first(), last() ops: Position of first/last token of a Match.

libwccl/CMakeLists.txt

@@ -44,6 +44,8 @@ SET(libwccl_STAT_SRC
 	ops/functions/bool/predicates/regex.cpp
 	ops/functions/bool/predicates/strongagreement.cpp
 	ops/functions/bool/predicates/weakagreement.cpp
+	ops/functions/position/firsttoken.cpp
+	ops/functions/position/lasttoken.cpp
 	ops/functions/position/relativeposition.cpp
 	ops/functions/strset/affix.cpp
 	ops/functions/strset/getlemmas.cpp

libwccl/ops/functions/position/firsttoken.cpp

+#include <libwccl/ops/functions/position/firsttoken.h>
+#include <libwccl/ops/functions/constant.h>
+#include <libwccl/exception.h>
+
+#include <sstream>
+
+namespace Wccl {
+
+FirstToken::BaseRetValPtr FirstToken::apply_internal(
+	const FunExecContext &context) const
+{
+	boost::shared_ptr<Corpus2::AnnotatedSentence> s = 
+		boost::dynamic_pointer_cast<Corpus2::AnnotatedSentence>(
+			context.sentence_context().get_sentence_ptr());
+	if (!s) {
+		throw InvalidArgument(
+			"context",
+			"Supplied context does not have valid Corpus2::AnnotatedSentence.");
+	}
+	const Function<Match>::RetValPtr match = match_expr_->apply(context);
+	if(match->empty()) {
+		return detail::DefaultFunction<Position>()->apply(context);
+	}
+	return boost::make_shared<Position>(match->first_token(s));
+}
+
+std::string FirstToken::to_string(const Corpus2::Tagset &tagset) const
+{
+	std::ostringstream ostream;
+	ostream << name(tagset) << "(" << match_expr_->to_string(tagset) << ")";
+	return ostream.str();
+}
+
+std::ostream& FirstToken::write_to(std::ostream& ostream) const
+{
+	return ostream << raw_name() << "(" << *match_expr_ << ")";
+}
+
+} /* end ns Wccl */

libwccl/ops/functions/position/firsttoken.h

+#ifndef LIBWCCL_OPS_FUNCTIONS_POSITION_FIRSTTOKEN_H
+#define LIBWCCL_OPS_FUNCTIONS_POSITION_FIRSTTOKEN_H
+
+#include <libwccl/ops/function.h>
+#include <libwccl/values/match.h>
+
+namespace Wccl {
+
+/**
+ * Operator that takes a Match and returns Position of the first token
+ * within the given Match.
+ */
+class FirstToken : public Function<Position>
+{
+public:
+	typedef boost::shared_ptr<Function<Match> > MatchFunctionPtr;
+
+	FirstToken(const MatchFunctionPtr& match_expr)
+		: match_expr_(match_expr)
+	{
+		BOOST_ASSERT(match_expr_);
+	}
+
+	/**
+	 * @returns String representation of the function
+	 */
+	std::string to_string(const Corpus2::Tagset& tagset) const;
+
+	/**
+	 * @returns Name of the function
+	 */
+	std::string raw_name() const {
+		return "first";
+	}
+
+protected:
+	const MatchFunctionPtr match_expr_;
+
+	/**
+	 * Writes string representation of the function
+	 * @returns Stream written to.
+	 */
+	std::ostream& write_to(std::ostream& ostream) const;
+
+	/**
+	 * Takes the value of a Match from argument expression, and returns
+	 * a Position that corresponds to first token of the given Match.
+	 * @returns Position corresponding to first token of a Match.
+	 *			Nowhere in case of an empty Match.
+	 */
+	BaseRetValPtr apply_internal(const FunExecContext& context) const;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_OPS_FUNCTIONS_POSITION_FIRSTTOKEN_H

libwccl/ops/functions/position/lasttoken.cpp

+#include <libwccl/ops/functions/position/lasttoken.h>
+#include <libwccl/ops/functions/constant.h>
+#include <libwccl/exception.h>
+#include <sstream>
+
+namespace Wccl {
+
+LastToken::BaseRetValPtr LastToken::apply_internal(
+	const FunExecContext &context) const
+{
+	boost::shared_ptr<Corpus2::AnnotatedSentence> s = 
+		boost::dynamic_pointer_cast<Corpus2::AnnotatedSentence>(
+			context.sentence_context().get_sentence_ptr());
+	if (!s) {
+		throw InvalidArgument(
+			"context",
+			"Supplied context does not have valid Corpus2::AnnotatedSentence.");
+	}
+	const Function<Match>::RetValPtr match = match_expr_->apply(context);
+	if(match->empty()) {
+		return detail::DefaultFunction<Position>()->apply(context);
+	}
+	return boost::make_shared<Position>(match->last_token(s));
+}
+
+std::string LastToken::to_string(const Corpus2::Tagset &tagset) const
+{
+	std::ostringstream ostream;
+	ostream << name(tagset) << "(" << match_expr_->to_string(tagset) << ")";
+	return ostream.str();
+}
+
+std::ostream& LastToken::write_to(std::ostream& ostream) const
+{
+	return ostream << raw_name() << "(" << *match_expr_ << ")";
+}
+
+} /* end ns Wccl */

libwccl/ops/functions/position/lasttoken.h

+#ifndef LIBWCCL_OPS_FUNCTIONS_POSITION_LASTTOKEN_H
+#define LIBWCCL_OPS_FUNCTIONS_POSITION_LASTTOKEN_H
+
+#include <libwccl/ops/function.h>
+#include <libwccl/values/match.h>
+
+namespace Wccl {
+
+/**
+ * Operator that takes a Match and returns Position of the last token
+ * within the given Match.
+ */
+class LastToken : public Function<Position>
+{
+public:
+	typedef boost::shared_ptr<Function<Match> > MatchFunctionPtr;
+
+	LastToken(const MatchFunctionPtr& match_expr)
+		: match_expr_(match_expr)
+	{
+		BOOST_ASSERT(match_expr_);
+	}
+
+	/**
+	 * @returns String representation of the function
+	 */
+	std::string to_string(const Corpus2::Tagset& tagset) const;
+
+	/**
+	 * @returns Name of the function
+	 */
+	std::string raw_name() const {
+		return "last";
+	}
+
+protected:
+	const MatchFunctionPtr match_expr_;
+
+	/**
+	 * Writes string representation of the function
+	 * @returns Stream written to.
+	 */
+	std::ostream& write_to(std::ostream& ostream) const;
+
+	/**
+	 * Takes the value of a Match from argument expression, and returns
+	 * a Position that corresponds to the last token of the given Match.
+	 * @returns Position corresponding to the last token of a Match.
+	 *			Nowhere in case of an empty Match.
+	 */
+	BaseRetValPtr apply_internal(const FunExecContext& context) const;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_OPS_FUNCTIONS_POSITION_LASTTOKEN_H