From a5a4401fe9d4ca9ad99f2c2ea9fc1120b88ac753 Mon Sep 17 00:00:00 2001
From: Matthias Huck <mhuck@inf.ed.ac.uk>
Date: Mon, 11 Jan 2016 20:14:28 +0000
Subject: [PATCH] TargetPreferencesPhraseProperty

---
 moses/PP/Factory.cpp                         |   2 +
 moses/PP/TargetPreferencesPhraseProperty.cpp | 123 +++++++++++++++++++
 moses/PP/TargetPreferencesPhraseProperty.h   |  71 +++++++++++
 3 files changed, 196 insertions(+)
 create mode 100644 moses/PP/TargetPreferencesPhraseProperty.cpp
 create mode 100644 moses/PP/TargetPreferencesPhraseProperty.h

diff --git a/moses/PP/Factory.cpp b/moses/PP/Factory.cpp
index cc393b18d..72c927072 100644
--- a/moses/PP/Factory.cpp
+++ b/moses/PP/Factory.cpp
@@ -6,6 +6,7 @@
 
 #include "moses/PP/CountsPhraseProperty.h"
 #include "moses/PP/SourceLabelsPhraseProperty.h"
+#include "moses/PP/TargetPreferencesPhraseProperty.h"
 #include "moses/PP/TreeStructurePhraseProperty.h"
 #include "moses/PP/SpanLengthPhraseProperty.h"
 #include "moses/PP/NonTermContextProperty.h"
@@ -57,6 +58,7 @@ PhrasePropertyFactory::PhrasePropertyFactory()
 
