Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * uuid.c
4 : * Functions for the built-in type "uuid".
5 : *
6 : * Copyright (c) 2007-2025, PostgreSQL Global Development Group
7 : *
8 : * IDENTIFICATION
9 : * src/backend/utils/adt/uuid.c
10 : *
11 : *-------------------------------------------------------------------------
12 : */
13 :
14 : #include "postgres.h"
15 :
16 : #include <time.h> /* for clock_gettime() */
17 :
18 : #include "common/hashfn.h"
19 : #include "lib/hyperloglog.h"
20 : #include "libpq/pqformat.h"
21 : #include "port/pg_bswap.h"
22 : #include "utils/fmgrprotos.h"
23 : #include "utils/guc.h"
24 : #include "utils/sortsupport.h"
25 : #include "utils/timestamp.h"
26 : #include "utils/uuid.h"
27 :
28 : /* helper macros */
29 : #define NS_PER_S INT64CONST(1000000000)
30 : #define NS_PER_MS INT64CONST(1000000)
31 : #define NS_PER_US INT64CONST(1000)
32 :
33 : /*
34 : * UUID version 7 uses 12 bits in "rand_a" to store 1/4096 (or 2^12) fractions of
35 : * sub-millisecond. While most Unix-like platforms provide nanosecond-precision
36 : * timestamps, some systems only offer microsecond precision, limiting us to 10
37 : * bits of sub-millisecond information. For example, on macOS, real time is
38 : * truncated to microseconds. Additionally, MSVC uses the ported version of
39 : * gettimeofday() that returns microsecond precision.
40 : *
41 : * On systems with only 10 bits of sub-millisecond precision, we still use
42 : * 1/4096 parts of a millisecond, but fill lower 2 bits with random numbers
43 : * (see generate_uuidv7() for details).
44 : *
45 : * SUBMS_MINIMAL_STEP_NS defines the minimum number of nanoseconds that guarantees
46 : * an increase in the UUID's clock precision.
47 : */
48 : #if defined(__darwin__) || defined(_MSC_VER)
49 : #define SUBMS_MINIMAL_STEP_BITS 10
50 : #else
51 : #define SUBMS_MINIMAL_STEP_BITS 12
52 : #endif
53 : #define SUBMS_BITS 12
54 : #define SUBMS_MINIMAL_STEP_NS ((NS_PER_MS / (1 << SUBMS_MINIMAL_STEP_BITS)) + 1)
55 :
56 : /* sortsupport for uuid */
57 : typedef struct
58 : {
59 : int64 input_count; /* number of non-null values seen */
60 : bool estimating; /* true if estimating cardinality */
61 :
62 : hyperLogLogState abbr_card; /* cardinality estimator */
63 : } uuid_sortsupport_state;
64 :
65 : static void string_to_uuid(const char *source, pg_uuid_t *uuid, Node *escontext);
66 : static int uuid_internal_cmp(const pg_uuid_t *arg1, const pg_uuid_t *arg2);
67 : static int uuid_fast_cmp(Datum x, Datum y, SortSupport ssup);
68 : static bool uuid_abbrev_abort(int memtupcount, SortSupport ssup);
69 : static Datum uuid_abbrev_convert(Datum original, SortSupport ssup);
70 : static inline void uuid_set_version(pg_uuid_t *uuid, unsigned char version);
71 : static inline int64 get_real_time_ns_ascending();
72 :
73 : Datum
74 585920 : uuid_in(PG_FUNCTION_ARGS)
75 : {
76 585920 : char *uuid_str = PG_GETARG_CSTRING(0);
77 : pg_uuid_t *uuid;
78 :
79 585920 : uuid = (pg_uuid_t *) palloc(sizeof(*uuid));
80 585920 : string_to_uuid(uuid_str, uuid, fcinfo->context);
81 585884 : PG_RETURN_UUID_P(uuid);
82 : }
83 :
84 : Datum
85 4704 : uuid_out(PG_FUNCTION_ARGS)
86 : {
87 4704 : pg_uuid_t *uuid = PG_GETARG_UUID_P(0);
88 : static const char hex_chars[] = "0123456789abcdef";
89 : char *buf,
90 : *p;
91 : int i;
92 :
93 : /* counts for the four hyphens and the zero-terminator */
94 4704 : buf = palloc(2 * UUID_LEN + 5);
95 4704 : p = buf;
96 79968 : for (i = 0; i < UUID_LEN; i++)
97 : {
98 : int hi;
99 : int lo;
100 :
101 : /*
102 : * We print uuid values as a string of 8, 4, 4, 4, and then 12
103 : * hexadecimal characters, with each group is separated by a hyphen
104 : * ("-"). Therefore, add the hyphens at the appropriate places here.
