TLA Line data Source code
1 : /*
2 : * contrib/pg_trgm/trgm_gin.c
3 : */
4 : #include "postgres.h"
5 :
6 : #include "access/gin.h"
7 : #include "access/stratnum.h"
8 : #include "fmgr.h"
9 : #include "trgm.h"
10 : #include "varatt.h"
11 :
12 UIC 0 : PG_FUNCTION_INFO_V1(gin_extract_trgm);
13 GBC 4 : PG_FUNCTION_INFO_V1(gin_extract_value_trgm);
14 CBC 4 : PG_FUNCTION_INFO_V1(gin_extract_query_trgm);
15 4 : PG_FUNCTION_INFO_V1(gin_trgm_consistent);
16 4 : PG_FUNCTION_INFO_V1(gin_trgm_triconsistent);
17 ECB :
18 : /*
19 : * This function can only be called if a pre-9.1 version of the GIN operator
20 : * class definition is present in the catalogs (probably as a consequence
21 : * of upgrade-in-place). Cope.
22 : */
23 : Datum
24 UIC 0 : gin_extract_trgm(PG_FUNCTION_ARGS)
25 EUB : {
26 UIC 0 : if (PG_NARGS() == 3)
27 UBC 0 : return gin_extract_value_trgm(fcinfo);
28 0 : if (PG_NARGS() == 7)
29 0 : return gin_extract_query_trgm(fcinfo);
30 0 : elog(ERROR, "unexpected number of arguments to gin_extract_trgm");
31 EUB : PG_RETURN_NULL();
32 : }
33 :
34 : Datum
35 GIC 2404 : gin_extract_value_trgm(PG_FUNCTION_ARGS)
36 ECB : {
37 GIC 2404 : text *val = (text *) PG_GETARG_TEXT_PP(0);
38 CBC 2404 : int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
39 2404 : Datum *entries = NULL;
40 ECB : TRGM *trg;
41 : int32 trglen;
42 :
43 GIC 2404 : *nentries = 0;
44 ECB :
45 GIC 2404 : trg = generate_trgm(VARDATA_ANY(val), VARSIZE_ANY_EXHDR(val));
46 CBC 2404 : trglen = ARRNELEM(trg);
47 ECB :
48 GIC 2404 : if (trglen > 0)
49 ECB : {
50 : trgm *ptr;
51 : int32 i;
52 :
53 GIC 2404 : *nentries = trglen;
54 CBC 2404 : entries = (Datum *) palloc(sizeof(Datum) * trglen);
55 ECB :
56 GIC 2404 : ptr = GETARR(trg);
57 CBC 35631 : for (i = 0; i < trglen; i++)
58 ECB : {
59 GIC 33227 : int32 item = trgm2int(ptr);
60 ECB :
61 GIC 33227 : entries[i] = Int32GetDatum(item);
62 CBC 33227 : ptr++;
63 ECB : }
64 : }
65 :
66 GIC 2404 : PG_RETURN_POINTER(entries);
67 ECB : }
68 :
69 : Datum
70 GIC 172 : gin_extract_query_trgm(PG_FUNCTION_ARGS)
71 ECB : {
72 GIC 172 : text *val = (text *) PG_GETARG_TEXT_PP(0);
73 CBC 172 : int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
74 172 : StrategyNumber strategy = PG_GETARG_UINT16(2);
75 ECB :
76 : /* bool **pmatch = (bool **) PG_GETARG_POINTER(3); */
77 GIC 172 : Pointer **extra_data = (Pointer **) PG_GETARG_POINTER(4);
78 ECB :
79 : /* bool **nullFlags = (bool **) PG_GETARG_POINTER(5); */
80 GIC 172 : int32 *searchMode = (int32 *) PG_GETARG_POINTER(6);
81 CBC 172 : Datum *entries = NULL;
82 ECB : TRGM *trg;
83 : int32 trglen;
84 : trgm *ptr;
85 : TrgmPackedGraph *graph;
86 : int32 i;
87 :
88 GIC 172 : switch (strategy)
89 ECB : {
90 GIC 80 : case SimilarityStrategyNumber:
91 ECB : case WordSimilarityStrategyNumber:
92 : case StrictWordSimilarityStrategyNumber:
93 : case EqualStrategyNumber:
94 GIC 80 : trg = generate_trgm(VARDATA_ANY(val), VARSIZE_ANY_EXHDR(val));
95 CBC 80 : break;
96 48 : case ILikeStrategyNumber:
97 ECB : #ifndef IGNORECASE
98 : elog(ERROR, "cannot handle ~~* with case-sensitive trigrams");
99 : #endif
100 : /* FALL THRU */
101 : case LikeStrategyNumber:
102 :
103 : /*
104 : * For wildcard search we extract all the trigrams that every
105 : * potentially-matching string must include.
