Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * tsginidx.c
4 : * GIN support functions for tsvector_ops
5 : *
6 : * Portions Copyright (c) 1996-2023, 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 "access/stratnum.h"
18 : #include "miscadmin.h"
19 : #include "tsearch/ts_type.h"
20 : #include "tsearch/ts_utils.h"
21 : #include "utils/builtins.h"
22 : #include "varatt.h"
23 :
24 :
25 : Datum
26 1809956 : gin_cmp_tslexeme(PG_FUNCTION_ARGS)
27 : {
28 1809956 : text *a = PG_GETARG_TEXT_PP(0);
29 1809956 : text *b = PG_GETARG_TEXT_PP(1);
30 : int cmp;
31 :
32 3619912 : cmp = tsCompareString(VARDATA_ANY(a), VARSIZE_ANY_EXHDR(a),
33 3619912 : VARDATA_ANY(b), VARSIZE_ANY_EXHDR(b),
34 : false);
35 :
36 1809956 : PG_FREE_IF_COPY(a, 0);
37 1809956 : PG_FREE_IF_COPY(b, 1);
38 1809956 : PG_RETURN_INT32(cmp);
39 : }
40 :
41 : Datum
42 444 : gin_cmp_prefix(PG_FUNCTION_ARGS)
43 : {
44 444 : text *a = PG_GETARG_TEXT_PP(0);
45 444 : text *b = PG_GETARG_TEXT_PP(1);
46 :
47 : #ifdef NOT_USED
48 : StrategyNumber strategy = PG_GETARG_UINT16(2);
49 : Pointer extra_data = PG_GETARG_POINTER(3);
50 : #endif
51 : int cmp;
52 :
53 888 : cmp = tsCompareString(VARDATA_ANY(a), VARSIZE_ANY_EXHDR(a),
54 888 : VARDATA_ANY(b), VARSIZE_ANY_EXHDR(b),
55 : true);
56 :
57 444 : if (cmp < 0)
58 12 : cmp = 1; /* prevent continue scan */
59 :
60 444 : PG_FREE_IF_COPY(a, 0);
61 444 : PG_FREE_IF_COPY(b, 1);
62 444 : PG_RETURN_INT32(cmp);
63 : }
64 :
65 : Datum
66 3096 : gin_extract_tsvector(PG_FUNCTION_ARGS)
67 : {
68 3096 : TSVector vector = PG_GETARG_TSVECTOR(0);
69 3096 : int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
70 3096 : Datum *entries = NULL;
71 :
72 3096 : *nentries = vector->size;
73 3096 : if (vector->size > 0)
74 : {
75 : int i;
76 3042 : WordEntry *we = ARRPTR(vector);
77 :
78 3042 : entries = (Datum *) palloc(sizeof(Datum) * vector->size);
79 :
80 175974 : for (i = 0; i < vector->size; i++)
81 : {
82 : text *txt;
83 :
84 172932 : txt = cstring_to_text_with_len(STRPTR(vector) + we->pos, we->len);
85 172932 : entries[i] = PointerGetDatum(txt);
86 :
87 172932 : we++;
88 : }
89 : }
90 :
91 3096 : PG_FREE_IF_COPY(vector, 0);
92 3096 : PG_RETURN_POINTER(entries);
93 : }
94 :
95 : Datum
96 450 : gin_extract_tsquery(PG_FUNCTION_ARGS)
97 : {
98 450 : TSQuery query = PG_GETARG_TSQUERY(0);
99 450 : int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
100 :
101 : /* StrategyNumber strategy = PG_GETARG_UINT16(2); */
102 450 : bool **ptr_partialmatch = (bool **) PG_GETARG_POINTER(3);
103 450 : Pointer **extra_data = (Pointer **) PG_GETARG_POINTER(4);
104 :
105 : /* bool **nullFlags = (bool **) PG_GETARG_POINTER(5); */
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 = (Datum *) palloc(sizeof(Datum) * j);
139 450 : partialmatch = *ptr_partialmatch = (bool *) palloc(sizeof(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 = (Pointer *) palloc(sizeof(Pointer) * j);
147 450 : map_item_operand = (int *) palloc0(sizeof(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 :
220 : /* StrategyNumber strategy = PG_GETARG_UINT16(1); */
221 24 : TSQuery query = PG_GETARG_TSQUERY(2);
222 :
223 : /* int32 nkeys = PG_GETARG_INT32(3); */
224 24 : Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4);
225 24 : bool *recheck = (bool *) PG_GETARG_POINTER(5);
226 24 : bool res = false;
227 :
228 : /* Initially assume query doesn't require recheck */
229 24 : *recheck = false;
230 :
231 24 : if (query->size > 0)
232 : {
233 : GinChkVal gcv;
234 :
235 : /*
236 : * check-parameter array has one entry for each value (operand) in the
237 : * query.
