public FA ToFA(int match)
{
Ast ast = this;
if (ast.IsLazy)
{
throw new NotSupportedException("The AST node cannot be lazy");
}
switch (ast.Kind)
{
case Ast.Alt:
return(FA.Or(_ToFAs(ast.Exprs, match), match));
case Ast.Cat:
if (1 == ast.Exprs.Length)
{
return(ast.Exprs[0].ToFA(match));
}
return(FA.Concat(_ToFAs(ast.Exprs, match), match));
case Ast.Dot:
return(FA.Set(new int[] { 0, 0xd7ff, 0xe000, 0x10ffff }, match));
case Ast.Lit:
return(FA.Literal(new int[] { ast.Value }, match));
case Ast.NSet:
var pairs = RangeUtility.ToPairs(ast.Ranges);
RangeUtility.NormalizeRangeList(pairs);
var pairl = new List <KeyValuePair <int, int> >(RangeUtility.NotRanges(pairs));
return(FA.Set(RangeUtility.FromPairs(pairl), match));
case Ast.NUCode:
pairs = RangeUtility.ToPairs(CharacterClasses.UnicodeCategories[ast.Value]);
RangeUtility.NormalizeRangeList(pairs);
pairl = new List <KeyValuePair <int, int> >(RangeUtility.NotRanges(pairs));
return(FA.Set(RangeUtility.FromPairs(pairl), match));
case Ast.Opt:
return(FA.Optional(ast.Exprs[0].ToFA(match), match));
case Ast.Plus:
return(FA.Repeat(ast.Exprs[0].ToFA(match), 1, 0, match));
case Ast.Rep:
return(FA.Repeat(ast.Exprs[0].ToFA(match), ast.Min, ast.Max, match));
case Ast.Set:
return(FA.Set(ast.Ranges, match));
case Ast.Star:
return(FA.Repeat(ast.Exprs[0].ToFA(match), 0, 0, match));
case Ast.UCode:
return(FA.Set(CharacterClasses.UnicodeCategories[ast.Value], match));
default:
throw new NotImplementedException();
}
}
static int[] _GetFirsts(int[][] part, int index)
{
if (part.Length <= index)
{
return(new int[0]);
}
int idx;
List <int> resl;
int[] result;
var pc = part[index];
switch (pc[0])
{
case Char:
return(new int[] { pc[1], pc[1] });
case Set:
result = new int[pc.Length - 1];
Array.Copy(pc, 1, result, 0, result.Length);
return(result);
case NSet:
result = new int[pc.Length - 1];
Array.Copy(pc, 1, result, 0, result.Length);
return(RangeUtility.FromPairs(new List <KeyValuePair <int, int> >(RangeUtility.NotRanges(RangeUtility.ToPairs(result)))));
case Any:
return(new int[] { 0, 0x10ffff });
case UCode:
result = CharacterClasses.UnicodeCategories[pc[1]];
return(result);
case NUCode:
result = CharacterClasses.UnicodeCategories[pc[1]];
Array.Copy(pc, 1, result, 0, result.Length);
return(RangeUtility.FromPairs(new List <KeyValuePair <int, int> >(RangeUtility.NotRanges(RangeUtility.ToPairs(result)))));
case Switch:
resl = new List <int>();
idx = 1;
while (pc.Length > idx && -2 != pc[idx])
{
if (-1 == pc[idx])
{
idx += 2;
continue;
}
resl.Add(pc[idx]);
}
if (pc.Length > idx && -2 == pc[idx])
{
++idx;
while (pc.Length > idx)
{
resl.AddRange(_GetFirsts(part, pc[idx]));
++idx;
}
}
return(resl.ToArray());
case Jmp:
resl = new List <int>();
idx = 1;
while (pc.Length > idx)
{
resl.AddRange(_GetFirsts(part, pc[idx]));
++idx;
}
return(resl.ToArray());
case Match:
return(new int[0]);
case Save:
return(_GetFirsts(part, index + 1));
}
// should never get here
throw new NotImplementedException();
}