third_party/hunspell_new/src/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 #ifdef HUNSPELL_CHROME_CLIENT
  22
  23 #include <vector>
  24
  25 // This class provides an implementation of the contclasses array in AffixMgr
  26 // that is normally a large static array. We should almost never need more than
  27 // 256 elements, so this class only allocates that much to start off with. If
  28 // elements higher than that are actually used, we'll automatically expand.
  29 class ContClasses {
  30  public:
  31   ContClasses() {
  32     // Pre-allocate a buffer so that typically, we'll never have to resize.
  33     EnsureSizeIs(256);
  34   }
  35
  36   char& operator[](size_t index) {
  37     EnsureSizeIs(index + 1);
  38     return data[index];
  39   }
  40
  41   void EnsureSizeIs(size_t new_size) {
  42     if (data.size() >= new_size)
  43       return;  // Nothing to do.
  44
  45     size_t old_size = data.size();
  46     data.resize(new_size);
  47     memset(&data[old_size], 0, new_size - old_size);
  48   }
  49
  50   std::vector<char> data;
  51 };
  52
  53 #endif  // HUNSPELL_CHROME_CLIENT
  54
  55 class LIBHUNSPELL_DLL_EXPORTED AffixMgr
  56 {
  57
  58   PfxEntry *          pStart[SETSIZE];
  59   SfxEntry *          sStart[SETSIZE];
  60   PfxEntry *          pFlag[SETSIZE];
  61   SfxEntry *          sFlag[SETSIZE];
  62   HashMgr *           pHMgr;
  63   HashMgr **          alldic;
  64   int *               maxdic;
  65   char *              keystring;
  66   char *              trystring;
  67   char *              encoding;
  68   struct cs_info *    csconv;
  69   int                 utf8;
  70   int                 complexprefixes;
  71   FLAG                compoundflag;
  72   FLAG                compoundbegin;
  73   FLAG                compoundmiddle;
  74   FLAG                compoundend;
  75   FLAG                compoundroot;
  76   FLAG                compoundforbidflag;
  77   FLAG                compoundpermitflag;
  78   int                 compoundmoresuffixes;
  79   int                 checkcompounddup;
  80   int                 checkcompoundrep;
  81   int                 checkcompoundcase;
  82   int                 checkcompoundtriple;
  83   int                 simplifiedtriple;
  84   FLAG                forbiddenword;
  85   FLAG                nosuggest;
  86   FLAG                nongramsuggest;
  87   FLAG                needaffix;
  88   int                 cpdmin;
  89   int                 numrep;
  90   replentry *         reptable;
  91   RepList *           iconvtable;
  92   RepList *           oconvtable;
  93   int                 nummap;
  94   mapentry *          maptable;
  95   int                 numbreak;
  96   char **             breaktable;
  97   int                 numcheckcpd;
  98   patentry *          checkcpdtable;
  99   int                 simplifiedcpd;
 100   int                 numdefcpd;
 101   flagentry *         defcpdtable;
 102   phonetable *        phone;
 103   int                 maxngramsugs;
 104   int                 maxcpdsugs;
 105   int                 maxdiff;
 106   int                 onlymaxdiff;
 107   int                 nosplitsugs;
 108   int                 sugswithdots;
 109   int                 cpdwordmax;
 110   int                 cpdmaxsyllable;
 111   char *              cpdvowels;
 112   w_char *            cpdvowels_utf16;
 113   int                 cpdvowels_utf16_len;
 114   char *              cpdsyllablenum;
 115   const char *        pfxappnd; // BUG: not stateless
 116   const char *        sfxappnd; // BUG: not stateless
 117   FLAG                sfxflag;  // BUG: not stateless
 118   char *              derived;  // BUG: not stateless
 119   SfxEntry *          sfx;      // BUG: not stateless
 120   PfxEntry *          pfx;      // BUG: not stateless
 121   int                 checknum;
 122   char *              wordchars;
 123   unsigned short *    wordchars_utf16;
 124   int                 wordchars_utf16_len;
 125   char *              ignorechars;
 126   unsigned short *    ignorechars_utf16;
 127   int                 ignorechars_utf16_len;
 128   char *              version;
 129   char *              lang;
