src/include/tsearch/ts_type.h

   1 /*-------------------------------------------------------------------------
   2  *
   3  * ts_type.h
   4  *        Definitions for the tsvector and tsquery types
   5  *
   6  * Copyright (c) 1998-2009, PostgreSQL Global Development Group
   7  *
   8  * $PostgreSQL$
   9  *
  10  *-------------------------------------------------------------------------
  11  */
  12 #ifndef _PG_TSTYPE_H_
  13 #define _PG_TSTYPE_H_
  14
  15 #include "fmgr.h"
  16 #include "utils/pg_crc.h"
  17
  18
  19 /*
  20  * TSVector type.
  21  * Note, tsvectorsend/recv believe that sizeof(WordEntry) == 4
  22  */
  23
  24 typedef struct
  25 {
  26         uint32
  27                                 haspos:1,
  28                                 len:11,                 /* MAX 2Kb */
  29                                 pos:20;                 /* MAX 1Mb */
  30 } WordEntry;
  31
  32 #define MAXSTRLEN ( (1<<11) - 1)
  33 #define MAXSTRPOS ( (1<<20) - 1)
  34
  35 /*
  36  * Equivalent to
  37  * typedef struct {
  38  *              uint16
  39  *                      weight:2,
  40  *                      pos:14;
  41  * }
  42  */
  43
  44 typedef uint16 WordEntryPos;
  45
  46 typedef struct
  47 {
  48         uint16          npos;
  49         WordEntryPos pos[1];            /* var length */
  50 } WordEntryPosVector;
  51
  52
  53 #define WEP_GETWEIGHT(x)        ( (x) >> 14 )
  54 #define WEP_GETPOS(x)           ( (x) & 0x3fff )
  55
  56 #define WEP_SETWEIGHT(x,v)      ( (x) = ( (v) << 14 ) | ( (x) & 0x3fff ) )
  57 #define WEP_SETPOS(x,v)         ( (x) = ( (x) & 0xc000 ) | ( (v) & 0x3fff ) )
  58
  59 #define MAXENTRYPOS (1<<14)
  60 #define MAXNUMPOS       (256)
  61 #define LIMITPOS(x) ( ( (x) >= MAXENTRYPOS ) ? (MAXENTRYPOS-1) : (x) )
  62
  63 /*
  64  * Structure of tsvector datatype:
  65  * 1) standard varlena header
  66  * 2) int4              size - number of lexemes or WordEntry array, which is the same
  67  * 3) Array of WordEntry - sorted array, comparison based on word's length
  68  *                                                      and strncmp(). WordEntry->pos points number of
  69  *                                                      bytes from end of WordEntry array to start of
  70  *                                                      corresponding lexeme.
  71  * 4) Lexeme's storage:
  72  *        lexeme (without null-terminator)
  73  *        if haspos is true:
  74  *              padding byte if necessary to make the number of positions 2-byte aligned
  75  *              uint16          number of positions that follow.
  76  *              uint16[]        positions
  77  *
  78  * The positions must be sorted.
  79  */
  80
  81 typedef struct
  82 {
  83         int32           vl_len_;                /* varlena header (do not touch directly!) */
  84         int32           size;
  85         WordEntry       entries[1];             /* var size */
  86         /* lexemes follow */
  87 } TSVectorData;
  88
  89 typedef TSVectorData *TSVector;
  90
  91 #define DATAHDRSIZE (offsetof(TSVectorData, entries))
  92 #define CALCDATASIZE(x, lenstr) (DATAHDRSIZE + (x) * sizeof(WordEntry) + (lenstr) )
  93 #define ARRPTR(x)       ( (x)->entries )
  94
  95 /* returns a pointer to the beginning of lexemes */
  96 #define STRPTR(x)       ( (char *) &(x)->entries[x->size] )
  97
  98 #define _POSVECPTR(x, e)        ((WordEntryPosVector *)(STRPTR(x) + SHORTALIGN((e)->pos + (e)->len)))
  99 #define POSDATALEN(x,e) ( ( (e)->haspos ) ? (_POSVECPTR(x,e)->npos) : 0 )
 100 #define POSDATAPTR(x,e) (_POSVECPTR(x,e)->pos)
 101
 102 /*
 103  * fmgr interface macros
 104  */
 105
 106 #define DatumGetTSVector(X)                     ((TSVector) PG_DETOAST_DATUM(X))
 107 #define DatumGetTSVectorCopy(X)         ((TSVector) PG_DETOAST_DATUM_COPY(X))
 108 #define TSVectorGetDatum(X)                     PointerGetDatum(X)
 109 #define PG_GETARG_TSVECTOR(n)           DatumGetTSVector(PG_GETARG_DATUM(n))
 110 #define PG_GETARG_TSVECTOR_COPY(n)      DatumGetTSVectorCopy(PG_GETARG_DATUM(n))
 111 #define PG_RETURN_TSVECTOR(x)           return TSVectorGetDatum(x)
 112
 113 /*
 114  * I/O
 115  */
 116 extern Datum tsvectorin(PG_FUNCTION_ARGS);
 117 extern Datum tsvectorout(PG_FUNCTION_ARGS);
