A timer context manager in python

When writing code you may want to quickly do performance analysis on your algorithms. The most likely tool that you need is obviously a profiler but if that is too much of a hassle to set up, the quickest way is to write your own time measuring tool.

The easiest way of doing it is:

import time
 
 actual_time = time.perf_counter()
 comput_time = time.process_time()
 
 result = f(*args, **kw)
 
 actual_time = time.perf_counter() - actual_time
 comput_time = time.process_time() - comput_time

For a better-looking code, you may wish to use a decorator if all you need is to print the execution times on `stdout`

import time
 from functools import wraps
 
 def timing(f):
  """
  Decorator to be used to measure the time it takes for a single call
  """
 
  @wraps(f)
  def wrap(*args, **kw):
  actual_time = time.perf_counter()
  comput_time = time.process_time()
 
  result = f(*args, **kw)
 
  actual_time = time.perf_counter() - actual_time
  comput_time = time.process_time() - comput_time
 
  print(
  '[%2.4fs|%2.4fs] func:%r args:[%r, %r]' % (
  actual_time,
  comput_time,
  f.__name__,
  args if len(args) < 1000 else "<something large>",
  kw
  )
  )
  return result
 
  return wrap
 

Example:

import time
 
 @timing
 def my_heavy_function(a, b):
  time.sleep(1)
 
 my_heavy_function(1, "2")
 
 > [1.0011s|0.0014s] func:'my_heavy_function' args:[(1, '2'), {}]

But if you want to have access to the `actual_time` and `comput_time` you may want to use a context manager (inspired by this thread):

import time
 
 class timer(object):
  """
  A small timer context manager
  """
  def __enter__(self):
  self.actual_time = time.perf_counter()
  self.comput_time = time.process_time()
  return self
 
  def __exit__(self, type, value, traceback):
  self.actual_time = time.perf_counter() - self.actual_time
  self.comput_time = time.process_time() - self.comput_time

Example:

with timer() as t:
  time.sleep(4)
 
 print(t.actual_time, t.comput_time)
 
 > 4.004134978000366 0.0028222399996593595