//public List<string> GetAllMailbox(List<DeepWebPage> deepWebPages)
//{
// List<string> Mailboxes = new List<string>();
// foreach (var dwp in deepWebPages)
// {
// }
// return Mailboxes;
//}
public List <string> GetSingleMailbox(DeepWebPage deepWebPage)
{
List <string> smailboxes = new List <string>();
WebClient MyWebClient = new WebClient();
string htmlCode = "";
MyWebClient.Credentials = CredentialCache.DefaultCredentials; //获取或设置用于向Internet资源的请求进行身份验证的网络凭据
Byte[] pageData = MyWebClient.DownloadData(deepWebPage.webUrl); //从指定网站下载数据
if (IsUtf8(pageData))
{
string pageHtml = Encoding.UTF8.GetString(pageData); //如果获取网站页面采用的是UTF-8，则使用这句
htmlCode = pageHtml;
}
else
{
string pageHtml = Encoding.Default.GetString(pageData); //如果获取网站页面采用的是GB2312，则使用这句
htmlCode = pageHtml;
}
/*暂时认为一个页面只有一个电子邮箱*/
string key1 = "<li>电子邮箱：<span><strong>";//寻找url的第一个键
int pos = htmlCode.IndexOf(key1);
string key2 = "user-name";
int pos1 = htmlCode.IndexOf(key2);
smailboxes.Add(htmlCode.Substring(pos1 + 11, htmlCode.Remove(0, pos1 + 11).IndexOf("</h5>")) + "\t" + htmlCode.Substring(pos + 23, htmlCode.Remove(0, pos + 23).IndexOf("</strong>")) + "\n");//24可修改
return(smailboxes);
}
public List <DeepWebPage> getSonPage(String url)
{
List <DeepWebPage> deepWebPages = new List <DeepWebPage>();
string htmlCode = "";
WebClient MyWebClient = new WebClient();
MyWebClient.Credentials = CredentialCache.DefaultCredentials; //获取或设置用于向Internet资源的请求进行身份验证的网络凭据
Byte[] pageData = MyWebClient.DownloadData(url); //从指定网站下载数据
if (IsUtf8(pageData))
{
string pageHtml = Encoding.UTF8.GetString(pageData); //如果获取网站页面采用的是UTF-8，则使用这句
htmlCode = pageHtml;
}
else
{
string pageHtml = Encoding.Default.GetString(pageData); //如果获取网站页面采用的是GB2312，则使用这句
htmlCode = pageHtml;
}
string key1 = "http://pdc.hzau.edu.cn/jgfw/desktop/";//寻找url的第一个键
int pos = htmlCode.IndexOf(key1);
IList <int> list = new List <int>();
while (pos > -1)
{
list.Add(pos);
pos += key1.Length;
if (pos >= htmlCode.Length)
{
break;
}
pos = htmlCode.IndexOf(key1, pos);
}
foreach (var item in list)
{
DeepWebPage deepWebPage = new DeepWebPage();
deepWebPage.webUrl = htmlCode.Substring(item, 66);//截取网址长度可以配
deepWebPages.Add(deepWebPage);
//var l = htmlCode.Remove(0, item).IndexOf("\"");
//surfaceWebPage.sonUrls.Add(htmlCode.Substring(item, 66));
//Console.WriteLine(htmlCode.Substring(item, item + 10).Length);
}
//foreach (var item in surfaceWebPage.sonUrls)
//{
// Console.WriteLine(item);
//}
return(deepWebPages);
}