105 : */
106 75264 : if (i == 4 || i == 6 || i == 8 || i == 10)
107 18816 : *p++ = '-';
108 :
109 75264 : hi = uuid->data[i] >> 4;
110 75264 : lo = uuid->data[i] & 0x0F;
111 :
112 75264 : *p++ = hex_chars[hi];
113 75264 : *p++ = hex_chars[lo];
114 : }
115 4704 : *p = '\0';
116 :
117 4704 : PG_RETURN_CSTRING(buf);
118 : }
119 :
120 : /*
121 : * We allow UUIDs as a series of 32 hexadecimal digits with an optional dash
122 : * after each group of 4 hexadecimal digits, and optionally surrounded by {}.
123 : * (The canonical format 8x-4x-4x-4x-12x, where "nx" means n hexadecimal
124 : * digits, is the only one used for output.)
125 : */
126 : static void
127 585920 : string_to_uuid(const char *source, pg_uuid_t *uuid, Node *escontext)
128 : {
129 585920 : const char *src = source;
130 585920 : bool braces = false;
131 : int i;
132 :
133 585920 : if (src[0] == '{')
134 : {
135 24 : src++;
136 24 : braces = true;
137 : }
138 :
139 9960226 : for (i = 0; i < UUID_LEN; i++)
140 : {
141 : char str_buf[3];
142 :
143 9374342 : if (src[0] == '\0' || src[1] == '\0')
144 36 : goto syntax_error;
145 9374330 : memcpy(str_buf, src, 2);
146 9374330 : if (!isxdigit((unsigned char) str_buf[0]) ||
147 9374318 : !isxdigit((unsigned char) str_buf[1]))
148 24 : goto syntax_error;
149 :
150 9374306 : str_buf[2] = '\0';
151 9374306 : uuid->data[i] = (unsigned char) strtoul(str_buf, NULL, 16);
152 9374306 : src += 2;
153 9374306 : if (src[0] == '-' && (i % 2) == 1 && i < UUID_LEN - 1)
154 1935434 : src++;
155 : }
156 :
157 585884 : if (braces)
158 : {
159 18 : if (*src != '}')
160 6 : goto syntax_error;
161 12 : src++;
162 : }
163 :
164 585878 : if (*src != '\0')
165 6 : goto syntax_error;
166 :
167 585872 : return;
168 :
169 48 : syntax_error:
170 48 : ereturn(escontext,,
171 : (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION),
172 : errmsg("invalid input syntax for type %s: \"%s\"",
173 : "uuid", source)));
174 : }
175 :
176 : Datum
177 0 : uuid_recv(PG_FUNCTION_ARGS)
178 : {
179 0 : StringInfo buffer = (StringInfo) PG_GETARG_POINTER(0);
180 : pg_uuid_t *uuid;
181 :
182 0 : uuid = (pg_uuid_t *) palloc(UUID_LEN);
183 0 : memcpy(uuid->data, pq_getmsgbytes(buffer, UUID_LEN), UUID_LEN);
184 0 : PG_RETURN_POINTER(uuid);
185 : }
186 :
187 : Datum
188 0 : uuid_send(PG_FUNCTION_ARGS)
189 : {
190 0 : pg_uuid_t *uuid = PG_GETARG_UUID_P(0);
191 : StringInfoData buffer;
192 :
193 0 : pq_begintypsend(&buffer);
194 0 : pq_sendbytes(&buffer, uuid->data, UUID_LEN);
195 0 : PG_RETURN_BYTEA_P(pq_endtypsend(&buffer));
196 : }
197 :
198 : /* internal uuid compare function */
199 : static int
200 41777838 : uuid_internal_cmp(const pg_uuid_t *arg1, const pg_uuid_t *arg2)
201 : {
202 41777838 : return memcmp(arg1->data, arg2->data, UUID_LEN);
203 : }
204 :
205 : Datum
206 83806 : uuid_lt(PG_FUNCTION_ARGS)
207 : {
208 83806 : pg_uuid_t *arg1 = PG_GETARG_UUID_P(0);
209 83806 : pg_uuid_t *arg2 = PG_GETARG_UUID_P(1);
210 :
