/// <summary>
/// Recog the charset from byte array.
/// </summary>
/// <param name="bytes">the byte array.</param>
/// <param name="language">the language.</param>
/// <param name="maxLength">max length per time. the default is 1024, -1 to without limit.</param>
/// <returns>charset string, will be empty when can't recog.</returns>
public static string RecogCharset(byte[] bytes, NChardetLanguage language = NChardetLanguage.ALL, int maxLength = 1024)
{
if (bytes == null || bytes.Length == 0)
return null;
PSMDetector detector = new PSMDetector(language);
string charset = String.Empty;
if (maxLength > 0)
{
int count = 0;
do
{
var tempBytes = bytes.Skip(maxLength * count).Take(maxLength);
if (tempBytes == null || tempBytes.Count() == 0)
break;
detector.HandleData(tempBytes.ToArray(), tempBytes.Count(), ref charset);
if (!string.IsNullOrEmpty(charset))
break;
count++;
}
while (true);
}
else
detector.HandleData(bytes, bytes.Length, ref charset);
return charset;
}
/// <summary>
/// Recog the Encoding from byte array.
/// </summary>
/// <param name="bytes">the byte array.</param>
/// <param name="language">the language.</param>
/// <returns>charset string, will be empty when can't recog.</returns>
public static Encoding RecogEncoding(byte[] bytes, NChardetLanguage language = NChardetLanguage.ALL)
{
string charset = RecogCharset(bytes, language);
if (!string.IsNullOrEmpty(charset))
return Encoding.GetEncoding(charset);
return Encoding.Default;
}
/// <summary>
/// Recog the Encoding from byte array.
/// </summary>
/// <param name="bytes">the byte array.</param>
/// <param name="language">the language.</param>
/// <returns>charset string, will be empty when can't recog.</returns>
public static Encoding RecogEncoding(byte[] bytes, NChardetLanguage language = NChardetLanguage.ALL)
{
string charset = RecogCharset(bytes, language);
if (!string.IsNullOrEmpty(charset))
{
return(Encoding.GetEncoding(charset));
}
return(Encoding.Default);
}
/// <summary>
/// Recog the charset from byte array.
/// </summary>
/// <param name="bytes">the byte array.</param>
/// <param name="language">the language.</param>
/// <param name="maxLength">max length per time. the default is 1024, -1 to without limit.</param>
/// <returns>charset string, will be empty when can't recog.</returns>
public static string RecogCharset(byte[] bytes, NChardetLanguage language = NChardetLanguage.ALL, int maxLength = 1024)
{
if (bytes == null || bytes.Length == 0)
{
return(null);
}
PSMDetector detector = new PSMDetector(language);
string charset = String.Empty;
if (maxLength > 0)
{
int count = 0;
do
{
var tempBytes = bytes.Skip(maxLength * count).Take(maxLength);
if (tempBytes == null || tempBytes.Count() == 0)
{
break;
}
detector.HandleData(tempBytes.ToArray(), tempBytes.Count(), ref charset);
if (!string.IsNullOrEmpty(charset))
{
break;
}
count++;
}while (true);
}
else
{
detector.HandleData(bytes, bytes.Length, ref charset);
}
return(charset);
}
protected void initVerifiers(NChardetLanguage currVerSet)
{
//int idx = 0 ;
NChardetLanguage currVerifierSet;
if (currVerSet >= 0 && currVerSet < NChardetLanguage.NO_OF_LANGUAGES)
{
currVerifierSet = currVerSet;
}
else {
currVerifierSet = NChardetLanguage.ALL;
}
mVerifier = null;
mStatisticsData = null;
if (currVerifierSet == NChardetLanguage.TRADITIONAL_CHINESE)
{
mVerifier = new Verifier[] {
new UTF8Verifier(),
new BIG5Verifier(),
new ISO2022CNVerifier(),
new EUCTWVerifier(),
new CP1252Verifier(),
new UCS2BEVerifier(),
new UCS2LEVerifier()
};
mStatisticsData = new EUCStatistics[] {
null,
new Big5Statistics(),
null,
new EUCTWStatistics(),
null,
null,
null
};
}
//==========================================================
