| public SetText ( char array, int start, int length ) : void | ||
| array | char | text buffer to examine |
| start | int | offset into buffer |
| length | int | maximum length to examine |
| return | void | |
public override void Reset()
{
base.Reset();
wrapper.SetText(m_buffer, 0, 0);
iterator.SetText(new string(wrapper.Text, wrapper.Start, wrapper.Length));
length = usableLength = m_offset = 0;
}
public override void Reset()
{
base.Reset();
wrapper.SetText(buffer, 0, 0);
iterator.SetText(new string(buffer, 0, 0));
length = usableLength = offset = 0;
}
/// <summary>
/// Refill the buffer, accumulating the offset and setting usableLength to the
/// last unambiguous break position
/// </summary>
private void Refill()
{
offset += usableLength;
int leftover = length - usableLength;
Array.Copy(buffer, usableLength, buffer, 0, leftover);
int requested = buffer.Length - leftover;
int returned = read(input, buffer, leftover, requested);
length = returned < 0 ? leftover : returned + leftover;
if (returned < requested) // reader has been emptied, process the rest
{
usableLength = length;
}
else // still more data to be read, find a safe-stopping place
{
usableLength = FindSafeEnd();
if (usableLength < 0)
{
usableLength = length; /*
* }
* more than IOBUFFER of text without breaks,
* gonna possibly truncate tokens
*/
}
wrapper.SetText(buffer, 0, Math.Max(0, usableLength));
iterator.Text = wrapper;
}
}
/* run this to test if your JRE is buggy
* public void testSentenceInstanceJREBUG() {
* // we use the default locale, as its randomized by LuceneTestCase
* BreakIterator bi = BreakIterator.getSentenceInstance(Locale.getDefault());
* Segment ci = new Segment();
* for (int i = 0; i < 10000; i++) {
* char text[] = TestUtil.randomUnicodeString(random).toCharArray();
* ci.array = text;
* ci.offset = 0;
* ci.count = text.length;
* consume(bi, ci);
* }
* }
*/
private void DoTests(CharArrayIterator ci)
{
// basics
ci.SetText("testing".ToCharArray(), 0, "testing".Length);
assertEquals(0, ci.BeginIndex);
assertEquals(7, ci.EndIndex);
assertEquals(0, ci.Index);
assertEquals('t', ci.Current);
assertEquals('e', ci.Next());
assertEquals('g', ci.Last());
assertEquals('n', ci.Previous());
assertEquals('t', ci.First());
assertEquals(CharacterIterator.Done, ci.Previous());
// first()
ci.SetText("testing".ToCharArray(), 0, "testing".Length);
ci.Next();
// Sets the position to getBeginIndex() and returns the character at that position.
assertEquals('t', ci.First());
assertEquals(ci.BeginIndex, ci.Index);
// or DONE if the text is empty
ci.SetText(new char[] { }, 0, 0);
assertEquals(CharacterIterator.Done, ci.First());
// last()
ci.SetText("testing".ToCharArray(), 0, "testing".Length);
// Sets the position to getEndIndex()-1 (getEndIndex() if the text is empty)
// and returns the character at that position.
assertEquals('g', ci.Last());
assertEquals(ci.Index, ci.EndIndex - 1);
// or DONE if the text is empty
ci.SetText(new char[] { }, 0, 0);
assertEquals(CharacterIterator.Done, ci.Last());
assertEquals(ci.EndIndex, ci.Index);
// current()
// Gets the character at the current position (as returned by getIndex()).
ci.SetText("testing".ToCharArray(), 0, "testing".Length);
assertEquals('t', ci.Current);
ci.Last();
ci.Next();
// or DONE if the current position is off the end of the text.
assertEquals(CharacterIterator.Done, ci.Current);
// next()
ci.SetText("te".ToCharArray(), 0, 2);
// Increments the iterator's index by one and returns the character at the new index.
