void SetLine()
{
if (line != null)
lineSeq = line & new RepeatParser().Until(Terminals.Eol);
else
lineSeq = null;
}
void SetBlock()
{
if (start != null && end != null)
blockSeq = start & new RepeatParser().Until(end, true);
else
blockSeq = null;
}
Parser SingleDefinition(Match match, bool isTerminal)
{
var terms = match.Find("term").ToArray();
if (terms.Length == 1)
return Term(terms[0], isTerminal);
else
{
var sequence = new SequenceParser(terms.Select(r => Term(r, isTerminal)));
if (!isTerminal)
sequence.Separator = separator;
return sequence;
}
}
public EbnfGrammar(EbnfStyle style)
: base("ebnf")
{
Style = style;
DefineCommonNonTerminals = true;
GenerateSpecialSequences();
// terminals
var comment = style.HasFlag(EbnfStyle.BracketComments) ? new GroupParser("(*", "*)") : new GroupParser("/*", "*/");
var ows = -(Terminals.WhiteSpace | comment);
var rws = +(Terminals.WhiteSpace | comment);
var hex_character = ("#x" & +Terminals.HexDigit);
var character = (("\\" & Terminals.AnyChar) | hex_character | Terminals.AnyChar.Except("]")).WithName("character");
var character_range = (character & "-" & character).WithName("character range");
var character_set = ("[" & ~(Parser)"^" & +(character_range | character) & "]").WithName("character set");
var terminal_string = new StringParser { QuoteCharacters = new [] { '\"', '\'', '’' }, Name = "terminal string" };
var special_sequence = ("?" & (+Terminals.AnyChar).Until("?").WithName("name") & "?").WithName("special sequence");
var meta_identifier_terminal = Terminals.Letter & -(Terminals.LetterOrDigit | '_');
var integer = new NumberParser().WithName("integer");
// nonterminals
var definition_list = new RepeatParser(0).WithName("definition list");
var single_definition = new RepeatParser(1).WithName("single definition");
var term = new SequenceParser().WithName("term");
var primary = new AlternativeParser().WithName("primary");
var exception = new UnaryParser("exception");
var factor = new SequenceParser().WithName("factor");
var meta_identifier = new RepeatParser(1).WithName("meta identifier");
var syntax_rule = new SequenceParser().WithName("syntax rule");
var rule_equals = new AlternativeParser().WithName("equals");
Parser meta_reference = meta_identifier;
Parser grouped_sequence = ("(" & ows & definition_list & ows & ")").WithName("grouped sequence");
if (style.HasFlag(EbnfStyle.SquareBracketAsOptional))
{
primary.Add(("[" & ows & definition_list & ows & "]").WithName("optional sequence"));
}
if (!style.HasFlag(EbnfStyle.CardinalityFlags))
{
var repeated_sequence = ("{" & ows & definition_list & ows & "}").WithName("repeated sequence");
primary.Add(repeated_sequence);
}
// rules
meta_identifier.Inner = meta_identifier_terminal;
meta_identifier.Separator = +(Terminals.SingleLineWhiteSpace);
if (!style.HasFlag(EbnfStyle.CommaSeparator))
{
// w3c identifiers must be a single word
meta_identifier.Maximum = 1;
meta_reference = meta_reference.NotFollowedBy(ows & rule_equals);
}
primary.Add(grouped_sequence, meta_reference, terminal_string, special_sequence);
if (style.HasFlag(EbnfStyle.CharacterSets) && !style.HasFlag(EbnfStyle.SquareBracketAsOptional))
{
// w3c supports character sets
primary.Add(hex_character.Named("hex character"));
primary.Add(character_set);
}
if (style.HasFlag(EbnfStyle.NumericCardinality))
factor.Add(~(integer & ows & "*" & ows));
factor.Add(primary);
if (style.HasFlag(EbnfStyle.CardinalityFlags))
{
// w3c defines cardinality at the end of a factor
var flags = style.HasFlag(EbnfStyle.SquareBracketAsOptional) ? "*+" : "?*+";
factor.Add(~(ows & Terminals.Set(flags).WithName("cardinality")));
}
term.Add(factor, ~(ows & "-" & ows & exception));
exception.Inner = term;
single_definition.Inner = term;
single_definition.Separator = style.HasFlag(EbnfStyle.CommaSeparator) ? (Parser)(ows & "," & ows) : ows;
definition_list.Inner = single_definition;
definition_list.Separator = ows & "|" & ows;
rule_equals.Add(style.HasFlag(EbnfStyle.DoubleColonEquals) ? "::=" : "=", ":=");
syntax_rule.Add(meta_identifier, ows, rule_equals, ows, definition_list);
if (style.HasFlag(EbnfStyle.SemicolonTerminator))
syntax_rule.Add(ows, ";"); // iso rules are terminated by a semicolon
var syntax_rules = +syntax_rule;
syntax_rules.Separator = style.HasFlag(EbnfStyle.SemicolonTerminator) ? ows : rws;
Inner = ows & syntax_rules & ows;
AttachEvents();
}
void AttachEvents()
{
ruleName.Matched += m => {
Parser parser;
var name = m["name"].Text;
if (!parserLookup.TryGetValue(name, out parser) && !baseLookup.TryGetValue(name, out parser))
{
parser = Terminals.LetterOrDigit.Repeat();
parser.Name = name;
}
m.Tag = parser;
};
literal.Matched += m => m.Tag = new LiteralTerminal(m["value"].Text);
optionalRule.Matched += m => m.Tag = new OptionalParser((Parser)m["parser"].Tag);
repeatRule.Matched += m => m.Tag = new RepeatParser((Parser)m["parser"].Tag, 0) { Separator = sws };
list.Matched += m => {
if (m.Matches.Count > 1)
{
var parser = new SequenceParser();
foreach (var child in m.Matches)
{
if (child.Parser.Name == "ws")
parser.Items.Add(sws);
else if (child.Parser.Name == "term")
parser.Items.Add((Parser)child["parser"].Tag);
}
m.Tag = parser;
}
else
{
m.Tag = m["term"]["parser"].Tag;
}
};
listRepeat.Matched += m => {
// collapse alternatives to one alternative parser
var parser = (Parser)m["expression"]["parser"].Tag;
var alt = parser as AlternativeParser ?? new AlternativeParser(parser);
alt.Items.Insert(0, (Parser)m["list"]["parser"].Tag);
m.Tag = alt;
};
rule.Matched += m => {
var parser = (UnaryParser)m.Tag;
parser.Inner = (Parser)m["parser"].Tag;
m.Tag = parser;
};
rule.PreMatch += m => {
var name = m["ruleName"]["name"].Text;
Parser parser;
if (name == startParserName)
parser = new Grammar(name);
else
parser = new UnaryParser(name);
m.Tag = parser;
parserLookup[parser.Name] = parser;
};
}
protected SequenceParser(SequenceParser other, ParserCloneArgs chain)
: base(other, chain)
{
Separator = chain.Clone(other.Separator);
}