106 : */
107 GIC 48 : trg = generate_wildcard_trgm(VARDATA_ANY(val),
108 CBC 48 : VARSIZE_ANY_EXHDR(val));
109 48 : break;
110 44 : case RegExpICaseStrategyNumber:
111 ECB : #ifndef IGNORECASE
112 : elog(ERROR, "cannot handle ~* with case-sensitive trigrams");
113 : #endif
114 : /* FALL THRU */
115 : case RegExpStrategyNumber:
116 GIC 44 : trg = createTrgmNFA(val, PG_GET_COLLATION(),
117 ECB : &graph, CurrentMemoryContext);
118 GIC 44 : if (trg && ARRNELEM(trg) > 0)
119 ECB : {
120 : /*
121 : * Successful regex processing: store NFA-like graph as
122 : * extra_data. GIN API requires an array of nentries
123 : * Pointers, but we just put the same value in each element.
124 : */
125 GIC 34 : trglen = ARRNELEM(trg);
126 CBC 34 : *extra_data = (Pointer *) palloc(sizeof(Pointer) * trglen);
127 848 : for (i = 0; i < trglen; i++)
128 814 : (*extra_data)[i] = (Pointer) graph;
129 ECB : }
130 : else
131 : {
132 : /* No result: have to do full index scan. */
133 GIC 10 : *nentries = 0;
134 CBC 10 : *searchMode = GIN_SEARCH_MODE_ALL;
135 10 : PG_RETURN_POINTER(entries);
136 ECB : }
137 GIC 34 : break;
138 LBC 0 : default:
139 UBC 0 : elog(ERROR, "unrecognized strategy number: %d", strategy);
140 EUB : trg = NULL; /* keep compiler quiet */
141 : break;
142 : }
143 :
144 GIC 162 : trglen = ARRNELEM(trg);
145 CBC 162 : *nentries = trglen;
146 ECB :
147 GIC 162 : if (trglen > 0)
148 ECB : {
149 GIC 138 : entries = (Datum *) palloc(sizeof(Datum) * trglen);
150 CBC 138 : ptr = GETARR(trg);
151 1684 : for (i = 0; i < trglen; i++)
152 ECB : {
153 GIC 1546 : int32 item = trgm2int(ptr);
154 ECB :
155 GIC 1546 : entries[i] = Int32GetDatum(item);
156 CBC 1546 : ptr++;
157 ECB : }
158 : }
159 :
160 : /*
161 : * If no trigram was extracted then we have to scan all the index.
