/// <summary>
/// Creates a rule that matches a pair of rules, consuming all nested pairs within
/// as well.
/// </summary>
/// <param name="begin"></param>
/// <param name="end"></param>
/// <returns></returns>
public Rule Nested(Rule begin, Rule end)
{
RecursiveRule recursive = new RecursiveRule(() => { return(Nested(begin, end)); });
return(begin + NoFailSeq(Star(recursive | (Not(end) + Not(begin) + Anything())) + end));
}
/// <summary>
/// Constructor: initializes the public rule fields.
/// </summary>
public CppBaseGrammar()
{
#region numbers
digit = CharRange('0', '9');
octal_digit = CharRange('0', '7');
nonzero_digit = CharRange('1', '9');
hex_digit = digit | CharRange('a', 'f') | CharRange('A', 'F');
sign = CharSet("+-");
#endregion numbers
#region identifiers
lower_case_letter = CharRange('a', 'z');
upper_case_letter = CharRange('A', 'Z');
letter = lower_case_letter | upper_case_letter;
ident_first_char = CharSet("_") | letter;
ident_next_char = ident_first_char | digit;
identifier_extension = CharSeq("::") + Recursive(() => Opt(identifier));
identifier = Leaf(ident_first_char + Star(ident_next_char) + Star(identifier_extension));
#endregion
#region whitespace
tab = CharSeq("\t");
space = CharSeq(" ");
simple_ws = tab | space;
eol = Opt(CharSeq("\r")) + CharSeq("\n");
ext_line = CharSeq("\\") + Star(simple_ws) + eol;
multiline_ws = simple_ws | eol;
until_eol = Star(ext_line | AnythingBut(eol | EndOfInput()));
line_comment_content = until_eol;
line_comment = CharSeq("//") + NoFailSeq(line_comment_content + (eol | EndOfInput()));
full_comment_content = Until(CharSeq("*/"));
full_comment = CharSeq("/*") + NoFailSeq(full_comment_content + CharSeq("*/"));
comment = line_comment | full_comment;
ws = Eat(multiline_ws | comment);
#endregion
#region keyword rules
COND = Word("?");
DOT = Word(".");
COLON = Word(":");
AMP = Word("&");
PLUS = Word("+");
MINUS = Word("-");
STAR = Word("*");
SLASH = Word("/");
MOD = Word("%");
NOT = Word("!");
TILDE = Word("~");
CARET = Word("^");
PIPE = Word("|");
EQ = Word("=");
COMMA = Word(",");
SIZEOF = Word("sizeof");
PTR_OP = Word("->");
INC_OP = Word("++");
DEC_OP = Word("--");
LEFT_OP = Word("<<");
RIGHT_OP = Word(">>");
LT_OP = Word("<");
GT_OP = Word(">");
LE_OP = Word("<=");
GE_OP = Word(">=");
EQ_OP = Word("==");
NE_OP = Word("!=");
AND_OP = Word("&&");
OR_OP = Word("||");
MUL_ASSIGN = Word("*=");
DIV_ASSIGN = Word("/=");
MOD_ASSIGN = Word("%=");
ADD_ASSIGN = Word("+=");
SUB_ASSIGN = Word("-=");
LEFT_ASSIGN = Word("<<=");
RIGHT_ASSIGN = Word(">>=");
AND_ASSIGN = Word("&=");
XOR_ASSIGN = Word("^=");
OR_ASSIGN = Word("|=");
TYPEDEF = Word("typedef");
EXTERN = Word("extern");
STATIC = Word("static");
AUTO = Word("auto");
REGISTER = Word("register");
CHAR = Word("char");
SHORT = Word("short");
INT = Word("int");
LONG = Word("long");
SIGNED = Word("signed");
UNSIGNED = Word("unsigned");
FLOAT = Word("float");
DOUBLE = Word("double");
CONST = Word("const");
VOLATILE = Word("volatile");
VOID = Word("void");
STRUCT = Word("struct");
UNION = Word("union");
ENUM = Word("enum");
ELLIPSIS = Word("...");
CASE = Word("case");
DEFAULT = Word("default");
IF = Word("if");
ELSE = Word("else");
SWITCH = Word("switch");
WHILE = Word("while");
DO = Word("do");
FOR = Word("for");
GOTO = Word("goto");
CONTINUE = Word("continue");
BREAK = Word("break");
RETURN = Word("return");
CLASS = Word("class");
TYPENAME = Word("typename");
TYPEID = Word("typeid");
TEMPLATE = Word("template");
