340 lines
9.4 KiB
JavaScript
340 lines
9.4 KiB
JavaScript
/****
|
||
* 带碰撞处理的Hash表
|
||
* 实际上在js中,单独实现一个Hash表感觉不是很有实用价值
|
||
* 如果需要通常是直接将Object,Map,Set来当Hash表用
|
||
*
|
||
* 总结：
|
||
* 我写的这个实现把store 从Object换成Array不会有运行性能上的区别
|
||
* 把hash函数改成生成一定范围的值的类型,然后初始化一个指定长度的数组因该会有一定的性能提升
|
||
* 把store换成Map，然后修改相关实现会获得飞越性的提升，因为在js中Map的实现对这种类型的操作做了优化
|
||
*/
|
||
class HashTable {
|
||
constructor() {
|
||
//创建一个没有原型链的对象
|
||
this.store = Object.create(null);
|
||
}
|
||
/**
|
||
* Donald E. Knuth在“计算机编程艺术第3卷”中提出的算法，主题是排序和搜索第6.4章。
|
||
* @param {*} string
|
||
* 翻译自别的语言的实现
|
||
* 需要注意的是由于js中没有int类型，number是dobule的标准实现
|
||
* 所以返回前的位运算实际和本来的设想不一致，也就是同样的实现，在别的语言中返回可能不同
|
||
*/
|
||
hash(string) {
|
||
let len = string.length;
|
||
let hash = len;
|
||
for (let i = 0; i < len; i++) {
|
||
hash = ((hash << 5) ^ (hash >> 27)) ^ string.charCodeAt(i);
|
||
}
|
||
return hash & 0x7FFFFFFF;
|
||
}
|
||
|
||
isCresh(item) {
|
||
return Object.prototype.toString.call(item) === "[object Map]"
|
||
}
|
||
/**
|
||
* 约定item必须要有key
|
||
* @param {*} item
|
||
*/
|
||
put(item) {
|
||
if (typeof item.key !== 'string') {
|
||
throw 'item must have key!'
|
||
}
|
||
let hash = this.hash(item.key);
|
||
//碰撞处理
|
||
let cresh = this.store[hash];
|
||
if (cresh) {
|
||
if (cresh.key === item.key) {
|
||
this.store[hash] = item;
|
||
return
|
||
}
|
||
|
||
if (!this.isCresh(cresh)) {
|
||
this.store[hash] = new Map();
|
||
}
|
||
this.store[hash].set(item.key, item);
|
||
} else {
|
||
this.store[hash] = item;
|
||
}
|
||
}
|
||
get(key) {
|
||
let hash = this.hash(key);
|
||
let value = this.store[hash] || null;
|
||
if (this.isCresh(value)) {
|
||
return value.get(key);
|
||
} else {
|
||
return value
|
||
}
|
||
}
|
||
remove(key) {
|
||
let hash = this.hash(key);
|
||
let value = this.store[hash];
|
||
if (!value) {
|
||
return null;
|
||
}
|
||
if (this.isCresh(value)) {
|
||
value.delete(key);
|
||
} else {
|
||
delete this.store[hash];
|
||
}
|
||
}
|
||
clear() {
|
||
this.store = {};
|
||
}
|
||
print() {
|
||
let values = Object.values(this.store);
|
||
values.forEach(element => {
|
||
if (this.isCresh(element)) {
|
||
element.forEach(item => {
|
||
console.log(item);
|
||
});
|
||
} else {
|
||
console.log(element)
|
||
}
|
||
});
|
||
}
|
||
}
|
||
/**
|
||
* 相比使用Object和Array做store 运行时的性能提升了三分之一
|
||
* 但当前这种用法没有直接使用Map方便，而且直接使用Map会快的多
|
||
*/
|
||
class HashTableBaseMap {
|
||
constructor() {
|
||
this.store = new Map();
|
||
}
|
||
/**
|
||
* Donald E. Knuth在“计算机编程艺术第3卷”中提出的算法，主题是排序和搜索第6.4章。
|
||
* @param {*} string
|
||
* 翻译自别的语言的实现
|
||
* 需要注意的是由于js中没有int类型，number是dobule的标准实现
|
||
* 所以返回前的位运算实际和本来的设想不一致，也就是同样的实现，在别的语言中返回可能不同
|
||
*/
|
||
hash(string) {
|
||
let len = string.length;
|
||
let hash = len;
|
||
for (let i = 0; i < len; i++) {
|
||
hash = ((hash << 5) ^ (hash >> 27)) ^ string.charCodeAt(i);
|
||
}
|
||
return hash & 0x7FFFFFFF;
|
||
}
|
||
|
||
isCresh(item) {
|
||
return Object.prototype.toString.call(item) === "[object Map]"
|
||
}
|
||
/**
|
||
* 约定item必须要有key
|
||
* @param {*} item
|
||
*/
|
||
put(item) {
|
||
if (typeof item.key !== 'string') {
|
||
throw 'item must have key!'
