| 1 | /*------------------------------------------------------------------------- |
| 2 | * |
| 3 | * tsgistidx.c |
| 4 | * GiST support functions for tsvector_ops |
| 5 | * |
| 6 | * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group |
| 7 | * |
| 8 | * |
| 9 | * IDENTIFICATION |
| 10 | * src/backend/utils/adt/tsgistidx.c |
| 11 | * |
| 12 | *------------------------------------------------------------------------- |
| 13 | */ |
| 14 | |
| 15 | #include "postgres.h" |
| 16 | |
| 17 | #include "access/gist.h" |
| 18 | #include "access/tuptoaster.h" |
| 19 | #include "port/pg_bitutils.h" |
| 20 | #include "tsearch/ts_utils.h" |
| 21 | #include "utils/builtins.h" |
| 22 | #include "utils/pg_crc.h" |
| 23 | |
| 24 | |
| 25 | #define SIGLENINT 31 /* >121 => key will toast, so it will not work |
| 26 | * !!! */ |
| 27 | |
| 28 | #define SIGLEN ( sizeof(int32) * SIGLENINT ) |
| 29 | #define SIGLENBIT (SIGLEN * BITS_PER_BYTE) |
| 30 | |
| 31 | typedef char BITVEC[SIGLEN]; |
| 32 | typedef char *BITVECP; |
| 33 | |
| 34 | #define LOOPBYTE \ |
| 35 | for(i=0;i<SIGLEN;i++) |
| 36 | |
| 37 | #define GETBYTE(x,i) ( *( (BITVECP)(x) + (int)( (i) / BITS_PER_BYTE ) ) ) |
| 38 | #define GETBITBYTE(x,i) ( ((char)(x)) >> (i) & 0x01 ) |
| 39 | #define CLRBIT(x,i) GETBYTE(x,i) &= ~( 0x01 << ( (i) % BITS_PER_BYTE ) ) |
| 40 | #define SETBIT(x,i) GETBYTE(x,i) |= ( 0x01 << ( (i) % BITS_PER_BYTE ) ) |
| 41 | #define GETBIT(x,i) ( (GETBYTE(x,i) >> ( (i) % BITS_PER_BYTE )) & 0x01 ) |
| 42 | |
| 43 | #define HASHVAL(val) (((unsigned int)(val)) % SIGLENBIT) |
| 44 | #define HASH(sign, val) SETBIT((sign), HASHVAL(val)) |
| 45 | |
| 46 | #define GETENTRY(vec,pos) ((SignTSVector *) DatumGetPointer((vec)->vector[(pos)].key)) |
| 47 | |
| 48 | /* |
| 49 | * type of GiST index key |
| 50 | */ |
| 51 | |
| 52 | typedef struct |
| 53 | { |
| 54 | int32 vl_len_; /* varlena header (do not touch directly!) */ |
| 55 | int32 flag; |
| 56 | char data[FLEXIBLE_ARRAY_MEMBER]; |
| 57 | } SignTSVector; |
| 58 | |
| 59 | #define ARRKEY 0x01 |
| 60 | #define SIGNKEY 0x02 |
| 61 | #define ALLISTRUE 0x04 |
| 62 | |
| 63 | #define ISARRKEY(x) ( ((SignTSVector*)(x))->flag & ARRKEY ) |
| 64 | #define ISSIGNKEY(x) ( ((SignTSVector*)(x))->flag & SIGNKEY ) |
| 65 | #define ISALLTRUE(x) ( ((SignTSVector*)(x))->flag & ALLISTRUE ) |
| 66 | |
| 67 | #define GTHDRSIZE ( VARHDRSZ + sizeof(int32) ) |
| 68 | #define CALCGTSIZE(flag, len) ( GTHDRSIZE + ( ( (flag) & ARRKEY ) ? ((len)*sizeof(int32)) : (((flag) & ALLISTRUE) ? 0 : SIGLEN) ) ) |
| 69 | |
| 70 | #define GETSIGN(x) ( (BITVECP)( (char*)(x)+GTHDRSIZE ) ) |
| 71 | #define GETARR(x) ( (int32*)( (char*)(x)+GTHDRSIZE ) ) |
| 72 | #define ARRNELEM(x) ( ( VARSIZE(x) - GTHDRSIZE )/sizeof(int32) ) |
| 73 | |
| 74 | static int32 sizebitvec(BITVECP sign); |
| 75 | |
| 76 | Datum |
| 77 | gtsvectorin(PG_FUNCTION_ARGS) |
| 78 | { |
| 79 | ereport(ERROR, |
| 80 | (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), |
| 81 | errmsg("gtsvector_in not implemented" ))); |
| 82 | PG_RETURN_DATUM(0); |
