introduce more efficient cleanup_feature_vectors function

shogun-toolbox · Oct 13, 2011 · 12a5173 · 12a5173
1 parent a6e49bb
commit 12a5173
Show file tree

Hide file tree

Showing 2 changed files with 29 additions and 4 deletions.
diff --git a/src/shogun/features/StringFeatures.cpp b/src/shogun/features/StringFeatures.cpp
@@ -140,10 +140,7 @@ template<class ST> void CStringFeatures<ST>::cleanup()
 		single_string=NULL;
 	}
 	else
-	{
-		for (int32_t i=0; i<num_vectors; i++)
-			cleanup_feature_vector(i);
-	}
+		cleanup_feature_vectors(0, num_vectors-1);
 
 	num_vectors=0;
 	SG_FREE(features);
@@ -176,6 +173,24 @@ template<class ST> void CStringFeatures<ST>::cleanup_feature_vector(int32_t num)
 	}
 }
 
+template<class ST> void CStringFeatures<ST>::cleanup_feature_vectors(int32_t start, int32_t stop)
+{
+	if (features && get_num_vectors())
+	{
+		ASSERT(start<get_num_vectors());
+		ASSERT(stop<get_num_vectors());
+
+		for (int32_t i=start; i<=stop; i++)
+		{
+			int32_t real_num=subset_idx_conversion(i);
+			SG_FREE(features[real_num].string);
+			features[real_num].string=NULL;
+			features[real_num].slen=0;
+		}
+		determine_maximum_string_length();
+	}
+}
+
 template<class ST> EFeatureClass CStringFeatures<ST>::get_feature_class() { return C_STRING; }
 
 template<class ST> EFeatureType CStringFeatures<ST>::get_feature_type() { return F_UNKNOWN; }
@@ -1398,6 +1413,7 @@ template<class ST> void CStringFeatures<ST>::determine_maximum_string_length()
 {
 	max_string_length=0;
 	index_t num_str=get_num_vectors();
+
 	for (int32_t i=0; i<num_str; i++)
 	{
 		max_string_length=CMath::max(max_string_length,

diff --git a/src/shogun/features/StringFeatures.h b/src/shogun/features/StringFeatures.h
@@ -128,6 +128,15 @@ template <class ST> class CStringFeatures : public CFeatures
 		 * */
 		virtual void cleanup_feature_vector(int32_t num);
 
+		/** cleanup multiple feature vectors
+		 *
+		 * possible with subset
+		 *
+		 * @param start index of first vector to be cleaned
+		 * @param end last number of the vector to be cleaned
+		 * */
+		virtual void cleanup_feature_vectors(int32_t start, int32_t stop);
+
 		/** get feature class
 		 *
 		 * @return feature class STRING