Merge branch 'master' of nlp.pwr.wroc.pl:wccl

libwccl/CMakeLists.txt

@@ -35,6 +35,7 @@ SET(libwccl_STAT_SRC
 	ops/functions/bool/iterations/rightlook.cpp
 	ops/functions/bool/predicate.cpp
 	ops/functions/bool/predicates/and.cpp
+	ops/functions/bool/predicates/annsub.cpp
 	ops/functions/bool/predicates/debug.cpp
 	ops/functions/bool/predicates/isinside.cpp
 	ops/functions/bool/predicates/isoutside.cpp

libwccl/ops/functions/bool/predicates/annsub.cpp

+#include <libwccl/ops/functions/bool/predicates/annsub.h>
+#include <libwccl/values/match.h>
+#include <libcorpus2/ann/annotatedsentence.h>
+
+namespace Wccl {
+
+AnnSub::BaseRetValPtr AnnSub::apply_internal(const FunExecContext& context) const
+{
+	boost::shared_ptr<Corpus2::AnnotatedSentence> as
+		= boost::dynamic_pointer_cast<Corpus2::AnnotatedSentence>(
+			context.sentence_context().get_sentence_ptr());
+	if (!as) {
+		throw InvalidArgument("context", "Operator needs an annotated sentence.");
+	}
+
+	boost::shared_ptr<const Match> check_from = check_from_->apply(context);
+	boost::shared_ptr<const Match> check_to =
+		(check_from_ == check_to_) ? check_from : check_to_->apply(context);
+	int abs_left = check_from->first_token(as).get_value();
+	if (abs_left < 0) {
+		throw WcclError("Received starting match that points outside sentence.");
+	}
+	int abs_right = check_to->last_token(as).get_value();
+	if (abs_right >= context.sentence_context().size()) {
+		throw WcclError("Received ending match that points outside sentence.");
+	}
+	if (abs_left > abs_right) {
+		throw WcclError("Received starting match points after the received ending match.");
+	}
+	if (!as->has_channel(chan_name_)) {
+		as->create_channel(chan_name_);
+	}
+	Corpus2::AnnotationChannel& channel = as->get_channel(chan_name_);
+
+	int segment_idx = channel.get_segment_at(abs_left);
+	if (segment_idx == 0) {
+		return Predicate::False(context);
+	} else {
+		for (int i = abs_left + 1; i <= abs_right; ++i) {
+			if (segment_idx != channel.get_segment_at(i)) {
+				return Predicate::False(context);
+			}
+		}
+	}
+	return Predicate::True(context);
+}
+
+std::string AnnSub::to_string(const Corpus2::Tagset& tagset) const
+{
+	std::ostringstream ostream;
+	ostream << raw_name() << "(" << check_from_->to_string(tagset);
+	if (check_from_ != check_to_) {
+		ostream << ", " << check_to_->to_string(tagset);
+	}
+	ostream << ", \"" << chan_name_ << "\")";
+	return ostream.str();
+}
+
+std::ostream& AnnSub::write_to(std::ostream& ostream) const
+{
+	ostream << raw_name() << "(" << *check_from_;
+	if (check_from_ != check_to_) {
+		ostream << ", " << *check_to_;
+	}
+	ostream << ", \"" << chan_name_ << "\")";
+	return ostream;
+}
+
+} /* end ns Wccl */