| 83 | } |
| 84 | |
| 85 | #define SINGOUTSTR "%d true bits, %d false bits" |
| 86 | #define ARROUTSTR "%d unique words" |
| 87 | #define ( 2*13 ) |
| 88 | |
| 89 | static int outbuf_maxlen = 0; |
| 90 | |
| 91 | Datum |
| 92 | gtsvectorout(PG_FUNCTION_ARGS) |
| 93 | { |
| 94 | SignTSVector *key = (SignTSVector *) PG_DETOAST_DATUM(PG_GETARG_POINTER(0)); |
| 95 | char *outbuf; |
| 96 | |
| 97 | if (outbuf_maxlen == 0) |
| 98 | outbuf_maxlen = 2 * EXTRALEN + Max(strlen(SINGOUTSTR), strlen(ARROUTSTR)) + 1; |
| 99 | outbuf = palloc(outbuf_maxlen); |
| 100 | |
| 101 | if (ISARRKEY(key)) |
| 102 | sprintf(outbuf, ARROUTSTR, (int) ARRNELEM(key)); |
| 103 | else |
| 104 | { |
| 105 | int cnttrue = (ISALLTRUE(key)) ? SIGLENBIT : sizebitvec(GETSIGN(key)); |
| 106 | |
| 107 | sprintf(outbuf, SINGOUTSTR, cnttrue, (int) SIGLENBIT - cnttrue); |
| 108 | } |
| 109 | |
| 110 | PG_FREE_IF_COPY(key, 0); |
| 111 | PG_RETURN_POINTER(outbuf); |
| 112 | } |
| 113 | |
| 114 | static int |
| 115 | compareint(const void *va, const void *vb) |
| 116 | { |
| 117 | int32 a = *((const int32 *) va); |
| 118 | int32 b = *((const int32 *) vb); |
| 119 | |
| 120 | if (a == b) |
| 121 | return 0; |
| 122 | return (a > b) ? 1 : -1; |
| 123 | } |
| 124 | |
| 125 | /* |
| 126 | * Removes duplicates from an array of int32. 'l' is |
| 127 | * size of the input array. Returns the new size of the array. |
| 128 | */ |
| 129 | static int |
| 130 | uniqueint(int32 *a, int32 l) |
| 131 | { |
| 132 | int32 *ptr, |
| 133 | *res; |
| 134 | |
| 135 | if (l <= 1) |
| 136 | return l; |
| 137 | |
| 138 | ptr = res = a; |
| 139 | |
| 140 | qsort((void *) a, l, sizeof(int32), compareint); |
| 141 | |
| 142 | while (ptr - a < l) |
| 143 | if (*ptr != *res) |
| 144 | *(++res) = *ptr++; |
| 145 | else |
| 146 | ptr++; |
| 147 | return res + 1 - a; |
| 148 | } |
| 149 | |
| 150 | static void |
| 151 | makesign(BITVECP sign, SignTSVector *a) |
| 152 | { |
| 153 | int32 k, |
| 154 | len = ARRNELEM(a); |
| 155 | int32 *ptr = GETARR(a); |
| 156 | |
| 157 | MemSet((void *) sign, 0, sizeof(BITVEC)); |
| 158 | for (k = 0; k < len; k++) |
| 159 | HASH(sign, ptr[k]); |
| 160 | } |
| 161 | |
| 162 | Datum |
| 163 | gtsvector_compress(PG_FUNCTION_ARGS) |
| 164 | { |
| 165 | GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0); |
| 166 | GISTENTRY *retval = entry; |
| 167 | |
| 168 | if (entry->leafkey) |
| 169 | { /* tsvector */ |
| 170 | SignTSVector *res; |
| 171 | TSVector val = DatumGetTSVector(entry->key); |
| 172 | int32 len; |
| 173 | int32 *arr; |
| 174 | WordEntry *ptr = ARRPTR(val); |
| 175 | char *words = STRPTR(val); |
| 176 | |
| 177 | len = CALCGTSIZE(ARRKEY, val->size); |
| 178 | res = (SignTSVector *) palloc(len); |
| 179 | SET_VARSIZE(res, len); |
| 180 | res->flag = ARRKEY; |
| 181 | arr = GETARR(res); |
| 182 | len = val->size; |
| 183 | while (len--) |
| 184 | { |
| 185 | pg_crc32 c; |
| 186 | |
| 187 | INIT_LEGACY_CRC32(c); |
| 188 | COMP_LEGACY_CRC32(c, words + ptr->pos, ptr->len); |
| 189 | FIN_LEGACY_CRC32(c); |
| 190 | |
| 191 | *arr = *(int32 *) &c; |
