public void TestExhaustive()
{
int counter = 0;
CanonicalIterator it = new CanonicalIterator("");
/*
* CanonicalIterator slowIt = new CanonicalIterator("");
* slowIt.SKIP_ZEROS = false;
*/
//Transliterator name = Transliterator.getInstance("[^\\u0020-\\u007F] name");
//Set itSet = new TreeSet();
//Set slowItSet = new TreeSet();
for (int i = 0; i < 0x10FFFF; ++i)
{
// skip characters we know don't have decomps
UUnicodeCategory type = UChar.GetUnicodeCategory(i);
if (type == UUnicodeCategory.OtherNotAssigned || type == UUnicodeCategory.PrivateUse ||
type == UUnicodeCategory.Surrogate)
{
continue;
}
if ((++counter % 5000) == 0)
{
Logln("Testing " + Utility.Hex(i, 0));
}
string s = UTF16.ValueOf(i);
CharacterTest(s, i, it);
CharacterTest(s + "\u0345", i, it);
}
}
private static bool IsLNS(int c)
{
// Letter, number, symbol,
// or a private use code point because those are typically used as letters or numbers.
// Consider modifier letters only if they are cased.
UUnicodeCategory gc = UCharacterProperty.Instance.GetUnicodeCategory(c);
return(((1 << (int)gc) & LNS) != 0 ||
(gc == UUnicodeCategory.ModifierLetter &&
UCaseProperties.Instance.GetCaseType(c) != CaseType.None));
}
/// <summary>
/// Converts a <see cref="UUnicodeCategory"/> to an <see cref="int"/>.
/// Same as <c>(int)<paramref name="characterCategory"/></c>.
/// </summary>
/// <param name="characterCategory">This <see cref="UUnicodeCategory"/>.</param>
/// <returns>This category as <see cref="int"/>.</returns>
public static int ToInt32(this UUnicodeCategory characterCategory) // ICU4N TODO: Add this extension to all main enums
{
return((int)characterCategory);
}
/// <summary>
/// Gets the name of the argument category.
/// </summary>
/// <param name="category">Category to retrieve name.</param>
/// <returns>Category name.</returns>
/// <stable>ICU 2.1</stable>
// ICU4N NOTE: Since ToString() cannot be changed from the default on an Enum,
// we have renamed this method AsString().
public static string AsString(this UUnicodeCategory category)
{
switch (category)
{
case UUnicodeCategory.UppercaseLetter:
return("Letter, Uppercase");
case UUnicodeCategory.LowercaseLetter:
return("Letter, Lowercase");
case UUnicodeCategory.TitlecaseLetter:
return("Letter, Titlecase");
case UUnicodeCategory.ModifierLetter:
return("Letter, Modifier");
case UUnicodeCategory.OtherLetter:
return("Letter, Other");
case UUnicodeCategory.NonSpacingMark:
return("Mark, Non-Spacing");
case UUnicodeCategory.EnclosingMark:
return("Mark, Enclosing");
case UUnicodeCategory.SpacingCombiningMark:
return("Mark, Spacing Combining");
case UUnicodeCategory.DecimalDigitNumber:
return("Number, Decimal Digit");
case UUnicodeCategory.LetterNumber:
return("Number, Letter");
case UUnicodeCategory.OtherNumber:
return("Number, Other");
case UUnicodeCategory.SpaceSeparator:
return("Separator, Space");
case UUnicodeCategory.LineSeparator:
return("Separator, Line");
case UUnicodeCategory.ParagraphSeparator:
return("Separator, Paragraph");
case UUnicodeCategory.Control:
return("Other, Control");
case UUnicodeCategory.Format:
return("Other, Format");
case UUnicodeCategory.PrivateUse:
return("Other, Private Use");
case UUnicodeCategory.Surrogate:
return("Other, Surrogate");
case UUnicodeCategory.DashPunctuation:
return("Punctuation, Dash");
case UUnicodeCategory.OpenPunctuation:
return("Punctuation, Open");
case UUnicodeCategory.ClosePunctuation:
return("Punctuation, Close");
case UUnicodeCategory.ConnectorPunctuation:
return("Punctuation, Connector");
case UUnicodeCategory.OtherPunctuation:
return("Punctuation, Other");
case UUnicodeCategory.MathSymbol:
return("Symbol, Math");
case UUnicodeCategory.CurrencySymbol:
return("Symbol, Currency");
case UUnicodeCategory.ModifierSymbol:
return("Symbol, Modifier");
case UUnicodeCategory.OtherSymbol:
return("Symbol, Other");
case UUnicodeCategory.InitialQuotePunctuation:
return("Punctuation, Initial quote");
case UUnicodeCategory.FinalQuotePunctuation:
return("Punctuation, Final quote");
default:
return("Unassigned");
}
}
