granicus.if.org Git - postgresql/blob - src/include/tsearch/ts_utils.h

   1 /*-------------------------------------------------------------------------
   2  *
   3  * ts_utils.h
   4  *        helper utilities for tsearch
   5  *
   6  * Copyright (c) 1998-2007, PostgreSQL Global Development Group
   7  *
   8  * $PostgreSQL: pgsql/src/include/tsearch/ts_utils.h,v 1.4 2007/09/10 12:36:41 teodor Exp $
   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;
  26 typedef struct TSVectorParseStateData *TSVectorParseState;
  27
  28 extern TSVectorParseState init_tsvector_parser(char *input, bool oprisdelim);
  29 extern void reset_tsvector_parser(TSVectorParseState state, char *input);
  30 extern bool gettoken_tsvector(TSVectorParseState state,
  31                                                           char **token, int *len,
  32                                                           WordEntryPos **pos, int *poslen,
  33                                                           char **endptr);
  34 extern void close_tsvector_parser(TSVectorParseState state);
  35
  36 /* parse_tsquery */
  37
  38 struct TSQueryParserStateData;  /* private in backend/utils/adt/tsquery.c */
  39 typedef struct TSQueryParserStateData *TSQueryParserState;
  40
  41 typedef void (*PushFunction)(Datum opaque, TSQueryParserState state,
  42                                 char *token, int tokenlen,
  43                                 int2 tokenweights /* bitmap as described in QueryOperand struct */ );
  44
  45 extern TSQuery parse_tsquery(char *buf,
  46                           PushFunction pushval,
  47                           Datum opaque, bool isplain);
  48
  49 /* Functions for use by PushFunction implementations */
  50 extern void pushValue(TSQueryParserState state,
  51                          char *strval, int lenval, int2 weight);
  52 extern void pushStop(TSQueryParserState state);
  53 extern void pushOperator(TSQueryParserState state, int8 operator);
  54
  55 /*
  56  * parse plain text and lexize words
  57  */
  58 typedef struct
  59 {
  60         uint16          len;
  61         uint16          nvariant;
  62         union
  63         {
  64                 uint16          pos;
  65                 /*
  66                  * When apos array is used, apos[0] is the number of elements
  67                  * in the array (excluding apos[0]), and alen is the allocated
  68                  * size of the array.
  69                  */
  70                 uint16     *apos;
  71         }                       pos;
  72         char       *word;
  73         uint32          alen;
  74 } ParsedWord;
  75
  76 typedef struct
  77 {
  78         ParsedWord *words;
  79         int4            lenwords;
  80         int4            curwords;
  81         int4            pos;
  82 } ParsedText;
  83
  84 extern void parsetext(Oid cfgId, ParsedText * prs, char *buf, int4 buflen);
  85
  86 /*
  87  * headline framework, flow in common to generate:
  88  *      1 parse text with hlparsetext
  89  *      2 parser-specific function to find part
  90  *      3 generateHeadline to generate result text
  91  */
  92
  93 extern void hlparsetext(Oid cfgId, HeadlineParsedText * prs, TSQuery query,
  94                         char *buf, int4 buflen);
  95 extern text *generateHeadline(HeadlineParsedText * prs);
  96
  97 /*
  98  * Common check function for tsvector @@ tsquery
  99  */
 100
 101 extern bool TS_execute(QueryItem * curitem, void *checkval, bool calcnot,
 102                    bool (*chkcond) (void *checkval, QueryOperand * val));
 103
 104 /*
 105  * Useful conversion macros
 106  */
 107 #define TextPGetCString(t) DatumGetCString(DirectFunctionCall1(textout, PointerGetDatum(t)))
 108 #define CStringGetTextP(c) DatumGetTextP(DirectFunctionCall1(textin, CStringGetDatum(c)))
 109
 110 /*
 111  * to_ts* - text transformation to tsvector, tsquery
 112  */
 113 extern TSVector make_tsvector(ParsedText *prs);
 114
 115 extern Datum to_tsvector_byid(PG_FUNCTION_ARGS);
 116 extern Datum to_tsvector(PG_FUNCTION_ARGS);
 117 extern Datum to_tsquery_byid(PG_FUNCTION_ARGS);
 118 extern Datum to_tsquery(PG_FUNCTION_ARGS);
 119 extern Datum plainto_tsquery_byid(PG_FUNCTION_ARGS);
 120 extern Datum plainto_tsquery(PG_FUNCTION_ARGS);
 121
 122 /*
 123  * GiST support function
 124  */
 125
 126 extern Datum gtsvector_compress(PG_FUNCTION_ARGS);
