rust/src/libcore/send_map.rs

458 lines
15 KiB
Rust
Raw Normal View History

/*!
Sendable hash maps. Very much a work in progress.
*/
2012-09-04 13:12:17 -05:00
use cmp::Eq;
use hash::Hash;
use to_bytes::IterBytes;
trait SendMap<K:Eq Hash, V: Copy> {
// FIXME(#3148) ^^^^ once find_ref() works, we can drop V:copy
fn insert(&mut self, +k: K, +v: V) -> bool;
fn remove(&mut self, k: &K) -> bool;
fn clear(&mut self);
pure fn len(&const self) -> uint;
pure fn is_empty(&const self) -> bool;
2012-09-14 23:39:09 -05:00
pure fn contains_key(&const self, k: &K) -> bool;
pure fn each_ref(&self, blk: fn(k: &K, v: &V) -> bool);
pure fn each_key_ref(&self, blk: fn(k: &K) -> bool);
pure fn each_value_ref(&self, blk: fn(v: &V) -> bool);
pure fn find(&const self, k: &K) -> Option<V>;
pure fn get(&const self, k: &K) -> V;
pure fn find_ref(&self, k: &K) -> Option<&self/V>;
pure fn get_ref(&self, k: &K) -> &self/V;
}
/// Open addressing with linear probing.
mod linear {
2012-08-14 18:54:13 -05:00
export LinearMap, linear_map, linear_map_with_capacity, public_methods;
const initial_capacity: uint = 32u; // 2^5
struct Bucket<K:Eq Hash,V> {
hash: uint,
key: K,
value: V,
}
struct LinearMap<K:Eq Hash,V> {
k0: u64,
k1: u64,
resize_at: uint,
size: uint,
buckets: ~[Option<Bucket<K,V>>],
}
// FIXME(#3148) -- we could rewrite found_entry
// to have type option<&bucket<K,V>> which would be nifty
// However, that won't work until #3148 is fixed
2012-08-14 18:54:13 -05:00
enum SearchResult {
FoundEntry(uint), FoundHole(uint), TableFull
}
fn resize_at(capacity: uint) -> uint {
((capacity as float) * 3. / 4.) as uint
}
fn LinearMap<K:Eq Hash,V>() -> LinearMap<K,V> {
linear_map_with_capacity(32)
}
fn linear_map_with_capacity<K:Eq Hash,V>(
2012-08-14 18:54:13 -05:00
initial_capacity: uint) -> LinearMap<K,V> {
let r = rand::Rng();
linear_map_with_capacity_and_keys(r.gen_u64(), r.gen_u64(),
initial_capacity)
}
fn linear_map_with_capacity_and_keys<K:Eq Hash,V> (
k0: u64, k1: u64,
initial_capacity: uint) -> LinearMap<K,V> {
LinearMap {
k0: k0, k1: k1,
resize_at: resize_at(initial_capacity),
size: 0,
buckets: vec::from_fn(initial_capacity, |_i| None)
}
}
priv impl<K:Hash IterBytes Eq, V> LinearMap<K,V> {
#[inline(always)]
2012-08-22 23:01:30 -05:00
pure fn to_bucket(&const self,
h: uint) -> uint {
// FIXME(#3041) borrow a more sophisticated technique here from
// Gecko, for example borrowing from Knuth, as Eich so
// colorfully argues for here:
// https://bugzilla.mozilla.org/show_bug.cgi?id=743107#c22
h % self.buckets.len()
}
#[inline(always)]
2012-08-22 23:01:30 -05:00
pure fn next_bucket(&const self,
idx: uint,
len_buckets: uint) -> uint {
let n = (idx + 1) % len_buckets;
unsafe{ // argh. log not considered pure.
