Reduce memory overhead for constructing the intersector.

1 files changed, 9 insertions, 1 deletions

diff --git a/extractor/data_array.h b/extractor/data_array.h
index 5207366d..e3823d18 100644
--- a/extractor/data_array.h
+++ b/extractor/data_array.h
@@ -51,7 +51,7 @@ class DataArray {
   virtual ~DataArray();
 
   // Returns a vector containing the word ids.
-  virtual const vector<int>& GetData() const;
+  virtual vector<int> GetData() const;
 
   // Returns the word id at the specified position.
   virtual int AtIndex(int index) const;
@@ -59,6 +59,14 @@ class DataArray {
   // Returns the original word at the specified position.
   virtual string GetWordAtIndex(int index) const;
 
+  // Returns the substring of word ids starting at the specified position and
+  // having the specified length.
+  virtual vector<int> GetWordIds(int start_index, int size) const;
+
+  // Returns the substring of words starting at the specified position and
+  // having the specified length.
+  virtual vector<string> GetWords(int start_index, int size) const;
+
   // Returns the size of the data array.
   virtual int GetSize() const;