我刚刚以一种通用的方式看到了优先级队列的实现,其中任何
可以将满足接口(interface)的类型放入队列。这是go的搭配方式,还是会带来任何问题?
// Copyright 2012 Stefan Nilsson
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
// Package prio provides a priority queue.
// The queue can hold elements that implement the two methods of prio.Interface.
package prio
/*
A type that implements prio.Interface can be inserted into a priority queue.
The simplest use case looks like this:
type myInt int
func (x myInt) Less(y prio.Interface) bool { return x < y.(myInt) }
func (x myInt) Index(i int) {}
To use the Remove method you need to keep track of the index of elements
in the heap, e.g. like this:
type myType struct {
value int
index int // index in heap
}
func (x *myType) Less(y prio.Interface) bool { return x.value < y.(*myType).value }
func (x *myType) Index(i int) { x.index = i }
*/
type Interface interface {
// Less returns whether this element should sort before element x.
Less(x Interface) bool
// Index is called by the priority queue when this element is moved to index i.
Index(i int)
}
// Queue represents a priority queue.
// The zero value for Queue is an empty queue ready to use.
type Queue struct {
h []Interface
}
// New returns an initialized priority queue with the given elements.
// A call of the form New(x...) uses the underlying array of x to implement
// the queue and hence might change the elements of x.
// The complexity is O(n), where n = len(x).
func New(x ...Interface) Queue {
q := Queue{x}
heapify(q.h)
return q
}
// Push pushes the element x onto the queue.
// The complexity is O(log(n)) where n = q.Len().
func (q *Queue) Push(x Interface) {
n := len(q.h)
q.h = append(q.h, x)
up(q.h, n) // x.Index(n) is done by up.
}
// Pop removes a minimum element (according to Less) from the queue and returns it.
// The complexity is O(log(n)), where n = q.Len().
func (q *Queue) Pop() Interface {
h := q.h
n := len(h) - 1
x := h[0]
h[0], h[n] = h[n], nil
h = h[:n]
if n > 0 {
down(h, 0) // h[0].Index(0) is done by down.
}
q.h = h
x.Index(-1) // for safety
return x
}
// Peek returns, but does not remove, a minimum element (according to Less) of the queue.
func (q *Queue) Peek() Interface {
return q.h[0]
}
// Remove removes the element at index i from the queue and returns it.
// The complexity is O(log(n)), where n = q.Len().
func (q *Queue) Remove(i int) Interface {
h := q.h
n := len(h) - 1
x := h[i]
h[i], h[n] = h[n], nil
h = h[:n]
if i < n {
down(h, i) // h[i].Index(i) is done by down.
up(h, i)
}
q.h = h
x.Index(-1) // for safety
return x
}
// Len returns the number of elements in the queue.
func (q *Queue) Len() int {
return len(q.h)
}
// Establishes the heap invariant in O(n) time.
func heapify(h []Interface) {
n := len(h)
for i := n - 1; i >= n/2; i-- {
h[i].Index(i)
}
for i := n/2 - 1; i >= 0; i-- { // h[i].Index(i) is done by down.
down(h, i)
}
}
// Moves element at position i towards top of heap to restore invariant.
func up(h []Interface, i int) {
for {
parent := (i - 1) / 2
if i == 0 || h[parent].Less(h[i]) {
h[i].Index(i)
break
}
h[parent], h[i] = h[i], h[parent]
h[i].Index(i)
i = parent
}
}
// Moves element at position i towards bottom of heap to restore invariant.
func down(h []Interface, i int) {
for {
n := len(h)
left := 2*i + 1
if left >= n {
h[i].Index(i)
break
}
j := left
if right := left + 1; right < n && h[right].Less(h[left]) {
j = right
}
if h[i].Less(h[j]) {
h[i].Index(i)
break
}
h[i], h[j] = h[j], h[i]
h[i].Index(i)
i = j
}
}
最佳答案
通常,此软件包不是通用的方法,但是如果您喜欢它并且满足您的需求,请使用它。我看不到任何重大问题。
与容器/堆相比,此软件包的概念是将接口(interface)放在节点上而不是容器上。容器/堆可通过使用容器来提供更大的灵活性。 (您可能已经在容器中包含节点,并且该容器甚至可能不是片。它必须是可索引的。)另一方面,您可能并不在意容器,这很常见。很高兴让包裹为您管理。这个包的索引管理是容器/堆上的一个不错的功能,尽管即使不需要索引管理,它也会增加方法调用的开销。
总会有权衡取舍。容器/堆很笼统。该程序包使用较小的方法集(2而不是5),并在顶部添加了索引管理,但是只能通过牺牲一些通用性和某些情况下的性能来实现。 (如果您真的很在意,则需要进行基准测试。还有其他差异可能会使Index调用的开销相形见)。)
关于algorithm - Go中的优先队列实现,我们在Stack Overflow上找到一个类似的问题:https://stackoverflow.com/questions/13777298/