private static EbnfNode 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 in {string.Join(", ", tokens)}");
}
tokens.RemoveFirst();
GobbleUpComments(tokens, p);
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 EbnfNode {
Kind = EbnfKind.Terminal, Text = t
};
if (t.StartsWith("<") && t.EndsWith(">"))
{
p.Kind = EbnfKind.ExtendedTerminal;
p.Text = t.Substring(1, t.Length - 2);
if (p.Text.Contains("?"))
{
throw new Exception("A special-terminal may not contain a question-mark '?'");
}
if (p.Text == "")
{
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, p);
return(p);
}
else
{
var t = tokens.First.Value; tokens.RemoveFirst();
var p = new EbnfNode {
Kind = EbnfKind.Reference, Text = t
};
GobbleUpComments(tokens, p);
return(p);
}
}
private static void GobbleUpComments(LinkedList <string> tokens, EbnfNode node)
{
string followingWhitespace = node.FollowingWhitespace;
string followingComment = node.FollowingComment;
bool followingNewline = node.FollowingNewline;
GobbleUpComments(tokens, ref followingWhitespace, ref followingComment, ref followingNewline);
node.FollowingWhitespace = followingWhitespace;
node.FollowingComment = followingComment;
node.FollowingNewline = followingNewline;
}
private static EbnfNode 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 EbnfNode {
Kind = EbnfKind.OneOrMoreOf, Children = new[] { p }.ToList()
};
GobbleUpComments(tokens, p);
}
else if (tokens.First.Value == "*")
{
tokens.RemoveFirst();
p = new EbnfNode {
Kind = EbnfKind.ZeroOrMoreOf, Children = new[] { p }.ToList()
};
GobbleUpComments(tokens, p);
}
else if (tokens.First.Value == "?")
{
tokens.RemoveFirst();
p = new EbnfNode {
Kind = EbnfKind.ZeroOrOneOf, Children = new[] { p }.ToList()
};
GobbleUpComments(tokens, p);
}
else
{
break;
}
}
return(p);
}
public static string ToString(EbnfNode node)
{
var r = "";
EbnfNode prevElement = null;
switch (node.Kind)
{
case EbnfKind.Terminal:
r = $"'{node.Text.Replace("\\", "\\\\").Replace("'", "\\'")}'";
break;
case EbnfKind.ExtendedTerminal:
r = $"'<{node.Text.Replace("\\", "\\\\").Replace("'", "\\'")}>'";
break;
case EbnfKind.Reference:
r = node.Text;
break;
case EbnfKind.OneOrMoreOf:
case EbnfKind.ZeroOrMoreOf:
case EbnfKind.ZeroOrOneOf:
var op = (node.Kind == EbnfKind.OneOrMoreOf ? "+" : (node.Kind == EbnfKind.ZeroOrMoreOf ? "*" : "?"));
if (node.Children[0].Kind == EbnfKind.Choice || node.Children[0].Kind == EbnfKind.Sequence)
{
r = $"( {ToString(node.Children[0])} ){op}";
}
else
{
r = $"{ToString(node.Children[0])}{op}";
}
break;
case EbnfKind.Choice:
foreach (var c in node.Children)
{
if (prevElement != null)
{
r += (r.Last() == '\t' ? "| " : " | ");
}
r += ToString(c);
prevElement = c;
}
break;
case EbnfKind.Sequence:
foreach (var c in node.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(node.FollowingComment))
{
r += " //" + node.FollowingComment;
}
if (node.FollowingNewline)
{
r += "\r\n\t";
}
return(r);
}
public static IEnumerable <ColorizedWord> ColorizeAntlr(EbnfNode node)
{
var lastWasTab = false;
EbnfNode prevElement = null;
switch (node.Kind)
{
case EbnfKind.Terminal:
yield return(Col("'" + node.Text.Replace("\\", "\\\\").Replace("'", "\\'").Replace("\\\"", "\"") + "'", "Terminal"));
break;
case EbnfKind.ExtendedTerminal:
yield return(Col(node.Text, "ExtendedTerminal"));
break;
case EbnfKind.Reference:
yield return(Col(node.Text, "Production"));
break;
case EbnfKind.OneOrMoreOf:
case EbnfKind.ZeroOrMoreOf:
case EbnfKind.ZeroOrOneOf:
var op = (node.Kind == EbnfKind.OneOrMoreOf ? "+" : (node.Kind == EbnfKind.ZeroOrMoreOf ? "*" : "?"));
if (node.Children[0].Kind == EbnfKind.Choice || node.Children[0].Kind == EbnfKind.Sequence)
{
yield return(Col("( ", "PlainText"));
foreach (var word in ColorizeAntlr(node.Children[0]))
{
yield return(word);
}
yield return(Col(" )", "PlainText"));
yield return(Col(op, "PlainText"));
}
else
{
foreach (var word in ColorizeAntlr(node.Children[0]))
{
yield return(word);
}
yield return(Col(op, "PlainText"));
}
break;
case EbnfKind.Choice:
foreach (var c in node.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 node.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(node.FollowingWhitespace))
{
yield return(Col(node.FollowingWhitespace, "Comment"));
}
if (!string.IsNullOrEmpty(node.FollowingComment))
{
yield return(Col(" //" + node.FollowingComment, "Comment"));
}
if (node.FollowingNewline)
{
yield return(null); yield return(Col("\t", "PlainText"));
}
}