5. Date and Time in Python

Section 5.1: Parsing a string into a timezone aware datetime object

Python 3.2+ has support for %z format when parsing a string into a datetime object.

UTC oset in the form +HHMM or -HHMM (empty string if the object is naive).

Python 3.x Version ≥ 3.2

import datetime

dt datetime.datetime.strptime(“2016-04-15T08:27:18-0500”, “%Y-%m-%dT%H:%M:%S%z”)

For other versions of Python, you can use an external library such as dateutil, which makes parsing a string with timezone into a datetime object is quick.

import dateutil.parser

dt dateutil.parser.parse(“2016-04-15T08:27:18-0500”)

The dt variable is now a datetime object with the following value:


Section 5.2: Constructing timezone-aware datetimes

By default all datetime objects are naive. To make them timezone-aware, you must attach a tzinfo object, which provides the UTC oset and timezone abbreviation as a function of date and time.

Fixed Oset Time Zones

For time zones that are a fixed oset from UTC, in Python 3.2+, the datetime module provides the timezone class, a concrete implementation of tzinfo, which takes a timedelta and an (optional) name parameter:

Python 3.x Version ≥ 3.2

from datetime import datetimetimedeltatimezone JST timezone(timedelta(hours=+9))

dt datetime(2015111200tzinfo=JST) print(dt)

# 2015-01-01 12:00:00+09:00


# UTC+09:00

dt datetime(2015111200tzinfo=timezone(timedelta(hours=9)‘JST’)) print(dt.tzname)

# ‘JST’

For Python versions before 3.2, it is necessary to use a third party library, such as dateutildateutil provides an equivalent class, tzoffset, which (as of version 2.5.3) takes arguments of the form dateutil.tz.tzoffset(tznameoffset), where offset is specified in seconds:

Python 3.x Version < 3.2

Python 2.x Version < 2.7

from datetime import datetimetimedelta from dateutil import tz

JST tz.tzoffset(‘JST’3600# 3600 seconds per hour dt datetime(201511120tzinfo=JST)


#2015-01-01 12:00:00+09:00 print(dt.tzname)


Zones with daylight savings time

For zones with daylight savings time, python standard libraries do not provide a standard class, so it is necessary to use a third party library. pytz and dateutil are popular libraries providing time zone classes.

In addition to static time zones, dateutil provides time zone classes that use daylight savings time (see the documentation for the tz module). You can use the tz.gettz() method to get a time zone object, which can then be passed directly to the datetime constructor:

from datetime import datetime from dateutil import tz

local tz.gettz() # Local time

PT tz.gettz(‘US/Pacific’# Pacific time

dt_l datetime(20151112tzinfo=local) # I am in EST dt_pst datetime(20151112tzinfo=PT)

dt_pdt datetime(20157112tzinfo=PT) # DST is handled automatically print(dt_l)

#2015-01-01 12:00:00-05:00print(dt_pst)

#2015-01-01 12:00:00-08:00print(dt_pdt)

#2015-07-01 12:00:00-07:00

CAUTION: As of version 2.5.3, dateutil does not handle ambiguous datetimes correctly, and will always default to the later date. There is no way to construct an object with a dateutil timezone representing, for example 20151101 1:30 EDT-4, since this is during a daylight savings time transition.

All edge cases are handled properly when using pytz, but pytz time zones should not be directly attached to time zones through the constructor. Instead, a pytz time zone should be attached using the time zone’s localize method:

from datetime import datetimetimedelta import pytz

PT pytz.timezone(‘US/Pacific’)

dt_pst PT.localize(datetime(20151112))

dt_pdt PT.localize(datetime(2015111030)) print(dt_pst)

#2015-01-01 12:00:00-08:00print(dt_pdt)

#2015-11-01 00:30:00-07:00

Be aware that if you perform datetime arithmetic on a pytz-aware time zone, you must either perform the calculations in UTC (if you want absolute elapsed time), or you must call normalize() on the result:


dt_new dt_pdt + timedelta(hours=3# This should be 2:30 AM PST print(dt_new)

# 2015-11-01 03:30:00-07:00

dt_corrected PT.normalize(dt_new) print(dt_corrected)

# 2015-11-01 02:30:00-08:00

Section 5.3: Computing time dierences

the timedelta module comes in handy to compute dierences between times:

from datetime import datetimetimedelta now datetime.now()

then datetime(2016523# datetime.datetime(2016, 05, 23, 0, 0, 0)

Specifying time is optional when creating a new datetime object

delta now-then

delta is of type timedelta


# 60


# 40826

To get n day’s after and n day’s before date we could use:

n day’s after date:

def get_n_days_after_date(date_format=“%d %B %Y”add_days=120):

date_n_days_after datetime.datetime.now() + timedelta(days=add_days) return date_n_days_after.strftime(date_format)

n day’s before date:

def get_n_days_before_date(selfdate_format=“%d %B %Y”days_before=120):

date_n_days_ago datetime.datetime.now() – timedelta(days=days_before) return date_n_days_ago.strftime(date_format)

