Class to hold lexicon data.

libwccl/CMakeLists.txt

@@ -27,6 +27,7 @@ endif(WIN32)
 
 SET(libwccl_STAT_SRC
 	exception.cpp
+	lexicon/lexicon.cpp
 	ops/formatters.cpp
 	ops/functions/bool/iteration.cpp
 	ops/functions/bool/iterations/atleast.cpp

libwccl/lexicon/lexicon.cpp

+#include <libwccl/lexicon/lexicon.h>
+#include <libpwrutils/util.h>
+#include <libwccl/exception.h>
+#include <libpwrutils/foreach.h>
+
+#include <boost/make_shared.hpp>
+
+namespace Wccl {
+
+const UnicodeString& Lexicon::translate(const icu_44::UnicodeString &key) const
+{
+	static UnicodeString empty;
+	map_t::const_iterator i = map_.find(key);
+	if (i == map_.end()) {
+		return empty;
+	}
+	return i->second;
+}
+
+boost::shared_ptr<StrSet> Lexicon::translate(const StrSet& set) const
+{
+	boost::shared_ptr<StrSet> ret_set = boost::make_shared<StrSet>();
+	foreach(const UnicodeString& s, set.get_value()) {
+		const UnicodeString& v = translate(s);
+		if (!v.isEmpty()) {
+			ret_set->insert(v);
+		}
+	}
+	return ret_set;
+}
+
+void Lexicon::insert(const UnicodeString& key, const UnicodeString& value)
+{
+	if (has_key(key)) {
+		throw InvalidArgument("key", PwrNlp::to_utf8(key) + " - entry already added.");
+	}
+	if (key.isEmpty()) {
+		throw InvalidArgument("key", "Cannot add an empty string.");
+	}
+	if (value.isEmpty()) {
+		throw InvalidArgument("value", "Cannot add an empty string.");
+	}
+	map_[key] = value;
+}
+
+} /* end ns Wccl */

libwccl/lexicon/lexicon.h

+#ifndef LIBWCCL_LEXICON_LEXICON_H
+#define LIBWCCL_LEXICON_LEXICON_H
+
+#include <boost/unordered_map.hpp>
+#include <boost/noncopyable.hpp>
+#include <libcorpus2/lexeme.h> // for unicodestring hash
+
+#include <libwccl/values/strset.h>
+
+namespace Wccl {
+
+class Lexicon : boost::noncopyable
+{
+public: 
+	typedef boost::unordered_map<UnicodeString, UnicodeString> map_t;
+
+	Lexicon(const std::string& name)
+		: name_(name)
+	{
+		BOOST_ASSERT(!name_.empty());
+	}
+
+	/**
+	 * Translate given key to a value held in this lexicon.
+	 * @returns Value assigned to the given key, if present.
+	 *          Empty UnicodeString if the key was not present.
+	 */
+	const UnicodeString& translate(const UnicodeString& key) const;
+
+
+	/**
+	 * Translate given set of strings to corresponding values
+	 * from the lexicon. 
+	 * Nonexisting keys will translate to nothing (will be removed
+	 * from output).
+	 */
+	boost::shared_ptr<StrSet> translate(const StrSet& set) const;
+
+	std::string name() const {
+		return name_;
+	}
+
+	bool has_key(const UnicodeString& key) const {
+		return map_.find(key) != map_.end();
+	}
+
+	void insert(const UnicodeString& key, const UnicodeString& value);
+	void insert(const UnicodeString& key) {
+		insert(key, key);
+	}
+
+	const map_t& map() const {
+		return map_;
+	}
+private:
+	map_t map_;
+	const std::string name_;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_LEXICON_LEXICON_H