This module provides various time-related functions. For relatedfunctionality, see also the datetime
and calendar
modules.
Although this module is always available,not all functions are available on all platforms. Most of the functionsdefined in this module call platform C library functions with the same name. Itmay sometimes be helpful to consult the platform documentation, because thesemantics of these functions varies among platforms.
An explanation of some terminology and conventions is in order.
The epoch is the point where the time starts, the return value of
time.gmtime(0)
. It is January 1, 1970, 00:00:00 (UTC) on all platforms.
The term seconds since the epoch refers to the total numberof elapsed seconds since the epoch, typically excludingleap seconds. Leap seconds are excluded from this total on allPOSIX-compliant platforms.
The functions in this module may not handle dates and times before the epoch orfar in the future. The cut-off point in the future is determined by the Clibrary; for 32-bit systems, it is typically in 2038.
Function strptime() can parse 2-digit years when given
%y
formatcode. When 2-digit years are parsed, they are converted according to the POSIXand ISO C standards: values 69–99 are mapped to 1969–1999, and values 0–68are mapped to 2000–2068.
UTC is Coordinated Universal Time (formerly known as Greenwich Mean Time, orGMT). The acronym UTC is not a mistake but a compromise between English andFrench.
DST is Daylight Saving Time, an adjustment of the timezone by (usually) onehour during part of the year. DST rules are magic (determined by local law) andcan change from year to year. The C library has a table containing the localrules (often it is read from a system file for flexibility) and is the onlysource of True Wisdom in this respect.
The precision of the various real-time functions may be less than suggested bythe units in which their value or argument is expressed. E.g. on most Unixsystems, the clock “ticks” only 50 or 100 times a second.
On the other hand, the precision of time() and sleep() is betterthan their Unix equivalents: times are expressed as floating-point numbers,time() returns the most accurate time available (using Unix
gettimeofday()
where available), and sleep() will accept a timewith a nonzero fraction (Unixselect()
is used to implement this, whereavailable).The time value as returned by gmtime(), localtime(), andstrptime(), and accepted by asctime(), mktime() andstrftime(), is a sequence of 9 integers. The return values ofgmtime(), localtime(), and strptime() also offer attributenames for individual fields.
See struct_time for a description of these objects.
Changed in version 3.3: The struct_time type was extended to providethe tm_gmtoff and tm_zoneattributes when platform supports corresponding
struct tm
members.Changed in version 3.6: The struct_time attributestm_gmtoff and tm_zoneare now available on all platforms.
Use the following functions to convert between time representations:
From
To
Use
seconds since the epoch
struct_time inUTC
gmtime()
seconds since the epoch
struct_time inlocal time
localtime()
struct_time inUTC
seconds since the epoch
struct_time inlocal time
seconds since the epoch
mktime()
Functions¶
- time.asctime([t])¶
Convert a tuple or struct_time representing a time as returned bygmtime() or localtime() to a string of the followingform:
'Sun Jun 20 23:21:05 1993'
. The day field is two characters longand is space padded if the day is a single digit,e.g.:'Wed Jun 9 04:26:40 1993'
.If t is not provided, the current time as returned by localtime()is used. Locale information is not used by asctime().
Note
Unlike the C function of the same name, asctime() does not add atrailing newline.
- time.pthread_getcpuclockid(thread_id)¶
Return the clk_id of the thread-specific CPU-time clock for the specified thread_id.
Use
threading.get_ident()
or theident
attribute ofthreading.Thread
objects to get a suitable valuefor thread_id.Warning
Passing an invalid or expired thread_id may result inundefined behavior, such as segmentation fault.
Availability: Unix
See the man page for pthread_getcpuclockid(3) forfurther information.
Added in version 3.7.
- time.clock_getres(clk_id)¶
Return the resolution (precision) of the specified clock clk_id. Refer toClock ID Constants for a list of accepted values for clk_id.
Availability: Unix.
Added in version 3.3.
- time.clock_gettime(clk_id) → float¶
Return the time of the specified clock clk_id. Refer toClock ID Constants for a list of accepted values for clk_id.