| 192 | arr++; |
| 193 | ptr++; |
| 194 | } |
| 195 | |
| 196 | len = uniqueint(GETARR(res), val->size); |
| 197 | if (len != val->size) |
| 198 | { |
| 199 | /* |
| 200 | * there is a collision of hash-function; len is always less than |
| 201 | * val->size |
| 202 | */ |
| 203 | len = CALCGTSIZE(ARRKEY, len); |
| 204 | res = (SignTSVector *) repalloc((void *) res, len); |
| 205 | SET_VARSIZE(res, len); |
| 206 | } |
| 207 | |
| 208 | /* make signature, if array is too long */ |
| 209 | if (VARSIZE(res) > TOAST_INDEX_TARGET) |
| 210 | { |
| 211 | SignTSVector *ressign; |
| 212 | |
| 213 | len = CALCGTSIZE(SIGNKEY, 0); |
| 214 | ressign = (SignTSVector *) palloc(len); |
| 215 | SET_VARSIZE(ressign, len); |
| 216 | ressign->flag = SIGNKEY; |
| 217 | makesign(GETSIGN(ressign), res); |
| 218 | res = ressign; |
| 219 | } |
| 220 | |
| 221 | retval = (GISTENTRY *) palloc(sizeof(GISTENTRY)); |
| 222 | gistentryinit(*retval, PointerGetDatum(res), |
| 223 | entry->rel, entry->page, |
| 224 | entry->offset, false); |
| 225 | } |
| 226 | else if (ISSIGNKEY(DatumGetPointer(entry->key)) && |
| 227 | !ISALLTRUE(DatumGetPointer(entry->key))) |
| 228 | { |
| 229 | int32 i, |
| 230 | len; |
| 231 | SignTSVector *res; |
| 232 | BITVECP sign = GETSIGN(DatumGetPointer(entry->key)); |
| 233 | |
| 234 | LOOPBYTE |
| 235 | { |
| 236 | if ((sign[i] & 0xff) != 0xff) |
| 237 | PG_RETURN_POINTER(retval); |
| 238 | } |
| 239 | |
| 240 | len = CALCGTSIZE(SIGNKEY | ALLISTRUE, 0); |
| 241 | res = (SignTSVector *) palloc(len); |
| 242 | SET_VARSIZE(res, len); |
| 243 | res->flag = SIGNKEY | ALLISTRUE; |
| 244 | |
| 245 | retval = (GISTENTRY *) palloc(sizeof(GISTENTRY)); |
| 246 | gistentryinit(*retval, PointerGetDatum(res), |
| 247 | entry->rel, entry->page, |
| 248 | entry->offset, false); |
| 249 | } |
| 250 | PG_RETURN_POINTER(retval); |
| 251 | } |
| 252 | |
| 253 | Datum |
| 254 | gtsvector_decompress(PG_FUNCTION_ARGS) |
| 255 | { |
| 256 | /* |
| 257 | * We need to detoast the stored value, because the other gtsvector |
| 258 | * support functions don't cope with toasted values. |
| 259 | */ |
| 260 | GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0); |
| 261 | SignTSVector *key = (SignTSVector *) PG_DETOAST_DATUM(entry->key); |
| 262 | |
| 263 | if (key != (SignTSVector *) DatumGetPointer(entry->key)) |
| 264 | { |
| 265 | GISTENTRY *retval = (GISTENTRY *) palloc(sizeof(GISTENTRY)); |
| 266 | |
| 267 | gistentryinit(*retval, PointerGetDatum(key), |
| 268 | entry->rel, entry->page, |
| 269 | entry->offset, false); |
| 270 | |
| 271 | PG_RETURN_POINTER(retval); |
| 272 | } |
| 273 | |
| 274 | PG_RETURN_POINTER(entry); |
| 275 | } |
| 276 | |
| 277 | typedef struct |
| 278 | { |
| 279 | int32 *arrb; |
| 280 | int32 *arre; |
| 281 | } CHKVAL; |
| 282 | |
| 283 | /* |
| 284 | * is there value 'val' in array or not ? |
| 285 | */ |
| 286 | static bool |
| 287 | checkcondition_arr(void *checkval, QueryOperand *val, ExecPhraseData *data) |
| 288 | { |
| 289 | int32 *StopLow = ((CHKVAL *) checkval)->arrb; |