|
||
}
|
||
let hash = this.hash(item.key);
|
||
//碰撞处理
|
||
let cresh = this.store.get(hash);
|
||
if (cresh) {
|
||
if (cresh.key === item.key) {
|
||
this.store.set(hash, item);
|
||
return
|
||
}
|
||
|
||
if (!this.isCresh(cresh)) {
|
||
this.store[hash] = new Map();
|
||
}
|
||
this.store[hash].set(item.key, item);
|
||
} else {
|
||
this.store.set(hash, item);
|
||
}
|
||
}
|
||
get(key) {
|
||
let hash = this.hash(key);
|
||
let value = this.store.get(hash);
|
||
if (this.isCresh(value)) {
|
||
return value.get(key);
|
||
} else {
|
||
return value
|
||
}
|
||
}
|
||
remove(key) {
|
||
let hash = this.hash(key);
|
||
let value = this.store.get(hash);
|
||
if (!value) {
|
||
return null;
|
||
}
|
||
if (this.isCresh(value)) {
|
||
value.delete(key);
|
||
} else {
|
||
this.store.delete(hash)
|
||
}
|
||
}
|
||
clear() {
|
||
this.store = {};
|
||
}
|
||
print() {
|
||
this.store.forEach(element => {
|
||
if (this.isCresh(element)) {
|
||
element.forEach(item => {
|
||
console.log(item);
|
||
});
|
||
} else {
|
||
console.log(element)
|
||
}
|
||
});
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 基础测试
|
||
*/
|
||
function baseTest() {
|
||
let hashTable = new HashTable();
|
||
for (let i = 0; i < 10; i++) {
|
||
hashTable.put({
|
||
key: 'test' + i,
|
||
value: 'some value' + i
|
||
});
|
||
}
|
||
console.log('step1:')
|
||
//随机获取5次
|
||
for (let j = 0; j < 5; j++) {
|
||
let key = 'test' + Math.floor(Math.random() * 10);
|
||
console.log(key);
|
||
console.log(hashTable.get(key))
|
||
}
|
||
//获得一次空值
|
||
console.log('get null:', hashTable.get('test10'))
|
||
//修改一次值
|
||
hashTable.put({
|
||
key: 'test1',
|
||
value: 'change'
|
||
});
|
||
//删除一次值
|
||
hashTable.remove('test2');
|
||
console.log('step2:')
|
||
//输出修改后所有的
|
||
hashTable.print();
|
||
}
|
||
|
||
/**
|
||
* 有序key存取，性能测试
|
||
*/
|
||
function ordKeyTest() {
|
||
let length = 1000000;
|
||
console.time('create')
|
||
let hashTable = new HashTable();
|
||
for (let i = 0; i < length; i++) {
|
||
//24位长度有序key
|
||
hashTable.put({
|
||
key: 'someTestSoSoSoSoLongKey' + i,
|
||
value: 'some value' + i
|
||
});
|
||
}
|
||
console.timeEnd('create')
|
||
|
||
let get = 100000;
|
||
console.time('get')
|
||
for (let j = 0; j < get; j++) {
|
||
let key = 'test' + Math.floor(Math.random() * 999999);
|
||
hashTable.get(key)
|
||
}
|
||
console.timeEnd('get')
|
||
}
|
||
|
||
/**
|
||
* 无序key性能测试
|
||
* 这个查找稍微有点不准，会有一定量随机字符串重复