Use clock_gettime_ns() to avoid the precision loss caused by the
float
type.Availability: Unix.
Added in version 3.3.
- time.clock_gettime_ns(clk_id) → int¶
Similar to clock_gettime() but return time as nanoseconds.
Availability: Unix.
Added in version 3.7.
- time.clock_settime(clk_id, time: float)¶
Set the time of the specified clock clk_id. Currently,CLOCK_REALTIME is the only accepted value for clk_id.
Use clock_settime_ns() to avoid the precision loss caused by the
float
type.Availability: Unix.
Added in version 3.3.
- time.clock_settime_ns(clk_id, time: int)¶
Similar to clock_settime() but set time with nanoseconds.
Availability: Unix.
Added in version 3.7.
- time.ctime([secs])¶
Convert a time expressed in seconds since the epoch to a string of a form:
'Sun Jun 20 23:21:05 1993'
representing local time. The day fieldis two characters long and is space padded if the day is a single digit,e.g.:'Wed Jun 9 04:26:40 1993'
.See AlsoBatch learning vs Online learningWhat is Model Training | Oden TechnologiesRequest for Explanation on steps_per_epoch parameterExploring the Complex Relationship between Batch Size and Learning Rate in Machine LearningIf secs is not provided or
None
, the current time asreturned by time() is used.ctime(secs)
is equivalent toasctime(localtime(secs))
. Locale information is not used byctime().
- time.get_clock_info(name)¶
Get information on the specified clock as a namespace object.Supported clock names and the corresponding functions to read their valueare:
'monotonic'
: time.monotonic()'perf_counter'
: time.perf_counter()'process_time'
: time.process_time()'thread_time'
: time.thread_time()'time'
: time.time()
The result has the following attributes:
adjustable:
True
if the clock can be changed automatically (e.g. bya NTP daemon) or manually by the system administrator,False
otherwiseimplementation: The name of the underlying C function used to getthe clock value. Refer to Clock ID Constants for possible values.
monotonic:
True
if the clock cannot go backward,False
otherwiseresolution: The resolution of the clock in seconds (
float
)
Added in version 3.3.
- time.gmtime([secs])¶
Convert a time expressed in seconds since the epoch to a struct_time inUTC in which the dst flag is always zero. If secs is not provided or
None
, the current time as returned by time() is used. Fractionsof a second are ignored. See above for a description of thestruct_time object. Seecalendar.timegm()
for the inverse of thisfunction.
- time.localtime([secs])¶
Like gmtime() but converts to local time. If secs is not provided or
None
, the current time as returned by time() is used. The dstflag is set to1
when DST applies to the given time.localtime() may raise
OverflowError
, if the timestamp isoutside the range of values supported by the platform Clocaltime()
orgmtime()
functions, andOSError
onlocaltime()
orgmtime()
failure. It’s common for this to be restricted to yearsbetween 1970 and 2038.
- time.mktime(t)¶
This is the inverse function of localtime(). Its argument is thestruct_time or full 9-tuple (since the dst flag is needed; use
-1
as the dst flag if it is unknown) which expresses the time in local time, notUTC. It returns a floating-point number, for compatibility with time().If the input value cannot be represented as a valid time, eitherOverflowError
orValueError
will be raised (which depends onwhether the invalid value is caught by Python or the underlying C libraries).The earliest date for which it can generate a time is platform-dependent.
- time.monotonic() → float¶
Return the value (in fractional seconds) of a monotonic clock, i.e. a clockthat cannot go backwards. The clock is not affected by system clock updates.The reference point of the returned value is undefined, so that only thedifference between the results of two calls is valid.
Use monotonic_ns() to avoid the precision loss caused by the
float
type.Added in version 3.3.
Changed in version 3.5: The function is now always available and always system-wide.
Changed in version 3.10: On macOS, the function is now system-wide.
- time.monotonic_ns() → int¶
Similar to monotonic(), but return time as nanoseconds.
Added in version 3.7.
- time.perf_counter() → float¶
Return the value (in fractional seconds) of a performance counter, i.e. aclock with the highest available resolution to measure a short duration. Itdoes include time elapsed during sleep and is system-wide. The referencepoint of the returned value is undefined, so that only the difference betweenthe results of two calls is valid.
