From 261c5b03ea34460359b6b4742dc3db4f2262af85 Mon Sep 17 00:00:00 2001
From: Adam Wardynski <award@.(win7-laptop)>
Date: Thu, 21 Apr 2011 19:02:47 +0200
Subject: [PATCH] text() match condition, matching given text.
---
libwccl/CMakeLists.txt | 1 +
libwccl/ops/match/conditions/matchtext.cpp | 49 +++++++++++++++++++++
libwccl/ops/match/conditions/matchtext.h | 51 ++++++++++++++++++++++
3 files changed, 101 insertions(+)
create mode 100644 libwccl/ops/match/conditions/matchtext.cpp
create mode 100644 libwccl/ops/match/conditions/matchtext.h
diff --git a/libwccl/CMakeLists.txt b/libwccl/CMakeLists.txt
index 92bfe50..eff2894 100644
--- a/libwccl/CMakeLists.txt
+++ b/libwccl/CMakeLists.txt
@@ -65,6 +65,7 @@ SET(libwccl_STAT_SRC
ops/match/conditions/conjconditions.cpp
ops/match/conditions/isannotatedas.cpp
ops/match/conditions/longest.cpp
+ ops/match/conditions/matchtext.cpp
ops/match/conditions/oneof.cpp
ops/match/conditions/optionalmatch.cpp
ops/match/conditions/repeatedmatch.cpp
diff --git a/libwccl/ops/match/conditions/matchtext.cpp b/libwccl/ops/match/conditions/matchtext.cpp
new file mode 100644
index 0000000..60e7c37
--- /dev/null
+++ b/libwccl/ops/match/conditions/matchtext.cpp
@@ -0,0 +1,49 @@
+#include <libwccl/ops/match/conditions/matchtext.h>
+#include <sstream>
+#include <libpwrutils/util.h>
+
+
+namespace Wccl {
+
+MatchResult MatchText::apply(const ActionExecContext& context) const
+{
+ SentenceContext& sc = context.sentence_context();
+
+ int orig_iter = sc.get_position();
+
+ UnicodeString sent_frag(sc.at(orig_iter)->orth());
+ int iter_pos = orig_iter + 1;
+ while(sent_frag.length() <= text_.length() && iter_pos < sc.size()) {
+ if (sc.at(iter_pos)->wa() != PwrNlp::Whitespace::None) {
+ sent_frag += " ";
+ }
+ sent_frag += sc.at(iter_pos)->orth();
+ ++iter_pos;
+ }
+
+ if (sent_frag != text_) {
+ return MatchResult();
+ }
+
+ boost::shared_ptr<MatchVector> v(new MatchVector());
+ for(int i = orig_iter; i < iter_pos; ++i) {
+ v->append(boost::shared_ptr<TokenMatch>(new TokenMatch(i));
+ }
+ // increase current sentence position to the point after the matched tokens.
+ sc.set_position(orig_iter + v->size());
+ return MatchResult(v);
+}
+
+std::string MatchText::to_string(const Corpus2::Tagset&) const
+{
+ std::ostringstream os;
+ os << name() << "(\"" << PwrNlp::to_utf8(text_) << "\")";
+ return os.str();
+}
+
+std::ostream& MatchText::write_to(std::ostream& os) const
+{
+ return os << name() << "(\"" << PwrNlp::to_utf8(text_) << "\")";
+}
+
+} /* end ns Wccl */
diff --git a/libwccl/ops/match/conditions/matchtext.h b/libwccl/ops/match/conditions/matchtext.h
new file mode 100644
index 0000000..f565540
--- /dev/null
+++ b/libwccl/ops/match/conditions/matchtext.h
@@ -0,0 +1,51 @@
+#ifndef LIBWCCL_OPS_MATCH_CONDITIONS_MATCHTEXT_H
+#define LIBWCCL_OPS_MATCH_CONDITIONS_MATCHTEXT_H
+
+#include <libwccl/ops/match/matchcondition.h>
+#include <libwccl/ops/function.h>
+#include <unicode/unistr.h>
+
+namespace Wccl {
+/**
+ * text() match condition - matches a text
+ */
+class MatchText : public MatchCondition
+{
+public:
+ MatchText(const UnicodeString& text)
+ : text_(text) {
+ BOOST_ASSERT(!text_.isEmpty());
+ }
+ /**
+ * @returns Name of the Condition.
+ */
+ std::string name() const {
+ return "text";
+ }
+ /**
+ * Applies the condition to the given execution context.
+ * If a match is found, the current sentence Position is increased
+ * by the amount of matched tokens.
+ */
+ 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 UnicodeString text_;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_OPS_MATCH_CONDITIONS_MATCHTEXT_H
--
GitLab