2012-08-22 19:24:52 -05:00
debug!("next_bucket(%?, %?) = %?", idx, len_buckets, n);
}
2012-08-01 19:30:05 -05:00
return n;
}
#[inline(always)]
2012-08-22 23:01:30 -05:00
pure fn bucket_sequence(&const self,
hash: uint,
op: fn(uint) -> bool) -> uint {
let start_idx = self.to_bucket(hash);
let len_buckets = self.buckets.len();
let mut idx = start_idx;
loop {
if !op(idx) {
2012-08-01 19:30:05 -05:00
return idx;
}
idx = self.next_bucket(idx, len_buckets);
if idx == start_idx {
2012-08-01 19:30:05 -05:00
return start_idx;
}
}
}
#[inline(always)]
2012-08-22 23:01:30 -05:00
pure fn bucket_for_key(&const self,
2012-08-20 14:23:37 -05:00
buckets: &[Option<Bucket<K,V>>],
2012-08-22 23:01:30 -05:00
k: &K) -> SearchResult {
let hash = k.hash_keyed(self.k0, self.k1) as uint;
self.bucket_for_key_with_hash(buckets, hash, k)
}
#[inline(always)]
2012-08-22 23:01:30 -05:00
pure fn bucket_for_key_with_hash(&const self,
2012-08-20 14:23:37 -05:00
buckets: &[Option<Bucket<K,V>>],
2012-08-22 23:01:30 -05:00
hash: uint,
k: &K) -> SearchResult {
let _ = for self.bucket_sequence(hash) |i| {
2012-08-06 14:34:08 -05:00
match buckets[i] {
Some(bkt) => if bkt.hash == hash && *k == bkt.key {
return FoundEntry(i);
},
None => return FoundHole(i)
}
};
2012-08-14 18:54:13 -05:00
return TableFull;
}
/// Expands the capacity of the array and re-inserts each
/// of the existing buckets.
2012-08-22 23:01:30 -05:00
fn expand(&mut self) {
let old_capacity = self.buckets.len();
let new_capacity = old_capacity * 2;
self.resize_at = ((new_capacity as float) * 3.0 / 4.0) as uint;
2012-08-20 14:23:37 -05:00
let mut old_buckets = vec::from_fn(new_capacity, |_i| None);
self.buckets <-> old_buckets;
for uint::range(0, old_capacity) |i| {
2012-08-20 14:23:37 -05:00
let mut bucket = None;
bucket <-> old_buckets[i];
2012-09-10 14:14:14 -05:00
self.insert_opt_bucket(move bucket);
}
}
fn insert_opt_bucket(&mut self, +bucket: Option<Bucket<K,V>>) {
match move bucket {
Some(Bucket {hash: move hash,
key: move key,
value: move value}) => {
self.insert_internal(hash, move key, move value);
}
None => {}
}
}
/// Inserts the key value pair into the buckets.
/// Assumes that there will be a bucket.
/// True if there was no previous entry with that key
2012-08-22 23:01:30 -05:00
fn insert_internal(&mut self, hash: uint, +k: K, +v: V) -> bool {
match self.bucket_for_key_with_hash(self.buckets, hash, &k) {
TableFull => { fail ~"Internal logic error"; }
FoundHole(idx) => {
debug!("insert fresh (%?->%?) at idx %?, hash %?",
k, v, idx, hash);
self.buckets[idx] = Some(Bucket {hash: hash,
key: k,
value: v});
self.size += 1;
true
}
FoundEntry(idx) => {
debug!("insert overwrite (%?->%?) at idx %?, hash %?",
k, v, idx, hash);
self.buckets[idx] = Some(Bucket {hash: hash,
key: k,
value: v});
false
}
}
}
2012-08-22 23:01:30 -05:00
fn search(&self,
hash: uint,
2012-08-20 14:23:37 -05:00
op: fn(x: &Option<Bucket<K,V>>) -> bool) {
2012-08-22 23:01:30 -05:00
let _ = self.bucket_sequence(hash, |i| op(&self.buckets[i]));
}
}
impl<K:Hash IterBytes Eq,V> LinearMap<K,V> {
2012-08-22 23:01:30 -05:00
fn insert(&mut self, +k: K, +v: V) -> bool {
if self.size >= self.resize_at {
// n.b.: We could also do this after searching, so
// that we do not resize if this call to insert is
// simply going to update a key in place. My sense
// though is that it's worse to have to search through
// buckets to find the right spot twice than to just
// resize in this corner case.
self.expand();
}
let hash = k.hash_keyed(self.k0, self.k1) as uint;
2012-09-10 14:14:14 -05:00
self.insert_internal(hash, move k, move v)
}
2012-08-22 23:01:30 -05:00
fn remove(&mut self, k: &K) -> bool {
// Removing from an open-addressed hashtable
// is, well, painful. The problem is that
// the entry may lie on the probe path for other
// entries, so removing it would make you think that
// those probe paths are empty.
