public void TestRuleStatus()
{
BreakIterator bi = BreakIterator.GetWordInstance(ULocale.ENGLISH);
bi.SetText("# ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= RuleBasedBreakIterator.WORD_NONE);
assertTrue(null, bi.RuleStatus < RuleBasedBreakIterator.WORD_NONE_LIMIT);
bi.SetText("3 ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= RuleBasedBreakIterator.WORD_NUMBER);
assertTrue(null, bi.RuleStatus < RuleBasedBreakIterator.WORD_NUMBER_LIMIT);
bi.SetText("a ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= RuleBasedBreakIterator.WORD_LETTER);
assertTrue(null, bi.RuleStatus < RuleBasedBreakIterator.WORD_LETTER_LIMIT);
bi.SetText("イ ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= RuleBasedBreakIterator.WORD_KANA);
// TODO: ticket #10261, Kana is not returning the correct status.
// assertTrue(null, bi.getRuleStatus() < RuleBasedBreakIterator.WORD_KANA_LIMIT);
// System.out.println("\n" + bi.getRuleStatus());
bi.SetText("退 ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= RuleBasedBreakIterator.WORD_IDEO);
assertTrue(null, bi.RuleStatus < RuleBasedBreakIterator.WORD_IDEO_LIMIT);
}
public void TestRuleStatus()
{
BreakIterator bi = BreakIterator.GetWordInstance(new UCultureInfo("en"));
bi.SetText("# ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= BreakIterator.WordNone);
assertTrue(null, bi.RuleStatus < BreakIterator.WordNoneLimit);
bi.SetText("3 ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= BreakIterator.WordNumber);
assertTrue(null, bi.RuleStatus < BreakIterator.WordNumberLimit);
bi.SetText("a ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= BreakIterator.WordLetter);
assertTrue(null, bi.RuleStatus < BreakIterator.WordLetterLimit);
bi.SetText("イ ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= BreakIterator.WordKana);
// TODO: ticket #10261, Kana is not returning the correct status.
// assertTrue(null, bi.RuleStatus < RuleStatus.WordKanaLimit);
// Console.Out.WriteLine("\n" + bi.RuleStatus);
bi.SetText("退 ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= BreakIterator.WordIdeo);
assertTrue(null, bi.RuleStatus < BreakIterator.WordIdeoLimit);
}
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;
}
public void TestRuleStatus()
{
BreakIterator bi = BreakIterator.GetWordInstance(ULocale.ENGLISH);
bi.SetText("# ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= RuleStatus.WordNone);
assertTrue(null, bi.RuleStatus < RuleStatus.WordNoneLimit);
bi.SetText("3 ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= RuleStatus.WordNumber);
assertTrue(null, bi.RuleStatus < RuleStatus.WordNumberLimit);
bi.SetText("a ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= RuleStatus.WordLetter);
assertTrue(null, bi.RuleStatus < RuleStatus.WordLetterLimit);
bi.SetText("イ ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= RuleStatus.WordKana);
// TODO: ticket #10261, Kana is not returning the correct status.
// assertTrue(null, bi.getRuleStatus() < RuleStatus.WordKanaLimit);
// System.out.println("\n" + bi.getRuleStatus());
bi.SetText("退 ");
assertEquals(null, bi.Next(), 1);
assertTrue(null, bi.RuleStatus >= RuleStatus.WordIdeo);
assertTrue(null, bi.RuleStatus < RuleStatus.WordIdeoLimit);
}
public virtual int FindStartOffset(StringBuilder buffer, int start)
{
// avoid illegal start offset
if (start > buffer.Length || start < 1) return start;
bi.SetText(buffer.ToString(0, start - 0));
bi.Last();
return bi.Previous();
}
/// <summary>
/// Updates this <code>LineBreakMeasurer</code> after a single
/// character is inserted into the text, and sets the current
/// position to the beginning of the paragraph.