238 : */
239 24 : gcv.first_item = GETQUERY(query);
240 24 : gcv.check = (GinTernaryValue *) check;
241 24 : gcv.map_item_operand = (int *) (extra_data[0]);
242 :
243 24 : switch (TS_execute_ternary(GETQUERY(query),
244 : &gcv,
245 : TS_EXEC_PHRASE_NO_POS,
246 : checkcondition_gin))
247 : {
248 0 : case TS_NO:
249 0 : res = false;
250 0 : break;
251 24 : case TS_YES:
252 24 : res = true;
253 24 : break;
254 0 : case TS_MAYBE:
255 0 : res = true;
256 0 : *recheck = true;
257 0 : break;
258 : }
259 24 : }
260 :
261 24 : PG_RETURN_BOOL(res);
262 : }
263 :
264 : Datum
265 36918 : gin_tsquery_triconsistent(PG_FUNCTION_ARGS)
266 : {
267 36918 : GinTernaryValue *check = (GinTernaryValue *) PG_GETARG_POINTER(0);
268 :
269 : /* StrategyNumber strategy = PG_GETARG_UINT16(1); */
270 36918 : TSQuery query = PG_GETARG_TSQUERY(2);
271 :
272 : /* int32 nkeys = PG_GETARG_INT32(3); */
273 36918 : Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4);
274 36918 : GinTernaryValue res = GIN_FALSE;
275 :
276 36918 : if (query->size > 0)
277 : {
278 : GinChkVal gcv;
279 :
280 : /*
281 : * check-parameter array has one entry for each value (operand) in the
282 : * query.
283 : */
284 36918 : gcv.first_item = GETQUERY(query);
285 36918 : gcv.check = check;
286 36918 : gcv.map_item_operand = (int *) (extra_data[0]);
287 :
288 36918 : res = TS_execute_ternary(GETQUERY(query),
289 : &gcv,
290 : TS_EXEC_PHRASE_NO_POS,
291 : checkcondition_gin);
292 : }
293 :
294 36918 : PG_RETURN_GIN_TERNARY_VALUE(res);
295 : }
296 :
297 : /*
298 : * Formerly, gin_extract_tsvector had only two arguments. Now it has three,
299 : * but we still need a pg_proc entry with two args to support reloading
300 : * pre-9.1 contrib/tsearch2 opclass declarations. This compatibility
301 : * function should go away eventually. (Note: you might say "hey, but the
302 : * code above is only *using* two args, so let's just declare it that way".
303 : * If you try that you'll find the opr_sanity regression test complains.)
304 : */
305 : Datum
306 0 : gin_extract_tsvector_2args(PG_FUNCTION_ARGS)
307 : {
308 0 : if (PG_NARGS() < 3) /* should not happen */
309 0 : elog(ERROR, "gin_extract_tsvector requires three arguments");
310 0 : return gin_extract_tsvector(fcinfo);
311 : }
312 :
313 : /*
314 : * Likewise, we need a stub version of gin_extract_tsquery declared with
315 : * only five arguments.
316 : */
317 : Datum
318 0 : gin_extract_tsquery_5args(PG_FUNCTION_ARGS)
319 : {
320 0 : if (PG_NARGS() < 7) /* should not happen */
321 0 : elog(ERROR, "gin_extract_tsquery requires seven arguments");
322 0 : return gin_extract_tsquery(fcinfo);
323 : }
324 :
325 : /*
326 : * Likewise, we need a stub version of gin_tsquery_consistent declared with
327 : * only six arguments.
328 : */
329 : Datum
330 0 : gin_tsquery_consistent_6args(PG_FUNCTION_ARGS)
331 : {
332 0 : if (PG_NARGS() < 8) /* should not happen */
333 0 : elog(ERROR, "gin_tsquery_consistent requires eight arguments");
334 0 : return gin_tsquery_consistent(fcinfo);
335 : }
336 :
337 : /*
338 : * Likewise, a stub version of gin_extract_tsquery declared with argument
339 : * types that are no longer considered appropriate.
340 : */
341 : Datum
342 0 : gin_extract_tsquery_oldsig(PG_FUNCTION_ARGS)
343 : {
344 0 : return gin_extract_tsquery(fcinfo);
345 : }
346 :
347 : /*
348 : * Likewise, a stub version of gin_tsquery_consistent declared with argument
349 : * types that are no longer considered appropriate.
350 : */
351 : Datum
352 0 : gin_tsquery_consistent_oldsig(PG_FUNCTION_ARGS)
353 : {
354 0 : return gin_tsquery_consistent(fcinfo);
355 : }
|
