/// ------------------------------------------------------------------------------------
/// <summary>
/// Gets the natural type (i.e., not overridden) of the given character code point. This
/// is essentially based on ICU, but we make one exception to force superscripted
/// numbers to be considered as word-forming since these are generally used to mark
/// tone in the vernaculars we work with (see TE-8384).
/// </summary>
/// <param name="codepoint">The codepoint.</param>
/// ------------------------------------------------------------------------------------
protected virtual ValidCharacterType GetNaturalCharType(int codepoint)
{
if (m_cpe.get_IsWordForming(codepoint))
{
return(ValidCharacterType.WordForming);
}
if (Icu.IsNumeric(codepoint))
{
foreach (string chr in s_defaultWordformingChars)
{
if ((int)chr[0] == codepoint)
{
return(ValidCharacterType.WordForming);
}
}
return(ValidCharacterType.Numeric);
}
return(ValidCharacterType.Other);
}
private static void VerifyNonexistentChars()
{
Icu.InitIcuDataDir();
Assert.IsFalse(Icu.IsAlphabetic(kChar1));
Assert.IsFalse(Icu.IsAlphabetic(kChar2));
Assert.IsFalse(Icu.IsAlphabetic(kChar3));
Assert.IsFalse(Icu.IsAlphabetic(kChar4));
Assert.IsFalse(Icu.IsControl(kChar1));
Assert.IsFalse(Icu.IsControl(kChar2));
Assert.IsFalse(Icu.IsControl(kChar3));
Assert.IsFalse(Icu.IsControl(kChar4));
Assert.IsFalse(Icu.IsDiacritic(kChar1));
Assert.IsFalse(Icu.IsDiacritic(kChar2));
Assert.IsFalse(Icu.IsDiacritic(kChar3));
Assert.IsFalse(Icu.IsDiacritic(kChar4));
Assert.IsFalse(Icu.IsIdeographic(kChar1));
Assert.IsFalse(Icu.IsIdeographic(kChar2));
Assert.IsFalse(Icu.IsIdeographic(kChar3));
Assert.IsFalse(Icu.IsIdeographic(kChar4));
Assert.IsFalse(Icu.IsNumeric(kChar1));
Assert.IsFalse(Icu.IsNumeric(kChar2));
Assert.IsFalse(Icu.IsNumeric(kChar3));
Assert.IsFalse(Icu.IsNumeric(kChar4));
Assert.IsFalse(Icu.IsPunct(kChar1));
Assert.IsFalse(Icu.IsPunct(kChar2));
Assert.IsFalse(Icu.IsPunct(kChar3));
Assert.IsFalse(Icu.IsPunct(kChar4));
Assert.IsFalse(Icu.IsSpace(kChar1));
Assert.IsFalse(Icu.IsSpace(kChar2));
Assert.IsFalse(Icu.IsSpace(kChar3));
Assert.IsFalse(Icu.IsSpace(kChar4));
Assert.IsFalse(Icu.IsSymbol(kChar1));
Assert.IsFalse(Icu.IsSymbol(kChar2));
Assert.IsFalse(Icu.IsSymbol(kChar3));
Assert.IsFalse(Icu.IsSymbol(kChar4));
Assert.AreEqual(Icu.UCharCategory.U_PRIVATE_USE_CHAR, Icu.GetCharType(kChar1));
Assert.AreEqual(Icu.UCharCategory.U_PRIVATE_USE_CHAR, Icu.GetCharType(kChar2));
Assert.AreEqual(Icu.UCharCategory.U_UNASSIGNED, Icu.GetCharType(kChar3));
Assert.AreEqual(Icu.UCharCategory.U_UNASSIGNED, Icu.GetCharType(kChar4));
var decompositionType = Icu.GetDecompositionType(kChar1);
Assert.AreEqual("[none]", decompositionType.Description);
decompositionType = Icu.GetDecompositionType(kChar2);
Assert.AreEqual("[none]", decompositionType.Description);
decompositionType = Icu.GetDecompositionType(kChar3);
Assert.AreEqual("[none]", decompositionType.Description);
decompositionType = Icu.GetDecompositionType(kChar4);
Assert.AreEqual("[none]", decompositionType.Description);
var numericType = Icu.GetNumericType(kChar1);
Assert.AreEqual("[none]", numericType.Description);
numericType = Icu.GetNumericType(kChar2);
Assert.AreEqual("[none]", numericType.Description);
numericType = Icu.GetNumericType(kChar3);
Assert.AreEqual("[none]", numericType.Description);
numericType = Icu.GetNumericType(kChar4);
Assert.AreEqual("[none]", numericType.Description);
var prettyName = Icu.GetPrettyICUCharName("\xE000");
Assert.IsNull(prettyName);
prettyName = Icu.GetPrettyICUCharName("\xE001");
Assert.IsNull(prettyName);
prettyName = Icu.GetPrettyICUCharName(kChar3S);
Assert.IsNull(prettyName);
prettyName = Icu.GetPrettyICUCharName("\xDDDDD");
Assert.IsNull(prettyName);
}