Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * tsginidx.c
4 : * GIN support functions for tsvector_ops
5 : *
6 : * Portions Copyright (c) 1996-2026, PostgreSQL Global Development Group
7 : *
8 : *
9 : * IDENTIFICATION
10 : * src/backend/utils/adt/tsginidx.c
11 : *
12 : *-------------------------------------------------------------------------
13 : */
14 : #include "postgres.h"
15 :
16 : #include "access/gin.h"
17 : #include "tsearch/ts_type.h"
18 : #include "tsearch/ts_utils.h"
19 : #include "utils/builtins.h"
20 : #include "varatt.h"
21 :
22 :
23 : Datum
24 1885022 : gin_cmp_tslexeme(PG_FUNCTION_ARGS)
25 : {
26 1885022 : text *a = PG_GETARG_TEXT_PP(0);
27 1885022 : text *b = PG_GETARG_TEXT_PP(1);
28 : int cmp;
29 :
30 1885022 : cmp = tsCompareString(VARDATA_ANY(a), VARSIZE_ANY_EXHDR(a),
31 1885022 : VARDATA_ANY(b), VARSIZE_ANY_EXHDR(b),
32 : false);
33 :
34 1885022 : PG_FREE_IF_COPY(a, 0);
35 1885022 : PG_FREE_IF_COPY(b, 1);
36 1885022 : PG_RETURN_INT32(cmp);
37 : }
38 :
39 : Datum
40 444 : gin_cmp_prefix(PG_FUNCTION_ARGS)
41 : {
42 444 : text *a = PG_GETARG_TEXT_PP(0);
43 444 : text *b = PG_GETARG_TEXT_PP(1);
44 :
45 : #ifdef NOT_USED
46 : StrategyNumber strategy = PG_GETARG_UINT16(2);
47 : Pointer extra_data = PG_GETARG_POINTER(3);
48 : #endif
49 : int cmp;
50 :
51 444 : cmp = tsCompareString(VARDATA_ANY(a), VARSIZE_ANY_EXHDR(a),
52 444 : VARDATA_ANY(b), VARSIZE_ANY_EXHDR(b),
53 : true);
54 :
55 444 : if (cmp < 0)
56 12 : cmp = 1; /* prevent continue scan */
57 :
58 444 : PG_FREE_IF_COPY(a, 0);
59 444 : PG_FREE_IF_COPY(b, 1);
60 444 : PG_RETURN_INT32(cmp);
61 : }
62 :
63 : Datum
64 3096 : gin_extract_tsvector(PG_FUNCTION_ARGS)
65 : {
66 3096 : TSVector vector = PG_GETARG_TSVECTOR(0);
67 3096 : int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
68 3096 : Datum *entries = NULL;
69 :
70 3096 : *nentries = vector->size;
71 3096 : if (vector->size > 0)
72 : {
73 : int i;
74 3042 : WordEntry *we = ARRPTR(vector);
75 :
76 3042 : entries = palloc_array(Datum, vector->size);
77 :
78 175974 : for (i = 0; i < vector->size; i++)
79 : {
80 : text *txt;
81 :
82 172932 : txt = cstring_to_text_with_len(STRPTR(vector) + we->pos, we->len);
83 172932 : entries[i] = PointerGetDatum(txt);
84 :
85 172932 : we++;
86 : }
87 : }
88 :
89 3096 : PG_FREE_IF_COPY(vector, 0);
90 3096 : PG_RETURN_POINTER(entries);
91 : }
92 :
93 : Datum
94 450 : gin_extract_tsquery(PG_FUNCTION_ARGS)
95 : {
96 450 : TSQuery query = PG_GETARG_TSQUERY(0);
97 450 : int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
98 : #ifdef NOT_USED
99 : StrategyNumber strategy = PG_GETARG_UINT16(2);
100 : #endif
101 450 : bool **ptr_partialmatch = (bool **) PG_GETARG_POINTER(3);
102 450 : Pointer **extra_data = (Pointer **) PG_GETARG_POINTER(4);
103 : #ifdef NOT_USED
104 : bool **nullFlags = (bool **) PG_GETARG_POINTER(5);
105 : #endif
106 450 : int32 *searchMode = (int32 *) PG_GETARG_POINTER(6);
107 450 : Datum *entries = NULL;
108 :
109 450 : *nentries = 0;
110 :
111 450 : if (query->size > 0)
112 : {
113 450 : QueryItem *item = GETQUERY(query);
114 : int32 i,
115 : j;
116 : bool *partialmatch;
117 : int *map_item_operand;
118 :
119 : /*
120 : * If the query doesn't have any required positive matches (for
121 : * instance, it's something like '! foo'), we have to do a full index
122 : * scan.
