ext/hunspell/affixmgr.hxx

   1 #ifndef _AFFIXMGR_HXX_
   2 #define _AFFIXMGR_HXX_
   3
   4 #include "hunvisapi.h"
   5
   6 #include <stdio.h>
   7
   8 #include "atypes.hxx"
   9 #include "baseaffix.hxx"
  10 #include "hashmgr.hxx"
  11 #include "phonet.hxx"
  12 #include "replist.hxx"
  13
  14 // check flag duplication
  15 #define dupSFX        (1 << 0)
  16 #define dupPFX        (1 << 1)
  17
  18 class PfxEntry;
  19 class SfxEntry;
  20
  21 class LIBHUNSPELL_DLL_EXPORTED AffixMgr
  22 {
  23
  24   PfxEntry *          pStart[SETSIZE];
  25   SfxEntry *          sStart[SETSIZE];
  26   PfxEntry *          pFlag[SETSIZE];
  27   SfxEntry *          sFlag[SETSIZE];
  28   HashMgr *           pHMgr;
  29   HashMgr **          alldic;
  30   int *               maxdic;
  31   char *              keystring;
  32   char *              trystring;
  33   char *              encoding;
  34   struct cs_info *    csconv;
  35   int                 utf8;
  36   int                 complexprefixes;
  37   FLAG                compoundflag;
  38   FLAG                compoundbegin;
  39   FLAG                compoundmiddle;
  40   FLAG                compoundend;
  41   FLAG                compoundroot;
  42   FLAG                compoundforbidflag;
  43   FLAG                compoundpermitflag;
  44   int                 checkcompounddup;
  45   int                 checkcompoundrep;
  46   int                 checkcompoundcase;
  47   int                 checkcompoundtriple;
  48   int                 simplifiedtriple;
  49   FLAG                forbiddenword;
  50   FLAG                nosuggest;
  51   FLAG                nongramsuggest;
  52   FLAG                needaffix;
  53   int                 cpdmin;
  54   int                 numrep;
  55   replentry *         reptable;
  56   RepList *           iconvtable;
  57   RepList *           oconvtable;
  58   int                 nummap;
  59   mapentry *          maptable;
  60   int                 numbreak;
  61   char **             breaktable;
  62   int                 numcheckcpd;
  63   patentry *          checkcpdtable;
  64   int                 simplifiedcpd;
  65   int                 numdefcpd;
  66   flagentry *         defcpdtable;
  67   phonetable *        phone;
  68   int                 maxngramsugs;
  69   int                 maxcpdsugs;
  70   int                 maxdiff;
  71   int                 onlymaxdiff;
  72   int                 nosplitsugs;
  73   int                 sugswithdots;
  74   int                 cpdwordmax;
  75   int                 cpdmaxsyllable;
  76   char *              cpdvowels;
  77   w_char *            cpdvowels_utf16;
  78   int                 cpdvowels_utf16_len;
  79   char *              cpdsyllablenum;
  80   const char *        pfxappnd; // BUG: not stateless
  81   const char *        sfxappnd; // BUG: not stateless
  82   FLAG                sfxflag;  // BUG: not stateless
  83   char *              derived;  // BUG: not stateless
  84   SfxEntry *          sfx;      // BUG: not stateless
  85   PfxEntry *          pfx;      // BUG: not stateless
  86   int                 checknum;
  87   char *              wordchars;
  88   unsigned short *    wordchars_utf16;
  89   int                 wordchars_utf16_len;
  90   char *              ignorechars;
  91   unsigned short *    ignorechars_utf16;
  92   int                 ignorechars_utf16_len;
  93   char *              version;
  94   char *              lang;
  95   int                 langnum;
  96   FLAG                lemma_present;
  97   FLAG                circumfix;
  98   FLAG                onlyincompound;
  99   FLAG                keepcase;
 100   FLAG                forceucase;
 101   FLAG                warn;
 102   int                 forbidwarn;
 103   FLAG                substandard;
