internal SymbolicRegexNode <S> Parse(string s, int i, out int i_next)
{
switch (s[i])
{
case '.':
{
#region character class of all characters
i_next = i + 1;
return(this.dot);
#endregion
}
case '[':
{
#region parse character class
if (s[i + 1] == ']')
{
i_next = i + 2;
return(this.nothing);
}
else
{
int j = s.IndexOf(']', i);
int[] atomIds = Array.ConvertAll(s.Substring(i + 1, j - (i + 1)).Split(','), x => int.Parse(x));
S[] bva = Array.ConvertAll(atomIds, id => this.solver.GetPartition()[id]);
var bv = this.solver.MkOr(bva);
SymbolicRegexNode <S> node;
if (!this.singletonCache.TryGetValue(bv, out node))
{
node = SymbolicRegexNode <S> .MkSingleton(this, bv);
this.singletonCache[bv] = node;
}
i_next = j + 1;
return(node);
}
#endregion
}
case 'E':
{
#region Epsilon
i_next = i + 1;
return(this.epsilon);
#endregion
}
case 'L': //L(l,u,body) for body{l,u} u may be *
{
#region Loop
int j = s.IndexOf(',', i + 2);
int lower = int.Parse(s.Substring(i + 2, j - (i + 2)));
int upper = int.MaxValue;
if (s[j + 1] == '*')
{
j = j + 3;
}
else
{
int k = s.IndexOf(',', j + 1);
upper = int.Parse(s.Substring(j + 1, k - (j + 1)));
j = k + 1;
}
int n;
var body = Parse(s, j, out n);
var node = SymbolicRegexNode <S> .MkLoop(this, body, lower, upper);
i_next = n + 1;
return(node);
#endregion
}
case 'S': //binary concat S(R1,R2)
{
#region concatenation
int n;
var first = Parse(s, i + 2, out n);
int m;
var second = Parse(s, n + 1, out m);
var seq = SymbolicRegexNode <S> .MkConcat(this, first, second);
i_next = m + 1;
return(seq);
#endregion
}
case 'C': //conjunction C(R1,R2,...,Rk)
{
#region conjunction
int n;
SymbolicRegexNode <S>[] nodes = ParseSequence(s, i + 2, out n);
var conj = SymbolicRegexNode <S> .MkAnd(this, nodes);
i_next = n;
return(conj);
#endregion
}
case 'D': //Disjunction D(R1,R2,...,Rk)
{
#region disjunction
int n;
SymbolicRegexNode <S>[] nodes = ParseSequence(s, i + 2, out n);
var disj = SymbolicRegexNode <S> .MkOr(this, nodes);
i_next = n;
return(disj);
#endregion
}
case 'I': //if then else I(x,y,z)
{
#region ITE
int n;
var cond = Parse(s, i + 2, out n);
int m;
var first = Parse(s, n + 1, out m);
int k;
var second = Parse(s, m + 1, out k);
var ite = SymbolicRegexNode <S> .MkIfThenElse(this, cond, first, second);
i_next = k + 1;
return(ite);
#endregion
}
case '^':
{
#region start anchor
i_next = i + 1;
return(this.startAnchor);
#endregion
}
case '$':
{
#region start anchor
i_next = i + 1;
return(this.endAnchor);
#endregion
}
default:
throw new NotImplementedException();
}
}
/// <summary>
/// Make an if-then-else regex (?(cond)left|right),
/// or create it as conjuction if right is false
/// </summary>
/// <param name="cond">condition</param>
/// <param name="left">true case</param>
/// <param name="right">false case</param>
/// <returns></returns>
public SymbolicRegexNode <S> MkIfThenElse(SymbolicRegexNode <S> cond, SymbolicRegexNode <S> left, SymbolicRegexNode <S> right)
{
return(SymbolicRegexNode <S> .MkIfThenElse(this, cond, left, right));
}
internal SymbolicRegexNode <S> Parse(string s, int i, out int i_next)
{
switch (s[i])
{
case '.':
{
#region .
