.net中网页抓取数据（提取html中的数据，提取table中的数据）

最新推荐文章于 2021-06-15 11:04:09 发布

TianGaojie123abc

最新推荐文章于 2021-06-15 11:04:09 发布

阅读量5.2k

点赞数

文章标签： html .net table string

本文链接：https://blog.csdn.net/TianGaojie123abc/article/details/8101268

版权

方法一：

WebRequest request = WebRequest.Create("http://www.cftea.com/");

WebResponse response = request.GetResponse();
StreamReader reader = new StreamReader(response.GetResponseStream(), Encoding.GetEncoding("gb2312"));

var contents = reader.ReadToEnd();

Console.WriteLine(StripHTML(contents));

reader.Close();
reader.Dispose();
response.Close();

Console.Read();

方法二：(抓取html中table里面的数据)

string html = @"<html>
<head>
<title></title>
</head>
<body>
<table>a</table>
<table>b</table>
<table>c</table>
<table>d</table>
<table>e</table>
</body>
</html>
";

var strReg = @"(?is)(?<=<table>).+?(?=</table>)";
List<string> result = new List<string>();
MatchCollection mc = Regex.Matches(html, strReg);
foreach (Match m in mc)
{
//result.Add(m.Value);
Console.WriteLine(m.Value);
}

方法三：

var htmlDocument = new HtmlDocument();
htmlDocument.LoadHtml(orgStr);
var tables = htmlDocument.DocumentNode.SelectNodes("//table");//xpath的写法

foreach (var table in tables)
{
foreach (var tr in table.SelectNodes("//tr"))
{
var collegeName = tr.SelectNodes("//td").Skip(1).FirstOrDefault().InnerText;
Console.WriteLine(collegeName);
}
}

http://www.cnblogs.com/xinzhyu/archive/2008/12/09/1351434.html

http://msdn.microsoft.com/zh-tw/ee787055.aspx

方法四：（对网页中table里面的数据提取）

#region http://www.gaokao.com/e/20120109/4f0a8e1773aa0.shtml
//var url = "http://www.gaokao.com/e/20120109/4f0a8e1773aa0.shtml";
//var url = "http://www.gaokao.com/e/20120109/4f0a8e1773aa0_2.shtml";
//var url = "http://www.gaokao.com/e/20120109/4f0a8e1773aa0_3.shtml";
//var url = "http://www.gaokao.com/e/20120109/4f0a8e1773aa0_4.shtml";
//var url = "http://www.gaokao.com/e/20120109/4f0a8e1773aa0_5.shtml";
//var url = "http://www.gaokao.com/e/20120109/4f0a8e1773aa0_6.shtml";

//var orgStr = ChinaEduSp.Crawl.HttpUtility.GetContentByUrl(url, "gb2312");

//var htmlDocument = new HtmlDocument();
//htmlDocument.LoadHtml(orgStr);
//var rows = htmlDocument.DocumentNode.SelectNodes("//table//tr");

//foreach (var item in rows)
//{
// var pos = item.SelectSingleNode("td[1]").InnerText;
// var school = item.SelectSingleNode("td[2]").InnerText;
// var province = item.SelectSingleNode("td[3]").InnerText;
// var type = item.SelectSingleNode("td[4]").InnerText;
// var totalScore = item.SelectSingleNode("td[5]").InnerText;
// var seq = db.RankingDescriptions.Count();

// //Response.Write("名次：" + pos + " 学校名称：" + school + " 所在省份：" + province + " 类型：" + type + " 总分：" + totalScore + "/r/n");
// //Response.Write("名次：" + pos + " 学校名称：" + school);

// try
// {
// db.RankingDescriptions.Add(new RankingDescription
// {
// POS = Convert.ToInt32(pos),
// SchoolName = school,
// Province = province,
// Area = province,
// Type = type,
// TotalScore = totalScore,
// IsShow = true,
// IsDelete = false,
// RankId = 0,
// Seq = seq
// });

// db.SaveChanges();

// seq++;
// }
// catch (Exception ex)
// {
// string msg = ex.Message;
// }
//}
#endregion

#region http://www.gaokao.com/e/20120109/4f0a914934baa_2.shtml

//var url = "http://www.gaokao.com/e/20120109/4f0a914934baa_2.shtml";

//var orgStr = ChinaEduSp.Crawl.HttpUtility.GetContentByUrl(url, "gb2312");

//var htmlDocument = new HtmlDocument();
//htmlDocument.LoadHtml(orgStr);
//var rows = htmlDocument.DocumentNode.SelectNodes("//table//tr//td//table//tr");

//foreach (var item in rows)
//{
// var pos = item.SelectSingleNode("td[1]").InnerText;
// var school = item.SelectSingleNode("td[2]").InnerText;
// var province = item.SelectSingleNode("td[3]").InnerText;
// var totalScore = item.SelectSingleNode("td[4]").InnerText;
// var seq = db.RankingDescriptions.Count();

// //Response.Write("名次：" + pos + " 学校名称：" + school + " 所在省份：" + province + " 总分：" + totalScore + "/r/n");
// //Response.Write("名次：" + pos + " 学校名称：" + school);

// try
// {
// db.RankingDescriptions.Add(new RankingDescription
// {
// POS = Convert.ToInt32(pos),
// SchoolName = school,
// Province = province,
// Area = province,
// TotalScore = totalScore,
// IsShow = true,
// IsDelete = false,
// RankId = 0,
// Seq = seq
// });

// db.SaveChanges();

// seq++;
// }
// catch (Exception ex)
// {
// string msg = ex.Message;
// }
//}

#endregion