/// </summary>
/// <param name="newParagraph"> the text after the insertion </param>
/// <param name="insertPos"> the position in the text at which the character
/// is inserted </param>
/// <exception cref="IndexOutOfBoundsException"> if <code>insertPos</code> is less
/// than the start of <code>newParagraph</code> or greater than
/// or equal to the end of <code>newParagraph</code> </exception>
/// <exception cref="NullPointerException"> if <code>newParagraph</code> is
/// <code>null</code> </exception>
/// <seealso cref= #deleteChar </seealso>
public void InsertChar(AttributedCharacterIterator newParagraph, int insertPos)
{
Measurer.InsertChar(newParagraph, insertPos);
Limit = newParagraph.EndIndex;
Pos = Start = newParagraph.BeginIndex;
CharIter.Reset(Measurer.Chars, newParagraph.BeginIndex);
BreakIter.SetText(CharIter);
}
/// <summary>
/// Implements the "Final_Cased" condition
///
/// Specification: Within the closest word boundaries containing C, there is a cased
/// letter before C, and there is no cased letter after C.
///
/// Regular Expression:
/// Before C: [{cased==true}][{wordBoundary!=true}]*
/// After C: !([{wordBoundary!=true}]*[{cased}])
/// </summary>
private static bool IsFinalCased(String src, int index, Locale locale)
{
BreakIterator wordBoundary = BreakIterator.GetWordInstance(locale);
wordBoundary.SetText(src);
int ch;
// Look for a preceding 'cased' letter
for (int i = index; (i >= 0) && !wordBoundary.IsBoundary(i); i -= Character.CharCount(ch))
{
ch = src.CodePointBefore(i);
if (IsCased(ch))
{
int len = src.Length();
// Check that there is no 'cased' letter after the index
for (i = index + Character.CharCount(src.CodePointAt(index)); (i < len) && !wordBoundary.IsBoundary(i); i += Character.CharCount(ch))
{
ch = src.CodePointAt(i);
if (IsCased(ch))
{
return(false);
}
}
return(true);
}
}
return(false);
}
public void TestBug4153072()
{
BreakIterator iter = BreakIterator.GetWordInstance();
String str = "...Hello, World!...";
int begin = 3;
int end = str.Length - 3;
// not used boolean gotException = false;
iter.SetText(new StringCharacterIterator(str, begin, end, begin));
for (int index = -1; index < begin + 1; ++index)
{
try
{
iter.IsBoundary(index);
if (index < begin)
{
Errln("Didn't get exception with offset = " + index +
" and begin index = " + begin);
}
}
catch (ArgumentException e)
{
if (index >= begin)
{
Errln("Got exception with offset = " + index +
" and begin index = " + begin);
}
}
}
}
private void Consume(BreakIterator bi, CharacterIterator ci)
{
bi.SetText(ci);
while (bi.Next() != BreakIterator.Done)
{
}
}
public void TestGetTitleInstance()
{
BreakIterator bi = BreakIterator.GetTitleInstance(new CultureInfo("en-CA"));
TestFmwk.assertNotEquals("Title instance break iterator not correctly instantiated", bi.First(), null);
bi.SetText("Here is some Text");
TestFmwk.assertEquals("Title instance break iterator not correctly instantiated", bi.First(), 0);
}
/**
* @param filteredBI
* @param text
*/
private void assertFrenchBreakBehavior(BreakIterator filteredBI, String text)
{
Logln("Testing French behavior:");
filteredBI.SetText(text);
assertEquals("6th next", 20, filteredBI.Next());
assertEquals("6th next", 84, filteredBI.Next());
filteredBI.First();
}
private void Consume(BreakIterator bi, CharacterIterator ci)
{
bi.SetText(ci.toString());
while (bi.Next() != BreakIterator.DONE)
{
;
}
}
/**
* @param filteredBI
* @param text
*/
private void assertEnglishBreakBehavior(BreakIterator filteredBI, String text)
{
Logln("Testing English filtered behavior:");
filteredBI.SetText(text);
assertEquals("5th next", 84, filteredBI.Next());
assertEquals("5th next", 278, filteredBI.Next());
filteredBI.First();
}
/**
* @param filteredBI
* @param text
*/
private void assertDefaultBreakBehavior(BreakIterator filteredBI, String text)
