public void TestSpecialCase()
{
RegExp re = new RegExp(".?");
Automaton automaton = re.ToAutomaton();
CharacterRunAutomaton cra = new CharacterRunAutomaton(automaton);
ByteRunAutomaton bra = new ByteRunAutomaton(automaton);
// make sure character dfa accepts empty string
Assert.IsTrue(cra.IsAccept(cra.InitialState));
Assert.IsTrue(cra.Run(""));
Assert.IsTrue(cra.Run(new char[0], 0, 0));
// make sure byte dfa accepts empty string
Assert.IsTrue(bra.IsAccept(bra.InitialState));
Assert.IsTrue(bra.Run(new byte[0], 0, 0));
}
public void TestSpecialCase3()
{
RegExp re = new RegExp("(\\鯺)*(.)*\\Ӕ");
string input = "\u5cfd\ufffd\ub2f7\u0033\ue304\u51d7\u3692\udb50\udfb3\u0576\udae2\udc62\u0053\u0449\u04d4";
Automaton automaton = re.ToAutomaton();
CharacterRunAutomaton cra = new CharacterRunAutomaton(automaton);
ByteRunAutomaton bra = new ByteRunAutomaton(automaton);
Assert.IsTrue(cra.Run(input));
sbyte[] bytes = input.GetBytes(Encoding.UTF8);
Assert.IsTrue(bra.Run(bytes, 0, bytes.Length));
}
public void TestSpecialCase2()
{
RegExp re = new RegExp(".+\u0775");
string input = "\ufadc\ufffd\ub80b\uda5a\udc68\uf234\u0056\uda5b\udcc1\ufffd\ufffd\u0775";
Automaton automaton = re.ToAutomaton();
CharacterRunAutomaton cra = new CharacterRunAutomaton(automaton);
ByteRunAutomaton bra = new ByteRunAutomaton(automaton);
Assert.IsTrue(cra.Run(input));
sbyte[] bytes = input.GetBytes(Encoding.UTF8);
Assert.IsTrue(bra.Run(bytes, 0, bytes.Length)); // this one fails!
}
private void AssertBruteForceT(string input, Automaton dfa, int distance)
{
CharacterRunAutomaton ra = new CharacterRunAutomaton(dfa);
int maxLen = input.Length + distance + 1;
int maxNum = (int)Math.Pow(2, maxLen);
for (int i = 0; i < maxNum; i++)
{
string encoded = Convert.ToString(i, 2);
bool accepts = ra.Run(encoded);
if (accepts)
{
Assert.IsTrue(GetTDistance(input, encoded) <= distance);
}
else
{
Assert.IsTrue(GetTDistance(input, encoded) > distance);
}
}
}
private void AssertBruteForceT(string input, Automaton dfa, int distance)
{
CharacterRunAutomaton ra = new CharacterRunAutomaton(dfa);
int maxLen = input.Length + distance + 1;
int maxNum = (int)Math.Pow(2, maxLen);
for (int i = 0; i < maxNum; i++)
{
string encoded = Convert.ToString(i, 2);
bool accepts = ra.Run(encoded);
if (accepts)
{
Assert.IsTrue(GetTDistance(input, encoded) <= distance);
}
else
{
Assert.IsTrue(GetTDistance(input, encoded) > distance);
}
}
}
private static void AssertAutomaton(Automaton automaton)
{
var cra = new CharacterRunAutomaton(automaton);
var bra = new ByteRunAutomaton(automaton);
var ras = new AutomatonTestUtil.RandomAcceptedStrings(automaton);
int num = AtLeast(1000);
for (int i = 0; i < num; i++)
{
string s;
if (Random().NextBoolean())
{
// likely not accepted
s = TestUtil.RandomUnicodeString(Random());
}
else
{
// will be accepted
int[] codepoints = ras.GetRandomAcceptedString(Random());
try
{
s = UnicodeUtil.NewString(codepoints, 0, codepoints.Length);
}
catch (Exception e)
{
Console.WriteLine(codepoints.Length + " codepoints:");
for (int j = 0; j < codepoints.Length; j++)
{
Console.WriteLine(" " + codepoints[j].ToString("x"));
}
throw e;
}
}
var bytes = s.GetBytes(Encoding.UTF8);
Assert.AreEqual(cra.Run(s), bra.Run(bytes, 0, bytes.Length));
}
}
private static void AssertAutomaton(Automaton automaton)
{
var cra = new CharacterRunAutomaton(automaton);
