Merge ../mosesdecoder into perf_moses2

cruise-control/test_all_new_commits.sh

@@ -148,7 +148,7 @@ function run_single_test () {
       cd ..
       touch giza-pp.ok
     fi
-    ./bjam $MCC_CONFIGURE_ARGS --with-giza="$(pwd)/giza-pp/bin" || err="bjam with-giza"
+    ./bjam $MCC_CONFIGURE_ARGS" || err="bjam"
     srilm_dir=$(echo $MCC_CONFIGURE_ARGS | sed -r 's/.*--with-srilm=([^ ]+) .*/\1/')
     mach_type=$($srilm_dir/sbin/machine-type)
     mkdir -p "$WORKDIR/ems_workdir"

moses/ExportInterface.cpp

@@ -161,7 +161,9 @@ int
 run_as_server()
 {
 #ifdef HAVE_XMLRPC_C
-  kill(getppid(),SIGALRM);
+  if (params.GetParam("daemon")) {
+    kill(getppid(),SIGALRM);
+  }
   MosesServer::Server server(params);
   return server.run(); // actually: don't return. see Server::run()
 #else

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);

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
+

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
+

phrase-extract/extract-rules-main.cpp

@@ -697,14 +697,13 @@ void ExtractTask::saveTargetSyntacticPreference( const HoleCollection &holeColl,
     const Hole &hole = *iterHoleList;
 
     int labelI = labelIndex[ 2+holeCount ];
-    string targetLabel = "X";
     int startT = hole.GetStart(1);
     int endT = hole.GetEnd(1);
     if (m_sentence.targetTree.HasNode(startT,endT)) {
       rule.targetSyntacticPreference += m_sentence.targetTree.GetNodes(startT,endT)[labelI]->label;
       rule.targetSyntacticPreference += " ";
     } else {
-      rule.targetSyntacticPreference += "X ";
+      rule.targetSyntacticPreference += "XRHS ";
     }
     ++holeCount;
   }
@@ -815,7 +814,7 @@ void ExtractTask::saveHieroPhrase( int startT, int endT, int startS, int endS
       rule.targetSyntacticPreference += " ";
       rule.targetSyntacticPreference += m_sentence.targetTree.GetNodes(startT,endT)[labelIndex[0] ]->label;
     } else {
-      rule.targetSyntacticPreference += " X";
+      rule.targetSyntacticPreference += " XLHS";
     }
   }
 
@@ -1098,7 +1097,7 @@ void ExtractTask::addRule( int startT, int endT, int startS, int endS, int count
     if (m_sentence.targetTree.HasNode(startT,endT)) {
       rule.targetSyntacticPreference += m_sentence.targetTree.GetNodes(startT,endT)[0]->label;
     } else {
-      rule.targetSyntacticPreference += "X";
+      rule.targetSyntacticPreference += "XLHS";
     }
   }
 
@@ -1233,14 +1232,30 @@ void writeGlueGrammar( const string & fileName, RuleExtractionOptions &options,
 {
   ofstream grammarFile;
   grammarFile.open(fileName.c_str());
+
   std::string glueRulesPhraseProperty = "";
   if (options.phraseOrientation) {
-    glueRulesPhraseProperty.append(" ||| ||| {{Orientation 1 1 0.5 0.5 1 1 0.5 0.5}}");
+    glueRulesPhraseProperty.append(" {{Orientation 1 1 0.5 0.5 1 1 0.5 0.5}}");
   }
+  const size_t targetSyntacticPreferencesLabelGlueTop = 0;
+  const size_t targetSyntacticPreferencesLabelGlueX = 1;
+
   if (!options.targetSyntax || options.targetSyntacticPreferences) {
-    grammarFile << "<s> [X] ||| <s> [S] ||| 1 ||| 0-0 ||| 0" << glueRulesPhraseProperty << endl
-                << "[X][S] </s> [X] ||| [X][S] </s> [S] ||| 1 ||| 0-0 1-1 ||| 0" << glueRulesPhraseProperty << endl
-                << "[X][S] [X][X] [X] ||| [X][S] [X][X] [S] ||| 2.718 ||| 0-0 1-1 ||| 0" << glueRulesPhraseProperty << endl;
+    grammarFile << "<s> [X] ||| <s> [S] ||| 1 ||| 0-0 ||| 0 ||| |||" << glueRulesPhraseProperty;
+    if (options.targetSyntacticPreferences) {
+      grammarFile << " {{TargetPreferences 1 1 " << targetSyntacticPreferencesLabelGlueTop << " 1}}";
+    }
+    grammarFile << std::endl;
+    grammarFile << "[X][S] </s> [X] ||| [X][S] </s> [S] ||| 1 ||| 0-0 1-1 ||| 0 ||| |||" << glueRulesPhraseProperty;
+    if (options.targetSyntacticPreferences) {
+      grammarFile << " {{TargetPreferences 2 1 " << targetSyntacticPreferencesLabelGlueTop << " 1 1 " << targetSyntacticPreferencesLabelGlueTop << " 1}}";
+    }
+    grammarFile << std::endl;
+    grammarFile << "[X][S] [X][X] [X] ||| [X][S] [X][X] [S] ||| 2.718 ||| 0-0 1-1 ||| 0 ||| |||" << glueRulesPhraseProperty;
+    if (options.targetSyntacticPreferences) {
+      grammarFile << " {{TargetPreferences 3 1 " << targetSyntacticPreferencesLabelGlueTop << " " << targetSyntacticPreferencesLabelGlueX << " 1 1 " << targetSyntacticPreferencesLabelGlueTop << " 1}}";
+    }
+    grammarFile << std::endl;
   } else {
     // choose a top label that is not already a label
     string topLabel = "QQQQQQ";
@@ -1251,21 +1266,21 @@ void writeGlueGrammar( const string & fileName, RuleExtractionOptions &options,
       }
     }
     // basic rules
-    grammarFile << "<s> [X] ||| <s> [" << topLabel << "] ||| 1  ||| 0-0" << endl
-                << "[X][" << topLabel << "] </s> [X] ||| [X][" << topLabel << "] </s> [" << topLabel << "] ||| 1 ||| 0-0 1-1" << endl;
+    grammarFile << "<s> [X] ||| <s> [" << topLabel << "] ||| 1  ||| 0-0" << std::endl
+                << "[X][" << topLabel << "] </s> [X] ||| [X][" << topLabel << "] </s> [" << topLabel << "] ||| 1 ||| 0-0 1-1" << std::endl;
 
