|
| 1 | +use std::marker::PhantomData; |
| 2 | +use std::hash::Hash; |
| 3 | +use std::hash::Hasher; |
| 4 | +use std::hash::BuildHasher; |
| 5 | +use std::mem::{self, size_of}; |
| 6 | +use std::ptr::{Unique, NonNull}; |
| 7 | +use std::alloc::{Global, Alloc}; |
| 8 | +use std::collections::hash_map::RandomState; |
| 9 | +use std::borrow::Borrow; |
| 10 | + |
| 11 | +const ENTRIES_PER_GROUP: usize = 5; |
| 12 | + |
| 13 | +#[repr(align(64), C)] |
| 14 | +pub struct Group { |
| 15 | + hashes: [u32; ENTRIES_PER_GROUP], |
| 16 | + size: u32, |
| 17 | + values: [u64; ENTRIES_PER_GROUP], |
| 18 | +} |
| 19 | + |
| 20 | +impl Group { |
| 21 | + #[inline(always)] |
| 22 | + fn search_for_empty(&self) -> Option<usize> { |
| 23 | + if self.size != ENTRIES_PER_GROUP as u32 { |
| 24 | + Some(self.size as usize) |
| 25 | + } else { |
| 26 | + None |
| 27 | + } |
| 28 | + } |
| 29 | + |
| 30 | + #[inline(always)] |
| 31 | + fn search_with<K, F: FnMut(&K) -> bool>(&self, eq: &mut F, hash: u32) -> Option<(usize, bool)> { |
| 32 | + for i in 0..ENTRIES_PER_GROUP { |
| 33 | + let h = unsafe { *self.hashes.get_unchecked(i) }; |
| 34 | + if h == hash && eq(unsafe { mem::transmute(self.values.get_unchecked(i)) }) { |
| 35 | + return Some((i, false)) |
| 36 | + } |
| 37 | + } |
| 38 | + self.search_for_empty().map(|i| (i, true)) |
| 39 | + } |
| 40 | + |
| 41 | + #[inline(always)] |
| 42 | + fn set(&mut self, pos: usize, hash: u32, value: u64) { |
| 43 | + unsafe { |
| 44 | + *self.hashes.get_unchecked_mut(pos) = hash; |
| 45 | + *self.values.get_unchecked_mut(pos) = value; |
| 46 | + } |
| 47 | + } |
| 48 | + |
| 49 | + #[inline(always)] |
| 50 | + fn iter<F: FnMut(u32, u64)>(&self, f: &mut F) { |
| 51 | + for i in 0..ENTRIES_PER_GROUP { |
| 52 | + unsafe { |
| 53 | + let h = *self.hashes.get_unchecked(i); |
| 54 | + if h != 0 { |
| 55 | + f(h, *self.values.get_unchecked(i)) |
| 56 | + } |
| 57 | + } |
| 58 | + } |
| 59 | + } |
| 60 | +} |
| 61 | + |
| 62 | +pub struct Table { |
| 63 | + group_mask: usize, |
| 64 | + size: usize, |
| 65 | + capacity: usize, |
| 66 | + groups: Unique<Group>, |
| 67 | +} |
| 68 | + |
| 69 | +pub struct RawEntry { |
| 70 | + group: *mut Group, |
| 71 | + pos: usize, |
| 72 | + empty: bool |
| 73 | +} |
| 74 | + |
| 75 | +impl Drop for Table { |
| 76 | + fn drop(&mut self) { |
| 77 | + if self.group_mask == 0 { |
| 78 | + return; |
| 79 | + } |
| 80 | + |
| 81 | + unsafe { |
| 82 | + Global.dealloc_array( |
| 83 | + NonNull::new_unchecked(self.groups.as_ptr()), |
| 84 | + self.group_mask + 1 |
| 85 | + ).unwrap(); |
| 86 | + } |
| 87 | + } |
| 88 | +} |
| 89 | + |
| 90 | +impl Table { |
| 91 | + unsafe fn new_uninitialized(group_count: usize) -> Table { |
| 92 | + assert!(size_of::<Group>() == 64); |
| 93 | + let groups: NonNull<Group> = Global.alloc_array(group_count).unwrap(); |
| 94 | + let capacity2 = group_count * ENTRIES_PER_GROUP; |
| 95 | + let capacity1 = capacity2 - 1; |
| 96 | + //let capacity = (capacity1 * 10 + 10 - 1) / 11; |
| 97 | + let capacity = (capacity1 * 10 + 10 - 1) / 13; |
| 98 | + //println!("capacity1 {} capacity {}", capacity1, capacity); |
| 99 | + assert!(capacity < capacity2); |
| 100 | + |
| 101 | + for i in 0..group_count { |
| 102 | + let group = unsafe { |
| 103 | + &mut (*groups.as_ptr().offset(i as isize)) |
| 104 | + }; |
| 105 | + group.hashes = [0; ENTRIES_PER_GROUP]; |
| 106 | + group.size = 0; |
| 107 | + } |
| 108 | + |
| 109 | + Table { |
| 110 | + group_mask: group_count.wrapping_sub(1), |
| 111 | + size: 0, |
| 112 | + capacity, |
| 113 | + groups: Unique::new_unchecked(groups.as_ptr()), |
| 114 | + } |
| 115 | + } |
| 116 | + |
| 117 | + fn search_for_empty(&self, hash: u64) -> RawEntry { |
| 118 | + let group_idx = hash as u32 as usize; |
| 119 | + let mask = self.group_mask; |
| 120 | + let mut group_idx = group_idx & mask; |
| 121 | + |
| 122 | + loop { |
| 123 | + let group_ptr = unsafe { |
| 124 | + self.groups.as_ptr().offset(group_idx as isize) |
| 125 | + }; |
| 126 | + let group = unsafe { |
| 127 | + &(*group_ptr) |
| 128 | + }; |
