static ReElement ParseRepitition(ReElement element, ReParseContext context)
{
var min = ParseInt(context) ?? 0;
if (context.Current == ',')
{
if (!context.MoveNext())
{
ThrowUnexpectedEnd();
}
var max = ParseInt(context);
return(ReFactory.NewRepetition(element, min, max));
}
else
{
return(ReFactory.NewRepetition(element, min, min));
}
}
static ReElement ParseSequence(ReParseContext context)
{
var choice = new List <ReElement>();
var sequence = new List <ReElement>();
var abort = false;
while (!context.End && !abort)
{
switch (context.Current)
{
case '[':
if (!context.MoveNext())
{
ThrowUnexpectedEnd();
}
sequence.Add(ParseCharSet(context));
if (context.End)
{
ThrowUnexpectedEnd();
}
if (context.Current != ']')
{
ThrowUnexpectedChar(context);
}
context.MoveNext();
break;
case '(':
if (!context.MoveNext())
{
ThrowUnexpectedEnd();
}
sequence.Add(ParseSequence(context));
if (context.End)
{
ThrowUnexpectedEnd();
}
if (context.Current != ')')
{
ThrowUnexpectedChar(context);
}
context.MoveNext();
break;
case ')':
abort = true;
break;
case '|':
choice.Add(ReFactory.NewConcatenation(sequence));
sequence.Clear();
context.MoveNext();
break;
case '.':
sequence.Add(ReFactory.NewSingleton(CharSet.LineBody));
context.MoveNext();
break;
case '?':
if (sequence.Count == 0)
{
ThrowUnexpectedChar(context);
}
sequence[sequence.Count - 1] = ReFactory.NewRepetition(sequence[sequence.Count - 1], 0, 1);
context.MoveNext();
break;
case '+':
if (sequence.Count == 0)
{
ThrowUnexpectedChar(context);
}
sequence[sequence.Count - 1] = ReFactory.NewRepetition(sequence[sequence.Count - 1], 1, null);
context.MoveNext();
break;
case '*':
if (sequence.Count == 0)
{
ThrowUnexpectedChar(context);
}
sequence[sequence.Count - 1] = ReFactory.NewRepetition(sequence[sequence.Count - 1], 0, null);
context.MoveNext();
break;
case '{':
if (sequence.Count == 0)
{
ThrowUnexpectedChar(context);
}
if (!context.MoveNext())
{
ThrowUnexpectedEnd();
}
sequence[sequence.Count - 1] = ParseRepitition(sequence[sequence.Count - 1], context);
if (context.Current != '}')
{
ThrowUnexpectedChar(context);
}
context.MoveNext();
break;
case '\\':
if (!context.MoveNext())
{
ThrowUnexpectedEnd();
}
sequence.Add(ReFactory.NewSingleton(CharSet.New(ParseEscapeSequence(context))));
break;
default:
sequence.Add(ReFactory.NewSingleton(CharSet.New(context.Current)));
context.MoveNext();
break;
}
}
choice.Add(ReFactory.NewConcatenation(sequence));
return(ReFactory.NewUnion(choice));
}
static ReElement ParseCharSet(ReParseContext context)
{
var ranges = new List <CharRange>();
var invert = false;
if (context.Current == '^')
{
invert = true;
if (!context.MoveNext())
{
ThrowUnexpectedEnd();
}
}
if (context.Current == ']')
{
ranges.Add(new CharRange(']', ']'));
if (!context.MoveNext())
{
ThrowUnexpectedEnd();
}
}
var abort = false;
while (!context.End && !abort)
{
switch (context.Current)
{
case ']':
abort = true;
break;
case '-':
if (ranges.Count == 0)
{
ranges.Add(new CharRange('-', '-'));
}
else
{
var from = ranges[ranges.Count - 1].To;
if (!context.MoveNext())
{
ThrowUnexpectedEnd();
}
var to = context.Current;
if (to == ']')
{
ranges.Add(new CharRange('-', '-'));
}
else
{
if (!context.MoveNext())
{
ThrowUnexpectedEnd();
}
if (to == '\\')
{
to = ParseEscapeSequence(context);
if (context.End)
{
ThrowUnexpectedEnd();
}
}
if (to < from)
{
ThrowBrokenCharRange(from, to);
}
ranges.Add(new CharRange(from, to));
}
}
break;
case '\\':
if (!context.MoveNext())
{
ThrowUnexpectedEnd();
}
var c = ParseEscapeSequence(context);
ranges.Add(new CharRange(c, c));
break;
default:
ranges.Add(new CharRange(context.Current, context.Current));
if (!context.MoveNext())
{
ThrowUnexpectedEnd();
}
break;
}
}
var set = CharSet.New(ranges.ToArray());
if (invert)
{
set = CharSet.Universal.Subtract(set);
}
return(ReFactory.NewSingleton(set));
}
