日期:2010-03-05 浏览次数:20375 次
//先按照下面的字段创建一个数据表
以下为引用的内容:
public partial class Form2 : Form
{
public Form2()
{
InitializeComponent();
}
//姓名
public static string XM = "";
//年龄
public static string nl = "";
//性别
public static string XB = "";
//身高
public static string SG = "";
//政治面貌
public static string mm = "";
//民族
public static string MZ = "";
//学历
public static string XL = "";
//婚姻状况
public static string HK = "";
//所学专业
public static string ZY = "";
//工作经验
public static string GZJY = "";
//在职单位
public static string ZZDW = "";
//在职职位
public static string ZZZW = "";
//工作经历
public static string GZJL = "";
//要求月薪
public static string YX = "";
//工作性质
public static string GZXZ = "";
//求职意向
public static string QZYX = "";
//具体职务
public static string JTZW = "";
//期望工作地
public static string QWGZD = "";
//教育情况,语言水平,技术专长
public static string QT = "";
private void button1_Click(object sender, EventArgs e)
{
label1.Text = "正在采集数据……";
//遍历数据的页数
for (int i = 1; i <=50; i++)
{
CJ("http://www.xcjob.cn/renli.asp?pageno=" + i);
}
label1.Text = "恭喜你采集完成!";
MessageBox.Show("恭喜你采集完成!");
}
//采集数据
private void CJ(string Url)
{
//获得页面源文件(Html)
string strWebContent = YM(Url);
//按照Html里面的标签 取出和数据有关的那段源码
int iBodyStart = strWebContent.IndexOf("<body", 0);
int aaa = strWebContent.IndexOf("关键字:", iBodyStart);
int iTableStart = strWebContent.IndexOf("<table", aaa);
int iTableEnd = strWebContent.IndexOf("</table>", iTableStart);
string strWeb = strWebContent.Substring(iTableStart, iTableEnd - iTableStart);
//生成HtmlDocument
HtmlElementCollection htmlTR = HtmlTR_Content(strWeb, "tr");
foreach (HtmlElement tr in htmlTR)
{
try
{
//姓名
XM = tr.GetElementsByTagName("a")[0].InnerText;
//获得详细信息页面的网址
string a = tr.GetElementsByTagName("a")[0].GetAttribute("href").ToString();
a = "http://www.xcjob.cn" + a.Substring(11);
Content(a);
}
catch { }
}
}
//采集详细数据
private void Content(string URL)
{
try
{
string strWebContent = YM(URL);
//按照Html里面的标签 取出和数据有关的那段源码
int iBodyStart = strWebContent.IndexOf("<body", 0);
int iTableStart = strWebContent.IndexOf("浏览次数", iBodyStart);
int iTableEnd = strWebContent.IndexOf("<table", iTableStart);
int dd = strWebContent.IndexOf("</table>", iTableEnd);
string strWeb = strWebContent.Substring(iTableEnd, dd - iTableEnd + 8);
HtmlElementCollection htmlTR = HtmlTR_Content(strWeb, "table");
foreach (HtmlElement tr in htmlTR)
{
try
{
//年龄
nl = tr.GetElementsByTagName("tr")[1].GetElementsByTagName("td")[1].InnerText;
//性别
string XB_SG = tr.GetElementsByTagName("tr")[1].GetElementsByTagName("td")[3].InnerText;
XB = XB_SG.Substring(0, 1);
//身高
SG = XB_SG.Substring(11);
//政治面貌
mm = tr.GetElementsByTagName("tr")[2].GetElementsByTagName("td")[1].InnerText;
//民族
MZ = tr.GetElementsByTagName("tr")[2].GetElementsByTagName("td")[3].InnerText;
//学历
XL = tr.GetElementsByTagName("tr")[3].GetElementsByTagName("td")[1].InnerText;
//婚烟状况
HK = tr.GetElementsByTagName("tr