{
Logln("Testing Default Behavior:");
filteredBI.SetText(text);
assertEquals("1st next", 20, filteredBI.Next());
assertEquals("1st next", 84, filteredBI.Next());
assertEquals("1st next", 90, filteredBI.Next());
assertEquals("1st next", 181, filteredBI.Next());
assertEquals("1st next", 278, filteredBI.Next());
filteredBI.First();
}
private void UxBreakClick(object sender, EventArgs e)
{
using (BreakIterator bi = BreakIterator.CreateWordInstance(Locale.GetUS()))
{
bi.SetText(this.uxText.Text);
var words = bi.Enumerate().ToList();
MessageBox.Show(string.Join("-", words.ToArray()));
}
}
public void TestFilteredJapanese()
{
ULocale loc = ULocale.JAPANESE;
BreakIterator brk = FilteredBreakIteratorBuilder
.GetInstance(loc)
.WrapIteratorWithFilter(BreakIterator.GetSentenceInstance(loc));
brk.SetText("ＯＫです。");
assertEquals("Starting point", 0, brk.Current);
assertEquals("Next point", 5, brk.Next());
assertEquals("Last point", BreakIterator.Done, brk.Next());
}
/** Asserts that two breakiterators break the text the same way */
public void assertSameBreaks(CharacterIterator one, CharacterIterator two, BreakIterator expected, BreakIterator actual)
{
expected.SetText(one);
actual.SetText(two);
assertEquals(expected.Current, actual.Current);
// next()
int v = expected.Current;
while (v != BreakIterator.DONE)
{
assertEquals(v = expected.Next(), actual.Next());
assertEquals(expected.Current, actual.Current);
}
// first()
assertEquals(expected.First(), actual.First());
assertEquals(expected.Current, actual.Current);
// last()
assertEquals(expected.Last(), actual.Last());
assertEquals(expected.Current, actual.Current);
// previous()
v = expected.Current;
while (v != BreakIterator.DONE)
{
assertEquals(v = expected.Previous(), actual.Previous());
assertEquals(expected.Current, actual.Current);
}
// following()
for (int i = one.BeginIndex; i <= one.EndIndex; i++)
{
expected.First();
actual.First();
assertEquals(expected.Following(i), actual.Following(i));
assertEquals(expected.Current, actual.Current);
}
// preceding()
for (int i = one.BeginIndex; i <= one.EndIndex; i++)
{
expected.Last();
actual.Last();
assertEquals(expected.Preceding(i), actual.Preceding(i));
assertEquals(expected.Current, actual.Current);
}
}
private void UxOccurrenceCountClick(object sender, EventArgs e)
{
using (BreakIterator bi = BreakIterator.CreateWordInstance(Locale.GetUS()))
{
bi.SetText(this.uxText.Text);
var words = bi.Enumerate()
.GroupBy(w => w)
.OrderBy(x => x.Count())
.Reverse()
.Select(x => x.Key + " : " + x.Count())
.Take(10)
.ToArray();
MessageBox.Show(string.Join(Environment.NewLine, words));
}
}
private void MakeLayoutWindow(int localStart)
{
int compStart = localStart;
int compLimit = FChars.Length;
// If we've already gone past the layout window, format to end of paragraph
if (LayoutCount > 0 && !HaveLayoutWindow)
{
float avgLineLength = System.Math.Max(LayoutCharCount / LayoutCount, 1);
compLimit = System.Math.Min(localStart + (int)(avgLineLength * EST_LINES), FChars.Length);
}
if (localStart > 0 || compLimit < FChars.Length)
{
if (CharIter == null)
{
CharIter = new CharArrayIterator(FChars);
}
else
{
CharIter.Reset(FChars);
}
if (FLineBreak == null)
{
FLineBreak = BreakIterator.LineInstance;
}
FLineBreak.SetText(CharIter);
if (localStart > 0)
{
if (!FLineBreak.IsBoundary(localStart))
{
compStart = FLineBreak.Preceding(localStart);
}
}
if (compLimit < FChars.Length)
{
if (!FLineBreak.IsBoundary(compLimit))
{
compLimit = FLineBreak.Following(compLimit);
}
}
}
EnsureComponents(compStart, compLimit);
HaveLayoutWindow = true;
}
public void IsBoundary(BreakIterator.UBreakIteratorType type,
string text,
int[] offsetsToTest,
bool[] expectedIsBoundary,
int[] expectedOffsets) // expected BreakIterator.Current after calling IsBoundary.
