asp.net(c#)做一個網頁數據采集工具

代碼如下:

//提取產品列表頁中產品最終頁的網頁 
private void button1_Click(object sender, EventArgs e) 
{ 
if (textBox1.Text.Trim() == "" || textBox2.Text.Trim() == "") 
{ 
MessageBox.Show("網址和域名不能為空！", "信息提示", MessageBoxButtons.OK, MessageBoxIcon.Information); 
return; 
} 
try 
{ 
string Html = inc.GetHtml("http://study.pctoday.net.cn"); 
//ArrayList al = inc.GetMatchesStr(Html, "<a[^>]*?>.*?</a>"); 
ArrayList al = inc.GetMatchesStr(Html, @"href\s*=\s*(?:[\'\""\s](?<1>[^\""\']*)[\'\""])");//提取鏈接 

" title="Replica Watches:">Replica Watches Buy Full Quality Popular Luxury Watches at Amazing Price, Your One Stop Discount Swiss Watches StoreExclusive Replica Rolex Watches, Tag Heuer Watches Replica, Cartier Watches online Sale! 
StringBuilder sb = new StringBuilder(); 
foreach (object var in al) 
{ 
string a = var.ToString().Replace("\"", "").Replace("'", ""); 
a = Regex.Replace(a, "href=", "", RegexOptions.IgnoreCase | RegexOptions.Multiline); 
if (a.StartsWith("/")) 
a = textBox2.Text.Trim() + a; 
if (!a.StartsWith("http://")) 
a = "http://" + a; 
sb.Append(a + "\r\n"); 
} 
textBox5.Text = sb.ToString();//把提取到網址

MessageBox.Show("共提取" + al.Count.ToString() + "個鏈接", "信息提示", MessageBoxButtons.OK, MessageBoxIcon.Information);

}
catch (Exception err)
{
MessageBox.Show("提取出錯！原因：" + err.Message, "信息提示", MessageBoxButtons.OK, MessageBoxIcon.Information);
}

}

//把采集的產品頁面html代碼進行字符串處理，提取需要的代碼，最后保存到本地一個access數據庫中，同時提取產品圖片地址并自動現在圖片到本地images文件夾下

private void backgroundWorker1_DoWork(object sender, DoWorkEventArgs e)
{
//填充產品表
Database.ExecuteNonQuery("delete from Tb_Product");
DataTable dt2 = new DataTable();
OleDbConnection conn = new OleDbConnection(Database.ConnectionStrings);
OleDbDataAdapter da = new OleDbDataAdapter("select * from Tb_Product", conn);
OleDbCommandBuilder cb = new OleDbCommandBuilder(da);
da.Fill(dt2);
dt2.Rows.Clear();

BackgroundWorker worker = (BackgroundWorker)sender;//這個是做一個進度條

string[] Urls = textBox5.Text.Trim().ToLower().Replace("\r\n", ",").Split(',');
DataTable dt = new DataTable();
StringBuilder ErrorStr = new StringBuilder();
string html = "", ImageDir = AppDomain.CurrentDomain.BaseDirectory + "Images\\";

//循環每次采集網址
for (int i = 0; i < Urls.Length; i++)
{
try
{
if (!worker.CancellationPending)
{
if (Urls[i] == "")
return;
html = inc.GetHtml(Urls[i]);//獲取該url的html代碼
DataRow NewRow = dt2.NewRow();

//產品名
string ProductName = html.Substring(html.IndexOf("<title>") + 7);
NewRow["ProductName"] = ProductName.Remove(ProductName.IndexOf("</title>")).Trim();

//產品編號
NewRow["ModelId"] = NewRow["ProductName"].ToString().Substring(NewRow["ProductName"].ToString().IndexOf("Model:") + 6).Trim();

//產品介紹，這些都是根據不同網站的html做相應的修改
string Introduce = html.Substring(html.IndexOf("Product Details") + 26);
Introduce = Introduce.Remove(Introduce.IndexOf("</table>") + 8).Trim()

NewRow["Introduce"] = Introduce;

" title="Replica Watches:">Replica Watches Buy Full Quality Popular Luxury Watches at Amazing Price, Your One Stop Discount Swiss Watches StoreExclusive Replica Rolex Watches, Tag Heuer Watches Replica, Cartier Watches online Sale!
//下載圖片
string ProductImage = html.Substring(html.IndexOf("align=center><img") + 17);
ProductImage = textBox2.Text.Trim() + ProductImage.Substring(ProductImage.IndexOf("src=\"") + 5);
ProductImage = ProductImage.Remove(ProductImage.IndexOf("\""));
try
{
inc.DownFile(ProductImage, ImageDir + ProductImage.Substring(ProductImage.LastIndexOf("/") + 1));
}
catch (Exception)
{
ErrorStr.Append("下載圖片失敗，圖片地址：" + ImageDir + ProductImage.Substring(ProductImage.LastIndexOf("/") + 1) + "\r\n");
}

dt2.Rows.Add(NewRow);

//Thread.Sleep(100);
worker.ReportProgress((i + 1) * 100 / Urls.Length, i);
toolStripStatusLabel1.Text = "處理進度:" + (i + 1).ToString() + "/" + Urls.Length.ToString();//進度條
}

}
catch (Exception err)
{
ErrorStr.Append("采集錯誤：" + err.Message + ";網址：" + Urls[i] + "\r\n");
}
}
da.Update(dt2);
DataBind(dt2);
ShowError(ErrorStr.ToString());
}

/// <summary>
/// ASPX頁面生成靜態Html頁面，作者：鄭少群
/// </summary>
public static string GetHtml(string url)
{
StreamReader sr = null;
string str = null;
//讀取遠程路徑
WebRequest request = WebRequest.Create(url);
HttpWebResponse response = (HttpWebResponse)request.GetResponse();
sr = new StreamReader(response.GetResponseStream(), Encoding.GetEncoding(response.CharacterSet));
str = sr.ReadToEnd();
sr.Close();
return str;
}

// 提取HTML代碼中的網址
public static ArrayList GetMatchesStr(string htmlCode, string strRegex)
{
ArrayList al = new ArrayList();

Regex r = new Regex(strRegex, RegexOptions.IgnoreCase | RegexOptions.Multiline);
MatchCollection m = r.Matches(htmlCode);

for (int i = 0; i < m.Count; i++)
{
bool rep = false;
string strNew = m[i].ToString();

// 過濾重復的URL
foreach (string str in al)
{
if (strNew == str)
{
rep = true;
break;
}
}

if (!rep) al.Add(strNew);
}

al.Sort();

return al;
}

public static void DownFile(string Url, string Path)
{

HttpWebRequest request = (HttpWebRequest)WebRequest.Create(Url);
HttpWebResponse response = (HttpWebResponse)request.GetResponse();
Stream stream = response.GetResponseStream();
long size = response.ContentLength;
//創建文件流對象
using (FileStream fs = new FileStream(Path, FileMode.OpenOrCreate, FileAccess.Write))
{
byte[] b = new byte[1025];
int n = 0;
while ((n = stream.Read(b, 0, 1024)) > 0)
{
fs.Write(b, 0, n);
}
}
}

聲明：本網頁內容旨在傳播知識，若有侵權等問題請及時與本網聯系，我們將在第一時間刪除處理。TEL:177 7030 7066 E-MAIL:11247931@qq.com