static FA _Determinize(FA fa, IProgress <FAProgress> progress = null)
{
if (null != progress)
{
progress.Report(new FAProgress(FAStatus.DfaTransform, 0));
}
var p = new HashSet <int>();
var closure = new List <FA>();
fa.FillClosure(closure);
for (int ic = closure.Count, i = 0; i < ic; ++i)
{
var ffa = closure[i];
p.Add(0);
foreach (var t in ffa.InputTransitions)
{
p.Add(t.Key.Key);
if (t.Key.Value < 0x10ffff)
{
p.Add((t.Key.Value + 1));
}
}
}
var points = new int[p.Count];
p.CopyTo(points, 0);
Array.Sort(points);
var comparer = _SetComparer.Default;
var sets = new Dictionary <HashSet <FA>, HashSet <FA> >(comparer);
var working = new Queue <HashSet <FA> >();
var dfaMap = new Dictionary <HashSet <FA>, FA>(comparer);
var initial = new HashSet <FA>();
fa.FillEpsilonClosure(initial);
sets.Add(initial, initial);
working.Enqueue(initial);
var result = new FA();
foreach (var afa in initial)
{
if (afa.IsAccepting)
{
result.IsAccepting = true;
result.AcceptSymbol = afa.AcceptSymbol;
break;
}
}
dfaMap.Add(initial, result);
var j = 1;
while (working.Count > 0)
{
var s = working.Dequeue();
var ecs = FillEpsilonClosure(s);
FA dfa;
dfaMap.TryGetValue(s, out dfa);
foreach (FA q in ecs)
{
if (q.IsAccepting)
{
dfa.IsAccepting = true;
dfa.AcceptSymbol = q.AcceptSymbol;
break;
}
}
for (var i = 0; i < points.Length; i++)
{
var set = new HashSet <FA>();
foreach (FA c in ecs)
{
foreach (var trns in c.InputTransitions)
{
if (trns.Key.Key <= points[i] && points[i] <= trns.Key.Value)
{
foreach (var efa in trns.Value.FillEpsilonClosure())
{
set.Add(efa);
}
}
}
}
if (!sets.ContainsKey(set))
{
sets.Add(set, set);
working.Enqueue(set);
dfaMap.Add(set, new FA());
}
FA dst;
dfaMap.TryGetValue(set, out dst);
int first = points[i];
int last;
if (i + 1 < points.Length)
{
last = (points[i + 1] - 1);
}
else
{
last = 0x10ffff;
}
dfa.InputTransitions.Add(new KeyValuePair <int, int>(first, last), dst);
}
if (null != progress)
{
progress.Report(new FAProgress(FAStatus.DfaTransform, j));
}
++j;
}
// remove dead transitions
foreach (var ffa in result.FillClosure())
{
var itrns = new List <KeyValuePair <KeyValuePair <int, int>, FA> >(ffa.InputTransitions);
foreach (var trns in itrns)
{
if (null == trns.Value.FirstAcceptingState)
{
ffa.InputTransitions.Remove(trns.Key);
}
}
if (null != progress)
{
progress.Report(new FAProgress(FAStatus.DfaTransform, j));
}
++j;
}
return(result);
}
/// <summary>
/// Returns a DFA table that can be used to lex or match
/// </summary>
/// <param name="symbolTable">The symbol table to use, or null to just implicitly tag symbols with integer ids</param>
/// <returns>A DFA table that can be used to efficiently match or lex input</returns>
public DfaEntry[] ToDfaStateTable(IList <int> symbolTable = null, IProgress <FAProgress> progress = null)
{
// only convert to a DFA if we haven't already
// ToDfa() already checks but it always copies
// the state information so this performs better
FA dfa = null;
if (!IsDfa)
{
dfa = ToDfa(progress);
dfa.TrimDuplicates(progress);
}
else
{
dfa = this;
}
var closure = new List <FA>();
dfa.FillClosure(closure);
var symbolLookup = new Dictionary <int, int>();
// if we don't have a symbol table, build
// the symbol lookup from the states.
if (null == symbolTable)
{
// go through each state, looking for accept symbols
// and then add them to the new symbol table is we
// haven't already
var i = 0;
for (int jc = closure.Count, j = 0; j < jc; ++j)
{
var fa = closure[j];
if (fa.IsAccepting && !symbolLookup.ContainsKey(fa.AcceptSymbol))
{
if (0 > fa.AcceptSymbol)
{
throw new InvalidOperationException("An accept symbol was never specified for state q" + jc.ToString());
}
symbolLookup.Add(fa.AcceptSymbol, i);
++i;
}
}
}
else // build the symbol lookup from the symbol table
{
for (int ic = symbolTable.Count, i = 0; i < ic; ++i)
{
symbolLookup.Add(symbolTable[i], i);
}
}
// build the root array
var result = new DfaEntry[closure.Count];
for (var i = 0; i < result.Length; i++)
{
var fa = closure[i];
#if DEBUG
if (fa.IsAccepting)
{
System.Diagnostics.Debug.Assert(-1 < fa.AcceptSymbol, "Illegal accept symbol " + fa.AcceptSymbol.ToString() + " was found on state state q" + i.ToString());
}
#endif
// get all the transition ranges for each destination state
var trgs = fa.FillInputTransitionRangesGroupedByState();
// make a new transition entry array for our DFA state table
var trns = new DfaTransitionEntry[trgs.Count];
var j = 0;
// for each transition range
foreach (var trg in trgs)
{
// add the transition entry using
// the packed ranges from CharRange
trns[j] = new DfaTransitionEntry(
trg.Value,
closure.IndexOf(trg.Key));
++j;
}
// now add the state entry for the state above
#if DEBUG
if (fa.IsAccepting && !symbolLookup.ContainsKey(fa.AcceptSymbol))
{
try
{
dfa.RenderToFile(@"dfastatetable_crashdump_dfa.jpg");
}
catch
{
}
System.Diagnostics.Debug.Assert(false, "The symbol table did not contain an entry for state q" + i.ToString());
}
#endif
result[i] = new DfaEntry(
fa.IsAccepting ? symbolLookup[fa.AcceptSymbol] : -1,
trns);
}
return(result);
}