assertEquals('e', ci.Next());
assertEquals(1, ci.Index);
// or DONE if the new position is off the end of the text range.
assertEquals(CharacterIterator.Done, ci.Next());
assertEquals(ci.EndIndex, ci.Index);
// setIndex()
ci.SetText("test".ToCharArray(), 0, "test".Length);
try
{
ci.SetIndex(5);
fail();
}
catch (Exception e)
{
assertTrue(e is System.ArgumentException);
}
// clone()
var text = "testing".ToCharArray();
ci.SetText(text, 0, text.Length);
ci.Next();
var ci2 = ci.Clone() as CharArrayIterator;
assertEquals(ci.Index, ci2.Index);
assertEquals(ci.Next(), ci2.Next());
assertEquals(ci.Last(), ci2.Last());
}
/* run this to test if your JRE is buggy
public void testSentenceInstanceJREBUG() {
// we use the default locale, as its randomized by LuceneTestCase
BreakIterator bi = BreakIterator.getSentenceInstance(Locale.getDefault());
Segment ci = new Segment();
for (int i = 0; i < 10000; i++) {
char text[] = TestUtil.randomUnicodeString(random).toCharArray();
ci.array = text;
ci.offset = 0;
ci.count = text.length;
consume(bi, ci);
}
}
*/
private void DoTests(CharArrayIterator ci)
{
// basics
ci.SetText("testing".ToCharArray(), 0, "testing".Length);
assertEquals(0, ci.BeginIndex);
assertEquals(7, ci.EndIndex);
assertEquals(0, ci.Index);
assertEquals('t', ci.Current());
assertEquals('e', ci.Next());
assertEquals('g', ci.Last());
assertEquals('n', ci.Previous());
assertEquals('t', ci.First());
assertEquals(CharacterIterator.DONE, ci.Previous());
// first()
ci.SetText("testing".ToCharArray(), 0, "testing".Length);
ci.Next();
// Sets the position to getBeginIndex() and returns the character at that position.
assertEquals('t', ci.First());
assertEquals(ci.BeginIndex, ci.Index);
// or DONE if the text is empty
ci.SetText(new char[] { }, 0, 0);
assertEquals(CharacterIterator.DONE, ci.First());
// last()
ci.SetText("testing".ToCharArray(), 0, "testing".Length);
// Sets the position to getEndIndex()-1 (getEndIndex() if the text is empty)
// and returns the character at that position.
assertEquals('g', ci.Last());
assertEquals(ci.Index, ci.EndIndex - 1);
// or DONE if the text is empty
ci.SetText(new char[] { }, 0, 0);
assertEquals(CharacterIterator.DONE, ci.Last());
assertEquals(ci.EndIndex, ci.Index);
// current()
// Gets the character at the current position (as returned by getIndex()).
ci.SetText("testing".ToCharArray(), 0, "testing".Length);
assertEquals('t', ci.Current());
ci.Last();
ci.Next();
// or DONE if the current position is off the end of the text.
assertEquals(CharacterIterator.DONE, ci.Current());
// next()
ci.SetText("te".ToCharArray(), 0, 2);
// Increments the iterator's index by one and returns the character at the new index.
assertEquals('e', ci.Next());
assertEquals(1, ci.Index);
// or DONE if the new position is off the end of the text range.
assertEquals(CharacterIterator.DONE, ci.Next());
assertEquals(ci.EndIndex, ci.Index);
// setIndex()
ci.SetText("test".ToCharArray(), 0, "test".Length);
try
{
ci.SetIndex(5);
fail();
}
catch (Exception e)
{
assertTrue(e is System.ArgumentException);
}
// clone()
var text = "testing".ToCharArray();
ci.SetText(text, 0, text.Length);
ci.Next();
var ci2 = ci.Clone() as CharArrayIterator;
assertEquals(ci.Index, ci2.Index);
assertEquals(ci.Next(), ci2.Next());
assertEquals(ci.Last(), ci2.Last());
}