| 290 | int32 *StopHigh = ((CHKVAL *) checkval)->arre; |
| 291 | int32 *StopMiddle; |
| 292 | |
| 293 | /* Loop invariant: StopLow <= val < StopHigh */ |
| 294 | |
| 295 | /* |
| 296 | * we are not able to find a prefix by hash value |
| 297 | */ |
| 298 | if (val->prefix) |
| 299 | return true; |
| 300 | |
| 301 | while (StopLow < StopHigh) |
| 302 | { |
| 303 | StopMiddle = StopLow + (StopHigh - StopLow) / 2; |
| 304 | if (*StopMiddle == val->valcrc) |
| 305 | return true; |
| 306 | else if (*StopMiddle < val->valcrc) |
| 307 | StopLow = StopMiddle + 1; |
| 308 | else |
| 309 | StopHigh = StopMiddle; |
| 310 | } |
| 311 | |
| 312 | return false; |
| 313 | } |
| 314 | |
| 315 | static bool |
| 316 | checkcondition_bit(void *checkval, QueryOperand *val, ExecPhraseData *data) |
| 317 | { |
| 318 | /* |
| 319 | * we are not able to find a prefix in signature tree |
| 320 | */ |
| 321 | if (val->prefix) |
| 322 | return true; |
| 323 | return GETBIT(checkval, HASHVAL(val->valcrc)); |
| 324 | } |
| 325 | |
| 326 | Datum |
| 327 | gtsvector_consistent(PG_FUNCTION_ARGS) |
| 328 | { |
| 329 | GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0); |
| 330 | TSQuery query = PG_GETARG_TSQUERY(1); |
| 331 | |
| 332 | /* StrategyNumber strategy = (StrategyNumber) PG_GETARG_UINT16(2); */ |
| 333 | /* Oid subtype = PG_GETARG_OID(3); */ |
| 334 | bool *recheck = (bool *) PG_GETARG_POINTER(4); |
| 335 | SignTSVector *key = (SignTSVector *) DatumGetPointer(entry->key); |
| 336 | |
| 337 | /* All cases served by this function are inexact */ |
| 338 | *recheck = true; |
| 339 | |
| 340 | if (!query->size) |
| 341 | PG_RETURN_BOOL(false); |
| 342 | |
| 343 | if (ISSIGNKEY(key)) |
| 344 | { |
| 345 | if (ISALLTRUE(key)) |
| 346 | PG_RETURN_BOOL(true); |
| 347 | |
| 348 | /* since signature is lossy, cannot specify CALC_NOT here */ |
| 349 | PG_RETURN_BOOL(TS_execute(GETQUERY(query), |
| 350 | (void *) GETSIGN(key), |
| 351 | TS_EXEC_PHRASE_NO_POS, |
| 352 | checkcondition_bit)); |
| 353 | } |
| 354 | else |
| 355 | { /* only leaf pages */ |
| 356 | CHKVAL chkval; |
| 357 | |
| 358 | chkval.arrb = GETARR(key); |
| 359 | chkval.arre = chkval.arrb + ARRNELEM(key); |
| 360 | PG_RETURN_BOOL(TS_execute(GETQUERY(query), |
| 361 | (void *) &chkval, |
| 362 | TS_EXEC_PHRASE_NO_POS | TS_EXEC_CALC_NOT, |
| 363 | checkcondition_arr)); |
| 364 | } |
| 365 | } |
| 366 | |
| 367 | static int32 |
| 368 | unionkey(BITVECP sbase, SignTSVector *add) |
| 369 | { |
| 370 | int32 i; |
| 371 | |
| 372 | if (ISSIGNKEY(add)) |
| 373 | { |
| 374 | BITVECP sadd = GETSIGN(add); |
| 375 | |
| 376 | if (ISALLTRUE(add)) |
| 377 | return 1; |
| 378 | |
| 379 | LOOPBYTE |
| 380 | sbase[i] |= sadd[i]; |
| 381 | } |
| 382 | else |
| 383 | { |
| 384 | int32 *ptr = GETARR(add); |
| 385 | |
| 386 | for (i = 0; i < ARRNELEM(add); i++) |
| 387 | HASH(sbase, ptr[i]); |
| 388 | } |
| 389 | return 0; |
| 390 | } |
| 391 | |
| 392 | |
| 393 | Datum |
| 394 | gtsvector_union(PG_FUNCTION_ARGS) |
| 395 | { |
| 396 | GistEntryVector *entryvec = (GistEntryVector *) PG_GETARG_POINTER(0); |
