span length property

2 files changed, 57 insertions, 2 deletions

diff --git a/moses/PP/SpanLengthPhraseProperty.cpp b/moses/PP/SpanLengthPhraseProperty.cpp
new file mode 100644
index 000000000..273637170
--- /dev/null
+++ b/moses/PP/SpanLengthPhraseProperty.cpp
@@ -0,0 +1,45 @@
+#include <vector>
+#include "SpanLengthPhraseProperty.h"
+#include "moses/Util.h"
+#include "util/exception.hh"
+
+using namespace std;
+
+namespace Moses
+{
+SpanLengthPhraseProperty::SpanLengthPhraseProperty(const std::string &value)
+: PhraseProperty(value)
+{
+  vector<string> toks;
+  Tokenize(toks, value);
+
+  for (size_t i = 0; i < toks.size(); i = i + 2) {
+	  const string &span = toks[i];
+	  float count = Scan<float>(toks[i + 1]);
+	  Populate(span, count);
+  }
+}
+
+void SpanLengthPhraseProperty::Populate(const string &span, float count)
+{
+  vector<size_t> toks;
+  Tokenize<size_t>(toks, span, ",");
+  UTIL_THROW_IF2(toks.size() != 3, "Incorrect format for SpanLength: " << span);
+  Populate(toks, count);
+}
+
+void SpanLengthPhraseProperty::Populate(const std::vector<size_t> &toks, float count)
+{
+  size_t ntInd = toks[0];
+  size_t sourceLength = toks[1];
+  size_t targetLength = toks[2];
+  if (ntInd >=  m_source.size() ) {
+	  m_source.resize(ntInd + 1);
+	  m_target.resize(ntInd + 1);
+  }
+  m_source[ntInd][sourceLength] = count;
+  m_target[ntInd][targetLength] = count;
+
+}
+
+}
diff --git a/moses/PP/SpanLengthPhraseProperty.h b/moses/PP/SpanLengthPhraseProperty.h
index e25d70515..94e7ab391 100644
--- a/moses/PP/SpanLengthPhraseProperty.h
+++ b/moses/PP/SpanLengthPhraseProperty.h
@@ -1,8 +1,10 @@
 
 #pragma once
 
-#include "moses/PP/PhraseProperty.h"
 #include <string>
+#include <map>
+#include <vector>
+#include "moses/PP/PhraseProperty.h"
 
 namespace Moses
 {
@@ -10,8 +12,16 @@ namespace Moses
 class SpanLengthPhraseProperty : public PhraseProperty
 {
 public:
-	SpanLengthPhraseProperty(const std::string &value) :  PhraseProperty(value) {};
+	SpanLengthPhraseProperty(const std::string &value);
+
+protected:
+	// fractional counts
+	typedef std::map<size_t, float> Map;
+	typedef std::vector<Map> Vec;
+	Vec m_source, m_target;
 
+	void Populate(const std::string &span, float count);
+	void Populate(const std::vector<size_t> &toks, float count);
 };
 
 } // namespace Moses