 118 extern Datum tsvectorsend(PG_FUNCTION_ARGS);
 119 extern Datum tsvectorrecv(PG_FUNCTION_ARGS);
 120
 121 /*
 122  * operations with tsvector
 123  */
 124 extern Datum tsvector_lt(PG_FUNCTION_ARGS);
 125 extern Datum tsvector_le(PG_FUNCTION_ARGS);
 126 extern Datum tsvector_eq(PG_FUNCTION_ARGS);
 127 extern Datum tsvector_ne(PG_FUNCTION_ARGS);
 128 extern Datum tsvector_ge(PG_FUNCTION_ARGS);
 129 extern Datum tsvector_gt(PG_FUNCTION_ARGS);
 130 extern Datum tsvector_cmp(PG_FUNCTION_ARGS);
 131
 132 extern Datum tsvector_length(PG_FUNCTION_ARGS);
 133 extern Datum tsvector_strip(PG_FUNCTION_ARGS);
 134 extern Datum tsvector_setweight(PG_FUNCTION_ARGS);
 135 extern Datum tsvector_concat(PG_FUNCTION_ARGS);
 136 extern Datum tsvector_update_trigger_byid(PG_FUNCTION_ARGS);
 137 extern Datum tsvector_update_trigger_bycolumn(PG_FUNCTION_ARGS);
 138
 139 extern Datum ts_match_vq(PG_FUNCTION_ARGS);
 140 extern Datum ts_match_qv(PG_FUNCTION_ARGS);
 141 extern Datum ts_match_tt(PG_FUNCTION_ARGS);
 142 extern Datum ts_match_tq(PG_FUNCTION_ARGS);
 143
 144 extern Datum ts_stat1(PG_FUNCTION_ARGS);
 145 extern Datum ts_stat2(PG_FUNCTION_ARGS);
 146
 147 extern Datum ts_rank_tt(PG_FUNCTION_ARGS);
 148 extern Datum ts_rank_wtt(PG_FUNCTION_ARGS);
 149 extern Datum ts_rank_ttf(PG_FUNCTION_ARGS);
 150 extern Datum ts_rank_wttf(PG_FUNCTION_ARGS);
 151 extern Datum ts_rankcd_tt(PG_FUNCTION_ARGS);
 152 extern Datum ts_rankcd_wtt(PG_FUNCTION_ARGS);
 153 extern Datum ts_rankcd_ttf(PG_FUNCTION_ARGS);
 154 extern Datum ts_rankcd_wttf(PG_FUNCTION_ARGS);
 155
 156 extern Datum tsmatchsel(PG_FUNCTION_ARGS);
 157 extern Datum tsmatchjoinsel(PG_FUNCTION_ARGS);
 158
 159 extern Datum ts_typanalyze(PG_FUNCTION_ARGS);
 160
 161
 162 /*
 163  * TSQuery
 164  *
 165  *
 166  */
 167
 168 typedef int8 QueryItemType;
 169
 170 /* Valid values for QueryItemType: */
 171 #define QI_VAL 1
 172 #define QI_OPR 2
 173 #define QI_VALSTOP 3                    /* This is only used in an intermediate stack
 174                                                                  * representation in parse_tsquery. It's not a
 175                                                                  * legal type elsewhere. */
 176
 177 /*
 178  * QueryItem is one node in tsquery - operator or operand.
 179  */
 180 typedef struct
 181 {
 182         QueryItemType type;                     /* operand or kind of operator (ts_tokentype) */
 183         uint8           weight;                 /* weights of operand to search. It's a
 184                                                                  * bitmask of allowed weights. if it =0 then
 185                                                                  * any weight are allowed. Weights and bit
 186                                                                  * map: A: 1<<3 B: 1<<2 C: 1<<1 D: 1<<0 */
 187         bool            prefix;                 /* true if it's a prefix search */
 188         int32           valcrc;                 /* XXX: pg_crc32 would be a more appropriate
 189                                                                  * data type, but we use comparisons to signed
 190                                                                  * integers in the code. They would need to be
 191                                                                  * changed as well. */
 192
 193         /* pointer to text value of operand, must correlate with WordEntry */
 194         uint32
 195                                 length:12,
 196                                 distance:20;
 197 } QueryOperand;
 198
 199
 200 /* Legal values for QueryOperator.operator */
 201 #define OP_NOT  1
 202 #define OP_AND  2
 203 #define OP_OR   3
 204
 205 typedef struct
 206 {
 207         QueryItemType type;
 208         int8            oper;                   /* see above */
 209         uint32          left;                   /* pointer to left operand. Right operand is
 210                                                                  * item + 1, left operand is placed
 211                                                                  * item+item->left */
 212 } QueryOperator;
 213
 214 /*
 215  * Note: TSQuery is 4-bytes aligned, so make sure there's no fields
 216  * inside QueryItem requiring 8-byte alignment, like int64.
