diff --git a/libwccl/CMakeLists.txt b/libwccl/CMakeLists.txt
index 5fd29ec6ab2cc2a299698609a846698601a893df..92bfe506d4a7dd9ebd1467a471c1c31432cbbb64 100644
--- a/libwccl/CMakeLists.txt
+++ b/libwccl/CMakeLists.txt
@@ -63,6 +63,7 @@ SET(libwccl_STAT_SRC
ops/match/actions/unmarkmatch.cpp
ops/match/applyoperator.cpp
ops/match/conditions/conjconditions.cpp
+ ops/match/conditions/isannotatedas.cpp
ops/match/conditions/longest.cpp
ops/match/conditions/oneof.cpp
ops/match/conditions/optionalmatch.cpp
diff --git a/libwccl/ops/match/conditions/isannotatedas.cpp b/libwccl/ops/match/conditions/isannotatedas.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..0b6defb03b410872818d899f7faab0d496919533
--- /dev/null
+++ b/libwccl/ops/match/conditions/isannotatedas.cpp
@@ -0,0 +1,58 @@
+#include <libwccl/ops/match/conditions/isannotatedas.h>
+#include <sstream>
+
+namespace Wccl {
+
+MatchResult IsAnnotatedAs::apply(const ActionExecContext& context) const
+{
+ SentenceContext& sc = context.sentence_context();
+ boost::shared_ptr<Corpus2::AnnotatedSentence> as;
+ as = boost::dynamic_pointer_cast<Corpus2::AnnotatedSentence>(sc.get_sentence_ptr());
+ if (!as) {
+ throw InvalidArgument("context", "Operator needs an annotated sentence.");
+ }
+ if (!as->has_channel(chan_name_)) {
+ return MatchResult();
+ }
+
+ int orig_iter = sc.get_position();
+ const Corpus2::AnnotationChannel& channel = as->get_channel(chan_name_);
+ int segment_idx = channel.get_segment_at(orig_iter);
+
+ // are we even within a segment annotaded with given annotation?
+ if (segment_idx == 0) {
+ return MatchResult();
+ }
+ // ok, so are we at the beginning of the segment?
+ if (channel.get_segment_at(orig_iter - 1) != segment_idx) {
+ return MatchResult();
+ }
+ // we are at the beginning of a segment with given annotation, so match it (continuous fragment at least).
+ boost::shared_ptr<AnnotationMatch> ann_match(
+ new AnnotationMatch(orig_iter, chan_name_));
+ int segment_length = 1;
+ for (
+ int i = orig_iter + 1;
+ (i < sc.size()) && (channel.get_segment_at(i) == segment_idx);
+ ++i
+ ) {
+ ++segment_length;
+ }
+ // increase current sentence position to point after the matched segment
+ sc.set_position(orig_iter + segment_length);
+ return MatchResult(ann_match);
+}
+
+std::string IsAnnotatedAs::to_string(const Corpus2::Tagset& tagset) const
+{
+ std::ostringstream os;
+ os << name() << "(" << chan_name_ << ")";
+ return os.str();
+}
+
+std::ostream& IsAnnotatedAs::write_to(std::ostream& os) const
+{
+ return os << name() << "(" << chan_name_ << ")";
+}
+
+} /* end ns Wccl */
diff --git a/libwccl/ops/match/conditions/isannotatedas.h b/libwccl/ops/match/conditions/isannotatedas.h
new file mode 100644
index 0000000000000000000000000000000000000000..f2dcfbbb0cf57c4365836f7ed6446512fe1a45a7
--- /dev/null
+++ b/libwccl/ops/match/conditions/isannotatedas.h
@@ -0,0 +1,51 @@
+#ifndef LIBWCCL_OPS_MATCH_CONDITIONS_ISANNOTATEDAS_H
+#define LIBWCCL_OPS_MATCH_CONDITIONS_ISANNOTATEDAS_H
+
+#include <libwccl/ops/match/matchcondition.h>
+#include <libwccl/ops/function.h>
+
+namespace Wccl {
+
+/**
+ * is() match condition - matches an annotation
+ */
+class IsAnnotatedAs : public MatchCondition
+{
+public:
+ IsAnnotatedAs(const std::string& annotation_name)
+ : chan_name_(annotation_name) {
+ BOOST_ASSERT(!chan_name_.empty());
+ }
+ /**
+ * @returns Name of the Condition.
+ */
+ std::string name() const {
+ return "is";
+ }
+ /**
+ * Applies the condition to the given execution context.
+ * If a match is found, the current sentence Position is increased
+ * by the lenght of matched annotation segment.
+ */
+ MatchResult apply(const ActionExecContext& context) const;
+
+ /**
+ * @returns String representation of the Condition
+ */
+ std::string to_string(const Corpus2::Tagset& tagset) const;
+
+protected:
+ /**
+ * Writes the string representation of the Condition 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 std::string chan_name_;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_OPS_MATCH_CONDITIONS_ISANNOTATEDAS_H
diff --git a/libwccl/parser/grammar.g b/libwccl/parser/grammar.g
index 0473540241f6b7ac236a752ded0ee4e2347398f2..95d0c3b88ef8f6840461f34e5a3178e50fe9f733 100644
--- a/libwccl/parser/grammar.g
+++ b/libwccl/parser/grammar.g
@@ -54,6 +54,8 @@ header {
#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>
@@ -831,8 +833,10 @@ position_operator
[const Corpus2::Tagset& tagset, Variables& vars]
returns [boost::shared_ptr<Function<Position> > ret]
:
- ( ret = position_var_val [vars]
- | ret = position_condition [tagset, vars]
+ ( ret = position_var_val [vars]
+ | ret = position_condition [tagset, vars]
+ | ret = position_first_token [tagset, vars]
+ | ret = position_last_token [tagset, vars]
| LPAREN ret = position_operator [tagset, vars] RPAREN
)
( // if there is SIGNED_INT after the position, it is actually a relative position
@@ -881,6 +885,32 @@ position_condition
}
;
+// ----------------------------------------------------------------------------
+// Taking position of a first token in a match
+// first(Match)
+position_first_token [const Corpus2::Tagset& tagset, Variables& vars]
+ returns [boost::shared_ptr<Function<Position> > ret]
+{
+ boost::shared_ptr<Function<Match> > m;
+}
+ : "first" LPAREN m = match_fit [tagset, vars] RPAREN {
+ ret.reset(new FirstToken(m));
+ }
+;
+
+// ----------------------------------------------------------------------------
+// Taking position of a first token in a match
+// last(Match)
+position_last_token [const Corpus2::Tagset& tagset, Variables& vars]
+ returns [boost::shared_ptr<Function<Position> > ret]
+{
+ boost::shared_ptr<Function<Match> > m;
+}
+ : "last" LPAREN m = match_fit [tagset, vars] RPAREN {
+ ret.reset(new LastToken(m));
+ }
+;
+
///////////////////////////////////////////////////////////////////////////////
// Stiring operator
// Returns boost::shared_ptr<Function<StrSet> >