grammar.g

}
	: "if" LPAREN test = bool_operator [tagset, vars] COMMA 
							p_true = bool_operator [tagset, vars] 
							(COMMA p_false = bool_operator [tagset, vars])? 
	RPAREN {
		if (p_false) {
			op.reset(new Conditional<Bool>(test, p_true, p_false));
		}
		else {
			op.reset(new Conditional<Bool>(test, p_true));
		}
	}
	| Q_MARK 
			p_true = bool_operator [tagset, vars]
		Q_MARK 
			test = bool_operator [tagset, vars] {
			op.reset(new Conditional<Bool>(test, p_true));
		}
;

// ----------------------------------------------------------------------------
// Equal operator
equal_operator
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Function<Bool> > op]
{
	boost::shared_ptr<Function<TSet> > t1, t2;
	boost::shared_ptr<Function<Bool> > b1, b2;
	boost::shared_ptr<Function<StrSet>  > s1, s2;
	boost::shared_ptr<Function<Position> > p1, p2;
}
	: "equal" LPAREN
	(
		(position_operator [tagset, vars]) =>
		(
			p1 = position_operator [tagset, vars] COMMA 
			p2 = position_operator [tagset, vars] {
				op.reset(new Equals<Position>(p1, p2));
			}
		)
	|
		(symset_operator [tagset, vars]) =>
		(
			t1 = symset_operator [tagset, vars] COMMA  
			t2 = symset_operator [tagset, vars] {
				op.reset(new Equals<TSet>(t1, t2));
			}
		)
	|
		(strset_operator [tagset, vars]) =>
		(
			s1 = strset_operator [tagset, vars] COMMA  
			s2 = strset_operator [tagset, vars] {
				op.reset(new Equals<StrSet>(s1, s2));
			}
		)
	|
		(
			b1 = bool_operator [tagset, vars] COMMA
			b2 = bool_operator [tagset, vars] {
				op.reset(new Equals<Bool>(b1, b2));
			}
		)
	)
	RPAREN
;

// ----------------------------------------------------------------------------
// In operator
in_operator
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Function<Bool> > op]
{
	boost::shared_ptr<Function<TSet> > t1, t2;
	boost::shared_ptr<Function<StrSet> > s1, s2;
}
:
	"in" LPAREN
	(
		(symset_operator [tagset, vars]) =>
		(
			t1 = symset_operator [tagset, vars] COMMA 
			t2 = symset_operator [tagset, vars] {
				op.reset(new IsSubsetOf<TSet>(t1, t2));
			}
		)
		|
		(
			s1 = strset_operator [tagset, vars] COMMA
			s2 = strset_operator [tagset, vars] {
				op.reset(new IsSubsetOf<StrSet>(s1, s2));
			}
		)
	)
	RPAREN
;
// ----------------------------------------------------------------------------
// Inter operator
inter_operator
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Function<Bool> > op]
{
	boost::shared_ptr<Function<TSet> > t1, t2;
	boost::shared_ptr<Function<StrSet> > s1, s2;
}
	:
	"inter" LPAREN
	(
		(symset_operator [tagset, vars]) =>
		(
			t1 = symset_operator [tagset, vars] COMMA  
			t2 = symset_operator [tagset, vars]  {
				op.reset(new Intersects<TSet>(t1, t2));
			}
		)
	|
		(
			s1 = strset_operator [tagset, vars] COMMA  
			s2 = strset_operator [tagset, vars]  {
				op.reset(new Intersects<StrSet>(s1, s2));
			}
		)
	)
	RPAREN
;

// ----------------------------------------------------------------------------
// Iterations:
bool_iteration
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Function<Bool> > ret]
{
	int min_match = 0;
	boost::shared_ptr<Function<Bool> > expr;
	boost::shared_ptr<Function<Position> > lpos, rpos;
	boost::shared_ptr<VariableAccessor<Position> > pacc;
}
	: "only" LPAREN 
			lpos = position_operator [tagset, vars] COMMA 
			rpos = position_operator [tagset, vars] COMMA
			pacc = position_variable_acc [vars]     COMMA
			expr = bool_operator     [tagset, vars]
		RPAREN {
			ret.reset(new Only(lpos, rpos, *pacc, expr));
		}

