grammar.g

header {
	//don't try to add all the headers inside our namespace
	ANTLR_END_NAMESPACE

	#include <libwccl/parser/ParserException.h>

	#include <cstdio>
	#include <antlr/Token.hpp>
	#include <boost/lexical_cast.hpp>

	// values/variables
	#include <libwccl/variables.h>
	#include <libwccl/values/bool.h>
	#include <libwccl/values/tset.h>
	#include <libwccl/values/strset.h>
	#include <libwccl/values/position.h>
	
	// sentence context
	#include <libwccl/sentencecontext.h>

	// operators
	#include <libwccl/ops/operator.h>

	#include <libwccl/ops/functions/constant.h>
	#include <libwccl/ops/functions/vargetter.h>
	#include <libwccl/ops/functions/conditional.h>

	#include <libwccl/ops/functions/bool/varsetter.h>
	#include <libwccl/ops/functions/bool/predicates/debug.h>
	#include <libwccl/ops/functions/bool/predicates/or.h>
	#include <libwccl/ops/functions/bool/predicates/nor.h>
	#include <libwccl/ops/functions/bool/predicates/and.h>
	#include <libwccl/ops/functions/bool/predicates/regex.h>
	#include <libwccl/ops/functions/bool/predicates/intersects.h>
	#include <libwccl/ops/functions/bool/predicates/issubsetof.h>
	#include <libwccl/ops/functions/bool/predicates/isinside.h>
	#include <libwccl/ops/functions/bool/predicates/isoutside.h>
	#include <libwccl/ops/functions/bool/predicates/equals.h>
	#include <libwccl/ops/functions/bool/predicates/weakagreement.h>
	#include <libwccl/ops/functions/bool/predicates/pointagreement.h>
	#include <libwccl/ops/functions/bool/predicates/strongagreement.h>
	#include <libwccl/ops/functions/bool/predicates/annsub.h>

	#include <libwccl/ops/functions/strset/affix.h>
	#include <libwccl/ops/functions/strset/getorth.h>
	#include <libwccl/ops/functions/strset/toupper.h>
	#include <libwccl/ops/functions/strset/tolower.h>
	#include <libwccl/ops/functions/strset/getlemmas.h>
	
	#include <libwccl/ops/functions/tset/agrfilter.h>
	#include <libwccl/ops/functions/tset/catfilter.h>
	#include <libwccl/ops/functions/tset/getsymbols.h>
	#include <libwccl/ops/functions/tset/getwordclass.h>
	#include <libwccl/ops/functions/tset/getsymbolsinrange.h>

	#include <libwccl/ops/functions/position/relativeposition.h>
	#include <libwccl/ops/functions/position/lasttoken.h>
	#include <libwccl/ops/functions/position/firsttoken.h>

	#include <libwccl/ops/functions/bool/iterations/only.h>
	#include <libwccl/ops/functions/bool/iterations/atleast.h>
	#include <libwccl/ops/functions/bool/iterations/leftlook.h>
	#include <libwccl/ops/functions/bool/iterations/rightlook.h>
	
	// Rules, actions
	#include <libwccl/ops/tagrule.h>
	#include <libwccl/ops/matchrule.h>
	#include <libwccl/ops/rulesequence.h>
	//
	#include <libwccl/ops/tagactions/unify.h>
	#include <libwccl/ops/tagactions/delete.h>
	#include <libwccl/ops/tagactions/select.h>
	#include <libwccl/ops/tagactions/relabel.h>
	#include <libwccl/ops/tagactions/mark.h>
	#include <libwccl/ops/tagactions/unmark.h>

	// Match operators
	#include <libwccl/values/tokenmatch.h>
	#include <libwccl/values/annotationmatch.h>
	#include <libwccl/values/matchvector.h>
	#include <libwccl/ops/match/applyoperator.h>
	#include <libwccl/ops/match/conditions/optionalmatch.h>
	#include <libwccl/ops/match/conditions/repeatedmatch.h>
	#include <libwccl/ops/match/conditions/isannotatedas.h>
	#include <libwccl/ops/match/conditions/matchtext.h>
	#include <libwccl/ops/match/conditions/conjconditions.h>
	#include <libwccl/ops/match/conditions/tokencondition.h>
	#include <libwccl/ops/match/conditions/oneof.h>
	#include <libwccl/ops/match/actions/markmatch.h>
	#include <libwccl/ops/match/actions/unmarkmatch.h>
	#include <libwccl/ops/functions/match/submatch.h>

