bool _HasExpression(EbnfExpression search, EbnfExpression match)
{
if (Equals(search, match))
{
return(true);
}
var b = search as EbnfBinaryExpression;
if (null != b)
{
if (_HasExpression(b.Left, match) || _HasExpression(b.Right, match))
{
return(true);
}
}
var u = search as EbnfUnaryExpression;
if (null != u)
{
if (_HasExpression(u.Expression, match))
{
return(true);
}
}
return(false);
}
void _VisitFetchTerminals(EbnfExpression expr, HashSet <EbnfExpression> terms)
{
var l = expr as EbnfLiteralExpression;
if (null != l)
{
terms.Add(l);
return;
}
var r = expr as EbnfRegexExpression;
if (null != r)
{
terms.Add(r);
return;
}
var u = expr as EbnfUnaryExpression;
if (null != u)
{
_VisitFetchTerminals(u.Expression, terms);
return;
}
var b = expr as EbnfBinaryExpression;
if (null != b)
{
_VisitFetchTerminals(b.Left, terms);
_VisitFetchTerminals(b.Right, terms);
return;
}
}
public string GetIdForExpression(EbnfExpression expression)
{
foreach (var prod in Productions)
{
if (Equals(prod.Value.Expression, expression))
{
return(prod.Key);
}
}
return("");
}
public string GetContainingIdForExpression(EbnfExpression expression)
{
foreach (var prod in Productions)
{
if (_HasExpression(prod.Value.Expression, expression))
{
return(prod.Key);
}
}
return("");
}
public EbnfOrExpression(EbnfExpression left, params EbnfExpression[] right)
{
if (null == right)
{
right = new EbnfExpression[] { null }
}
;
Left = left;
for (var i = 0; i < right.Length; ++i)
{
if (Right == null)
{
Right = right[i];
}
else
{
Right = new EbnfOrExpression(Right, right[i]);
}
}
}
public EbnfOptionalExpression(EbnfExpression expression)
{
Expression = expression;
}
public EbnfProduction(EbnfExpression expression)
{
Expression = expression;
}
static EbnfExpression _ParseExpression(EbnfDocument doc, ParseContext pc)
{
EbnfExpression current = null;
EbnfExpression e;
long position;
int line;
int column;
pc.TrySkipCCommentsAndWhiteSpace();
position = pc.Position; line = pc.Line; column = pc.Column;
while (-1 != pc.Current && ']' != pc.Current && ')' != pc.Current && '}' != pc.Current && ';' != pc.Current)
{
pc.TrySkipCCommentsAndWhiteSpace();
position = pc.Position; line = pc.Line; column = pc.Column;
switch (pc.Current)
{
case '|':
pc.Advance();
current = new EbnfOrExpression(current, _ParseExpression(doc, pc));
current.SetLocationInfo(line, column, position);
break;
case '(':
pc.Advance();
e = _ParseExpression(doc, pc);
current.SetLocationInfo(line, column, position);
pc.Expecting(')');
pc.Advance();
e.SetLocationInfo(line, column, position);
if (null == current)
{
current = e;
}
else
{
current = new EbnfConcatExpression(current, e);
}
break;
case '[':
pc.Advance();
e = new EbnfOptionalExpression(_ParseExpression(doc, pc));
e.SetLocationInfo(line, column, position);
pc.TrySkipCCommentsAndWhiteSpace();
pc.Expecting(']');
pc.Advance();
if (null == current)
{
current = e;
}
else
{
current = new EbnfConcatExpression(current, e);
}
break;
case '{':
pc.Advance();
e = new EbnfRepeatExpression(_ParseExpression(doc, pc));
e.SetLocationInfo(line, column, position);
pc.TrySkipCCommentsAndWhiteSpace();
pc.Expecting('}');
pc.Advance();
if ('+' == pc.Current)
{
pc.Advance();
((EbnfRepeatExpression)e).IsOptional = false;
}
if (null == current)
{
current = e;
}
else
{
current = new EbnfConcatExpression(current, e);
}
break;
case '\"':
e = new EbnfLiteralExpression(pc.ParseJsonString());
if (null == current)
{
current = e;
}
else
{
current = new EbnfConcatExpression(current, e);
}
e.SetLocationInfo(line, column, position);
break;
case '\'':
pc.Advance();
pc.ClearCapture();
pc.TryReadUntil('\'', '\\', false);
pc.Expecting('\'');
pc.Advance();
e = new EbnfRegexExpression(pc.Capture);
if (null == current)
{
current = e;
}
else
{
current = new EbnfConcatExpression(current, e);
}
e.SetLocationInfo(line, column, position);
break;
case ';':
case ']':
case ')':
case '}':
return(current);
default:
e = new EbnfRefExpression(_ParseIdentifier(pc));
if (null == current)
{
current = e;
}
else
{
current = new EbnfConcatExpression(current, e);
}
e.SetLocationInfo(line, column, position);
break;
}
}
pc.TrySkipCCommentsAndWhiteSpace();
return(current);
}
void _ValidateExpression(EbnfExpression expr, IDictionary <string, int> refCounts, IList <EbnfMessage> messages)
{
var l = expr as EbnfLiteralExpression;
if (null != l)
{
var i = GetIdForExpression(l);
