src/include/tsearch/ts_utils.h

   1 /*-------------------------------------------------------------------------
   2  *
   3  * ts_utils.h
   4  *        helper utilities for tsearch
   5  *
   6  * Copyright (c) 1998-2009, PostgreSQL Global Development Group
   7  *
   8  * $PostgreSQL$
   9  *
  10  *-------------------------------------------------------------------------
  11  */
  12 #ifndef _PG_TS_UTILS_H_
  13 #define _PG_TS_UTILS_H_
  14
  15 #include "tsearch/ts_type.h"
  16 #include "tsearch/ts_public.h"
  17 #include "nodes/pg_list.h"
  18
  19 /*
  20  * Common parse definitions for tsvector and tsquery
  21  */
  22
  23 /* tsvector parser support. */
  24
  25 struct TSVectorParseStateData;  /* opaque struct in tsvector_parser.c */
  26 typedef struct TSVectorParseStateData *TSVectorParseState;
  27
  28 extern TSVectorParseState init_tsvector_parser(char *input,
  29                                          bool oprisdelim,
  30                                          bool is_tsquery);
  31 extern void reset_tsvector_parser(TSVectorParseState state, char *input);
  32 extern bool gettoken_tsvector(TSVectorParseState state,
  33                                   char **token, int *len,
  34                                   WordEntryPos **pos, int *poslen,
  35                                   char **endptr);
  36 extern void close_tsvector_parser(TSVectorParseState state);
  37
  38 /* parse_tsquery */
  39
  40 struct TSQueryParserStateData;  /* private in backend/utils/adt/tsquery.c */
  41 typedef struct TSQueryParserStateData *TSQueryParserState;
  42
  43 typedef void (*PushFunction) (Datum opaque, TSQueryParserState state,
  44                                                                                   char *token, int tokenlen,
  45                                                                                   int2 tokenweights,    /* bitmap as described
  46                                                                                                                                  * in QueryOperand
  47                                                                                                                                  * struct */
  48                                                                                   bool prefix);
  49
  50 extern TSQuery parse_tsquery(char *buf,
  51                           PushFunction pushval,
  52                           Datum opaque, bool isplain);
  53
  54 /* Functions for use by PushFunction implementations */
  55 extern void pushValue(TSQueryParserState state,
  56                   char *strval, int lenval, int2 weight, bool prefix);
  57 extern void pushStop(TSQueryParserState state);
  58 extern void pushOperator(TSQueryParserState state, int8 operator);
  59
  60 /*
  61  * parse plain text and lexize words
  62  */
  63 typedef struct
  64 {
  65         uint16          len;
  66         uint16          nvariant;
  67         union
  68         {
  69                 uint16          pos;
  70
  71                 /*
  72                  * When apos array is used, apos[0] is the number of elements in the
  73                  * array (excluding apos[0]), and alen is the allocated size of the
  74                  * array.
