Added contexts_corpus for reading text data files.

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@36 ec762483-ff6d-05da-a07a-a48fb63a330f

1 files changed, 39 insertions, 0 deletions

diff --git a/gi/pyp-topics/src/contexts_corpus.hh b/gi/pyp-topics/src/contexts_corpus.hh
new file mode 100644
index 00000000..e680cef5
--- /dev/null
+++ b/gi/pyp-topics/src/contexts_corpus.hh
@@ -0,0 +1,39 @@
+#ifndef _CONTEXTS_CORPUS_HH
+#define _CONTEXTS_CORPUS_HH
+
+#include <vector>
+#include <string>
+#include <map>
+
+#include <boost/ptr_container/ptr_vector.hpp>
+
+#include "corpus.hh"
+#include "contexts_lexer.h"
+#include "../../../decoder/dict.h"
+
+////////////////////////////////////////////////////////////////
+// ContextsCorpus
+////////////////////////////////////////////////////////////////
+
+class ContextsCorpus : public Corpus {
+  friend void read_callback(const ContextsLexer::PhraseContextsType&, void*);
+
+public:
+    typedef boost::ptr_vector<Document>::const_iterator const_iterator;
+
+public:
+    ContextsCorpus() {}
+    virtual ~ContextsCorpus() {}
+
+    unsigned read_contexts(const std::string &filename);
+
+    TermBackoffPtr backoff_index() {
+      return m_backoff;
+    }
+
+private:
+    TermBackoffPtr m_backoff;
+    Dict m_dict;
+};
+
+#endif // _CONTEXTS_CORPUS_HH