|
||||
home: usize,
|
||||
displacement: Displacement,
|
||||
}
|
||||
|
||||
impl<E: TableEntry> Entry<E> {
|
||||
const VACANT: Entry<E> = Entry {
|
||||
entry: E::VACANT,
|
||||
home: 0,
|
||||
displacement: 0,
|
||||
};
|
||||
}
|
||||
|
||||
#[derive(Debug)]
|
||||
pub struct DisplacementSet<E: TableEntry> {
|
||||
// default representing vacancy!
|
||||
pub map: Vec<Entry<E>>, // todo: make this private and use an iterator
|
||||
map_bounds: std::ops::Range<usize>,
|
||||
map_count: usize, // number of elements in table
|
||||
}
|
||||
|
||||
pub fn get<E: TableEntry + Equivalent<K> ,K: Hashes>(map: &[Entry<E>], index: K) -> Option<&E> {
|
||||
if map.len() == 0 {
|
||||
return None;
|
||||
}
|
||||
let location = index.hashed() as usize % map.len();
|
||||
let home = &map[location];
|
||||
if home.entry.matches(&index) {
|
||||
Some(&home.entry)
|
||||
} else {
|
||||
for i in 1..home.displacement as usize {
|
||||
let neighbour = &map[(location + i) % map.len()];
|
||||
if neighbour.entry.matches(&index) {
|
||||
return Some(&neighbour.entry);
|
||||
}
|
||||
}
|
||||
None
|
||||
}
|
||||
}
|
||||
|
||||
impl<E: TableEntry> DisplacementSet<E> {
|
||||
fn exchange(&mut self, len: usize) {
|
||||
let old = std::mem::replace(
|
||||
&mut self.map,
|
||||
(0..len).map(|_| Entry::VACANT).collect(), // todo: I don't like that this may go over all memory
|
||||
);
|
||||
self.map_bounds = self.lower()..self.upper();
|
||||
for Entry { entry, .. } in old {
|
||||
if !entry.is_vacant() {
|
||||
self.set(entry)
|
||||
}
|
||||
}
|
||||
}
|
||||
pub fn resize(&mut self, len: usize) {
|
||||
self.exchange(len.max(self.map_count + (self.map_count / 3)));
|
||||
self.map_bounds = 0..self.upper();
|
||||
}
|
||||
fn upper(&self) -> usize {
|
||||
(self.map.len() / 4) * 3
|
||||
}
|
||||
fn lower(&self) -> usize {
|
||||
self.map.len() / 3
|
||||
}
|
||||
fn ensure(&mut self) {
|
||||
if self.map_count + 1 > self.upper() {
|
||||
// free space is short
|
||||
self.exchange((self.map.len() + 4) * 2);
|
||||
} else if self.map_count < self.lower() {
|
||||
// too much free space
|
||||
self.exchange(self.map.len() / 2)
|
||||
}
|
||||
}
|
||||
pub fn set(&mut self, entry: E) {
|
||||
#[cfg(feature = "assertions")]
|
||||
assert!(!entry.is_vacant());
|
||||
if !self.map_bounds.contains(&(self.map_count + 1)) {
|
||||
// make sure the table is appropriately sized
|
||||
self.ensure();
|
||||
}
|
||||
let range = self.map.len();
|
||||
let home = entry.hashed() as usize % range;
|
||||
if self.map[home].entry.is_vacant() {
|
||||
// attempt to place it directly in an empty space
|
||||
// INSERT
|
||||
self.map[home].home = home;
|
||||
self.map[home].entry = entry;
|
||||
self.map[home].displacement = self.map[home].displacement.max(0);
|
||||
self.map_count += 1;
|
||||
} else if self.map[home].entry.matches(&entry) {
|
||||
// attempt to replace it directly
|
||||
// REPLACE
|
||||
self.map[home].entry = entry; // home.item.soft_drop()
|
||||
} else {
|
||||
// attempt to replace it in a collided neighbour location
|
||||
for i in 1..self.map[home].displacement + 1 {
|
||||
let neighbour = &mut self.map[(home + i as usize) % range]; // todo: is mod expensive?
|
||||
if neighbour.entry.matches(&entry) {
|
||||
// found where it was displaced to
|
||||
// REPLACE
|
||||
neighbour.entry = entry;
|
||||
return;
|
||||
}
|
||||
}
|
||||
// at this point it must be added, probe to place in an empty space
|
||||
for j in self.map[home].displacement + 1..Displacement::MAX - 1 {
|
||||
let neighbour = &mut self.map[(home + j as usize) % range];
|
||||
if neighbour.entry.is_vacant() {
|
||||
// new empty slot hooray!