var bra = new ByteRunAutomaton(automaton);
var ras = new RandomAcceptedStrings(automaton);
int num = AtLeast(1000);
for (int i = 0; i < num; i++)
{
string s;
if (Random.NextBoolean())
{
// likely not accepted
s = TestUtil.RandomUnicodeString(Random);
}
else
{
// will be accepted
int[] codepoints = ras.GetRandomAcceptedString(Random);
try
{
s = UnicodeUtil.NewString(codepoints, 0, codepoints.Length);
}
catch (Exception /*e*/)
{
Console.WriteLine(codepoints.Length + " codepoints:");
for (int j = 0; j < codepoints.Length; j++)
{
Console.WriteLine(" " + codepoints[j].ToString("x"));
}
throw; // LUCENENET: CA2200: Rethrow to preserve stack details (https://docs.microsoft.com/en-us/visualstudio/code-quality/ca2200-rethrow-to-preserve-stack-details)
}
}
var bytes = s.GetBytes(Encoding.UTF8);
Assert.AreEqual(cra.Run(s), bra.Run(bytes, 0, bytes.Length));
}
}
public void TestSpecialCase()
{
RegExp re = new RegExp(".?");
Automaton automaton = re.ToAutomaton();
CharacterRunAutomaton cra = new CharacterRunAutomaton(automaton);
ByteRunAutomaton bra = new ByteRunAutomaton(automaton);
// make sure character dfa accepts empty string
Assert.IsTrue(cra.IsAccept(cra.InitialState));
Assert.IsTrue(cra.Run(""));
Assert.IsTrue(cra.Run(new char[0], 0, 0));
// make sure byte dfa accepts empty string
Assert.IsTrue(bra.IsAccept(bra.InitialState));
Assert.IsTrue(bra.Run(new byte[0], 0, 0));
}
private static void AssertAutomaton(Automaton automaton)
{
var cra = new CharacterRunAutomaton(automaton);
var bra = new ByteRunAutomaton(automaton);
var ras = new AutomatonTestUtil.RandomAcceptedStrings(automaton);
int num = AtLeast(1000);
for (int i = 0; i < num; i++)
{
string s;
if (Random().NextBoolean())
{
// likely not accepted
s = TestUtil.RandomUnicodeString(Random());
}
else
{
// will be accepted
int[] codepoints = ras.GetRandomAcceptedString(Random());
try
{
s = UnicodeUtil.NewString(codepoints, 0, codepoints.Length);
}
catch (Exception e)
{
Console.WriteLine(codepoints.Length + " codepoints:");
for (int j = 0; j < codepoints.Length; j++)
{
Console.WriteLine(" " + codepoints[j].ToString("x"));
}
throw e;
}
}
var bytes = s.GetBytes(Encoding.UTF8);
Assert.AreEqual(cra.Run(s), bra.Run(bytes, 0, bytes.Length));
}
}
public void TestSpecialCase3()
{
RegExp re = new RegExp("(\\鯺)*(.)*\\Ӕ");
string input = "\u5cfd\ufffd\ub2f7\u0033\ue304\u51d7\u3692\udb50\udfb3\u0576\udae2\udc62\u0053\u0449\u04d4";
Automaton automaton = re.ToAutomaton();
CharacterRunAutomaton cra = new CharacterRunAutomaton(automaton);
ByteRunAutomaton bra = new ByteRunAutomaton(automaton);
Assert.IsTrue(cra.Run(input));
var bytes = input.GetBytes(Encoding.UTF8);
Assert.IsTrue(bra.Run(bytes, 0, bytes.Length));
}
public void TestSpecialCase2()
{
RegExp re = new RegExp(".+\u0775");
string input = "\ufadc\ufffd\ub80b\uda5a\udc68\uf234\u0056\uda5b\udcc1\ufffd\ufffd\u0775";
Automaton automaton = re.ToAutomaton();
CharacterRunAutomaton cra = new CharacterRunAutomaton(automaton);
ByteRunAutomaton bra = new ByteRunAutomaton(automaton);
Assert.IsTrue(cra.Run(input));
var bytes = input.GetBytes(Encoding.UTF8);
Assert.IsTrue(bra.Run(bytes, 0, bytes.Length)); // this one fails!
}
protected override AcceptStatus Accept(BytesRef term)
{
UnicodeUtil.UTF8toUTF16(term.Bytes, term.Offset, term.Length, utf16);
return runAutomaton.Run(utf16.Chars, 0, utf16.Length) ? AcceptStatus.YES : AcceptStatus.NO;
}