| 397 | int *size = (int *) PG_GETARG_POINTER(1); |
| 398 | BITVEC base; |
| 399 | int32 i, |
| 400 | len; |
| 401 | int32 flag = 0; |
| 402 | SignTSVector *result; |
| 403 | |
| 404 | MemSet((void *) base, 0, sizeof(BITVEC)); |
| 405 | for (i = 0; i < entryvec->n; i++) |
| 406 | { |
| 407 | if (unionkey(base, GETENTRY(entryvec, i))) |
| 408 | { |
| 409 | flag = ALLISTRUE; |
| 410 | break; |
| 411 | } |
| 412 | } |
| 413 | |
| 414 | flag |= SIGNKEY; |
| 415 | len = CALCGTSIZE(flag, 0); |
| 416 | result = (SignTSVector *) palloc(len); |
| 417 | *size = len; |
| 418 | SET_VARSIZE(result, len); |
| 419 | result->flag = flag; |
| 420 | if (!ISALLTRUE(result)) |
| 421 | memcpy((void *) GETSIGN(result), (void *) base, sizeof(BITVEC)); |
| 422 | |
| 423 | PG_RETURN_POINTER(result); |
| 424 | } |
| 425 | |
| 426 | Datum |
| 427 | gtsvector_same(PG_FUNCTION_ARGS) |
| 428 | { |
| 429 | SignTSVector *a = (SignTSVector *) PG_GETARG_POINTER(0); |
| 430 | SignTSVector *b = (SignTSVector *) PG_GETARG_POINTER(1); |
| 431 | bool *result = (bool *) PG_GETARG_POINTER(2); |
| 432 | |
| 433 | if (ISSIGNKEY(a)) |
| 434 | { /* then b also ISSIGNKEY */ |
| 435 | if (ISALLTRUE(a) && ISALLTRUE(b)) |
| 436 | *result = true; |
| 437 | else if (ISALLTRUE(a)) |
| 438 | *result = false; |
| 439 | else if (ISALLTRUE(b)) |
| 440 | *result = false; |
| 441 | else |
| 442 | { |
| 443 | int32 i; |
| 444 | BITVECP sa = GETSIGN(a), |
| 445 | sb = GETSIGN(b); |
| 446 | |
| 447 | *result = true; |
| 448 | LOOPBYTE |
| 449 | { |
| 450 | if (sa[i] != sb[i]) |
| 451 | { |
| 452 | *result = false; |
| 453 | break; |
| 454 | } |
| 455 | } |
| 456 | } |
| 457 | } |
| 458 | else |
| 459 | { /* a and b ISARRKEY */ |
| 460 | int32 lena = ARRNELEM(a), |
| 461 | lenb = ARRNELEM(b); |
| 462 | |
| 463 | if (lena != lenb) |
| 464 | *result = false; |
| 465 | else |
| 466 | { |
| 467 | int32 *ptra = GETARR(a), |
| 468 | *ptrb = GETARR(b); |
| 469 | int32 i; |
| 470 | |
| 471 | *result = true; |
| 472 | for (i = 0; i < lena; i++) |
| 473 | if (ptra[i] != ptrb[i]) |
| 474 | { |
| 475 | *result = false; |
| 476 | break; |
| 477 | } |
| 478 | } |
| 479 | } |
| 480 | |
| 481 | PG_RETURN_POINTER(result); |
| 482 | } |
| 483 | |
| 484 | static int32 |
| 485 | sizebitvec(BITVECP sign) |
| 486 | { |
| 487 | return pg_popcount(sign, SIGLEN); |
| 488 | } |
| 489 | |
| 490 | static int |
| 491 | hemdistsign(BITVECP a, BITVECP b) |
| 492 | { |
| 493 | int i, |
| 494 | diff, |
| 495 | dist = 0; |
| 496 | |
| 497 | LOOPBYTE |
| 498 | { |
| 499 | diff = (unsigned char) (a[i] ^ b[i]); |
| 500 | /* Using the popcount functions here isn't likely to win */ |
| 501 | dist += pg_number_of_ones[diff]; |
| 502 | } |
| 503 | return dist; |
| 504 | } |
| 505 | |
| 506 | static int |
| 507 | hemdist(SignTSVector *a, SignTSVector *b) |
| 508 | { |
| 509 | if (ISALLTRUE(a)) |
| 510 | { |
| 511 | if (ISALLTRUE(b)) |
| 512 | return 0; |
| 513 | else |
| 514 | return SIGLENBIT - sizebitvec(GETSIGN(b)); |
| 515 | } |
| 516 | else if (ISALLTRUE(b)) |
