public void MatchTestUtf8(string regex, bool caseSensitive, byte[] input, bool match)
{
var inputString = Encoding.UTF8.GetString(input);
this.output.WriteLine(inputString);
this.output.WriteLine(string.Join(" ", inputString.Select(c => ((int)c).ToString("x4"))));
var expression = RegexParser.Parse(regex, 0);
var mapper = new UnicodeUtf8Mapper(false, false);
var rxCodepoint = expression.ToInvariant(mapper, new UnicodeCharSetProvider(), caseSensitive);
var nfa = NfaBuilder <byte> .Build(rxCodepoint.Optimize());
var dfa = DfaBuilder <byte> .Build(nfa, Utf8Bytes.EOF);
var state = dfa.StartState;
foreach (var inputByte in input.Append(Utf8Bytes.EOF))
{
var nextStateId = state.GetTransition(inputByte);
if (nextStateId == Dfa <byte> .Reject)
{
Assert.False(match, "Regex has not matched input, but expected a match");
break;
}
if (nextStateId == Dfa <byte> .Accept)
{
Assert.True(match, "Regex has matched input, but expected a non-match");
break;
}
state = dfa.GetState(nextStateId);
Assert.NotNull(state);
}
}
public void CreateUtf8Diagram(string regex, bool caseSensitive)
{
var expression = RegexParser.Parse(regex, 0);
var mapper = new UnicodeUtf8Mapper(false, false);
var rxCodepoint = expression.ToInvariant(mapper, new UnicodeCharSetProvider(), caseSensitive);
var nfa = NfaBuilder <byte> .Build(rxCodepoint.Optimize());
var dfa = DfaBuilder <byte> .Build(nfa, 0xFF);
NfaFactoryTest.WriteDiagram(this.output, nfa);
WriteDiagram(this.output, dfa);
}
