BisTiming
A logging-friendly stopwatch and profiling tool for Python.
When we search the stopwatch or timing module for Python on the internet, we can find a lot of code snippets, but none of them is powerful or convenient enough to do our daily jobs. BisTiming aims at implementing all the missing functions in those code snippets and prevents us from reinventing the wheel. It is very useful when we want to log something with some timing information or optimize the performance of our code.
This package is tested with Python 2.7, 3.5, 3.6 and 3.7, but might also work in other Python versions.
Installation
pip install bistimingGetting Started
BisTiming has a context manager interface that logs the running time of a code block easily, and it also offers a low-level API to help time multiple segments or loops of code easily.
See examples for all the useful examples.
Context Manager
The simplest way to use BisTiming is by using the context manager Stopwatch
and include the code we want to evaluate:
>>> from bistiming import Stopwatch
>>> from time import sleep
>>> with Stopwatch("Waiting"):
... print("do something")
... sleep(0.1)
... print("finished something")
...
...Waiting
do something
finished something
...Waiting done in 0:00:00.100330We can use the parameter logger and logging_level to tell the stopwatch to output using a logger:
>>> import logging
>>> logging.basicConfig(
... level=logging.DEBUG,
... format="[%(asctime)s] %(levelname)s: %(name)s: %(message)s")
>>> logger = logging.getLogger(__name__)
>>> with Stopwatch("Waiting", logger=logger, logging_level=logging.DEBUG):
... print("do something")
... sleep(0.1)
... print("finished something")
...
[2019-04-24 22:27:52,347] DEBUG: __main__: ...Waiting
do something
finished something
[2019-04-24 22:27:52,448] DEBUG: __main__: ...Waiting done in 0:00:00.100344Another common use case is to evaluate the running time of a specific code segment in a loop, we can initialize the stopwatch outside the loop, and reuse it in the loop:
>>> timer = Stopwatch("Waiting")
>>> for i in range(2):
... with timer:
... print("do something 1")
... sleep(0.1)
... print("finished something 1")
... print("do something 2")
... sleep(0.1)
... print("finished something 2")
...
...Waiting
do something 1
finished something 1
...Waiting done in 0:00:00.100468
do something 2
finished something 2
...Waiting
do something 1
finished something 1
...Waiting done in 0:00:00.100440
do something 2
finished something 2
>>> timer.split_elapsed_time
[datetime.timedelta(microseconds=100468),
datetime.timedelta(microseconds=100440)]
>>> timer.get_cumulative_elapsed_time()
datetime.timedelta(microseconds=200908)Each item in split_elapsed_time is the running time of
the code segment in each iteration, and we can use
get_cumulative_elapsed_time()
to get the total running time of the code segment.
Low-level API
The low-level API is similar to a stopwatch in real life. A simple use case using the low-level API is:
>>> from time import sleep
>>> from bistiming import Stopwatch
>>> timer = Stopwatch("Waiting").start()
...Waiting
>>> sleep(0.2) # do the first step of my program
>>> timer.split()
...Waiting done in 0:00:00.201457
>>> sleep(0.1) # do the second step of my program
>>> timer.split()
...Waiting done in 0:00:00.100982The context manager
>>> with Stopwatch("Waiting"):
... sleep(0.1)
...Waiting
...Waiting done in 0:00:00.100330is actually equivalent to the low-level API:
>>> timer = Stopwatch("Waiting").start()
...Waiting
>>> sleep(0.1)
>>> timer.pause()
>>> timer.split()
...Waiting done in 0:00:00.100330Advance Profiling
MultiStopwatch in this package contains multiple
Stopwatch, so we can use them to define each code segment
we want to evaluate and compare easily:
>>> from time import sleep
>>> from bistiming import MultiStopwatch
>>> timers = MultiStopwatch(2, verbose=False)
>>> for i in range(5):
... for i in range(2):
... with timers[0]:
... sleep(0.1)
... with timers[1]:
... sleep(0.1)
...
>>> print(timers.format_statistics())
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโคโโโโโโโโโโโโโโโคโโโโโโโโโโโโโคโโโโโโโโโโโโโโโโโโโ
โ cumulative_elapsed_time โ percentage โ n_splits โ mean_per_split โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโชโโโโโโโโโโโโโโโชโโโโโโโโโโโโโชโโโโโโโโโโโโโโโโโโโก
โ 0:00:01.002417 โ 0.666377 โ 10 โ 0:00:00.100242 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโผโโโโโโโโโโโโโโโผโโโโโโโโโโโโโผโโโโโโโโโโโโโโโโโโโค
โ 0:00:00.501861 โ 0.333623 โ 5 โ 0:00:00.100372 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโงโโโโโโโโโโโโโโโงโโโโโโโโโโโโโงโโโโโโโโโโโโโโโโโโโDocumentation
There are a lot more ways to use this package. See the documentation for more information.
