web.js

var page = require(‘webpage‘).create(),
 system = require(‘system‘),
 address,output,csvPath,nodePathFile,outOriginalimg,PCSPuserAgent;
var fs = require("fs");
if (system.args.length  -1)
{
     page.settings = {
       userAgent:PCSPuserAgent,
  	  javascriptEnabled: true,
      loadImages: true
};
}
else
{
     page.settings = {
      javascriptEnabled: true,
      loadImages: true
};
}
    page.viewportSize = {width:414,height:30};
    page.open(address, function (status) {
      //  fs.write(‘test.txt‘, "childNodes Num{_}Node Name{_}NodeIndexPath{_}Width{_}Height{_}x{_}y{_}Dispaly{_}ImgUrlOrBackImgUrl" + "\r\n", ‘a‘);
      if(fs.exists(csvPath))
      {
         fs.remove(csvPath);
       }
       window.setTimeout(function () { 
       VisiteHtmlDom("0",BeforecsvPath); 
 console.log("---------------------FS MOVE------------------------------");
       fs.move(BeforecsvPath,csvPath);
            }, 40000);
 waitFor2(
 function csvCreate()
{  
  if(fs.exists(csvPath)){
if(csvindex -1) || (htmlNode.nodeName.indexOf("NOSCRIPT") > -1)|| (htmlNode.nodeName.indexOf("HEAD") > -1))
            {
            htmlNode = htmlNode;
            }
       else if(htmlNode.nodeName.indexOf("IFRAME") > -1&&window.getComputedStyle(htmlNode).display!="none")
        {
        var iframdom = htmlNode.contentWindow.document;
    
        var roots = iframdom.getElementsByTagName(‘html‘);
        htmlNode=roots[0];
      }
            
            else
            {
            htmlNode = htmlNode.childNodes[index];
            } 
        }

        if ((htmlNode.nodeName.indexOf("#text") > -1)) 
        {
                    return htmlNode.childNodes.length + "{_}" + htmlNode.nodeName+ "{_}" + str+"{_}"+"{_}"+"{_}"+"{_}"+"{_}"+"{_}"+"{_}"+ ("{_}" + htmlNode.nodeValue||"").replace(/\r|\n/ig, "");
        
        }
        else if((htmlNode.nodeName.indexOf("SCRIPT") > -1) || (htmlNode.nodeName.indexOf("NOSCRIPT") > -1)|| (htmlNode.nodeName.indexOf("HEAD") > -1)|| (htmlNode.nodeName.indexOf("#comment") > -1))
        {
        	
            return "0"+"{_}"+htmlNode.nodeName + "{_}" + str+"{_}"+"{_}"+"{_}"+"{_}"+"{_}"+"{_}"+"{_}"+"{_}";
         }
        else {
            // 
            var bgImgUrl= htmlNode.style.getPropertyValue("background-image");
          
            // var bgImgUrl=htmlNode.style.getPropertyValue("background-image").replace(/^(url)\(|\)/g, ‘‘);
            if(bgImgUrl!=null)
                bgImgUrl=bgImgUrl.replace(/^(url)\(|\)/g, ‘‘);
          
            var width = window.getComputedStyle(htmlNode).width;
            var height = window.getComputedStyle(htmlNode).height;
            var left = window.getComputedStyle(htmlNode).left;
            var top = window.getComputedStyle(htmlNode).top;
            var display = window.getComputedStyle(htmlNode).display;
            switch (htmlNode.nodeName) {
                case "IMG":
                    var imgUrl = htmlNode.src; //???????
                    return htmlNode.childNodes.length + "{_}" + htmlNode.nodeName + "{_}" + str + "{_}" + width + "{_}" + height + "{_}" + left + "{_}" + top + "{_}" + display + "{_}" + imgUrl+ "{_}"+ "{_}"; //img bgurl txt
                    break;
                case "EMBED":
                case "#COMMENT":
                    return htmlNode.childNodes.length + "{_}" +htmlNode.nodeName+"{_}" + str+"{_}"+"{_}"+"{_}"+"{_}"+"{_}"+"{_}"+"{_}"+"{_}";
                    break;
                case "IFRAME":
                     var iframeUrl = htmlNode.src; 
                if(window.getComputedStyle(htmlNode).display!="none")
                   {
                     return "1" + "{_}" + htmlNode.nodeName + "{_}" + str + "{_}" + width + "{_}" + height + "{_}" + left + "{_}" + top + "{_}" + display + "{_}" + iframeUrl+ "{_}"+ "{_}"; //img bgurl txt
                     }
                     else
                     { 
                 		   return "0" + "{_}" + htmlNode.nodeName + "{_}" + str + "{_}" + width + "{_}" + height + "{_}" + left + "{_}" + top + "{_}" + display + "{_}" + iframeUrl+ "{_}"+ "{_}"; //img bgurl txt
                   }
                     default:
                     return htmlNode.childNodes.length + "{_}" + htmlNode.nodeName + "{_}" + str + "{_}" + width + "{_}" + height + "{_}" + left + "{_}" + top + "{_}" + display + "{_}" +"{_}"+ bgImgUrl+"{_}";
                     break;
            }
        }
    }, nodePath);

console.log("create CSV");
    fs.write(inputcsvPath, nodeinfo + "\r\n", ‘a‘);
   

    // if(!(nodeinfo.indexOf("undefined")>-1))

    var childNodesCount = nodeinfo.split("{_}")[0].split("/")[0];

    for (var childIndex = 0; childIndex  -1)
        {
        var iframdom = htmlNode.contentWindow.document;
        var roots = iframdom.getElementsByTagName(‘html‘);
        htmlNode=roots[0];
        i++;
        }
    }
     
       if (htmlNode.nodeName.indexOf("text")>-1)
          {
            htmlNode.parentNode.style.boxSizing = "border-box";
            htmlNode.parentNode.style.border = "5px solid #ff0000";
          }
          else
             {
              htmlNode.style.boxSizing = "border-box";
              htmlNode.style.border = "5px solid #ff0000";
              } 
      },line);
  }
  
stream.close();
}
 function waitFor(testFx, onReady, timeOutMillis) {
    var maxtimeOutMillis = timeOutMillis ? timeOutMillis : 120000, //