| 517 | return SIGLENBIT - sizebitvec(GETSIGN(a)); |
| 518 | |
| 519 | return hemdistsign(GETSIGN(a), GETSIGN(b)); |
| 520 | } |
| 521 | |
| 522 | Datum |
| 523 | gtsvector_penalty(PG_FUNCTION_ARGS) |
| 524 | { |
| 525 | GISTENTRY *origentry = (GISTENTRY *) PG_GETARG_POINTER(0); /* always ISSIGNKEY */ |
| 526 | GISTENTRY *newentry = (GISTENTRY *) PG_GETARG_POINTER(1); |
| 527 | float *penalty = (float *) PG_GETARG_POINTER(2); |
| 528 | SignTSVector *origval = (SignTSVector *) DatumGetPointer(origentry->key); |
| 529 | SignTSVector *newval = (SignTSVector *) DatumGetPointer(newentry->key); |
| 530 | BITVECP orig = GETSIGN(origval); |
| 531 | |
| 532 | *penalty = 0.0; |
| 533 | |
| 534 | if (ISARRKEY(newval)) |
| 535 | { |
| 536 | BITVEC sign; |
| 537 | |
| 538 | makesign(sign, newval); |
| 539 | |
| 540 | if (ISALLTRUE(origval)) |
| 541 | *penalty = ((float) (SIGLENBIT - sizebitvec(sign))) / (float) (SIGLENBIT + 1); |
| 542 | else |
| 543 | *penalty = hemdistsign(sign, orig); |
| 544 | } |
| 545 | else |
| 546 | *penalty = hemdist(origval, newval); |
| 547 | PG_RETURN_POINTER(penalty); |
| 548 | } |
| 549 | |
| 550 | typedef struct |
| 551 | { |
| 552 | bool allistrue; |
| 553 | BITVEC sign; |
| 554 | } CACHESIGN; |
| 555 | |
| 556 | static void |
| 557 | fillcache(CACHESIGN *item, SignTSVector *key) |
| 558 | { |
| 559 | item->allistrue = false; |
| 560 | if (ISARRKEY(key)) |
| 561 | makesign(item->sign, key); |
| 562 | else if (ISALLTRUE(key)) |
| 563 | item->allistrue = true; |
| 564 | else |
| 565 | memcpy((void *) item->sign, (void *) GETSIGN(key), sizeof(BITVEC)); |
| 566 | } |
| 567 | |
| 568 | #define WISH_F(a,b,c) (double)( -(double)(((a)-(b))*((a)-(b))*((a)-(b)))*(c) ) |
| 569 | typedef struct |
| 570 | { |
| 571 | OffsetNumber pos; |
| 572 | int32 cost; |
| 573 | } SPLITCOST; |
| 574 | |
| 575 | static int |
| 576 | comparecost(const void *va, const void *vb) |
| 577 | { |
| 578 | const SPLITCOST *a = (const SPLITCOST *) va; |
| 579 | const SPLITCOST *b = (const SPLITCOST *) vb; |
| 580 | |
| 581 | if (a->cost == b->cost) |
| 582 | return 0; |
| 583 | else |
| 584 | return (a->cost > b->cost) ? 1 : -1; |
| 585 | } |
| 586 | |
| 587 | |
| 588 | static int |
| 589 | hemdistcache(CACHESIGN *a, CACHESIGN *b) |
| 590 | { |
| 591 | if (a->allistrue) |
| 592 | { |
| 593 | if (b->allistrue) |
| 594 | return 0; |
| 595 | else |
| 596 | return SIGLENBIT - sizebitvec(b->sign); |
| 597 | } |
| 598 | else if (b->allistrue) |
| 599 | return SIGLENBIT - sizebitvec(a->sign); |
| 600 | |
| 601 | return hemdistsign(a->sign, b->sign); |
| 602 | } |
| 603 | |
| 604 | Datum |
| 605 | gtsvector_picksplit(PG_FUNCTION_ARGS) |
| 606 | { |
| 607 | GistEntryVector *entryvec = (GistEntryVector *) PG_GETARG_POINTER(0); |
| 608 | GIST_SPLITVEC *v = (GIST_SPLITVEC *) PG_GETARG_POINTER(1); |
| 609 | OffsetNumber k, |
| 610 | j; |
| 611 | SignTSVector *datum_l, |
| 612 | *datum_r; |
| 613 | BITVECP union_l, |
| 614 | union_r; |
| 615 | int32 size_alpha, |
| 616 | size_beta; |
| 617 | int32 size_waste, |