Use perf_counter_ns() to avoid the precision loss caused by the
float
type.Added in version 3.3.
Changed in version 3.10: On Windows, the function is now system-wide.
- time.perf_counter_ns() → int¶
Similar to perf_counter(), but return time as nanoseconds.
Added in version 3.7.
- time.process_time() → float¶
Return the value (in fractional seconds) of the sum of the system and userCPU time of the current process. It does not include time elapsed duringsleep. It is process-wide by definition. The reference point of thereturned value is undefined, so that only the difference between the resultsof two calls is valid.
Use process_time_ns() to avoid the precision loss caused by the
float
type.Added in version 3.3.
- time.process_time_ns() → int¶
Similar to process_time() but return time as nanoseconds.
Added in version 3.7.
- time.sleep(secs)¶
Suspend execution of the calling thread for the given number of seconds.The argument may be a floating-point number to indicate a more precise sleeptime.
If the sleep is interrupted by a signal and no exception is raised by thesignal handler, the sleep is restarted with a recomputed timeout.
The suspension time may be longer than requested by an arbitrary amount,because of the scheduling of other activity in the system.
On Windows, if secs is zero, the thread relinquishes the remainder of itstime slice to any other thread that is ready to run. If there are no otherthreads ready to run, the function returns immediately, and the threadcontinues execution. On Windows 8.1 and newer the implementation usesa high-resolution timerwhich provides resolution of 100 nanoseconds. If secs is zero,
Sleep(0)
is used.Unix implementation:
Use
clock_nanosleep()
if available (resolution: 1 nanosecond);Or use
nanosleep()
if available (resolution: 1 nanosecond);Or use
select()
(resolution: 1 microsecond).
Changed in version 3.5: The function now sleeps at least secs even if the sleep is interruptedby a signal, except if the signal handler raises an exception (seePEP 475 for the rationale).
Changed in version 3.11: On Unix, the
clock_nanosleep()
andnanosleep()
functions are nowused if available. On Windows, a waitable timer is now used.
- time.strftime(format[, t])¶
Convert a tuple or struct_time representing a time as returned bygmtime() or localtime() to a string as specified by the formatargument. If t is not provided, the current time as returned bylocaltime() is used. format must be a string.
ValueError
israised if any field in t is outside of the allowed range.0 is a legal argument for any position in the time tuple; if it is normallyillegal the value is forced to a correct one.
The following directives can be embedded in the format string. They are shownwithout the optional field width and precision specification, and are replacedby the indicated characters in the strftime() result:
Directive
Meaning
Notes
%a
Locale’s abbreviated weekday name.
%A
Locale’s full weekday name.
%b
Locale’s abbreviated month name.
%B
Locale’s full month name.
%c
Locale’s appropriate date and timerepresentation.
%d
Day of the month as a decimal number [01,31].
%f
- Microseconds as a decimal number
[000000,999999].
(1)
%H
Hour (24-hour clock) as a decimal number[00,23].
%I
Hour (12-hour clock) as a decimal number[01,12].
%j
Day of the year as a decimal number [001,366].
%m
Month as a decimal number [01,12].
%M
Minute as a decimal number [00,59].
%p
Locale’s equivalent of either AM or PM.
(2)
%S
Second as a decimal number [00,61].
(3)
%U
Week number of the year (Sunday as the firstday of the week) as a decimal number [00,53].All days in a new year preceding the firstSunday are considered to be in week 0.
(4)
%w
Weekday as a decimal number [0(Sunday),6].
%W
Week number of the year (Monday as the firstday of the week) as a decimal number [00,53].All days in a new year preceding the firstMonday are considered to be in week 0.
(4)
%x
Locale’s appropriate date representation.
%X
Locale’s appropriate time representation.
%y
Year without century as a decimal number[00,99].
%Y
Year with century as a decimal number.