Section 5.4: Basic datetime objects usage

The datetime module contains three primary types of objects – date, time, and datetime.

import datetime

# Date object

today datetime.date.today()

new_year datetime.date(20170101#datetime.date(2017, 1, 1)

# Time object

noon datetime.time(1200#datetime.time(12, 0)

# Current datetime

now datetime.datetime.now()

# Datetime object

millenium_turn datetime.datetime(200011000#datetime.datetime(2000, 1, 1, 0, 0)

Arithmetic operations for these objects are only supported within same datatype and performing simple arithmetic with instances of dierent types will result in a TypeError.

#subtraction of noon from todaynoon-today

Traceback (most recent call last): File “<stdin>”line 1in <module>

TypeError: unsupported operand type(s) for -: ‘datetime.time’ and ‘datetime.date’ Howeverit is straightforward to convert between types.

#Do this instead

print(‘Time since the millenium at midnight: ‘datetime.datetime(today.yeartoday.monthtoday.day) – millenium_turn)

# Or this

print(‘Time since the millenium at noon: ‘datetime.datetime.combine(todaynoon) – millenium_turn)

Section 5.5: Switching between time zones

To switch between time zones, you need datetime objects that are timezone-aware.

from datetime import datetime from dateutil import tz

utc tz.tzutc() local tz.tzlocal()

utc_now datetime.utcnow() utc_now # Not timezone-aware.

utc_now utc_now.replace(tzinfo=utc) utc_now Timezone-aware.

local_now utc_now.astimezone(local) local_now # Converted to local time.

Section 5.6: Simple date arithmetic

Dates don’t exist in isolation. It is common that you will need to find the amount of time between dates or determine what the date will be tomorrow. This can be accomplished using timedelta objects

import datetime

today datetime.date.today() print(‘Today:’today)

yesterday today – datetime.timedelta(days=1print(‘Yesterday:’yesterday)

tomorrow today + datetime.timedelta(days=1print(‘Tomorrow:’tomorrow)

print(‘Time between tomorrow and yesterday:’tomorrow – yesterday)

This will produce results similar to:

Today: 2016-04-15

Yesterday: 2016-04-14

Tomorrow: 2016-04-16

Difference between tomorrow and yesterday: 2 days, 0:00:00

Section 5.7: Converting timestamp to datetime

The datetime module can convert a POSIX timestamp to a ITC datetime object.

The Epoch is January 1st, 1970 midnight.

import time

from datetime import datetime

seconds_since_epoch=time.time() #1469182681.709

utc_date=datetime.utcfromtimestamp(seconds_since_epoch) #datetime.datetime(2016, 7, 22, 10, 18, 1, 709000)

Section 5.8: Subtracting months from a date accurately

Using the calendar module

import calendar

from datetime import date

def monthdelta(datedelta):

m(date.month+delta) % 12date.year + ((date.month)+delta-1) // 12 if not m: m 12

min(date.daycalendar.monthrange(ym)[1]) return date.replace(day=d,month=myear=y)

next_month monthdelta(date.today()1#datetime.date(2016, 10, 23)

Using the dateutils module

import datetime

import dateutil.relativedelta

datetime.datetime.strptime(“2013-03-31”, “%Y-%m-%d”)

d2 d – dateutil.relativedelta.relativedelta(months=1#datetime.datetime(2013, 2, 28, 0, 0)

Section 5.9: Parsing an arbitrary ISO 8601 timestamp with minimal libraries

Python has only limited support for parsing ISO 8601 timestamps. For strptime you need to know exactly what format it is in. As a complication the stringification of a datetime is an ISO 8601 timestamp, with space as a separator and 6 digit fraction:


# ‘2016-07-22 09:25:59.555555’

but if the fraction is 0, no fractional part is output



# ‘2016-07-22 09:25:59’

But these 2 forms need a dierent format for strptime. Furthermore, strptime‘ does not support at all parsing minute timezones that have a:in itthus2016-07-22 09:25:59+0300can be parsedbut the standard format2016-07-2209:25:59+03:00` cannot.

