rust/src/libcollections/priority_queue.rs

401 lines
12 KiB
Rust
Raw Normal View History

Deprecate the rev_iter pattern in all places where a DoubleEndedIterator is provided (everywhere but treemap) This commit deprecates rev_iter, mut_rev_iter, move_rev_iter everywhere (except treemap) and also deprecates related functions like rsplit, rev_components, and rev_str_components. In every case, these functions can be replaced with the non-reversed form followed by a call to .rev(). To make this more concrete, a translation table for all functional changes necessary follows: * container.rev_iter() -> container.iter().rev() * container.mut_rev_iter() -> container.mut_iter().rev() * container.move_rev_iter() -> container.move_iter().rev() * sliceorstr.rsplit(sep) -> sliceorstr.split(sep).rev() * path.rev_components() -> path.components().rev() * path.rev_str_components() -> path.str_components().rev() In terms of the type system, this change also deprecates any specialized reversed iterator types (except in treemap), opting instead to use Rev directly if any type annotations are needed. However, since methods directly returning reversed iterators are now discouraged, the need for such annotations should be small. However, in those cases, the general pattern for conversion is to take whatever follows Rev in the original reversed name and surround it with Rev<>: * RevComponents<'a> -> Rev<Components<'a>> * RevStrComponents<'a> -> Rev<StrComponents<'a>> * RevItems<'a, T> -> Rev<Items<'a, T>> * etc. The reasoning behind this change is that it makes the standard API much simpler without reducing readability, performance, or power. The presence of functions such as rev_iter adds more boilerplate code to libraries (all of which simply call .iter().rev()), clutters up the documentation, and only helps code by saving two characters. Additionally, the numerous type synonyms that were used to make the type signatures look nice like RevItems add even more boilerplate and clutter up the docs even more. With this change, all that cruft goes away. [breaking-change]
2014-04-20 23:59:12 -05:00
// Copyright 2013-2014 The Rust Project Developers. See the COPYRIGHT
2013-01-18 01:28:42 -06:00
// file at the top-level directory of this distribution and at
// http://rust-lang.org/COPYRIGHT.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.
2012-12-16 21:53:24 -06:00
//! A priority queue implemented with a binary heap
#![allow(missing_doc)]
std: Recreate a `collections` module As with the previous commit with `librand`, this commit shuffles around some `collections` code. The new state of the world is similar to that of librand: * The libcollections crate now only depends on libcore and liballoc. * The standard library has a new module, `std::collections`. All functionality of libcollections is reexported through this module. I would like to stress that this change is purely cosmetic. There are very few alterations to these primitives. There are a number of notable points about the new organization: * std::{str, slice, string, vec} all moved to libcollections. There is no reason that these primitives shouldn't be necessarily usable in a freestanding context that has allocation. These are all reexported in their usual places in the standard library. * The `hashmap`, and transitively the `lru_cache`, modules no longer reside in `libcollections`, but rather in libstd. The reason for this is because the `HashMap::new` contructor requires access to the OSRng for initially seeding the hash map. Beyond this requirement, there is no reason that the hashmap could not move to libcollections. I do, however, have a plan to move the hash map to the collections module. The `HashMap::new` function could be altered to require that the `H` hasher parameter ascribe to the `Default` trait, allowing the entire `hashmap` module to live in libcollections. The key idea would be that the default hasher would be different in libstd. Something along the lines of: // src/libstd/collections/mod.rs pub type HashMap<K, V, H = RandomizedSipHasher> = core_collections::HashMap<K, V, H>; This is not possible today because you cannot invoke static methods through type aliases. If we modified the compiler, however, to allow invocation of static methods through type aliases, then this type definition would essentially be switching the default hasher from `SipHasher` in libcollections to a libstd-defined `RandomizedSipHasher` type. This type's `Default` implementation would randomly seed the `SipHasher` instance, and otherwise perform the same as `SipHasher`. This future state doesn't seem incredibly far off, but until that time comes, the hashmap module will live in libstd to not compromise on functionality. * In preparation for the hashmap moving to libcollections, the `hash` module has moved from libstd to libcollections. A previously snapshotted commit enables a distinct `Writer` trait to live in the `hash` module which `Hash` implementations are now parameterized over. Due to using a custom trait, the `SipHasher` implementation has lost its specialized methods for writing integers. These can be re-added backwards-compatibly in the future via default methods if necessary, but the FNV hashing should satisfy much of the need for speedier hashing. A list of breaking changes: * HashMap::{get, get_mut} no longer fails with the key formatted into the error message with `{:?}`, instead, a generic message is printed. With backtraces, it should still be not-too-hard to track down errors. * The HashMap, HashSet, and LruCache types are now available through std::collections instead of the collections crate. * Manual implementations of hash should be parameterized over `hash::Writer` instead of just `Writer`. [breaking-change]