 217  */
 218 typedef union
 219 {
 220         QueryItemType type;
 221         QueryOperator operator;
 222         QueryOperand operand;
 223 } QueryItem;
 224
 225 /*
 226  * Storage:
 227  *      (len)(size)(array of QueryItem)(operands as '\0'-terminated c-strings)
 228  */
 229
 230 typedef struct
 231 {
 232         int32           vl_len_;                /* varlena header (do not touch directly!) */
 233         int4            size;                   /* number of QueryItems */
 234         char            data[1];
 235 } TSQueryData;
 236
 237 typedef TSQueryData *TSQuery;
 238
 239 #define HDRSIZETQ       ( VARHDRSZ + sizeof(int4) )
 240
 241 /* Computes the size of header and all QueryItems. size is the number of
 242  * QueryItems, and lenofoperand is the total length of all operands
 243  */
 244 #define COMPUTESIZE(size, lenofoperand) ( HDRSIZETQ + (size) * sizeof(QueryItem) + (lenofoperand) )
 245
 246 /* Returns a pointer to the first QueryItem in a TSQuery */
 247 #define GETQUERY(x)  ((QueryItem*)( (char*)(x)+HDRSIZETQ ))
 248
 249 /* Returns a pointer to the beginning of operands in a TSQuery */
 250 #define GETOPERAND(x)   ( (char*)GETQUERY(x) + ((TSQuery)(x))->size * sizeof(QueryItem) )
 251
 252 /*
 253  * fmgr interface macros
 254  * Note, TSQuery type marked as plain storage, so it can't be toasted
 255  * but PG_DETOAST_DATUM_COPY is used for simplicity
 256  */
 257
 258 #define DatumGetTSQuery(X)                      ((TSQuery) DatumGetPointer(X))
 259 #define DatumGetTSQueryCopy(X)          ((TSQuery) PG_DETOAST_DATUM_COPY(X))
 260 #define TSQueryGetDatum(X)                      PointerGetDatum(X)
 261 #define PG_GETARG_TSQUERY(n)            DatumGetTSQuery(PG_GETARG_DATUM(n))
 262 #define PG_GETARG_TSQUERY_COPY(n)       DatumGetTSQueryCopy(PG_GETARG_DATUM(n))
 263 #define PG_RETURN_TSQUERY(x)            return TSQueryGetDatum(x)
 264
 265 /*
 266  * I/O
 267  */
 268 extern Datum tsqueryin(PG_FUNCTION_ARGS);
 269 extern Datum tsqueryout(PG_FUNCTION_ARGS);
 270 extern Datum tsquerysend(PG_FUNCTION_ARGS);
 271 extern Datum tsqueryrecv(PG_FUNCTION_ARGS);
 272
 273 /*
 274  * operations with tsquery
 275  */
 276 extern Datum tsquery_lt(PG_FUNCTION_ARGS);
 277 extern Datum tsquery_le(PG_FUNCTION_ARGS);
 278 extern Datum tsquery_eq(PG_FUNCTION_ARGS);
 279 extern Datum tsquery_ne(PG_FUNCTION_ARGS);
 280 extern Datum tsquery_ge(PG_FUNCTION_ARGS);
 281 extern Datum tsquery_gt(PG_FUNCTION_ARGS);
 282 extern Datum tsquery_cmp(PG_FUNCTION_ARGS);
 283
 284 extern Datum tsquerytree(PG_FUNCTION_ARGS);
 285 extern Datum tsquery_numnode(PG_FUNCTION_ARGS);
 286
 287 extern Datum tsquery_and(PG_FUNCTION_ARGS);
 288 extern Datum tsquery_or(PG_FUNCTION_ARGS);
 289 extern Datum tsquery_not(PG_FUNCTION_ARGS);
 290
 291 extern Datum tsquery_rewrite(PG_FUNCTION_ARGS);
 292 extern Datum tsquery_rewrite_query(PG_FUNCTION_ARGS);
 293
 294 extern Datum tsq_mcontains(PG_FUNCTION_ARGS);
 295 extern Datum tsq_mcontained(PG_FUNCTION_ARGS);
 296
 297 #endif   /* _PG_TSTYPE_H_ */