//
// To address this we basically have to keep walking,
// re-inserting entries we find until we reach an empty
// bucket. We know we will eventually reach one because
// we insert one ourselves at the beginning (the removed
// entry).
//
// I found this explanation elucidating:
// http://www.maths.lse.ac.uk/Courses/MA407/del-hash.pdf
2012-08-06 14:34:08 -05:00
let mut idx = match self.bucket_for_key(self.buckets, k) {
TableFull | FoundHole(_) => return false,
FoundEntry(idx) => idx
};
let len_buckets = self.buckets.len();
2012-08-20 14:23:37 -05:00
self.buckets[idx] = None;
idx = self.next_bucket(idx, len_buckets);
while self.buckets[idx].is_some() {
2012-08-20 14:23:37 -05:00
let mut bucket = None;
bucket <-> self.buckets[idx];
2012-09-10 14:14:14 -05:00
self.insert_opt_bucket(move bucket);
idx = self.next_bucket(idx, len_buckets);
}
self.size -= 1;
2012-08-01 19:30:05 -05:00
return true;
}
2012-08-22 23:01:30 -05:00
fn clear(&mut self) {
for uint::range(0, self.buckets.len()) |idx| {
2012-08-20 14:23:37 -05:00
self.buckets[idx] = None;
}
self.size = 0;
}
2012-08-22 23:01:30 -05:00
pure fn len(&const self) -> uint {
self.size
}
2012-08-22 23:01:30 -05:00
pure fn is_empty(&const self) -> bool {
self.len() == 0
}
2012-09-14 23:39:09 -05:00
pure fn contains_key(&const self,
2012-08-22 23:01:30 -05:00
k: &K) -> bool {
2012-08-06 14:34:08 -05:00
match self.bucket_for_key(self.buckets, k) {
FoundEntry(_) => {true}
TableFull | FoundHole(_) => {false}
}
}
2012-09-14 23:39:09 -05:00
pure fn find_ref(&self, k: &K) -> Option<&self/V> {
match self.bucket_for_key(self.buckets, k) {
FoundEntry(idx) => {
match self.buckets[idx] {
Some(ref bkt) => {
let ptr = unsafe {
// FIXME(#3148)--region inference
// fails to capture needed deps.
// Here, the bucket value is known to
// live as long as self, because self
// is immutable. But the region
// inference stupidly infers a
// lifetime for `ref bkt` that is
// shorter than it needs to be.
unsafe::copy_lifetime(self, &bkt.value)
};
Some(ptr)
}
None => {
fail ~"LinearMap::find: internal logic error"
}
}
}
TableFull | FoundHole(_) => {
None
}
}
}
2012-09-14 23:39:09 -05:00
pure fn get_ref(&self, k: &K) -> &self/V {
match self.find_ref(k) {
Some(v) => v,
None => fail fmt!("No entry found for key: %?", k),
}
}
2012-09-14 23:39:09 -05:00
pure fn each_ref(&self, blk: fn(k: &K, v: &V) -> bool) {
2012-07-31 16:00:17 -05:00
for vec::each(self.buckets) |slot| {
let mut broke = false;
do slot.iter |bucket| {
if !blk(&bucket.key, &bucket.value) {
broke = true; // FIXME(#3064) just write "break;"
}
}
if broke { break; }
}
}
2012-08-22 23:01:30 -05:00
2012-09-14 23:39:09 -05:00
pure fn each_key_ref(&self, blk: fn(k: &K) -> bool) {
self.each_ref(|k, _v| blk(k))
2012-07-31 16:00:17 -05:00
}
2012-08-22 23:01:30 -05:00
2012-09-14 23:39:09 -05:00
pure fn each_value_ref(&self, blk: fn(v: &V) -> bool) {
