codetiming - A flexible, customizable timer for your Python code

Install codetiming from PyPI:

$ python -m pip install codetiming

The source code is available on GitHub.

For a complete tutorial on codetiming, see Python Timer Functions: Three Ways to Monitor Your Code on Real Python.

Basic Usage

You can use codetiming.Timer in several different ways:

1. As a class:

   t = Timer(name="class")
   t.start()
   # Do something
   t.stop()

2. As a context manager:

   with Timer(name="context manager"):
       # Do something

3. As a decorator:

   @Timer(name="decorator")
   def stuff():
       # Do something

Arguments

Timer accepts the following arguments when it's created. All arguments are optional:

• name: An optional name for your timer
• text: The text shown when your timer ends. It should contain a {} placeholder that will be filled by the elapsed time in seconds (default: "Elapsed time: {:.4f} seconds")
• logger: A function/callable that takes a string argument and will report the elapsed time when the logger is stopped (default: print())

You can turn off explicit reporting of the elapsed time by setting logger=None.

In the template text, you can also use explicit attributes to refer to the name of the timer or log the elapsed time in milliseconds, seconds (the default), or minutes. For example:

t1 = Timer(name="NamedTimer", text="{name}: {minutes:.1f} minutes")
t2 = Timer(text="Elapsed time: {milliseconds:.0f} ms")

Note that the strings used by text are not f-strings. Instead, they are used as templates that will be populated using .format() behind the scenes. If you want to combine the text template with an f-string, you need to use double braces for the template values:

t = Timer(text=f"{__file__}: {{:.4f}}")

text is also allowed to be a callable like a function or a class. If text is a callable, it is expected to require one argument: the number of seconds elapsed. It should return a text string that will be logged using logger:

t = Timer(text=lambda secs: f"{secs / 86400:.0f} days")

This allows you to use third-party libraries like humanfriendly to do the text formatting:

from humanfriendly import format_timespan

t1 = Timer(text=format_timespan)
t2 = Timer(text=lambda secs: f"Elapsed time: {format_timespan(secs)}")

Capturing the Elapsed Time

When using Timer as a class, you can capture the elapsed time when calling .stop():

elapsed_time = t.stop()

You can also find the last measured elapsed time in the .last attribute. The following code will have the same effect as the previous example:

t.stop()
elapsed_time = t.last

Named Timers

Named timers are made available in the class dictionary Timer.timers. The elapsed time will accumulate if the same name or same timer is used several times. Consider the following example:

>>> import logging
>>> from codetiming import Timer

>>> t = Timer("example", text="Time spent: {:.2f}", logger=logging.warning)

>>> t.start()
>>> t.stop()
WARNING:root:Time spent: 3.58
3.5836678670002584

>>> with t:
...     _ = list(range(100_000_000))
... 
WARNING:root:Time spent: 1.73

>>> Timer.timers
{'example': 5.312697440000193}

The example shows how you can redirect the timer output to the logging module. Note that the elapsed time spent in the two different uses of t has been accumulated in Timer.timers.

You can also get simple statistics about your named timers. Continuing from the example above:

>>> Timer.timers.max("example")
3.5836678670002584

>>> Timer.timers.mean("example")
2.6563487200000964

>>> Timer.timers.stdev("example")
1.311427314335879

timers support .count(), .total(), .min(), .max(), .mean(), .median(), and .stdev().

Acknowledgments

codetiming is based on a similar module initially developed for the Midgard Geodesy library at the Norwegian Mapping Authority.