211 83806 : PG_RETURN_BOOL(uuid_internal_cmp(arg1, arg2) < 0);
212 : }
213 :
214 : Datum
215 17046 : uuid_le(PG_FUNCTION_ARGS)
216 : {
217 17046 : pg_uuid_t *arg1 = PG_GETARG_UUID_P(0);
218 17046 : pg_uuid_t *arg2 = PG_GETARG_UUID_P(1);
219 :
220 17046 : PG_RETURN_BOOL(uuid_internal_cmp(arg1, arg2) <= 0);
221 : }
222 :
223 : Datum
224 154444 : uuid_eq(PG_FUNCTION_ARGS)
225 : {
226 154444 : pg_uuid_t *arg1 = PG_GETARG_UUID_P(0);
227 154444 : pg_uuid_t *arg2 = PG_GETARG_UUID_P(1);
228 :
229 154444 : PG_RETURN_BOOL(uuid_internal_cmp(arg1, arg2) == 0);
230 : }
231 :
232 : Datum
233 12554 : uuid_ge(PG_FUNCTION_ARGS)
234 : {
235 12554 : pg_uuid_t *arg1 = PG_GETARG_UUID_P(0);
236 12554 : pg_uuid_t *arg2 = PG_GETARG_UUID_P(1);
237 :
238 12554 : PG_RETURN_BOOL(uuid_internal_cmp(arg1, arg2) >= 0);
239 : }
240 :
241 : Datum
242 16286 : uuid_gt(PG_FUNCTION_ARGS)
243 : {
244 16286 : pg_uuid_t *arg1 = PG_GETARG_UUID_P(0);
245 16286 : pg_uuid_t *arg2 = PG_GETARG_UUID_P(1);
246 :
247 16286 : PG_RETURN_BOOL(uuid_internal_cmp(arg1, arg2) > 0);
248 : }
249 :
250 : Datum
251 18 : uuid_ne(PG_FUNCTION_ARGS)
252 : {
253 18 : pg_uuid_t *arg1 = PG_GETARG_UUID_P(0);
254 18 : pg_uuid_t *arg2 = PG_GETARG_UUID_P(1);
255 :
256 18 : PG_RETURN_BOOL(uuid_internal_cmp(arg1, arg2) != 0);
257 : }
258 :
259 : /* handler for btree index operator */
260 : Datum
261 9308 : uuid_cmp(PG_FUNCTION_ARGS)
262 : {
263 9308 : pg_uuid_t *arg1 = PG_GETARG_UUID_P(0);
264 9308 : pg_uuid_t *arg2 = PG_GETARG_UUID_P(1);
265 :
266 9308 : PG_RETURN_INT32(uuid_internal_cmp(arg1, arg2));
267 : }
268 :
269 : /*
270 : * Sort support strategy routine
271 : */
272 : Datum
273 388 : uuid_sortsupport(PG_FUNCTION_ARGS)
274 : {
275 388 : SortSupport ssup = (SortSupport) PG_GETARG_POINTER(0);
276 :
277 388 : ssup->comparator = uuid_fast_cmp;
278 388 : ssup->ssup_extra = NULL;
279 :
280 388 : if (ssup->abbreviate)
281 : {
282 : uuid_sortsupport_state *uss;
283 : MemoryContext oldcontext;
284 :
285 308 : oldcontext = MemoryContextSwitchTo(ssup->ssup_cxt);
286 :
287 308 : uss = palloc(sizeof(uuid_sortsupport_state));
288 308 : uss->input_count = 0;
289 308 : uss->estimating = true;
290 308 : initHyperLogLog(&uss->abbr_card, 10);
291 :
292 308 : ssup->ssup_extra = uss;
293 :
294 308 : ssup->comparator = ssup_datum_unsigned_cmp;
295 308 : ssup->abbrev_converter = uuid_abbrev_convert;
296 308 : ssup->abbrev_abort = uuid_abbrev_abort;
297 308 : ssup->abbrev_full_comparator = uuid_fast_cmp;
298 :
299 308 : MemoryContextSwitchTo(oldcontext);
300 : }
301 :
302 388 : PG_RETURN_VOID();
303 : }
304 :
305 : /*
306 : * SortSupport comparison func
307 : */
308 : static int
309 41484376 : uuid_fast_cmp(Datum x, Datum y, SortSupport ssup)
310 : {
311 41484376 : pg_uuid_t *arg1 = DatumGetUUIDP(x);
312 41484376 : pg_uuid_t *arg2 = DatumGetUUIDP(y);
313 :
314 41484376 : return uuid_internal_cmp(arg1, arg2);
315 : }
316 :
317 : /*
318 : * Callback for estimating effectiveness of abbreviated key optimization.
