From 0d1f0af81ad2a368bc3216451a87111be8b9f6f5 Mon Sep 17 00:00:00 2001
From: Paul Baltescu <pauldb89@gmail.com>
Date: Sun, 24 Nov 2013 13:19:28 +0000
Subject: Reduce memory overhead for constructing the intersector.

---
 extractor/data_array.cc | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

(limited to 'extractor/data_array.cc')

diff --git a/extractor/data_array.cc b/extractor/data_array.cc
index 82efcd51..dacc4283 100644
--- a/extractor/data_array.cc
+++ b/extractor/data_array.cc
@@ -78,7 +78,7 @@ void DataArray::CreateDataArray(const vector<string>& lines) {
 
 DataArray::~DataArray() {}
 
-const vector<int>& DataArray::GetData() const {
+vector<int> DataArray::GetData() const {
   return data;
 }
 
@@ -90,6 +90,18 @@ string DataArray::GetWordAtIndex(int index) const {
   return id2word[data[index]];
 }
 
+vector<int> DataArray::GetWordIds(int index, int size) const {
+  return vector<int>(data.begin() + index, data.begin() + index + size);
+}
+
+vector<string> DataArray::GetWords(int start_index, int size) const {
+  vector<string> words;
+  for (int word_id: GetWordIds(start_index, size)) {
+    words.push_back(id2word[word_id]);
+  }
+  return words;
+}
+
 int DataArray::GetSize() const {
   return data.size();
 }
-- 
cgit v1.2.3