codekingpro/portable-devtools

1/*-------------------------------------------------------------------------2 *3 * timestamp.h4 *	  Timestamp and Interval typedefs and related macros.5 *6 * Note: this file must be includable in both frontend and backend contexts.7 *8 * Portions Copyright (c) 1996-2023, PostgreSQL Global Development Group9 * Portions Copyright (c) 1994, Regents of the University of California10 *11 * src/include/datatype/timestamp.h12 *13 *-------------------------------------------------------------------------14 */15#ifndef DATATYPE_TIMESTAMP_H16#define DATATYPE_TIMESTAMP_H17 18/*19 * Timestamp represents absolute time.20 *21 * Interval represents delta time. Keep track of months (and years), days,22 * and hours/minutes/seconds separately since the elapsed time spanned is23 * unknown until instantiated relative to an absolute time.24 *25 * Note that Postgres uses "time interval" to mean a bounded interval,26 * consisting of a beginning and ending time, not a time span - thomas 97/03/2027 *28 * Timestamps, as well as the h/m/s fields of intervals, are stored as29 * int64 values with units of microseconds.  (Once upon a time they were30 * double values with units of seconds.)31 *32 * TimeOffset and fsec_t are convenience typedefs for temporary variables.33 * Do not use fsec_t in values stored on-disk.34 * Also, fsec_t is only meant for *fractional* seconds; beware of overflow35 * if the value you need to store could be many seconds.36 */37 38typedef int64 Timestamp;39typedef int64 TimestampTz;40typedef int64 TimeOffset;41typedef int32 fsec_t;			/* fractional seconds (in microseconds) */42 43 44/*45 * Storage format for type interval.46 */47typedef struct48{49	TimeOffset	time;			/* all time units other than days, months and50								 * years */51	int32		day;			/* days, after time for alignment */52	int32		month;			/* months and years, after time for alignment */53} Interval;54 55/*56 * Data structure representing a broken-down interval.57 *58 * For historical reasons, this is modeled on struct pg_tm for timestamps.59 * Unlike the situation for timestamps, there's no magic interpretation60 * needed for months or years: they're just zero or not.  Note that fields61 * can be negative; however, because of the divisions done while converting62 * from struct Interval, only tm_mday could be INT_MIN.  This is important63 * because we may need to negate the values in some code paths.64 */65struct pg_itm66{67	int			tm_usec;68	int			tm_sec;69	int			tm_min;70	int64		tm_hour;		/* needs to be wide */71	int			tm_mday;72	int			tm_mon;73	int			tm_year;74};75 76/*77 * Data structure for decoding intervals.  We could just use struct pg_itm,78 * but then the requirement for tm_usec to be 64 bits would propagate to79 * places where it's not really needed.  Also, omitting the fields that80 * aren't used during decoding seems like a good error-prevention measure.81 */82struct pg_itm_in83{84	int64		tm_usec;		/* needs to be wide */85	int			tm_mday;86	int			tm_mon;87	int			tm_year;88};89 90 91/* Limits on the "precision" option (typmod) for these data types */92#define MAX_TIMESTAMP_PRECISION 693#define MAX_INTERVAL_PRECISION 694 95/*96 *	Round off to MAX_TIMESTAMP_PRECISION decimal places.97 *	Note: this is also used for rounding off intervals.98 */99#define TS_PREC_INV 1000000.0100#define TSROUND(j) (rint(((double) (j)) * TS_PREC_INV) / TS_PREC_INV)101 102 103/*104 * Assorted constants for datetime-related calculations105 */106 107#define DAYS_PER_YEAR	365.25	/* assumes leap year every four years */108#define MONTHS_PER_YEAR 12109/*110 *	DAYS_PER_MONTH is very imprecise.  The more accurate value is111 *	365.2425/12 = 30.436875, or '30 days 10:29:06'.  Right now we only112 *	return an integral number of days, but someday perhaps we should113 *	also return a 'time' value to be used as well.  ISO 8601 suggests114 *	30 days.115 */116#define DAYS_PER_MONTH	30		/* assumes exactly 30 days per month */117#define HOURS_PER_DAY	24		/* assume no daylight savings time changes */118 119/*120 *	This doesn't adjust for uneven daylight savings time intervals or leap121 *	seconds, and it crudely estimates leap years.  A more accurate value122 *	for days per years is 365.2422.123 */124#define SECS_PER_YEAR	(36525 * 864)	/* avoid floating-point computation */125#define SECS_PER_DAY	86400126#define SECS_PER_HOUR	3600127#define SECS_PER_MINUTE 60128#define MINS_PER_HOUR	60129 130#define USECS_PER_DAY	INT64CONST(86400000000)131#define USECS_PER_HOUR	INT64CONST(3600000000)132#define USECS_PER_MINUTE INT64CONST(60000000)133#define USECS_PER_SEC	INT64CONST(1000000)134 135/*136 * We allow numeric timezone offsets up to 15:59:59 either way from Greenwich.137 * Currently, the record holders for wackiest offsets in actual use are zones138 * Asia/Manila, at -15:56:00 until 1844, and America/Metlakatla, at +15:13:42139 * until 1867.  