您现在的位置是：网站首页> 编程资料编程资料
hta作品笨狼CSDN爬虫_hta_
2023-05-26
457人已围观
简介 hta作品笨狼CSDN爬虫_hta_
帖子号码
.xml?temp=.XXXXXXX
/
.xml?temp=.XXXXXXX
爬行范围
说明：
1,将CSDN帖子分类保存在爬虫身边/*.XML文件;
2,也可以选择保留XML文件,进行深加工;
3,请根据当前的帖子号码来设定爬行距离;
4,欢迎光临
window.resizeTo(600,400);
window.moveTo(window.screen.availWidth/2-300,window.screen.availHeight/2-150);
var xmlURL=new String();
var headURL ="http://community.csdn.net/Expert/topic/";
var tailURL =".xml?temp=";
var xmlhttp = new ActiveXObject("Microsoft.XMLHTTP");
var stream = new ActiveXObject("ADODB.Stream");
var fso = new ActiveXObject("Scripting.FileSystemObject")
//定义本文件绝对路径
var thisFileDir =getthisFileDir();
//定义本文件名
var thisFileName = LastOne(thisFileDir,"\\");
//定义本文件夹路径
var thisFileFolder = getFolderDir(thisFileDir);
InitBigClass();
InitSmallClass();
var xmlID;
var xmlFolder = setFolder("xml");
function gogogo()
{ //主程序
wormClick.innerText ="正在爬...";
//window.resizeTo(100,200);
//window.scrollTo(10,20);
//window.moveTo(window.screen.availWidth-200,window.screen.availHeight-250)
for(i=parseInt(numStart1.value) ;i
for(j=parseInt(numStart2.value);j
xmlURL = headURL + i + "/" + j + tailURL + Math.random().toString().substr(1,8);
try
{
saveToFile(xmlURL,smallclassid.value);
xmlID = j;
}
catch(e)
{
}
}
}
alert("爬行完毕！！");
wormClick.innerText = "开始爬...";
}
function saveToFile(URL,classID)
{
xmlhttp.Open("GET",URL, false);
try
{
xmlhttp.Send();
var result = xmlhttp.status;
}
catch(e)
{
return(false);
}
if (xmlhttp.readyState == 4) // 调用完毕
{
if(result==200)
{
var tempXML=xmlhttp.responseBody;
stream.type=1;
stream.Open() ;
stream.Write(tempXML);
stream.SaveToFile(xmlFolder + xmlID + ".xml")
stream.Close();
}
}
}
function setFolder(str)
{
var folder = thisFileFolder + str;
if(!fso.FolderExists( folder))
{
fso.CreateFolder(folder)
}
return(folder + "\\")
}
function getthisFileDir()
{//得到本文件完全路径
var temp = window.location.href;
temp = temp.substr(8);
temp = unescape(temp);
temp = temp.replace(/[\/]/ig,"\\")
return(temp);
}
function LastOne(Str,splitStr)
{//输入字符和分隔符，得到最后一部分
var Arr=Str.split(splitStr)
return(Arr[Arr.length-1]);
}
function getFolderDir(fullDir)
{//输入得到全路径，得到文件夹路径
var s=LastOne(fullDir,"\\");
return(fullDir.substr(0,fullDir.length-s.length));
}
function InitBigClass()
{
bigclass = new Array();
bigclass[0] = new Array();
bigclass[0][0] = "0";
bigclass[0][1] = "全部社区";
bigclass[1] = new Array();
bigclass[1][0] = "3";
bigclass[1][1] = "Web 开发";
bigclass[2] = new Array();
bigclass[2][0] = "4";
bigclass[2][1] = "软件工程/管理";
bigclass[3] = new Array();
bigclass[3][0] = "5";
bigclass[3][1] = "企业开发";
bigclass[4] = new Array();
bigclass[4][0] = "6";
bigclass[4][1] = "专题开发";
bigclass[5] = new Array();
bigclass[5][0] = "7";
bigclass[5][1] = "多媒体/设计";
bigclass[6] = new Array();
bigclass[6][0] = "8";
bigclass[6][1] = "Linux/Unix社区";
bigclass[7] = new Array();
bigclass[7][0] = "9";
bigclass[7][1] = "硬件/嵌入开发";
bigclass[8] = new Array();
bigclass[8][0] = "10";
bigclass[8][1] = "Windows专区";
bigclass[9] = new Array();
bigclass[9][0] = "11";
bigclass[9][1] = "硬件使用";
bigclass[10] = new Array();
bigclass[10][0] = "13";
bigclass[10][1] = "软件培训/认证/考试";
bigclass[11] = new Array();
bigclass[11][0] = "14";
bigclass[11][1] = "扩充话题";
bigclass[12] = new Array();
bigclass[12][0] = "15";
bigclass[12][1] = "社区支持";
bigclass[13] = new Array();
bigclass[13][0] = "16";
bigclass[13][1] = "产品/厂家";
bigclass[14] = new Array();
bigclass[14][0] = "17";
bigclass[14][1] = "其他开发语言";
bigclass[15] = new Array();
bigclass[15][0] = "18";
bigclass[15][1] = "其他数据库开发";
bigclass[16] = new Array();
bigclass[16][0] = "20";
bigclass[16][1] = "计算机图书";
bigclass[17] = new Array();
bigclass[17][0] = "50";
bigclass[17][1] = "VC/MFC";
bigclass[18] = new Array();
bigclass[18][0] = "51";
bigclass[18][1] = "VB";
bigclass[19] = new Array();
bigclass[19][0] = "52";
bigclass[19][1] = ".NET技术";
bigclass[20] = new Array();
bigclass[20][0] = "53";
bigclass[20][1] = "Delphi";
bigclass[21] = new Array();
bigclass[21][0] = "54";
bigclass[21][1] = "Java";
bigclass[22] = new Array();
bigclass[22][0] = "55";
bigclass[22][1] = "C++ Builder";
bigclass[23] = new Array();
bigclass[23][0] = "56";
bigclass[23][1] = "C/C++";
bigclass[24] = new Array();
bigclass[24][0] = "57";
bigclass[24][1] = "MS-SQL Server";
bigclass[25] = new Array();
bigclass[25][0] = "58";
bigclass[25][1] = "PowerBuilder";
bigclass[26] = new Array();
bigclass[26][0] = "59";
bigclass[26][1] = "WebSphere";
bigclass[27] = new Array();
bigclass[27][0] = "60";
bigclass[27][1] = "DB2";
bigclass[28] = new Array();
bigclass[28][0] = "61";
bigclass[28][1] = "Oracle";
bigclass[29] = new Array();
bigclass[29][0] = "62";
bigclass[29][1] = "程序员杂志";
bigclass[30] = new Array();
bigclass[30
