爬取慕课网课程信息实例教程

npm install cheerio

//接受一个url爬取整个网页，返回一个Promise对象function getPageAsync(url){return new Promise((resolve,reject)=>{
        console.log(`正在爬取${url}的内容`);
        http.get(url,function(res){
            let html = '';

            res.on('data',function(data){
                html += data;
            });

            res.on('end',function(){
                resolve(html);
            });

            res.on('error',function(err){
                reject(err);
                console.log('错误信息：' + err);
            })
        });
    })
}

const baseUrl = 'http://www.imooc.com/learn/';
const baseNuUrl = 'http://www.imooc.com/course/AjaxCourseMembers?ids=';//获取课程的IDconst videosId = [773,371];

Promise//当所有网页的内容爬取完毕    .all(courseArray)
    .then((pages)=>{//所有页面需要的内容let courseData = [];//遍历每个网页提取出所需要的内容pages.forEach((html)=>{
            let courses = filterChapter(html);
            courseData.push(courses);
        });//给每个courseMenners.number赋值for(let i=0;i<videosId.length;i++){for(let j=0;j<videosId.length;j++){if(courseMembers[i].id +'' == videosId[j]){
                    courseData[j].number = courseMembers[i].numbers;
                }
            }
        }//对所需要的内容进行排序courseData.sort((a,b)=>{return a.number > b.number;
        });//在重新将爬取内容写入文件中前，清空文件fs.writeFileSync(outputFile,'###爬取慕课网课程信息###',(err)=>{if(err){
                console.log(err)
            }
        });
        printfData(courseData);
    });

//接受一个爬取下来的网页内容，查找网页中需要的信息function filterChapter(html){
    const $ = cheerio.load(html);//所有章const chapters = $('.chapter');//课程的标题和学习人数let title = $('.hd>h2').text();
    let number = 0;//最后返回的数据//每个网页需要的内容的结构let courseData = {'title':title,'number':number,'videos':[]
    };

    chapters.each(function(item){
        let chapter = $(this);//文章标题let chapterTitle = Trim(chapter.find('strong').text(),'g');//每个章节的结构let chapterdata = {'chapterTitle':chapterTitle,'video':[]
        };//一个网页中的所有视频let videos = chapter.find('.video').children('li');
        videos.each(function(item){//视频标题let videoTitle = Trim($(this).find('a.J-media-item').text(),'g');//视频IDlet id = $(this).find('a').attr('href').split('video/')[1];
            chapterdata.video.push({'title':videoTitle,'id':id
            })
        });

        courseData.videos.push(chapterdata);

    });return courseData;
}

//获取上课人数function getNumber(url){

    let datas = '';

    http.get(url,(res)=>{
        res.on('data',(chunk)=>{
            datas += chunk;
        });

        res.on('end',()=>{
            datas = JSON.parse(datas);
            courseMembers.push({'id':datas.data[0].id,'numbers':parseInt(datas.data[0].numbers,10)});
        });
    });
}

        //给每个courseMenners.number赋值for(let i=0;i<videosId.length;i++){for(let j=0;j<videosId.length;j++){if(courseMembers[i].id +'' == videosId[j]){
                    courseData[j].number = courseMembers[i].numbers;
                }
            }
        }

//写入文件function writeFile(file,string) {
    fs.appendFileSync(file,string,(err)=>{if(err){
                console.log(err);
            }
        })
}//打印信息function printfData(coursesData){

    coursesData.forEach((courseData)=>{       // console.log(`${courseData.number}人学习过${courseData.title}\n`);       writeFile(outputFile,`\n\n${courseData.number}人学习过${courseData.title}\n\n`);

        courseData.videos.forEach(function(item){
            let chapterTitle = item.chapterTitle;// console.log(chapterTitle + '\n');            writeFile(outputFile,`\n  ${chapterTitle}\n`);

            item.video.forEach(function(item){// console.log('     【' + item.id + '】' + item.title + '\n');                writeFile(outputFile,`     【${item.id}】  ${item.title}\n`);
            })
        });

    });


}

/**
 * Created by hp-pc on 2017/6/7 0007. */const http = require('http');
const fs = require('fs');
const cheerio = require('cheerio');
const baseUrl = 'http://www.imooc.com/learn/';
const baseNuUrl = 'http://www.imooc.com/course/AjaxCourseMembers?ids=';//获取课程的IDconst videosId = [773,371];//