%z
Time zone offset indicating a positive ornegative time difference from UTC/GMT of theform +HHMM or -HHMM, where H represents decimalhour digits and M represents decimal minutedigits [-23:59, +23:59]. [1]
%Z
Time zone name (no characters if no time zoneexists). Deprecated. [1]
%%
A literal
'%'
character.Notes:
The
%f
format directive only applies to strptime(),not to strftime(). However, see alsodatetime.datetime.strptime()
anddatetime.datetime.strftime()
where the%f
format directiveapplies to microseconds.When used with the strptime() function, the
%p
directive only affectsthe output hour field if the%I
directive is used to parse the hour.
The range really is
0
to61
; value60
is valid intimestamps representing leap seconds and value61
is supportedfor historical reasons.When used with the strptime() function,
%U
and%W
are only used incalculations when the day of the week and the year are specified.
Here is an example, a format for dates compatible with that specified in theRFC 2822 Internet email standard. [1]
>>> from time import gmtime, strftime>>> strftime("%a, %d %b %Y %H:%M:%S +0000", gmtime())'Thu, 28 Jun 2001 14:17:15 +0000'
Additional directives may be supported on certain platforms, but only theones listed here have a meaning standardized by ANSI C. To see the full setof format codes supported on your platform, consult the strftime(3)documentation.
On some platforms, an optional field width and precision specification canimmediately follow the initial
'%'
of a directive in the following order;this is also not portable. The field width is normally 2 except for%j
whereit is 3.
- time.strptime(string[, format])¶
Parse a string representing a time according to a format. The return valueis a struct_time as returned by gmtime() orlocaltime().
The format parameter uses the same directives as those used bystrftime(); it defaults to
"%a %b %d %H:%M:%S %Y"
which matches theformatting returned by ctime(). If string cannot be parsed accordingto format, or if it has excess data after parsing,ValueError
israised. The default values used to fill in any missing data when moreaccurate values cannot be inferred are(1900, 1, 1, 0, 0, 0, 0, 1, -1)
.Both string and format must be strings.For example:
>>> import time>>> time.strptime("30 Nov 00", "%d %b %y") time.struct_time(tm_year=2000, tm_mon=11, tm_mday=30, tm_hour=0, tm_min=0, tm_sec=0, tm_wday=3, tm_yday=335, tm_isdst=-1)
Support for the
%Z
directive is based on the values contained intzname
and whetherdaylight
is true. Because of this, it is platform-specificexcept for recognizing UTC and GMT which are always known (and are considered tobe non-daylight savings timezones).Only the directives specified in the documentation are supported. Because
strftime()
is implemented per platform it can sometimes offer moredirectives than those listed. Butstrptime()
is independent of any platformand thus does not necessarily support all directives available that are notdocumented as supported.
- class time.struct_time¶
The type of the time value sequence returned by gmtime(),localtime(), and strptime(). It is an object with a namedtuple interface: values can be accessed by index and by attribute name. Thefollowing values are present:
Index
Attribute
Values
- tm_year¶
(for example, 1993)
1
- tm_mon¶
range [1, 12]
2
- tm_mday¶
range [1, 31]
3
- tm_hour¶
range [0, 23]
4
- tm_min¶
range [0, 59]
5
- tm_sec¶
range [0, 61]; see Note (2) in strftime()
6
- tm_wday¶
range [0, 6]; Monday is 0
7
- tm_yday¶
range [1, 366]
8
- tm_isdst¶
0, 1 or -1; see below
N/A
- tm_zone¶
abbreviation of timezone name
N/A
- tm_gmtoff¶
offset east of UTC in seconds
Note that unlike the C structure, the month value is a range of [1, 12], not[0, 11].
In calls to mktime(), tm_isdst may be set to 1 when daylightsavings time is in effect, and 0 when it is not. A value of -1 indicates thatthis is not known, and will usually result in the correct state being filled in.
When a tuple with an incorrect length is passed to a function expecting astruct_time, or having elements of the wrong type, a
TypeError
is raised.
- time.time() → float¶
Return the time in seconds since the epoch as a floating-pointnumber. The handling of leap seconds is platform dependent.On Windows and most Unix systems, the leap seconds are not counted towardsthe time in seconds since the epoch. This is commonly referred to as Unixtime.