     // top rules
     for( map<string,int>::const_iterator i =  targetTopLabelCollection.begin();
          i !=  targetTopLabelCollection.end(); i++ ) {
-      grammarFile << "<s> [X][" << i->first << "] </s> [X] ||| <s> [X][" << i->first << "] </s> [" << topLabel << "] ||| 1 ||| 0-0 1-1 2-2" << endl;
+      grammarFile << "<s> [X][" << i->first << "] </s> [X] ||| <s> [X][" << i->first << "] </s> [" << topLabel << "] ||| 1 ||| 0-0 1-1 2-2" << std::endl;
     }
 
     // glue rules
     for( set<string>::const_iterator i =  targetLabelCollection.begin();
          i !=  targetLabelCollection.end(); i++ ) {
-      grammarFile << "[X][" << topLabel << "] [X][" << *i << "] [X] ||| [X][" << topLabel << "] [X][" << *i << "] [" << topLabel << "] ||| 2.718 ||| 0-0 1-1" << endl;
+      grammarFile << "[X][" << topLabel << "] [X][" << *i << "] [X] ||| [X][" << topLabel << "] [X][" << *i << "] [" << topLabel << "] ||| 2.718 ||| 0-0 1-1" << std::endl;
     }
-    grammarFile << "[X][" << topLabel << "] [X][X] [X] ||| [X][" << topLabel << "] [X][X] [" << topLabel << "] ||| 2.718 |||  0-0 1-1 " << endl; // glue rule for unknown word...
+    grammarFile << "[X][" << topLabel << "] [X][X] [X] ||| [X][" << topLabel << "] [X][X] [" << topLabel << "] ||| 2.718 |||  0-0 1-1 " << std::endl; // glue rule for unknown word...
   }
   grammarFile.close();
 }

scripts/generic/score-parallel.perl

@@ -314,7 +314,7 @@ if (!$inverse && defined($partsOfSpeechFile))
 # merge target syntactic preferences labels files
 if (!$inverse && defined($targetSyntacticPreferencesLabelsFile))
 {
-  my $cmd = "(echo \"GlueTop 0\"; echo \"GlueX 1\"; echo \"SSTART 2\"; echo \"SEND 3\"; cat $TMPDIR/phrase-table.half.*.gz.syntaxLabels.tgtpref | LC_ALL=C sort | uniq | perl -pe \"s/\$/ \@{[\$.+3]}/\") > $targetSyntacticPreferencesLabelsFile";
+  my $cmd = "(echo \"GlueTop 0\"; echo \"GlueX 1\"; cat $TMPDIR/phrase-table.half.*.gz.syntaxLabels.tgtpref | LC_ALL=C sort | uniq | perl -pe \"s/\$/ \@{[\$.+3]}/\") > $targetSyntacticPreferencesLabelsFile";
   print STDERR "Merging target syntactic preferences labels files: $cmd \n";
   `$cmd`;
 }