/// <summary>
/// 将Unicode字串\u.\u.格式字串转换为原始字符串
/// </summary>
/// <param name="str">待转换字符串</param>
/// <returns></returns>
public static string UnicodeToString(string str)
{
string outStr = "";
str = CRegex.Replace(str, "[\r\n]", "", 0);
if (!string.IsNullOrEmpty(str))
{
string[] strlist = str.Replace("\\u", "㊣").Split('㊣');
try
{
outStr += strlist[0];
for (int i = 1; i < strlist.Length; i++)
{
string strTemp = strlist[i];
if (!string.IsNullOrEmpty(strTemp) && strTemp.Length >= 4)
{
strTemp = strlist[i].Substring(0, 4);
//将unicode字符转为10进制整数，然后转为char中文字符
outStr += (char)int.Parse(strTemp, System.Globalization.NumberStyles.HexNumber);
outStr += strlist[i].Substring(4);
}
}
}
catch (FormatException ex)
{
outStr += ex.Message;
LogHelper.WriteLog(LogLevel.LOG_LEVEL_CRIT, ex, typeof(UnicodeHelper));
}
}
return(outStr);
}
public static string GetContent(string sOriContent, string sOtherRemoveReg, string sPageUrl, DataTable dtAntiLink)
{
string sFormartted = sOriContent;
//去掉有危险的标记
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);
Regex r = new Regex(@"<input[\s\S]+?>|<form[\s\S]+?>|</form[\s\S]*?>|<select[\s\S]+?>?</select>|<textarea[\s\S]*?>?</textarea>|<file[\s\S]*?>|<noscript>|</noscript>", RegexOptions.IgnoreCase);
sFormartted = r.Replace(sFormartted, "");
string[] sOtherReg = sOtherRemoveReg.Split(new string[] { "\r\n" }, StringSplitOptions.RemoveEmptyEntries);
foreach (string sRemoveReg in sOtherReg)
{
sFormartted = CRegex.Replace(sFormartted, sRemoveReg, "", 0);
}
//图片路径
//sFormartted = _ReplaceUrl("<img[^>]+src\\s*=\\s*(?:'(?<src>[^']+)'|\"(?<src>[^\"]+)\"|(?<src>[^>\\s]+))\\s*[^>]*>", "src", sFormartted,sPageUrl);
sFormartted = _ReplaceUrl("<img[\\s\\S]+?src\\s*=\\s*(?:'(?<src>[^']+)'|\"(?<src>[^\"]+)\"|(?<src>[^>\\s]+))\\s*[^>]*>", "src", sFormartted, sPageUrl);
//反防盗链
string domain = GetDomain(sPageUrl);
DataRow[] drs = dtAntiLink.Select("Domain='" + domain + "'");
if (drs.Length > 0)
{
foreach (DataRow dr in drs)
{
switch (Convert.ToInt32(dr["Type"]))
{
case 1: //置换
sFormartted = sFormartted.Replace(dr["imgUrl"].ToString(), "http://stat.580k.com/t.asp?url=");
break;
default: //附加
sFormartted = sFormartted.Replace(dr["imgUrl"].ToString(), "http://stat.580k.com/t.asp?url=" + dr["imgUrl"].ToString());
break;
}
}
}
//A链接
sFormartted = _ReplaceUrl(@"<a[^>]+href\s*=\s*(?:'(?<href>[^']+)'|""(?<href>[^""]+)""|(?<href>[^>\s]+))\s*[^>]*>", "href", sFormartted, sPageUrl);
//CSS
sFormartted = _ReplaceUrl(@"<link[^>]+href\s*=\s*(?:'(?<href>[^']+)'|""(?<href>[^""]+)""|(?<href>[^>\s]+))\s*[^>]*>", "href", sFormartted, sPageUrl);
//BACKGROUND
sFormartted = _ReplaceUrl(@"background\s*=\s*(?:'(?<img>[^']+)'|""(?<img>[^""]+)""|(?<img>[^>\s]+))", "img", sFormartted, sPageUrl);
//style方式的背景：background-image:url(...)
sFormartted = _ReplaceUrl(@"background-image\s*:\s*url\s*\x28(?<img>[^\x29]+)\x29", "img", sFormartted, sPageUrl);
//FLASH
sFormartted = _ReplaceUrl(@"<param\s[^>]+""movie""[^>]+value\s*=\s*""(?<flash>[^"">]+\x2eswf)""[^>]*>", "flash", sFormartted, sPageUrl);
//XSL
if (IsXml(sFormartted))
{
sFormartted = _ReplaceUrl(@"<\x3fxml-stylesheet\s+[^\x3f>]+href=\s*(?:'(?<href>[^']+)'|""(?<href>[^""]+)"")\s*[^\x3f>]*\x3f>", "href", sFormartted, sPageUrl);
}
//script
//sFormartted = _ReplaceUrl(@"<script[^>]+src\s*=\s*(?:'(?<src>[^']+)'|""(?<src>[^""]+)""|(?<src>[^>\s]+))\s*[^>]*>", "src", sFormartted,sPageUrl);
return(sFormartted);
}
/// <summary>
/// 网页内容
/// </summary>
/// <param name="sInput">输入内容</param>
public static string GetHtml(string sInput)
{
return(CRegex.Replace(sInput, @"(?<Head>[^<]+)<", "", "Head"));
}