|
||||
// INSERT
|
||||
neighbour.entry = entry;
|
||||
neighbour.home = home;
|
||||
self.map[home].displacement = j;
|
||||
self.map_count += 1;
|
||||
return;
|
||||
}
|
||||
}
|
||||
// impossible but this still needs to be complete
|
||||
#[cfg(feature = "messages")]
|
||||
eprintln!("Large table collision, are the hashes ok?\n\tBrute-force probing...");
|
||||
let mut free: Option<usize> = None;
|
||||
for k in 0..self.map.len() {
|
||||
let neighbour = &mut self.map[k];
|
||||
if neighbour.entry.is_vacant() && free.is_none() {
|
||||
free = Some(k);
|
||||
} else if neighbour.entry.matches(&entry) {
|
||||
// REPLACE
|
||||
neighbour.entry = entry;
|
||||
return;
|
||||
}
|
||||
}
|
||||
if let Some(k) = free {
|
||||
// INSERT
|
||||
self.map[k].home = home;
|
||||
self.map[k].entry = entry;
|
||||
self.map[home].displacement = Displacement::MAX;
|
||||
self.map_count += 1;
|
||||
return;
|
||||
}
|
||||
// must resize
|
||||
#[cfg(feature = "messages")]
|
||||
eprintln!("\n\tResizing...");
|
||||
self.resize((self.map.len() + 4) * 2);
|
||||
self.set(entry);
|
||||
}
|
||||
}
|
||||
pub fn remove<K>(&mut self, index: K)
|
||||
where
|
||||
E: Equivalent<K>,
|
||||
K: Hashes,
|
||||
{
|
||||
if self.map.len() == 0 {
|
||||
return;
|
||||
}
|
||||
let range = self.map.len();
|
||||
let home = index.hashed() as usize % range;
|
||||
if self.map[home].entry.matches(&index) {
|
||||
self.map[home].entry = E::VACANT;
|
||||
self.map_count -= 1;
|
||||
} else {
|
||||
if self.map[home].displacement != Displacement::MAX {
|
||||
let mut largest = 0;
|
||||
for i in 1..self.map[home].displacement {
|
||||
let neighbour = &mut self.map[(home + i as usize) % range]; // todo: maybe mod slows this down
|
||||
if neighbour.entry.matches(&index) {
|
||||
neighbour.entry = E::VACANT;
|
||||
self.map_count -= 1;
|
||||
if i == self.map[home].displacement {
|
||||
self.map[home].displacement = largest
|
||||
}
|
||||
return;
|
||||
} else if !neighbour.entry.is_vacant() && neighbour.home == home {
|
||||
largest = i;
|
||||
}
|
||||
}
|
||||
} else {
|
||||
let mut largest = 0;
|
||||
#[cfg(feature = "assertions")]
|
||||
let mut finished = false; // what
|
||||
for k in 0..self.map.len() {
|
||||
let neighbour = &mut self.map[k];
|
||||
if neighbour.entry.matches(&index) {
|
||||
neighbour.entry = E::VACANT;
|
||||
self.map_count -= 1;
|
||||
#[cfg(feature = "assertions")]
|
||||
{
|
||||
assert!(!finished);
|
||||
finished = true;
|
||||
}
|
||||
}
|
||||
if neighbour.home == home && !neighbour.entry.is_vacant() {
|
||||
largest = largest.max(if k < home {
|
||||
k + self.map.len() - home - 1 // ?
|
||||
} else {
|
||||
k - home
|
||||
})
|
||||
}
|
||||
}
|
||||
self.map[home].displacement = if largest > Displacement::MAX as usize {
|
||||
Displacement::MAX
|
||||
} else {
|
||||
largest as Displacement
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
pub fn get<K>(&mut self, index: K) -> Option<&E>
|
||||
where
|
||||
E: Equivalent<K>,
|
||||
K: Hashes,
|
||||
{
|
||||
get(self.map.as_slice(),index)
|
||||
}
|
||||
pub fn new() -> DisplacementSet<E> {
|
||||
DisplacementSet {
|
||||
map: Vec::new(),
|
||||
map_count: 0,
|
||||
map_bounds: (0..0).into(),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Clone)]
|
||||
pub struct MapEntry<K: Hashes + PartialEq, V>(pub Option<(K, V)>);
|
||||
|
||||
impl<K: PartialEq + Hashes, V> Equivalent<K> for MapEntry<K, V> {
|
||||
fn matches(&self, other: &K) -> bool {
|
||||
&self.0.as_ref().unwrap().0 == other
|
||||
}
|
||||
}
|
||||
|
||||
impl<K: Hashes + PartialEq, V> Hashes for MapEntry<K, V> {
|
||||
fn hashed(&self) -> Hashed {
|
||||
self.0.as_ref().unwrap().0.hashed()
|
||||
}
|
||||
}
|
||||
|
||||
impl<K: Hashes + PartialEq, V> PartialEq for MapEntry<K, V> {
|
||||
fn eq(&self, other: &Self) -> bool {
|
||||
self.0.as_ref().unwrap().0 == other.0.as_ref().unwrap().0
|
||||
}
|
||||
}
|
||||
|
||||
impl<K: Hashes + PartialEq, V> TableEntry for MapEntry<K, V> {
|
||||
const VACANT: Self = MapEntry(None);
|
||||
fn is_vacant(&self) -> bool {
|
||||
self.0.is_none()
|
||||
}
|
||||
}
|
||||
|
||||
pub struct Map<K: Hashes + PartialEq, V>(pub DisplacementSet<MapEntry<K, V>>);
|
||||
|
||||
impl<K: Hashes + PartialEq, V> Map<K, V> {
|
||||
pub fn resize(&mut self, len: usize) {
|
||||
self.0.resize(len)
|
||||
}
|
||||
pub fn set(&mut self, index: K, value: V) {
|
||||
self.0.set(MapEntry(Some((index, value))));
|
||||
}
|
||||
pub fn remove(&mut self, index: K) {
|
||||
self.0.remove(index);
|
||||
}
|
||||
pub fn get(&mut self, index: K) -> Option<&V> {
|
||||
if let Some(ref entry) = self.0.get(index) {
|
||||
Some(&entry.0.as_ref().unwrap().1)
|
||||
} else {
|
||||
None
|
||||
}
|
||||
}
|
||||
pub fn new() -> Map<K, V> {
|
||||
Map(DisplacementSet::new())
|
||||
}
|
||||
}
|
||||
|
||||
impl<K: Debug + Hashes + PartialEq, V: Debug> Debug for Map<K, V> {
|
||||
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
|
||||
f.write_fmt(format_args!("Map")) // todo: debug and print the entire thing?
|
||||
}
|
||||
}
|
||||
Loading…
Add table
Add a link
Reference in a new issue