Promise的诞生与Javascript中异步编程息息相关，js中异步编程主要指的是setTimout/setInterval、DOM事件机制、ajax，通过传入回调函数实现控制反转。异步编程为js带来强大灵活性的同时，也带来了嵌套回调的问题。详细来说主要有两点，第一嵌套太深代码可读性太差，第二并行逻辑必须串行执行。
request = function(url, cb, eb) {
var xhr = new XMLHttpRequest();
xhr.onreadystatechange = function() {
if (xhr.readyState === 4) {
if ((xhr.status >=200 && xhr.status < 300) || xhr.status === 304) {
cb(xhr.responseText);
} else {
eb(new Error({
message: xhr.status
}));
}
}
};
xhr.open('get', url, true);
xhr.send(null);
}
这个例子中程序要依次处理data1、data2、data3，嵌套太多可读性太差
request = function(url) {
var def = new Deferred();
var xhr = new XMLHttpRequest();
xhr.onreadystatechange = function() {
if (xhr.readyState === 4) {
if ((xhr.status >=200 && xhr.status < 300) || xhr.status === 304) {
def.resolve(xhr.responseText)
} else {//简化ajax，没有提供错误回调
def.reject(new Error({
message: xhr.status
}));
}
}
};
xhr.open('get', url, true);
xhr.send(null);
return def.promise;
}
request('data1.json').then(function(data1) {
console.log(data1);//处理data1
return request('data2.json');
}).then(function(data2) {
console.log(data2);//处理data2
return request('data3.json');
}, function(err) {
console.error(err);
}).then(function(data3) {
console.log(data3);
alert('success');
}, function(err) {
console.error(err);
});
这个例子中程序需要请求data1、data2、data3数据，得到三个数据后才进行下一步处理。数据并不需要串行请求，但我们的代码却需要串行执行，增加了等待时间。
//并行逻辑串行执行
request('data1', function(data1) {
request('data2', function(data2) {
request('data3', function(data3) {
console.log(data1, data2, data3);//处理全部数据
alert('success');
}, function(err) {
console.error(err);
});
}, function(err) {
console.error(err);
});
}, function(err) {
console.error(err);
});
Promise机制
Promise机制便是上述问题的一种解决方案。与他相关的规范有PromiseA和PromiseA+，PromiseA中对Promise进行了整体描述，PromiseA+对A进行了补充，在then函数的行为方面进行了更加详尽的阐述。
PromiseA+规范
A promise represents the eventual result of an asynchronous operation.
一个promise代表了一个异步操作的最终结果
The primary way of interacting with a promise is through its
then method, which registers callbacks to receive either a promise’s eventual value or the reason why the promise cannot be fulfilled.跟promise交互的主要方式是通过他的then方法来注册回调函数去接收promise的最终结果值或者是promise不能完成的原因。我们可以简单总结一下规范。每个promise都有三个状态：pending(默认)、fulfilled(完成)、rejected(失败)；默认状态可以转变为完成态或失败态，完成态与失败态之间无法相互转换，转变的过程是不可逆的，转变一旦完成promise对象就不能被修改。通过promise提供的then函数注册onFulfill(成功回调)、onReject(失败回调)、onProgres(进度回调)来与promise交互。Then函数返回一个promise对象（称为promise2，前者成为promise1），promise2受promise1状态的影响，具体请查看A+规范。
上两个规范中并没有说明promise的状态如何改变，大部分前端框架中使用Deferred来改变promise的状态（resolve()、reject（））。二者关系请看下图。
这里根据规范，我们实现一下promise
Promise = function() {
this.queue = [];
this.value = null;
this.status = 'pending';// pending fulfilled rejected
};
Promise.prototype.getQueue = function() {
return this.queue;
};
Promise.prototype.getStatus = function() {
return this.status;
};
Promise.prototype.setStatus = function(s, value) {
if (s === 'fulfilled' || s === 'rejected') {
this.status = s;
this.value = value || null;
this.queue = [];
var freezeObject = Object.freeze || function(){};
freezeObject(this);// promise的状态是不可逆的
} else {
throw new Error({
message: "doesn't support status: " + s
});
}
};
Promise.prototype.isFulfilled = function() {
return this.status === 'fulfilled';
};
Promise.prototype.isRejected = function() {
return this.status === 'rejected';
}
Promise.prototype.isPending = function() {
return this.status === 'pending';
}
Promise.prototype.then = function(onFulfilled, onRejected) {
var handler = {
'fulfilled': onFulfilled,
'rejected': onRejected
};
handler.deferred = new Deferred();
if (!this.isPending()) {//这里允许先改变promise状态后添加回调
utils.procedure(this.status, handler, this.value);
} else {
this.queue.push(handler);//then may be called multiple times on the same promise;规范2.2.6
}
return handler.deferred.promise;//then must return a promise;规范2.2.7
};
var utils = (function(){
var makeSignaler = function(deferred, type) {
return function(result) {
transition(deferred, type, result);
}
};
var procedure = function(type, handler, result) {
var func = handler[type];
var def = handler.deferred;
if (func) {
try {
var newResult = func(result);
if (newResult && typeof newResult.then === 'function') {//thenable
// 此种写法存在闭包容易造成内存泄露，我们通过高阶函数解决
// newResult.then(function(data) {
// def.resolve(data);
// }, function(err) {
// def.reject(err);
// });
//PromiseA+规范，x代表newResult，promise代表def.promise
//If x is a promise, adopt its state [3.4]:
//If x is pending, promise must remain pending until x is fulfilled or rejected.
//If/when x is fulfilled, fulfill promise with the same value.