There is a single-file library called iso8601 which properly parses ISO 8601 timestamps and only them. It supports fractions and timezones, and the T separator all with a single function:

import iso8601

iso8601.parse_date(‘2016-07-22 09:25:59’)

#datetime.datetime(2016, 7, 22, 9, 25, 59, tzinfo=<iso8601.Utc>)iso8601.parse_date(‘2016-07-22 09:25:59+03:00′)

#datetime.datetime(2016, 7, 22, 9, 25, 59, tzinfo=<FixedOffset ‘+03:00’ …>)iso8601.parse_date(‘2016-07-22 09:25:59Z’)

#datetime.datetime(2016, 7, 22, 9, 25, 59, tzinfo=<iso8601.Utc>)iso8601.parse_date(‘2016-07-22T09:25:59.000111+03:00’)

#datetime.datetime(2016, 7, 22, 9, 25, 59, 111, tzinfo=<FixedOffset ‘+03:00’ …>)

If no timezone is set, iso8601.parse_date defaults to UTC. The default zone can be changed with default_zone keyword argument. Notably, if this is None instead of the default, then those timestamps that do not have an explicit timezone are returned as naive datetimes instead:

iso8601.parse_date(‘2016-07-22T09:25:59’, default_timezone=None)

# datetime.datetime(2016, 7, 22, 9, 25, 59)

iso8601.parse_date(‘2016-07-22T09:25:59Z’, default_timezone=None)

# datetime.datetime(2016, 7, 22, 9, 25, 59, tzinfo=<iso8601.Utc>)

Section 5.10: Get an ISO 8601 timestamp

Without timezone, with microseconds

from datetime import datetime


#Out: ‘2016-07-31T23:08:20.886783’

With timezone, with microseconds

from datetime import datetime from dateutil.tz import tzlocal


#Out: ‘2016-07-31T23:09:43.535074-07:00’

With timezone, without microseconds

from datetime import datetime from dateutil.tz import tzlocal


# Out: ‘2016-07-31T23:10:30-07:00’

See ISO 8601 for more information about the ISO 8601 format.

Section 5.11: Parsing a string with a short time zone name into

a timezone aware datetime object

Using the dateutil library as in the previous example on parsing timezone-aware timestamps, it is also possible to parse timestamps with a specified “short” time zone name.

For dates formatted with short time zone names or abbreviations, which are generally ambiguous (e.g. CST, which could be Central Standard Time, China Standard Time, Cuba Standard Time, etc – more can be found here) or not necessarily available in a standard database, it is necessary to specify a mapping between time zone abbreviation and tzinfo object.

from dateutil import tz

from dateutil.parser import parse

ET tz.gettz(‘US/Eastern’)

CT tz.gettz(‘US/Central’)

MT tz.gettz(‘US/Mountain’)

PT tz.gettz(‘US/Pacific’)

us_tzinfos {‘CST’: CT‘CDT’: CT‘EST’: ET‘EDT’: ET‘MST’: MT‘MDT’: MT‘PST’: PT‘PDT’: PT}

dt_est parse(‘2014-01-02 04:00:00 EST’tzinfos=us_tzinfos) dt_pst parse(‘2016-03-11 16:00:00 PST’tzinfos=us_tzinfos)

After running this:


#datetime.datetime(2014, 1, 2, 4, 0, tzinfo=tzfile(‘/usr/share/zoneinfo/US/Eastern’)) dt_pst

#datetime.datetime(2016, 3, 11, 16, 0, tzinfo=tzfile(‘/usr/share/zoneinfo/US/Pacific’))

It is worth noting that if using a pytz time zone with this method, it will not be properly localized:

from dateutil.parser import parse import pytz

EST pytz.timezone(‘America/New_York’)

dt parse(‘2014-02-03 09:17:00 EST’tzinfos={‘EST’: EST})

This simply attaches the pytz time zone to the datetime:

dt.tzinfo # Will be in Local Mean Time!

# <DstTzInfo ‘America/New_York’ LMT-1 day, 19:04:00 STD>

If using this method, you should probably re-localize the naive portion of the datetime after parsing:

dt_fixed dt.tzinfo.localize(dt.replace(tzinfo=None)) dt_fixed.tzinfo # Now it’s EST.

# <DstTzInfo ‘America/New_York’ EST-1 day, 19:00:00 STD>)

Section 5.12: Fuzzy datetime parsing (extracting datetime out of a text)

It is possible to extract a date out of a text using the dateutil parser in a “fuzzy” mode, where components of the

string not recognized as being part of a date are ignored.

from dateutil.parser import parse

dt parse(“Today is January 1, 2047 at 8:21:00AM”fuzzy=Trueprint(dt)

dt is now a datetime object and you would see datetime.datetime(204711821) printed.

Section 5.13: Iterate over dates

Sometimes you want to iterate over a range of dates from a start date to some end date. You can do it using datetimelibrary and timedelta object:

import datetime

# The size of each step in days

day_delta datetime.timedelta(days=1)

start_date datetime.date.today() end_date start_date + 7*day_delta

for in range((end_date – start_date).days): print(start_date + i*day_delta)

Which produces:










  *This content is compiled from Stack Overflow Documentation, and the content is written by the beautiful people at Stack Overflow.  This work is licensed under cc by-sa.



*This content is compiled from Stack Overflow Documentation, and the content is written by the beautiful people at Stack Overflow.