convenience functions for token disamb lexemes

libcorpus2/token.cpp

@@ -15,9 +15,10 @@ or FITNESS FOR A PARTICULAR PURPOSE.
 */
 
 #include <libcorpus2/token.h>
-#include <sstream>
 #include <libpwrutils/foreach.h>
 #include <libcorpus2/tokenmetadata.h>
+#include <sstream>
+#include <boost/bind.hpp>
 
 namespace Corpus2 {
 
@@ -79,6 +80,25 @@ size_t Token::get_preferred_lexeme_index(const Tagset& tagset) const
 	return std::distance(lexemes_.begin(), pref);
 }
 
+bool Token::has_disamb_lexeme() const
+{
+	return std::find_if(lexemes().begin(), lexemes().end(),
+		boost::bind(&Lexeme::is_disamb, _1)) != lexemes().end();
+}
+
+int Token::count_disamb_lexemes() const
+{
+	return std::count_if(lexemes().begin(), lexemes().end(),
+		boost::bind(&Lexeme::is_disamb, _1));
+}
+
+std::pair<Token::lexeme_filter_iterator, Token::lexeme_filter_iterator> Token::disamb_lexemes() const
+{
+	lexeme_filter_iterator f1(boost::bind(&Lexeme::is_disamb, _1), lexemes().begin(), lexemes().end());
+	lexeme_filter_iterator f2(boost::bind(&Lexeme::is_disamb, _1), lexemes().end(), lexemes().end());
+	return std::make_pair(f1, f2);
+}
+
 void Token::make_ign(const Tagset& tagset)
 {
 	lexemes_.clear();

libcorpus2/token.h

@@ -23,6 +23,9 @@ or FITNESS FOR A PARTICULAR PURPOSE.
 #include <libpwrutils/util.h>
 #include <libpwrutils/whitespace.h>
 
+#include <boost/iterator/filter_iterator.hpp>
+#include <boost/function.hpp>
+
 #include <unicode/unistr.h>
 #include <memory>
 #include <string>
@@ -128,6 +131,28 @@ public:
 	 */
 	bool remove_duplicate_lexemes();
 
+	/**
+	 * Return true if there is at least one disambed lexeme in this token.
+	 */
+	bool has_disamb_lexeme() const;
+
+	/**
+	 * Return the count of disambed lexemes in this token.
+	 */
+	int count_disamb_lexemes() const;
+
+	/// Helper iterator typedef
+	typedef boost::filter_iterator<
+		boost::function<bool (const Lexeme&)>,
+		std::vector<Lexeme>::const_iterator
+	> lexeme_filter_iterator;
+
+	/**
+	 * Helper to iterate through disamb-marked lexemes only (lexemes with
+	 * the disamb flag false are skipped).
+	 */
+	std::pair<lexeme_filter_iterator, lexeme_filter_iterator> disamb_lexemes() const;
+
 	/**
 	 * Check if all the lexemes of the token have the given POS, and and
 	 * the token orth caselessly matches the given orth.