Initial working commit.

1 files changed, 71 insertions, 0 deletions

diff --git a/extractor/data_array.h b/extractor/data_array.h
new file mode 100644
index 00000000..6d3e99d5
--- /dev/null
+++ b/extractor/data_array.h
@@ -0,0 +1,71 @@
+#ifndef _DATA_ARRAY_H_
+#define _DATA_ARRAY_H_
+
+#include <string>
+#include <tr1/unordered_map>
+#include <vector>
+
+#include <boost/filesystem.hpp>
+
+namespace fs = boost::filesystem;
+using namespace std;
+using namespace tr1;
+
+enum Side {
+  SOURCE,
+  TARGET
+};
+
+class DataArray {
+ public:
+  static int END_OF_FILE;
+  static int END_OF_LINE;
+  static string END_OF_FILE_STR;
+  static string END_OF_LINE_STR;
+
+  DataArray();
+
+  DataArray(const string& filename);
+
+  DataArray(const string& filename, const Side& side);
+
+  virtual ~DataArray();
+
+  virtual const vector<int>& GetData() const;
+
+  virtual int AtIndex(int index) const;
+
+  virtual int GetSize() const;
+
+  virtual int GetVocabularySize() const;
+
+  virtual bool HasWord(const string& word) const;
+
+  virtual int GetWordId(const string& word) const;
+
+  string GetWord(int word_id) const;
+
+  int GetNumSentences() const;
+
+  int GetSentenceStart(int position) const;
+
+  virtual int GetSentenceId(int position) const;
+
+  void WriteBinary(const fs::path& filepath) const;
+
+  void WriteBinary(FILE* file) const;
+
+ private:
+  void InitializeDataArray();
+  void CreateDataArray(const vector<string>& lines);
+
+  unordered_map<string, int> word2id;
+  vector<string> id2word;
+  vector<int> data;
+  // TODO(pauldb): We only need sentence_id for the source language. Maybe we
+  // can save some memory here.
+  vector<int> sentence_id;
+  vector<int> sentence_start;
+};
+
+#endif