See below for the formatting function that converts from a formatted string to TIMESTAMPTZ.

Alias: FROM_UNIXTIME - only accepts one parameter of type DOUBLE PRECISION

​
Conversion function

Converts the number of seconds since the Unix epoch (1970-01-01 00:00:00 UTC) to a TIMESTAMPTZ value.

​
Syntax

TO_TIMESTAMP(<seconds>)

​
Parameters

ParameterDescription
<seconds>A value expression of type DOUBLE PRECISION representing the number of seconds before or after the Unix epoch. If present, the fractional part is interpreted as subseconds.

​
Return Type

TIMESTAMPTZ

​
Remarks

TO_TIMESTAMP(<seconds>) is the inverse function of EXTRACT(EPOCH FROM TIMESTAMPTZ).

​
Example

SET time_zone = 'Europe/Berlin';
SELECT TO_TIMESTAMP(42.123456::DOUBLE PRECISION);  --> 1970-01-01 01:00:42.123456+01

​
Formatting function

Converts a string to TIMESTAMPTZ type (i.e., timestamp with time zone) using format.

​
Syntax

TO_TIMESTAMP(<expression> [,'<format>'])

​
Parameters

ParameterDescriptionSupported input types
<expression>The text to convert to a timestamp with time zone. If no optional <format> argument is given that can be used to parse the <expression>, the following format is required: any supported date format directly followed by ( |T)[H]H:[m]m:[S]S[.F]. (For supported date formats see TO_DATETEXT
<format>Optional. A string literal that specifies the format of the <expression> to convert.TEXT (see below)

Accepted <format> patterns include the following specifications:

Format optionDescriptionExample
YYYYYear (4 or more digits)TO_TIMESTAMP('2023', 'YYYY'); --> '2023-01-01 00:00:00+00'
YYYLast 3 digits of yearTO_TIMESTAMP('2023', 'YYY'); --> '2023-01-01 00:00:00+00'
YYLast 2 digits of yearTO_TIMESTAMP('2023', 'YY'); --> '2023-01-01 00:00:00+00'
YLast digit of yearTO_TIMESTAMP('2023', 'Y'); --> '2023-01-01 00:00:00+00'
MONTHFull month name (case insensitive)TO_TIMESTAMP('august', 'MONTH'); --> '0001-08-01 00:00:00+00'
MONabbreviated month name (3 chars, case insensitive)TO_TIMESTAMP('dec', 'MON'); --> '0001-12-01 00:00:00+00'
MMMonth number (01–12)TO_TIMESTAMP('7', 'MM'); --> '0001-07-01 00:00:00+00'
DDDay of month (01–31)TO_TIMESTAMP('15', 'DD'); --> '0001-01-15 00:00:00+00'
HH or HH12Hour of day (01–12)TO_TIMESTAMP('8', 'hh'); --> '0001-01-01 08:00:00+00'
HH24Hour of day (00–23)TO_TIMESTAMP('18', 'hh24'); --> '0001-01-01 18:00:00+00'
MIMinute (00–59)TO_TIMESTAMP('35', 'hh'); --> '0001-01-01 00:35:00+00'
SSSecond (00–59)TO_TIMESTAMP('52', 'hh'); --> '0001-01-01 00:00:52+00'
MSMillisecond (000–999)TO_TIMESTAMP('89', 'ms'); --> '0001-01-01 08:00:00.89+00'
USMicrosecond (000000–999999)TO_TIMESTAMP('04852', 'hh'); --> '0001-01-01 08:00:00.04852+00'
AM or PM meridiem indicator (without periods)TO_TIMESTAMP('5 am', 'hh PM'); --> '0001-01-01 05:00:00+00+00'
A.M. or P.M.meridiem indicator (with periods)TO_TIMESTAMP('5 p.m.', 'hh A.M.'); --> '0001-01-01 17:00:00+00'
TZHTime zone hoursTO_TIMESTAMP('12 -2', 'hh24 TZH'); --> '0001-01-01 14:00:00+00+00'
TZMTime zone minutesTO_TIMESTAMP('12 45', 'hh24 TZM'); --> '0001-01-01 11:15:00+00

Usage notes for formatting

  • Case letters in the input <expression> are ignored
  • A separator (non-digit and non-letter) in the <format> string will match exactly one separator or is skipped
  • Any non-separator in the <format> that is not part of a format option will match exactly one other character.
  • Any character in quotes " will match exactly one other character.
  • If the year format specification is 'YYY', 'YY', or 'Y' and the supplied year is less than four digits, the year will be adjusted to be nearest to the year 2020, (e.g., 80 becomes 1980).
  • Milliseconds and microseconds are interpreted as subseconds in the form ss.xxx[xxx]. This means that TO_TIMESTAMP('30.7', 'SS.MS') is not 7 milliseconds, but 700, because it will be treated as 30 + 0.7 seconds. To achieve 7 milliseconds, one ust write 30.007 instead.
  • Modifiers (e.g., 'FM') are not supported.

Some additional format patterns are reserved but currently not implemented: FF1, FF2, FF3, FF4, FF5, FF6, SSSS, SSSSS, IYYY, IYY, IY, I, BC, AD, B_DOT_C_DOT, A_DOT_D_DOT, DAY, DY, DDD, IDDD, D, ID, W, WW, IW, CC, J, Q. Using them in the format string raises an error.

​
Examples

The example below shows our separators and non-separators can cause skips. The separator ' ' (space) in the <format> matches the other separator '/' in the <expression>. The non-separator 'x' will match any other character, in this case the 'a'. Lastly, the two separators '++' will match up to two other separators, here the first 'x' matches '.' while the second 'x' will simply be ignored as no other separators follow.

The example below shows how the year is adjusted to be nearest to 2020 because YYY was used to match a less than four digit number. To receive the exact year '180' use YYYY instead. Furthermore, as the three separators are quotes "..." they will match any character (separator or non-separator) which in this case is 'ar '.

Due to rounding it can happen that the conversion from an extracted epoch back to a timestamp does not generate the original timestamp.