public virtual void TestFromChars()
{
for (int i = 0; i < 100; i++)
{
string s = TestUtil.RandomUnicodeString(Random());
string s2 = (new BytesRef(s)).Utf8ToString();
Assert.AreEqual(s, s2);
}
// only for 4.x
Assert.AreEqual("\uFFFF", (new BytesRef("\uFFFF")).Utf8ToString());
}
public virtual void TestFromCharSequence()
{
for (int i = 0; i < 100; i++)
{
ICharSequence s = new StringCharSequence(TestUtil.RandomUnicodeString(Random));
ICharSequence s2 = (new BytesRef(s)).Utf8ToString().AsCharSequence();
Assert.AreEqual(s, s2);
}
// only for 4.x
Assert.AreEqual("\uFFFF", (new BytesRef("\uFFFF")).Utf8ToString());
}
public virtual void TestUTF8toUTF32()
{
BytesRef utf8 = new BytesRef(20);
Int32sRef utf32 = new Int32sRef(20);
int[] codePoints = new int[20];
int num = AtLeast(50000);
for (int i = 0; i < num; i++)
{
string s = TestUtil.RandomUnicodeString(Random);
UnicodeUtil.UTF16toUTF8(s.ToCharArray(), 0, s.Length, utf8);
UnicodeUtil.UTF8toUTF32(utf8, utf32);
int charUpto = 0;
int intUpto = 0;
while (charUpto < s.Length)
{
int cp = Character.CodePointAt(s, charUpto);
codePoints[intUpto++] = cp;
charUpto += Character.CharCount(cp);
}
if (!ArrayUtil.Equals(codePoints, 0, utf32.Int32s, utf32.Offset, intUpto))
{
Console.WriteLine("FAILED");
for (int j = 0; j < s.Length; j++)
{
Console.WriteLine(" char[" + j + "]=" + ((int)s[j]).ToString("x"));
}
Console.WriteLine();
Assert.AreEqual(intUpto, utf32.Length);
for (int j = 0; j < intUpto; j++)
{
Console.WriteLine(" " + utf32.Int32s[j].ToString("x") + " vs " + codePoints[j].ToString("x"));
}
Assert.Fail("mismatch");
}
}
}
public virtual void TestUTF16InUTF8Order()
{
int numStrings = AtLeast(1000);
BytesRef[] utf8 = new BytesRef[numStrings];
CharsRef[] utf16 = new CharsRef[numStrings];
for (int i = 0; i < numStrings; i++)
{
string s = TestUtil.RandomUnicodeString(Random());
utf8[i] = new BytesRef(s);
utf16[i] = new CharsRef(s);
}
Array.Sort(utf8);
Array.Sort(utf16, CharsRef.UTF16SortedAsUTF8Comparer);
for (int i = 0; i < numStrings; i++)
{
Assert.AreEqual(utf8[i].Utf8ToString(), utf16[i].ToString());
}
}
public static string RandomSubString(Random random, int wordLength, bool simple)
{
if (wordLength == 0)
{
return("");
}
int evilness = TestUtil.NextInt32(random, 0, 20);
StringBuilder sb = new StringBuilder();
while (sb.Length < wordLength)
{
;
if (simple)
{
sb.Append(random.NextBoolean() ? TestUtil.RandomSimpleString(random, wordLength) : TestUtil.RandomHtmlishString(random, wordLength));
}
else
{
if (evilness < 10)
{
sb.Append(TestUtil.RandomSimpleString(random, wordLength));
}
else if (evilness < 15)
{
Assert.AreEqual(0, sb.Length); // we should always get wordLength back!
sb.Append(TestUtil.RandomRealisticUnicodeString(random, wordLength, wordLength));
}
else if (evilness == 16)
{
sb.Append(TestUtil.RandomHtmlishString(random, wordLength));
}
else if (evilness == 17)
{
// gives a lot of punctuation
sb.Append(TestUtil.RandomRegexpishString(random, wordLength));
}
else
{
sb.Append(TestUtil.RandomUnicodeString(random, wordLength));
}
}
}
if (sb.Length > wordLength)
{
sb.Length = wordLength;
if (char.IsHighSurrogate(sb[wordLength - 1]))
{
sb.Length = wordLength - 1;
}
}
if (random.Next(17) == 0)
{
// mix up case
string mixedUp = TestUtil.RandomlyRecaseString(random, sb.ToString());
Assert.True(mixedUp.Length == sb.Length, "Lengths are not the same: mixedUp = " + mixedUp + ", length = " + mixedUp.Length + ", sb = " + sb + ", length = " + sb.Length);
return(mixedUp);
}
else
{
return(sb.ToString());
}
}