monkeytime 0.2.1

Patch datetime for functions and increased performance of strptime

Patch datetime to use a much quicker strptime implementation.

This ONLY improves strptime when the format string uses padded directives.

This is the list of supported directives (by far the most common):

SUPPORTED_DIRECTIVES = ('%d', '%m', '%Y', '%H', '%M', '%S', '%f')

Check if your date/time format is supported by calling:

datetime.supported_strptime(format_string)
# True if supported, False otherwise

On average, 4 times quicker than the stdlib version, and 40 times quicker when using pypy!

Usage:

# insert this one line
from monkeytime import datetime
# Done migrating code to monkeytime!

from datetime import datetime

# Call it once with the specific format string
# It will build a new function to quickly parse the string, then run it on
# the string you passed it.
dt = datetime.strptime('2014-05-30T12:14:15.123456', '%Y-%m-%dT%H:%M:%S.%f')

# Now, whenever you use that format again, it will use the function it
# built already! All constructed functions are memoized.
# Since we usually only use one or two formats in our code to parse logs
# and such, the performance of constructing it is negligible.

All you need is that one line “from monkeytime import datetime”, and your strptime performance will increase dramatically.

Make sure it is ABOVE your “from datetime import datetime” import line!

Performance example:

$ python timeit_test.py
Testing builtin strptime
testing ('2015-01-02 03:04:05.001234', '%Y-%m-%d %H:%M:%S.%f')
6.80288290977 seconds
testing ('05-06 12:15:18', '%m-%d %H:%M:%S')
5.81049013138 seconds
testing ('2010', '%Y')
4.29107117653 seconds
testing ('1905/08/05', '%Y/%m/%d')
4.92634987831 seconds
testing ('14:05:03.123456', '%H:%M:%S.%f')
5.6812889576 seconds
Testing monkeytime strptime
testing ('2015-01-02 03:04:05.001234', '%Y-%m-%d %H:%M:%S.%f')
1.83126211166 seconds
testing ('05-06 12:15:18', '%m-%d %H:%M:%S')
1.5586848259 seconds
testing ('2010', '%Y')
0.877351999283 seconds
testing ('1905/08/05', '%Y/%m/%d')
1.24154901505 seconds
testing ('14:05:03.123456', '%H:%M:%S.%f')
1.3871409893 seconds
3.714860 times as fast
3.727816 times as fast
4.890935 times as fast
3.967906 times as fast
4.095682 times as fast
Average: 4.079440 times as fast

$ pypy timeit_test.py
Testing builtin strptime
testing ('2015-01-02 03:04:05.001234', '%Y-%m-%d %H:%M:%S.%f')
2.19319605827 seconds
testing ('05-06 12:15:18', '%m-%d %H:%M:%S')
1.60669994354 seconds
testing ('2010', '%Y')
0.858637809753 seconds
testing ('1905/08/05', '%Y/%m/%d')
1.19449591637 seconds
testing ('14:05:03.123456', '%H:%M:%S.%f')
1.38721394539 seconds
Testing monkeytime strptime
testing ('2015-01-02 03:04:05.001234', '%Y-%m-%d %H:%M:%S.%f')
0.0362730026245 seconds
testing ('05-06 12:15:18', '%m-%d %H:%M:%S')
0.0450730323792 seconds
testing ('2010', '%Y')
0.0332229137421 seconds
testing ('1905/08/05', '%Y/%m/%d')
0.0321681499481 seconds
testing ('14:05:03.123456', '%H:%M:%S.%f')
0.0355319976807 seconds
60.463593 times as fast
35.646591 times as fast
25.844747 times as fast
37.132876 times as fast
39.041260 times as fast
Average: 39.625813 times as fast