123 : */
124 450 : if (tsquery_requires_match(item))
125 330 : *searchMode = GIN_SEARCH_MODE_DEFAULT;
126 : else
127 120 : *searchMode = GIN_SEARCH_MODE_ALL;
128 :
129 : /* count number of VAL items */
130 450 : j = 0;
131 1704 : for (i = 0; i < query->size; i++)
132 : {
133 1254 : if (item[i].type == QI_VAL)
134 768 : j++;
135 : }
136 450 : *nentries = j;
137 :
138 450 : entries = palloc_array(Datum, j);
139 450 : partialmatch = *ptr_partialmatch = palloc_array(bool, j);
140 :
141 : /*
142 : * Make map to convert item's number to corresponding operand's (the
143 : * same, entry's) number. Entry's number is used in check array in
144 : * consistent method. We use the same map for each entry.
145 : */
146 450 : *extra_data = palloc_array(Pointer, j);
147 450 : map_item_operand = palloc0_array(int, query->size);
148 :
149 : /* Now rescan the VAL items and fill in the arrays */
150 450 : j = 0;
151 1704 : for (i = 0; i < query->size; i++)
152 : {
153 1254 : if (item[i].type == QI_VAL)
154 : {
155 768 : QueryOperand *val = &item[i].qoperand;
156 : text *txt;
157 :
158 768 : txt = cstring_to_text_with_len(GETOPERAND(query) + val->distance,
159 768 : val->length);
160 768 : entries[j] = PointerGetDatum(txt);
161 768 : partialmatch[j] = val->prefix;
162 768 : (*extra_data)[j] = (Pointer) map_item_operand;
163 768 : map_item_operand[i] = j;
164 768 : j++;
165 : }
166 : }
167 : }
168 :
169 450 : PG_FREE_IF_COPY(query, 0);
170 :
171 450 : PG_RETURN_POINTER(entries);
172 : }
173 :
174 : typedef struct
175 : {
176 : QueryItem *first_item;
177 : GinTernaryValue *check;
178 : int *map_item_operand;
179 : } GinChkVal;
180 :
181 : /*
182 : * TS_execute callback for matching a tsquery operand to GIN index data
183 : */
184 : static TSTernaryValue
185 48366 : checkcondition_gin(void *checkval, QueryOperand *val, ExecPhraseData *data)
186 : {
187 48366 : GinChkVal *gcv = (GinChkVal *) checkval;
188 : int j;
189 : GinTernaryValue result;
190 :
191 : /* convert item's number to corresponding entry's (operand's) number */
192 48366 : j = gcv->map_item_operand[((QueryItem *) val) - gcv->first_item];
193 :
194 : /* determine presence of current entry in indexed value */
195 48366 : result = gcv->check[j];
196 :
197 : /*
198 : * If any val requiring a weight is used or caller needs position
199 : * information then we must recheck, so replace TRUE with MAYBE.
200 : */
201 48366 : if (result == GIN_TRUE)
202 : {
203 15654 : if (val->weight != 0 || data != NULL)
204 6498 : result = GIN_MAYBE;
205 : }
206 :
207 : /*
208 : * We rely on GinTernaryValue and TSTernaryValue using equivalent value
209 : * assignments. We could use a switch statement to map the values if that
210 : * ever stops being true, but it seems unlikely to happen.
211 : */
212 48366 : return (TSTernaryValue) result;
213 : }
214 :
215 : Datum
216 24 : gin_tsquery_consistent(PG_FUNCTION_ARGS)
217 : {
218 24 : bool *check = (bool *) PG_GETARG_POINTER(0);
219 : #ifdef NOT_USED
220 : StrategyNumber strategy = PG_GETARG_UINT16(1);
221 : #endif
222 24 : TSQuery query = PG_GETARG_TSQUERY(2);
223 : #ifdef NOT_USED
224 : int32 nkeys = PG_GETARG_INT32(3);
225 : #endif
226 24 : Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4);
227 24 : bool *recheck = (bool *) PG_GETARG_POINTER(5);
228 24 : bool res = false;
229 :
230 : /* Initially assume query doesn't require recheck */
231 24 : *recheck = false;
232 :
233 24 : if (query->size > 0)
234 : {
235 : GinChkVal gcv;
236 :
237 : /*
238 : * check-parameter array has one entry for each value (operand) in the
239 : * query.