libwccl/ops/functions/bool/predicates/annsub.h

+#ifndef LIBWCCL_OPS_FUNCTIONS_BOOL_PREDICATES_ANNSUB_H
+#define LIBWCCL_OPS_FUNCTIONS_BOOL_PREDICATES_ANNSUB_H
+
+#include <libwccl/ops/functions/bool/predicate.h>
+
+namespace Wccl {
+
+/**
+ * An annotation-checking match condition
+ */
+class AnnSub : public Predicate
+{
+public:
+	AnnSub(
+		const boost::shared_ptr<Function<Match> >& check_from,
+		const boost::shared_ptr<Function<Match> >& check_to,
+		const std::string& annotation_name)
+		: check_from_(check_from),
+		  check_to_(check_to),
+		  chan_name_(annotation_name)
+	{
+		BOOST_ASSERT(check_from);
+		BOOST_ASSERT(check_to);
+	}
+
+	AnnSub(
+		const boost::shared_ptr<Function<Match> >& check_from_to,
+		const std::string& annotation_name)
+		: check_from_(check_from_to),
+		  check_to_(check_from_to),
+		  chan_name_(annotation_name)
+	{
+		BOOST_ASSERT(check_from_);
+		BOOST_ASSERT(check_to_);
+	}
+
+
+	/**
+	 * @returns Name of the function
+	 */
+	std::string raw_name() const {
+		return "annsub";
+	}
+
+	/**
+	 * @returns String representation of the predicate
+	 */
+	std::string to_string(const Corpus2::Tagset& tagset) const;
+
+protected:
+	/**
+	 * Outputs the string value of the returned value
+	 * @returns True
+	 */
+	BaseRetValPtr apply_internal(const FunExecContext& context) const;
+
+	/**
+	 * Writes string representation of the AnnSub to
+	 * an output stream.
+	 * @returns Stream written to.
+	 * @note May be incomplete and/or containt internal info.
+	 */
+	std::ostream& write_to(std::ostream& ostream) const;
+
+private:
+	const boost::shared_ptr< const Function<Match> > check_from_;
+	const boost::shared_ptr< const Function<Match> > check_to_;
+	const std::string chan_name_;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_OPS_FUNCTIONS_BOOL_PREDICATES_ANNSUB_H

libwccl/ops/match/actions/markmatch.cpp

@@ -17,9 +17,9 @@ void MarkMatch::execute(const ActionExecContext& context) const
 
 	boost::shared_ptr<const Match> match_from = match_from_->apply(context);
 	boost::shared_ptr<const Match> match_to =
-		(match_from_.get() == match_to_.get()) ? match_from : match_to_->apply(context);
+		(match_from_ == match_to_) ? match_from : match_to_->apply(context);
 	boost::shared_ptr<const Match> head_match =
-		(match_from_.get() == head_match_.get()) ? match_from : head_match_->apply(context);
+		(match_from_ == head_match_) ? match_from : head_match_->apply(context);
 
 	int abs_left = match_from->first_token(as).get_value();
 	if (abs_left < 0) {

libwccl/ops/match/applyoperator.cpp

@@ -20,8 +20,10 @@ ApplyOperator::ApplyOperator(
 
 void ApplyOperator::execute(const ActionExecContext &context) const
 {
-	boost::shared_ptr<MatchVector> matches =
-		boost::dynamic_pointer_cast<MatchVector>(context.variables()->get_fast(_matches));
+	MatchVector* matches =
+		dynamic_cast<MatchVector*>(
+			&context.variables()->get_fast(_matches)->get_value());
+	BOOST_ASSERT(matches);
 	context.sentence_context().goto_start();
 	while(context.sentence_context().is_current_inside()) {
 		int orig_pos = context.sentence_context().get_position();

libwccl/parser/grammar.g

@@ -39,6 +39,7 @@ header {
 	#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>
@@ -690,7 +691,7 @@ symset_operator
 ;
 
 // ----------------------------------------------------------------------------
-// It's wrapper for symset variable and symset value.
+// A wrapper for symset variable and symset value.
 symset_var_val
 	[const Corpus2::Tagset& tagset, Variables& vars]
 	returns [boost::shared_ptr<Function<TSet> > op]
@@ -1060,6 +1061,8 @@ bool_operator
 	| ret = bool_agreement  [tagset, vars]
 	//
 	| ret = bool_phrase     [tagset, vars]
+	//
+	| ret = bool_annsub     [tagset, vars]
 	// debug operators
 	| ret = debug_print_operator [tagset, vars]
 	//
@@ -1302,6 +1305,30 @@ inter_operator
 	RPAREN
 ;
 
+// ----------------------------------------------------------------------------
+// Annotation-sub operator.
+bool_annsub
+	[const Corpus2::Tagset& tagset, Variables& vars]
+	returns [boost::shared_ptr<Function<Bool> > op]
+{
+	boost::shared_ptr< Function<Match> > match_from;
+	boost::shared_ptr< Function<Match> > match_to;
+	std::string chan_name;
+}
+	: "annsub" LPAREN
+		match_from = match_fit [tagset, vars] COMMA
+		(match_to = match_fit [tagset, vars] COMMA)?
+		name : STRING
+		RPAREN
+		{
+			if (match_to) {
+				op.reset(new AnnSub(match_from, match_to, chan_name));
+			} else {
+				op.reset(new AnnSub(match_from, chan_name));
+			}
+		}
+;
+
 // ----------------------------------------------------------------------------
 // Debug printing:
 debug_print_operator