 130   int                 langnum;
 131   FLAG                lemma_present;
 132   FLAG                circumfix;
 133   FLAG                onlyincompound;
 134   FLAG                keepcase;
 135   FLAG                forceucase;
 136   FLAG                warn;
 137   int                 forbidwarn;
 138   FLAG                substandard;
 139   int                 checksharps;
 140   int                 fullstrip;
 141
 142   int                 havecontclass; // boolean variable
 143 #ifdef HUNSPELL_CHROME_CLIENT
 144   ContClasses         contclasses;
 145 #else
 146   char                contclasses[CONTSIZE]; // flags of possible continuing classes (twofold affix)
 147 #endif
 148
 149 public:
 150
 151 #ifdef HUNSPELL_CHROME_CLIENT
 152   AffixMgr(hunspell::BDictReader* reader, HashMgr** ptr, int * md);
 153 #else
 154   AffixMgr(const char * affpath, HashMgr** ptr, int * md,
 155     const char * key = NULL);
 156 #endif
 157   ~AffixMgr();
 158   struct hentry *     affix_check(const char * word, int len,
 159             const unsigned short needflag = (unsigned short) 0,
 160             char in_compound = IN_CPD_NOT);
 161   struct hentry *     prefix_check(const char * word, int len,
 162             char in_compound, const FLAG needflag = FLAG_NULL);
 163   inline int isSubset(const char * s1, const char * s2);
 164   struct hentry *     prefix_check_twosfx(const char * word, int len,
 165             char in_compound, const FLAG needflag = FLAG_NULL);
 166   inline int isRevSubset(const char * s1, const char * end_of_s2, int len);
 167   struct hentry *     suffix_check(const char * word, int len, int sfxopts,
 168             PfxEntry* ppfx, char ** wlst, int maxSug, int * ns,
 169             const FLAG cclass = FLAG_NULL, const FLAG needflag = FLAG_NULL,
 170             char in_compound = IN_CPD_NOT);
 171   struct hentry *     suffix_check_twosfx(const char * word, int len,
 172             int sfxopts, PfxEntry* ppfx, const FLAG needflag = FLAG_NULL);
 173
 174   char * affix_check_morph(const char * word, int len,
 175             const FLAG needflag = FLAG_NULL, char in_compound = IN_CPD_NOT);
 176   char * prefix_check_morph(const char * word, int len,
 177             char in_compound, const FLAG needflag = FLAG_NULL);
 178   char * suffix_check_morph (const char * word, int len, int sfxopts,
 179             PfxEntry * ppfx, const FLAG cclass = FLAG_NULL,
 180             const FLAG needflag = FLAG_NULL, char in_compound = IN_CPD_NOT);
 181
 182   char * prefix_check_twosfx_morph(const char * word, int len,
 183             char in_compound, const FLAG needflag = FLAG_NULL);
 184   char * suffix_check_twosfx_morph(const char * word, int len,
 185             int sfxopts, PfxEntry * ppfx, const FLAG needflag = FLAG_NULL);
 186
 187   char * morphgen(char * ts, int wl, const unsigned short * ap,
 188             unsigned short al, char * morph, char * targetmorph, int level);
 189
 190   int    expand_rootword(struct guessword * wlst, int maxn, const char * ts,
 191             int wl, const unsigned short * ap, unsigned short al, char * bad,
 192             int, char *);
 193
 194   short       get_syllable (const char * word, int wlen);
 195   int         cpdrep_check(const char * word, int len);
 196   int         cpdpat_check(const char * word, int len, hentry * r1, hentry * r2,
 197                     const char affixed);
 198   int         defcpd_check(hentry *** words, short wnum, hentry * rv,
 199                     hentry ** rwords, char all);
 200   int         cpdcase_check(const char * word, int len);
 201   inline int  candidate_check(const char * word, int len);
 202   void        setcminmax(int * cmin, int * cmax, const char * word, int len);
 203   struct hentry * compound_check(const char * word, int len, short wordnum,
 204             short numsyllable, short maxwordnum, short wnum, hentry ** words,
 205             char hu_mov_rule, char is_sug, int * info);
 206
 207   int compound_check_morph(const char * word, int len, short wordnum,
 208             short numsyllable, short maxwordnum, short wnum, hentry ** words,
 209             char hu_mov_rule, char ** result, char * partresult);
 210
