Nodejs. HTTP module, Cheerio module implementation of small crawler.

Code:

1 varHTTP = require ("http");2 3 varCheerio = require ("Cheerio");4 5 6 varurl = ' http://www.imooc.com/learn/348 ';7 8 9Http.get (URL,function(res) {Ten     varhtml = ' '; One  ARes.on (' Data ',function(data) { -HTML + =data; -     }); the  -Res.on (' End ',function(){ -         varCoursedata =filterchapters (HTML); -  + Printcourseinfo (coursedata); - Console.log (coursedata); +     }); A}). On (' Error ',function(){ atConsole.log ("Error getting course data!")); - }); -  - functionfilterchapters (HTML) - { -     var$ =cheerio.load (HTML); in  -     //All Chapters to     varChapters = $ ('. Chapter ')); +  -     varCoursedata = []; the  *Chapters.each (function(item) { $         varChapter = $ ( This);Panax Notoginseng         varChaptertitle = Chapter.find (' h3 strong '). Text (). Replace (/\r|\n/ig, ""). Trim (); -         varVideos = Chapter.find (". Video"). Children (' Li ')); the  +         varChapterdata = { A Chaptertitle:chaptertitle, the videos: [] +         }; -  $Videos.each (function(index, item2) { $             varVideo = $ ( This). Find ('. J-media-item '); -             varVideotitle = Video.text (). Replace (/\r|\n/ig, ""). Trim (); -             varid = video.attr (' href '). Split (' video/') [1]; the  - ChapterData.videos.push ({Wuyi Title:videotitle, the Id:id -             }) Wu         }); -  About Coursedata.push (chapterdata); $     }); -  -     returnCoursedata; - } A  +  the functionprintcourseinfo (coursedata) - { $Coursedata.foreach (function(item) { the         varChaptertitle =Item.chaptertitle; theConsole.log (chaptertitle + ' \ n ')); the  theItem.videos.forEach (function(video) { -Console.log (' [' + video.id+ '] ' +video.title); in         }); the     }); the}

Run:

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

Reference Links:

• HTTP small Reptile
• node. JS Learning-Crawl web data using Cheerio

Nodejs. HTTP module, Cheerio module implementation of small crawler.