	| "atleast" LPAREN
			lpos = position_operator [tagset, vars] COMMA 
			rpos = position_operator [tagset, vars] COMMA
			pacc = position_variable_acc [vars]     COMMA
			expr = bool_operator     [tagset, vars] COMMA
			min_match = number
		RPAREN {
			ret.reset(new AtLeast(lpos, rpos, *pacc, expr, min_match));
		}
	| "llook" LPAREN //note inverted rpos/lpos order
			rpos = position_operator [tagset, vars] COMMA 
			lpos = position_operator [tagset, vars] COMMA
			pacc = position_variable_acc [vars]     COMMA
			expr = bool_operator     [tagset, vars] 
		RPAREN {
			ret.reset(new LeftLook(lpos, rpos, *pacc, expr));
		}
	| "rlook" LPAREN
			lpos = position_operator [tagset, vars] COMMA 
			rpos = position_operator [tagset, vars] COMMA
			pacc = position_variable_acc [vars]     COMMA
			expr = bool_operator     [tagset, vars] 
		RPAREN {
			ret.reset(new RightLook(lpos, rpos, *pacc, expr));
		}
;

// ----------------------------------------------------------------------------
// Agreement operator: agr, agrpp, wagr
bool_agreement
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Function<Bool> > ret]
{
	boost::shared_ptr<Function<TSet> > expr;
	boost::shared_ptr<Function<Position> > lpos, rpos;
}
	: "agr"   LPAREN 
			lpos = position_operator [tagset, vars] COMMA 
			rpos = position_operator [tagset, vars] COMMA 
			expr = symset_operator [tagset, vars]
		RPAREN {
			ret.reset(new StrongAgreement(lpos, rpos, expr, tagset));
		}
	| "agrpp" LPAREN
			lpos = position_operator [tagset, vars] COMMA 
			rpos = position_operator [tagset, vars] COMMA 
			expr = symset_operator [tagset, vars]
		RPAREN {
			ret.reset(new PointAgreement(lpos, rpos, expr, tagset));
		}
	| "wagr"  LPAREN
			lpos = position_operator [tagset, vars] COMMA 
			rpos = position_operator [tagset, vars] COMMA 
			expr = symset_operator [tagset, vars]
		RPAREN {
			ret.reset(new WeakAgreement(lpos, rpos, expr, tagset));
		}
;

// ----------------------------------------------------------------------------
// Parse operator on L1 level
bool_phrase
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Function<Bool> > ret]
	: ret = bool_phrase_annotation [tagset, vars]
	| ret = bool_phrase_iteration  [tagset, vars]
;

// ----------------------------------------------------------------------------
// Annotation operator: phrase, phrase_beg, phrase_end, phrase_whole, phrase_pp
bool_phrase_annotation
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Function<Bool> > ret]
{
	boost::shared_ptr<Function<Position> > lpos, rpos;
}
	: "phrase" LPAREN 
			lpos = position_operator [tagset, vars] COMMA n1: STRING 
		RPAREN {
			// TODO
		}
	| "phrase_beg" LPAREN 
			lpos = position_operator [tagset, vars] COMMA n2: STRING 
		RPAREN {
			// TODO
		}
	| "phrase_end" LPAREN 
			lpos = position_operator [tagset, vars] COMMA n3: STRING 
		RPAREN {
			// TODO
		}
	| "phrase_whole" LPAREN 
			lpos = position_operator [tagset, vars] COMMA 
			rpos = position_operator [tagset, vars] COMMA n4: STRING 
		RPAREN {
			// TODO
		}
	| "phrase_pp" LPAREN 
			lpos = position_operator [tagset, vars] COMMA 
			rpos = position_operator [tagset, vars] COMMA n5: STRING 
		RPAREN {
			// TODO
		}
;

// ----------------------------------------------------------------------------
// Phrase iteration operator: lphrase, rphrase
bool_phrase_iteration
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Function<Bool> > ret]
{
	boost::shared_ptr<Function<Position> > position;
	boost::shared_ptr<VarGetter<Position> > var_position;
}
	: "lphrase" LPAREN
			position     = position_operator [tagset, vars] COMMA
			var_position = position_variable [vars]         COMMA
			n1: STRING
		RPAREN {
			// TODO
		}
	| "rphrase" LPAREN 
			position     = position_operator [tagset, vars] COMMA
			var_position = position_variable [vars]         COMMA
			n2: STRING
		RPAREN {
			// TODO
		}
;


