public void TestNormalizerAPI()
{
// instantiate a Normalizer from a CharacterIterator
string s = Utility.Unescape("a\u0308\uac00\\U0002f800");
// make s a bit longer and more interesting
CharacterIterator iter = new StringCharacterIterator(s + s);
//test deprecated constructors
Normalizer norm = new Normalizer(iter, NormalizerMode.NFC, 0);
if (norm.Next() != 0xe4)
{
Errln("error in Normalizer(CharacterIterator).next()");
}
Normalizer norm2 = new Normalizer(s, NormalizerMode.NFC, 0);
if (norm2.Next() != 0xe4)
{
Errln("error in Normalizer(CharacterIterator).next()");
}
// test clone(), ==, and hashCode()
Normalizer clone = (Normalizer)norm.Clone();
if (clone.GetBeginIndex() != norm.GetBeginIndex())
{
Errln("error in Normalizer.getBeginIndex()");
}
if (clone.GetEndIndex() != norm.GetEndIndex())
{
Errln("error in Normalizer.getEndIndex()");
}
// test setOption() and getOption()
clone.SetOption(0xaa0000, true);
clone.SetOption(0x20000, false);
if (clone.GetOption(0x880000) == 0 || clone.GetOption(0x20000) == 1)
{
Errln("error in Normalizer::setOption() or Normalizer::getOption()");
}
// ICU4N specific - test setting normalizer options via enum
clone.UnicodeVersion = NormalizerUnicodeVersion.Unicode3_2;
assertEquals("error in Normalizer.UnicodeVersion property", NormalizerUnicodeVersion.Unicode3_2, clone.UnicodeVersion);
clone.UnicodeVersion = NormalizerUnicodeVersion.Default;
assertEquals("error in Normalizer.UnicodeVersion property", NormalizerUnicodeVersion.Default, clone.UnicodeVersion);
//test deprecated normalize method
Normalizer.Normalize(s, NormalizerMode.NFC, 0);
//test deprecated compose method
Normalizer.Compose(s, false, 0);
//test deprecated decompose method
Normalizer.Decompose(s, false, 0);
}
public void TestRoundTrip()
{
int options = Normalizer.IGNORE_HANGUL;
bool compat = false;
ComposedCharIter iter = new ComposedCharIter(false, options);
while (iter.HasNext)
{
char ch = iter.Next();
string chStr = "" + ch;
string decomp = iter.Decomposition();
string comp = Normalizer.Compose(decomp, compat);
if (UChar.HasBinaryProperty(ch, UProperty.Full_Composition_Exclusion))
{
Logln("Skipped excluded char " + Hex(ch) + " (" + UChar.GetName(ch) + ")");
continue;
}
// Avoid disparaged characters
if (decomp.Length == 4)
{
continue;
}
if (!comp.Equals(chStr))
{
Errln("ERROR: Round trip invalid: " + Hex(chStr) + " --> " + Hex(decomp)
+ " --> " + Hex(comp));
Errln(" char decomp is '" + decomp + "'");
}
}
}
private void CharacterTest(string s, int ch, CanonicalIterator it)
{
int mixedCounter = 0;
int lastMixedCounter = -1;
bool gotDecomp = false;
bool gotComp = false;
bool gotSource = false;
string decomp = Normalizer.Decompose(s, false);
string comp = Normalizer.Compose(s, false);
// skip characters that don't have either decomp.
// need quick test for this!
if (s.Equals(decomp) && s.Equals(comp))
{
return;
}
it.SetSource(s);
while (true)
{
string item = it.Next();
if (item == null)
{
break;
}
if (item.Equals(s))
{
gotSource = true;
}
if (item.Equals(decomp))
{
gotDecomp = true;
}
if (item.Equals(comp))
{
gotComp = true;
}
if ((mixedCounter & 0x7F) == 0 && (ch < 0xAD00 || ch > 0xAC00 + 11172))
{
if (lastMixedCounter != mixedCounter)
{
Logln("");
lastMixedCounter = mixedCounter;
}
Logln("\t" + mixedCounter + "\t" + Hex(item)
+ (item.Equals(s) ? "\t(*original*)" : "")
+ (item.Equals(decomp) ? "\t(*decomp*)" : "")
+ (item.Equals(comp) ? "\t(*comp*)" : "")
);
}
}
// check that zeros optimization doesn't mess up.
/*
* if (true) {
* it.reset();
* itSet.clear();
* while (true) {
* String item = it.next();
* if (item == null) break;
* itSet.add(item);
* }
* slowIt.setSource(s);
* slowItSet.clear();
* while (true) {
* String item = slowIt.next();
* if (item == null) break;
* slowItSet.add(item);
* }
* if (!itSet.equals(slowItSet)) {
* errln("Zero optimization failure with " + getReadable(s));
* }
* }
*/
mixedCounter++;
if (!gotSource || !gotDecomp || !gotComp)
{
Errln("FAIL CanonicalIterator: " + s + " decomp: " + decomp + " comp: " + comp);
it.Reset();
for (string item = it.Next(); item != null; item = it.Next())
{
Err(item + " ");
}
Errln("");
}
}
