把做工程过程中比较常用的内容记录起来，如下资料是关于C#通过正则表达式抓取网页信息的类的内容。
using System; using System.Data; using System.Configuration; using System.Net; using System.IO; using System.Text; using System.Collections.Generic; using System.Text.RegularExpressions; using System.Threading; using System.Web; using System.Web.UI.MobileControls; public class WebPage { #region 私有成员
#endregion
#region 属性
public string URL
{
get
{
return m_uri.AbsoluteUri;
}
}
public string Title
{
get
{
if (m_title == "")
{
Match mc = reg.Match(m_html);
if (mc.Success)
m_title = mc.Groups["title"].Value.Trim();
}
return m_title;
}
}
public string M_html
{
get
{
if (m_html == null)
{
m_html = "";
}
return m_html;
}
}
public List<Link> Links
{
get
{
if (m_links.Count == 0) getLinks();
return m_links;
}
}
public string Context
{
get
{
if (m_outstr == "") getContext(Int16.MaxValue);
return m_outstr;
}
}
public int PageSize
{
get
{
return m_pagesize;
}
}
public List<Link> InsiteLinks
{
get
{
}
}
public bool IsGood
{
get
{
return m_good;
}
}
public string Host
{
get
{
return m_uri.Host;
}
}
#endregion
private List<Link> getLinks()
{
if (m_links.Count == 0)
{
Regex[] regex = new Regex[2];
for (int i = 0; i < 2; i++)
{
Match match = regex[i].Match(m_html);
while (match.Success)
{
try
{
string url = HttpUtility.UrlDecode(new Uri(m_uri, match.Groups["URL"].Value).AbsoluteUri);
string text = "";
if (i == 0) text = new Regex("(<[^>]+>)|(\s)|( )|&|"", RegexOptions.Multiline | RegexOptions.IgnoreCase).Replace(match.Groups["text"].Value, "");
Link link = new Link();
link.Text = text;
link.NavigateUrl = url;
m_links.Add(link);
}
catch (Exception ex) { Console.WriteLine(ex.Message); };
match = match.NextMatch();
}
}
}
return m_links;
}
private string getFirstNchar(string instr, int firstN, bool withLink)
{
if (m_outstr == "")
{
m_outstr = instr.Clone() as string;
Regex objReg = new System.Text.RegularExpressions.Regex("(<[^>]+?>)| ", RegexOptions.Multiline | RegexOptions.IgnoreCase);
m_outstr = objReg.Replace(m_outstr, "");
Regex objReg2 = new System.Text.RegularExpressions.Regex("(\s)+", RegexOptions.Multiline | RegexOptions.IgnoreCase);
m_outstr = objReg2.Replace(m_outstr, " ");
}
return m_outstr.Length > firstN ? m_outstr.Substring(0, firstN) : m_outstr;
}
#region 公有文法
public string getContext(int firstN)
{
return getFirstNchar(m_html, firstN, true);
}
public List<Link> getSpecialLinksByUrl(string pattern, int count)
{
if (m_links.Count == 0) getLinks();
List<Link> SpecialLinks = new List<Link>();
List<Link>.Enumerator i;
i = m_links.GetEnumerator();
int cnt = 0;
while (i.MoveNext() && cnt < count)
{
if (new Regex(pattern, RegexOptions.Multiline | RegexOptions.IgnoreCase).Match(i.Current.NavigateUrl).Success)
{
SpecialLinks.Add(i.Current);
cnt++;
}
}
return SpecialLinks;
}
public List<Link> getSpecialLinksByText(string pattern, int count)
{
if (m_links.Count == 0) getLinks();
List<Link> SpecialLinks = new List<Link>();
List<Link>.Enumerator i;
i = m_links.GetEnumerator();
int cnt = 0;
while (i.MoveNext() && cnt < count)
{
if (new Regex(pattern, RegexOptions.Multiline | RegexOptions.IgnoreCase).Match(i.Current.Text).Success)
{
SpecialLinks.Add(i.Current);
cnt++;
}
}
return SpecialLinks;
}
public string getSpecialWords(string pattern)
{
if (m_outstr == "") getContext(Int16.MaxValue);
Regex regex = new Regex(pattern, RegexOptions.Multiline | RegexOptions.IgnoreCase);
Match mc = regex.Match(m_outstr);
if (mc.Success)
return mc.Groups[1].Value;
return string.Empty;
}
#endregion
#region 构造函数
private void Init(string _url)
{
try
{
m_uri = new Uri(_url);
m_links = new List<Link>();
m_html = "";
m_outstr = "";
m_title = "";
m_good = true;
if (_url.EndsWith(".rar") || _url.EndsWith(".dat") || _url.EndsWith(".msi"))
{
m_good = false;
return;
}
HttpWebRequest rqst = (HttpWebRequest)WebRequest.Create(m_uri);
rqst.AllowAutoRedirect = true;
rqst.MaximumAutomaticRedirections = 3;
rqst.UserAgent = "Mozilla/4.0 (compatible; MSIE 5.01; Windows NT 5.0)";
rqst.KeepAlive = true;
rqst.Timeout = 10000;
lock (WebPage.webcookies)
{
if (WebPage.webcookies.ContainsKey(m_uri.Host))
rqst.CookieContainer = WebPage.webcookies[m_uri.Host];
else
{
CookieContainer cc = new CookieContainer();
WebPage.webcookies[m_uri.Host] = cc;
rqst.CookieContainer = cc;
}
}
HttpWebResponse rsps = (HttpWebResponse)rqst.GetResponse();
Stream sm = rsps.GetResponseStream();
if (!rsps.ContentType.ToLower().StartsWith("text/") || rsps.ContentLength > 1 << 22)
{
rsps.Close();
m_good = false;
return;
}
Encoding cding = System.Text.Encoding.Default;
string contenttype = rsps.ContentType.ToLower();
int ix = contenttype.IndexOf("charset=");
if (ix != -1)
{
try
{
cding = System.Text.Encoding.GetEncoding(rsps.ContentType.Substring(ix + "charset".Length + 1));
}
catch
{
cding = Encoding.Default;
}
m_html = new StreamReader(sm, cding).ReadToEnd();
}
else
{
m_html = new StreamReader(sm, cding).ReadToEnd();
Regex regex = new Regex("charset=(?<cding>[^=]+)?"", RegexOptions.IgnoreCase);
string strcding = regex.Match(m_html).Groups["cding"].Value;
try
{
cding = Encoding.GetEncoding(strcding);
}
catch
{
cding = Encoding.Default;
}
byte[] bytes = Encoding.Default.GetBytes(m_html.ToCharArray());
m_html = cding.GetString(bytes);
if (m_html.Split('?').Length > 100)
{
m_html = Encoding.Default.GetString(bytes);
}
}
m_pagesize = m_html.Length;
m_uri = rsps.ResponseUri;
rsps.Close();
}
catch (Exception ex)
{
}
}
public WebPage(string _url)
{
string uurl = "";
try
{
uurl = Uri.UnescapeDataString(_url);
_url = uurl;
}
catch { };
Init(_url);
}
#endregion
}
调用演示
WebPage webInfo = new WebPage("网址");
...参考属性