// ----------------------------------------------------------------------------
// Setvar operator
// Returns boost::shared_ptr<Function<Bool> >
// ----------------------------------------------------------------------------
setvar_operator 
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Function<Bool> > ret]
	: "setvar" LPAREN
	  (
		  ret = position_setvar [tagset, vars]
		| ret = bool_setvar     [tagset, vars]
		| ret = strset_setvar   [tagset, vars]
		| ret = symset_setvar   [tagset, vars]
	  )
	  RPAREN
;

// ----------------------------------------------------------------------------
// Setvar for position
position_setvar 
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Function<Bool> > op]
{
	boost::shared_ptr<Function<Position> > ret_op;
	boost::shared_ptr<VariableAccessor<Position> > ret_acc;
}
	:	ret_acc = position_variable_acc [vars]
		COMMA
		ret_op  = position_operator [tagset, vars] {
			op.reset(new VarSetter<Position>(*ret_acc, ret_op));
		}
;

// ----------------------------------------------------------------------------
// Setvar for bool
bool_setvar 
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Function<Bool> > op]
{
	boost::shared_ptr<Function<Bool> > ret_op;
	boost::shared_ptr<VariableAccessor<Bool> > ret_acc;
}
	:	ret_acc = bool_variable_acc [vars]
		COMMA
		ret_op  = bool_operator [tagset, vars] {
			op.reset(new VarSetter<Bool>(*ret_acc, ret_op));
		}
;

// ----------------------------------------------------------------------------
// Setvar for strset
strset_setvar
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Function<Bool> > op]
{
	boost::shared_ptr<Function<StrSet> > ret_op;
	boost::shared_ptr<VariableAccessor<StrSet> > ret_acc;
}
	: ret_acc = strset_variable_acc [vars]
	  COMMA 
		ret_op  = strset_operator [tagset, vars] {
			op.reset(new VarSetter<StrSet>(*ret_acc, ret_op));
		}
;

// ----------------------------------------------------------------------------
// Setvar for symset
symset_setvar 
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Function<Bool> > op]
{
	boost::shared_ptr<Function<TSet> > ret_op;
	boost::shared_ptr<VariableAccessor<TSet> > ret_acc;
}
	: ret_acc = symset_variable_acc [vars]
	  COMMA 
	  ret_op  = symset_operator [tagset, vars] {
			op.reset(new VarSetter<TSet>(*ret_acc, ret_op));
		}
;

// ----------------------------------------------------------------------------
// ----------------------------------------------------------------------------
// Tagging actions and rules:
// ----------------------------------------------------------------------------

// ----------------------------------------------------------------------------
// Single action such as select, delete, relabel or unify
action
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<TagAction> act]
	: act = action_select  [tagset, vars]
	| act = action_delete  [tagset, vars]
	| act = action_relabel [tagset, vars]
	//
	| act = action_unify [tagset, vars]
	//
;

// Action sequence - the actions are separated with commas:
// 	select(...), select(...), delete(...)
action_sequence
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<std::vector<boost::shared_ptr<TagAction> > > v_act]
{
	boost::shared_ptr<TagAction> act;
	v_act.reset(new std::vector<boost::shared_ptr<TagAction> >);
}
	: act = action[tagset, vars] {
		v_act->push_back(act);
	} 
	(
		COMMA act = action[tagset, vars] {
			v_act->push_back(act);
		}
	)*
;

// ----------------------------------------------------------------------------
// Single rule:
// 	rule(NAME, ACTIONS) or rule(NAME, COND, ACTIONS)
rule
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<TagRule> rle]
{
	boost::shared_ptr<Function<Bool> > condition;
	boost::shared_ptr<std::vector<boost::shared_ptr<TagAction> > > actions;
}
	: "rule" LPAREN name: STRING COMMA 
				(condition = bool_operator [tagset, vars] COMMA)?
				actions = action_sequence [tagset, vars] 
		RPAREN {
			if (condition) {
				rle.reset(
					new TagRule(token_ref_to_std_string(name), vars, actions, condition));
			}
			else {
				rle.reset(
					new TagRule(token_ref_to_std_string(name), vars, actions));
			}
		}
/*
	: "rule" LPAREN name: STRING COMMA
	(
		(bool_operator[tagset, vars]) =>
		(
			condition = bool_operator [tagset, vars] COMMA
			actions = action_sequence [tagset, vars] {
				// rule(NAME, COND, ACTIONS)
				rle.reset(
					new TagRule(token_ref_to_std_string(name), vars, actions, condition));
			}
		)
	|
		(
			actions = action_sequence [tagset, vars] {
				// rule(NAME, ACTIONS)
				rle.reset(new TagRule(token_ref_to_std_string(name), vars, actions));
			}
		)
	)
	RPAREN
*/
;

