├── package.json
├── README.md
└── app.js


/package.json:
--------------------------------------------------------------------------------
 1 | {
 2 |   "name": "imooc-video-download",
 3 |   "version": "0.0.1",
 4 |   "description": "可以批量下载慕课网的视频",
 5 |   "main": "app.js",
 6 |   "dependencies": {
 7 |     "cheerio": "^0.17.0",
 8 |     "superagent": "^0.20.0"
 9 |   },
10 |   "scripts": {
11 |     "test": "node app courseid"
12 |   },
13 |   "keywords": [
14 |     "imooc"
15 |   ],
16 |   "author": "Awai",
17 |   "license": "ISC"
18 | }
19 | 


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
 1 | # imooc-video-download
 2 | 
 3 | 该项目已经年久失修了~
 4 | 
 5 | 如果你有兴趣，可以 fork 之后尝试修复~
 6 | 
 7 | > 使用 :
 8 | 
 9 | > 基于 Node.js 写的爬虫, 可以批量下载慕课网上的视频
10 | 
11 | > - 因为基于 Node.js, 所以使用前确保已经安装好 Node.js 环境
12 | 
13 | > - 这个爬虫用了 superagent 和 cheerio 这两个模块, 已经加入到项目中
14 | 
15 | > - 使用前先在源码里面修改 ***储存路径*** 和 自己的 ***cookies***
16 | 
17 | 直接终端输入
18 | ---
19 | 
20 | ``` bash 
21 | node app id  // id 就是课程id
22 | ```
23 | `例如: http://www.imooc.com/view/442  id 就是 442`
24 | 
25 | 


--------------------------------------------------------------------------------
/app.js:
--------------------------------------------------------------------------------
  1 | /**
  2 |  * imooc
  3 |  *
  4 |  * Created by Awai on 15/9/14.
  5 |  */
  6 | 
  7 | /**
  8 |  *  修改储存目录 /xxx/xxx/ (最后要加一个 '/' )
  9 |  */
 10 | var savePath = ''
 11 | /**
 12 |  * 自己在慕课网上登录后 获取cookies 并替换下面的
 13 |  */
 14 | var cookies = ''
 15 | 
 16 | var superagent = require('superagent');
 17 | var cheerio = require('cheerio');
 18 | var fs = require('fs');
 19 | 
 20 | var courseId = process.argv.splice(2, 1);
 21 | 
 22 | var courseTitle = '';
 23 | var courseTotalCount = 0
 24 | var finishCount = 0
 25 | 
 26 | var download = function(url, filename, callback) {
 27 | 
 28 |     filename = filename.replace(/\(.*\)/,'') + '.mp4';
 29 | 
 30 |     var dirPath = savePath + courseTitle + '/'
 31 |     if (!fs.existsSync(dirPath)) {
 32 |         fs.mkdirSync(dirPath);
 33 |     }
 34 | 
 35 |     console.log('开始下载第' + courseTotalCount + '个视频' + filename + ' 地址: ' + url);
 36 |     var writeStream = fs.createWriteStream(dirPath + filename);
 37 |     writeStream.on('close', function() {
 38 | 
 39 |         callback(filename);
 40 |     })
 41 | 
 42 |     var req = superagent.get(url)
 43 |     req.pipe(writeStream);
 44 | 
 45 | }
 46 | 
 47 | var getMovieUrl = function(id, callback) {
 48 |     superagent.get('http://www.imooc.com/course/ajaxmediainfo/?mid=' + id + '&mode=flash')
 49 |         .end(function(err, res) {
 50 |             var url = JSON.parse(res.text);
 51 | 
 52 |             if(url.result == 0) {
 53 |                 url = url.data.result.mpath[0];
 54 |                 callback(url);
 55 |             }
 56 |         })
 57 | }
 58 | 
 59 | var headers = {
 60 |     "Cache-Control": "max-age=0",
 61 |     "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8",
 62 |     "Referer": "http://www.imooc.com/",
 63 |     "Accept-Encoding": "gzip, deflate, sdch",
 64 |     "Accept-Language": "zh-CN,zh;q=0.8",
 65 |     "Cookie": cookies
 66 | };
 67 | 
 68 | superagent
 69 |     .get('www.imooc.com/learn/' + courseId)
 70 |     .set(headers)
 71 |     .end(function(err, res) {
 72 | 
 73 |         var $ = cheerio.load(res.text);
 74 | 
 75 |         $('.course-infos .hd').find('h2').each(function(item) {
 76 |             courseTitle = $(this).text();
 77 |         })
 78 | 
 79 |         $('.chapter').each(function(item) {
 80 | 
 81 |             var videos = $(this).find('.video').children('li')
 82 | 
 83 |             videos.each(function(item) {
 84 |                 var video = $(this).find('a')
 85 |                 var filename = video.text().replace(/(^\s+)|(\s+$)/g,"");
 86 |                 var id = video.attr('href').split('video/')[1]
 87 |                 if (!id) {
 88 |                     console.log('跳过下载: ' + filename)
 89 |                 }else {
 90 | 
 91 |                     // 获取下载地址
 92 |                     getMovieUrl(id, function(url) {
 93 | 
 94 |                         courseTotalCount ++;
 95 | 
 96 |                         // 开始下载
 97 |                         download(url, filename, function(_filename) {
 98 |                             finishCount++;
 99 |                             console.log('第' + finishCount + '个视频' + _filename + '下载完成')
100 |                         })
101 |                     })
102 |                 }
103 |             })
104 |         })
105 |     })
106 | 


--------------------------------------------------------------------------------