From f528ac27dab11770f01595b043675dba2947a263 Mon Sep 17 00:00:00 2001 From: Paul Baltescu Date: Sun, 24 Nov 2013 13:19:28 +0000 Subject: Reduce memory overhead for constructing the intersector. --- extractor/data_array.h | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) (limited to 'extractor/data_array.h') 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& GetData() const; + virtual vector 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 GetWordIds(int start_index, int size) const; + + // Returns the substring of words starting at the specified position and + // having the specified length. + virtual vector GetWords(int start_index, int size) const; + // Returns the size of the data array. virtual int GetSize() const; -- cgit v1.2.3