public LanguageProfile FindMatchingLanguage(string str)
{
Utils.ThrowException(languageProfiles.Count == 0 ? new InvalidOperationException() : null);
Utils.ThrowException(str == null ? new ArgumentNullException("str") : null);
NGramProfile p = new NGramProfile(n);
p.AddTokensFromString(str);
if (p.IsEmpty)
{
return(null);
}
p.DoRanking();
return(FindMatchingLanguage(p));
}
public ArrayList <KeyDat <double, LanguageProfile> > DetectLanguageAll(string str)
{
Utils.ThrowException(mLanguageProfiles.Count == 0 ? new InvalidOperationException() : null);
Utils.ThrowException(str == null ? new ArgumentNullException("str") : null);
NGramProfile p = new NGramProfile(mN);
p.AddTokensFromString(str);
if (p.IsEmpty)
{
return(null);
}
p.DoRanking();
return(DetectLanguageAll(p));
}
public ArrayList <KeyDat <double, LanguageProfile> > DetectLanguageAll(NGramProfile p, int cutOff)
{
Utils.ThrowException(mLanguageProfiles.Count == 0 ? new InvalidOperationException() : null);
Utils.ThrowException(p == null ? new ArgumentNullException("p") : null);
Utils.ThrowException((!p.IsRanked) ? new ArgumentValueException("p") : null);
Utils.ThrowException(cutOff < 1 ? new ArgumentOutOfRangeException("cutOff") : null);
ArrayList <KeyDat <double, LanguageProfile> > result = new ArrayList <KeyDat <double, LanguageProfile> >();
foreach (LanguageProfile l in mLanguageProfiles)
{
double dist = p.CalcOutOfPlace(l, cutOff);
result.Add(new KeyDat <double, LanguageProfile>(dist, l));
}
return(result);
}
public LanguageProfile DetectLanguage(NGramProfile p, int cutOff)
{
Utils.ThrowException(mLanguageProfiles.Count == 0 ? new InvalidOperationException() : null);
Utils.ThrowException(p == null ? new ArgumentNullException("p") : null);
Utils.ThrowException((!p.IsRanked) ? new ArgumentValueException("p") : null);
Utils.ThrowException(cutOff < 1 ? new ArgumentOutOfRangeException("cutOff") : null);
LanguageProfile matchingLang = null;
double minDist = Double.MaxValue;
foreach (LanguageProfile l in mLanguageProfiles)
{
double dist = p.CalcOutOfPlace(l, cutOff);
if (dist < minDist)
{
matchingLang = l;
minDist = dist;
}
}
return(matchingLang);
}
public uint CalcOutOfPlace(NGramProfile p, int cutOff)
{
Utils.ThrowException(p == null ? new ArgumentNullException("p") : null);
Utils.ThrowException(cutOff < 1 ? new ArgumentOutOfRangeException("cutOff") : null);
Utils.ThrowException(!isRanked ? new InvalidOperationException() : null);
int dist = 0;
uint v = 0;
uint sum = 0;
int penalty = Math.Min(cutOff, p.hist.Count);
// sums up distances of each N-gram in this profile to
// the same N-gram in profile 'p'
foreach (KeyValuePair <string, uint> kvp in hist)
{
if (kvp.Value > cutOff)
{
continue;
}
if (p.hist.TryGetValue(kvp.Key, out v))
{
if (v > cutOff)
{
dist = penalty;
}
else
{
dist = (int)Math.Abs(v - kvp.Value);
}
}
else
{
dist = penalty;
}
sum += (uint)dist;
}
return(sum);
}
public uint CalcOutOfPlace(NGramProfile p)
{
return CalcOutOfPlace(p, /*cutOff=*/500); // throws ArgumentNullException, ArgumentOutOfRangeException, InvalidOperationException
}
public uint CalcOutOfPlace(NGramProfile p, int cutOff)
{
Utils.ThrowException(p == null ? new ArgumentNullException("p") : null);