{
var locale = new Locale("zh");
BreakIterator bi = default(BreakIterator);
try
{
switch (type)
{
case BreakIterator.UBreakIteratorType.SENTENCE:
bi = BreakIterator.CreateSentenceInstance(locale);
break;
case BreakIterator.UBreakIteratorType.WORD:
bi = BreakIterator.CreateWordInstance(locale);
break;
default:
throw new NotSupportedException("This iterator type is not supported in this test yet. [" + type + "]");
}
bi.SetText(text);
for (int i = 0; i < offsetsToTest.Length; i++)
{
var isBoundary = bi.IsBoundary(offsetsToTest[i]);
Assert.AreEqual(expectedIsBoundary[i], isBoundary, "Expected IsBoundary was not equal at i: {0}, offset: {1}", i, offsetsToTest[i]);
Assert.AreEqual(expectedOffsets[i], bi.Current);
}
}
finally
{
if (bi != default(BreakIterator))
{
bi.Dispose();
}
}
}
public void TestGetSetText()
{
Logln("Testing getText setText ");
String str1 = "first string.";
String str2 = "Second string.";
//RuleBasedBreakIterator charIter1 = (RuleBasedBreakIterator) BreakIterator.getCharacterInstance(Locale.getDefault());
RuleBasedBreakIterator wordIter1 = (RuleBasedBreakIterator)BreakIterator.GetWordInstance(CultureInfo.CurrentCulture);
CharacterIterator text1 = new StringCharacterIterator(str1);
//CharacterIterator text1Clone = (CharacterIterator) text1.Clone();
//CharacterIterator text2 = new StringCharacterIterator(str2);
wordIter1.SetText(str1);
if (!wordIter1.Text.Equals(text1))
{
Errln("ERROR:1 error in setText or getText ");
}
if (wordIter1.Current != 0)
{
Errln("ERROR:1 setText did not set the iteration position to the beginning of the text, it is"
+ wordIter1.Current + "\n");
}
wordIter1.Next(2);
wordIter1.SetText(str2);
if (wordIter1.Current != 0)
{
Errln("ERROR:2 setText did not reset the iteration position to the beginning of the text, it is"
+ wordIter1.Current + "\n");
}
// Test the CharSequence overload of setText() for a simple case.
BreakIterator lineIter = BreakIterator.GetLineInstance(new CultureInfo("en"));
ICharSequence csText = "Hello, World. ".ToCharSequence();
// Expected Line Brks ^ ^ ^
// 0123456789012345
List <int> expected = new List <int>();
expected.Add(0); expected.Add(7); expected.Add(14);
lineIter.SetText(csText);
for (int pos = lineIter.First(); pos != BreakIterator.Done; pos = lineIter.Next())
{
assertTrue("", expected.Contains(pos));
}
assertEquals("", csText.Length, lineIter.Current);
}
public void TestEndBehavior()
{
String testString = "boo.";
BreakIterator wb = BreakIterator.GetWordInstance();
wb.SetText(testString);
if (wb.First() != 0)
{
Errln("Didn't get break at beginning of string.");
}
if (wb.Next() != 3)
{
Errln("Didn't get break before period in \"boo.\"");
}
if (wb.Current != 4 && wb.Next() != 4)
{
Errln("Didn't get break at end of string.");
}
}
public void TestPreceding()
{
String words3 = "aaa bbb ccc";
BreakIterator e = BreakIterator.GetWordInstance(CultureInfo.CurrentCulture);
e.SetText(words3);
e.First();
int p1 = e.Next();
int p2 = e.Next();
int p3 = e.Next();
int p4 = e.Next();
int f = e.Following(p2 + 1);
int p = e.Preceding(p2 + 1);
if (f != p3)
{
Errln("IntlTestTextBoundary::TestPreceding: f!=p3");
}
if (p != p2)
{
Errln("IntlTestTextBoundary::TestPreceding: p!=p2");
}
if (p1 + 1 != p2)
{
Errln("IntlTestTextBoundary::TestPreceding: p1+1!=p2");
}
if (p3 + 1 != p4)
{
Errln("IntlTestTextBoundary::TestPreceding: p3+1!=p4");
}
if (!e.IsBoundary(p2) || e.IsBoundary(p2 + 1) || !e.IsBoundary(p3))
{
Errln("IntlTestTextBoundary::TestPreceding: isBoundary err");
}
}
public void TestBug12918()
{
// This test triggered an assertion failure in ICU4C, in dictbe.cpp
// The equivalent code in ICU4J is structured slightly differently,
// and does not appear vulnerable to the same issue.
//
// \u3325 decomposes with normalization, then the CJK dictionary
// finds a break within the decomposition.