319 : *
320 : * We pay no attention to the cardinality of the non-abbreviated data, because
321 : * there is no equality fast-path within authoritative uuid comparator.
322 : */
323 : static bool
324 2322 : uuid_abbrev_abort(int memtupcount, SortSupport ssup)
325 : {
326 2322 : uuid_sortsupport_state *uss = ssup->ssup_extra;
327 : double abbr_card;
328 :
329 2322 : if (memtupcount < 10000 || uss->input_count < 10000 || !uss->estimating)
330 2130 : return false;
331 :
332 192 : abbr_card = estimateHyperLogLog(&uss->abbr_card);
333 :
334 : /*
335 : * If we have >100k distinct values, then even if we were sorting many
336 : * billion rows we'd likely still break even, and the penalty of undoing
337 : * that many rows of abbrevs would probably not be worth it. Stop even
338 : * counting at that point.
339 : */
340 192 : if (abbr_card > 100000.0)
341 : {
342 0 : if (trace_sort)
343 0 : elog(LOG,
344 : "uuid_abbrev: estimation ends at cardinality %f"
345 : " after " INT64_FORMAT " values (%d rows)",
346 : abbr_card, uss->input_count, memtupcount);
347 0 : uss->estimating = false;
348 0 : return false;
349 : }
350 :
351 : /*
352 : * Target minimum cardinality is 1 per ~2k of non-null inputs. 0.5 row
353 : * fudge factor allows us to abort earlier on genuinely pathological data
354 : * where we've had exactly one abbreviated value in the first 2k
355 : * (non-null) rows.
356 : */
357 192 : if (abbr_card < uss->input_count / 2000.0 + 0.5)
358 : {
359 96 : if (trace_sort)
360 0 : elog(LOG,
361 : "uuid_abbrev: aborting abbreviation at cardinality %f"
362 : " below threshold %f after " INT64_FORMAT " values (%d rows)",
363 : abbr_card, uss->input_count / 2000.0 + 0.5, uss->input_count,
364 : memtupcount);
365 96 : return true;
366 : }
367 :
368 96 : if (trace_sort)
369 0 : elog(LOG,
370 : "uuid_abbrev: cardinality %f after " INT64_FORMAT
371 : " values (%d rows)", abbr_card, uss->input_count, memtupcount);
372 :
373 96 : return false;
374 : }
375 :
376 : /*
377 : * Conversion routine for sortsupport. Converts original uuid representation
378 : * to abbreviated key representation. Our encoding strategy is simple -- pack
379 : * the first `sizeof(Datum)` bytes of uuid data into a Datum (on little-endian
380 : * machines, the bytes are stored in reverse order), and treat it as an
381 : * unsigned integer.
382 : */
383 : static Datum
384 3384150 : uuid_abbrev_convert(Datum original, SortSupport ssup)
385 : {
386 3384150 : uuid_sortsupport_state *uss = ssup->ssup_extra;
387 3384150 : pg_uuid_t *authoritative = DatumGetUUIDP(original);
388 : Datum res;
389 :
390 3384150 : memcpy(&res, authoritative->data, sizeof(Datum));
391 3384150 : uss->input_count += 1;
392 :
393 3384150 : if (uss->estimating)
394 : {
395 : uint32 tmp;
396 :
397 : #if SIZEOF_DATUM == 8
398 3384150 : tmp = (uint32) res ^ (uint32) ((uint64) res >> 32);
399 : #else /* SIZEOF_DATUM != 8 */
400 : tmp = (uint32) res;
401 : #endif
402 :
403 3384150 : addHyperLogLog(&uss->abbr_card, DatumGetUInt32(hash_uint32(tmp)));
404 : }
405 :
406 : /*
407 : * Byteswap on little-endian machines.
408 : *
409 : * This is needed so that ssup_datum_unsigned_cmp() (an unsigned integer
410 : * 3-way comparator) works correctly on all platforms. If we didn't do
411 : * this, the comparator would have to call memcmp() with a pair of
412 : * pointers to the first byte of each abbreviated key, which is slower.
