datetime library

Time is linear

progresses as a straightline trajectory from the big bag

to now and into the future

日期库官方说明 https://docs.python.org/3.5/library/datetime.html

Reasoning about time is important in data analysis

Analyzing financial timeseries data

Looking at commuter transit passenger flows by time of day

Understanding web traffic by time of day

Examining seaonality in department store purchases

The datetime library

understands the relationship between different points of time

understands how to do operations on time

Example:

Which is greater? "10/24/2017" or "11/24/2016"



In [1]:

    
d1 = "10/24/2017"
d2 = "11/24/2016"
max(d1,d2)









    Out[1]:





'11/24/2016'

How much time has passed?



In [2]:

    
d1 - d2









    



---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-2-8e72eafee703> in <module>()
----> 1 d1 - d2

TypeError: unsupported operand type(s) for -: 'str' and 'str'

Obviously that's not going to work.

We can't do date operations on strings

Let's see what happens with datetime



In [1]:

    
import datetime
d1 = datetime.date(2016,11,24)
d2 = datetime.date(2017,10,24)
max(d1,d2)









    



<class 'datetime.date'>



In [4]:

    
print(d2 - d1)









    



334 days, 0:00:00

datetime objects understand time

The datetime library contains several useful types

date: stores the date (month,day,year)

time: stores the time (hours,minutes,seconds)

datetime: stores the date as well as the time (month,day,year,hours,minutes,seconds)

timedelta: duration between two datetime or date objects

datetime.date



In [2]:

    
import datetime
century_start = datetime.date(2000,1,1)
today = datetime.date.today()
print(century_start,today)
print("We are",today-century_start,"days into this century")
print(type(century_start))
print(type(today))









    



2000-01-01 2017-08-25
We are 6446 days, 0:00:00 days into this century
<class 'datetime.date'>
<class 'datetime.date'>

For a cleaner output



In [6]:

    
print("We are",(today-century_start).days,"days into this century")









    



We are 6445 days into this century

datetime.datetime



In [7]:

    
century_start = datetime.datetime(2000,1,1,0,0,0)
time_now = datetime.datetime.now()
print(century_start,time_now)
print("we are",time_now - century_start,"days, hour, minutes and seconds into this century")









    



2000-01-01 00:00:00 2017-08-24 18:09:48.245052
we are 6445 days, 18:09:48.245052 days, hour, minutes and seconds into this century

datetime objects can check validity

A ValueError exception is raised if the object is invalid



In [8]:

    
some_date=datetime.date(2015,2,29)
#some_date =datetime.date(2016,2,29)
#some_time=datetime.datetime(2015,2,28,23,60,0)









    



---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-8-16e372a33db1> in <module>()
----> 1 some_date=datetime.date(2015,2,29)
      2 #some_date =datetime.date(2016,2,29)
      3 #some_time=datetime.datetime(2015,2,28,23,60,0)

ValueError: day is out of range for month

datetime.timedelta

Used to store the duration between two points in time



In [3]:

    
century_start = datetime.datetime(2050,1,1,0,0,0)
time_now = datetime.datetime.now()
time_since_century_start = time_now - century_start
print("days since century start",time_since_century_start.days)
print("seconds since century start",time_since_century_start.total_seconds())
print("minutes since century start",time_since_century_start.total_seconds()/60)
print("hours since century start",time_since_century_start.total_seconds()/60/60)









    



days since century start -11817
seconds since century start -1020947887.555188
minutes since century start -17015798.1259198
hours since century start -283596.63543199666

datetime.time



In [10]:

    
date_and_time_now = datetime.datetime.now()
time_now = date_and_time_now.time()
print(time_now)









    



19:44:37.142884

You can do arithmetic operations on datetime objects

You can use timedelta objects to calculate new dates or times from a given date



In [11]:

    
today=datetime.date.today()
five_days_later=today+datetime.timedelta(days=5)
print(five_days_later)



In [12]:

    
now=datetime.datetime.today()
five_minutes_and_five_seconds_later = now + datetime.timedelta(minutes=5,seconds=5)
print(five_minutes_and_five_seconds_later)









    



2017-08-24 19:50:09.630242



In [13]:

    
now=datetime.datetime.today()
five_minutes_and_five_seconds_earlier = now+datetime.timedelta(minutes=-5,seconds=-5)
print(five_minutes_and_five_seconds_earlier)









    



2017-08-24 21:39:23.763762

But you can't use timedelta on time objects. If you do, you'll get a TypeError exception



In [14]:

    
time_now=datetime.datetime.now().time() #Returns the time component (drops the day)
print(time_now)
thirty_seconds=datetime.timedelta(seconds=30)
time_later=time_now+thirty_seconds
#Bug or feature?









    



22:02:21.552801






    



---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-14-2dea49afbdf4> in <module>()
      2 print(time_now)
      3 thirty_seconds=datetime.timedelta(seconds=30)
----> 4 time_later=time_now+thirty_seconds
      5 #Bug or feature?

TypeError: unsupported operand type(s) for +: 'datetime.time' and 'datetime.timedelta'



In [17]:

    
#But this is Python
#And we can always get around something by writing a new function!
#Let's write a small function to get around this problem
def add_to_time(time_object,time_delta):
    import datetime
    temp_datetime_object = datetime.datetime(500,1,1,time_object.hour,time_object.minute,time_object.second)
    print(temp_datetime_object)
    return (temp_datetime_object+time_delta).time()



In [18]:

    
#And test it
time_now=datetime.datetime.now().time()
thirty_seconds=datetime.timedelta(seconds=30)
print(time_now,add_to_time(time_now,thirty_seconds))









    



0500-01-01 22:37:07
22:37:07.239431 22:37:37

datetime and strings

More often than not, the program will need to get the date or time from a string: From a website (bus/train timings) From a file (date or datetime associated with a stock price) From the user (from the input statement) Python needs to parse the string so that it correctly creates a date or time object

datetime.strptime

datetime.strptime(): grabs time from a string and creates a date or datetime or time object

The programmer needs to tell the function what format the string is using

See http://pubs.opengroup.org/onlinepubs/009695399/functions/strptime.html for how to specify the format



In [30]:

    
date='01-Apr-03'
date_object=datetime.datetime.strptime(date,'%d-%b-%y')
print(date_object)









    



2003-04-01 00:00:00



In [26]:

    
#Unfortunately, there is no similar thing for time delta
#So we have to be creative!
bus_travel_time='2:15:30'
hours,minutes,seconds=bus_travel_time.split(':')
x=datetime.timedelta(hours=int(hours),minutes=int(minutes),seconds=int(seconds))
print(x)



In [27]:

    
#Or write a function that will do this for a particular format
def get_timedelta(time_string):
    hours,minutes,seconds = time_string.split(':')
    import datetime
    return datetime.timedelta(hours=int(hours),minutes=int(minutes),seconds=int(seconds))

datetime.strftime

The strftime function flips the strptime function. It converts a datetime object to a string

with the specified format



In [28]:

    
now = datetime.datetime.now()
string_now = datetime.datetime.strftime(now,'%m/%d/%y %H:%M:%S')
print(now,string_now)
print(str(now)) #Or you can use the default conversion









    



2017-08-24 23:03:35.197581 08/24/17 23:03:35
2017-08-24 23:03:35.197581



In [ ]: