merge paul's extractor code

1 files changed, 38 insertions, 0 deletions

diff --git a/extractor/sampler.h b/extractor/sampler.h
new file mode 100644
index 00000000..be4aa1bb
--- /dev/null
+++ b/extractor/sampler.h
@@ -0,0 +1,38 @@
+#ifndef _SAMPLER_H_
+#define _SAMPLER_H_
+
+#include <memory>
+
+using namespace std;
+
+namespace extractor {
+
+class PhraseLocation;
+class SuffixArray;
+
+/**
+ * Provides uniform sampling for a PhraseLocation.
+ */
+class Sampler {
+ public:
+  Sampler(shared_ptr<SuffixArray> suffix_array, int max_samples);
+
+  virtual ~Sampler();
+
+  // Samples uniformly at most max_samples phrase occurrences.
+  virtual PhraseLocation Sample(const PhraseLocation& location) const;
+
+ protected:
+  Sampler();
+
+ private:
+  // Round floating point number to the nearest integer.
+  int Round(double x) const;
+
+  shared_ptr<SuffixArray> suffix_array;
+  int max_samples;
+};
+
+} // namespace extractor
+
+#endif