PUBLIC = Word("public");
PROTECTED = Word("protected");
PRIVATE = Word("private");
VIRTUAL = Word("virtual");
OPERATOR = Word("operator");
USING = Word("using");
#endregion
#region literals
dot
= CharSeq(".");
dbl_quote
= CharSeq("\"");
quote
= CharSeq("\'");
simple_escape
= CharSeq("\\") + CharSet("abfnrtv'\"?\\");
octal_escape
= CharSeq("\\") + octal_digit + Opt(octal_digit + Opt(octal_digit));
hex_escape
= CharSeq("\\x") + Star(hex_digit);
escape_sequence
= simple_escape
| octal_escape
| hex_escape;
c_char
= escape_sequence | Not(quote) + Anything();
s_char
= escape_sequence | Not(dbl_quote) + Anything();
long_suffix
= CharSet("Ll");
unsigned_suffix
= CharSet("Uu");
digit_sequence
= Plus(digit);
exponent
= Opt(sign) + digit_sequence;
exponent_prefix
= CharSet("Ee");
exponent_part
= exponent_prefix + exponent;
float_suffix
= CharSet("LlFf");
simple_float
= CharSeq(".") + digit_sequence
| digit_sequence + dot + Opt(digit_sequence);
exponential_float
= digit_sequence + exponent_part
| simple_float + exponent_part;
unsigned_float
= simple_float
| exponential_float;
hex_prefix
= CharSeq("0X") | CharSeq("0x");
hex_literal
= hex_prefix + Plus(hex_digit);
octal_literal
= CharSeq("0") + Star(octal_digit);
decimal_literal
= nonzero_digit + Star(digit);
unsigned_literal
= hex_literal
| octal_literal
| decimal_literal;
integer_suffix
= unsigned_suffix + Opt(long_suffix) + Opt(long_suffix)
| long_suffix + Opt(long_suffix) + Opt(unsigned_suffix);
int_literal
= unsigned_literal + Not(dot) + Opt(integer_suffix);
float_literal
= unsigned_float + Opt(float_suffix);
char_literal
= Opt(CharSeq("L")) + quote + Star(c_char) + quote;
string_literal
= Opt(CharSeq("L")) + dbl_quote + Star(s_char) + dbl_quote;
boolean_literal
= Word("true") | Word("false");
literal
= (int_literal
| char_literal
| float_literal
| string_literal
| boolean_literal)
+ NoFail(Not(ident_next_char)) + ws;
#endregion
#region pre-processor directives
pragma = Word("#") + Word("pragma") + until_eol;
included_file = string_literal | CharSeq("<") + Star(Not(CharSeq(">")) + Anything()) + CharSeq(">");
include = Word("#") + Word("include") + included_file;
ifdef_macro = Word("#") + Word("if") + until_eol + eol;
endif_macro = Word("#") + Word("endif") + until_eol + eol;
elif_macro = Word("#") + Word("elif") + until_eol + eol;
else_macro = Word("#") + Word("else") + until_eol + eol;
#endregion
#region symbols
semicolon = CharSeq(";");
eos = Word(";");
#endregion
#region operators
operator_char = CharSet("=<>!-+*/[]%&|^~?");
operator_prefix = Leaf(Star(ident_first_char + Star(ident_next_char) + CharSeq("::")));
operator_seq = Leaf(operator_char + Opt(operator_char));
@operator = operator_prefix + OPERATOR + operator_seq;
#endregion
InitializeRules <CppBaseGrammar>();
}
public LeafRule(Rule x)
{
AddRule(x);
}
/// <summary>
/// Creates a rule that matches the rule R multiple times, delimited by commas.
/// </summary>
/// <param name="r"></param>
/// <returns></returns>
public Rule CommaList(Rule r)
{
return(r + Star(COMMA + r));
}
public NoFailRule(Rule x)
{
AddRule(x);
}
public SkipRule(Rule x)
{
AddRule(x);
}
public OptRule(Rule x)
{
AddRule(x);
}
protected override bool InternalMatch(ParserState p)
{
Rule r = func();
return(r.Match(p));
}
public NotRule(Rule x)
{
AddRule(x);
}
public PlusRule(Rule x)
{
AddRule(x);
}
public StarRule(Rule x)
{
AddRule(x);
}
public SeqRule(Rule a, Rule b)
{
AddRule(a);
AddRule(b);
}
public ChoiceRule(Rule a, Rule b)
{
AddRule(a);
AddRule(b);
}