162 : */
163 GIC 162 : if (trglen == 0)
164 CBC 24 : *searchMode = GIN_SEARCH_MODE_ALL;
165 ECB :
166 GIC 162 : PG_RETURN_POINTER(entries);
167 ECB : }
168 :
169 : Datum
170 GIC 8 : gin_trgm_consistent(PG_FUNCTION_ARGS)
171 ECB : {
172 GIC 8 : bool *check = (bool *) PG_GETARG_POINTER(0);
173 CBC 8 : StrategyNumber strategy = PG_GETARG_UINT16(1);
174 ECB :
175 : /* text *query = PG_GETARG_TEXT_PP(2); */
176 GIC 8 : int32 nkeys = PG_GETARG_INT32(3);
177 CBC 8 : Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4);
178 8 : bool *recheck = (bool *) PG_GETARG_POINTER(5);
179 ECB : bool res;
180 : int32 i,
181 : ntrue;
182 : double nlimit;
183 :
184 : /* All cases served by this function are inexact */
185 GIC 8 : *recheck = true;
186 ECB :
187 GIC 8 : switch (strategy)
188 ECB : {
189 UIC 0 : case SimilarityStrategyNumber:
190 EUB : case WordSimilarityStrategyNumber:
191 : case StrictWordSimilarityStrategyNumber:
192 UIC 0 : nlimit = index_strategy_get_limit(strategy);
193 EUB :
194 : /* Count the matches */
195 UIC 0 : ntrue = 0;
196 UBC 0 : for (i = 0; i < nkeys; i++)
197 EUB : {
198 UIC 0 : if (check[i])
199 UBC 0 : ntrue++;
200 EUB : }
201 :
202 : /*--------------------
203 : * If DIVUNION is defined then similarity formula is:
204 : * c / (len1 + len2 - c)
205 : * where c is number of common trigrams and it stands as ntrue in
206 : * this code. Here we don't know value of len2 but we can assume
207 : * that c (ntrue) is a lower bound of len2, so upper bound of
208 : * similarity is:
209 : * c / (len1 + c - c) => c / len1
210 : * If DIVUNION is not defined then similarity formula is:
211 : * c / max(len1, len2)
212 : * And again, c (ntrue) is a lower bound of len2, but c <= len1
213 : * just by definition and, consequently, upper bound of
214 : * similarity is just c / len1.
215 : * So, independently on DIVUNION the upper bound formula is the same.
216 : */
217 UIC 0 : res = (nkeys == 0) ? false :
218 UBC 0 : (((((float4) ntrue) / ((float4) nkeys))) >= nlimit);
219 0 : break;
220 GBC 8 : case ILikeStrategyNumber:
221 ECB : #ifndef IGNORECASE
222 : elog(ERROR, "cannot handle ~~* with case-sensitive trigrams");
223 : #endif
224 : /* FALL THRU */
225 : case LikeStrategyNumber:
226 : case EqualStrategyNumber:
227 : /* Check if all extracted trigrams are presented. */
228 GIC 8 : res = true;
229 CBC 16 : for (i = 0; i < nkeys; i++)
230 ECB : {
231 GIC 8 : if (!check[i])
232 ECB : {
233 UIC 0 : res = false;
234 UBC 0 : break;
235 EUB : }
236 : }
237 GIC 8 : break;
238 LBC 0 : case RegExpICaseStrategyNumber:
239 EUB : #ifndef IGNORECASE
240 : elog(ERROR, "cannot handle ~* with case-sensitive trigrams");
241 : #endif
242 : /* FALL THRU */
243 : case RegExpStrategyNumber:
244 UIC 0 : if (nkeys < 1)
245 EUB : {
246 : /* Regex processing gave no result: do full index scan */
247 UIC 0 : res = true;
248 EUB : }
249 : else
250 UIC 0 : res = trigramsMatchGraph((TrgmPackedGraph *) extra_data[0],
251 EUB : check);
252 UIC 0 : break;
253 UBC 0 : default:
254 0 : elog(ERROR, "unrecognized strategy number: %d", strategy);
255 EUB : res = false; /* keep compiler quiet */
256 : break;
257 : }
258 :
259 GIC 8 : PG_RETURN_BOOL(res);
260 ECB : }
261 :
262 : /*
263 : * In all cases, GIN_TRUE is at least as favorable to inclusion as
264 : * GIN_MAYBE. If no better option is available, simply treat
265 : * GIN_MAYBE as if it were GIN_TRUE and apply the same test as the binary
266 : * consistent function.