413 : */
414 3384150 : res = DatumBigEndianToNative(res);
415 :
416 3384150 : return res;
417 : }
418 :
419 : /* hash index support */
420 : Datum
421 2426 : uuid_hash(PG_FUNCTION_ARGS)
422 : {
423 2426 : pg_uuid_t *key = PG_GETARG_UUID_P(0);
424 :
425 2426 : return hash_any(key->data, UUID_LEN);
426 : }
427 :
428 : Datum
429 60 : uuid_hash_extended(PG_FUNCTION_ARGS)
430 : {
431 60 : pg_uuid_t *key = PG_GETARG_UUID_P(0);
432 :
433 60 : return hash_any_extended(key->data, UUID_LEN, PG_GETARG_INT64(1));
434 : }
435 :
436 : /*
437 : * Set the given UUID version and the variant bits
438 : */
439 : static inline void
440 126 : uuid_set_version(pg_uuid_t *uuid, unsigned char version)
441 : {
442 : /* set version field, top four bits */
443 126 : uuid->data[6] = (uuid->data[6] & 0x0f) | (version << 4);
444 :
445 : /* set variant field, top two bits are 1, 0 */
446 126 : uuid->data[8] = (uuid->data[8] & 0x3f) | 0x80;
447 126 : }
448 :
449 : /*
450 : * Generate UUID version 4.
451 : *
452 : * All UUID bytes are filled with strong random numbers except version and
453 : * variant bits.
454 : */
455 : Datum
456 42 : gen_random_uuid(PG_FUNCTION_ARGS)
457 : {
458 42 : pg_uuid_t *uuid = palloc(UUID_LEN);
459 :
460 42 : if (!pg_strong_random(uuid, UUID_LEN))
461 0 : ereport(ERROR,
462 : (errcode(ERRCODE_INTERNAL_ERROR),
463 : errmsg("could not generate random values")));
464 :
465 : /*
466 : * Set magic numbers for a "version 4" (pseudorandom) UUID and variant,
467 : * see https://datatracker.ietf.org/doc/html/rfc9562#name-uuid-version-4
468 : */
469 42 : uuid_set_version(uuid, 4);
470 :
471 42 : PG_RETURN_UUID_P(uuid);
472 : }
473 :
474 : /*
475 : * Get the current timestamp with nanosecond precision for UUID generation.
476 : * The returned timestamp is ensured to be at least SUBMS_MINIMAL_STEP greater
477 : * than the previous returned timestamp (on this backend).
478 : */
479 : static inline int64
480 84 : get_real_time_ns_ascending()
481 : {
482 : static int64 previous_ns = 0;
483 : int64 ns;
484 :
485 : /* Get the current real timestamp */
486 :
487 : #ifdef _MSC_VER
488 : struct timeval tmp;
489 :
490 : gettimeofday(&tmp, NULL);
491 : ns = tmp.tv_sec * NS_PER_S + tmp.tv_usec * NS_PER_US;
492 : #else
493 : struct timespec tmp;
494 :
495 : /*
496 : * We don't use gettimeofday(), instead use clock_gettime() with
497 : * CLOCK_REALTIME where available in order to get a high-precision
498 : * (nanoseconds) real timestamp.
499 : *
500 : * Note while a timestamp returned by clock_gettime() with CLOCK_REALTIME
501 : * is nanosecond-precision on most Unix-like platforms, on some platforms
502 : * such as macOS it's restricted to microsecond-precision.
503 : */
504 84 : clock_gettime(CLOCK_REALTIME, &tmp);
505 84 : ns = tmp.tv_sec * NS_PER_S + tmp.tv_nsec;
506 : #endif
507 :
508 : /* Guarantee the minimal step advancement of the timestamp */
509 84 : if (previous_ns + SUBMS_MINIMAL_STEP_NS >= ns)
510 0 : ns = previous_ns + SUBMS_MINIMAL_STEP_NS;
511 84 : previous_ns = ns;
512 :
513 84 : return ns;
514 : }
515 :
516 : /*
517 : * Generate UUID version 7 per RFC 9562, with the given timestamp.
518 : *
519 : * UUID version 7 consists of a Unix timestamp in milliseconds (48 bits) and
520 : * 74 random bits, excluding the required version and variant bits. To ensure
521 : * monotonicity in scenarios of high-frequency UUID generation, we employ the
522 : * method "Replace Leftmost Random Bits with Increased Clock Precision (Method 3)",
523 : * described in the RFC. This method utilizes 12 bits from the "rand_a" bits
524 : * to store a 1/4096 (or 2^12) fraction of sub-millisecond precision.
