/// <summary>
/// Initializes this Unicode block
/// </summary>
/// <param name="name">Block's name</param>
/// <param name="begin">Beginning character (included)</param>
/// <param name="end">End character (included)</param>
public UnicodeBlock(string name, int begin, int end)
{
this.name = name;
span = new UnicodeSpan(begin, end);
}
/// <summary>
/// Adds a unicode character span to an existing NFA automaton
/// </summary>
/// <param name="automata">The target NFA</param>
/// <param name="span">The unicode span to add</param>
private static void AddUnicodeSpanToNFA(NFA automata, UnicodeSpan span)
{
char[] b = span.Begin.GetUTF16();
char[] e = span.End.GetUTF16();
if (span.IsPlane0)
{
// this span is entirely in plane 0
automata.StateEntry.AddTransition(new CharSpan(b[0], e[0]), automata.StateExit);
}
else if (span.Begin.IsPlane0)
{
// this span has only a part in plane 0
if (b[0] < 0xD800)
{
automata.StateEntry.AddTransition(new CharSpan(b[0], (char)0xD7FF), automata.StateExit);
automata.StateEntry.AddTransition(new CharSpan((char)0xE000, (char)0xFFFF), automata.StateExit);
}
else
{
automata.StateEntry.AddTransition(new CharSpan(b[0], (char)0xFFFF), automata.StateExit);
}
NFAState intermediate = automata.AddNewState();
automata.StateEntry.AddTransition(new CharSpan((char)0xD800, e[0]), intermediate);
intermediate.AddTransition(new CharSpan((char)0xDC00, e[1]), automata.StateExit);
}
else
{
// this span has no part in plane 0
if (b[0] == e[0])
{
// same first surrogate
NFAState intermediate = automata.AddNewState();
automata.StateEntry.AddTransition(new CharSpan(b[0], b[0]), intermediate);
intermediate.AddTransition(new CharSpan(b[1], e[1]), automata.StateExit);
}
else if (e[0] == b[0] + 1)
{
// the first surrogates are consecutive encodings
// build lower half
NFAState i1 = automata.AddNewState();
automata.StateEntry.AddTransition(new CharSpan(b[0], b[0]), i1);
i1.AddTransition(new CharSpan(b[1], (char)0xDFFF), automata.StateExit);
// build upper half
NFAState i2 = automata.AddNewState();
automata.StateEntry.AddTransition(new CharSpan(e[0], e[0]), i2);
i2.AddTransition(new CharSpan((char)0xDC00, e[1]), automata.StateExit);
}
else
{
// there is at least one surrogate value between the first surrogates of begin and end
// build lower part
NFAState ia = automata.AddNewState();
automata.StateEntry.AddTransition(new CharSpan(b[0], b[0]), ia);
ia.AddTransition(new CharSpan(b[1], (char)0xDFFF), automata.StateExit);
// build intermediate part
NFAState im = automata.AddNewState();
automata.StateEntry.AddTransition(new CharSpan((char)(b[0] + 1), (char)(e[0] - 1)), im);
im.AddTransition(new CharSpan((char)0xDC00, (char)0xDFFF), automata.StateExit);
// build upper part
NFAState iz = automata.AddNewState();
automata.StateEntry.AddTransition(new CharSpan(e[0], e[0]), iz);
iz.AddTransition(new CharSpan((char)0xDC00, e[1]), automata.StateExit);
}
}
}
