span length

2 files changed, 44 insertions, 3 deletions

diff --git a/moses/PP/SpanLengthPhraseProperty.cpp b/moses/PP/SpanLengthPhraseProperty.cpp
index 273637170..20acb17e4 100644
--- a/moses/PP/SpanLengthPhraseProperty.cpp
+++ b/moses/PP/SpanLengthPhraseProperty.cpp
@@ -18,6 +18,10 @@ SpanLengthPhraseProperty::SpanLengthPhraseProperty(const std::string &value)
 	  float count = Scan<float>(toks[i + 1]);
 	  Populate(span, count);
   }
+
+  // totals
+  CalcTotals(m_source);
+  CalcTotals(m_target);
 }
 
 void SpanLengthPhraseProperty::Populate(const string &span, float count)
@@ -37,9 +41,44 @@ void SpanLengthPhraseProperty::Populate(const std::vector<size_t> &toks, float c
 	  m_source.resize(ntInd + 1);
 	  m_target.resize(ntInd + 1);
   }
-  m_source[ntInd][sourceLength] = count;
-  m_target[ntInd][targetLength] = count;
+  m_source[ntInd].first[sourceLength] = count;
+  m_target[ntInd].first[targetLength] = count;
+
+}
+
+void SpanLengthPhraseProperty::CalcTotals(Vec &vec)
+{
+	for (size_t i = 0; i < vec.size(); i = i + 2) {
+		float total = 0;
+
+		const Map &map = vec[i].first;
+		Map::const_iterator iter;
+		for (iter = map.begin(); iter != map.end(); ++iter) {
+			float count = iter->second;
+			total += count;
+		}
+
+		vec[i].second = total;
+	}
+}
+
+float SpanLengthPhraseProperty::GetProb(size_t ntInd, size_t sourceWidth, float smoothing) const
+{
+	float count;
+
+	const std::pair<Map, float> &data = m_source[ntInd];
+	const Map &map = data.first;
+	Map::const_iterator iter = map.find(sourceWidth);
+	if (iter == map.end()) {
+	  count = 0;
+	}
+	else {
+      count = iter->second;
+	}
+	count += smoothing;
 
+	float ret = count / (data.second + smoothing * map.size());
+	return ret;
 }
 
 }
diff --git a/moses/PP/SpanLengthPhraseProperty.h b/moses/PP/SpanLengthPhraseProperty.h
index 94e7ab391..a45974802 100644
--- a/moses/PP/SpanLengthPhraseProperty.h
+++ b/moses/PP/SpanLengthPhraseProperty.h
@@ -14,14 +14,16 @@ class SpanLengthPhraseProperty : public PhraseProperty
 public:
 	SpanLengthPhraseProperty(const std::string &value);
 
+	float GetProb(size_t ntInd, size_t sourceWidth, float smoothing) const;
 protected:
 	// fractional counts
 	typedef std::map<size_t, float> Map;
-	typedef std::vector<Map> Vec;
+	typedef std::vector<std::pair<Map, float> > Vec;
 	Vec m_source, m_target;
 
 	void Populate(const std::string &span, float count);
 	void Populate(const std::vector<size_t> &toks, float count);
+	void CalcTotals(Vec &vec);
 };
 
 } // namespace Moses