fixings about include files

include/phrase-extract/ScoreFeature.h

@@ -35,7 +35,7 @@
 
 #include "util/exception.hh"
 
-#include "PhraseAlignment.h"
+#include "ExtractionPhrasePair.h"
 
 namespace MosesTraining
 {
@@ -64,17 +64,14 @@ public:
 /** Passed to each feature to be used to calculate its values */
 struct ScoreFeatureContext {
   ScoreFeatureContext(
-    const PhraseAlignmentCollection &thePhrasePair,
-    float theCount, /* Total counts of all phrase pairs*/
+    const ExtractionPhrasePair &thePhrasePair,
     const MaybeLog& theMaybeLog
   ) :
     phrasePair(thePhrasePair),
-    count(theCount),
     maybeLog(theMaybeLog) {
   }
 
-  const PhraseAlignmentCollection& phrasePair;
-  float count;
+  const ExtractionPhrasePair &phrasePair;
   MaybeLog maybeLog;
 };
 
@@ -85,16 +82,19 @@ struct ScoreFeatureContext {
 class ScoreFeature
 {
 public:
+
+  /** Some features might need to store properties in ExtractionPhrasePair,
+   *  e.g. to pass along external information loaded by a feature
+   *  which may distinguish several phrase occurrences based on sentence ID */
+  virtual void addPropertiesToPhrasePair(ExtractionPhrasePair &phrasePair,
+                                         float count,
+                                         int sentenceId) const {};
+
   /** Add the values for this feature function. */
   virtual void add(const ScoreFeatureContext& context,
                    std::vector<float>& denseValues,
                    std::map<std::string,float>& sparseValues) const = 0;
 
-  /** Return true if the two phrase pairs are equal from the point of this feature. Assume
-      that they already compare true according to PhraseAlignment.equals()
-   **/
-  virtual bool equals(const PhraseAlignment& lhs, const PhraseAlignment& rhs) const = 0;
-
   virtual ~ScoreFeature() {}
 
 };
@@ -112,17 +112,18 @@ public:
   /** Pass the unused command-line arguments to configure the extra features */
   void configure(const std::vector<std::string> args);
 
+  /** Some features might need to store properties in ExtractionPhrasePair,
+   *  e.g. to pass along external information loaded by a feature
+   *  which may distinguish several phrase occurrences based on sentence ID */
+  void addPropertiesToPhrasePair(ExtractionPhrasePair &phrasePair,
+                                 float count,
+                                 int sentenceId) const;
+
   /** Add all the features */
   void addFeatures(const ScoreFeatureContext& context,
                    std::vector<float>& denseValues,
                    std::map<std::string,float>& sparseValues) const;
 
-  /**
-    * Used to tell if the PhraseAlignment should be considered the same by all
-    * extended features.
-   **/
-  bool equals(const PhraseAlignment& lhs, const PhraseAlignment& rhs) const;
-
   const std::vector<ScoreFeaturePtr>& getFeatures() const {
     return m_features;
   }

include/phrase-extract/SentenceAlignment.h

@@ -33,7 +33,7 @@ public:
   std::vector<std::string> target;
   std::vector<std::string> source;
   std::vector<int> alignedCountS;
-  std::vector<std::vector<int> > alignedToT;
+  std::vector<std::vector<int> > alignedToT, alignedToS;
   int sentenceID;
   std::string weightString;
 
@@ -46,6 +46,8 @@ public:
   bool create(char targetString[], char sourceString[],
               char alignmentString[], char weightString[], int sentenceID, bool boundaryRules);
 
+  void invertAlignment();
+
 };
 
 }

include/phrase-extract/extract-ghkm/ExtractGHKM.h

@@ -60,6 +60,8 @@ private:
                              const std::map<std::string, std::string> &,
                              const Options &,
                              std::ostream &);
+  void WriteUnknownWordSoftMatches(const std::set<std::string> &,
+                                   std::ostream &);
   void WriteGlueGrammar(const std::set<std::string> &,
                         const std::map<std::string, int> &,
                         std::ostream &);

include/phrase-extract/extract-ghkm/Options.h

@@ -40,6 +40,7 @@ public:
     , maxScope(3)
     , minimal(false)
     , pcfg(false)
+    , treeFragments(false)
     , sentenceOffset(0)
     , unpairedExtractFormat(false)
     , unknownWordMinRelFreq(0.03f)
@@ -62,9 +63,11 @@ public:
   int maxScope;
   bool minimal;
   bool pcfg;
+  bool treeFragments;
   int sentenceOffset;
   bool unpairedExtractFormat;
   std::string unknownWordFile;
+  std::string unknownWordSoftMatchesFile;
   float unknownWordMinRelFreq;
   bool unknownWordUniform;
 };

include/phrase-extract/extract-ghkm/ScfgRuleWriter.h

@@ -21,6 +21,8 @@
 #ifndef EXTRACT_GHKM_RULE_WRITER_H_
 #define EXTRACT_GHKM_RULE_WRITER_H_
 
+#include "Subgraph.h"
+
 #include <ostream>
 
 namespace Moses
@@ -40,7 +42,9 @@ public:
     , m_inv(inv)
     , m_options(options) {}
 
-  void Write(const ScfgRule &);
+  void Write(const ScfgRule &rule, bool printEndl=true);
+
+  void Write(const ScfgRule &rule, const Subgraph &g);
 
 private:
   // Disallow copying

include/util/exception.hh

@@ -83,6 +83,9 @@ template <class Except, class Data> typename Except::template ExceptionTag<Excep
 #define UTIL_THROW(Exception, Modify) \
   UTIL_THROW_BACKEND(NULL, Exception, , Modify);
 
+#define UTIL_THROW2(Modify) \
+		UTIL_THROW_BACKEND(NULL, util::Exception, , Modify);
+
 #if __GNUC__ >= 3
 #define UTIL_UNLIKELY(x) __builtin_expect (!!(x), 0)
 #else
@@ -98,6 +101,9 @@ template <class Except, class Data> typename Except::template ExceptionTag<Excep
 #define UTIL_THROW_IF(Condition, Exception, Modify) \
   UTIL_THROW_IF_ARG(Condition, Exception, , Modify)
 
+#define UTIL_THROW_IF2(Condition, Modify) \
+  UTIL_THROW_IF_ARG(Condition, util::Exception, , Modify)
+
 // Exception that records errno and adds it to the message.
 class ErrnoException : public Exception {
   public:
@@ -111,6 +117,13 @@ class ErrnoException : public Exception {
     int errno_;
 };
 
+// file wasn't there, or couldn't be open for some reason
+class FileOpenException : public Exception {
+  public:
+	FileOpenException() throw() {}
+    ~FileOpenException() throw() {}
+};
+
 // Utilities for overflow checking.  
 class OverflowException : public Exception {
   public:

include/util/read_compressed.hh

@@ -1,5 +1,5 @@
-#ifndef UTIL_READ_COMPRESSED__
-#define UTIL_READ_COMPRESSED__
+#ifndef UTIL_READ_COMPRESSED_H
+#define UTIL_READ_COMPRESSED_H
 
 #include "util/exception.hh"
 #include "util/scoped.hh"
@@ -78,4 +78,4 @@ class ReadCompressed {
 
 } // namespace util
 
-#endif // UTIL_READ_COMPRESSED__
+#endif // UTIL_READ_COMPRESSED_H