  75                  */
  76                 uint16     *apos;
  77         }                       pos;
  78         uint16          flags;                  /* currently, only TSL_PREFIX */
  79         char       *word;
  80         uint32          alen;
  81 } ParsedWord;
  82
  83 typedef struct
  84 {
  85         ParsedWord *words;
  86         int4            lenwords;
  87         int4            curwords;
  88         int4            pos;
  89 } ParsedText;
  90
  91 extern void parsetext(Oid cfgId, ParsedText *prs, char *buf, int4 buflen);
  92
  93 /*
  94  * headline framework, flow in common to generate:
  95  *      1 parse text with hlparsetext
  96  *      2 parser-specific function to find part
  97  *      3 generateHeadline to generate result text
  98  */
  99
 100 extern void hlparsetext(Oid cfgId, HeadlineParsedText *prs, TSQuery query,
 101                         char *buf, int4 buflen);
 102 extern text *generateHeadline(HeadlineParsedText *prs);
 103
 104 /*
 105  * Common check function for tsvector @@ tsquery
 106  */
 107
 108 extern bool TS_execute(QueryItem *curitem, void *checkval, bool calcnot,
 109                    bool (*chkcond) (void *checkval, QueryOperand *val));
 110
 111 /*
 112  * to_ts* - text transformation to tsvector, tsquery
 113  */
 114 extern TSVector make_tsvector(ParsedText *prs);
 115 extern int32 tsCompareString(char *a, int lena, char *b, int lenb, bool prefix);
 116
 117 extern Datum to_tsvector_byid(PG_FUNCTION_ARGS);
 118 extern Datum to_tsvector(PG_FUNCTION_ARGS);
 119 extern Datum to_tsquery_byid(PG_FUNCTION_ARGS);
 120 extern Datum to_tsquery(PG_FUNCTION_ARGS);
 121 extern Datum plainto_tsquery_byid(PG_FUNCTION_ARGS);
 122 extern Datum plainto_tsquery(PG_FUNCTION_ARGS);
 123
 124 /*
 125  * GiST support function
 126  */
 127
 128 extern Datum gtsvector_compress(PG_FUNCTION_ARGS);
 129 extern Datum gtsvector_decompress(PG_FUNCTION_ARGS);
 130 extern Datum gtsvector_consistent(PG_FUNCTION_ARGS);
 131 extern Datum gtsvector_union(PG_FUNCTION_ARGS);
 132 extern Datum gtsvector_same(PG_FUNCTION_ARGS);
 133 extern Datum gtsvector_penalty(PG_FUNCTION_ARGS);
 134 extern Datum gtsvector_picksplit(PG_FUNCTION_ARGS);
 135
 136 /*
 137  * IO functions for pseudotype gtsvector
 138  * used internally in tsvector GiST opclass
 139  */
 140 extern Datum gtsvectorin(PG_FUNCTION_ARGS);
 141 extern Datum gtsvectorout(PG_FUNCTION_ARGS);
 142
 143 /*
 144  * GIN support function
 145  */
 146
 147 extern Datum gin_extract_tsvector(PG_FUNCTION_ARGS);
 148 extern Datum gin_cmp_tslexeme(PG_FUNCTION_ARGS);
 149 extern Datum gin_cmp_prefix(PG_FUNCTION_ARGS);
 150 extern Datum gin_extract_tsquery(PG_FUNCTION_ARGS);
 151 extern Datum gin_tsquery_consistent(PG_FUNCTION_ARGS);
 152
 153 /*
 154  * Possible strategy numbers for indexes
 155  *        TSearchStrategyNumber  - (tsvector|text) @@ tsquery
 156  *        TSearchWithClassStrategyNumber  - tsvector @@@ tsquery
 157  */
 158 #define TSearchStrategyNumber                   1
 159 #define TSearchWithClassStrategyNumber  2
 160
 161 /*
 162  * TSQuery Utilities
 163  */
 164 extern QueryItem *clean_NOT(QueryItem *ptr, int4 *len);
 165 extern QueryItem *clean_fakeval(QueryItem *ptr, int4 *len);
 166
 167 typedef struct QTNode
 168 {
 169         QueryItem  *valnode;
 170         uint32          flags;
 171         int4            nchild;
 172         char       *word;