240 : */
241 24 : gcv.first_item = GETQUERY(query);
242 24 : gcv.check = (GinTernaryValue *) check;
243 24 : gcv.map_item_operand = (int *) (extra_data[0]);
244 :
245 24 : switch (TS_execute_ternary(GETQUERY(query),
246 : &gcv,
247 : TS_EXEC_PHRASE_NO_POS,
248 : checkcondition_gin))
249 : {
250 0 : case TS_NO:
251 0 : res = false;
252 0 : break;
253 24 : case TS_YES:
254 24 : res = true;
255 24 : break;
256 0 : case TS_MAYBE:
257 0 : res = true;
258 0 : *recheck = true;
259 0 : break;
260 : }
261 : }
262 :
263 24 : PG_RETURN_BOOL(res);
264 : }
265 :
266 : Datum
267 36918 : gin_tsquery_triconsistent(PG_FUNCTION_ARGS)
268 : {
269 36918 : GinTernaryValue *check = (GinTernaryValue *) PG_GETARG_POINTER(0);
270 : #ifdef NOT_USED
271 : StrategyNumber strategy = PG_GETARG_UINT16(1);
272 : #endif
273 36918 : TSQuery query = PG_GETARG_TSQUERY(2);
274 : #ifdef NOT_USED
275 : int32 nkeys = PG_GETARG_INT32(3);
276 : #endif
277 36918 : Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4);
278 36918 : GinTernaryValue res = GIN_FALSE;
279 :
280 36918 : if (query->size > 0)
281 : {
282 : GinChkVal gcv;
283 :
284 : /*
285 : * check-parameter array has one entry for each value (operand) in the
286 : * query.
287 : */
288 36918 : gcv.first_item = GETQUERY(query);
289 36918 : gcv.check = check;
290 36918 : gcv.map_item_operand = (int *) (extra_data[0]);
291 :
292 36918 : res = TS_execute_ternary(GETQUERY(query),
293 : &gcv,
294 : TS_EXEC_PHRASE_NO_POS,
295 : checkcondition_gin);
296 : }
297 :
298 36918 : PG_RETURN_GIN_TERNARY_VALUE(res);
299 : }
300 :
301 : /*
302 : * Formerly, gin_extract_tsvector had only two arguments. Now it has three,
303 : * but we still need a pg_proc entry with two args to support reloading
304 : * pre-9.1 contrib/tsearch2 opclass declarations. This compatibility
305 : * function should go away eventually. (Note: you might say "hey, but the
306 : * code above is only *using* two args, so let's just declare it that way".
307 : * If you try that you'll find the opr_sanity regression test complains.)
308 : */
309 : Datum
310 0 : gin_extract_tsvector_2args(PG_FUNCTION_ARGS)
311 : {
312 0 : if (PG_NARGS() < 3) /* should not happen */
313 0 : elog(ERROR, "gin_extract_tsvector requires three arguments");
314 0 : return gin_extract_tsvector(fcinfo);
315 : }
316 :
317 : /*
318 : * Likewise, we need a stub version of gin_extract_tsquery declared with
319 : * only five arguments.
320 : */
321 : Datum
322 0 : gin_extract_tsquery_5args(PG_FUNCTION_ARGS)
323 : {
324 0 : if (PG_NARGS() < 7) /* should not happen */
325 0 : elog(ERROR, "gin_extract_tsquery requires seven arguments");
326 0 : return gin_extract_tsquery(fcinfo);
327 : }
328 :
329 : /*
330 : * Likewise, we need a stub version of gin_tsquery_consistent declared with
331 : * only six arguments.
332 : */
333 : Datum
334 0 : gin_tsquery_consistent_6args(PG_FUNCTION_ARGS)
335 : {
336 0 : if (PG_NARGS() < 8) /* should not happen */
337 0 : elog(ERROR, "gin_tsquery_consistent requires eight arguments");
338 0 : return gin_tsquery_consistent(fcinfo);
339 : }
340 :
341 : /*
342 : * Likewise, a stub version of gin_extract_tsquery declared with argument
343 : * types that are no longer considered appropriate.
344 : */
345 : Datum
346 0 : gin_extract_tsquery_oldsig(PG_FUNCTION_ARGS)
347 : {
348 0 : return gin_extract_tsquery(fcinfo);
349 : }
350 :
351 : /*
352 : * Likewise, a stub version of gin_tsquery_consistent declared with argument
353 : * types that are no longer considered appropriate.
354 : */
355 : Datum
356 0 : gin_tsquery_consistent_oldsig(PG_FUNCTION_ARGS)
357 : {
358 0 : return gin_tsquery_consistent(fcinfo);
359 : }
|