Utils.ThrowException(cutOff < 1 ? new ArgumentOutOfRangeException("cutOff") : null);
Utils.ThrowException(!mIsRanked ? new InvalidOperationException() : null);
int dist = 0;
uint v = 0;
uint sum = 0;
int penalty = Math.Min(cutOff, p.mHist.Count);
// sum up distances from each n-gram in this profile to the same n-gram in profile p
foreach (KeyValuePair<string, uint> kvp in mHist)
{
if (kvp.Value > cutOff) { continue; }
if (p.mHist.TryGetValue(kvp.Key, out v))
{
if (v > cutOff) { dist = penalty; }
else { dist = (int)Math.Abs(v - kvp.Value); }
}
else { dist = penalty; }
sum += (uint)dist;
}
return sum;
}
public ArrayList<KeyDat<double, LanguageProfile>> DetectLanguageAll(string str)
{
Utils.ThrowException(mLanguageProfiles.Count == 0 ? new InvalidOperationException() : null);
Utils.ThrowException(str == null ? new ArgumentNullException("str") : null);
NGramProfile p = new NGramProfile(mN);
p.AddTokensFromString(str);
if (p.IsEmpty) { return null; }
p.DoRanking();
return DetectLanguageAll(p);
}
public ArrayList<KeyDat<double, LanguageProfile>> DetectLanguageAll(NGramProfile p)
{
return DetectLanguageAll(p, /*cutOff=*/500); // throws ArgumentNullException, ArgumentValueException, InvalidOperationException
}
public ArrayList<KeyDat<double, LanguageProfile>> DetectLanguageAll(NGramProfile p, int cutOff)
{
Utils.ThrowException(mLanguageProfiles.Count == 0 ? new InvalidOperationException() : null);
Utils.ThrowException(p == null ? new ArgumentNullException("p") : null);
Utils.ThrowException((!p.IsRanked) ? new ArgumentValueException("p") : null);
Utils.ThrowException(cutOff < 1 ? new ArgumentOutOfRangeException("cutOff") : null);
ArrayList<KeyDat<double, LanguageProfile>> result = new ArrayList<KeyDat<double, LanguageProfile>>();
foreach (LanguageProfile l in mLanguageProfiles)
{
double dist = p.CalcOutOfPlace(l, cutOff);
result.Add(new KeyDat<double, LanguageProfile>(dist, l));
}
return result;
}
public LanguageProfile DetectLanguage(NGramProfile p)
{
return DetectLanguage(p, /*cutOff=*/500); // throws ArgumentNullException, ArgumentValueException, InvalidOperationException
}
public LanguageProfile DetectLanguage(NGramProfile p, int cutOff)
{
Utils.ThrowException(mLanguageProfiles.Count == 0 ? new InvalidOperationException() : null);
Utils.ThrowException(p == null ? new ArgumentNullException("p") : null);
Utils.ThrowException((!p.IsRanked) ? new ArgumentValueException("p") : null);
Utils.ThrowException(cutOff < 1 ? new ArgumentOutOfRangeException("cutOff") : null);
LanguageProfile matchingLang = null;
double minDist = Double.MaxValue;
foreach (LanguageProfile l in mLanguageProfiles)
{
double dist = p.CalcOutOfPlace(l, cutOff);
if (dist < minDist)
{
matchingLang = l;
minDist = dist;
}
}
return matchingLang;
}
public ArrayList <KeyDat <double, LanguageProfile> > DetectLanguageAll(NGramProfile p)
{
return(DetectLanguageAll(p, /*cutOff=*/ 500)); // throws ArgumentNullException, ArgumentValueException, InvalidOperationException
}
public LanguageProfile DetectLanguage(NGramProfile p)
{
return(DetectLanguage(p, /*cutOff=*/ 500)); // throws ArgumentNullException, ArgumentValueException, InvalidOperationException
}