From b07720658f5a5f22ababa7bef04f8738de22ee27 Mon Sep 17 00:00:00 2001
From: Adam Wardynski <award@.(win7-laptop)>
Date: Wed, 20 Apr 2011 13:45:35 +0200
Subject: [PATCH] unmark() action for match rules.
---
libwccl/CMakeLists.txt | 1 +
libwccl/ops/match/actions/unmarkmatch.cpp | 52 ++++++++++++++++++++
libwccl/ops/match/actions/unmarkmatch.h | 58 +++++++++++++++++++++++
libwccl/parser/grammar.g | 18 +++++--
4 files changed, 126 insertions(+), 3 deletions(-)
create mode 100644 libwccl/ops/match/actions/unmarkmatch.cpp
create mode 100644 libwccl/ops/match/actions/unmarkmatch.h
diff --git a/libwccl/CMakeLists.txt b/libwccl/CMakeLists.txt
index dae2955..5466587 100644
--- a/libwccl/CMakeLists.txt
+++ b/libwccl/CMakeLists.txt
@@ -58,6 +58,7 @@ SET(libwccl_STAT_SRC
ops/functions/tset/getsymbols.cpp
ops/functions/tset/getsymbolsinrange.cpp
ops/match/actions/markmatch.cpp
+ ops/match/actions/unmarkmatch.cpp
ops/match/applyoperator.cpp
ops/match/conditions/conjconditions.cpp
ops/match/conditions/longest.cpp
diff --git a/libwccl/ops/match/actions/unmarkmatch.cpp b/libwccl/ops/match/actions/unmarkmatch.cpp
new file mode 100644
index 0000000..f1ac2a9
--- /dev/null
+++ b/libwccl/ops/match/actions/unmarkmatch.cpp
@@ -0,0 +1,52 @@
+#include <libwccl/values/match.h>
+#include <libwccl/ops/match/actions/unmarkmatch.h>
+#include <libcorpus2/ann/annotatedsentence.h>
+
+#include <sstream>
+
+namespace Wccl {
+
+void UnmarkMatch::execute(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_)) {
+ throw InvalidArgument("context", "Sentence does not have annotation channel \"" + chan_name_ + "\".");
+ }
+
+ int abs_pos = match_->apply(context)->first_token(as).get_value();
+ if(sc.is_outside(abs_pos)) {
+ throw WcclError("Received starting match that points outside sentence.");
+ }
+
+ Corpus2::AnnotationChannel& channel = as->get_channel(chan_name_);
+
+ int segment_idx = channel.get_segment_at(abs_pos);
+ if (segment_idx == 0) {
+ throw WcclError("No annotation \"" + chan_name_ + "\" to delete at the specified position.");
+ }
+
+ for (int i = 0; i < channel.size(); ++i) {
+ if (channel.segments()[i] == segment_idx) {
+ channel.set_segment_at(i, 0);
+ }
+ }
+}
+
+std::string UnmarkMatch::to_string(const Corpus2::Tagset& tagset) const
+{
+ std::ostringstream os;
+ os << name() << "(" << match_->to_string(tagset) << ", \"" << chan_name_ << "\")";
+ return os.str();
+}
+
+std::ostream& UnmarkMatch::write_to(std::ostream& os) const
+{
+ return os << name() << "(" << *match_ << ", \"" << chan_name_ << "\")";
+}
+
+} /* end ns Wccl */
diff --git a/libwccl/ops/match/actions/unmarkmatch.h b/libwccl/ops/match/actions/unmarkmatch.h
new file mode 100644
index 0000000..b0cfd31
--- /dev/null
+++ b/libwccl/ops/match/actions/unmarkmatch.h
@@ -0,0 +1,58 @@
+#ifndef LIBWCCL_OPS_MATCH_ACTIONS_UNMARKMATCH_H
+#define LIBWCCL_OPS_MATCH_ACTIONS_UNMARKMATCH_H
+
+#include <libwccl/ops/match/matchaction.h>
+#include <libwccl/ops/function.h>
+
+namespace Wccl {
+
+/**
+ * Action to unmark (delete) an annotation passing through a first
+ * token of a Match.
+ */
+class UnmarkMatch : public MatchAction
+{
+public:
+ UnmarkMatch(
+ const boost::shared_ptr<Function<Match> >& match,
+ const std::string& annotation_name)
+ : match_(match_),
+ chan_name_(annotation_name)
+ {
+ BOOST_ASSERT(match_);
+ BOOST_ASSERT(!chan_name_.empty());
+ }
+
+ /**
+ * @returns Name of the action.
+ */
+ std::string name() const {
+ return "unmark";
+ }
+
+ /**
+ * Executes the action for the given execution context.
+ */
+ void execute(const ActionExecContext& context) const;
+
+ /**
+ * @returns String representation of the expression.
+ */
+ std::string to_string(const Corpus2::Tagset& tagset) const;
+
+protected:
+ /**
+ * Writes string representation of the MatchAction to
+ * an output stream.
+ * @returns Stream written to.
+ * @note May be incomplete and/or containt internal info.
+ */
+ virtual std::ostream& write_to(std::ostream& ostream) const;
+private:
+ const boost::shared_ptr<Function<Match> >& match_;
+ const std::string chan_name_;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_OPS_MATCH_ACTIONS_UNMARKMATCH_H
diff --git a/libwccl/parser/grammar.g b/libwccl/parser/grammar.g
index 9bea745..3e553af 100644
--- a/libwccl/parser/grammar.g
+++ b/libwccl/parser/grammar.g
@@ -79,6 +79,7 @@ header {
#include <libwccl/ops/match/conditions/repeatedmatch.h>
#include <libwccl/ops/match/conditions/conjconditions.h>
#include <libwccl/ops/match/actions/markmatch.h>
+ #include <libwccl/ops/match/actions/unmarkmatch.h>
// Unicode String
#include <unicode/uniset.h>
@@ -1981,11 +1982,22 @@ match_mark_action
;
// Match unmark action
-// Returns ???
+// Returns boost::shared_ptr<UnmarkMatch>
match_unmark_action
[const Corpus2::Tagset& tagset, Variables& vars]
- returns [boost::shared_ptr<MatchAction> m_act]
- : "unmark" LPAREN /* TODO */ RPAREN
+ returns [boost::shared_ptr<UnmarkMatch> m_act]
+{
+ boost::shared_ptr<Function<Match> > match_at;
+}
+ : "unmark" LPAREN
+ match_at = match_fit[tagset, vars] COMMA
+ annotation_name : STRING
+ RPAREN {
+ m_act.reset(
+ new UnmarkMatch(
+ match_at,
+ ((antlr::Token*)annotation_name)->getText()));
+ }
;
// Match action separated by comma
--
GitLab