TLA 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 GIC 904978 : gin_cmp_tslexeme(PG_FUNCTION_ARGS)
27 ECB : {
28 GIC 904978 : text *a = PG_GETARG_TEXT_PP(0);
29 CBC 904978 : text *b = PG_GETARG_TEXT_PP(1);
30 ECB : int cmp;
31 :
32 GIC 1809956 : cmp = tsCompareString(VARDATA_ANY(a), VARSIZE_ANY_EXHDR(a),
33 CBC 1809956 : VARDATA_ANY(b), VARSIZE_ANY_EXHDR(b),
34 ECB : false);
35 :
36 GIC 904978 : PG_FREE_IF_COPY(a, 0);
37 CBC 904978 : PG_FREE_IF_COPY(b, 1);
38 904978 : PG_RETURN_INT32(cmp);
39 ECB : }
40 :
41 : Datum
42 GIC 222 : gin_cmp_prefix(PG_FUNCTION_ARGS)
43 ECB : {
44 GIC 222 : text *a = PG_GETARG_TEXT_PP(0);
45 CBC 222 : text *b = PG_GETARG_TEXT_PP(1);
46 ECB :
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 GIC 444 : cmp = tsCompareString(VARDATA_ANY(a), VARSIZE_ANY_EXHDR(a),
54 CBC 444 : VARDATA_ANY(b), VARSIZE_ANY_EXHDR(b),
55 ECB : true);
56 :
57 GIC 222 : if (cmp < 0)
58 CBC 6 : cmp = 1; /* prevent continue scan */
59 ECB :
60 GIC 222 : PG_FREE_IF_COPY(a, 0);
61 CBC 222 : PG_FREE_IF_COPY(b, 1);
62 222 : PG_RETURN_INT32(cmp);
63 ECB : }
64 :
65 : Datum
66 GIC 1548 : gin_extract_tsvector(PG_FUNCTION_ARGS)
67 ECB : {
68 GIC 1548 : TSVector vector = PG_GETARG_TSVECTOR(0);
69 CBC 1548 : int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
70 1548 : Datum *entries = NULL;
71 ECB :
72 GIC 1548 : *nentries = vector->size;
73 CBC 1548 : if (vector->size > 0)
74 ECB : {
75 : int i;
76 GIC 1521 : WordEntry *we = ARRPTR(vector);
77 ECB :
78 GIC 1521 : entries = (Datum *) palloc(sizeof(Datum) * vector->size);
79 ECB :
80 GIC 87987 : for (i = 0; i < vector->size; i++)
81 ECB : {
82 : text *txt;
83 :
84 GIC 86466 : txt = cstring_to_text_with_len(STRPTR(vector) + we->pos, we->len);
85 CBC 86466 : entries[i] = PointerGetDatum(txt);
86 ECB :
87 GIC 86466 : we++;
88 ECB : }
89 : }
90 :
91 GIC 1548 : PG_FREE_IF_COPY(vector, 0);
92 CBC 1548 : PG_RETURN_POINTER(entries);
93 ECB : }
94 :
95 : Datum
96 GIC 225 : gin_extract_tsquery(PG_FUNCTION_ARGS)
97 ECB : {
98 GIC 225 : TSQuery query = PG_GETARG_TSQUERY(0);
99 CBC 225 : int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
100 ECB :
101 : /* StrategyNumber strategy = PG_GETARG_UINT16(2); */
102 GIC 225 : bool **ptr_partialmatch = (bool **) PG_GETARG_POINTER(3);
103 CBC 225 : Pointer **extra_data = (Pointer **) PG_GETARG_POINTER(4);
104 ECB :
105 : /* bool **nullFlags = (bool **) PG_GETARG_POINTER(5); */
106 GIC 225 : int32 *searchMode = (int32 *) PG_GETARG_POINTER(6);
107 CBC 225 : Datum *entries = NULL;
108 ECB :
109 GIC 225 : *nentries = 0;
110 ECB :
111 GIC 225 : if (query->size > 0)
112 ECB : {
113 GIC 225 : QueryItem *item = GETQUERY(query);
114 ECB : 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 GIC 225 : if (tsquery_requires_match(item))
125 CBC 165 : *searchMode = GIN_SEARCH_MODE_DEFAULT;
126 ECB : else
127 GIC 60 : *searchMode = GIN_SEARCH_MODE_ALL;
128 ECB :
129 : /* count number of VAL items */
130 GIC 225 : j = 0;
131 CBC 852 : for (i = 0; i < query->size; i++)
132 ECB : {
133 GIC 627 : if (item[i].type == QI_VAL)
134 CBC 384 : j++;
135 ECB : }
136 GIC 225 : *nentries = j;
137 ECB :
138 GIC 225 : entries = (Datum *) palloc(sizeof(Datum) * j);
139 CBC 225 : partialmatch = *ptr_partialmatch = (bool *) palloc(sizeof(bool) * j);
140 ECB :
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 GIC 225 : *extra_data = (Pointer *) palloc(sizeof(Pointer) * j);
147 CBC 225 : map_item_operand = (int *) palloc0(sizeof(int) * query->size);
148 ECB :
149 : /* Now rescan the VAL items and fill in the arrays */
150 GIC 225 : j = 0;
151 CBC 852 : for (i = 0; i < query->size; i++)
152 ECB : {
153 GIC 627 : if (item[i].type == QI_VAL)
154 ECB : {
155 GIC 384 : QueryOperand *val = &item[i].qoperand;
156 ECB : text *txt;
157 :
158 GIC 384 : txt = cstring_to_text_with_len(GETOPERAND(query) + val->distance,
159 CBC 384 : val->length);
160 384 : entries[j] = PointerGetDatum(txt);
161 384 : partialmatch[j] = val->prefix;
162 384 : (*extra_data)[j] = (Pointer) map_item_operand;
163 384 : map_item_operand[i] = j;
164 384 : j++;
165 ECB : }
166 : }
167 : }
168 :
169 GIC 225 : PG_FREE_IF_COPY(query, 0);
170 ECB :
171 GIC 225 : PG_RETURN_POINTER(entries);
172 ECB : }