String crasherString = "\u3325\u4a16";
BreakIterator iter = BreakIterator.GetWordInstance(ULocale.ENGLISH);
iter.SetText(crasherString);
iter.First();
int pos = 0;
int lastPos = -1;
while ((pos = iter.Next()) != BreakIterator.Done)
{
assertTrue("", pos > lastPos);
}
}
public void SetText(string text)
{
this.text = text;
wordBreaker.SetText(text);
}
public void TestLineIteration()
{
BreakIterator bi = GetLineInstance(System.Globalization.CultureInfo.InvariantCulture);
// Test empty
Assert.AreEqual(0, bi.Current);
Assert.AreEqual(BreakIterator.Done, bi.Next());
Assert.AreEqual(0, bi.Current);
bi.SetText(LINE_TEXT);
// Ensure position starts at 0 when initialized
Assert.AreEqual(0, bi.Current);
// Check first boundary (Apache\t^Lucene) - Ensure we break on \t
Assert.AreEqual(7, bi.Next());
// Ensure Current returns the most recent boundary
Assert.AreEqual(7, bi.Current);
// Check next boundary (Lucene^(TM))
Assert.AreEqual(13, bi.Next());
// Ensure Current returns the most recent boundary
Assert.AreEqual(13, bi.Current);
// Check next boundary (Lucene(TM) ^is a)
Assert.AreEqual(18, bi.Next());
// Ensure Current returns the most recent boundary
Assert.AreEqual(18, bi.Current);
// Move to start of high-performance
bi.Next();
bi.Next();
// Check next boundary (high-\n^performance)
Assert.AreEqual(29, bi.Next());
// Check last boundary (in Java.^)
Assert.AreEqual(108, bi.Last());
// Check move past last boundary
Assert.AreEqual(BreakIterator.Done, bi.Next());
// Ensure we are still at last boundary
Assert.AreEqual(108, bi.Current);
// Check MovePrevious
Assert.AreEqual(103, bi.Previous());
// Ensure we get the same value for Current as the last move
Assert.AreEqual(103, bi.Current);
// Check MoveFirst
Assert.AreEqual(0, bi.First());
// Ensure we get the same value for Current as the last move
Assert.AreEqual(0, bi.Current);
// Check moving beyond first boundary
Assert.AreEqual(BreakIterator.Done, bi.Previous());
// Ensure we are still at first boundary
Assert.AreEqual(0, bi.Current);
// Check MoveLast()
Assert.AreEqual(108, bi.Last());
}
public void TestWordIteration()
{
BreakIterator bi = GetWordInstance(System.Globalization.CultureInfo.InvariantCulture);
bi.SetText("");
// Test empty
Assert.AreEqual(0, bi.Current);
Assert.AreEqual(BreakIterator.Done, bi.Next());
Assert.AreEqual(0, bi.Current);
bi.SetText(TEXT);
// Ensure position starts at 0 when initialized
Assert.AreEqual(0, bi.Current);
// Check first boundary (Apache^)
Assert.AreEqual(6, bi.Next());
// Ensure Current returns the last boundary iterated to
Assert.AreEqual(6, bi.Current);
// Check second boundary (^Lucene)
Assert.AreEqual(7, bi.Next());
// Ensure Current returns the last boundary iterated to
Assert.AreEqual(7, bi.Current);
// Check third boundary (Lucene^)
Assert.AreEqual(13, bi.Next());
// Ensure Current returns the last boundary iterated to
Assert.AreEqual(13, bi.Current);
// Check fourth boundary (^TM)
Assert.AreEqual(14, bi.Next());
// Check fifth boundary (TM^)
Assert.AreEqual(16, bi.Next());
// Check sixth boundary (TM)^
Assert.AreEqual(17, bi.Next());
// Check seventh boundary (^is)
Assert.AreEqual(18, bi.Next());
// Move to (^high-performance)
bi.Next();
bi.Next();
bi.Next();
// Check next boundary (^high-performance)
Assert.AreEqual(23, bi.Next());
// Ensure we don't break on hyphen (high-performance^)
Assert.AreEqual(39, bi.Next());
// Check MoveLast()
Assert.AreEqual(107, bi.Last());
// Check going past last boundary
Assert.AreEqual(BreakIterator.Done, bi.Next());
// Check we are still at last boundary
Assert.AreEqual(107, bi.Current);
// Check MoveFirst()
Assert.AreEqual(0, bi.First());
// Check going past first boundary