 173         uint32          sign;
 174         struct QTNode **child;
 175 } QTNode;
 176
 177 /* bits in QTNode.flags */
 178 #define QTN_NEEDFREE    0x01
 179 #define QTN_NOCHANGE    0x02
 180 #define QTN_WORDFREE    0x04
 181
 182 typedef uint64 TSQuerySign;
 183
 184 #define TSQS_SIGLEN  (sizeof(TSQuerySign)*BITS_PER_BYTE)
 185
 186 #define TSQuerySignGetDatum(X)          Int64GetDatum((int64) (X))
 187 #define DatumGetTSQuerySign(X)          ((TSQuerySign) DatumGetInt64(X))
 188 #define PG_RETURN_TSQUERYSIGN(X)        return TSQuerySignGetDatum(X)
 189 #define PG_GETARG_TSQUERYSIGN(n)        DatumGetTSQuerySign(PG_GETARG_DATUM(n))
 190
 191
 192 extern QTNode *QT2QTN(QueryItem *in, char *operand);
 193 extern TSQuery QTN2QT(QTNode *in);
 194 extern void QTNFree(QTNode *in);
 195 extern void QTNSort(QTNode *in);
 196 extern void QTNTernary(QTNode *in);
 197 extern void QTNBinary(QTNode *in);
 198 extern int      QTNodeCompare(QTNode *an, QTNode *bn);
 199 extern QTNode *QTNCopy(QTNode *in);
 200 extern void QTNClearFlags(QTNode *in, uint32 flags);
 201 extern bool QTNEq(QTNode *a, QTNode *b);
 202 extern TSQuerySign makeTSQuerySign(TSQuery a);
 203 extern QTNode *findsubquery(QTNode *root, QTNode *ex, QTNode *subs,
 204                          bool *isfind);
 205
 206 /*
 207  * TSQuery GiST support
 208  */
 209 extern Datum gtsquery_compress(PG_FUNCTION_ARGS);
 210 extern Datum gtsquery_decompress(PG_FUNCTION_ARGS);
 211 extern Datum gtsquery_consistent(PG_FUNCTION_ARGS);
 212 extern Datum gtsquery_union(PG_FUNCTION_ARGS);
 213 extern Datum gtsquery_same(PG_FUNCTION_ARGS);
 214 extern Datum gtsquery_penalty(PG_FUNCTION_ARGS);
 215 extern Datum gtsquery_picksplit(PG_FUNCTION_ARGS);
 216
 217 /*
 218  * Parser interface to SQL
 219  */
 220 extern Datum ts_token_type_byid(PG_FUNCTION_ARGS);
 221 extern Datum ts_token_type_byname(PG_FUNCTION_ARGS);
 222 extern Datum ts_parse_byid(PG_FUNCTION_ARGS);
 223 extern Datum ts_parse_byname(PG_FUNCTION_ARGS);
 224
 225 /*
 226  * Default word parser
 227  */
 228
 229 extern Datum prsd_start(PG_FUNCTION_ARGS);
 230 extern Datum prsd_nexttoken(PG_FUNCTION_ARGS);
 231 extern Datum prsd_end(PG_FUNCTION_ARGS);
 232 extern Datum prsd_headline(PG_FUNCTION_ARGS);
 233 extern Datum prsd_lextype(PG_FUNCTION_ARGS);
 234
 235 /*
 236  * Dictionary interface to SQL
 237  */
 238 extern Datum ts_lexize(PG_FUNCTION_ARGS);
 239
 240 /*
 241  * Simple built-in dictionary
 242  */
 243 extern Datum dsimple_init(PG_FUNCTION_ARGS);
 244 extern Datum dsimple_lexize(PG_FUNCTION_ARGS);
 245
 246 /*
 247  * Synonym built-in dictionary
 248  */
 249 extern Datum dsynonym_init(PG_FUNCTION_ARGS);
 250 extern Datum dsynonym_lexize(PG_FUNCTION_ARGS);
 251
 252 /*
 253  * ISpell dictionary
 254  */
 255 extern Datum dispell_init(PG_FUNCTION_ARGS);
 256 extern Datum dispell_lexize(PG_FUNCTION_ARGS);
 257
 258 /*
 259  * Thesaurus
 260  */
 261 extern Datum thesaurus_init(PG_FUNCTION_ARGS);
 262 extern Datum thesaurus_lexize(PG_FUNCTION_ARGS);
 263
 264 /*
 265  * headline
 266  */
 267 extern Datum ts_headline_byid_opt(PG_FUNCTION_ARGS);
 268 extern Datum ts_headline_byid(PG_FUNCTION_ARGS);
 269 extern Datum ts_headline(PG_FUNCTION_ARGS);
 270 extern Datum ts_headline_opt(PG_FUNCTION_ARGS);
 271
 272 /*
 273  * current cfg
 274  */
 275 extern Datum get_current_ts_config(PG_FUNCTION_ARGS);
 276
 277 #endif   /* _PG_TS_UTILS_H_ */