ext/hunspell/affixmgr.hxx

   1 #ifndef _AFFIXMGR_HXX_
   2 #define _AFFIXMGR_HXX_
   3
   4 #ifdef MOZILLA_CLIENT
   5 #ifdef __SUNPRO_CC // for SunONE Studio compiler
   6 using namespace std;
   7 #endif
   8 #include <stdio.h>
   9 #else
  10 #include <cstdio>
  11 #endif
  12
  13 #include "atypes.hxx"
  14 #include "baseaffix.hxx"
  15 #include "hashmgr.hxx"
  16
  17 // check flag duplication
  18 #define dupSFX        (1 << 0)
  19 #define dupPFX        (1 << 1)
  20
  21 class AffixMgr
  22 {
  23
  24   AffEntry *          pStart[SETSIZE];
  25   AffEntry *          sStart[SETSIZE];
  26   AffEntry *          pFlag[CONTSIZE];
  27   AffEntry *          sFlag[CONTSIZE];
  28   HashMgr *           pHMgr;
  29   char *              trystring;
  30   char *              encoding;
  31   struct cs_info *    csconv;
  32   int                 utf8;
  33   int                 complexprefixes;
  34   FLAG                compoundflag;
  35   FLAG                compoundbegin;
  36   FLAG                compoundmiddle;
  37   FLAG                compoundend;
  38   FLAG                compoundroot;
  39   FLAG                compoundforbidflag;
  40   FLAG                compoundpermitflag;
  41   int                 checkcompounddup;
  42   int                 checkcompoundrep;
  43   int                 checkcompoundcase;
  44   int                 checkcompoundtriple;
  45   FLAG                forbiddenword;
  46   FLAG                nosuggest;
  47   FLAG                pseudoroot;
  48   int                 cpdmin;
  49   int                 numrep;
  50   replentry *         reptable;
  51   int                 nummap;
  52   mapentry *          maptable;
  53   int                 numbreak;
  54   char **             breaktable;
  55   int                 numcheckcpd;
  56   replentry *         checkcpdtable;
  57   int                 numdefcpd;
  58   flagentry *         defcpdtable;
  59   int                 maxngramsugs;
  60   int                 nosplitsugs;
  61   int                 sugswithdots;
  62   int                 cpdwordmax;
  63   int                 cpdmaxsyllable;
  64   char *              cpdvowels;
  65   w_char *            cpdvowels_utf16;
  66   int                 cpdvowels_utf16_len;
  67   char *              cpdsyllablenum;
  68   const char *        pfxappnd; // BUG: not stateless
  69   const char *        sfxappnd; // BUG: not stateless
  70   FLAG                sfxflag;  // BUG: not stateless
  71   char *              derived;  // BUG: not stateless
  72   AffEntry *          sfx;      // BUG: not stateless
  73   AffEntry *          pfx;      // BUG: not stateless
  74   int                 checknum;
  75   char *              wordchars;
  76   unsigned short *    wordchars_utf16;
  77   int                 wordchars_utf16_len;
  78   char *              ignorechars;
  79   unsigned short *    ignorechars_utf16;
  80   int                 ignorechars_utf16_len;
  81   char *              version;
  82   char *              lang;
  83   int                 langnum;
  84   FLAG                lemma_present;
  85   FLAG                circumfix;
  86   FLAG                onlyincompound;
  87   FLAG                keepcase;
  88   int                 checksharps;
  89
  90   int                 havecontclass; // boolean variable
  91   char                contclasses[CONTSIZE]; // flags of possible continuing classes (twofold affix)
  92   flag                flag_mode;
  93
  94 public:
  95
  96   AffixMgr(const char * affpath, HashMgr * ptr);
  97   ~AffixMgr();
  98   struct hentry *     affix_check(const char * word, int len,
  99             const unsigned short needflag = (unsigned short) 0, char in_compound = IN_CPD_NOT);
 100   struct hentry *     prefix_check(const char * word, int len,
 101             char in_compound, const FLAG needflag = FLAG_NULL);
 102   inline int isSubset(const char * s1, const char * s2);
 103   struct hentry *     prefix_check_twosfx(const char * word, int len,
 104             char in_compound, const FLAG needflag = FLAG_NULL);
 105   inline int isRevSubset(const char * s1, const char * end_of_s2, int len);
 106   struct hentry *     suffix_check(const char * word, int len, int sfxopts, AffEntry* ppfx,
 107                         char ** wlst, int maxSug, int * ns, const FLAG cclass = FLAG_NULL,
 108                         const FLAG needflag = FLAG_NULL, char in_compound = IN_CPD_NOT);
 109   struct hentry *     suffix_check_twosfx(const char * word, int len,
 110             int sfxopts, AffEntry* ppfx, const FLAG needflag = FLAG_NULL);
 111
 112   char * affix_check_morph(const char * word, int len,
 113                     const FLAG needflag = FLAG_NULL, char in_compound = IN_CPD_NOT);
 114   char * prefix_check_morph(const char * word, int len,
 115                     char in_compound, const FLAG needflag = FLAG_NULL);