2014-05-29 20:50:12 -05:00
use core::prelude::*;
use core::mem::{zeroed, replace, swap};
use core::ptr;
std: Recreate a `collections` module As with the previous commit with `librand`, this commit shuffles around some `collections` code. The new state of the world is similar to that of librand: * The libcollections crate now only depends on libcore and liballoc. * The standard library has a new module, `std::collections`. All functionality of libcollections is reexported through this module. I would like to stress that this change is purely cosmetic. There are very few alterations to these primitives. There are a number of notable points about the new organization: * std::{str, slice, string, vec} all moved to libcollections. There is no reason that these primitives shouldn't be necessarily usable in a freestanding context that has allocation. These are all reexported in their usual places in the standard library. * The `hashmap`, and transitively the `lru_cache`, modules no longer reside in `libcollections`, but rather in libstd. The reason for this is because the `HashMap::new` contructor requires access to the OSRng for initially seeding the hash map. Beyond this requirement, there is no reason that the hashmap could not move to libcollections. I do, however, have a plan to move the hash map to the collections module. The `HashMap::new` function could be altered to require that the `H` hasher parameter ascribe to the `Default` trait, allowing the entire `hashmap` module to live in libcollections. The key idea would be that the default hasher would be different in libstd. Something along the lines of: // src/libstd/collections/mod.rs pub type HashMap<K, V, H = RandomizedSipHasher> = core_collections::HashMap<K, V, H>; This is not possible today because you cannot invoke static methods through type aliases. If we modified the compiler, however, to allow invocation of static methods through type aliases, then this type definition would essentially be switching the default hasher from `SipHasher` in libcollections to a libstd-defined `RandomizedSipHasher` type. This type's `Default` implementation would randomly seed the `SipHasher` instance, and otherwise perform the same as `SipHasher`. This future state doesn't seem incredibly far off, but until that time comes, the hashmap module will live in libstd to not compromise on functionality. * In preparation for the hashmap moving to libcollections, the `hash` module has moved from libstd to libcollections. A previously snapshotted commit enables a distinct `Writer` trait to live in the `hash` module which `Hash` implementations are now parameterized over. Due to using a custom trait, the `SipHasher` implementation has lost its specialized methods for writing integers. These can be re-added backwards-compatibly in the future via default methods if necessary, but the FNV hashing should satisfy much of the need for speedier hashing. A list of breaking changes: * HashMap::{get, get_mut} no longer fails with the key formatted into the error message with `{:?}`, instead, a generic message is printed. With backtraces, it should still be not-too-hard to track down errors. * The HashMap, HashSet, and LruCache types are now available through std::collections instead of the collections crate. * Manual implementations of hash should be parameterized over `hash::Writer` instead of just `Writer`. [breaking-change]
2014-05-29 20:50:12 -05:00
use slice;
use vec::Vec;
/// A priority queue implemented with a binary heap
2013-07-11 14:05:17 -05:00
#[deriving(Clone)]
pub struct PriorityQueue<T> {
data: Vec<T>,
}
impl<T: Ord> Container for PriorityQueue<T> {
2013-01-21 20:59:19 -06:00
/// Returns the length of the queue
2013-06-15 01:20:06 -05:00
fn len(&self) -> uint { self.data.len() }
2013-01-21 20:59:19 -06:00
}
impl<T: Ord> Mutable for PriorityQueue<T> {
/// Drop all items from the queue
fn clear(&mut self) { self.data.truncate(0) }
}
impl<T: Ord> PriorityQueue<T> {
/// An iterator visiting all values in underlying vector, in
/// arbitrary order.
pub fn iter<'a>(&'a self) -> Items<'a, T> {
Items { iter: self.data.iter() }
}
/// Returns the greatest item in a queue or None if it is empty
pub fn top<'a>(&'a self) -> Option<&'a T> {
if self.is_empty() { None } else { Some(self.data.get(0)) }
}
#[deprecated="renamed to `top`"]
pub fn maybe_top<'a>(&'a self) -> Option<&'a T> { self.top() }
/// Returns the number of elements the queue can hold without reallocating
pub fn capacity(&self) -> uint { self.data.capacity() }
/// Reserve capacity for exactly n elements in the PriorityQueue.
/// Do nothing if the capacity is already sufficient.
pub fn reserve_exact(&mut self, n: uint) { self.data.reserve_exact(n) }
/// Reserve capacity for at least n elements in the PriorityQueue.