If we were to reject such values we would fail to dump and140 * restore old timestamptz values with these zone settings.141 */142#define MAX_TZDISP_HOUR		15	/* maximum allowed hour part */143#define TZDISP_LIMIT		((MAX_TZDISP_HOUR + 1) * SECS_PER_HOUR)144 145/*146 * We reserve the minimum and maximum integer values to represent147 * timestamp (or timestamptz) -infinity and +infinity.148 */149#define TIMESTAMP_MINUS_INFINITY	PG_INT64_MIN150#define TIMESTAMP_INFINITY	PG_INT64_MAX151 152/*153 * Historically these alias for infinity have been used.154 */155#define DT_NOBEGIN		TIMESTAMP_MINUS_INFINITY156#define DT_NOEND		TIMESTAMP_INFINITY157 158#define TIMESTAMP_NOBEGIN(j)	\159	do {(j) = DT_NOBEGIN;} while (0)160 161#define TIMESTAMP_IS_NOBEGIN(j) ((j) == DT_NOBEGIN)162 163#define TIMESTAMP_NOEND(j)		\164	do {(j) = DT_NOEND;} while (0)165 166#define TIMESTAMP_IS_NOEND(j)	((j) == DT_NOEND)167 168#define TIMESTAMP_NOT_FINITE(j) (TIMESTAMP_IS_NOBEGIN(j) || TIMESTAMP_IS_NOEND(j))169 170 171/*172 * Julian date support.173 *174 * date2j() and j2date() nominally handle the Julian date range 0..INT_MAX,175 * or 4714-11-24 BC to 5874898-06-03 AD.  In practice, date2j() will work and176 * give correct negative Julian dates for dates before 4714-11-24 BC as well.177 * We rely on it to do so back to 4714-11-01 BC.  Allowing at least one day's178 * slop is necessary so that timestamp rotation doesn't produce dates that179 * would be rejected on input.  For example, '4714-11-24 00:00 GMT BC' is a180 * legal timestamptz value, but in zones east of Greenwich it would print as181 * sometime in the afternoon of 4714-11-23 BC; if we couldn't process such a182 * date we'd have a dump/reload failure.  So the idea is for IS_VALID_JULIAN183 * to accept a slightly wider range of dates than we really support, and184 * then we apply the exact checks in IS_VALID_DATE or IS_VALID_TIMESTAMP,185 * after timezone rotation if any.  To save a few cycles, we can make186 * IS_VALID_JULIAN check only to the month boundary, since its exact cutoffs187 * are not very critical in this scheme.188 *189 * It is correct that JULIAN_MINYEAR is -4713, not -4714; it is defined to190 * allow easy comparison to tm_year values, in which we follow the convention191 * that tm_year <= 0 represents abs(tm_year)+1 BC.192 */193 194#define JULIAN_MINYEAR (-4713)195#define JULIAN_MINMONTH (11)196#define JULIAN_MINDAY (24)197#define JULIAN_MAXYEAR (5874898)198#define JULIAN_MAXMONTH (6)199#define JULIAN_MAXDAY (3)200 201#define IS_VALID_JULIAN(y,m,d) \202	(((y) > JULIAN_MINYEAR || \203	  ((y) == JULIAN_MINYEAR && ((m) >= JULIAN_MINMONTH))) && \204	 ((y) < JULIAN_MAXYEAR || \205	  ((y) == JULIAN_MAXYEAR && ((m) < JULIAN_MAXMONTH))))206 207/* Julian-date equivalents of Day 0 in Unix and Postgres reckoning */208#define UNIX_EPOCH_JDATE		2440588 /* == date2j(1970, 1, 1) */209#define POSTGRES_EPOCH_JDATE	2451545 /* == date2j(2000, 1, 1) */210 211/*212 * Range limits for dates and timestamps.213 *214 * We have traditionally allowed Julian day zero as a valid datetime value,215 * so that is the lower bound for both dates and timestamps.216 *217 * The upper limit for dates is 5874897-12-31, which is a bit less than what218 * the Julian-date code can allow.  For timestamps, the upper limit is219 * 294276-12-31.  The int64 overflow limit would be a few days later; again,220 * leaving some slop avoids worries about corner-case overflow, and provides221 * a simpler user-visible definition.222 */223 224/* First allowed date, and first disallowed date, in Julian-date form */225#define DATETIME_MIN_JULIAN (0)226#define DATE_END_JULIAN (2147483494)	/* == date2j(JULIAN_MAXYEAR, 1, 1) */227#define TIMESTAMP_END_JULIAN (109203528)	/* == date2j(294277, 1, 1) */228 229/* Timestamp limits */230#define MIN_TIMESTAMP	INT64CONST(-211813488000000000)231/* == (DATETIME_MIN_JULIAN - POSTGRES_EPOCH_JDATE) * USECS_PER_DAY */232#define END_TIMESTAMP	INT64CONST(9223371331200000000)233/* == (TIMESTAMP_END_JULIAN - POSTGRES_EPOCH_JDATE) * USECS_PER_DAY */234 235/* Range-check a date (given in Postgres, not Julian, numbering) */236#define IS_VALID_DATE(d) \237	((DATETIME_MIN_JULIAN - POSTGRES_EPOCH_JDATE) <= (d) && \238	 (d) < (DATE_END_JULIAN - POSTGRES_EPOCH_JDATE))239 240/* Range-check a timestamp */241#define IS_VALID_TIMESTAMP(t)  (MIN_TIMESTAMP <= (t) && (t) < END_TIMESTAMP)242 243#endif							/* DATATYPE_TIMESTAMP_H */244