// don't count itself. only things just like itself
if (!string.IsNullOrEmpty(i) && !ReferenceEquals(Productions[i].Expression, l))
{
refCounts[i] += 1;
}
}
var rx = expr as EbnfRegexExpression;
if (null != rx)
{
try
{
CharFA.Parse(rx.Value);
}
catch (ExpectingException)
{
messages.Add(
new EbnfMessage(
EbnfErrorLevel.Error, 12,
"Invalid regular expression",
expr.Line, expr.Column, expr.Position));
}
var i = GetIdForExpression(rx);
if (!string.IsNullOrEmpty(i) && !ReferenceEquals(Productions[i].Expression, l))
{
refCounts[i] += 1;
}
}
var r = expr as EbnfRefExpression;
if (null != r)
{
int rc;
if (null == r.Symbol)
{
messages.Add(
new EbnfMessage(
EbnfErrorLevel.Error, 4,
"Null reference expression",
expr.Line, expr.Column, expr.Position));
return;
}
if (!refCounts.TryGetValue(r.Symbol, out rc))
{
messages.Add(
new EbnfMessage(
EbnfErrorLevel.Error, 1,
string.Concat(
"Reference to undefined symbol \"",
r.Symbol,
"\""),
expr.Line, expr.Column, expr.Position));
return;
}
refCounts[r.Symbol] = rc + 1;
return;
}
var b = expr as EbnfBinaryExpression;
if (null != b)
{
if (null == b.Left && null == b.Right)
{
messages.Add(
new EbnfMessage(
EbnfErrorLevel.Warning, 3,
"Nil expression",
expr.Line, expr.Column, expr.Position));
return;
}
_ValidateExpression(b.Left, refCounts, messages);
_ValidateExpression(b.Right, refCounts, messages);
return;
}
var u = expr as EbnfUnaryExpression;
if (null != u)
{
if (null == u.Expression)
{
messages.Add(
new EbnfMessage(
EbnfErrorLevel.Warning, 3,
"Nil expression",
expr.Line, expr.Column, expr.Position));
return;
}
_ValidateExpression(u.Expression, refCounts, messages);
}
}
static IList <EbnfMessage> _TryParseExpressions(ParseNode parent, int firstChildIndex, out EbnfExpression result)
{
var msgs = new List <EbnfMessage>();
result = null;
var or = new List <IList <EbnfExpression> >();
var seq = new List <EbnfExpression>();
for (int ic = parent.Children.Count, i = firstChildIndex; i < ic; ++i)
{
EbnfExpression expr;
var pn = parent.Children[i];
switch (pn.SymbolId)
{
case EbnfParser.expression:
msgs.AddRange(_TryParseExpression(parent.Children[i], out expr));
seq.Add(expr);
break;
case EbnfParser.or:
or.Add(seq);
seq = new List <EbnfExpression>();
break;
default:
break;
}
}
or.Add(seq);
result = null;
switch (or.Count)
{
case 0:
result = null;
break;
case 1:
result = _SeqToExpression(or[0]);
break;
default:
var state = 0;
var oe = new EbnfOrExpression();
for (int ic = or.Count, i = 0; i < ic; ++i)
{
switch (state)
{
case 0:
oe.Left = _SeqToExpression(or[i]);
state = 1;
break;
case 1:
oe.Right = _SeqToExpression(or[i]);
state = 2;
break;
case 2:
oe = new EbnfOrExpression(oe, null);
oe.Right = _SeqToExpression(or[i]);
break;
}
}
//if (0 == state || 2 == state)
// result = oe.Left;
//else
result = oe;
break;
}
return(msgs);
}
static IList <EbnfMessage> _TryParseExpression(ParseNode pn, out EbnfExpression result)
{
result = null;
Debug.Assert(EbnfParser.expression == pn.SymbolId, "Not positioned on expression");
var msgs = new List <EbnfMessage>();
if (1 == pn.Children.Count)
{
var c = pn.Children[0];
if (EbnfParser.symbol == c.SymbolId)
{
ParseContext pc;
var cc = c.Children[0];
//TODO: parse the regular expressions and literals to make sure they're valid.
switch (cc.SymbolId)
{
case EbnfParser.identifier:
result = new EbnfRefExpression(cc.Value);
return(msgs);
case EbnfParser.regex:
pc = ParseContext.Create(cc.Value);
pc.EnsureStarted();
pc.Advance();
pc.TryReadUntil('\'', '\\', false);
result = new EbnfRegexExpression(pc.GetCapture());
return(msgs);
case EbnfParser.literal:
pc = ParseContext.Create(cc.Value);
pc.EnsureStarted();
pc.Advance();
pc.TryReadUntil('\"', '\\', false);
result = new EbnfLiteralExpression(pc.GetCapture());
return(msgs);
case EbnfParser.lbrace:
msgs.AddRange(_TryParseExpressions(c, 1, out result));
result = new EbnfRepeatExpression(result);
if (EbnfParser.rbracePlus == c.Children[c.Children.Count - 1].SymbolId)
{
((EbnfRepeatExpression)result).IsOptional = false;
}
return(msgs);
case EbnfParser.lbracket:
msgs.AddRange(_TryParseExpressions(c, 1, out result));
result = new EbnfOptionalExpression(result);
return(msgs);
case EbnfParser.lparen:
msgs.AddRange(_TryParseExpressions(c, 1, out result));
return(msgs);
default:
break;
}
}
}
return(msgs);
}
public EbnfRepeatExpression(EbnfExpression expression)
{
Expression = expression;
}