 116   char * suffix_check_morph (const char * word, int len, int sfxopts, AffEntry * ppfx,
 117             const FLAG cclass = FLAG_NULL, const FLAG needflag = FLAG_NULL, char in_compound = IN_CPD_NOT);
 118
 119   char * prefix_check_twosfx_morph(const char * word, int len,
 120             char in_compound, const FLAG needflag = FLAG_NULL);
 121   char * suffix_check_twosfx_morph(const char * word, int len,
 122             int sfxopts, AffEntry * ppfx, const FLAG needflag = FLAG_NULL);
 123
 124   int                 expand_rootword(struct guessword * wlst, int maxn, const char * ts,
 125                         int wl, const unsigned short * ap, unsigned short al, char * bad, int);
 126
 127   short               get_syllable (const char * word, int wlen);
 128   int                 cpdrep_check(const char * word, int len);
 129   int                 cpdpat_check(const char * word, int len);
 130   int                 defcpd_check(hentry *** words, short wnum, hentry * rv, hentry ** rwords, char all);
 131   int                 cpdcase_check(const char * word, int len);
 132   inline int                 candidate_check(const char * word, int len);
 133   struct hentry *     compound_check(const char * word, int len,
 134                               short wordnum, short numsyllable, short maxwordnum, short wnum, hentry ** words,
 135                               char hu_mov_rule, int * cmpdstemnum, int * cmpdstem, char is_sug);
 136
 137   int compound_check_morph(const char * word, int len,
 138                               short wordnum, short numsyllable, short maxwordnum, short wnum, hentry ** words,
 139                               char hu_mov_rule, char ** result, char * partresult);
 140
 141   struct hentry *     lookup(const char * word);
 142   int                 get_numrep();
 143   struct replentry *  get_reptable();
 144   int                 get_nummap();
 145   struct mapentry *   get_maptable();
 146   int                 get_numbreak();
 147   char **             get_breaktable();
 148   char *              get_encoding();
 149   int                 get_langnum();
 150   char *              get_try_string();
 151   const char *        get_wordchars();
 152   unsigned short *    get_wordchars_utf16(int * len);
 153   char *              get_ignore();
 154   unsigned short *    get_ignore_utf16(int * len);
 155   int                 get_compound();
 156   FLAG                get_compoundflag();
 157   FLAG                get_compoundbegin();
 158   FLAG                get_forbiddenword();
 159   FLAG                get_nosuggest();
 160 //  FLAG                get_circumfix();
 161   FLAG                get_pseudoroot();
 162   FLAG                get_onlyincompound();
 163   FLAG                get_compoundroot();
 164   FLAG                get_lemma_present();
 165   int                 get_checknum();
 166   char *              get_possible_root();
 167   const char *        get_prefix();
 168   const char *        get_suffix();
 169   const char *        get_derived();
 170   const char *        get_version();
 171   const int           have_contclass();
 172   int                 get_utf8();
 173   int                 get_complexprefixes();
 174   char *              get_suffixed(char );
 175   int                 get_maxngramsugs();
 176   int                 get_nosplitsugs();
 177   int                 get_sugswithdots(void);
 178   FLAG                get_keepcase(void);
 179   int                 get_checksharps(void);
 180
 181 private:
 182   int  parse_file(const char * affpath);
 183 //  int  parse_string(char * line, char ** out, const char * name);
 184   int  parse_flag(char * line, unsigned short * out, const char * name);
 185   int  parse_num(char * line, int * out, const char * name);
 186 //  int  parse_array(char * line, char ** out, unsigned short ** out_utf16,
 187 //            int * out_utf16_len, const char * name);
 188   int  parse_cpdsyllable(char * line);
 189   int  parse_reptable(char * line, FILE * af);
 190   int  parse_maptable(char * line, FILE * af);
 191   int  parse_breaktable(char * line, FILE * af);
 192   int  parse_checkcpdtable(char * line, FILE * af);
 193   int  parse_defcpdtable(char * line, FILE * af);
 194   int  parse_affix(char * line, const char at, FILE * af, char * dupflags);
 195
 196   int encodeit(struct affentry * ptr, char * cs);
 197   int build_pfxtree(AffEntry* pfxptr);
 198   int build_sfxtree(AffEntry* sfxptr);
 199   int process_pfx_order();
 200   int process_sfx_order();
 201   AffEntry * process_pfx_in_order(AffEntry * ptr, AffEntry * nptr);
 202   AffEntry * process_sfx_in_order(AffEntry * ptr, AffEntry * nptr);
 203   int process_pfx_tree_to_list();
 204   int process_sfx_tree_to_list();
 205   int redundant_condition(char, char * strip, int stripl, const char * cond, char *);
 206 };
 207
 208 #endif
 209