 104   int                 checksharps;
 105   int                 fullstrip;
 106
 107   int                 havecontclass; // boolean variable
 108   char                contclasses[CONTSIZE]; // flags of possible continuing classes (twofold affix)
 109
 110 public:
 111
 112   AffixMgr(const char * affpath, HashMgr** ptr, int * md,
 113     const char * key = NULL);
 114   ~AffixMgr();
 115   struct hentry *     affix_check(const char * word, int len,
 116             const unsigned short needflag = (unsigned short) 0,
 117             char in_compound = IN_CPD_NOT);
 118   struct hentry *     prefix_check(const char * word, int len,
 119             char in_compound, const FLAG needflag = FLAG_NULL);
 120   inline int isSubset(const char * s1, const char * s2);
 121   struct hentry *     prefix_check_twosfx(const char * word, int len,
 122             char in_compound, const FLAG needflag = FLAG_NULL);
 123   inline int isRevSubset(const char * s1, const char * end_of_s2, int len);
 124   struct hentry *     suffix_check(const char * word, int len, int sfxopts,
 125             PfxEntry* ppfx, char ** wlst, int maxSug, int * ns,
 126             const FLAG cclass = FLAG_NULL, const FLAG needflag = FLAG_NULL,
 127             char in_compound = IN_CPD_NOT);
 128   struct hentry *     suffix_check_twosfx(const char * word, int len,
 129             int sfxopts, PfxEntry* ppfx, const FLAG needflag = FLAG_NULL);
 130
 131   char * affix_check_morph(const char * word, int len,
 132             const FLAG needflag = FLAG_NULL, char in_compound = IN_CPD_NOT);
 133   char * prefix_check_morph(const char * word, int len,
 134             char in_compound, const FLAG needflag = FLAG_NULL);
 135   char * suffix_check_morph (const char * word, int len, int sfxopts,
 136             PfxEntry * ppfx, const FLAG cclass = FLAG_NULL,
 137             const FLAG needflag = FLAG_NULL, char in_compound = IN_CPD_NOT);
 138
 139   char * prefix_check_twosfx_morph(const char * word, int len,
 140             char in_compound, const FLAG needflag = FLAG_NULL);
 141   char * suffix_check_twosfx_morph(const char * word, int len,
 142             int sfxopts, PfxEntry * ppfx, const FLAG needflag = FLAG_NULL);
 143
 144   char * morphgen(char * ts, int wl, const unsigned short * ap,
 145             unsigned short al, char * morph, char * targetmorph, int level);
 146
 147   int    expand_rootword(struct guessword * wlst, int maxn, const char * ts,
 148             int wl, const unsigned short * ap, unsigned short al, char * bad,
 149             int, char *);
 150
 151   short       get_syllable (const char * word, int wlen);
 152   int         cpdrep_check(const char * word, int len);
 153   int         cpdpat_check(const char * word, int len, hentry * r1, hentry * r2,
 154                     const char affixed);
 155   int         defcpd_check(hentry *** words, short wnum, hentry * rv,
 156                     hentry ** rwords, char all);
 157   int         cpdcase_check(const char * word, int len);
 158   inline int  candidate_check(const char * word, int len);
 159   void        setcminmax(int * cmin, int * cmax, const char * word, int len);
 160   struct hentry * compound_check(const char * word, int len, short wordnum,
 161             short numsyllable, short maxwordnum, short wnum, hentry ** words,
 162             char hu_mov_rule, char is_sug, int * info);
 163
 164   int compound_check_morph(const char * word, int len, short wordnum,
 165             short numsyllable, short maxwordnum, short wnum, hentry ** words,
 166             char hu_mov_rule, char ** result, char * partresult);
 167
 168   struct hentry * lookup(const char * word);
 169   int                 get_numrep() const;
 170   struct replentry *  get_reptable() const;
 171   RepList *           get_iconvtable() const;
