A Pythonic way to time your code block

A simple utility to time your python code block.

Get this python package to use this utility (and others) with ease..

Pre-requisites

1. Context Manager - an object that defines runtime context for the desired code block.
   • __enter__() method handles the entry into the code block - as in execute this before executing the code block..
   • __exit__(exc_type, exc_val, exc_tb) method handles the exit from the runtime context - as in execute this after executing the code block..
   • Context manager objects find usage in saving and restoring various kinds of global state, locking and unlocking resources, closing opened files, etc.
2. The with statement - This statement encapsulates try…except…finally usage patterns with the help of context manager objects. The desired behavior is defined in the context manager and used in the with statement to execute it.

Timer context manager

Using @contextmanager decorator, we will define a timer context manager. Once we define this we can reuse this as required. The desired behavior is to -

1. note start time before executing the code block
2. yield to execute the code block
3. note the end time and print the execution time

We will define a generator-iterator to do this. There is no need to return a value but if desired in we can return the start time.

from contextlib import contextmanager
import time

# Simplest
@contextmanager
def timer(code_block_name: str) -> None:
    start_time = time.time()
    # yields the control to the with statement
    yield
    # executed after with code block execution
    execution_time = time.time() - start_time
    print(f"[{code_block_name}] completed in {execution_time:.0f} seconds..")

Usage

import pandas as pd
import numpy as np

df = pd.DataFrame({'X' : np.random.randint(low=-10, high=10, size=1000000),
                   'Y' : np.random.normal(0.0, 1.0, size=1000000)})

with timer("Square root of abs col 'X'"):
    df['X_square_root'] = df['X'].apply(lambda x: np.sqrt(abs(x)))

[Square root of abs col 'X'] completed in 1 seconds..

Another version

Context manager yielding start time.

from contextlib import contextmanager
import time
from datetime import datetime

# Return Start Date and time
# just to show the use of `as` assignment in the `with` statement
@contextmanager
def timer_v2(code_block_name: str) -> None:
    start_time = time.time()
    # yields the value to the with statement
    yield datetime.now()
    # executed after with code block execution
    execution_time = time.time() - start_time
    print(f"[{code_block_name}] completed in {execution_time:.0f} seconds..")

Using the v2 version - which yields the start time. The start time can be assigned to a variable in the with statement as follows..

import pandas as pd
import numpy as np

df = pd.DataFrame({'X' : np.random.randint(low=-10, high=10, size=1000000),
                   'Y' : np.random.normal(0.0, 1.0, size=1000000)})

with timer_v2("Square root of abs col 'X'") as current_time:
    print(f"Started at {current_time}")
    df['X_square_root'] = df['X'].apply(lambda x: np.sqrt(abs(x)))

Started at 2020-07-11 13:45:45.194144
[Square root of abs col 'X'] completed in 1 seconds..

Class based implementation of timer

import time

class Timer:
    def __init__(self, code_block_name):
        # init the context manager with a name
        self.code_block_name = code_block_name
    def __enter__(self):
        # executes and yields the control to the code block
        self.start_time = time.time()
    def __exit__(self, type, value, traceback):
        # executed after with code block execution
        execution_time = time.time() - self.start_time
        print(f"[{self.code_block_name}] completed in {execution_time:.0f} seconds..")

with Timer("for loop 100,000,000"):
    _ = [x for x in range(100_000_000)]

[for loop 100,000,000] completed in 4 seconds..