|
||
* 实际结果，创建没有区别，大数据量下由于无序key有一些会碰撞，get的总体用的时间会多不少。
|
||
*/
|
||
function randKeyTest() {
|
||
let length = 1000000;
|
||
let keyList = [];
|
||
for (let i = 0; i < length; i++) {
|
||
keyList.push(randomString());
|
||
}
|
||
console.time('create')
|
||
let hashTable = new HashTable();
|
||
for (let i = 0; i < length; i++) {
|
||
hashTable.put({
|
||
key: keyList[i],
|
||
value: 'some value' + i
|
||
});
|
||
}
|
||
console.timeEnd('create')
|
||
let get = 100000;
|
||
console.time('get')
|
||
for (let j = 0; j < get; j++) {
|
||
let key = keyList[Math.floor(Math.random() * 999999)];
|
||
hashTable.get(key)
|
||
}
|
||
console.timeEnd('get')
|
||
}
|
||
|
||
/**
|
||
* 直接使用Object的性能测试
|
||
* 有序就不测了，估计不会有区别，只看不使用hash的无序key
|
||
* 结果:想达到同样的结果创建会比hash后的慢接近四分之三,获取用时差不多
|
||
*/
|
||
function randKeyTestFromObj() {
|
||
let length = 1000000;
|
||
let keyList = [];
|
||
for (let i = 0; i < length; i++) {
|
||
keyList.push(randomString());
|
||
}
|
||
console.time('create')
|
||
let hashTable = {};
|
||
for (let i = 0; i < length; i++) {
|
||
let key = keyList[i];
|
||
hashTable[key] = {
|
||
key: key,
|
||
value: 'some value' + i
|
||
}
|
||
}
|
||
console.timeEnd('create')
|
||
let get = 100000;
|
||
console.time('get')
|
||
for (let j = 0; j < get; j++) {
|
||
let key = keyList[Math.floor(Math.random() * 999999)];
|
||
hashTable[key]
|
||
}
|
||
console.timeEnd('get')
|
||
}
|
||
/**
|
||
* 直接使用Map的性能测试
|
||
* 结果：创建用时差不多，但是获取快了一个数量级(十倍不止)
|
||
*/
|
||
function randKeyTestFromMap() {
|
||
let length = 1000000;
|
||
let keyList = [];
|
||
for (let i = 0; i < length; i++) {
|
||
keyList.push(randomString());
|
||
}
|
||
console.time('create')
|
||
let hashTable = new Map();
|
||
for (let i = 0; i < length; i++) {
|
||
let key = keyList[i];
|
||
hashTable.set(key, {
|
||
key: key,
|
||
value: 'some value' + i
|
||
})
|
||
}
|
||
console.timeEnd('create')
|
||
let get = 100000;
|
||
console.time('get')
|
||
for (let j = 0; j < get; j++) {
|
||
let key = keyList[Math.floor(Math.random() * 999999)];
|
||
hashTable.get(key);
|
||
}
|
||
console.timeEnd('get')
|
||
}
|
||
|
||
//生成指定长度的字符串
|
||
function randomString(len) {
|
||
len = len || 24;
|
||
var chars = 'ABCDEFGHJKMNPQRSTWXYZabcdefhijkmnprstwxyz2345678';
|
||
var maxPos = chars.length;
|
||
var pwd = '';
|
||
for (i = 0; i < len; i++) {
|
||
pwd += chars.charAt(Math.floor(Math.random() * maxPos));
|
||
}
|
||
return pwd;
|
||
} |