| 129 | + match unsafe { group.search_for_empty() } { |
| 130 | + Some(pos) => return RawEntry { |
| 131 | + group: group_ptr, |
| 132 | + pos, |
| 133 | + empty: true, |
| 134 | + }, |
| 135 | + None => (), |
| 136 | + } |
| 137 | + group_idx = (group_idx + 1) & mask; |
| 138 | + } |
| 139 | + } |
| 140 | + |
| 141 | + fn search_with<K, F: FnMut(&K) -> bool>(&self, mut eq: F, hash: u64) -> RawEntry { |
| 142 | + let group_idx = hash as u32 as usize; |
| 143 | + let mask = self.group_mask; |
| 144 | + let mut group_idx = group_idx & mask; |
| 145 | + |
| 146 | + loop { |
| 147 | + let group_ptr = unsafe { |
| 148 | + self.groups.as_ptr().offset(group_idx as isize) |
| 149 | + }; |
| 150 | + let group = unsafe { |
| 151 | + &(*group_ptr) |
| 152 | + }; |
| 153 | + let r = unsafe { group.search_with(&mut eq, hash as u32) } ; |
| 154 | + match r { |
| 155 | + Some((pos, empty)) => return RawEntry { |
| 156 | + group: group_ptr, |
| 157 | + pos, |
| 158 | + empty, |
| 159 | + }, |
| 160 | + None => (), |
| 161 | + } |
| 162 | + group_idx = (group_idx + 1) & mask; |
| 163 | + } |
| 164 | + } |
| 165 | + |
| 166 | + fn iter<F: FnMut(u32, u64)>(&self, mut f: F) { |
| 167 | + if self.group_mask == 0 { |
| 168 | + return; |
| 169 | + } |
| 170 | + for i in 0..(self.group_mask + 1) { |
| 171 | + let group = unsafe { |
| 172 | + &(*self.groups.as_ptr().offset(i as isize)) |
| 173 | + }; |
| 174 | + group.iter(&mut f); |
| 175 | + } |
| 176 | + } |
| 177 | +} |
| 178 | + |
| 179 | +pub struct Interner<K: Eq + Hash, S = RandomState> { |
| 180 | + hash_builder: S, |
| 181 | + table: Table, |
| 182 | + marker: PhantomData<K>, |
| 183 | +} |
| 184 | + |
| 185 | +impl<K: Eq + Hash, S: Default> Default for Interner<K, S> { |
| 186 | + fn default() -> Self { |
| 187 | + assert!(size_of::<K>() == 8); |
| 188 | + Interner { |
| 189 | + hash_builder: S::default(), |
| 190 | + table: Table { |
| 191 | + group_mask: 0, |
| 192 | + size: 0, |
| 193 | + capacity: 0, |
| 194 | + groups: unsafe { Unique::new_unchecked(NonNull::dangling().as_ptr()) }, |
| 195 | + }, |
| 196 | + marker: PhantomData, |
| 197 | + } |
| 198 | + } |
| 199 | +} |
| 200 | + |
| 201 | +pub fn make_hash<T: ?Sized, S>(hash_state: &S, t: &T) -> u64 |
| 202 | + where T: Hash, |
| 203 | + S: BuildHasher |
| 204 | +{ |
| 205 | + let mut state = hash_state.build_hasher(); |
| 206 | + t.hash(&mut state); |
| 207 | + state.finish() | (1 << 31) |
| 208 | +} |
| 209 | + |
| 210 | +impl<K: Eq + Hash, S: BuildHasher> Interner<K, S> { |
| 211 | + #[inline(never)] |
| 212 | + #[cold] |
| 213 | + fn expand(&mut self) { |
| 214 | + let mut new_table = unsafe { |
| 215 | + Table::new_uninitialized((self.table.group_mask + 1) << 1) |
| 216 | + }; |
| 217 | + new_table.size = self.table.size; |
| 218 | + self.table.iter(|h, v| { |
| 219 | + let spot = new_table.search_for_empty(h as u64); |
| 220 | + unsafe { |
| 221 | + (*spot.group).size += 1; |
| 222 | + (*spot.group).set(spot.pos, h, v); |
| 223 | + } |
| 224 | + }); |
| 225 | + self.table = new_table; |
| 226 | + } |
| 227 | + |
| 228 | + #[inline(always)] |
| 229 | + fn incr(&mut self) { |
| 230 | + if self.table.size + 1 > self.table.capacity { |
| 231 | + self.expand() |
| 232 | + } |
| 233 | + } |
| 234 | + |
| 235 | + pub fn len(&self) -> usize { |
| 236 | + self.table.size |
| 237 | + } |
| 238 | + |
| 239 | + #[inline] |
| 240 | + pub fn intern<Q, F: FnOnce(Q) -> K>(&mut self, value: Q, make: F) -> &K |
| 241 | + where K: Borrow<Q>, |
| 242 | + Q: Hash + Eq |
| 243 | + { |
| 244 | + self.incr(); |
| 245 | + let hash = make_hash(&self.hash_builder, &value); |
| 246 | + let spot = self.table.search_with::<K, _>(|k| value.eq(k.borrow()), hash); |
| 247 | + unsafe { |
| 248 | + if spot.empty { |
| 249 | + self.table.size += 1; |
| 250 | + (*spot.group).size += 1; |
| 251 | + let key = make(value); |
| 252 | + (*spot.group).set(spot.pos, hash as u32, *(&key as *const _ as *const u64)); |
| 253 | + } |
| 254 | + &*((*spot.group).values.get_unchecked(spot.pos) as *const _ as *const K) |
| 255 | + } |
| 256 | + } |
| 257 | +} |
0 commit comments