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 __MAP__
   8 #include <map>
   9 #endif
  10 #include <libsrilm/Vocab.h>
  11 #include <libsrilm/Ngram.h>
  12 #include "debug.h"
  13
  14 #define sarch get_sarch()
  15 /*
  16         namespace Dictionary {
  17 */
  18
  19 typedef VocabIndex strid;                               // for easy migration
  20 #ifndef __MYSTRING_H__
  21 #include "mystring.h"
  22 #endif
  23
  24 typedef struct {
  25         strid id;                                                                                       // real id
  26         strid cid;                                                                              // case-insensitive comparision
  27 } strpair;
  28
  29 class StringArchive {
  30 private:
  31         Vocab dict;
  32         Vocab *rest;
  33         bool blocked;
  34
  35 public:
  36         StringArchive():rest(NULL),blocked(false) {}
  37         void set_blocked(bool _blocked);
  38         Vocab& get_dict() { return dict; }
  39         void clear_rest();
  40
  41         strid operator[] (VocabString s);
  42         strid operator[] (const std::string &s) { return (*this)[s.c_str()]; }
  43         VocabString operator[] (strid i);
  44         bool in_dict(VocabString s);
  45         bool in_dict(const std::string &s) { return in_dict(s.c_str()); }
  46         bool in_dict(strid i) { return i < dict.numWords(); }
  47
  48         void dump();
  49 };
  50
  51 //strpair make_strpair(strid id);
  52
  53 int viet_toupper(int ch);
  54 int viet_tolower(int ch);
  55 bool viet_isupper(int ch);
  56 bool viet_islower(int ch);
  57 bool viet_isalpha(int ch);
  58 bool viet_isdigit (int ch);
  59 bool viet_isxdigit(int ch);
  60 bool viet_isspace(int ch);
  61 bool viet_ispunct(int ch);
  62
  63 bool dic_init();
  64 void dic_clean();
  65 bool is_syllable_exist(strid);
  66 bool is_syllable_exist(const std::string &syll);
  67 float get_syllable(const std::string &syll);
  68 bool is_word_exist(const std::string &word);
  69 float get_word(const std::string &word);
  70
  71 #define UNK_ID 0
  72 #define PUNCT_ID 1
  73 #define PROPER_NAME_ID 2
  74 #define START_ID 3
  75 #define STOP_ID 4
  76 #define POEM_ID 5
  77 #define NUMBER_ID 6
  78 #define LEAF_ID 7
  79 #define TOTAL_ID 8
  80 StringArchive& get_sarch();
  81 const std::map<strid,strid_string>& get_pnames();
  82 Ngram& get_ngram();
  83 Ngram& get_syngram();
  84 strid get_id(int id);
  85
  86 inline bool is_syllable_exist(const std::string &syll) {
  87         return is_syllable_exist(get_sarch()[syll]);
  88 }
  89
  90 /*
  91         };
  92 */
  93
  94 #endif