/// Do nothing if the capacity is already sufficient.
pub fn reserve(&mut self, n: uint) {
self.data.reserve(n)
}
/// Remove the greatest item from a queue and return it, or `None` if it is
/// empty.
pub fn pop(&mut self) -> Option<T> {
match self.data.pop() {
None => { None }
Some(mut item) => {
if !self.is_empty() {
swap(&mut item, self.data.get_mut(0));
self.siftdown(0);
}
Some(item)
}
2013-05-05 23:42:54 -05:00
}
}
#[deprecated="renamed to `pop`"]
pub fn maybe_pop(&mut self) -> Option<T> { self.pop() }
/// Push an item onto the queue
pub fn push(&mut self, item: T) {
self.data.push(item);
let new_len = self.len() - 1;
self.siftup(0, new_len);
}
/// Optimized version of a push followed by a pop
pub fn push_pop(&mut self, mut item: T) -> T {
if !self.is_empty() && *self.top().unwrap() > item {
swap(&mut item, self.data.get_mut(0));
self.siftdown(0);
}
item
}
/// Optimized version of a pop followed by a push. The push is done
/// regardless of whether the queue is empty.
pub fn replace(&mut self, mut item: T) -> Option<T> {
if !self.is_empty() {
swap(&mut item, self.data.get_mut(0));
self.siftdown(0);
Some(item)
} else {
self.push(item);
None
}
}
2014-05-18 18:33:30 -05:00
#[allow(dead_code)]
2014-05-18 17:04:43 -05:00
#[deprecated="renamed to `into_vec`"]
fn to_vec(self) -> Vec<T> { self.into_vec() }
2014-05-18 18:33:30 -05:00
#[allow(dead_code)]
2014-05-18 17:04:43 -05:00
#[deprecated="renamed to `into_sorted_vec`"]
fn to_sorted_vec(self) -> Vec<T> { self.into_sorted_vec() }
/// Consume the PriorityQueue and return the underlying vector
2014-05-18 17:04:43 -05:00
pub fn into_vec(self) -> Vec<T> { let PriorityQueue{data: v} = self; v }
2012-12-16 21:56:09 -06:00
/// Consume the PriorityQueue and return a vector in sorted
/// (ascending) order
2014-05-18 17:04:43 -05:00
pub fn into_sorted_vec(self) -> Vec<T> {
let mut q = self;
let mut end = q.len();
while end > 1 {
end -= 1;
q.data.as_mut_slice().swap(0, end);
2013-03-22 17:07:09 -05:00
q.siftdown_range(0, end)
}
q.into_vec()
}
2012-12-17 17:08:59 -06:00
/// Create an empty PriorityQueue
pub fn new() -> PriorityQueue<T> { PriorityQueue{data: vec!(),} }
2012-12-17 17:08:59 -06:00
/// Create an empty PriorityQueue with capacity `capacity`
pub fn with_capacity(capacity: uint) -> PriorityQueue<T> {
PriorityQueue { data: Vec::with_capacity(capacity) }
}
/// Create a PriorityQueue from a vector (heapify)
pub fn from_vec(xs: Vec<T>) -> PriorityQueue<T> {
let mut q = PriorityQueue{data: xs,};
let mut n = q.len() / 2;
while n > 0 {
n -= 1;
2013-03-22 17:07:09 -05:00
q.siftdown(n)
}
q
}
2012-12-16 21:56:09 -06:00
// The implementations of siftup and siftdown use unsafe blocks in
// order to move an element out of the vector (leaving behind a
// zeroed element), shift along the others and move it back into the
2012-12-16 21:56:09 -06:00
// vector over the junk element. This reduces the constant factor
// compared to using swaps, which involves twice as many moves.