   MOSES_PNAME2("Counts", CountsPhraseProperty);
   MOSES_PNAME2("SourceLabels", SourceLabelsPhraseProperty);
+  MOSES_PNAME2("TargetPreferences", TargetPreferencesPhraseProperty);
   MOSES_PNAME2("Tree",TreeStructurePhraseProperty);
   MOSES_PNAME2("SpanLength", SpanLengthPhraseProperty);
   MOSES_PNAME2("NonTermContext", NonTermContextProperty);
diff --git a/moses/PP/TargetPreferencesPhraseProperty.cpp b/moses/PP/TargetPreferencesPhraseProperty.cpp
new file mode 100644
index 000000000..9358ee4bf
--- /dev/null
+++ b/moses/PP/TargetPreferencesPhraseProperty.cpp
@@ -0,0 +1,123 @@
+#include "moses/PP/TargetPreferencesPhraseProperty.h"
+#include <iostream>
+#include <cstdio>
+#include <cstdlib>
+#include <sstream>
+#include <string>
+#include <queue>
+#include <assert.h>
+#include <limits>
+
+namespace Moses
+{
+
+void TargetPreferencesPhraseProperty::ProcessValue(const std::string &value)
+{
+  std::istringstream tokenizer(value);
+
+  if (! (tokenizer >> m_nNTs)) { // first token: number of non-terminals (incl. left-hand side)
+    UTIL_THROW2("TargetPreferencesPhraseProperty: Not able to read number of non-terminals. Flawed property?");
+  }
+  assert( m_nNTs > 0 );
+
+  if (! (tokenizer >> m_totalCount)) { // second token: overall rule count
+    UTIL_THROW2("TargetPreferencesPhraseProperty: Not able to read overall rule count. Flawed property?");
+  }
+  assert( m_totalCount > 0.0 );
+
+
+  // read labelled rule items
+
+  std::priority_queue<float> ruleLabelledCountsPQ;
+
+  while (tokenizer.peek() != EOF) {
+    try {
+
+      TargetPreferencesPhrasePropertyItem item;
+      size_t numberOfLHSsGivenRHS = std::numeric_limits<std::size_t>::max();
+
+      if (m_nNTs == 1) {
+
+        item.m_labelsRHSCount = m_totalCount;
+
+      } else { // rule has right-hand side non-terminals, i.e. it's a hierarchical rule
+
+        for (size_t i=0; i<m_nNTs-1; ++i) { // RHS non-terminal labels
+          size_t labelRHS;
+          if (! (tokenizer >> labelRHS) ) { // RHS non-terminal label
+            UTIL_THROW2("TargetPreferencesPhraseProperty: Not able to read right-hand side label index. Flawed property?");
+          }
+          item.m_labelsRHS.push_back(labelRHS);
+        }
+
+        if (! (tokenizer >> item.m_labelsRHSCount)) {
+          UTIL_THROW2("TargetPreferencesPhraseProperty: Not able to read right-hand side count. Flawed property?");
+        }
+
+        if (! (tokenizer >> numberOfLHSsGivenRHS)) {
+          UTIL_THROW2("TargetPreferencesPhraseProperty: Not able to read number of left-hand sides. Flawed property?");
+        }
+      }
+
+      for (size_t i=0; i<numberOfLHSsGivenRHS && tokenizer.peek()!=EOF; ++i) { // LHS non-terminal labels seen with this RHS
+        size_t labelLHS;
+        if (! (tokenizer >> labelLHS)) { // LHS non-terminal label
+          UTIL_THROW2("TargetPreferencesPhraseProperty: Not able to read left-hand side label index. Flawed property?");
+        }
+        float ruleLabelledCount;
+        if (! (tokenizer >> ruleLabelledCount)) {
+          UTIL_THROW2("TargetPreferencesPhraseProperty: Not able to read count. Flawed property?");
+        }
+        item.m_labelsLHSList.push_back( std::make_pair(labelLHS,ruleLabelledCount) );
+        ruleLabelledCountsPQ.push(ruleLabelledCount);
+      }
+
+      m_labelItems.push_back(item);
+
+    } catch (const std::exception &e) {
+      UTIL_THROW2("TargetPreferencesPhraseProperty: Read error. Flawed property?");
+    }
+  }
+
+  // keep only top N label vectors
+  const size_t N=50;
+
+  if (ruleLabelledCountsPQ.size() > N) {
+
+    float topNRuleLabelledCount = std::numeric_limits<int>::max();
+    for (size_t i=0; !ruleLabelledCountsPQ.empty() && i<N; ++i) {
+      topNRuleLabelledCount = ruleLabelledCountsPQ.top();
+      ruleLabelledCountsPQ.pop();
+    }
+
+    size_t nKept=0;
+    std::list<TargetPreferencesPhrasePropertyItem>::iterator itemIter=m_labelItems.begin();
+    while (itemIter!=m_labelItems.end()) {
+      if (itemIter->m_labelsRHSCount < topNRuleLabelledCount) {
+        itemIter = m_labelItems.erase(itemIter);
+      } else {
+        std::list< std::pair<size_t,float> >::iterator itemLHSIter=(itemIter->m_labelsLHSList).begin();
+        while (itemLHSIter!=(itemIter->m_labelsLHSList).end()) {
+          if (itemLHSIter->second < topNRuleLabelledCount) {
+            itemLHSIter = (itemIter->m_labelsLHSList).erase(itemLHSIter);
+          } else {
+            if (nKept >= N) {
+              itemLHSIter = (itemIter->m_labelsLHSList).erase(itemLHSIter,(itemIter->m_labelsLHSList).end());
+            } else {
+              ++nKept;
+              ++itemLHSIter;
+            }
+          }
+        }
+        if ((itemIter->m_labelsLHSList).empty()) {
+          itemIter = m_labelItems.erase(itemIter);
+        } else {
+          ++itemIter;
+        }
+      }
+    }
+  }
+};
+
+} // namespace Moses
+
diff --git a/moses/PP/TargetPreferencesPhraseProperty.h b/moses/PP/TargetPreferencesPhraseProperty.h
new file mode 100644
index 000000000..84ef9b3c5
--- /dev/null
+++ b/moses/PP/TargetPreferencesPhraseProperty.h
@@ -0,0 +1,71 @@
+
+#pragma once
+
+#include "moses/PP/PhraseProperty.h"
+#include "util/exception.hh"
+#include <string>
+#include <list>
+
+namespace Moses
+{
+
+class TargetPreferencesPhrasePropertyItem
+{
+  friend class TargetPreferencesPhraseProperty;
+
+public:
+  TargetPreferencesPhrasePropertyItem() {};
+
+  float GetTargetPreferencesRHSCount() const {
+    return m_labelsRHSCount;
+  };
+
+  const std::list<size_t> &GetTargetPreferencesRHS() const {
+    return m_labelsRHS;
+  };
+
+  const std::list< std::pair<size_t,float> > &GetTargetPreferencesLHSList() const {
+    return m_labelsLHSList;
+  };
+
+private:
+  float m_labelsRHSCount;
+  std::list<size_t> m_labelsRHS; // should be of size nNTs-1 (empty if initial rule, i.e. no right-hand side non-terminals)
+  std::list< std::pair<size_t,float> > m_labelsLHSList; // list of left-hand sides for this right-hand side, with counts
+};
+
+
+class TargetPreferencesPhraseProperty : public PhraseProperty
+{
+public:
+  TargetPreferencesPhraseProperty() {};
+
+  virtual void ProcessValue(const std::string &value);
+
+  size_t GetNumberOfNonTerminals() const {
+    return m_nNTs;
+  }
+
+  float GetTotalCount() const {
+    return m_totalCount;
+  }
+
+  const std::list<TargetPreferencesPhrasePropertyItem> &GetTargetPreferencesItems() const {
+    return m_labelItems;
+  };
+
+  virtual const std::string *GetValueString() const {
+    UTIL_THROW2("TargetPreferencesPhraseProperty: value string not available in this phrase property");
+    return NULL;
+  };
+
+protected:
+
+  size_t m_nNTs;
+  float m_totalCount;
+
+  std::list<TargetPreferencesPhrasePropertyItem> m_labelItems;
+};
+
+} // namespace Moses
+