Assert.AreEqual(BreakIterator.Done, bi.Previous());
// Check we are still at first boundary
Assert.AreEqual(0, bi.Current);
}
public void TestSentenceIteration()
{
BreakIterator bi = GetSentenceInstance(System.Globalization.CultureInfo.InvariantCulture);
bi.SetText("");
// Test empty
Assert.AreEqual(0, bi.Current);
Assert.AreEqual(BreakIterator.Done, bi.Next());
Assert.AreEqual(0, bi.Current);
bi.SetText(SENTENCE_TEXT);
// Ensure position starts at 0 when initialized
Assert.AreEqual(0, bi.Current);
// Check first boundary (in Java.^) - Ensure we don't break on \n
Assert.AreEqual(108, bi.Next());
// Ensure Current returns the most recent boundary
Assert.AreEqual(108, bi.Current);
// Check next boundary (especially cross-platform.^)
Assert.AreEqual(221, bi.Next());
// Check next boundary (free download.^)
Assert.AreEqual(290, bi.Next());
// Check next boundary (things easy.^)
Assert.AreEqual(324, bi.Next());
// Check next boundary (is powerful.^)
Assert.AreEqual(344, bi.Next());
// Check next boundary (is exciting.^)
Assert.AreEqual(364, bi.Next());
// Check next boundary (is cool.^)
Assert.AreEqual(380, bi.Next());
// Check last boundary (Lucene now?^)
Assert.AreEqual(400, bi.Next());
// Check move past last boundary
Assert.AreEqual(BreakIterator.Done, bi.Next());
// Ensure we are still at last boundary
Assert.AreEqual(400, bi.Current);
// Check MovePrevious
Assert.AreEqual(380, bi.Previous());
// Ensure we get the same value for Current as the last move
Assert.AreEqual(380, bi.Current);
// Check MoveFirst
Assert.AreEqual(0, bi.First());
// Ensure we get the same value for Current as the last move
Assert.AreEqual(0, bi.Current);
// Check moving beyond first boundary
Assert.AreEqual(BreakIterator.Done, bi.Previous());
// Ensure we are still at first boundary
Assert.AreEqual(0, bi.Current);
// Check MoveLast()
Assert.AreEqual(400, bi.Last());
}
public void TestWordIterationThai()
{
BreakIterator bi = GetWordInstance(new System.Globalization.CultureInfo("th"));
bi.SetText("");
// Test empty
Assert.AreEqual(0, bi.Current);
Assert.AreEqual(BreakIterator.Done, bi.Next());
Assert.AreEqual(0, bi.Current);
bi.SetText("บริษัทMicrosoftบริการดีที่สุด");
// Ensure position starts at 0 when initialized
Assert.AreEqual(0, bi.Current);
// Check first boundary (บริษัท^Microsoft)
Assert.AreEqual(6, bi.Next());
// Ensure Current returns the last boundary iterated to
Assert.AreEqual(6, bi.Current);
// Check second boundary (Microsoft^บริการ)
Assert.AreEqual(15, bi.Next());
// Ensure Current returns the last boundary iterated to
Assert.AreEqual(15, bi.Current);
// Check third boundary (บริการ^ดี)
Assert.AreEqual(21, bi.Next());
// Ensure Current returns the last boundary iterated to
Assert.AreEqual(21, bi.Current);
// Check fourth boundary (ดี^ที่สุด)
Assert.AreEqual(23, bi.Next());
// Check fifth boundary (ดีที่สุด^)
Assert.AreEqual(29, bi.Next());
// Check beyond last boundary (ดีที่สุด)^
Assert.AreEqual(BreakIterator.Done, bi.Next());
// Check we are still at last boundary
Assert.AreEqual(29, bi.Current);
// Check MovePrevious() (ดี^ที่สุด)
Assert.AreEqual(23, bi.Previous());
// Check MoveFirst()
Assert.AreEqual(0, bi.First());
// Check going past first boundary
Assert.AreEqual(BreakIterator.Done, bi.Previous());
// Check we are still at first boundary
Assert.AreEqual(0, bi.Current);
// Check Numerals
bi.SetText("๑23๔๕๖7");
// Ensure position starts at 0 when initialized
Assert.AreEqual(0, bi.Current);
// Ensure Hindu and Thai numerals stay in one group
Assert.AreEqual(7, bi.Next());
}
private void Consume(BreakIterator bi, CharacterIterator ci)
{
bi.SetText(ci.toString());
while (bi.Next() != BreakIterator.DONE)
{
;
}
}
