public void pbWithKleene() // http://www.codeproject.com/csharp/spart.asp?df=100&forumid=30315&select=1812016#xx1812016xx
{
Rule opQt = new Rule();
Rule cnd_qt = new Rule();
//([aA-zZ0-9 ]+)
opQt.Parser = Ops.OneOrMore(Ops.Choice(Prims.LetterOrDigit, ' '));
//CND_QT -> [OPQT] *(
cnd_qt.Parser = Ops.Sequence(
'[',
opQt,
']',
Ops.ZeroOrMore(' '),
'('
);
Assert.IsTrue(cnd_qt.Parse(new StringScanner("[all] (")).Success);
Assert.IsTrue(cnd_qt.Parse(new StringScanner("[all](")).Success);
}
private void DefineParsingRules()
{
// someWhiteSpace ::= WS+
// optionalWhiteSpace ::= WS*
_someWhiteSpace = Ops.OneOrMore(Prims.WhiteSpace);
_optionalWhiteSpace = Ops.ZeroOrMore(Prims.WhiteSpace);
// Valid escaping formats (from http://www.icu-project.org/userguide/Collate_Customization.html )
//
// Most of the characters can be used as parts of rules.
// However, whitespace characters will be skipped over,
// and all ASCII characters that are not digits or letters
// are considered to be part of syntax. In order to use
// these characters in rules, they need to be escaped.
// Escaping can be done in several ways:
// * Single characters can be escaped using backslash \ (U+005C).
// * Strings can be escaped by putting them between single quotes 'like this'.
// * Single quote can be quoted using two single quotes ''.
// because Unicode escape sequences are allowed in LDML we need to handle those also,
// escapeSequence ::= '\' U[A-F0-9]{8} | u[A-F0-9]{4} | anyChar
_escapeSequence = Ops.Choice(new Parser[] {
Ops.Sequence('\\', Ops.Sequence('U', Prims.HexDigit, Prims.HexDigit, Prims.HexDigit,
Prims.HexDigit, Prims.HexDigit, Prims.HexDigit, Prims.HexDigit,
Prims.HexDigit)),
Ops.Sequence('\\', Ops.Sequence('u', Prims.HexDigit, Prims.HexDigit, Prims.HexDigit,
Prims.HexDigit)),
Ops.Sequence('\\', Ops.Expect("icu0002", "Invalid escape sequence.", Prims.AnyChar))
});
// singleQuoteLiteral ::= "''"
// quotedStringCharacter ::= AllChars - "'"
// quotedString ::= "'" (singleQuoteLiteral | quotedStringCharacter)+ "'"
_singleQuoteLiteral = Prims.Str("''");
_quotedStringCharacter = Prims.AnyChar - '\'';
_quotedString = Ops.Sequence('\'', Ops.OneOrMore(_singleQuoteLiteral | _quotedStringCharacter),
Ops.Expect("icu0003", "Quoted string without matching end-quote.", '\''));
// Any alphanumeric ASCII character and all characters above the ASCII range are valid data characters
// normalCharacter ::= [A-Za-z0-9] | [U+0080-U+1FFFFF]
// dataCharacter ::= normalCharacter | singleQuoteLiteral | escapeSequence
// dataString ::= (dataCharacter | quotedString) (WS? (dataCharacter | quotedString))*
_normalCharacter = Prims.LetterOrDigit | Prims.Range('\u0080', char.MaxValue);
_dataCharacter = _normalCharacter | _singleQuoteLiteral | _escapeSequence;
_dataString = new Spart.Parsers.NonTerminal.Rule(Ops.List(_dataCharacter | _quotedString, _optionalWhiteSpace));
// firstOrLast ::= 'first' | 'last'
// primarySecondaryTertiary ::= 'primary' | 'secondary' | 'tertiary'
// indirectOption ::= (primarySecondaryTertiary WS 'ignorable') | 'variable' | 'regular' | 'implicit' | 'trailing'
// indirectPosition ::= '[' WS? firstOrLast WS indirectOption WS? ']'
// According to the LDML spec, "implicit" should not be allowed in a reset element, but we're not going to check that
_firstOrLast = Ops.Choice("first", "last");
_primarySecondaryTertiary = Ops.Choice("primary", "secondary", "tertiary");
_indirectOption = Ops.Choice(Ops.Sequence(_primarySecondaryTertiary, _someWhiteSpace, "ignorable"),
"variable", "regular", "implicit", "trailing");
_indirectPosition = new Spart.Parsers.NonTerminal.Rule(Ops.Sequence('[', _optionalWhiteSpace,
Ops.Expect("icu0004", "Invalid indirect position specifier: unknown option",
Ops.Sequence(_firstOrLast, _someWhiteSpace, _indirectOption)), _optionalWhiteSpace,
Ops.Expect("icu0005", "Indirect position specifier missing closing ']'", ']')));
// top ::= '[' WS? 'top' WS? ']'
// [top] is a deprecated element in ICU and should be replaced by indirect positioning.