| 618 | waste = -1; |
| 619 | int32 nbytes; |
| 620 | OffsetNumber seed_1 = 0, |
| 621 | seed_2 = 0; |
| 622 | OffsetNumber *left, |
| 623 | *right; |
| 624 | OffsetNumber maxoff; |
| 625 | BITVECP ptr; |
| 626 | int i; |
| 627 | CACHESIGN *cache; |
| 628 | SPLITCOST *costvector; |
| 629 | |
| 630 | maxoff = entryvec->n - 2; |
| 631 | nbytes = (maxoff + 2) * sizeof(OffsetNumber); |
| 632 | v->spl_left = (OffsetNumber *) palloc(nbytes); |
| 633 | v->spl_right = (OffsetNumber *) palloc(nbytes); |
| 634 | |
| 635 | cache = (CACHESIGN *) palloc(sizeof(CACHESIGN) * (maxoff + 2)); |
| 636 | fillcache(&cache[FirstOffsetNumber], GETENTRY(entryvec, FirstOffsetNumber)); |
| 637 | |
| 638 | for (k = FirstOffsetNumber; k < maxoff; k = OffsetNumberNext(k)) |
| 639 | { |
| 640 | for (j = OffsetNumberNext(k); j <= maxoff; j = OffsetNumberNext(j)) |
| 641 | { |
| 642 | if (k == FirstOffsetNumber) |
| 643 | fillcache(&cache[j], GETENTRY(entryvec, j)); |
| 644 | |
| 645 | size_waste = hemdistcache(&(cache[j]), &(cache[k])); |
| 646 | if (size_waste > waste) |
| 647 | { |
| 648 | waste = size_waste; |
| 649 | seed_1 = k; |
| 650 | seed_2 = j; |
| 651 | } |
| 652 | } |
| 653 | } |
| 654 | |
| 655 | left = v->spl_left; |
| 656 | v->spl_nleft = 0; |
| 657 | right = v->spl_right; |
| 658 | v->spl_nright = 0; |
| 659 | |
| 660 | if (seed_1 == 0 || seed_2 == 0) |
| 661 | { |
| 662 | seed_1 = 1; |
| 663 | seed_2 = 2; |
| 664 | } |
| 665 | |
| 666 | /* form initial .. */ |
| 667 | if (cache[seed_1].allistrue) |
| 668 | { |
| 669 | datum_l = (SignTSVector *) palloc(CALCGTSIZE(SIGNKEY | ALLISTRUE, 0)); |
| 670 | SET_VARSIZE(datum_l, CALCGTSIZE(SIGNKEY | ALLISTRUE, 0)); |
| 671 | datum_l->flag = SIGNKEY | ALLISTRUE; |
| 672 | } |
| 673 | else |
| 674 | { |
| 675 | datum_l = (SignTSVector *) palloc(CALCGTSIZE(SIGNKEY, 0)); |
| 676 | SET_VARSIZE(datum_l, CALCGTSIZE(SIGNKEY, 0)); |
| 677 | datum_l->flag = SIGNKEY; |
| 678 | memcpy((void *) GETSIGN(datum_l), (void *) cache[seed_1].sign, sizeof(BITVEC)); |
| 679 | } |
| 680 | if (cache[seed_2].allistrue) |
| 681 | { |
| 682 | datum_r = (SignTSVector *) palloc(CALCGTSIZE(SIGNKEY | ALLISTRUE, 0)); |
| 683 | SET_VARSIZE(datum_r, CALCGTSIZE(SIGNKEY | ALLISTRUE, 0)); |
| 684 | datum_r->flag = SIGNKEY | ALLISTRUE; |
| 685 | } |
| 686 | else |
| 687 | { |
| 688 | datum_r = (SignTSVector *) palloc(CALCGTSIZE(SIGNKEY, 0)); |
| 689 | SET_VARSIZE(datum_r, CALCGTSIZE(SIGNKEY, 0)); |
| 690 | datum_r->flag = SIGNKEY; |
| 691 | memcpy((void *) GETSIGN(datum_r), (void *) cache[seed_2].sign, sizeof(BITVEC)); |
| 692 | } |
| 693 | |
| 694 | union_l = GETSIGN(datum_l); |
| 695 | union_r = GETSIGN(datum_r); |
| 696 | maxoff = OffsetNumberNext(maxoff); |
| 697 | fillcache(&cache[maxoff], GETENTRY(entryvec, maxoff)); |
| 698 | /* sort before ... */ |
| 699 | costvector = (SPLITCOST *) palloc(sizeof(SPLITCOST) * maxoff); |
| 700 | for (j = FirstOffsetNumber; j <= maxoff; j = OffsetNumberNext(j)) |
| 701 | { |
| 702 | costvector[j - 1].pos = j; |
| 703 | size_alpha = hemdistcache(&(cache[seed_1]), &(cache[j])); |
