public static string GetContent(string sOriContent, string sOtherRemoveReg, string sPageUrl)
{
string sFormartted = sOriContent;
sFormartted = Regex.Replace(sFormartted, @"<script[\s\S]*?</script>", "", RegexOptions.Multiline | RegexOptions.IgnorePatternWhitespace | RegexOptions.IgnoreCase);
sFormartted = Regex.Replace(sFormartted, @"<iframe[^>]*>[\s\S]*?</iframe>", "", RegexOptions.Multiline | RegexOptions.IgnorePatternWhitespace | RegexOptions.IgnoreCase);
string[] sOtherReg = sOtherRemoveReg.Split(new string[] { "\r\n" }, StringSplitOptions.RemoveEmptyEntries);
foreach (string sRemoveReg in sOtherReg)
{
sFormartted = CRegex.Replace(sFormartted, sRemoveReg, "", 0);
}
Regex re = new Regex("<img[^>]+src=\\s*(?:'(?<src>[^']+)'|\"(?<src>[^\"]+)\"|(?<src>[^>\\s]+))\\s*[^>]*>", RegexOptions.Multiline | RegexOptions.IgnorePatternWhitespace | RegexOptions.IgnoreCase);
MatchCollection mcs = re.Matches(sFormartted);
string sOriStr = "";
string sReplaceStr = "";
foreach (Match mc in mcs)
{
sOriStr = mc.Value;
sReplaceStr = sOriStr.Replace(mc.Groups["src"].Value, CRegex.GetUrl(sPageUrl, mc.Groups["src"].Value));
sFormartted = sFormartted.Replace(sOriStr, sReplaceStr);
}
re = new Regex(@"<a[^>]+href=\s*(?:'(?<href>[^']+)'|""(?<href>[^""]+)""|(?<href>[^>\s]+))\s*[^>]*>", RegexOptions.Multiline | RegexOptions.IgnorePatternWhitespace | RegexOptions.IgnoreCase);
mcs = re.Matches(sFormartted);
sOriStr = "";
sReplaceStr = "";
foreach (Match mc in mcs)
{
sOriStr = mc.Value;
sReplaceStr = CRegex.Replace(sOriStr, @"<[^>]*\ba\b[^>]*>", "", 0);
sFormartted = sFormartted.Replace(sOriStr, sReplaceStr);
}
return(sFormartted);
}
public static List <CookieObj> GetCookieList(string sInput, List <CookieObj> listInput)
{
string strCookie = CRegex.GetText(sInput, @"Set-Cookie:(?<Cookie>[\s\S]+?)\n", "Cookie");
strCookie = CRegex.Replace(strCookie, @"expires=([^;]+)GMT;", "", 0);
strCookie = strCookie.Replace("path=/", "");
List <string> list = CRegex.GetList(strCookie, @"(?<cookie>[\w\d\&\.=]+[\w\d\._-]+);", "cookie");
string cookieName, cookieValue;
CookieObj mCookie = null;
List <CookieObj> listCookie = new List <CookieObj>();
foreach (string s in list)
{
if (s.IndexOf('=') < 1)
{
continue;
}
cookieName = s.Substring(0, s.IndexOf('='));
if (cookieName == "domain")
{
continue;
}
if (s.Length < s.IndexOf('=') + 1)
{
cookieValue = "";
}
else
{
cookieValue = s.Substring(s.IndexOf('=') + 1);
}
mCookie = new CookieObj();
mCookie.cookieName = cookieName;
mCookie.cookieValue = cookieValue;
listCookie.Add(mCookie);
}
bool blExists;
CookieObj mInput;
foreach (CookieObj model in listCookie)
{
blExists = false;
for (int i = 0; i < listInput.Count; i++)
{
mInput = listInput[i];
if (mInput.cookieName == model.cookieName)
{
blExists = true;
mInput.cookieValue = model.cookieValue;
}
}
if (!blExists)
{
listInput.Add(model);
}
}
return(listInput);
}
public static string ClearScript(string sDetail, bool blLink)
{
sDetail = CRegex.Replace(sDetail, @"<script([\s\S]+?)((</script>)|(/>))", "", 0);
sDetail = CRegex.Replace(sDetail, @"<iframe([\s\S]+?)((</iframe>)|(/>))", "", 0);
if (blLink)
{
sDetail = CRegex.Replace(sDetail, @"<a([\s\S]+?)((</a>)|(/>))", "", 0);
}
return(sDetail);
}
public static string GetCookieByHead(string sInput)
{
string strCookie = CRegex.GetText(sInput, @"Set-Cookie:(?<Cookie>[\s\S]+?)\n", "Cookie");
List <string> list = CRegex.GetList(strCookie, @"(?<cookie>[\w\d\.]+=[\w\d\._-]+);", "cookie");
string sReturnCookie = "";
foreach (string s in list)
{
sReturnCookie += s + "; ";
}
return(sReturnCookie);
}
/// <summary>
///
/// </summary>
/// <param name="sHtml"></param>
/// <param name="bClear"></param>
public static string ClearTag(string sHtml, bool bClear)
{
if (sHtml == "")
{
return("");
}
sHtml = CRegex.Replace(sHtml, CRegex.sIFrameReg, "", 0);
sHtml = CRegex.Replace(sHtml, CRegex.sScriptReg, "", 0);
sHtml = CRegex.Replace(sHtml, @"(<[^>\s]*\b(\w)+\b[^>]*>)|([\s]+)|(<>)|( )", "", 0);