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 GIC 24183 : checkcondition_gin(void *checkval, QueryOperand *val, ExecPhraseData *data)
186 ECB : {
187 GIC 24183 : GinChkVal *gcv = (GinChkVal *) checkval;
188 ECB : int j;
189 : GinTernaryValue result;
190 :
191 : /* convert item's number to corresponding entry's (operand's) number */
192 GIC 24183 : j = gcv->map_item_operand[((QueryItem *) val) - gcv->first_item];
193 ECB :
194 : /* determine presence of current entry in indexed value */
195 GIC 24183 : result = gcv->check[j];
196 ECB :
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 GIC 24183 : if (result == GIN_TRUE)
202 ECB : {
203 GIC 7827 : if (val->weight != 0 || data != NULL)
204 CBC 3249 : result = GIN_MAYBE;
205 ECB : }
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 GIC 24183 : return (TSTernaryValue) result;
213 ECB : }
214 :
215 : Datum
216 GIC 12 : gin_tsquery_consistent(PG_FUNCTION_ARGS)
217 ECB : {
218 GIC 12 : bool *check = (bool *) PG_GETARG_POINTER(0);
219 ECB :
220 : /* StrategyNumber strategy = PG_GETARG_UINT16(1); */
221 GIC 12 : TSQuery query = PG_GETARG_TSQUERY(2);
222 ECB :
223 : /* int32 nkeys = PG_GETARG_INT32(3); */
224 GIC 12 : Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4);
225 CBC 12 : bool *recheck = (bool *) PG_GETARG_POINTER(5);
226 12 : bool res = false;
227 ECB :
228 : /* Initially assume query doesn't require recheck */
229 GIC 12 : *recheck = false;
230 ECB :
231 GIC 12 : if (query->size > 0)
232 ECB : {
233 : GinChkVal gcv;
234 :
235 : /*
236 : * check-parameter array has one entry for each value (operand) in the
237 : * query.
238 : */
239 GIC 12 : gcv.first_item = GETQUERY(query);
240 CBC 12 : gcv.check = (GinTernaryValue *) check;
241 GIC 12 : gcv.map_item_operand = (int *) (extra_data[0]);
242 ECB :
243 GIC 12 : switch (TS_execute_ternary(GETQUERY(query),
244 : &gcv,
245 : TS_EXEC_PHRASE_NO_POS,
246 : checkcondition_gin))
247 EUB : {
248 UBC 0 : case TS_NO:
249 0 : res = false;
250 LBC 0 : break;
251 CBC 12 : case TS_YES:
252 12 : res = true;
253 GBC 12 : break;
254 UBC 0 : case TS_MAYBE:
255 0 : res = true;
256 0 : *recheck = true;
257 UIC 0 : break;
258 : }
259 : }
260 ECB :
261 GIC 12 : PG_RETURN_BOOL(res);
262 : }
263 :
264 ECB : Datum
265 GIC 18459 : gin_tsquery_triconsistent(PG_FUNCTION_ARGS)
266 ECB : {
267 GIC 18459 : GinTernaryValue *check = (GinTernaryValue *) PG_GETARG_POINTER(0);
268 :
269 ECB : /* StrategyNumber strategy = PG_GETARG_UINT16(1); */
270 GIC 18459 : TSQuery query = PG_GETARG_TSQUERY(2);
271 :
272 ECB : /* int32 nkeys = PG_GETARG_INT32(3); */
273 CBC 18459 : Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4);
274 GIC 18459 : GinTernaryValue res = GIN_FALSE;
275 ECB :
276 GIC 18459 : 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 ECB : */
284 CBC 18459 : gcv.first_item = GETQUERY(query);
285 18459 : gcv.check = check;
286 GIC 18459 : gcv.map_item_operand = (int *) (extra_data[0]);
287 ECB :
288 GIC 18459 : res = TS_execute_ternary(GETQUERY(query),
289 : &gcv,
290 : TS_EXEC_PHRASE_NO_POS,
291 : checkcondition_gin);
292 : }
293 ECB :
294 GIC 18459 : 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 EUB : Datum
306 UIC 0 : gin_extract_tsvector_2args(PG_FUNCTION_ARGS)
307 EUB : {
308 UBC 0 : if (PG_NARGS() < 3) /* should not happen */
309 0 : elog(ERROR, "gin_extract_tsvector requires three arguments");
310 UIC 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 EUB : Datum
318 UIC 0 : gin_extract_tsquery_5args(PG_FUNCTION_ARGS)
319 EUB : {
320 UBC 0 : if (PG_NARGS() < 7) /* should not happen */
321 0 : elog(ERROR, "gin_extract_tsquery requires seven arguments");
322 UIC 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 EUB : Datum
330 UIC 0 : gin_tsquery_consistent_6args(PG_FUNCTION_ARGS)
331 EUB : {
332 UBC 0 : if (PG_NARGS() < 8) /* should not happen */
333 0 : elog(ERROR, "gin_tsquery_consistent requires eight arguments");
334 UIC 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 EUB : Datum
342 UIC 0 : gin_extract_tsquery_oldsig(PG_FUNCTION_ARGS)
343 EUB : {
344 UIC 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 EUB : Datum
352 UIC 0 : gin_tsquery_consistent_oldsig(PG_FUNCTION_ARGS)
353 EUB : {
354 UIC 0 : return gin_tsquery_consistent(fcinfo);
355 : }
|