fn siftup(&mut self, start: uint, mut pos: uint) {
unsafe {
let new = replace(self.data.get_mut(pos), zeroed());
while pos > start {
let parent = (pos - 1) >> 1;
if new > *self.data.get(parent) {
let x = replace(self.data.get_mut(parent), zeroed());
ptr::write(self.data.get_mut(pos), x);
pos = parent;
continue
}
break
}
ptr::write(self.data.get_mut(pos), new);
}
}
fn siftdown_range(&mut self, mut pos: uint, end: uint) {
unsafe {
let start = pos;
let new = replace(self.data.get_mut(pos), zeroed());
let mut child = 2 * pos + 1;
while child < end {
let right = child + 1;
if right < end && !(*self.data.get(child) > *self.data.get(right)) {
child = right;
}
let x = replace(self.data.get_mut(child), zeroed());
ptr::write(self.data.get_mut(pos), x);
pos = child;
child = 2 * pos + 1;
}
ptr::write(self.data.get_mut(pos), new);
self.siftup(start, pos);
}
}
fn siftdown(&mut self, pos: uint) {
let len = self.len();
self.siftdown_range(pos, len);
}
}
/// PriorityQueue iterator
pub struct Items <'a, T> {
iter: slice::Items<'a, T>,
}
impl<'a, T> Iterator<&'a T> for Items<'a, T> {
#[inline]
fn next(&mut self) -> Option<(&'a T)> { self.iter.next() }
#[inline]
fn size_hint(&self) -> (uint, Option<uint>) { self.iter.size_hint() }
}
impl<T: Ord> FromIterator<T> for PriorityQueue<T> {
fn from_iter<Iter: Iterator<T>>(iter: Iter) -> PriorityQueue<T> {
2013-07-29 19:06:49 -05:00
let mut q = PriorityQueue::new();
q.extend(iter);
q
}
}
impl<T: Ord> Extendable<T> for PriorityQueue<T> {
fn extend<Iter: Iterator<T>>(&mut self, mut iter: Iter) {
let (lower, _) = iter.size_hint();
2013-07-29 19:06:49 -05:00
let len = self.capacity();
self.reserve(len + lower);
for elem in iter {
2013-07-29 19:06:49 -05:00
self.push(elem);
}
}
}
#[cfg(test)]
mod tests {
use std::prelude::*;
use priority_queue::PriorityQueue;
use vec::Vec;
#[test]
fn test_iterator() {
let data = vec!(5, 9, 3);
let iterout = [9, 5, 3];
let pq = PriorityQueue::from_vec(data);
let mut i = 0;
for el in pq.iter() {
assert_eq!(*el, iterout[i]);
i += 1;
}
}
#[test]
fn test_top_and_pop() {
let data = vec!(2u, 4, 6, 2, 1, 8, 10, 3, 5, 7, 0, 9, 1);
let mut sorted = data.clone();
sorted.sort();
let mut heap = PriorityQueue::from_vec(data);
while !heap.is_empty() {
assert_eq!(heap.top().unwrap(), sorted.last().unwrap());
assert_eq!(heap.pop().unwrap(), sorted.pop().unwrap());
}
}
#[test]
fn test_push() {
let mut heap = PriorityQueue::from_vec(vec!(2, 4, 9));
assert_eq!(heap.len(), 3);
assert!(*heap.top().unwrap() == 9);
heap.push(11);
assert_eq!(heap.len(), 4);
assert!(*heap.top().unwrap() == 11);
heap.push(5);
assert_eq!(heap.len(), 5);
assert!(*heap.top().unwrap() == 11);
heap.push(27);
assert_eq!(heap.len(), 6);
assert!(*heap.top().unwrap() == 27);
heap.push(3);
assert_eq!(heap.len(), 7);
assert!(*heap.top().unwrap() == 27);
heap.push(103);
assert_eq!(heap.len(), 8);
assert!(*heap.top().unwrap() == 103);
}
#[test]
fn test_push_unique() {
let mut heap = PriorityQueue::from_vec(vec!(box 2, box 4, box 9));
assert_eq!(heap.len(), 3);
assert!(*heap.top().unwrap() == box 9);
2014-04-25 03:08:02 -05:00
heap.push(box 11);
assert_eq!(heap.len(), 4);
assert!(*heap.top().unwrap() == box 11);
2014-04-25 03:08:02 -05:00
heap.push(box 5);
assert_eq!(heap.len(), 5);
assert!(*heap.top().unwrap() == box 11);
2014-04-25 03:08:02 -05:00
heap.push(box 27);
assert_eq!(heap.len(), 6);
assert!(*heap.top().unwrap() == box 27);
2014-04-25 03:08:02 -05:00
heap.push(box 3);
assert_eq!(heap.len(), 7);
assert!(*heap.top().unwrap() == box 27);
2014-04-25 03:08:02 -05:00
heap.push(box 103);
assert_eq!(heap.len(), 8);
assert!(*heap.top().unwrap() == box 103);
}
#[test]
fn test_push_pop() {
let mut heap = PriorityQueue::from_vec(vec!(5, 5, 2, 1, 3));
assert_eq!(heap.len(), 5);
assert_eq!(heap.push_pop(6), 6);
assert_eq!(heap.len(), 5);
assert_eq!(heap.push_pop(0), 5);
assert_eq!(heap.len(), 5);
assert_eq!(heap.push_pop(4), 5);
assert_eq!(heap.len(), 5);
assert_eq!(heap.push_pop(1), 4);
