|
|
復(fù)制代碼 代碼如下:
//提取產(chǎn)品列表頁(yè)中產(chǎn)品最終頁(yè)的網(wǎng)頁(yè)
private void button1_Click(object sender, EventArgs e)
{
if (textBox1.Text.Trim() == "" || textBox2.Text.Trim() == "")
{
MessageBox.Show("網(wǎng)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(wǎng)址輸出到一個(gè)textBox，每個(gè)鏈接占一行
MessageBox.Show("共提取" + al.Count.ToString() + "個(gè)鏈接", "信息提示", MessageBoxButtons.OK, MessageBoxIcon.Information);
}
catch (Exception err)
{
MessageBox.Show("提取出錯(cuò)！原因：" + err.Message, "信息提示", MessageBoxButtons.OK, MessageBoxIcon.Information);
}
}
//把采集的產(chǎn)品頁(yè)面html代碼進(jìn)行字符串處理，提取需要的代碼，最后保存到本地一個(gè)access數(shù)據(jù)庫(kù)中，同時(shí)提取產(chǎn)品圖片地址并自動(dòng)現(xiàn)在圖片到本地images文件夾下
private void backgroundWorker1_DoWork(object sender, DoWorkEventArgs e)
{
//填充產(chǎn)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(gè)是做一個(gè)進(jìn)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(huán)每次采集網(wǎng)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(chǎn)品名
string ProductName = html.Substring(html.IndexOf("<title>") + 7);
NewRow["ProductName"] = ProductName.Remove(ProductName.IndexOf("</title>")).Trim();
//產(chǎn)品編號(hào)
NewRow["ModelId"] = NewRow["ProductName"].ToString().Substring(NewRow["ProductName"].ToString().IndexOf("Model:") + 6).Trim();
//產(chǎn)品介紹，這些都是根據(jù)不同網(wǎng)站的html做相應(yīng)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(jìn)度:" + (i + 1).ToString() + "/" + Urls.Length.ToString();//進(jìn)度條
}
}
catch (Exception err)
{
ErrorStr.Append("采集錯(cuò)誤：" + err.Message + ";網(wǎng)址：" + Urls[i] + "/r/n");
}
}
da.Update(dt2);
DataBind(dt2);
ShowError(ErrorStr.ToString());
}
/// <summary>
/// ASPX頁(yè)面生成靜態(tài)Html頁(yè)面，作者：鄭少群
/// </summary>
public static string GetHtml(string url)
{
StreamReader sr = null;
string str = null;
//讀取遠(yuǎn)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(wǎng)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(guò)濾重復(fù)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(chuàng)建文件流對(duì)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);
}
}
}
AspNet技術(shù)：asp.net(c#)做一個(gè)網(wǎng)頁(yè)數(shù)據(jù)采集工具，轉(zhuǎn)載需保留來(lái)源！
鄭重聲明：本文版權(quán)歸原作者所有，轉(zhuǎn)載文章僅為傳播更多信息之目的，如作者信息標(biāo)記有誤，請(qǐng)第一時(shí)間聯(lián)系我們修改或刪除，多謝。