else if (currVerifierSet == NChardetLanguage.KOREAN)
{
mVerifier = new Verifier[] {
new UTF8Verifier(),
new EUCKRVerifier(),
new ISO2022KRVerifier(),
new CP1252Verifier(),
new UCS2BEVerifier(),
new UCS2LEVerifier()
};
}
//==========================================================
else if (currVerifierSet == NChardetLanguage.SIMPLIFIED_CHINESE)
{
mVerifier = new Verifier[] {
new UTF8Verifier(),
new GB2312Verifier(),
new GB18030Verifier(),
new ISO2022CNVerifier(),
new HZVerifier(),
new CP1252Verifier(),
new UCS2BEVerifier(),
new UCS2LEVerifier()
};
}
//==========================================================
else if (currVerifierSet == NChardetLanguage.JAPANESE)
{
mVerifier = new Verifier[] {
new UTF8Verifier(),
new SJISVerifier(),
new EUCJPVerifier(),
new ISO2022JPVerifier(),
new CP1252Verifier(),
new UCS2BEVerifier(),
new UCS2LEVerifier()
};
}
//==========================================================
else if (currVerifierSet == NChardetLanguage.CHINESE)
{
mVerifier = new Verifier[] {
new UTF8Verifier(),
new GB2312Verifier(),
new GB18030Verifier(),
new BIG5Verifier(),
new ISO2022CNVerifier(),
new HZVerifier(),
new EUCTWVerifier(),
new CP1252Verifier(),
new UCS2BEVerifier(),
new UCS2LEVerifier()
};
mStatisticsData = new EUCStatistics[] {
null,
new GB2312Statistics(),
null,
new Big5Statistics(),
null,
null,
new EUCTWStatistics(),
null,
null,
null
};
}
//==========================================================
else if (currVerifierSet == NChardetLanguage.ALL)
{
mVerifier = new Verifier[] {
new UTF8Verifier(),
new SJISVerifier(),
new EUCJPVerifier(),
new ISO2022JPVerifier(),
new EUCKRVerifier(),
new ISO2022KRVerifier(),
new BIG5Verifier(),
new EUCTWVerifier(),
new GB2312Verifier(),
new GB18030Verifier(),
new ISO2022CNVerifier(),
new HZVerifier(),
new CP1252Verifier(),
new UCS2BEVerifier(),
new UCS2LEVerifier()
};
mStatisticsData = new EUCStatistics[] {
null,
null,
new EUCJPStatistics(),
null,
new EUCKRStatistics(),
null,
new Big5Statistics(),
new EUCTWStatistics(),
new GB2312Statistics(),
null,
null,
null,
null,
null,
null
};
}
mClassRunSampler = (mStatisticsData != null);
mClassItems = mVerifier.Length;
}
public PSMDetector(NChardetLanguage langFlag)
{
initVerifiers(langFlag);
Reset();
}
protected void initVerifiers(NChardetLanguage currVerSet)
{
//int idx = 0 ;
NChardetLanguage currVerifierSet;
if (currVerSet >= 0 && currVerSet < NChardetLanguage.NO_OF_LANGUAGES)
{
currVerifierSet = currVerSet;
}
else
{
currVerifierSet = NChardetLanguage.ALL;
}
mVerifier = null;
mStatisticsData = null;
//==========================================================
if (currVerifierSet == NChardetLanguage.CHINESE)
{
mVerifier = new Verifier[] {
new UTF8Verifier(),
new GB2312Verifier(),
new UCS2BEVerifier(),
new UCS2LEVerifier()
};
mStatisticsData = new EUCStatistics[] {
null,
new GB2312Statistics(),
null,
null
};
}
//==========================================================
else if (currVerifierSet == NChardetLanguage.ALL)
{
mVerifier = new Verifier[] {
new UTF8Verifier(),
new SJISVerifier(),
new EUCJPVerifier(),
new ISO2022JPVerifier(),
new EUCKRVerifier(),
new ISO2022KRVerifier(),
new BIG5Verifier(),
new EUCTWVerifier(),
new GB2312Verifier(),
new GB18030Verifier(),
new ISO2022CNVerifier(),
new HZVerifier(),
new CP1252Verifier(),
new UCS2BEVerifier(),
new UCS2LEVerifier()
};
mStatisticsData = new EUCStatistics[] {
null,
null,
new EUCJPStatistics(),
null,
new EUCKRStatistics(),
null,
new Big5Statistics(),
new EUCTWStatistics(),
new GB2312Statistics(),
null,
null,
null,
null,
null,
null
};
}
mClassRunSampler = (mStatisticsData != null);
mClassItems = mVerifier.Length;
}
public PSMDetector(NChardetLanguage langFlag)
{
initVerifiers(langFlag);
Reset();
}
