public virtual void TestIsAllCaps()
{
NUnit.Framework.Assert.AreEqual("-allcap", FrenchUnknownWordSignatures.IsAllCaps("YO"));
NUnit.Framework.Assert.AreEqual(string.Empty, FrenchUnknownWordSignatures.IsAllCaps("\\\\"));
NUnit.Framework.Assert.AreEqual(string.Empty, FrenchUnknownWordSignatures.IsAllCaps("0D"));
NUnit.Framework.Assert.AreEqual(string.Empty, FrenchUnknownWordSignatures.IsAllCaps("×"));
NUnit.Framework.Assert.AreEqual("-allcap", FrenchUnknownWordSignatures.IsAllCaps("ÀÅÆÏÜÝÞ"));
}
public virtual void TestHasPunc()
{
NUnit.Framework.Assert.AreEqual("-hpunc", FrenchUnknownWordSignatures.HasPunc("Yes!"));
NUnit.Framework.Assert.AreEqual("-hpunc", FrenchUnknownWordSignatures.HasPunc("["));
NUnit.Framework.Assert.AreEqual("-hpunc", FrenchUnknownWordSignatures.HasPunc("40%"));
NUnit.Framework.Assert.AreEqual(string.Empty, FrenchUnknownWordSignatures.HasPunc("B"));
NUnit.Framework.Assert.AreEqual("-hpunc", FrenchUnknownWordSignatures.HasPunc("BQ_BD"));
NUnit.Framework.Assert.AreEqual(string.Empty, FrenchUnknownWordSignatures.HasPunc("BQBD"));
NUnit.Framework.Assert.AreEqual(string.Empty, FrenchUnknownWordSignatures.HasPunc("0"));
NUnit.Framework.Assert.AreEqual("-hpunc", FrenchUnknownWordSignatures.HasPunc("\\"));
NUnit.Framework.Assert.AreEqual("-hpunc", FrenchUnknownWordSignatures.HasPunc("]aeiou"));
NUnit.Framework.Assert.AreEqual("-hpunc", FrenchUnknownWordSignatures.HasPunc("]"));
NUnit.Framework.Assert.AreEqual("-hpunc", FrenchUnknownWordSignatures.HasPunc("÷"));
NUnit.Framework.Assert.AreEqual(string.Empty, FrenchUnknownWordSignatures.HasPunc("ø"));
}
/// <summary>TODO Can add various signatures, setting the signature via Options.</summary>
/// <param name="word">The word to make a signature for</param>
/// <param name="loc">
/// Its position in the sentence (mainly so sentence-initial
/// capitalized words can be treated differently)
/// </param>
/// <returns>A String that is its signature (equivalence class)</returns>
public override string GetSignature(string word, int loc)
{
string BaseLabel = "UNK";
StringBuilder sb = new StringBuilder(BaseLabel);
switch (unknownLevel)
{
case 1:
{
//Marie's initial attempt
sb.Append(FrenchUnknownWordSignatures.NounSuffix(word));
if (sb.ToString().Equals(BaseLabel))
{
sb.Append(FrenchUnknownWordSignatures.AdjSuffix(word));
if (sb.ToString().Equals(BaseLabel))
{
sb.Append(FrenchUnknownWordSignatures.VerbSuffix(word));
if (sb.ToString().Equals(BaseLabel))
{
sb.Append(FrenchUnknownWordSignatures.AdvSuffix(word));
}
}
}
sb.Append(FrenchUnknownWordSignatures.PossiblePlural(word));
string hasDigit = FrenchUnknownWordSignatures.HasDigit(word);
string isDigit = FrenchUnknownWordSignatures.IsDigit(word);
if (!hasDigit.Equals(string.Empty))
{
if (isDigit.Equals(string.Empty))
{
sb.Append(hasDigit);
}
else
{
sb.Append(isDigit);
}
}
// if(FrenchUnknownWordSignatures.isPunc(word).equals(""))
sb.Append(FrenchUnknownWordSignatures.HasPunc(word));
// else
// sb.append(FrenchUnknownWordSignatures.isPunc(word));
sb.Append(FrenchUnknownWordSignatures.IsAllCaps(word));
if (loc > 0)
{
if (FrenchUnknownWordSignatures.IsAllCaps(word).Equals(string.Empty))
{
sb.Append(FrenchUnknownWordSignatures.IsCapitalized(word));
}
}
//Backoff to suffix if we haven't matched anything else
if (unknownSuffixSize > 0 && sb.ToString().Equals(BaseLabel))
{
int min = word.Length < unknownSuffixSize ? word.Length : unknownSuffixSize;
sb.Append('-').Append(Sharpen.Runtime.Substring(word, word.Length - min));
}
break;
}
default:
{
System.Console.Error.Printf("%s: Invalid unknown word signature! (%d)%n", this.GetType().FullName, unknownLevel);
break;
}
}
return(sb.ToString());
}