Promise.all 本身不负责执行，执行过程在传递给Promise.all之前已经开始，Promise.all只等待全部执行完成，执行resolve，或碰到有执行失败，立即执行reject部分。Promise.all非常好用，唯一的问题是，不能限制并发数量，所有任务同时开始执行，因为Promise.all本身不负责执行具体任务，所以也无法实现并发控制。

实现一个简单的可以控制并发数量的Promise.allLimit函数，可以通过参数来控制并发数量。代码：

/* promise-limit.js */
/* jshint esversion: 6 */
/*jslint node: true */
Promise.allLimit = function(arr, wrap, limit, callback) {return new Promise((resolve, reject) => {var total = arr.length;var result = new Array(total);var rejected = false;var dones = 0;function run(n) {setTimeout(() => {wrap(n, arr.shift()).then(res => {return typeof callback === 'function' ? callback(n, res) : Promise.resolve(res);}).then(res => {dones++;result[n] = res;if (!rejected) {if (arr.length) {run(total - arr.length);} else if (dones === total) {resolve(result);}}}).catch(err => {rejected = true;reject(err);});}, 0);}arr.slice(0, limit).forEach((v, n) => {run(n);});});
};

同样返回一个Promise对象，可以直接替换Promise.all，不同的是，需要传递一个函数(wrap参数），用来包裹生成每一个具体执行的Promise对象，limit用来限定并发数量，在指定并发任务内，一个任务完成后，再吸入一个新任务继续执行。

callback用来解析每一次任务完成后所需要的后续动作，比如存储下载的内容、或将参数做变换，必须也返回一个Promise对象。

测试代码：

Promise.allLimit([2000, 1500, 2500, 3000, 1500], function(n, time) {return new Promise((resolve, reject) => {console.log("Start Job: ", n, time);// setTimeout(2500 === time ? reject : resolve, time, "Time: " + time); // 测试rejectsetTimeout(resolve, time, "Time: " + time);});
}, 2, (n, res) => {// log Job n doneconsole.log("Done Job: ", res);return Promise.resolve(n);
// log Job n done
}).then(result => {console.log("All Done: ", result);
}).catch(err => {console.log("Error: ", err);
});

最后贡献一个下载妹子图的简单代码，默认控制并发数量10个。

#!/usr/bin/env node/* jshint esversion: 6 */
/*jslint node: true */
require('./promise-limit.js');const FS = require('fs');
const PATH = require('path');
const UTIL = require('util');
const ARGV = require('yargs').argv;
const REQUEST = require('request');
const CHEERIO = require('cheerio');if (!ARGV.url || !ARGV.dir || !/\/$/.test(ARGV.dir) || !ARGV.img || !ARGV.total || !ARGV.from || !ARGV.to) {console.log("usage: --url http://example.com --dir ./imgs/ --img '.main-image img' --total '共(\d+)页' --from '.php' --to '_%d.php' ");process.exit();
}const parallel = ARGV.parallel || 10; //并发数量const headers = {'Referer': ARGV.url,'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8','Accept-Encoding': 'gzip, deflate, sdch','Accept-Language': 'en-US,en;q=0.8,zh-CN;q=0.6,zh;q=0.4','User-Agent': 'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/53.0.2785.104 Safari/537.36'
};function fetch(url, encoding = 'utf8') {return new Promise((resolve, reject) => {REQUEST({url: url,headers: headers,gzip: true,encoding: encoding}, (error, response, body) => {if (error) {reject(error);} else {resolve(body);}});});
}function write(file, content) {return new Promise(function(resolve, reject) {FS.writeFile(file, content, function(err) {if (err) {reject(err);} else {resolve();}});});
}fetch(ARGV.url).then(rsp => {const total = parseInt((rsp.match(new RegExp(ARGV.total)) || [0, 0])[1]);if (!total) {throw new Error('Match total error');}console.log("Total: %d, Parallel: %d", total, parallel);return Promise.allLimit(Array.from({length: total}, (v, k) => k + 1), (k, v) => {return fetch(v === 1 ? ARGV.url : ARGV.url.replace(ARGV.from, UTIL.format(ARGV.to, v)));}, parallel, (k, res) => {let src = CHEERIO.load(res)(ARGV.img).attr('src');let file = (k + 1) + PATH.extname(src);return fetch(src, null).then(img => {return write(ARGV.dir + file, img);}).then(() => {console.log("OK: [" + file + "]\t" + src);return file;}).catch(err => {console.log("ER: [" + file + "]\t" + src + " : " + err.toString());});});}).then(rsp => {console.log("All Jobs Done: ");console.log(rsp);}).catch(err => {console.log("Fetch failed: %s", err.toString());});

执行：

 ./request.js --total '(\d+)</span></a><a[^<>]+><span>下一页' --from '/71636' --to '/71636/%d' --img '.main-image img' --dir ./71636/ --url 'http://www.mzitu.com/71636' --parallel 10