	// Unicode String
	#include <unicode/uniset.h>
	#include <unicode/unistr.h>

	// start our namespace again
	ANTLR_BEGIN_NAMESPACE(Wccl)
}

options {
	language = "Cpp";
	genHashLines = false;
	namespace = "Wccl";
//	genHashLines = true;
}

// ----------------------------------------------------------------------------
// ANTLR PARSER
// ----------------------------------------------------------------------------
class ANTLRParser extends Parser;
options {
	k = 1;
	buildAST = false;
	exportVocab = ANTLRExpr;
	defaultErrorHandler = false;
}
{
private:
	// 
	const UnicodeString token_ref_to_ustring(antlr::RefToken& rstr) const { 
		return UnicodeString::fromUTF8(((antlr::Token*)rstr)->getText().c_str()).unescape();
	}
	/*
	const UnicodeString str_token_ref_to_ustring(antlr::RefToken& rstr) const { 
		UnicodeString ret_ustr, ustr = token_ref_to_ustring(rstr);

		if (ustr.length() < 3) {
			return "";
		}

		ustr.extract(1, ustr.length() - 2, ret_ustr);

		return ret_ustr;
	}
	*/
	//
	const std::string str_token_rem_grav(antlr::RefToken& rstr) const {
		size_t len = 0;
		std::string ret = token_ref_to_std_string(rstr);

		if ((len = ret.length()) < 2) {
			return ret;
		}

		if (ret[0] == '`' && ret[len - 1] == '`') {
			return ret.substr(1, len - 2);
		}

		return ret;
	}
	//
	const std::string token_ref_to_std_string(antlr::RefToken& rstr) const { 
		return (((antlr::Token*)rstr)->getText());
	}
	//
	int token_ref_to_int(antlr::RefToken& rstr) { 
		return atoi(((antlr::Token*)rstr)->getText().c_str());
	}
}

///////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////
// "GLOBAL" RULES
///////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////

// ----------------------------------------------------------------------------
// Rule for parsing string set operator with scope. 
// Returns boost::shared_ptr<Operator<StrSet> >
parse_strset_operator
	[const Corpus2::Tagset &tagset]
	returns [boost::shared_ptr<Operator<StrSet> > res]
{
	Variables vars;
	boost::shared_ptr<Function<StrSet> > body;
}
	: body = strset_operator [tagset, vars] {
			res.reset(new Operator<StrSet>(body, vars));
	}
	EOF
;

// ----------------------------------------------------------------------------
// Rule for parsing bool operator with scope. 
// Returns boost::shared_ptr<Operator<Bool> > 
parse_bool_operator 
	[const Corpus2::Tagset &tagset]
	returns [boost::shared_ptr<Operator<Bool> > res]
{
	Variables vars;
	boost::shared_ptr<Function<Bool> > body;
}
	: body = bool_operator [tagset, vars] {
			res.reset(new Operator<Bool>(body, vars));
	}
	EOF
;

// ----------------------------------------------------------------------------
// Rule for parsing symbol set operator with scope.
// Returns boost::shared_ptr<Operator<TSet> >
parse_symset_operator
	[const Corpus2::Tagset &tagset]
	returns [boost::shared_ptr<Operator<TSet> > res]
{
	Variables vars;
	boost::shared_ptr<Function<TSet> > body;
}
	: body = symset_operator [tagset, vars] {
			res.reset(new Operator<TSet>(body, vars));
	}
	EOF
;

// ----------------------------------------------------------------------------
// Rule for parsing position operator with scope.
// Returns boost::shared_ptr<Operator<Position> >
parse_position_operator
	[const Corpus2::Tagset &tagset]
	returns [boost::shared_ptr<Operator<Position> > res]
{	
	Variables vars;
	boost::shared_ptr<Function<Position> > body;
}
	: body = position_operator [tagset, vars] {
			res.reset(new Operator<Position>(body, vars));
	}
	EOF
;

// ----------------------------------------------------------------------------
// ----------------------------------------------------------------------------
// Rule for parsing single WCCL Rule
// Returns boost::shared_ptr<TagRule>
parse_single_rule
	[const Corpus2::Tagset &tagset]
	returns [boost::shared_ptr<TagRule> rle]
{	
	Variables vars;
}
	: rle = rule [tagset, vars]
;

// Rule for parsing rules section in the wccl file
// Returns boost::shared_ptr<RuleSequence>
parse_rule_sequence
	[const Corpus2::Tagset& tagset]
	returns [boost::shared_ptr<RuleSequence> rule_seq]
{
	Variables vars;
}
	: rule_seq = rules[tagset, vars]
;

