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

libwccl/CMakeLists.txt

@@ -45,6 +45,7 @@ SET(libwccl_STAT_SRC
 	ops/functions/bool/predicates/regex.cpp
 	ops/functions/bool/predicates/strongagreement.cpp
 	ops/functions/bool/predicates/weakagreement.cpp
+	ops/functions/match/submatch.cpp
 	ops/functions/position/firsttoken.cpp
 	ops/functions/position/lasttoken.cpp
 	ops/functions/position/relativeposition.cpp

libwccl/ops/functions/match/submatch.cpp

+#include <libwccl/ops/functions/match/submatch.h>
+#include <libwccl/ops/functions/vargetter.h>
+
+namespace Wccl {
+
+Submatch::BaseRetValPtr Submatch::apply_internal(
+	const FunExecContext &context) const
+{
+	const RetValPtr& match = match_expr_->apply(context);
+	const MatchData& md = match->get_value();
+	return RetValPtr(md.submatch(index_ ));
+}
+
+std::string Submatch::to_string(const Corpus2::Tagset &tagset) const
+{
+	std::ostringstream ss;
+	boost::shared_ptr<VarGetter<Match> > getvar =
+		boost::dynamic_pointer_cast<VarGetter<Match> >(match_expr_);
+	if (!getvar) {
+		ss << match_expr_->to_string(tagset);
+	} else {
+		std::string s = match_expr_->to_string(tagset);
+		if (s == Match::var_repr("_M")) {
+			ss << "M";
+		} else {
+			ss << s;
+		}
+	}
+	ss << " " << name(tagset) << " " << index_;
+	return ss.str();
+}
+
+std::ostream& Submatch::write_to(std::ostream& ostream) const
+{
+	boost::shared_ptr<VarGetter<Match> > getvar =
+		boost::dynamic_pointer_cast<VarGetter<Match> >(match_expr_);
+	if (!getvar) {
+		ostream << *match_expr_;
+	} else {
+		std::string s = match_expr_->to_raw_string();
+		if (s == Match::var_repr("_M")) {
+			ostream << "M";
+		} else {
+			ostream << s;
+		}
+	}
+	ostream << " " << raw_name() << " " << index_;
+	return ostream;
+}
+
+} /* end ns Wccl */

libwccl/ops/functions/match/submatch.h

+#ifndef LIBWCCL_OPS_FUNCTIONS_MATCH_SUBMATCH_H
+#define LIBWCCL_OPS_FUNCTIONS_MATCH_SUBMATCH_H
+
+#include <libwccl/ops/function.h>
+#include <libwccl/values/match.h>
+
+
+namespace Wccl {
+
+/**
+ * Operator that takes a Match and an index and returns 
+ * a submatch at given index (indices start at 1; operator
+ * works only on Matches that have MatchVector).
+ */
+class Submatch : public Function<Match>
+{
+public:
+	typedef boost::shared_ptr<Function<Match> > MatchFunctionPtr;
+
+	Submatch(const MatchFunctionPtr& match_expr, size_t index)
+		: match_expr_(match_expr),
+		  index_(index)
+	{
+		BOOST_ASSERT(match_expr_);
+		if (index < 1) {
+			throw InvalidArgument("index", "Submatch indices start from 1.");
+		}
+	}
+
+	/**
+	 * @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 "->";
+	}
+
+protected:
+	const MatchFunctionPtr match_expr_;
+	const size_t index_;
+
+	/**
+	 * 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 submatch at given index. Works only if Match has a MatchVector.
+	 * @throws WcclError if given Match does not hold a MatchVector, or
+	 *         if the index is outside boundaries of MatchVector.
+	 * @returns Match that is in underlying MatchVector at specified index
+	 *          (note - indexing starts from 1 not from 0).
+	 */
+	BaseRetValPtr apply_internal(const FunExecContext& context) const;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_OPS_FUNCTIONS_MATCH_SUBMATCH_H

