public void TestCharSequenceIterator()
{
String text = "abc123\ud800\udc01 "; // Includes a Unicode supplemental character
String vals = "LLLNNNX?S";
Trie2Writable tw = new Trie2Writable(0, 666);
tw.SetRange('a', 'z', 'L', false);
tw.SetRange('1', '9', 'N', false);
tw.Set(' ', 'S');
tw.Set(0x10001, 'X');
using (Trie2CharSequenceEnumerator it = tw.GetCharSequenceEnumerator(text, 0))
{
// Check forwards iteration.
Trie2CharSequenceValues ir;
int i;
for (i = 0; it.MoveNext(); i++)
{
ir = it.Current;
int expectedCP = Character.CodePointAt(text, i);
assertEquals("" + " i=" + i, expectedCP, ir.CodePoint);
assertEquals("" + " i=" + i, i, ir.Index);
assertEquals("" + " i=" + i, vals[i], ir.Value);
if (expectedCP >= 0x10000)
{
i++;
}
}
assertEquals("", text.Length, i);
// Check reverse iteration, starting at an intermediate point.
it.Set(5);
for (i = 5; it.MovePrevious();)
{
ir = it.Current;
int expectedCP = Character.CodePointBefore(text, i);
i -= (expectedCP < 0x10000 ? 1 : 2);
assertEquals("" + " i=" + i, expectedCP, ir.CodePoint);
assertEquals("" + " i=" + i, i, ir.Index);
assertEquals("" + " i=" + i, vals[i], ir.Value);
}
assertEquals("", 0, i);
}
}
private void doOtherInvariantTest(BreakIterator tb, String testChars)
{
StringBuffer work = new StringBuffer("a\r\na");
int errorCount = 0;
// a break should never occur between CR and LF
for (int i = 0; i < testChars.Length; i++)
{
work[0] = testChars[i];
for (int j = 0; j < testChars.Length; j++)
{
work[3] = testChars[j];
tb.SetText(work.ToString());
for (int k = tb.First(); k != BreakIterator.Done; k = tb.Next())
{
if (k == 2)
{
Errln("Break between CR and LF in string U+" +
(work[0]).ToHexString() + ", U+d U+a U+" +
(work[3]).ToHexString());
errorCount++;
if (errorCount >= 75)
{
return;
}
}
}
}
}
// a break should never occur before a non-spacing mark, unless it's preceded
// by a line terminator
work.Length = (0);
work.Append("aaaa");
for (int i = 0; i < testChars.Length; i++)
{
char c = testChars[i];
if (c == '\n' || c == '\r' || c == '\u2029' || c == '\u2028' || c == '\u0003')
{
continue;
}
work[1] = c;
for (int j = 0; j < testChars.Length; j++)
{
c = testChars[j];
if (Character.GetType(c) != UnicodeCategory.NonSpacingMark && Character.GetType(c)
!= UnicodeCategory.EnclosingMark)
{
continue;
}
work[2] = c;
tb.SetText(work.ToString());
for (int k = tb.First(); k != BreakIterator.Done; k = tb.Next())
{
if (k == 2)
{
Errln("Break between U+" + ((work[1])).ToHexString()
+ " and U+" + ((work[2])).ToHexString());
errorCount++;
if (errorCount >= 75)
{
return;
}
}
}
}
}
}