private static EBNF ParseAtom(LinkedList <string> tokens, ref string ExtraWhitespace, ref string ExtraComments)
{
if (tokens.First.Value == "(")
{
tokens.RemoveFirst();
var p = ParseProduction(tokens, ref ExtraWhitespace, ref ExtraComments);
if (tokens.Count == 0 || tokens.First.Value != ")")
{
throw new Exception("mismatched parentheses");
}
tokens.RemoveFirst();
GobbleUpComments(tokens, ref p.FollowingWhitespace, ref p.FollowingComment, ref p.FollowingNewline);
return(p);
}
else if (tokens.First.Value.StartsWith("'"))
{
var t = tokens.First.Value; tokens.RemoveFirst();
t = t.Substring(1, t.Length - 2);
var p = new EBNF {
Kind = EBNFKind.Terminal, s = t
};
if (t.StartsWith("<") && t.EndsWith(">"))
{
p.Kind = EBNFKind.ExtendedTerminal; p.s = t.Substring(1, t.Length - 2);
if (p.s.Contains("?"))
{
throw new Exception("A special-terminal may not contain a question-mark '?'");
}
if (p.s == "")
{
throw new Exception("A terminal may not be '<>'");
}
}
else
{
if (t.Contains("'") && t.Contains("\""))
{
throw new Exception("A terminal must either contain no ' or no \"");
}
}
GobbleUpComments(tokens, ref p.FollowingWhitespace, ref p.FollowingComment, ref p.FollowingNewline);
return(p);
}
else
{
var t = tokens.First.Value; tokens.RemoveFirst();
var p = new EBNF {
Kind = EBNFKind.Reference, s = t
};
GobbleUpComments(tokens, ref p.FollowingWhitespace, ref p.FollowingComment, ref p.FollowingNewline);
return(p);
}
}
private static EBNF ParseUnary(LinkedList <string> tokens, ref string ExtraWhitespace, ref string ExtraComments)
{
var p = ParseAtom(tokens, ref ExtraWhitespace, ref ExtraComments);
while (tokens.Any())
{
if (tokens.First.Value == "+")
{
tokens.RemoveFirst();
p = new EBNF {
Kind = EBNFKind.OneOrMoreOf, Children = new[] { p }.ToList()
};
GobbleUpComments(tokens, ref p.FollowingWhitespace, ref p.FollowingComment, ref p.FollowingNewline);
}
else if (tokens.First.Value == "*")
{
tokens.RemoveFirst();
p = new EBNF {
Kind = EBNFKind.ZeroOrMoreOf, Children = new[] { p }.ToList()
};
GobbleUpComments(tokens, ref p.FollowingWhitespace, ref p.FollowingComment, ref p.FollowingNewline);
}
else if (tokens.First.Value == "?")
{
tokens.RemoveFirst();
p = new EBNF {
Kind = EBNFKind.ZeroOrOneOf, Children = new[] { p }.ToList()
};
GobbleUpComments(tokens, ref p.FollowingWhitespace, ref p.FollowingComment, ref p.FollowingNewline);
}
else
{
break;
}
}
return(p);
}
public static string ToString(EBNF ebnf)
{
var r = "";
EBNF prevElement = null;
switch (ebnf.Kind)
{
case EBNFKind.Terminal:
r = $"'{ebnf.s.Replace("\\", "\\\\").Replace("'", "\\'")}'";
break;
case EBNFKind.ExtendedTerminal:
r = $"'<{ebnf.s.Replace("\\", "\\\\").Replace("'", "\\'")}>'";
break;
case EBNFKind.Reference:
r = ebnf.s;
break;
case EBNFKind.OneOrMoreOf:
case EBNFKind.ZeroOrMoreOf:
case EBNFKind.ZeroOrOneOf:
var op = (ebnf.Kind == EBNFKind.OneOrMoreOf ? "+" : (ebnf.Kind == EBNFKind.ZeroOrMoreOf ? "*" : "?"));