 127 extern Datum gtsvector_decompress(PG_FUNCTION_ARGS);
 128 extern Datum gtsvector_consistent(PG_FUNCTION_ARGS);
 129 extern Datum gtsvector_union(PG_FUNCTION_ARGS);
 130 extern Datum gtsvector_same(PG_FUNCTION_ARGS);
 131 extern Datum gtsvector_penalty(PG_FUNCTION_ARGS);
 132 extern Datum gtsvector_picksplit(PG_FUNCTION_ARGS);
 133
 134 /*
 135  * IO functions for pseudotype gtsvector
 136  * used internally in tsvector GiST opclass
 137  */
 138 extern Datum gtsvectorin(PG_FUNCTION_ARGS);
 139 extern Datum gtsvectorout(PG_FUNCTION_ARGS);
 140
 141 /*
 142  * GIN support function
 143  */
 144
 145 extern Datum gin_extract_tsvector(PG_FUNCTION_ARGS);
 146 extern Datum gin_extract_query(PG_FUNCTION_ARGS);
 147 extern Datum gin_ts_consistent(PG_FUNCTION_ARGS);
 148
 149 /*
 150  * Possible strategy numbers for indexes
 151  *        TSearchStrategyNumber  - (tsvector|text) @@ tsquery
 152  *        TSearchWithClassStrategyNumber  - tsvector @@@ tsquery
 153  */
 154 #define TSearchStrategyNumber                   1
 155 #define TSearchWithClassStrategyNumber  2
 156
 157 /*
 158  * TSQuery Utilities
 159  */
 160 extern QueryItem *clean_NOT(QueryItem * ptr, int4 *len);
 161 extern QueryItem *clean_fakeval(QueryItem * ptr, int4 *len);
 162
 163 typedef struct QTNode
 164 {
 165         QueryItem  *valnode;
 166         uint32          flags;
 167         int4            nchild;
 168         char       *word;
 169         uint32          sign;
 170         struct QTNode **child;
 171 } QTNode;
 172
 173 #define QTN_NEEDFREE    0x01
 174 #define QTN_NOCHANGE    0x02
 175 #define QTN_WORDFREE    0x04
 176
 177 typedef uint64 TSQuerySign;
 178
 179 #define TSQS_SIGLEN  (sizeof(TSQuerySign)*BITS_PER_BYTE)
 180
 181
 182 extern QTNode *QT2QTN(QueryItem * in, char *operand);
 183 extern TSQuery QTN2QT(QTNode *in);
 184 extern void QTNFree(QTNode * in);
 185 extern void QTNSort(QTNode * in);
 186 extern void QTNTernary(QTNode * in);
 187 extern void QTNBinary(QTNode * in);
 188 extern int      QTNodeCompare(QTNode * an, QTNode * bn);
 189 extern QTNode *QTNCopy(QTNode *in);
 190 extern bool QTNEq(QTNode * a, QTNode * b);
 191 extern TSQuerySign makeTSQuerySign(TSQuery a);
 192
 193 /*
 194  * TSQuery GiST support
 195  */
 196 extern Datum gtsquery_compress(PG_FUNCTION_ARGS);
 197 extern Datum gtsquery_decompress(PG_FUNCTION_ARGS);
 198 extern Datum gtsquery_consistent(PG_FUNCTION_ARGS);
 199 extern Datum gtsquery_union(PG_FUNCTION_ARGS);
 200 extern Datum gtsquery_same(PG_FUNCTION_ARGS);
 201 extern Datum gtsquery_penalty(PG_FUNCTION_ARGS);
 202 extern Datum gtsquery_picksplit(PG_FUNCTION_ARGS);
 203
 204 /*
 205  * Parser interface to SQL
 206  */
 207 extern Datum ts_token_type_byid(PG_FUNCTION_ARGS);
 208 extern Datum ts_token_type_byname(PG_FUNCTION_ARGS);
 209 extern Datum ts_parse_byid(PG_FUNCTION_ARGS);
 210 extern Datum ts_parse_byname(PG_FUNCTION_ARGS);
 211
 212 /*
 213  * Default word parser
 214  */
 215
 216 extern Datum prsd_start(PG_FUNCTION_ARGS);
 217 extern Datum prsd_nexttoken(PG_FUNCTION_ARGS);
 218 extern Datum prsd_end(PG_FUNCTION_ARGS);
 219 extern Datum prsd_headline(PG_FUNCTION_ARGS);
 220 extern Datum prsd_lextype(PG_FUNCTION_ARGS);
 221
 222 /*
 223  * Dictionary interface to SQL
 224  */
 225 extern Datum ts_lexize_byid(PG_FUNCTION_ARGS);
 226 extern Datum ts_lexize_byname(PG_FUNCTION_ARGS);
 227
 228 /*
 229  * Simple built-in dictionary
 230  */
 231 extern Datum dsimple_init(PG_FUNCTION_ARGS);
 232 extern Datum dsimple_lexize(PG_FUNCTION_ARGS);
 233
 234 /*
 235  * Synonym built-in dictionary
 236  */
 237 extern Datum dsynonym_init(PG_FUNCTION_ARGS);
 238 extern Datum dsynonym_lexize(PG_FUNCTION_ARGS);
 239
 240 /*
 241  * ISpell dictionary
 242  */
 243 extern Datum dispell_init(PG_FUNCTION_ARGS);
 244 extern Datum dispell_lexize(PG_FUNCTION_ARGS);
 245
 246 /*
 247  * Thesaurus
 248  */
 249 extern Datum thesaurus_init(PG_FUNCTION_ARGS);
 250 extern Datum thesaurus_lexize(PG_FUNCTION_ARGS);
 251
 252 /*
 253  * headline
 254  */
 255 extern Datum ts_headline_byid_opt(PG_FUNCTION_ARGS);
 256 extern Datum ts_headline_byid(PG_FUNCTION_ARGS);
 257 extern Datum ts_headline(PG_FUNCTION_ARGS);
 258 extern Datum ts_headline_opt(PG_FUNCTION_ARGS);
 259
 260 /*
 261  * current cfg
 262  */
 263 extern Datum get_current_ts_config(PG_FUNCTION_ARGS);
 264
 265 #endif   /* _PG_TS_UTILS_H_ */