525 : *
526 : * ns is a number of nanoseconds since start of the UNIX epoch. This value is
527 : * used for time-dependent bits of UUID.
528 : */
529 : static pg_uuid_t *
530 84 : generate_uuidv7(int64 ns)
531 : {
532 84 : pg_uuid_t *uuid = palloc(UUID_LEN);
533 : int64 unix_ts_ms;
534 : int32 increased_clock_precision;
535 :
536 84 : unix_ts_ms = ns / NS_PER_MS;
537 :
538 : /* Fill in time part */
539 84 : uuid->data[0] = (unsigned char) (unix_ts_ms >> 40);
540 84 : uuid->data[1] = (unsigned char) (unix_ts_ms >> 32);
541 84 : uuid->data[2] = (unsigned char) (unix_ts_ms >> 24);
542 84 : uuid->data[3] = (unsigned char) (unix_ts_ms >> 16);
543 84 : uuid->data[4] = (unsigned char) (unix_ts_ms >> 8);
544 84 : uuid->data[5] = (unsigned char) unix_ts_ms;
545 :
546 : /*
547 : * sub-millisecond timestamp fraction (SUBMS_BITS bits, not
548 : * SUBMS_MINIMAL_STEP_BITS)
549 : */
550 84 : increased_clock_precision = ((ns % NS_PER_MS) * (1 << SUBMS_BITS)) / NS_PER_MS;
551 :
552 : /* Fill the increased clock precision to "rand_a" bits */
553 84 : uuid->data[6] = (unsigned char) (increased_clock_precision >> 8);
554 84 : uuid->data[7] = (unsigned char) (increased_clock_precision);
555 :
556 : /* fill everything after the increased clock precision with random bytes */
557 84 : if (!pg_strong_random(&uuid->data[8], UUID_LEN - 8))
558 0 : ereport(ERROR,
559 : (errcode(ERRCODE_INTERNAL_ERROR),
560 : errmsg("could not generate random values")));
561 :
562 : #if SUBMS_MINIMAL_STEP_BITS == 10
563 :
564 : /*
565 : * On systems that have only 10 bits of sub-ms precision, 2 least
566 : * significant are dependent on other time-specific bits, and they do not
567 : * contribute to uniqueness. To make these bit random we mix in two bits
568 : * from CSPRNG. SUBMS_MINIMAL_STEP is chosen so that we still guarantee
569 : * monotonicity despite altering these bits.
570 : */
571 : uuid->data[7] = uuid->data[7] ^ (uuid->data[8] >> 6);
572 : #endif
573 :
574 : /*
575 : * Set magic numbers for a "version 7" (pseudorandom) UUID and variant,
576 : * see https://www.rfc-editor.org/rfc/rfc9562#name-version-field
577 : */
578 84 : uuid_set_version(uuid, 7);
579 :
580 84 : return uuid;
581 : }
582 :
583 : /*
584 : * Generate UUID version 7 with the current timestamp.
585 : */
586 : Datum
587 78 : uuidv7(PG_FUNCTION_ARGS)
588 : {
589 78 : pg_uuid_t *uuid = generate_uuidv7(get_real_time_ns_ascending());
590 :
591 78 : PG_RETURN_UUID_P(uuid);
592 : }
593 :
594 : /*
595 : * Similar to uuidv7() but with the timestamp adjusted by the given interval.
596 : */
597 : Datum
598 6 : uuidv7_interval(PG_FUNCTION_ARGS)
599 : {
600 6 : Interval *shift = PG_GETARG_INTERVAL_P(0);
601 : TimestampTz ts;
602 : pg_uuid_t *uuid;
603 6 : int64 ns = get_real_time_ns_ascending();
604 :
605 : /*
606 : * Shift the current timestamp by the given interval. To calculate time
607 : * shift correctly, we convert the UNIX epoch to TimestampTz and use
608 : * timestamptz_pl_interval(). Since this calculation is done with
609 : * microsecond precision, we carry nanoseconds from original ns value to
610 : * shifted ns value.