Note that even though the time is always returned as a floating-pointnumber, not all systems provide time with a better precision than 1 second.While this function normally returns non-decreasing values, it can return alower value than a previous call if the system clock has been set backbetween the two calls.
The number returned by time() may be converted into a more commontime format (i.e. year, month, day, hour, etc…) in UTC by passing it togmtime() function or in local time by passing it to thelocaltime() function. In both cases astruct_time object is returned, from which the componentsof the calendar date may be accessed as attributes.
Use time_ns() to avoid the precision loss caused by the
float
type.
- time.time_ns() → int¶
Similar to time() but returns time as an integer number ofnanoseconds since the epoch.
Added in version 3.7.
- time.thread_time() → float¶
Return the value (in fractional seconds) of the sum of the system and userCPU time of the current thread. It does not include time elapsed duringsleep. It is thread-specific by definition. The reference point of thereturned value is undefined, so that only the difference between the resultsof two calls in the same thread is valid.
Use thread_time_ns() to avoid the precision loss caused by the
float
type.Availability: Linux, Unix, Windows.
Unix systems supporting
CLOCK_THREAD_CPUTIME_ID
.Added in version 3.7.
- time.thread_time_ns() → int¶
Similar to thread_time() but return time as nanoseconds.
Added in version 3.7.
- time.tzset()¶
Reset the time conversion rules used by the library routines. The environmentvariable
TZ
specifies how this is done. It will also set the variablestzname
(from theTZ
environment variable),timezone
(non-DSTseconds West of UTC),altzone
(DST seconds west of UTC) anddaylight
(to 0 if this timezone does not have any daylight saving time rules, or tononzero if there is a time, past, present or future when daylight saving timeapplies).Availability: Unix.
Note
Although in many cases, changing the
TZ
environment variable mayaffect the output of functions like localtime() without callingtzset(), this behavior should not be relied on.The
TZ
environment variable should contain no whitespace.The standard format of the
TZ
environment variable is (whitespaceadded for clarity):std offset [dst [offset [,start[/time], end[/time]]]]
Where the components are:
std
anddst
Three or more alphanumerics giving the timezone abbreviations. These will bepropagated into time.tzname
offset
The offset has the form:
± hh[:mm[:ss]]
. This indicates the valueadded the local time to arrive at UTC. If preceded by a ‘-’, the timezoneis east of the Prime Meridian; otherwise, it is west. If no offset followsdst, summer time is assumed to be one hour ahead of standard time.start[/time], end[/time]
Indicates when to change to and back from DST. The format of thestart and end dates are one of the following:
Jn
The Julian day n (1 <= n <= 365). Leap days are not counted, so inall years February 28 is day 59 and March 1 is day 60.
n
The zero-based Julian day (0 <= n <= 365). Leap days are counted, andit is possible to refer to February 29.
Mm.n.d
The d’th day (0 <= d <= 6) of week n of month m of the year (1<= n <= 5, 1 <= m <= 12, where week 5 means “the last d day inmonth m” which may occur in either the fourth or the fifthweek). Week 1 is the first week in which the d’th day occurs. Dayzero is a Sunday.
time
has the same format asoffset
except that no leading sign(‘-’ or ‘+’) is allowed. The default, if time is not given, is 02:00:00.
>>> os.environ['TZ'] = 'EST+05EDT,M4.1.0,M10.5.0'>>> time.tzset()>>> time.strftime('%X %x %Z')'02:07:36 05/08/03 EDT'>>> os.environ['TZ'] = 'AEST-10AEDT-11,M10.5.0,M3.5.0'>>> time.tzset()>>> time.strftime('%X %x %Z')'16:08:12 05/08/03 AEST'
On many Unix systems (including *BSD, Linux, Solaris, and Darwin), it is moreconvenient to use the system’s zoneinfo (tzfile(5)) database tospecify the timezone rules. To do this, set the
TZ
environmentvariable to the path of the required timezone datafile, relative to the root ofthe systems ‘zoneinfo’ timezone database, usually located at/usr/share/zoneinfo
. For example,'US/Eastern'
,'Australia/Melbourne'
,'Egypt'
or'Europe/Amsterdam'
.>>> os.environ['TZ'] = 'US/Eastern'>>> time.tzset()>>> time.tzname('EST', 'EDT')>>> os.environ['TZ'] = 'Egypt'>>> time.tzset()>>> time.tzname('EET', 'EEST')
Clock ID Constants¶
These constants are used as parameters for clock_getres() andclock_gettime().