// Rule sequence
rule_sequence
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<RuleSequence> rule_seq]
{
	// FIXME czy tutaj przypadkiem nie powinno byc shared_ptr?
	boost::shared_ptr<TagRule> rle;

	rule_seq.reset(new RuleSequence());
}
	: rle = rule [tagset, vars] {
		rule_seq->push_back(*rle);
	}
	(
		COMMA rle = rule [tagset, vars] {
			rule_seq->push_back(*rle);
		}
	)*
;

// Temporary name. 
// This is wrapper for rule_sequence in rules section in the wccl file
rules
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<RuleSequence> rule_seq]
	: "rules" LPAREN rule_seq = rule_sequence [tagset, vars] RPAREN {
		//
	}
;

// ----------------------------------------------------------------------------
// Select action: 
//	select(position, predicate) or select(predicate);
action_select 
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Select> action]
{
	boost::shared_ptr<Function<Position> > pos;
	boost::shared_ptr<Function<Bool> > condition;
}
	: "select" LPAREN
	(
		(position_operator [tagset, vars]) =>
		(
			pos = position_operator [tagset, vars] COMMA
			condition = bool_operator [tagset, vars] {
				// select(positon, condition); 
				action.reset(new Select(condition, pos));
			}
		)
	|
		(
			condition = bool_operator [tagset, vars] {
				// select(condition); 
				action.reset(new Select(condition));
			}
		)
	) 
	RPAREN
;

// ----------------------------------------------------------------------------
// Delete action
//	delete(position, predicate) or delete(predicate);
action_delete
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Delete> action]
{
	boost::shared_ptr<Function<Position> > pos;
	boost::shared_ptr<Function<Bool> > condition;
}
	: "delete" LPAREN
	(
		(position_operator [tagset, vars]) =>
		(
			pos = position_operator [tagset, vars] COMMA
			condition = bool_operator [tagset, vars] {
				// delete(positon, condition); 
				action.reset(new Delete(condition, pos));
			}
		)
	|
		(
			condition = bool_operator [tagset, vars] {
				// delete(condition); 
				action.reset(new Delete(condition));
			}
		)
	) 
	RPAREN
;

// ----------------------------------------------------------------------------
// Relabel action
// 	relabel(pos, symset, predicate) or relabel(symset, predicate)
action_relabel
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Relabel> action]
{
	boost::shared_ptr<Function<Position> > pos;
	boost::shared_ptr<Function<Bool> > condition;
	boost::shared_ptr<Function<TSet> > replace_with;
}
	: "relabel" LPAREN
	(
		(position_operator [tagset, vars]) =>
		(
			pos = position_operator [tagset, vars] COMMA
			replace_with = symset_operator [tagset, vars] COMMA
			condition = bool_operator [tagset, vars] {
				// relabel(pos, symset, predicate)
				action.reset(new Relabel(replace_with, condition, pos));
			}
		)
	|
		(
			replace_with = symset_operator [tagset, vars] COMMA
			condition = bool_operator [tagset, vars] {
				// relabel(symset, predicate)
				action.reset(new Relabel(replace_with, condition));
			}
		)
	)
	RPAREN
;

// ----------------------------------------------------------------------------
// Unify action 
action_unify
	[const Corpus2::Tagset& tagset, Variables& vars]
	returns [boost::shared_ptr<Unify> action]
{
	boost::shared_ptr<Function<TSet> > attribs_expr;
	boost::shared_ptr<Function<Position> > pos_begin, pos_end;
}
	: "unify" LPAREN
			pos_begin    = position_operator [tagset, vars] COMMA
			pos_end      = position_operator [tagset, vars] COMMA
			attribs_expr = symset_operator   [tagset, vars]
		RPAREN  {
			action.reset(new Unify(pos_begin, pos_end, attribs_expr));
		}
;

