From c7f5fdc27a6eaef4541c7327bd0f281c5bacccd0 Mon Sep 17 00:00:00 2001
From: Adam Wardynski <award@.(win7-laptop)>
Date: Mon, 4 Apr 2011 14:27:43 +0200
Subject: [PATCH] longest - condition that takes longest possible variant.
---
libwccl/CMakeLists.txt | 1 +
libwccl/ops/match/conditions/longest.cpp | 61 ++++++++++++++++++++++++
libwccl/ops/match/conditions/longest.h | 52 ++++++++++++++++++++
3 files changed, 114 insertions(+)
create mode 100644 libwccl/ops/match/conditions/longest.cpp
create mode 100644 libwccl/ops/match/conditions/longest.h
diff --git a/libwccl/CMakeLists.txt b/libwccl/CMakeLists.txt
index d57040a..4e2017b 100644
--- a/libwccl/CMakeLists.txt
+++ b/libwccl/CMakeLists.txt
@@ -55,6 +55,7 @@ SET(libwccl_STAT_SRC
ops/functions/tset/getsymbols.cpp
ops/functions/tset/getsymbolsinrange.cpp
ops/match/conditions/conjconditions.cpp
+ ops/match/conditions/longest.cpp
ops/match/conditions/oneof.cpp
ops/match/conditions/optionalmatch.cpp
ops/match/conditions/repeatedmatch.cpp
diff --git a/libwccl/ops/match/conditions/longest.cpp b/libwccl/ops/match/conditions/longest.cpp
new file mode 100644
index 0000000..ce8b9aa
--- /dev/null
+++ b/libwccl/ops/match/conditions/longest.cpp
@@ -0,0 +1,61 @@
+#include <libwccl/ops/match/conditions/longest.h>
+#include <libwccl/values/matchvector.h>
+#include <sstream>
+#include <libpwrutils/foreach.h>
+
+namespace Wccl {
+
+Longest::Longest(const boost::shared_ptr<std::vector<ConjConditions> >& variants)
+ : _variants(variants)
+{
+ BOOST_ASSERT(_variants);
+ BOOST_ASSERT(!_variants->empty());
+}
+
+MatchResult Longest::apply(const ActionExecContext& context) const
+{
+ int orig_pos = context.sentence_context().get_position();
+ int longest_pos = orig_pos;
+ MatchResult longest;
+
+ foreach(const ConjConditions& variant, *_variants) {
+ MatchResult res = variant.apply(context);
+ int cur_pos = context.sentence_context().get_position();
+ if (res.matched() && longest_pos < cur_pos) {
+ longest_pos = cur_pos;
+ longest = res;
+ }
+ context.sentence_context().set_position(orig_pos);
+ }
+
+ context.sentence_context().set_position(longest_pos);
+ return longest;
+}
+
+std::string Longest::to_string(const Corpus2::Tagset& tagset) const
+{
+ std::ostringstream ostream;
+ ostream << name() << "(";
+ for (size_t i = 0; i < _variants->size(); ++i) {
+ if (i != 0) {
+ ostream << ", ";
+ }
+ ostream << "variant" << _variants->at(i).to_string(tagset);
+ }
+ ostream << ")";
+ return ostream.str();
+}
+
+std::ostream& Longest::write_to(std::ostream& ostream) const
+{
+ ostream << name() << "(";
+ for (size_t i = 0; i < _variants->size(); ++i) {
+ if (i != 0) {
+ ostream << ", ";
+ }
+ ostream << "variant" << _variants->at(i);
+ }
+ return ostream << ")";
+}
+
+} /* end ns Wccl */
diff --git a/libwccl/ops/match/conditions/longest.h b/libwccl/ops/match/conditions/longest.h
new file mode 100644
index 0000000..2aed55b
--- /dev/null
+++ b/libwccl/ops/match/conditions/longest.h
@@ -0,0 +1,52 @@
+#ifndef LIBWCCL_OPS_MATCH_CONDITIONS_LONGEST_H
+#define LIBWCCL_OPS_MATCH_CONDITIONS_LONGEST_H
+
+#include <libwccl/ops/match/conditions/conjconditions.h>
+
+namespace Wccl {
+
+/**
+ * Class for "longest" condition of match
+ */
+class Longest : public MatchCondition
+{
+public:
+ Longest(const boost::shared_ptr<std::vector<ConjConditions> >& variants);
+
+ /**
+ * @returns Name of the condition.
+ */
+ std::string name() const {
+ return "longest";
+ }
+ /**
+ * Applies the condition to the given execution context.
+ * Inner match variants are executed one by one to check which
+ * one is the longest one. If any match variants were found to be "true",
+ * the one that produced the longest match is returned with "true".
+ * If there was no "true" match variant, "false" is returned instead.
+ * If a match is found, the current sentence Position is increased
+ * as to point one token after all the matched tokens, otherwise
+ * it stays unchanged.
+ */
+ MatchResult apply(const ActionExecContext& context) const;
+
+ /**
+ * @returns String representation of the MatchCondition
+ */
+ std::string to_string(const Corpus2::Tagset& tagset) const;
+
+protected:
+ /**
+ * Writes string representation of the MatchCondition 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 boost::shared_ptr<std::vector<ConjConditions> > _variants;
+};
+} /* end ns Wccl */
+
+#endif // LIBWCCL_OPS_MATCH_CONDITIONS_LONGEST_H
--
GitLab