- time.CLOCK_BOOTTIME¶
Identical to CLOCK_MONOTONIC, except it also includes any time thatthe system is suspended.
This allows applications to get a suspend-aware monotonic clock withouthaving to deal with the complications of CLOCK_REALTIME, which mayhave discontinuities if the time is changed using
settimeofday()
orsimilar.Availability: Linux >= 2.6.39.
Added in version 3.7.
- time.CLOCK_HIGHRES¶
The Solaris OS has a
CLOCK_HIGHRES
timer that attempts to use an optimalhardware source, and may give close to nanosecond resolution.CLOCK_HIGHRES
is the nonadjustable, high-resolution clock.Availability: Solaris.
Added in version 3.3.
- time.CLOCK_MONOTONIC¶
Clock that cannot be set and represents monotonic time since some unspecifiedstarting point.
Availability: Unix.
Added in version 3.3.
- time.CLOCK_MONOTONIC_RAW¶
Similar to CLOCK_MONOTONIC, but provides access to a rawhardware-based time that is not subject to NTP adjustments.
Availability: Linux >= 2.6.28, macOS >= 10.12.
Added in version 3.3.
- time.CLOCK_PROCESS_CPUTIME_ID¶
High-resolution per-process timer from the CPU.
Availability: Unix.
Added in version 3.3.
- time.CLOCK_PROF¶
High-resolution per-process timer from the CPU.
Availability: FreeBSD, NetBSD >= 7, OpenBSD.
Added in version 3.7.
- time.CLOCK_TAI¶
The system must have a current leap second table in order for this to givethe correct answer. PTP or NTP software can maintain a leap second table.
Availability: Linux.
Added in version 3.9.
- time.CLOCK_THREAD_CPUTIME_ID¶
Thread-specific CPU-time clock.
Availability: Unix.
Added in version 3.3.
- time.CLOCK_UPTIME¶
Time whose absolute value is the time the system has been running and notsuspended, providing accurate uptime measurement, both absolute andinterval.
Availability: FreeBSD, OpenBSD >= 5.5.
Added in version 3.7.
- time.CLOCK_UPTIME_RAW¶
Clock that increments monotonically, tracking the time since an arbitrarypoint, unaffected by frequency or time adjustments and not incremented whilethe system is asleep.
Availability: macOS >= 10.12.
Added in version 3.8.
The following constant is the only parameter that can be sent toclock_settime().
- time.CLOCK_REALTIME¶
System-wide real-time clock. Setting this clock requires appropriateprivileges.
Availability: Unix.
Added in version 3.3.
Timezone Constants¶
- time.altzone¶
The offset of the local DST timezone, in seconds west of UTC, if one is defined.This is negative if the local DST timezone is east of UTC (as in Western Europe,including the UK). Only use this if
daylight
is nonzero. See note below.
- time.daylight¶
Nonzero if a DST timezone is defined. See note below.
- time.timezone¶
The offset of the local (non-DST) timezone, in seconds west of UTC (negative inmost of Western Europe, positive in the US, zero in the UK). See note below.
- time.tzname¶
A tuple of two strings: the first is the name of the local non-DST timezone, thesecond is the name of the local DST timezone. If no DST timezone is defined,the second string should not be used. See note below.
Note
For the above Timezone constants (altzone, daylight, timezone,and tzname), the value is determined by the timezone rules in effectat module load time or the last time tzset() is called and may be incorrectfor times in the past. It is recommended to use the tm_gmtoff andtm_zone results from localtime() to obtain timezone information.
See also
- Module
datetime
More object-oriented interface to dates and times.
- Module
locale
Internationalization services. The locale setting affects the interpretationof many format specifiers in strftime() and strptime().
- Module
calendar
General calendar-related functions.
timegm()
is theinverse of gmtime() from this module.
Footnotes