libwccl/parser/grammar.g

@@ -82,6 +82,7 @@ header {
 	#include <libwccl/ops/match/conditions/tokencondition.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>
@@ -2050,6 +2051,19 @@ match_fit
 {
 	//
 }
+	:
+	( ret = match_var_val [tagset, vars]
+	| "M" { ret.reset(new VarGetter<Match>(vars.create_accessor<Match>("_M")));	}
+	| LPAREN ret = match_fit [tagset, vars] RPAREN
+	)
+	( // if there's an arrow after the match, we have a submatch reference
+		ARROW i: UNSIGNED_INT { ret.reset(new Submatch(ret, token_ref_to_int(i))); }
+	)?
+;
+
+match_var_val
+  [const Corpus2::Tagset& tagset, Variables& vars]
+	returns [boost::shared_ptr<Function<Match> > ret]
 	: ret = match_vector_variable [vars]
 	| ret = match_value_const
 ;
@@ -2211,6 +2225,13 @@ options {
 	: ','
 ;
 
+ARROW
+options {
+	paraphrase = "->";
+}
+	: "->"
+;
+
 SYMBOL
 options { 
 	paraphrase = "Symbol"; 

libwccl/values/matchdata.h

@@ -2,10 +2,12 @@
 #define LIBWCCL_VALUES_MATCHDATA_H
 
 #include <libwccl/values/position.h>
+#include <libwccl/exception.h>
 #include <libcorpus2/ann/annotatedsentence.h>
 
 namespace Wccl {
 
+class Match;
 /**
  * Base abstract class for data held by a Match Value
  * - VectorMatch, TokenMatch or AnnotationMatch.
@@ -32,6 +34,20 @@ public:
 	 */
 	virtual Position last_token(const boost::shared_ptr<Corpus2::AnnotatedSentence>&) const = 0;
 
+	/**
+	 * Getter for a submatch at given index (indexing starts from 1).
+	 */
+	virtual const boost::shared_ptr<const Match> submatch(size_t) const {
+		throw WcclError("Getting a submatch is possible only for a MatchVector.");
+	}
+
+	/**
+	 * Getter for a submatch at given index (indexing starts from 1).
+	 */
+	virtual const boost::shared_ptr<Match>& submatch(size_t) {
+		throw WcclError("Getting a submatch is possible only for a MatchVector.");
+	}
+
 	boost::shared_ptr<MatchData> clone() const {
 		return boost::shared_ptr<MatchData>(clone_internal());
 	}

libwccl/values/matchvector.cpp

@@ -89,10 +89,18 @@ void MatchVector::append(const boost::shared_ptr<MatchData> &m)
 	matches_.push_back(boost::shared_ptr<Match>(new Match(m)));
 }
 
+const boost::shared_ptr<const Match> MatchVector::submatch(size_t idx) const {
+	if (idx + 1< matches_.size() || idx == 0) {
+		return matches_[idx - 1];
+	} else {
+		throw Wccl::WcclError("Match vector index out of range");
+	}
+}
+
 const boost::shared_ptr<Match>& MatchVector::submatch(size_t idx)
 {
-	if (idx < matches_.size()) {
-		return matches_[idx];
+	if (idx + 1 < matches_.size() || idx == 0) {
+		return matches_[idx - 1];
 	} else {
 		throw Wccl::WcclError("Match vector index out of range");
 	}

libwccl/values/matchvector.h

@@ -47,17 +47,17 @@ public:
 	}
 
 	/**
-	 * Submatch accesor with bounds checking, throws if out of bounds
+	 * Submatch accessor with bounds checking, throws if out of bounds.
+	 * @note Indexing is assumed from 1.
 	 */
 	const boost::shared_ptr<Match>& submatch(size_t idx);
 
 	/**
-	 * Submatch indexing operator. Per C++ container tradition, no bounds
-	 * checking is done.
+	 * Submatch accessor with bounds checking, throws if out of bounds.
+	 * Const version.
+	 * @note Indexing is assumed from 1.
 	 */
-	const boost::shared_ptr<Match>& operator[](size_t idx) const {
-		return matches_[idx];
-	}
+	const boost::shared_ptr<const Match> submatch(size_t idx) const;
 
 	void clear() {
 		matches_.clear();