libvspell/dictionary.h

   1 #ifndef __DICTIONARY_H__ // -*- tab-width: 2 mode: c++ -*-
   2 #define __DICTIONARY_H__
   3
   4 #ifndef __STRING__
   5 #include <string>
   6 #endif
   7 #ifndef __VECTOR__
   8 #include <vector>
   9 #endif
  10 #ifndef __MAP__
  11 #include <map>
  12 #endif
  13 extern "C" {
  14 #include "lm/lm_3g.h"
  15 }
  16 #include "debug.h"
  17
  18 #define sarch get_ngram()
  19 /*
  20         namespace Dictionary {
  21 */
  22
  23 typedef int32 VocabIndex;
  24 typedef int32 strid;                            // for easy migration
  25 #ifndef __MYSTRING_H__
  26 #include "mystring.h"
  27 #endif
  28
  29 typedef struct {
  30         strid id;                                                                                       // real id
  31         strid cid;                                                                              // case-insensitive comparision
  32 } strpair;
  33
  34
  35 class LM {
  36 private:
  37         lm_t *lm;
  38         hash_t oov_ht;
  39         std::vector<const char *> oov;
  40         VocabIndex blocked;
  41
  42 public:
  43         bool read(const char *f);
  44         double wordProb(VocabIndex w1, const VocabIndex *wcontext);
  45         const lm_t* get_lm() const { return lm; }
  46
  47 public:
  48         LM();
  49         ~LM();
  50         void set_blocked(bool _blocked);
  51         void clear_oov();
  52
  53         strid operator[] (const char* s);
  54         strid operator[] (const std::string &s) { return (*this)[s.c_str()]; }
  55         const char* operator[] (strid i);
  56         bool in_dict(const char* s);
  57         bool in_dict(const std::string &s) { return in_dict(s.c_str()); }
  58         bool in_dict(strid i) { return i < lm3g_word_str_size(lm); }
  59
  60         void dump();
  61 };
  62
  63 //strpair make_strpair(strid id);
  64
  65 double LogPtoProb(double v);
  66
  67 int viet_toupper(int ch);
  68 int viet_tolower(int ch);
  69 bool viet_isupper(int ch);
  70 bool viet_islower(int ch);
  71 bool viet_isalpha(int ch);
  72 bool viet_isdigit (int ch);
  73 bool viet_isxdigit(int ch);
  74 bool viet_isspace(int ch);
  75 bool viet_ispunct(int ch);
  76
  77 bool dic_init();
  78 void dic_clean();
  79 bool is_syllable_exist(strid);
  80 bool is_syllable_exist(const std::string &syll);
  81 float get_syllable(const std::string &syll);
  82 bool is_word_exist(const std::string &word);
  83 float get_word(const std::string &word);
  84
  85 #define UNK_ID 0
  86 #define PUNCT_ID 1
  87 #define PROPER_NAME_ID 2
  88 #define START_ID 3
  89 #define STOP_ID 4
  90 #define POEM_ID 5
  91 #define NUMBER_ID 6
  92 #define LEAF_ID 7
  93 #define TOTAL_ID 8
  94 const std::map<strid,strid_string>& get_pnames();
  95 LM& get_ngram();
  96 LM& get_syngram();
  97 strid get_id(int id);
  98
  99 inline bool is_syllable_exist(const std::string &syll) {
 100         return is_syllable_exist(get_ngram()[syll]);
 101 }
 102
 103 /*
 104         };
 105 */
 106
 107 #endif