if (ebnf.Children[0].Kind == EBNFKind.Choice || ebnf.Children[0].Kind == EBNFKind.Sequence)
{
r = $"( {ToString(ebnf.Children[0])} ){op}";
}
else
{
r = $"{ToString(ebnf.Children[0])}{op}";
}
break;
case EBNFKind.Choice:
foreach (var c in ebnf.Children)
{
if (prevElement != null)
{
r += (r.Last() == '\t' ? "| " : " | ");
}
r += ToString(c);
prevElement = c;
}
break;
case EBNFKind.Sequence:
foreach (var c in ebnf.Children)
{
if (prevElement != null)
{
r += (r == "" || r.Last() == '\t' ? "" : " ");
}
if (c.Kind == EBNFKind.Choice)
{
r += "( " + ToString(c) + " )";
}
else
{
r += ToString(c);
}
prevElement = c;
}
break;
default:
r = "???";
break;
}
if (!string.IsNullOrEmpty(ebnf.FollowingComment))
{
r += " //" + ebnf.FollowingComment;
}
if (ebnf.FollowingNewline)
{
r += "\r\n\t";
}
return(r);
}
public static IEnumerable <ColorizedWord> ColorizeAntlr(EBNF ebnf)
{
var lastWasTab = false;
EBNF prevElement = null;
switch (ebnf.Kind)
{
case EBNFKind.Terminal:
yield return(Col("'" + ebnf.s.Replace("\\", "\\\\").Replace("'", "\\'").Replace("\\\"", "\"") + "'", "Terminal"));
break;
case EBNFKind.ExtendedTerminal:
yield return(Col(ebnf.s, "ExtendedTerminal"));
break;
case EBNFKind.Reference:
yield return(Col(ebnf.s, "Production"));
break;
case EBNFKind.OneOrMoreOf:
case EBNFKind.ZeroOrMoreOf:
case EBNFKind.ZeroOrOneOf:
var op = (ebnf.Kind == EBNFKind.OneOrMoreOf ? "+" : (ebnf.Kind == EBNFKind.ZeroOrMoreOf ? "*" : "?"));
if (ebnf.Children[0].Kind == EBNFKind.Choice || ebnf.Children[0].Kind == EBNFKind.Sequence)
{
yield return(Col("( ", "PlainText"));
foreach (var word in ColorizeAntlr(ebnf.Children[0]))
{
yield return(word);
}
yield return(Col(" )", "PlainText"));
yield return(Col(op, "PlainText"));
}
else
{
foreach (var word in ColorizeAntlr(ebnf.Children[0]))
{
yield return(word);
}
yield return(Col(op, "PlainText"));
}
break;
case EBNFKind.Choice:
foreach (var c in ebnf.Children)
{
if (prevElement != null)
{
yield return(Col(lastWasTab ? "| " : "| ", "PlainText"));
}
foreach (var word in ColorizeAntlr(c))
{
yield return(word); lastWasTab = (word?.Text == "\t");
}
prevElement = c;
}
break;
case EBNFKind.Sequence:
foreach (var c in ebnf.Children)
{
if (lastWasTab)
{
yield return(Col(" ", "PlainText"));
}
if (c.Kind == EBNFKind.Choice)
{
yield return(Col("( ", "PlainText"));
foreach (var word in ColorizeAntlr(c))
{
yield return(word);
}
yield return(Col(" )", "PlainText"));
lastWasTab = false;
}
else
{
foreach (var word in ColorizeAntlr(c))
{
yield return(word); lastWasTab = (word?.Text == "\t");
}
}
prevElement = c;
}
break;
default:
throw new NotSupportedException("Unrecognized EBNF");
}
if (!string.IsNullOrEmpty(ebnf.FollowingWhitespace))
{
yield return(Col(ebnf.FollowingWhitespace, "Comment"));
}
if (!string.IsNullOrEmpty(ebnf.FollowingComment))
{
yield return(Col(" //" + ebnf.FollowingComment, "Comment"));
}
if (ebnf.FollowingNewline)
{
yield return(null); yield return(Col("\t", "PlainText"));
}
}