private void FindLeaves(RegExp exp, Kind kind, IList <Automaton> list, IDictionary <string, Automaton> automata, IAutomatonProvider automaton_provider)
{
if (exp.kind == kind)
{
FindLeaves(exp.exp1, kind, list, automata, automaton_provider);
FindLeaves(exp.exp2, kind, list, automata, automaton_provider);
}
else
{
list.Add(exp.ToAutomaton(automata, automaton_provider));
}
}
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!
}
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 virtual void TestGetRandomAcceptedString()
{
int ITER1 = AtLeast(100);
int ITER2 = AtLeast(100);
for (int i = 0; i < ITER1; i++)
{
RegExp re = new RegExp(AutomatonTestUtil.RandomRegexp(Random), RegExpSyntax.NONE);
Automaton a = re.ToAutomaton();
Assert.IsFalse(BasicOperations.IsEmpty(a));
RandomAcceptedStrings rx = new RandomAcceptedStrings(a);
for (int j = 0; j < ITER2; j++)
{
int[] acc = null;
try
{
acc = rx.GetRandomAcceptedString(Random);
string s = UnicodeUtil.NewString(acc, 0, acc.Length);
Assert.IsTrue(BasicOperations.Run(a, s));
}
catch (Exception /*t*/)
{
Console.WriteLine("regexp: " + re);
if (acc != null)
{
Console.WriteLine("fail acc re=" + re + " count=" + acc.Length);
for (int k = 0; k < acc.Length; k++)
{
Console.WriteLine(" " + acc[k].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)
}
}
}
}
public virtual void TestGetRandomAcceptedString()
{
int ITER1 = AtLeast(100);
int ITER2 = AtLeast(100);
for (int i = 0; i < ITER1; i++)
{
RegExp re = new RegExp(AutomatonTestUtil.RandomRegexp(Random()), RegExp.NONE);
Automaton a = re.ToAutomaton();
Assert.IsFalse(BasicOperations.IsEmpty(a));
AutomatonTestUtil.RandomAcceptedStrings rx = new AutomatonTestUtil.RandomAcceptedStrings(a);
for (int j = 0; j < ITER2; j++)
{
int[] acc = null;
try
{
acc = rx.GetRandomAcceptedString(Random());
string s = UnicodeUtil.NewString(acc, 0, acc.Length);
Assert.IsTrue(BasicOperations.Run(a, s));
}
catch (Exception t)
{
Console.WriteLine("regexp: " + re);
if (acc != null)
{
Console.WriteLine("fail acc re=" + re + " count=" + acc.Length);
for (int k = 0; k < acc.Length; k++)
{
Console.WriteLine(" " + acc[k].ToString("x"));
}
}
throw t;
}
}
}
}
public virtual void TestGetRandomAcceptedString()
{
int ITER1 = AtLeast(100);
int ITER2 = AtLeast(100);
for (int i = 0; i < ITER1; i++)
{
RegExp re = new RegExp(AutomatonTestUtil.RandomRegexp(Random()), RegExp.NONE);
Automaton a = re.ToAutomaton();
Assert.IsFalse(BasicOperations.IsEmpty(a));
AutomatonTestUtil.RandomAcceptedStrings rx = new AutomatonTestUtil.RandomAcceptedStrings(a);
for (int j = 0; j < ITER2; j++)
{
int[] acc = null;
try
{
acc = rx.GetRandomAcceptedString(Random());
string s = UnicodeUtil.NewString(acc, 0, acc.Length);
Assert.IsTrue(BasicOperations.Run(a, s));
}
catch (Exception t)
{
Console.WriteLine("regexp: " + re);
if (acc != null)
{
Console.WriteLine("fail acc re=" + re + " count=" + acc.Length);
for (int k = 0; k < acc.Length; k++)
{
Console.WriteLine(" " + acc[k].ToString("x"));
}
}
throw t;
}
}
}
}
private Automaton ToAutomaton(IDictionary <string, Automaton> automata, IAutomatonProvider automaton_provider)
