https://github.com/wilkeraziz/mosesdecoder
Raw File
Tip revision: 0d108d213f263b5b6442c94f9c4ca97b85050bf5 authored by Ales Tamchyna on 25 November 2013, 16:36:49 UTC
filter extract file based on dev/test set
Tip revision: 0d108d2
PhrasePairCollection.h
#pragma once

#include <vector>
#include <string>

class Alignment;
class PhrasePair;
class SuffixArray;
class TargetCorpus;
class Mismatch;

class PhrasePairCollection
{
public:
  typedef unsigned int INDEX;

private:
  SuffixArray *m_suffixArray;
  TargetCorpus *m_targetCorpus;
  Alignment *m_alignment;
  std::vector<std::vector<PhrasePair*> > m_collection;
  std::vector< Mismatch* > m_mismatch, m_unaligned;
  int m_size;
  int m_max_lookup;
  int m_max_pp_target;
  int m_max_pp;

  // No copying allowed.
  PhrasePairCollection(const PhrasePairCollection&);
  void operator=(const PhrasePairCollection&);

public:
  PhrasePairCollection ( SuffixArray *, TargetCorpus *, Alignment * );
  ~PhrasePairCollection ();

  bool GetCollection( const std::vector<std::string >& sourceString );
  void Print() const;
  void PrintHTML() const;
};

// sorting helper
struct CompareBySize {
  bool operator()(const std::vector<PhrasePair*>& a, const std::vector<PhrasePair*>& b ) const {
    return a.size() > b.size();
  }
};
back to top