self.each_ref(|_k, v| blk(v))
}
}
impl<K:Hash IterBytes Eq, V: Copy> LinearMap<K,V> {
2012-09-14 23:39:09 -05:00
pure fn find(&const self, k: &K) -> Option<V> {
2012-08-22 23:01:30 -05:00
match self.bucket_for_key(self.buckets, k) {
FoundEntry(idx) => {
// FIXME (#3148): Once we rewrite found_entry, this
// failure case won't be necessary
match self.buckets[idx] {
Some(bkt) => {Some(copy bkt.value)}
None => fail ~"LinearMap::find: internal logic error"
}
}
TableFull | FoundHole(_) => {
None
2012-08-22 23:01:30 -05:00
}
}
}
2012-09-14 23:39:09 -05:00
pure fn get(&const self, k: &K) -> V {
2012-08-22 23:01:30 -05:00
let value = self.find(k);
if value.is_none() {
2012-08-22 19:24:52 -05:00
fail fmt!("No entry found for key: %?", k);
2012-08-22 23:01:30 -05:00
}
2012-09-11 19:17:54 -05:00
option::unwrap(move value)
2012-08-22 23:01:30 -05:00
}
}
impl<K: Hash IterBytes Eq Copy, V: Copy> LinearMap<K,V> {
2012-09-14 23:39:09 -05:00
pure fn each(&self, blk: fn(+K,+V) -> bool) {
self.each_ref(|k,v| blk(copy *k, copy *v));
}
}
impl<K: Hash IterBytes Eq Copy, V> LinearMap<K,V> {
2012-09-14 23:39:09 -05:00
pure fn each_key(&self, blk: fn(+K) -> bool) {
self.each_key_ref(|k| blk(copy *k));
}
}
impl<K: Hash IterBytes Eq, V: Copy> LinearMap<K,V> {
2012-09-14 23:39:09 -05:00
pure fn each_value(&self, blk: fn(+V) -> bool) {
self.each_value_ref(|v| blk(copy *v));
2012-07-31 16:00:17 -05:00
}
}
}
#[test]
mod test {
use linear::LinearMap;
#[test]
fn inserts() {
2012-09-07 12:41:07 -05:00
let mut m = ~LinearMap();
assert m.insert(1, 2);
assert m.insert(2, 4);
assert m.get(&1) == 2;
assert m.get(&2) == 4;
}
#[test]
fn overwrite() {
2012-09-07 12:41:07 -05:00
let mut m = ~LinearMap();
assert m.insert(1, 2);
assert m.get(&1) == 2;
assert !m.insert(1, 3);
assert m.get(&1) == 3;
}
#[test]
fn conflicts() {
let mut m = linear::linear_map_with_capacity(4);
assert m.insert(1, 2);
assert m.insert(5, 3);
assert m.insert(9, 4);
assert m.get(&9) == 4;
assert m.get(&5) == 3;
assert m.get(&1) == 2;
}
#[test]
fn conflict_remove() {
let mut m = linear::linear_map_with_capacity(4);
assert m.insert(1, 2);
assert m.insert(5, 3);
assert m.insert(9, 4);
assert m.remove(&1);
assert m.get(&9) == 4;
assert m.get(&5) == 3;
}
#[test]
fn empty() {
let mut m = linear::linear_map_with_capacity(4);
assert m.insert(1, 2);
assert !m.is_empty();
assert m.remove(&1);
assert m.is_empty();
}
#[test]
fn iterate() {
let mut m = linear::linear_map_with_capacity(4);
for uint::range(0, 32) |i| {
assert m.insert(i, i*2);
}
let mut observed = 0;
for m.each |k, v| {
assert v == k*2;
observed |= (1 << k);
}
assert observed == 0xFFFF_FFFF;
}
#[test]
fn find_ref() {
let mut m = ~LinearMap();
assert m.find_ref(&1).is_none();
m.insert(1, 2);
match m.find_ref(&1) {
None => fail,
Some(v) => assert *v == 2
}
}
2012-07-31 16:00:17 -05:00
}