From dbb28904730f7c63a8971cd6aeff9f95081413de Mon Sep 17 00:00:00 2001
From: Adam Wardynski <award@.(win7-laptop)>
Date: Thu, 21 Apr 2011 17:35:33 +0200
Subject: [PATCH] is() match condition (matching an annotation at beginning of
a segment).
---
libwccl/CMakeLists.txt | 1 +
.../ops/match/conditions/isannotatedas.cpp | 58 +++++++++++++++++++
libwccl/ops/match/conditions/isannotatedas.h | 51 ++++++++++++++++
3 files changed, 110 insertions(+)
create mode 100644 libwccl/ops/match/conditions/isannotatedas.cpp
create mode 100644 libwccl/ops/match/conditions/isannotatedas.h
diff --git a/libwccl/CMakeLists.txt b/libwccl/CMakeLists.txt
index 05affc1..59dc656 100644
--- a/libwccl/CMakeLists.txt
+++ b/libwccl/CMakeLists.txt
@@ -62,6 +62,7 @@ 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 0000000..0b6defb
--- /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 0000000..f2dcfbb
--- /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
--
GitLab