i_next = i + 1;
return(this.dot);
#endregion
}
case '[':
{
#region parse singleton
int j = s.IndexOf(']', i);
var p = solver.DeserializePredicate(s.Substring(i + 1, j - (i + 1)));
var node = this.MkSingleton(p);
//SymbolicRegexNode<S> node;
//var seq_str = s.Substring(i + 1, j - (i + 1));
//var preds_str = seq_str.Split(';');
//var preds = Array.ConvertAll(preds_str, solver.DeserializePredicate);
//node = this.MkSequence(preds);
i_next = j + 1;
return(node);
#endregion
}
case 'E':
{
#region Epsilon
i_next = i + 1;
return(this.epsilon);
#endregion
}
case 'L': //L(l,u,body) for body{l,u} u may be *
{
#region Loop
int j = s.IndexOf(',', i + 2);
int lower = int.Parse(s.Substring(i + 2, j - (i + 2)));
int upper = int.MaxValue;
if (s[j + 1] == '*')
{
j = j + 3;
}
else
{
int k = s.IndexOf(',', j + 1);
upper = int.Parse(s.Substring(j + 1, k - (j + 1)));
j = k + 1;
}
int n;
var body = Parse(s, j, out n);
var node = SymbolicRegexNode <S> .MkLoop(this, body, lower, upper, false);
i_next = n + 1;
return(node);
#endregion
}
case 'Z': //Z(l,u,body) for body{l,u}? u may be *
{
#region Loop
int j = s.IndexOf(',', i + 2);
int lower = int.Parse(s.Substring(i + 2, j - (i + 2)));
int upper = int.MaxValue;
if (s[j + 1] == '*')
{
j = j + 3;
}
else
{
int k = s.IndexOf(',', j + 1);
upper = int.Parse(s.Substring(j + 1, k - (j + 1)));
j = k + 1;
}
int n;
var body = Parse(s, j, out n);
var node = SymbolicRegexNode <S> .MkLoop(this, body, lower, upper, true);
i_next = n + 1;
return(node);
#endregion
}
case 'S':
{
#region concatenation
int n;
SymbolicRegexNode <S>[] nodes = ParseSequence(s, i + 2, out n);
var concat = this.MkConcat(nodes, false);
i_next = n;
return(concat);
#endregion
}
case 'C': //conjunction C(R1,R2,...,Rk)
{
#region conjunction
int n;
SymbolicRegexNode <S>[] nodes = ParseSequence(s, i + 2, out n);
var conj = SymbolicRegexNode <S> .MkAnd(this, nodes);
i_next = n;
return(conj);
#endregion
}
case 'D': //Disjunction D(R1,R2,...,Rk)
{
#region disjunction
int n;
SymbolicRegexNode <S>[] nodes = ParseSequence(s, i + 2, out n);
var disj = SymbolicRegexNode <S> .MkOr(this, nodes);
i_next = n;
return(disj);
#endregion
}
case 'I': //if then else I(x,y,z)
{
#region ITE
int n;
var cond = Parse(s, i + 2, out n);
int m;
var first = Parse(s, n + 1, out m);
int k;
var second = Parse(s, m + 1, out k);
var ite = SymbolicRegexNode <S> .MkIfThenElse(this, cond, first, second);
i_next = k + 1;
return(ite);
#endregion
}
case '^':
{
#region start anchor
i_next = i + 1;
return(this.startAnchor);
#endregion
}
case '$':
{
#region end anchor
i_next = i + 1;
return(this.endAnchor);
#endregion
}
case '#':
{
#region end of sequence anchor
int j = s.IndexOf(')', i + 2);
int length = int.Parse(s.Substring(i + 2, j - (i + 2)));
i_next = j + 1;
return(SymbolicRegexNode <S> .MkWatchDog(this, length));
#endregion
}
default:
throw new NotImplementedException();
}
}