 172   RepList *           get_oconvtable() const;
 173   struct phonetable * get_phonetable() const;
 174   int                 get_nummap() const;
 175   struct mapentry *   get_maptable() const;
 176   int                 get_numbreak() const;
 177   char **             get_breaktable() const;
 178   char *              get_encoding();
 179   int                 get_langnum() const;
 180   char *              get_key_string();
 181   char *              get_try_string() const;
 182   const char *        get_wordchars() const;
 183   unsigned short *    get_wordchars_utf16(int * len) const;
 184   char *              get_ignore() const;
 185   unsigned short *    get_ignore_utf16(int * len) const;
 186   int                 get_compound() const;
 187   FLAG                get_compoundflag() const;
 188   FLAG                get_compoundbegin() const;
 189   FLAG                get_forbiddenword() const;
 190   FLAG                get_nosuggest() const;
 191   FLAG                get_nongramsuggest() const;
 192   FLAG                get_needaffix() const;
 193   FLAG                get_onlyincompound() const;
 194   FLAG                get_compoundroot() const;
 195   FLAG                get_lemma_present() const;
 196   int                 get_checknum() const;
 197   const char *        get_prefix() const;
 198   const char *        get_suffix() const;
 199   const char *        get_derived() const;
 200   const char *        get_version() const;
 201   int                 have_contclass() const;
 202   int                 get_utf8() const;
 203   int                 get_complexprefixes() const;
 204   char *              get_suffixed(char ) const;
 205   int                 get_maxngramsugs() const;
 206   int                 get_maxcpdsugs() const;
 207   int                 get_maxdiff() const;
 208   int                 get_onlymaxdiff() const;
 209   int                 get_nosplitsugs() const;
 210   int                 get_sugswithdots(void) const;
 211   FLAG                get_keepcase(void) const;
 212   FLAG                get_forceucase(void) const;
 213   FLAG                get_warn(void) const;
 214   int                 get_forbidwarn(void) const;
 215   int                 get_checksharps(void) const;
 216   char *              encode_flag(unsigned short aflag) const;
 217   int                 get_fullstrip() const;
 218
 219 private:
 220   int  parse_file(const char * affpath, const char * key);
 221   int  parse_flag(char * line, unsigned short * out, FileMgr * af);
 222   int  parse_num(char * line, int * out, FileMgr * af);
 223   int  parse_cpdsyllable(char * line, FileMgr * af);
 224   int  parse_reptable(char * line, FileMgr * af);
 225   int  parse_convtable(char * line, FileMgr * af, RepList ** rl, const char * keyword);
 226   int  parse_phonetable(char * line, FileMgr * af);
 227   int  parse_maptable(char * line, FileMgr * af);
 228   int  parse_breaktable(char * line, FileMgr * af);
 229   int  parse_checkcpdtable(char * line, FileMgr * af);
 230   int  parse_defcpdtable(char * line, FileMgr * af);
 231   int  parse_affix(char * line, const char at, FileMgr * af, char * dupflags);
 232
 233   void reverse_condition(char *);
 234   void debugflag(char * result, unsigned short flag);
 235   int condlen(char *);
 236   int encodeit(affentry &entry, char * cs);
 237   int build_pfxtree(PfxEntry* pfxptr);
 238   int build_sfxtree(SfxEntry* sfxptr);
 239   int process_pfx_order();
 240   int process_sfx_order();
 241   PfxEntry * process_pfx_in_order(PfxEntry * ptr, PfxEntry * nptr);
 242   SfxEntry * process_sfx_in_order(SfxEntry * ptr, SfxEntry * nptr);
 243   int process_pfx_tree_to_list();
 244   int process_sfx_tree_to_list();
 245   int redundant_condition(char, char * strip, int stripl,
 246       const char * cond, int);
 247 };
 248
 249 #endif
 250