{
IList <Automaton> list;
Automaton a = null;
switch (kind)
{
case Kind.REGEXP_UNION:
list = new List <Automaton>();
FindLeaves(exp1, Kind.REGEXP_UNION, list, automata, automaton_provider);
FindLeaves(exp2, Kind.REGEXP_UNION, list, automata, automaton_provider);
a = BasicOperations.Union(list);
MinimizationOperations.Minimize(a);
break;
case Kind.REGEXP_CONCATENATION:
list = new List <Automaton>();
FindLeaves(exp1, Kind.REGEXP_CONCATENATION, list, automata, automaton_provider);
FindLeaves(exp2, Kind.REGEXP_CONCATENATION, list, automata, automaton_provider);
a = BasicOperations.Concatenate(list);
MinimizationOperations.Minimize(a);
break;
case Kind.REGEXP_INTERSECTION:
a = exp1.ToAutomaton(automata, automaton_provider).Intersection(exp2.ToAutomaton(automata, automaton_provider));
MinimizationOperations.Minimize(a);
break;
case Kind.REGEXP_OPTIONAL:
a = exp1.ToAutomaton(automata, automaton_provider).Optional();
MinimizationOperations.Minimize(a);
break;
case Kind.REGEXP_REPEAT:
a = exp1.ToAutomaton(automata, automaton_provider).Repeat();
MinimizationOperations.Minimize(a);
break;
case Kind.REGEXP_REPEAT_MIN:
a = exp1.ToAutomaton(automata, automaton_provider).Repeat(min);
MinimizationOperations.Minimize(a);
break;
case Kind.REGEXP_REPEAT_MINMAX:
a = exp1.ToAutomaton(automata, automaton_provider).Repeat(min, max);
MinimizationOperations.Minimize(a);
break;
case Kind.REGEXP_COMPLEMENT:
a = exp1.ToAutomaton(automata, automaton_provider).Complement();
MinimizationOperations.Minimize(a);
break;
case Kind.REGEXP_CHAR:
a = BasicAutomata.MakeChar(c);
break;
case Kind.REGEXP_CHAR_RANGE:
a = BasicAutomata.MakeCharRange(from, to);
break;
case Kind.REGEXP_ANYCHAR:
a = BasicAutomata.MakeAnyChar();
break;
case Kind.REGEXP_EMPTY:
a = BasicAutomata.MakeEmpty();
break;
case Kind.REGEXP_STRING:
a = BasicAutomata.MakeString(s);
break;
case Kind.REGEXP_ANYSTRING:
a = BasicAutomata.MakeAnyString();
break;
case Kind.REGEXP_AUTOMATON:
Automaton aa = null;
if (automata != null)
{
aa = automata[s];
}
if (aa == null && automaton_provider != null)
{
try
{
aa = automaton_provider.GetAutomaton(s);
}
catch (Exception e) when(e.IsIOException())
{
throw new ArgumentException(e.ToString(), e);
}
}
if (aa == null)
{
throw new ArgumentException("'" + s + "' not found");
}
a = (Automaton)aa.Clone(); // always clone here (ignore allow_mutate)
break;
case Kind.REGEXP_INTERVAL:
a = BasicAutomata.MakeInterval(min, max, digits);
break;
}
return(a);
}
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));
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!
}
private void FindLeaves(RegExp exp, Kind kind, IList<Automaton> list, IDictionary<string, Automaton> automata, AutomatonProvider automaton_provider)
{
if (exp.kind == kind)
{
FindLeaves(exp.Exp1, kind, list, automata, automaton_provider);
FindLeaves(exp.Exp2, kind, list, automata, automaton_provider);
}
else
{
list.Add(exp.ToAutomaton(automata, automaton_provider));
}
}
internal DumbRegexpQuery(Term term, RegExpSyntax flags)
: base(term.Field)
{
RegExp re = new RegExp(term.Text, flags);
automaton = re.ToAutomaton();
}