// ----------------------------------------------------------------------------
// ----------------------------------------------------------------------------
// Rule for parsing the match rules
// Returns boost::shared_ptr<MatchRule>
parse_match_rule
	[const Corpus2::Tagset& tagset]
	returns [boost::shared_ptr<MatchRule> ret_match]
{
	Variables vars;
	vars.get_put<Match>("_M");
}
	: ret_match = match_rule_operator[tagset, vars] 
;

///////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////
// VALUES
///////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////

// ----------------------------------------------------------------------------
// Single or muliple (comma separated) elements in string set, may be:
//   'a' "a" [] ['a'] ['a', 'b'] ["a"] ["a", "b"] ['a', "b"]
// Parsing strset literal and returning plain strset value.
// Returns boost::shared_ptr<StrSet> 
strset_literal 
	returns [boost::shared_ptr<StrSet> s_set]
{
	s_set.reset(new StrSet());
}
	: s0: STRING { 
			s_set->insert(token_ref_to_ustring(s0)); 
		}
	| LBRACKET 
		(
			s1: STRING { 
				s_set->insert(token_ref_to_ustring(s1)); 
			}
	  	(
				COMMA s2: STRING { 
					s_set->insert(token_ref_to_ustring(s2)); 
				}
			)*
	  )? 
		RBRACKET
;
// String set value as constrant string set:
// Returns boost::shared_ptr<Constant<StrSet> >
strset_value 
	returns [boost::shared_ptr<Constant<StrSet> > val]
{
	boost::shared_ptr<StrSet> set;
}
	: set = strset_literal {
		val.reset(new Constant<StrSet>(*set));
	}
;

// ----------------------------------------------------------------------------
// Element of sym set. This rule, inserts element into symbol set 
// with corresponding tagset. 
// WARNING! This rule can throw ParserException! Be careful!
symset_elem
	[const Corpus2::Tagset& tagset, boost::shared_ptr<TSet>& t_set]
	: s1: SYMBOL {
		try {
			t_set->insert_symbol(tagset, str_token_rem_grav(s1));
		}
		catch(Corpus2::TagParseError &e) {
			throw(ParserException(e.info()));
		}
	}
;

// Symset literal. Symset element may be: 
//  a, `a ` (this is guaranteed by lexer rule - SYMBOL) or {a} {`a`} {a, b} 
//  {`a`, `b`} {a, `b`} {`a`, b}
// Parsing symset literal and returning plain symset value.
// Returns boost::shared_ptr<TSet>
symset_literal
	[const Corpus2::Tagset& tagset]
	returns [boost::shared_ptr<TSet> t_set]
{
	t_set.reset(new TSet());
}
	: symset_elem [tagset, t_set]
	| LCURLY 
		(
			symset_elem [tagset, t_set] (COMMA symset_elem [tagset, t_set])* 
		)?
	  RCURLY
;
// Symset value, as constant symbol set
// Returns boost::shared_ptr<Constant<TSet> >
symset_value 
	[const Corpus2::Tagset& tagset]
	returns [boost::shared_ptr<Constant<TSet> > val]
{
	boost::shared_ptr<TSet> set;
}
	: set = symset_literal [tagset] {
		val.reset(new Constant<TSet>(*set));
	}
;

// ----------------------------------------------------------------------------
// Bool literal. May be True or False. Parsing bool literal and returning 
// plain bool value.
// Returns boost::shared_ptr<Bool>
bool_literal
	returns [boost::shared_ptr<Bool> val]
	: "True"  { val.reset(new Bool(Bool(true ))); }
	| "False" { val.reset(new Bool(Bool(false))); }
;
// Bool value, as constat bool Value
// Returns boost::shared_ptr<Constant<Bool> >
bool_value 
	returns [boost::shared_ptr<Constant<Bool> > val]
{
	boost::shared_ptr<Bool> bool_lit;
}
	: bool_lit = bool_literal {
		val.reset(new Constant<Bool>(*bool_lit));
	}
;

// ----------------------------------------------------------------------------
// Position literal may be:
// 	(+|-)?(0-9)+ or  begin or end or nowhere
// Parsing position literal and returning plain position value.
// returns boost::shared_ptr<Position>
position_literal
	returns [boost::shared_ptr<Position> val]
{
	int i = 0;
}
	: i = number {
		val.reset(new Position(Position(i)));
	}
	| "begin" {
		val.reset(new Position(Position(Position::Begin)));
	}
	| "end" {
		val.reset(new Position(Position(Position::End)));
	}
	| "nowhere" {
		val.reset(new Position(Position(Position::Nowhere)));
	}
;

