src/include/tsearch/dicts/spell.h

   1 /*-------------------------------------------------------------------------
   2  *
   3  * spell.h
   4  *
   5  * Declarations for ISpell dictionary
   6  *
   7  * Portions Copyright (c) 1996-2009, PostgreSQL Global Development Group
   8  *
   9  * $PostgreSQL$
  10  *
  11  *-------------------------------------------------------------------------
  12  */
  13
  14 #ifndef __SPELL_H__
  15 #define __SPELL_H__
  16
  17 #include "regex/regex.h"
  18 #include "tsearch/dicts/regis.h"
  19 #include "tsearch/ts_public.h"
  20
  21 /*
  22  * Max length of a flag name. Names longer than this will be truncated
  23  * to the maximum.
  24  */
  25 #define MAXFLAGLEN 16
  26
  27 struct SPNode;
  28
  29 typedef struct
  30 {
  31         uint32          val:8,
  32                                 isword:1,
  33                                 compoundflag:4,
  34                                 affix:19;
  35         struct SPNode *node;
  36 } SPNodeData;
  37
  38 /*
  39  * Names of FF_ are correlated with Hunspell options in affix file
  40  * http://sourceforge.net/docman/display_doc.php?docid=29374&group_id=143754
  41  */
  42 #define FF_COMPOUNDONLY         0x01
  43 #define FF_COMPOUNDBEGIN        0x02
  44 #define FF_COMPOUNDMIDDLE       0x04
  45 #define FF_COMPOUNDLAST         0x08
  46 #define FF_COMPOUNDFLAG         ( FF_COMPOUNDBEGIN | FF_COMPOUNDMIDDLE | FF_COMPOUNDLAST )
  47 #define FF_DICTFLAGMASK         0x0f
  48
  49 typedef struct SPNode
  50 {
  51         uint32          length;
  52         SPNodeData      data[1];
  53 } SPNode;
  54
  55 #define SPNHDRSZ        (offsetof(SPNode,data))
  56
  57
  58 typedef struct spell_struct
  59 {
  60         union
  61         {
  62                 /*
  63                  * flag is filled in by NIImportDictionary. After NISortDictionary, d
  64                  * is valid and flag is invalid.
  65                  */
  66                 char            flag[MAXFLAGLEN];
  67                 struct
  68                 {
  69                         int                     affix;
  70                         int                     len;
  71                 }                       d;
  72         }                       p;
  73         char            word[1];                /* variable length, null-terminated */
  74 } SPELL;
  75
  76 #define SPELLHDRSZ      (offsetof(SPELL, word))
  77
  78 typedef struct aff_struct
  79 {
  80         uint32          flag:8,
  81                                 type:1,
  82                                 flagflags:7,
  83                                 issimple:1,
  84                                 isregis:1,
  85                                 replen:14;
  86         char       *find;
  87         char       *repl;
  88         union
  89         {
  90                 regex_t         regex;
  91                 Regis           regis;
  92         }                       reg;
  93 } AFFIX;
  94
  95 /*
  96  * affixes use dictionary flags too
  97  */
  98 #define FF_COMPOUNDPERMITFLAG   0x10
  99 #define FF_COMPOUNDFORBIDFLAG   0x20
 100 #define FF_CROSSPRODUCT                 0x40
 101
 102 /*
 103  * Don't change the order of these. Initialization sorts by these,
 104  * and expects prefixes to come first after sorting.
 105  */
 106 #define FF_SUFFIX                               1
 107 #define FF_PREFIX                               0
 108
 109 struct AffixNode;
 110
 111 typedef struct
 112 {
 113         uint32          val:8,
 114                                 naff:24;
 115         AFFIX     **aff;
 116         struct AffixNode *node;
 117 } AffixNodeData;
 118
 119 typedef struct AffixNode
 120 {
 121         uint32          isvoid:1,
 122                                 length:31;
 123         AffixNodeData data[1];
 124 } AffixNode;
 125
 126 #define ANHRDSZ            (offsetof(AffixNode, data))
 127
 128 typedef struct
 129 {
 130         char       *affix;
 131         int                     len;
 132         bool            issuffix;
 133 } CMPDAffix;
 134
 135 typedef struct
 136 {
 137         int                     maffixes;
 138         int                     naffixes;
 139         AFFIX      *Affix;
 140
 141         /*
 142          * Temporary array of all words in the dict file. Only used during
 143          * initialization
 144          */
 145         SPELL     **Spell;
 146         int                     nspell;                 /* number of valid entries in Spell array */
 147         int                     mspell;                 /* allocated length of Spell array */
 148
 149         AffixNode  *Suffix;
 150         AffixNode  *Prefix;
 151
 152         SPNode     *Dictionary;
 153         char      **AffixData;
 154         int                     lenAffixData;
 155         int                     nAffixData;
 156
 157         CMPDAffix  *CompoundAffix;
 158
 159         unsigned char flagval[256];
 160         bool            usecompound;
 161 } IspellDict;
 162
 163 extern TSLexeme *NINormalizeWord(IspellDict *Conf, char *word);
 164 extern void NIImportAffixes(IspellDict *Conf, const char *filename);
 165 extern void NIImportDictionary(IspellDict *Conf, const char *filename);
 166 extern void NISortDictionary(IspellDict *Conf);
 167 extern void NISortAffixes(IspellDict *Conf);
 168
 169 #endif