611 : */
612 :
613 6 : ts = (TimestampTz) (ns / NS_PER_US) -
614 : (POSTGRES_EPOCH_JDATE - UNIX_EPOCH_JDATE) * SECS_PER_DAY * USECS_PER_SEC;
615 :
616 : /* Compute time shift */
617 6 : ts = DatumGetTimestampTz(DirectFunctionCall2(timestamptz_pl_interval,
618 : TimestampTzGetDatum(ts),
619 : IntervalPGetDatum(shift)));
620 :
621 : /*
622 : * Convert a TimestampTz value back to an UNIX epoch and back nanoseconds.
623 : */
624 6 : ns = (ts + (POSTGRES_EPOCH_JDATE - UNIX_EPOCH_JDATE) * SECS_PER_DAY * USECS_PER_SEC)
625 6 : * NS_PER_US + ns % NS_PER_US;
626 :
627 : /* Generate an UUIDv7 */
628 6 : uuid = generate_uuidv7(ns);
629 :
630 6 : PG_RETURN_UUID_P(uuid);
631 : }
632 :
633 : /*
634 : * Start of a Gregorian epoch == date2j(1582,10,15)
635 : * We cast it to 64-bit because it's used in overflow-prone computations
636 : */
637 : #define GREGORIAN_EPOCH_JDATE INT64CONST(2299161)
638 :
639 : /*
640 : * Extract timestamp from UUID.
641 : *
642 : * Returns null if not RFC 9562 variant or not a version that has a timestamp.
643 : */
644 : Datum
645 24 : uuid_extract_timestamp(PG_FUNCTION_ARGS)
646 : {
647 24 : pg_uuid_t *uuid = PG_GETARG_UUID_P(0);
648 : int version;
649 : uint64 tms;
650 : TimestampTz ts;
651 :
652 : /* check if RFC 9562 variant */
653 24 : if ((uuid->data[8] & 0xc0) != 0x80)
654 6 : PG_RETURN_NULL();
655 :
656 18 : version = uuid->data[6] >> 4;
657 :
658 18 : if (version == 1)
659 : {
660 6 : tms = ((uint64) uuid->data[0] << 24)
661 6 : + ((uint64) uuid->data[1] << 16)
662 6 : + ((uint64) uuid->data[2] << 8)
663 6 : + ((uint64) uuid->data[3])
664 6 : + ((uint64) uuid->data[4] << 40)
665 6 : + ((uint64) uuid->data[5] << 32)
666 6 : + (((uint64) uuid->data[6] & 0xf) << 56)
667 6 : + ((uint64) uuid->data[7] << 48);
668 :
669 : /* convert 100-ns intervals to us, then adjust */
670 6 : ts = (TimestampTz) (tms / 10) -
671 : ((uint64) POSTGRES_EPOCH_JDATE - GREGORIAN_EPOCH_JDATE) * SECS_PER_DAY * USECS_PER_SEC;
672 6 : PG_RETURN_TIMESTAMPTZ(ts);
673 : }
674 :
675 12 : if (version == 7)
676 : {
677 6 : tms = (uuid->data[5])
678 6 : + (((uint64) uuid->data[4]) << 8)
679 6 : + (((uint64) uuid->data[3]) << 16)
680 6 : + (((uint64) uuid->data[2]) << 24)
681 6 : + (((uint64) uuid->data[1]) << 32)
682 6 : + (((uint64) uuid->data[0]) << 40);
683 :
684 : /* convert ms to us, then adjust */
685 6 : ts = (TimestampTz) (tms * NS_PER_US) -
686 : (POSTGRES_EPOCH_JDATE - UNIX_EPOCH_JDATE) * SECS_PER_DAY * USECS_PER_SEC;
687 :
688 6 : PG_RETURN_TIMESTAMPTZ(ts);
689 : }
690 :
691 : /* not a timestamp-containing UUID version */
692 6 : PG_RETURN_NULL();
693 : }
694 :
695 : /*
696 : * Extract UUID version.
697 : *
698 : * Returns null if not RFC 9562 variant.
699 : */
700 : Datum
701 30 : uuid_extract_version(PG_FUNCTION_ARGS)
702 : {
703 30 : pg_uuid_t *uuid = PG_GETARG_UUID_P(0);
704 : uint16 version;
705 :
706 : /* check if RFC 9562 variant */
707 30 : if ((uuid->data[8] & 0xc0) != 0x80)
708 6 : PG_RETURN_NULL();
709 :
710 24 : version = uuid->data[6] >> 4;
711 :
712 24 : PG_RETURN_UINT16(version);
713 : }
|