// Position as constant position value
// Returns boost::shared_ptr<Constant<Position> >
position_value
	returns [boost::shared_ptr<Constant<Position> > val]
{
	boost::shared_ptr<Position> pos_lit;
}
	: pos_lit = position_literal {
		val.reset(new Constant<Position>(*pos_lit));
	}
;

// ----------------------------------------------------------------------------
// Value used into match operator such as TOK[position] and ANN[position, name]
// Returns boost::shared_ptr<Match>
match_literal
	returns [boost::shared_ptr<Match> val]
{
	boost::shared_ptr<MatchData> m;
}
	: m = match_data_literal {
		val.reset(new Match(m));
	}
;

// Constant match value
// Returns boost::shared_ptr<Constant<Match> >
match_value_const
	returns [boost::shared_ptr<Constant<Match> > val]
{
	boost::shared_ptr<Match> m;
}
	: m = match_literal {
		val.reset(new Constant<Match>(*m));
	}
;

// ----------------------------------------------------------------------------
// Value used into match operator such as TOK[position] and ANN[position, name]
// Returns boost::shared_ptr<MatchData>
match_data_literal
	returns [boost::shared_ptr<MatchData> val]
	: val = token_match_literal
	| val = ann_match_literal
	| val = match_vector_literal
;

// token match literal - TOK[position]
// Returns boost::shared_ptr<TokenMatch>
token_match_literal
	returns [boost::shared_ptr<TokenMatch> val]
{
	boost::shared_ptr<Position> p;
}
	: "TOK" LBRACKET p = position_literal RBRACKET {
		val.reset(new TokenMatch(*p));
	}
;

// annotation match literal - ANN[position, name]
// Returns boost::shared_ptr<AnnotationMatch> 
ann_match_literal
	returns [boost::shared_ptr<AnnotationMatch> val]
{
	boost::shared_ptr<Position> p;
}
	: "ANN" LBRACKET p = position_literal COMMA channel : STRING RBRACKET {
		val.reset(new AnnotationMatch(*p, token_ref_to_std_string(channel)));
	}
;

// annotation match vector literal: MATCH() or MATCH(token, ann, MATCH())
// Returns boost::shared_ptr<MatchVector>
match_vector_literal
	returns [boost::shared_ptr<MatchVector> val]
{
	val.reset(new MatchVector());
}
	: "MATCH" LPAREN (match_vector_literal_item[val])? RPAREN
;

// Body of the MATCH value. It only adds vector items to the MatchVector
// Item may be single or multiple
match_vector_literal_item [boost::shared_ptr<MatchVector>& mvector]
{
	boost::shared_ptr<Match> m_val;
}
	: m_val = match_literal {
		mvector->append(m_val);
	} 
	(
		COMMA
		m_val = match_literal {
			mvector->append(m_val);
		}
	)*
;

// ----------------------------------------------------------------------------
// Number may be unsigned or signed: 1, +1, -1
number 
	returns [int ret]
{
	ret = 0;
}
	: s: SIGNED_INT   { ret = token_ref_to_int(s); }
	| u: UNSIGNED_INT { ret = token_ref_to_int(u); }
;

///////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////
// VARIABLES
///////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////

// ----------------------------------------------------------------------------
// Position: $Name
// Get position variable (however, before put into) from variables
// Returns boost::shared_ptr<VariableAccessor<Position> > 
position_variable_acc
	[Variables& vars] 
	returns [boost::shared_ptr<VariableAccessor<Position> > pos_acc]
	: POS_PREFIX n: SYMBOL { 
		vars.get_put<Position>(str_token_rem_grav(n));

		VariableAccessor<Position> acc = 
			vars.create_accessor<Position>(str_token_rem_grav(n));

		pos_acc.reset(new VariableAccessor<Position>(acc));
	}
;

// VarGetter for Position variable. This rule wrapped position_variable_acc.
// Returs boost::shared_ptr<VarGetter<Position> >
position_variable
	[Variables& vars] 
	returns [boost::shared_ptr<VarGetter<Position> > op]
{
	boost::shared_ptr<VariableAccessor<Position> > pos_acc;
}
	: pos_acc = position_variable_acc [vars] {
		op.reset(new VarGetter<Position>(*pos_acc));
	}
;

// ----------------------------------------------------------------------------