assert_eq!(heap.len(), 5);
}
#[test]
fn test_replace() {
let mut heap = PriorityQueue::from_vec(vec!(5, 5, 2, 1, 3));
assert_eq!(heap.len(), 5);
assert_eq!(heap.replace(6).unwrap(), 5);
assert_eq!(heap.len(), 5);
assert_eq!(heap.replace(0).unwrap(), 6);
assert_eq!(heap.len(), 5);
assert_eq!(heap.replace(4).unwrap(), 5);
assert_eq!(heap.len(), 5);
assert_eq!(heap.replace(1).unwrap(), 4);
assert_eq!(heap.len(), 5);
}
fn check_to_vec(mut data: Vec<int>) {
2013-07-02 14:47:32 -05:00
let heap = PriorityQueue::from_vec(data.clone());
let mut v = heap.clone().into_vec();
v.sort();
data.sort();
assert_eq!(v.as_slice(), data.as_slice());
assert_eq!(heap.into_sorted_vec().as_slice(), data.as_slice());
}
#[test]
fn test_to_vec() {
check_to_vec(vec!());
check_to_vec(vec!(5));
check_to_vec(vec!(3, 2));
check_to_vec(vec!(2, 3));
check_to_vec(vec!(5, 1, 2));
check_to_vec(vec!(1, 100, 2, 3));
check_to_vec(vec!(1, 3, 5, 7, 9, 2, 4, 6, 8, 0));
check_to_vec(vec!(2, 4, 6, 2, 1, 8, 10, 3, 5, 7, 0, 9, 1));
check_to_vec(vec!(9, 11, 9, 9, 9, 9, 11, 2, 3, 4, 11, 9, 0, 0, 0, 0));
check_to_vec(vec!(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10));
check_to_vec(vec!(10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0));
check_to_vec(vec!(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 0, 0, 0, 1, 2));
check_to_vec(vec!(5, 4, 3, 2, 1, 5, 4, 3, 2, 1, 5, 4, 3, 2, 1));
}
#[test]
fn test_empty_pop() {
let mut heap: PriorityQueue<int> = PriorityQueue::new();
assert!(heap.pop().is_none());
}
#[test]
fn test_empty_top() {
let empty: PriorityQueue<int> = PriorityQueue::new();
assert!(empty.top().is_none());
}
#[test]
fn test_empty_replace() {
let mut heap: PriorityQueue<int> = PriorityQueue::new();
heap.replace(5).is_none();
}
#[test]
fn test_from_iter() {
let xs = vec!(9u, 8, 7, 6, 5, 4, 3, 2, 1);
Deprecate the rev_iter pattern in all places where a DoubleEndedIterator is provided (everywhere but treemap) This commit deprecates rev_iter, mut_rev_iter, move_rev_iter everywhere (except treemap) and also deprecates related functions like rsplit, rev_components, and rev_str_components. In every case, these functions can be replaced with the non-reversed form followed by a call to .rev(). To make this more concrete, a translation table for all functional changes necessary follows: * container.rev_iter() -> container.iter().rev() * container.mut_rev_iter() -> container.mut_iter().rev() * container.move_rev_iter() -> container.move_iter().rev() * sliceorstr.rsplit(sep) -> sliceorstr.split(sep).rev() * path.rev_components() -> path.components().rev() * path.rev_str_components() -> path.str_components().rev() In terms of the type system, this change also deprecates any specialized reversed iterator types (except in treemap), opting instead to use Rev directly if any type annotations are needed. However, since methods directly returning reversed iterators are now discouraged, the need for such annotations should be small. However, in those cases, the general pattern for conversion is to take whatever follows Rev in the original reversed name and surround it with Rev<>: * RevComponents<'a> -> Rev<Components<'a>> * RevStrComponents<'a> -> Rev<StrComponents<'a>> * RevItems<'a, T> -> Rev<Items<'a, T>> * etc. The reasoning behind this change is that it makes the standard API much simpler without reducing readability, performance, or power. The presence of functions such as rev_iter adds more boilerplate code to libraries (all of which simply call .iter().rev()), clutters up the documentation, and only helps code by saving two characters. Additionally, the numerous type synonyms that were used to make the type signatures look nice like RevItems add even more boilerplate and clutter up the docs even more. With this change, all that cruft goes away. [breaking-change]
2014-04-20 23:59:12 -05:00
let mut q: PriorityQueue<uint> = xs.as_slice().iter().rev().map(|&x| x).collect();
for &x in xs.iter() {
assert_eq!(q.pop().unwrap(), x);
}
}
}