_top = Ops.Sequence('[', _optionalWhiteSpace, "top", _optionalWhiteSpace, ']');
// simpleElement ::= indirectPosition | dataString
_simpleElement = new Spart.Parsers.NonTerminal.Rule("simpleElement", _indirectPosition | _dataString);
// expansion ::= WS? '/' WS? simpleElement
_expansion = new Spart.Parsers.NonTerminal.Rule("extend", Ops.Sequence(_optionalWhiteSpace, '/', _optionalWhiteSpace,
Ops.Expect("icu0007", "Invalid expansion: Data missing after '/'", _simpleElement)));
// prefix ::= simpleElement WS? '|' WS?
_prefix = new Spart.Parsers.NonTerminal.Rule("context", Ops.Sequence(_simpleElement, _optionalWhiteSpace, '|', _optionalWhiteSpace));
// extendedElement ::= (prefix simpleElement expansion?) | (prefix? simpleElement expansion)
_extendedElement = Ops.Sequence(_prefix, _simpleElement, !_expansion) |
Ops.Sequence(!_prefix, _simpleElement, _expansion);
// beforeOption ::= '1' | '2' | '3'
// beforeSpecifier ::= '[' WS? 'before' WS beforeOption WS? ']'
_beforeOption = Ops.Choice('1', '2', '3');
_beforeSpecifier = Ops.Sequence('[', _optionalWhiteSpace, "before", _someWhiteSpace,
Ops.Expect("icu0010", "Invalid 'before' specifier: Invalid or missing option", _beforeOption), _optionalWhiteSpace,
Ops.Expect("icu0011", "Invalid 'before' specifier: Missing closing ']'", ']'));
// The difference operator initially caused some problems with parsing. The spart library doesn't
// handle situations where the first choice is the beginning of the second choice.
// Ex: differenceOperator = "<" | "<<" | "<<<" | "=" DOES NOT WORK!
// That will fail to parse bothe the << and <<< operators because it always thinks it should match <.
// However, differenceOperator = "<<<" | "<<" | "<" | "=" will work because it tries to match <<< first.
// I'm using this strange production with the option '<' characters because it also works and doesn't
// depend on order. It is less likely for someone to change it and unknowingly mess it up.
// differenceOperator ::= ('<' '<'? '<'?) | '='
_differenceOperator = Ops.Sequence('<', !Prims.Ch('<'), !Prims.Ch('<')) | Prims.Ch('=');
// simpleDifference ::= differenceOperator WS? simpleElement
// extendedDifference ::= differenceOperator WS? extendedElement
// difference ::= simpleDifference | extendedDifference
// NOTE: Due to the implementation of the parser, extendedDifference MUST COME BEFORE simpleDifference in the difference definition
_simpleDifference = new Spart.Parsers.NonTerminal.Rule("simpleDifference", Ops.Sequence(_differenceOperator,
_optionalWhiteSpace, _simpleElement));
_extendedDifference = new Spart.Parsers.NonTerminal.Rule("x", Ops.Sequence(_differenceOperator,
_optionalWhiteSpace, _extendedElement));
_difference = _extendedDifference | _simpleDifference;
// reset ::= '&' WS? ((beforeSpecifier? WS? simpleElement) | top)
_reset = new Spart.Parsers.NonTerminal.Rule("reset", Ops.Sequence('&', _optionalWhiteSpace,
_top | Ops.Sequence(!_beforeSpecifier, _optionalWhiteSpace, _simpleElement)));
// This option is a weird one, as it can come at any place in a rule and sets the preceding
// dataString as the variable top option in the settings element. So, it has to look at the
// data for the preceding element to know its own value, but leaves the preceding and any
// succeeding elements as if the variable top option wasn't there. Go figure.
// Also, it's really probably only valid following a simpleDifference or reset with a dataString
// and not an indirect position, but checking for all that in the grammar would be very convoluted, so
// we'll do it in the semantic action and throw. Yuck.
// optionVariableTop ::= '<' WS? '[' WS? 'variable' WS? 'top' WS? ']'
_optionVariableTop = Ops.Sequence('<', _optionalWhiteSpace, '[', _optionalWhiteSpace, "variable",
_optionalWhiteSpace, "top", _optionalWhiteSpace, ']');
// oneRule ::= reset (WS? (optionVariableTop | difference))*
_oneRule = new Spart.Parsers.NonTerminal.Rule("oneRule", Ops.Sequence(_reset, Ops.ZeroOrMore(Ops.Sequence(_optionalWhiteSpace,
_optionVariableTop | _difference))));
// Option notes:
// * The 'strength' option is specified in ICU as having valid values 1-4 and 'I'. In the LDML spec, it
// seems to indicate that valid values in ICU are 1-5, so I am accepting both and treating 'I' and '5'
// as the same. I'm also accepting 'I' and 'i', although my approach is, in general, to be case-sensitive.