| 704 | size_beta = hemdistcache(&(cache[seed_2]), &(cache[j])); |
| 705 | costvector[j - 1].cost = Abs(size_alpha - size_beta); |
| 706 | } |
| 707 | qsort((void *) costvector, maxoff, sizeof(SPLITCOST), comparecost); |
| 708 | |
| 709 | for (k = 0; k < maxoff; k++) |
| 710 | { |
| 711 | j = costvector[k].pos; |
| 712 | if (j == seed_1) |
| 713 | { |
| 714 | *left++ = j; |
| 715 | v->spl_nleft++; |
| 716 | continue; |
| 717 | } |
| 718 | else if (j == seed_2) |
| 719 | { |
| 720 | *right++ = j; |
| 721 | v->spl_nright++; |
| 722 | continue; |
| 723 | } |
| 724 | |
| 725 | if (ISALLTRUE(datum_l) || cache[j].allistrue) |
| 726 | { |
| 727 | if (ISALLTRUE(datum_l) && cache[j].allistrue) |
| 728 | size_alpha = 0; |
| 729 | else |
| 730 | size_alpha = SIGLENBIT - sizebitvec( |
| 731 | (cache[j].allistrue) ? GETSIGN(datum_l) : GETSIGN(cache[j].sign) |
| 732 | ); |
| 733 | } |
| 734 | else |
| 735 | size_alpha = hemdistsign(cache[j].sign, GETSIGN(datum_l)); |
| 736 | |
| 737 | if (ISALLTRUE(datum_r) || cache[j].allistrue) |
| 738 | { |
| 739 | if (ISALLTRUE(datum_r) && cache[j].allistrue) |
| 740 | size_beta = 0; |
| 741 | else |
| 742 | size_beta = SIGLENBIT - sizebitvec( |
| 743 | (cache[j].allistrue) ? GETSIGN(datum_r) : GETSIGN(cache[j].sign) |
| 744 | ); |
| 745 | } |
| 746 | else |
| 747 | size_beta = hemdistsign(cache[j].sign, GETSIGN(datum_r)); |
| 748 | |
| 749 | if (size_alpha < size_beta + WISH_F(v->spl_nleft, v->spl_nright, 0.1)) |
| 750 | { |
| 751 | if (ISALLTRUE(datum_l) || cache[j].allistrue) |
| 752 | { |
| 753 | if (!ISALLTRUE(datum_l)) |
| 754 | MemSet((void *) GETSIGN(datum_l), 0xff, sizeof(BITVEC)); |
| 755 | } |
| 756 | else |
| 757 | { |
| 758 | ptr = cache[j].sign; |
| 759 | LOOPBYTE |
| 760 | union_l[i] |= ptr[i]; |
| 761 | } |
| 762 | *left++ = j; |
| 763 | v->spl_nleft++; |
| 764 | } |
| 765 | else |
| 766 | { |
| 767 | if (ISALLTRUE(datum_r) || cache[j].allistrue) |
| 768 | { |
| 769 | if (!ISALLTRUE(datum_r)) |
| 770 | MemSet((void *) GETSIGN(datum_r), 0xff, sizeof(BITVEC)); |
| 771 | } |
| 772 | else |
| 773 | { |
| 774 | ptr = cache[j].sign; |
| 775 | LOOPBYTE |
| 776 | union_r[i] |= ptr[i]; |
| 777 | } |
| 778 | *right++ = j; |
| 779 | v->spl_nright++; |
| 780 | } |
| 781 | } |
| 782 | |
| 783 | *right = *left = FirstOffsetNumber; |
| 784 | v->spl_ldatum = PointerGetDatum(datum_l); |
| 785 | v->spl_rdatum = PointerGetDatum(datum_r); |
| 786 | |
| 787 | PG_RETURN_POINTER(v); |
| 788 | } |
| 789 | |
| 790 | /* |
| 791 | * Formerly, gtsvector_consistent was declared in pg_proc.h with arguments |
| 792 | * that did not match the documented conventions for GiST support functions. |
| 793 | * We fixed that, but we still need a pg_proc entry with the old signature |
| 794 | * to support reloading pre-9.6 contrib/tsearch2 opclass declarations. |
| 795 | * This compatibility function should go away eventually. |
| 796 | */ |
| 797 | Datum |
| 798 | gtsvector_consistent_oldsig(PG_FUNCTION_ARGS) |
| 799 | { |
| 800 | return gtsvector_consistent(fcinfo); |
| 801 | } |
| 802 | |