 211   struct hentry * lookup(const char * word);
 212   int                 get_numrep() const;
 213   struct replentry *  get_reptable() const;
 214   RepList *           get_iconvtable() const;
 215   RepList *           get_oconvtable() const;
 216   struct phonetable * get_phonetable() const;
 217   int                 get_nummap() const;
 218   struct mapentry *   get_maptable() const;
 219   int                 get_numbreak() const;
 220   char **             get_breaktable() const;
 221   char *              get_encoding();
 222   int                 get_langnum() const;
 223   char *              get_key_string();
 224   char *              get_try_string() const;
 225   const char *        get_wordchars() const;
 226   unsigned short *    get_wordchars_utf16(int * len) const;
 227   char *              get_ignore() const;
 228   unsigned short *    get_ignore_utf16(int * len) const;
 229   int                 get_compound() const;
 230   FLAG                get_compoundflag() const;
 231   FLAG                get_compoundbegin() const;
 232   FLAG                get_forbiddenword() const;
 233   FLAG                get_nosuggest() const;
 234   FLAG                get_nongramsuggest() const;
 235   FLAG                get_needaffix() const;
 236   FLAG                get_onlyincompound() const;
 237   FLAG                get_compoundroot() const;
 238   FLAG                get_lemma_present() const;
 239   int                 get_checknum() const;
 240   const char *        get_prefix() const;
 241   const char *        get_suffix() const;
 242   const char *        get_derived() const;
 243   const char *        get_version() const;
 244   int                 have_contclass() const;
 245   int                 get_utf8() const;
 246   int                 get_complexprefixes() const;
 247   char *              get_suffixed(char ) const;
 248   int                 get_maxngramsugs() const;
 249   int                 get_maxcpdsugs() const;
 250   int                 get_maxdiff() const;
 251   int                 get_onlymaxdiff() const;
 252   int                 get_nosplitsugs() const;
 253   int                 get_sugswithdots(void) const;
 254   FLAG                get_keepcase(void) const;
 255   FLAG                get_forceucase(void) const;
 256   FLAG                get_warn(void) const;
 257   int                 get_forbidwarn(void) const;
 258   int                 get_checksharps(void) const;
 259   char *              encode_flag(unsigned short aflag) const;
 260   int                 get_fullstrip() const;
 261
 262 private:
 263 #ifdef HUNSPELL_CHROME_CLIENT
 264   // Not owned by us, owned by the Hunspell object.
 265   hunspell::BDictReader* bdict_reader;
 266 #endif
 267   int  parse_file(const char * affpath, const char * key);
 268   int  parse_flag(char * line, unsigned short * out, FileMgr * af);
 269   int  parse_num(char * line, int * out, FileMgr * af);
 270   int  parse_cpdsyllable(char * line, FileMgr * af);
 271   int  parse_reptable(char * line, FileMgr * af);
 272   int  parse_convtable(char * line, FileMgr * af, RepList ** rl, const char * keyword);
 273   int  parse_phonetable(char * line, FileMgr * af);
 274   int  parse_maptable(char * line, FileMgr * af);
 275   int  parse_breaktable(char * line, FileMgr * af);
 276   int  parse_checkcpdtable(char * line, FileMgr * af);
 277   int  parse_defcpdtable(char * line, FileMgr * af);
 278   int  parse_affix(char * line, const char at, FileMgr * af, char * dupflags);
 279
 280   void reverse_condition(char *);
 281   void debugflag(char * result, unsigned short flag);
 282   int condlen(char *);
 283   int encodeit(affentry &entry, char * cs);
 284   int build_pfxtree(PfxEntry* pfxptr);
 285   int build_sfxtree(SfxEntry* sfxptr);
 286   int process_pfx_order();
 287   int process_sfx_order();
 288   PfxEntry * process_pfx_in_order(PfxEntry * ptr, PfxEntry * nptr);
 289   SfxEntry * process_sfx_in_order(SfxEntry * ptr, SfxEntry * nptr);
 290   int process_pfx_tree_to_list();
 291   int process_sfx_tree_to_list();
 292   int redundant_condition(char, char * strip, int stripl,
 293       const char * cond, int);
 294   void finishFileMgr(FileMgr *afflst);
 295 };
 296
 297 #endif