libvspell/wfst.h

   1 #ifndef __WFST_H__ // -*- tab-width: 2 mode: c++ -*-
   2 #define __WFST_H__
   3
   4 #ifndef __DICTIONARY_H__
   5 #include "dictionary.h"
   6 #endif
   7
   8 #ifndef __WORDNODE_H__
   9 #include "wordnode.h"
  10 #endif
  11
  12 #ifndef __SPELL_H__
  13 #include "spell.h"
  14 #endif
  15
  16 #ifndef __VECTOR__
  17 #include <vector>
  18 #endif
  19
  20 #ifndef BOOST_SHARED_PTR_HPP_INCLUDED
  21 #include <boost/shared_ptr.hpp>
  22 #endif
  23
  24 #define SEGM_SEPARATOR 1
  25 #define NGRAM_LENGTH 2
  26
  27 struct Section {
  28         uint segment;
  29         uint start;
  30         uint len;
  31         void segment_best(const Lattice&,Segmentation &final_seg);
  32 };
  33
  34 class Sections: public std::vector<Section> {
  35 public:
  36         Sentence const * st;                                    /// This is used for operator<< only
  37         void construct(const Lattice &words);
  38         friend std::ostream& operator << (std::ostream &os,const Sections &s);
  39 };
  40
  41
  42 /**
  43          Segmentor takes a Sentence, a Lattice and a range, then try to generate
  44          all possible Segmentation.
  45  */
  46
  47 class Segmentor
  48 {
  49 private:
  50         struct Trace
  51         {
  52                 Segmentation s;
  53                 int next_syllable;
  54                 Trace(boost::shared_ptr<WordEntries> _we):s(_we),next_syllable(0) {}
  55         };
  56         int nr_syllables;
  57         std::vector<Trace> segs;
  58         Lattice const *_words;
  59         int from,to;
  60
  61 public:
  62         void init(const Lattice &words,
  63                                                 int from,
  64                                                 int to);
  65         bool step(Segmentation &seg);
  66         void done();
  67 };
  68
  69 class WFST
  70 {
  71 protected:
  72         bool ngram_enabled;
  73
  74 public:
  75         WFST():ngram_enabled(false) {}
  76
  77         void enable_ngram(bool enable = true) { ngram_enabled = enable; }
  78
  79         void segment_best(const Lattice &words,Segmentation &seps);
  80         void segment_best_no_fuzzy(const Lattice &words,Segmentation &seps);
  81         void segment_all(const Sentence &sent,std::vector<Segmentation> &result);
  82
  83         //private:
  84 public:                                                                                                 // for testing purpose
  85         void generate_misspelled_words(const std::vector<uint> &pos,
  86                                                                                                                                  int len,
  87                                                                                                                                  Segmentation& final_seg);
  88         // variables needed when run wfst
  89         Lattice const *p_words;
  90 };
  91
  92 #endif