sHtml = sHtml.Replace("\"", "").Replace("<", "").Replace(">", "");
return(sHtml);
}
public static string JsToHtml(string strJS)
{
string sReturn = strJS.Replace("document.writeln(\"", "");
sReturn = sReturn.Replace("document.write(\"", "");
sReturn = sReturn.Replace("document.write('", "");
sReturn = CRegex.Replace(sReturn, @"(?<backslash>\\)[^\\]", "", "backslash");
sReturn = sReturn.Replace(@"\\", @"\");
sReturn = sReturn.Replace("/\\\\\\", "\\");
sReturn = sReturn.Replace("/\\\\\\'", "\\'");
sReturn = sReturn.Replace("/\\\\\\//", "\\/");
sReturn = sReturn.Replace("\");", "");
sReturn = sReturn.Replace("\")", "");
sReturn = sReturn.Replace("');", "");
return(sReturn);
}
public static string ClearTag(string sHtml)
{
if (sHtml == "")
{
return("");
}
sHtml = CRegex.Replace(sHtml, CRegex.sIFrameReg, "", 0);
sHtml = CRegex.Replace(sHtml, CRegex.sScriptReg, "", 0);
sHtml = Regex.Replace(sHtml, @"<[^>]*>| ", string.Empty, RegexOptions.IgnoreCase);
sHtml = Regex.Replace(sHtml, @"\s\s", " ", RegexOptions.IgnoreCase);
sHtml = Regex.Replace(sHtml, @"\s\s", " ", RegexOptions.IgnoreCase);
sHtml = Regex.Replace(sHtml, @"\s\s", " ", RegexOptions.IgnoreCase);
sHtml = Regex.Replace(sHtml, "“", "“", RegexOptions.IgnoreCase);
sHtml = Regex.Replace(sHtml, "”", "”", RegexOptions.IgnoreCase);
return(sHtml);
}
/// <summary>
/// 网页内容
/// </summary>
/// <param name="sInput">输入内容</param>
/// <param name="sRegex">表达式字符串</param>
public static string GetHtml(string sInput)
{
return(CRegex.Replace(sInput, @"(?<Head>[^<]+)<", "", "Head"));
}
public static string GetPost(RequestArgs mArgs)
{
//(1)创建IPEndPoint实例和套接字
IPAddress hostIp = Dns.GetHostEntry(mArgs.IpAddress).AddressList[0];
IPEndPoint ep = new IPEndPoint(hostIp, 80);
Socket client = new Socket(AddressFamily.InterNetwork, SocketType.Stream, ProtocolType.Tcp);
//(2)连接服务器端
try
{
client.Connect(ep);
}
catch (SocketException e)
{
Console.WriteLine(e.Message);
}
StringBuilder sbRequest = new StringBuilder();
Uri u = new Uri(mArgs.Url);
sbRequest.AppendLine(string.Format("{0} {1} HTTP/1.1", mArgs.Method, u.PathAndQuery));
if (string.IsNullOrEmpty(mArgs.Accept))
{
sbRequest.AppendLine("Accept: image/gif, image/jpeg, image/pjpeg, application/x-shockwave-flash, application/x-ms-application, application/vnd.ms-xpsdocument, application/xaml+xml, application/x-ms-xbap, application/vnd.ms-excel, application/vnd.ms-powerpoint, application/msword, */*");
}
else
{
sbRequest.AppendLine(string.Format("Accept: {0}", mArgs.Accept));
}
sbRequest.AppendLine("Accept-Language: zh-cn");
sbRequest.AppendLine("User-Agent: Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 6.0; Trident/4.0; SLCC1; .NET CLR 2.0.50727; .NET CLR 3.5.30729; .NET CLR 3.0.30618; InfoPath.2)");
sbRequest.AppendLine("Accept-Encoding: gzip, deflate");
sbRequest.AppendLine(string.Format("Host: {0}", u.Host));
if (!string.IsNullOrEmpty(mArgs.RefererUrl))
{
sbRequest.AppendLine(string.Format("Referer: {0}", mArgs.RefererUrl));
}
if (!string.IsNullOrEmpty(mArgs.ContentType))
{
sbRequest.AppendLine(string.Format("Content-Type: {0}", mArgs.ContentType));
}
sbRequest.AppendLine("Connection: Keep-Alive");
if (mArgs.Method == "POST")
{
sbRequest.AppendLine(string.Format("Content-Length: {0}", mArgs.postData.Length));
}
sbRequest.AppendLine("");
if (mArgs.Method == "POST")
{
sbRequest.AppendLine(mArgs.postData);
sbRequest.AppendLine("");
}
//(3)发送请求
client.Send(Encoding.ASCII.GetBytes(sbRequest.ToString()));
//(4)接收数据
StringBuilder recstr = new StringBuilder();
byte[] buff = new byte[1024 * 3];
int rCount = 0;
while (true)
{
rCount = client.Receive(buff, buff.Length, SocketFlags.None); //读取数据
if (rCount > 0)
{
recstr.Append(Encoding.GetEncoding(mArgs.Encode).GetString(buff, 0, rCount));
System.Threading.Thread.Sleep(50);
}
if (rCount <= buff.Length)
{
break;
}
}
client.Close();
if (mArgs.blGetHeaders)
{
return(recstr.ToString());
}
else
{
return(CRegex.Replace(recstr.ToString(), @"^HTTP[\s\S]+?(\r\n){2,}", "", 0));
}
}
