/// <summary>
/// 自动检测内容的charset,通过字符集和html中指定的contentType来获取
/// 如果无法获得字符集，则返回null
/// </summary>
/// <param name="stringcontent"></param>
/// <returns></returns>
public static Encoding DetectCharset(QuickWebResponse response, byte[] bytes)
{
string regexstr = "(text/html|text/xml).*charset=(?<charset>\\w+\\-*\\d*)";
Regex regex = new Regex(regexstr, RegexOptions.IgnoreCase);
string contentType = response.ContentTypeStr;
if (contentType != null)
{
Match match1 = regex.Match(contentType);
if (match1.Success)
{
string charset = match1.Groups["charset"].Value.ToUpper();
Encoding encoder = System.Text.Encoding.GetEncoding(charset);
if (encoder != null)
{
return encoder;
}
}
}
string ascii = System.Text.Encoding.ASCII.GetString(bytes);
//<META http-equiv="Content-Type" content="text/html; charset=GB2312">
//Content-Type=text/html;
Match match = regex.Match(ascii);
if ( match.Success )
{
string charset = match.Groups["charset"].Value.ToUpper();
return System.Text.Encoding.GetEncoding(charset);
}
else {
return null;
}
}
public QuickWebResponse GetResponse(int Timeout)
{
if (response == null || !response.SocketConnected)
{
response = new QuickWebResponse(Timeout);
response.ConnectAndGetHeader(this);
}
else
{
response.ConnectAndGetHeader(this);
}
return(response);
}
/// <summary>
/// 自动检测内容的charset,通过字符集和html中指定的contentType来获取
/// 如果无法获得字符集，则返回null
/// </summary>
/// <param name="stringcontent"></param>
/// <returns></returns>
public static Encoding DetectCharset(QuickWebResponse response, byte[] bytes)
{
string regexstr = "(text/html|text/xml).*charset=(?<charset>\\w+\\-*\\d*)";
Regex regex = new Regex(regexstr, RegexOptions.IgnoreCase);
string contentType = response.ContentTypeStr;
if (contentType != null)
{
Match match1 = regex.Match(contentType);
if (match1.Success)
{
string charset = match1.Groups["charset"].Value.ToUpper();
Encoding encoder = System.Text.Encoding.GetEncoding(charset);
if (encoder != null)
{
return(encoder);
}
}
}
string ascii = System.Text.Encoding.ASCII.GetString(bytes);
//<META http-equiv="Content-Type" content="text/html; charset=GB2312">
//Content-Type=text/html;
Match match = regex.Match(ascii);
if (match.Success)
{
string charset = match.Groups["charset"].Value.ToUpper();
return(System.Text.Encoding.GetEncoding(charset));
}
else
{
return(null);
}
}
public QuickWebResponse GetResponse(int Timeout)
{
if (response == null || !response.SocketConnected)
{
response = new QuickWebResponse(Timeout);
response.ConnectAndGetHeader(this);
}
else
{
response.ConnectAndGetHeader(this);
}
return response;
}
private string ReadUrlContent(ref string url, bool checkredirect, bool checkHtmlRedurect,int timeoutsecond)
{
string content = String.Empty;
try
{
Uri uri = new Uri(url);
gRequest = QuickWebRequest.Create(uri, null, gRequest, false);
gResponse = gRequest.GetResponse(timeoutsecond);
if (checkredirect)
{
if (gResponse.RedirectUri != null && gResponse.RedirectUri.AbsoluteUri != url)
{
return ReadUrlContent(gResponse.RedirectUri.AbsoluteUri, false, checkHtmlRedurect);
}
}
byte[] bytes = gResponse.ReadResponse();
if (bytes == null)
{
return null;
}
System.Text.Encoding charset = HttpUtils.DetectCharset(gResponse, bytes);
if (charset == null)
{ //默认使用GB2312
charset = System.Text.Encoding.GetEncoding("GB2312");
}
// 使用指定的编码
if (!string.IsNullOrEmpty(Encoding))
{
charset = System.Text.Encoding.GetEncoding(Encoding);
}
//完成到UTF8的编码转换
if (charset != System.Text.Encoding.UTF8)
{
bytes = System.Text.Encoding.Convert(charset, System.Text.Encoding.UTF8, bytes);
charset = System.Text.Encoding.UTF8;
}
content = charset.GetString(bytes);
//检查一次Html重定向
if (checkHtmlRedurect)
{
System.Text.RegularExpressions.Regex redirectRegex = new System.Text.RegularExpressions.Regex("<META\\s+HTTP-EQUIV\\s*=\\s*[\"]*Refresh[\"]*\\s+CONTENT=[\"\\s]*\\d+\\s*[;]\\s*URL=(?<match>.*?)[\\s\"]*>", System.Text.RegularExpressions.RegexOptions.Singleline | System.Text.RegularExpressions.RegexOptions.IgnoreCase);
System.Text.RegularExpressions.Match match = redirectRegex.Match(content);
if (match.Success)
{
if (!string.IsNullOrEmpty(match.Groups["match"].Value) && match.Groups["match"].Value != url)
{
url = match.Groups["match"].Value;
return ReadUrlContent(url, checkredirect, false);
}
}
}
}
catch
{
//暂时不做处理。
}
finally
{
gResponse.Close();
}
return content;
}
private string ReadUrlContent(ref string url, bool checkredirect, bool checkHtmlRedurect, int timeoutsecond)
{
string content = String.Empty;
try
{
Uri uri = new Uri(url);
gRequest = QuickWebRequest.Create(uri, null, gRequest, false);
gResponse = gRequest.GetResponse(timeoutsecond);
if (checkredirect)
{
if (gResponse.RedirectUri != null && gResponse.RedirectUri.AbsoluteUri != url)
{
return(ReadUrlContent(gResponse.RedirectUri.AbsoluteUri, false, checkHtmlRedurect));
}
}
byte[] bytes = gResponse.ReadResponse();
if (bytes == null)
{
return(null);
}
System.Text.Encoding charset = HttpUtils.DetectCharset(gResponse, bytes);
if (charset == null)
{ //默认使用GB2312
charset = System.Text.Encoding.GetEncoding("GB2312");
}
// 使用指定的编码
if (!string.IsNullOrEmpty(Encoding))
{
charset = System.Text.Encoding.GetEncoding(Encoding);
}
//完成到UTF8的编码转换
if (charset != System.Text.Encoding.UTF8)
{
bytes = System.Text.Encoding.Convert(charset, System.Text.Encoding.UTF8, bytes);
charset = System.Text.Encoding.UTF8;
}
content = charset.GetString(bytes);
//检查一次Html重定向
if (checkHtmlRedurect)
{
System.Text.RegularExpressions.Regex redirectRegex = new System.Text.RegularExpressions.Regex("<META\\s+HTTP-EQUIV\\s*=\\s*[\"]*Refresh[\"]*\\s+CONTENT=[\"\\s]*\\d+\\s*[;]\\s*URL=(?<match>.*?)[\\s\"]*>", System.Text.RegularExpressions.RegexOptions.Singleline | System.Text.RegularExpressions.RegexOptions.IgnoreCase);
System.Text.RegularExpressions.Match match = redirectRegex.Match(content);
if (match.Success)
{
if (!string.IsNullOrEmpty(match.Groups["match"].Value) && match.Groups["match"].Value != url)
{
url = match.Groups["match"].Value;
return(ReadUrlContent(url, checkredirect, false));
}
}
}
}
catch
{
//暂时不做处理。
}
finally
{
gResponse.Close();
}
return(content);
}
