private static IDfa <Optional <Rule <TLabel> >, TLabel> CompileRules(IList <Rule <TLabel> > rules)
{
var dfas = rules.ToDictionary(
keySelector: (rule) => Optional <Rule <TLabel> > .Some(rule),
elementSelector: (rule) => RegexToDfa(rule.Rhs));
var merged = DfaMerger <Optional <Rule <TLabel> >, TLabel> .Merge(
dfas,
conflictSolver : (matchesEnumerable) =>
{
var matches = matchesEnumerable.ToList();
if (matches.Count == 0)
{
return(Optional <Rule <TLabel> > .None());
}
else if (matches.Count == 1)
{
return(matches[0]);
}
else
{
throw new ArgumentException($"two rules have conflicting regexes (for symbol: {matches[0].Get().Lhs})");
}
});
return(DfaMinimizer <Optional <Rule <TLabel> >, TLabel> .Minimize(merged));
}
private static IDfa <bool, TLabel> RegexToDfa(Regex <TLabel> regex)
{
INfa <TLabel> nfa = RegexToNfaConverter <TLabel> .Convert(regex);
IDfa <bool, TLabel> dfa = NfaToDfaConverter <TLabel> .Convert(nfa);
return(DfaMinimizer <bool, TLabel> .Minimize(dfa));
}
public void TestMinimize(string[] strings, string expected)
{
var trie = new Trie();
trie.AddRange(strings);
var min = DfaMinimizer <char> .Minimize(trie);
var result = min.ToString();
Assert.AreEqual(expected, result);
}
private void CheckDfa(IMatchable regex, string resourceSection)
{
var nfa = new Nfa <int>();
var startStates = new[] { AddToNfa(nfa, regex, 1), AddToNfa(nfa, Pattern.Regex("a*(b|c)d+"), 2) };
var rawDfa = new DfaFromNfa <int>(nfa, startStates, null).GetDfa();
var minimalDfa = new DfaMinimizer <int>(rawDfa).GetMinimizedDfa();
PrintDot(minimalDfa);
// serializableDfa = new SerializableDfa<TResult>(minimalDfa);
// CheckNfa(nfa, state, $"NfaTests.out.txt#{resourceSection}", true);
}
private static void CheckMinimization <TLabel>(IDfa <TLabel, char> dfa, int expectedNumberOfStates = -1)
{
var minimalDfa = DfaMinimizer <TLabel, char> .Minimize(dfa);
var numberOfStates = ReachableStates(minimalDfa).Count;
if (expectedNumberOfStates != -1)
{
Assert.AreEqual(expectedNumberOfStates, numberOfStates);
}
CheckAutomatonEquivalence(dfa, minimalDfa);
CheckStateStability(minimalDfa);
}
// tokenCategories - List of pair (Token, Regex for thie Token)
public Lexer(
IEnumerable <KeyValuePair <TLabel, string> > tokenCategories,
TLabel eof,
TLabel noneValue,
Func <IEnumerable <TLabel>, TLabel> conflictSolver)
{
this.eof = eof;
this.noneValue = noneValue;
var converter = new StringToRegexConverterFactory().CreateConverter();
Dictionary <TLabel, IDfa <bool, char> > multipleDfa = tokenCategories.ToDictionary(
x => x.Key,
x =>
{
Regex <char> regex = converter.Convert(x.Value);
INfa <char> nfaPre = RegexToNfaConverter <char> .Convert(regex);
INfa <char> nfa = ConcreteNfa <char> .CreateFromNfa(nfaPre);
IDfa <bool, char> dfa = NfaToDfaConverter <char> .Convert(nfa);
return(DfaMinimizer <bool, char> .Minimize(dfa));
});
var mergedDfa = DfaMerger <TLabel, char> .Merge(multipleDfa, conflictSolver);
this.minimalizedDfa = DfaMinimizer <TLabel, char> .Minimize(mergedDfa);
}