///////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////
// ANTLR LEXER
///////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////
class ANTLRLexer extends Lexer;
options {
	exportVocab    = ANTLRExpr;
	charVocabulary = '\3'..'\377';
	testLiterals   = false;
	k              = 2;
}


// TODO
STRING
options {
	paraphrase = "a string";
}
	: '"'!  (~('"'  | '\n' | '\r'))* '"'!
	| '\''! (~('\'' | '\n' | '\r'))* '\''!
;

SIGNED_INT
options {
	paraphrase = "Signed integer";
}
	: ('-'|'+') (' '!|'\t'!)* ('0'..'9')+ 
;	

UNSIGNED_INT
options {
	paraphrase = "Unsigned integer";
}
	: ('0'..'9')+ 
;	


QUOT_MARK
options {
	paraphrase = "Quote";
} 
	: '\'' 
;

APOS_MARK
options {
	paraphrase = "Apostrophe";
}
	: '"' 
;

Q_MARK
options {
	paraphrase = "Question mark";
}
	: '?'
;

E_MARK
options {
	paraphrase = "Exclamation mark";
}
	: '!'
;

STR_PREFIX
options {
	paraphrase = "String prefix";
}
	: "$s:"
;

TST_PREFIX
options {
	paraphrase = "Symset prefix";
}
	: "$t:"
;

BOOL_PREFIX
options {
	paraphrase = "Bool prefix";
}
	: "$b:"
;

POS_PREFIX
options {
	paraphrase = "Position prefix";
}
	: '$'
;

LBRACKET 
options {
	paraphrase = "'['";
}
	: '[' 
;

RBRACKET 
options {
	paraphrase = "']'";
}
	: ']' 
;

LPAREN
options {
	paraphrase = "'('";
}   
	: '(' 
;

RPAREN 
options {
	paraphrase = "')'";
} 
	: ')' 
;

LCURLY 
options {
	paraphrase = "'{'";
} 
	: '{' 
;

RCURLY 
options {
	paraphrase = "'}'";
} 
	: '}' 
;

AT_MARK 
options {
	paraphrase = "'@'";
} 
	: '@' 
;

COMMA
options { 
	paraphrase = "','";
}
	: ','
;

SYMBOL
options { 
	paraphrase = "Symbol"; 
	testLiterals = true; 
}
	: ('a'..'z' | 'A'..'Z' | '_') ('a'..'z' | 'A'..'Z' | '_' | '0'..'9')*
	| '`' ('a'..'z' | 'A'..'Z' | '_') ('a'..'z' | 'A'..'Z' | '_' | '0'..'9')* '`'
;

WS
	: ( ' '
	  	| '\t'
	  	| '\f'
	  	| 
			( "\r\n"
				| '\r'
				| '\n'
			) { newline(); } 
		) { $setType(antlr::Token::SKIP); } 
;

COMMENT
options {
	paraphrase = "Single line comment";
}
	: "//" (~('\n'|'\r'))* { $setType(antlr::Token::SKIP);  }
;

ML_COMMENT
options {
	paraphrase = "Multi line comment";
}
  : "/*"
	(			// TODO: test it and add reference to the site it's taken from!
				/* This actually works OK despite the ambiguity that
				'\r' '\n' can be matched in one alternative or by matching
				'\r' in one iteration and '\n' in another.. But 
				this is really matched just by one rule per (...)* 
				loop iteration, so it's OK.
				This is exactly how they do it all over the web - just
				turn off the warning for this particular token.*/
		options { 
			generateAmbigWarnings = false; 
		}
      : { LA(2)!='/' }? '*'
      | '\r' '\n' { newline(); }
      | '\r' { newline(); }
      | '\n' { newline(); }
      | ~('*'|'\n'|'\r')
  	)*
    "*/"
    { $setType(antlr::Token::SKIP); }
;

HASH
options { 
	paraphrase = "'#'"; 
}
	: '#' 
;

//DSEPARATOR
//options { 
//	paraphrase = "':-'"; 
//}
//	: ":-" 
//;