//If/when x is rejected, reject promise with the same reason.
newResult.then(makeSignaler(def, 'fulfilled'), makeSignaler(def, 'rejected'));//此处的本质是利用了异步闭包
} else {
transition(def, type, newResult);
}
} catch(err) {
transition(def, 'rejected', err);
}
} else {
transition(def, type, result);
}
};
var transition = function(deferred, type, result) {
if (type === 'fulfilled') {
deferred.resolve(result);
} else if (type === 'rejected') {
deferred.reject(result);
} else if (type !== 'pending') {
throw new Error({
'message': "doesn't support type: " + type
});
}
};
return {
'procedure': procedure
}
})();
Deferred = function() {
this.promise = new Promise();
};
Deferred.prototype.resolve = function(result) {
if (!this.promise.isPending()) {
return;
}
var queue = this.promise.getQueue();
for (var i = 0, len = queue.length; i < len; i++) {
utils.procedure('fulfilled', queue[i], result);
}
this.promise.setStatus('fulfilled', result);
};
Deferred.prototype.reject = function(err) {
if (!this.promise.isPending()) {
return;
}
var queue = this.promise.getQueue();
for (var i = 0, len = queue.length; i < len; i++) {
utils.procedure('rejected', queue[i], err);
}
this.promise.setStatus('rejected', err);
}
通过Promise机制我们的编程方式可以变成这样：
request = function(url) {
var def = new Deferred();
var xhr = new XMLHttpRequest();
xhr.onreadystatechange = function() {
if (xhr.readyState === 4) {
if ((xhr.status >=200 && xhr.status < 300) || xhr.status === 304) {
def.resolve(xhr.responseText)
} else {//简化ajax，没有提供错误回调
def.reject(new Error({
message: xhr.status
}));
}
}
};
xhr.open('get', url, true);
xhr.send(null);
return def.promise;
}
request('data1.json').then(function(data1) {
console.log(data1);//处理data1
return request('data2.json');
}).then(function(data2) {
console.log(data2);//处理data2
return request('data3.json');
}, function(err) {
console.error(err);
}).then(function(data3) {
console.log(data3);
alert('success');
}, function(err) {
console.error(err);
});
对于并行逻辑串行执行问题我们可以这样解决
//所有异步操作都完成时，进入完成态，
//其中一项异步操作失败则进入失败态
all = function(requestArray) {
// var some = Array.prototype.some;
var def = new Deferred();
var results = [];
var total = 0;
requestArray.some(function(r, idx) {
//为数组中每一项注册回调函数
r.then(function(data) {
if (def.promise.isPending()) {
total++;
results[idx] = data;
if (total === requestArray.length) {
def.resolve(results);
}
}
}, function(err) {
def.reject(err);
});
//如果不是等待状态则停止，比如requestArray[0]失败的话，剩下数组则不用继续注册
return !def.promise.isPending();
});
return def.promise;
}
all(
[request('data1.json'),
request('data2.json'),
request('data3.json')]
).then(
function(results){
console.log(results);// 处理data1,data2,data3
alert('success');
}, function(err) {
console.error(err);
});
以下是几个测试案例
//链式调用
var p1 = new Deferred();
p1.promise.then(function(result) {
console.log('resolve: ', result);
return result;
}, function(err) {
console.log('reject: ', err);
return err;
}).then(function(result) {
console.log('resolve2: ', result);
return result;
}, function(err) {
console.log('reject2: ', err);
return err;
}).then(function(result) {
console.log('resolve3: ', result);
return result;
}, function(err) {
console.log('reject3: ', err);
return err;
});
p1.resolve('success');
//p1.reject('failed');
p1.promise.then(function(result) {
console.log('after resolve: ', result);
return result;
}, function(err) {
console.log('after reject: ', err);
return err;
}).then(function(result) {
console.log('after resolve2: ', result);
return result;
}, function(err) {
console.log('after reject2: ', err);
return err;
}).then(function(result) {
console.log('after resolve2: ', result);
return result;
}, function(err) {
console.log('after reject2: ', err);
return err;
});
//串行异步
var p2 = new Deferred();
p2.promise.then(function(result) {
var def = new Deferred();
setTimeout(function(){
console.log('resolve: ', result);
def.resolve(result);
})
return def.promise;
}, function(err) {
console.log('reject: ', err);
return err;
}).then(function(result) {
var def = new Deferred();
setTimeout(function(){
console.log('resolve2: ', result);
def.reject(result);
})
return def.promise;
}, function(err) {
console.log('reject2: ', err);
return err;
}).then(function(result) {
console.log('resolve3: ', result);
return result;
}, function(err) {
console.log('reject3: ', err);
return err;
});
p2.resolve('success');
//并行异步
var p1 = function(){
var def = new Deferred();
setTimeout(function() {
console.log('p1 success');
def.resolve('p1 success');
}, 20);
return def.promise;
}
var p2 = function(){
var def = new Deferred();
setTimeout(function() {
console.log('p2 failed');
def.reject('p2 failed');
}, 10);
return def.promise;
}
var p3 = function(){
var def = new Deferred();
setTimeout(function() {
console.log('p3 success');
def.resolve('p3 success');
}, 15);
return def.promise;
}
all([p1(), p2(), p3()]).then(function(results) {
console.log(results);
}, function(err) {
console.error(err);
});
Promise优点
对比使用Promise前后我们可以发现，传统异步编程通过嵌套回调函数的方式，等待异步操作结束后再执行下一步操作。过多的嵌套导致意大利面条式的代码，可读性差、耦合度高、扩展性低。通过Promise机制，扁平化的代码机构，大大提高了代码可读性；用同步编程的方式来编写异步代码，保存线性的代码逻辑，极大的降低了代码耦合性而提高了程序的可扩展性。
参考文章：