// * The 'numeric' option is not mentioned on the ICU website, but it is implied as being acceptable ICU
// in the LDML spec, so I am supporting it here.
// * There are LDML options 'match-boundaries' and 'match-style' that are not in ICU, so they are not listed here.
// * The UCA spec seems to indicate that there is a 'locale' option which is not mentioned in either the
// LDML or ICU specs, so I am not supporting it here. It could be referring to the 'base' element that
// is an optional part of the 'collation' element in LDML.
// optionOnOff ::= 'on' | 'off'
// optionAlternate ::= 'alternate' WS ('non-ignorable' | 'shifted')
// optionBackwards ::= 'backwards' WS ('1' | '2')
// optionNormalization ::= 'normalization' WS optionOnOff
// optionCaseLevel ::= 'caseLevel' WS optionOnOff
// optionCaseFirst ::= 'caseFirst' WS ('off' | 'upper' | 'lower')
// optionStrength ::= 'strength' WS ('1' | '2' | '3' | '4' | 'I' | 'i' | '5')
// optionHiraganaQ ::= 'hiraganaQ' WS optionOnOff
// optionNumeric ::= 'numeric' WS optionOnOff
// characterSet ::= '[' (AnyChar - ']')* ']'
// optionSuppressContractions ::= 'suppress' WS 'contractions' WS characterSet
// optionOptimize ::= 'optimize' WS characterSet
// option ::= '[' WS? (optionAlternate | optionBackwards | optionNormalization | optionCaseLevel
// | optionCaseFirst | optionStrength | optionHiraganaQ | optionNumeric
// | optionSuppressContractions | optionOptimize) WS? ']'
_optionOnOff = Ops.Choice("on", "off");
_optionAlternate = Ops.Sequence("alternate", _someWhiteSpace, Ops.Choice("non-ignorable", "shifted"));
_optionBackwards = Ops.Sequence("backwards", _someWhiteSpace, Ops.Choice('1', '2'));
_optionNormalization = Ops.Sequence("normalization", _someWhiteSpace, _optionOnOff);
_optionCaseLevel = Ops.Sequence("caseLevel", _someWhiteSpace, _optionOnOff);
_optionCaseFirst = Ops.Sequence("caseFirst", _someWhiteSpace, Ops.Choice("off", "upper", "lower"));
_optionStrength = Ops.Sequence("strength", _someWhiteSpace, Ops.Choice('1', '2', '3', '4', 'I', 'i', '5'));
_optionHiraganaQ = Ops.Sequence("hiraganaQ", _someWhiteSpace, _optionOnOff);
_optionNumeric = Ops.Sequence("numeric", _someWhiteSpace, _optionOnOff);
_characterSet = Ops.Sequence('[', Ops.ZeroOrMore(Prims.AnyChar - ']'), ']');
_optionSuppressContractions = Ops.Sequence("suppress", _someWhiteSpace, "contractions", _someWhiteSpace,
_characterSet);
_optionOptimize = Ops.Sequence("optimize", _someWhiteSpace, _characterSet);
_option = new Spart.Parsers.NonTerminal.Rule("option", Ops.Sequence('[', _optionalWhiteSpace, _optionAlternate |
_optionBackwards | _optionNormalization | _optionCaseLevel | _optionCaseFirst |
_optionStrength | _optionHiraganaQ | _optionNumeric | _optionSuppressContractions |
_optionOptimize, _optionalWhiteSpace, ']'));
// I don't know if ICU requires all options first (it's unclear), but I am. :)
// icuRules ::= WS? (option WS?)* (oneRule WS?)* EOF
_icuRules = new Spart.Parsers.NonTerminal.Rule("icuRules", Ops.Sequence(_optionalWhiteSpace,
Ops.ZeroOrMore(Ops.Sequence(_option, _optionalWhiteSpace)),
Ops.ZeroOrMore(Ops.Sequence(_oneRule, _optionalWhiteSpace)), Ops.Expect("icu0015", "Invalid ICU rules.", Prims.End)));
if (_useDebugger)
{
_debugger = new Spart.Debug.Debugger(Console.Out);
_debugger += _option;
_debugger += _oneRule;
_debugger += _reset;
_debugger += _simpleElement;
_debugger += _simpleDifference;
_debugger += _extendedDifference;
_debugger += _dataString;
}
}