267 : */
268 : Datum
269 GIC 14358 : gin_trgm_triconsistent(PG_FUNCTION_ARGS)
270 ECB : {
271 GIC 14358 : GinTernaryValue *check = (GinTernaryValue *) PG_GETARG_POINTER(0);
272 CBC 14358 : StrategyNumber strategy = PG_GETARG_UINT16(1);
273 ECB :
274 : /* text *query = PG_GETARG_TEXT_PP(2); */
275 GIC 14358 : int32 nkeys = PG_GETARG_INT32(3);
276 CBC 14358 : Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4);
277 14358 : GinTernaryValue res = GIN_MAYBE;
278 ECB : int32 i,
279 : ntrue;
280 : bool *boolcheck;
281 : double nlimit;
282 :
283 GIC 14358 : switch (strategy)
284 ECB : {
285 GIC 8191 : case SimilarityStrategyNumber:
286 ECB : case WordSimilarityStrategyNumber:
287 : case StrictWordSimilarityStrategyNumber:
288 GIC 8191 : nlimit = index_strategy_get_limit(strategy);
289 ECB :
290 : /* Count the matches */
291 GIC 8191 : ntrue = 0;
292 CBC 86909 : for (i = 0; i < nkeys; i++)
293 ECB : {
294 GIC 78718 : if (check[i] != GIN_FALSE)
295 CBC 33172 : ntrue++;
296 ECB : }
297 :
298 : /*
299 : * See comment in gin_trgm_consistent() about * upper bound
300 : * formula
301 : */
302 GIC 16382 : res = (nkeys == 0)
303 CBC 8191 : ? GIN_FALSE : (((((float4) ntrue) / ((float4) nkeys)) >= nlimit)
304 ECB : ? GIN_MAYBE : GIN_FALSE);
305 GIC 8191 : break;
306 CBC 4046 : case ILikeStrategyNumber:
307 ECB : #ifndef IGNORECASE
308 : elog(ERROR, "cannot handle ~~* with case-sensitive trigrams");
309 : #endif
310 : /* FALL THRU */
311 : case LikeStrategyNumber:
312 : case EqualStrategyNumber:
313 : /* Check if all extracted trigrams are presented. */
314 GIC 4046 : res = GIN_MAYBE;
315 CBC 8182 : for (i = 0; i < nkeys; i++)
316 ECB : {
317 GIC 4156 : if (check[i] == GIN_FALSE)
318 ECB : {
319 GIC 20 : res = GIN_FALSE;
320 CBC 20 : break;
321 ECB : }
322 : }
323 GIC 4046 : break;
324 CBC 2121 : case RegExpICaseStrategyNumber:
325 ECB : #ifndef IGNORECASE
326 : elog(ERROR, "cannot handle ~* with case-sensitive trigrams");
327 : #endif
328 : /* FALL THRU */
329 : case RegExpStrategyNumber:
330 GIC 2121 : if (nkeys < 1)
331 ECB : {
332 : /* Regex processing gave no result: do full index scan */
333 GIC 736 : res = GIN_MAYBE;
334 ECB : }
335 : else
336 : {
337 : /*
338 : * As trigramsMatchGraph implements a monotonic boolean
339 : * function, promoting all GIN_MAYBE keys to GIN_TRUE will
340 : * give a conservative result.
341 : */
342 GIC 1385 : boolcheck = (bool *) palloc(sizeof(bool) * nkeys);
343 CBC 319289 : for (i = 0; i < nkeys; i++)
344 317904 : boolcheck[i] = (check[i] != GIN_FALSE);
345 1385 : if (!trigramsMatchGraph((TrgmPackedGraph *) extra_data[0],
346 ECB : boolcheck))
347 GIC 6 : res = GIN_FALSE;
348 CBC 1385 : pfree(boolcheck);
349 ECB : }
350 GIC 2121 : break;
351 LBC 0 : default:
352 UBC 0 : elog(ERROR, "unrecognized strategy number: %d", strategy);
353 EUB : res = GIN_FALSE; /* keep compiler quiet */
354 : break;
355 : }
356 :
357 : /* All cases served by this function are inexact */
358 GIC 14358 : Assert(res != GIN_TRUE);
359 CBC 14358 : PG_RETURN_GIN_TERNARY_VALUE(res);
360 ECB : }
|