ext/hunspell/hunspell.hxx

   1 #include "license.hunspell"
   2 #include "license.myspell"
   3
   4 #include "hashmgr.hxx"
   5 #include "affixmgr.hxx"
   6 #include "suggestmgr.hxx"
   7 #include "csutil.hxx"
   8 #include "langnum.hxx"
   9 //#include "config.h"
  10
  11 #define  SPELL_COMPOUND  (1 << 0)
  12 #define  SPELL_FORBIDDEN (1 << 1)
  13
  14 #define NOCAP   0
  15 #define INITCAP 1
  16 #define ALLCAP  2
  17 #define HUHCAP  3
  18 #define HUHINITCAP  4
  19
  20 #define MAXSUGGESTION 15
  21 #define MAXSHARPS 5
  22
  23 #ifdef W32
  24 #define DLLTEST2_API __declspec(dllexport)
  25 #endif
  26
  27 #ifndef _MYSPELLMGR_HXX_
  28 #define _MYSPELLMGR_HXX_
  29
  30 #ifdef W32
  31 class DLLTEST2_API Hunspell
  32 #else
  33 class Hunspell
  34 #endif
  35 {
  36   AffixMgr*       pAMgr;
  37   HashMgr*        pHMgr;
  38   SuggestMgr*     pSMgr;
  39   char *          encoding;
  40   struct cs_info * csconv;
  41   int             langnum;
  42   int             utf8;
  43   int             complexprefixes;
  44   char**          wordbreak;
  45
  46 public:
  47
  48   /* Hunspell(aff, dic) - constructor of Hunspell class
  49    * input: path of affix file and dictionary file
  50    */
  51
  52   Hunspell(const char * affpath, const char * dpath);
  53
  54   ~Hunspell();
  55
  56   /* spell(word) - spellcheck word
  57    * output: 0 = bad word, not 0 = good word
  58    *
  59    * plus output:
  60    *   info: information bit array, fields:
  61    *     SPELL_COMPOUND  = a compound word
  62    *     SPELL_FORBIDDEN = an explicit forbidden word
  63    *   root: root (stem), when input is a word with affix(es)
  64    */
  65
  66   int spell(const char * word, int * info = NULL, char ** root = NULL);
  67
  68   /* suggest(suggestions, word) - search suggestions
  69    * input: pointer to an array of strings pointer and the (bad) word
  70    *   array of strings pointer (here *slst) may not be initialized
  71    * output: number of suggestions in string array, and suggestions in
  72    *   a newly allocated array of strings (*slts will be NULL when number
  73    *   of suggestion equals 0.)
  74    */
  75
  76   int suggest(char*** slst, const char * word);
  77   char * get_dic_encoding();
  78
  79   /* handling custom dictionary */
  80
  81   int put_word(const char * word);
  82
  83   /* pattern is a sample dictionary word
  84    * put word into custom dictionary with affix flags of pattern word
  85    */
  86
  87   int put_word_pattern(const char * word, const char * pattern);
  88
  89   /* other */
  90
  91   /* get extra word characters definied in affix file for tokenization */
  92   const char * get_wordchars();
  93   unsigned short * get_wordchars_utf16(int * len);
  94
  95   struct cs_info * get_csconv();
  96   const char * get_version();
  97
  98   /* experimental functions */
  99
 100 #ifdef HUNSPELL_EXPERIMENTAL
 101   /* suffix is an affix flag string, similarly in dictionary files */
 102
 103   int put_word_suffix(const char * word, const char * suffix);
 104
 105   /* morphological analysis */
 106
 107   char * morph(const char * word);
 108   int analyze(char*** out, const char *word);
 109
 110   char * morph_with_correction(const char * word);
 111
 112   /* stemmer function */
 113
 114   int stem(char*** slst, const char * word);
 115
 116   /* spec. suggestions */
 117   int suggest_auto(char*** slst, const char * word);
 118   int suggest_pos_stems(char*** slst, const char * word);
 119   char * get_possible_root();
 120 #endif
 121
 122 private:
 123    int    cleanword(char *, const char *, int * pcaptype, int * pabbrev);
 124    int    cleanword2(char *, const char *, w_char *, int * w_len, int * pcaptype, int * pabbrev);
 125    void   mkinitcap(char *);
 126    int    mkinitcap2(char * p, w_char * u, int nc);
 127    int    mkinitsmall2(char * p, w_char * u, int nc);
 128    void   mkallcap(char *);
 129    int    mkallcap2(char * p, w_char * u, int nc);
 130    void   mkallsmall(char *);
 131    int    mkallsmall2(char * p, w_char * u, int nc);
 132    struct hentry * checkword(const char *, int * info, char **root);
 133    char * sharps_u8_l1(char * dest, char * source);
 134    hentry * spellsharps(char * base, char *, int, int, char * tmp, int * info, char **root);
 135    int    is_keepcase(const hentry * rv);
 136    int    insert_sug(char ***slst, char * word, int *ns);
 137
 138 };
 139
 140 #endif