line-profiler-decorator

line-profiler-decorator

Wrapper around line-profiler adding a few features as an decorator

Install

pip install line-profiler-decorator

Usage

from time import sleep
from line_profiler_decorator import profiler


@profiler
def slow_function():
    print("quick line")
    sleep(1)  # slow one
    sleep(0.3)  # not that slow

Output

quick line
Timer unit: 1e-06 s

Total time: 6.52823 s
File: /Users/fabio/projects/line-profiler-decorator/examples/example.py
Function: slow_function at line 5

Line #      Hits         Time  Per Hit   % Time  Line Contents
==============================================================
     5                                           @profiler(aggregate=True)
     6                                           def slow_function():
     7         5         69.0     13.8      0.0      print("quick line")
     8         5    5016944.0 1003388.8     76.8     sleep(1)  # slow one
     9         5    1511218.0 302243.6      23.1     sleep(0.3)  # not that slow

Options

Save output to a file

Save the output to a file instead of stdout

@profiler("output_file.txt")
def slow_function():
    print("quick line")
    sleep(1)  # slow one
    sleep(0.3)  # not that slow

Aggregate results

Aggregate the results of multiple calls to the decorated function

@profiler(aggregate=True)
def slow_function():
    print("quick line")
    sleep(1)  # slow one
    sleep(0.3)  # not that slow

def run_slow(n=100):
    for _ in range(n):
        slow_function()

Follow function

from time import sleep
from line_profiler_decorator import profiler


def other_function():
    sleep(1)  # slow one
    sleep(0.3)  # not that slow


@profiler(follow=[other_function])
def function_that_calls_other_function(n):
    for _ in range(n):
        other_function()

Output

In [3]: function_that_calls_other_function(10)                                                                                                                                      
Timer unit: 1e-06 s

Total time: 13.0668 s
File: /Users/fabio/projects/line-profiler-decorator/examples/example.py
Function: other_function at line 17

Line #      Hits         Time  Per Hit   % Time  Line Contents
==============================================================
    17                                           def other_function():
    18        10   10037030.0 1003703.0     76.8      sleep(1)  # slow one
    19        10    3029725.0 302972.5     23.2      sleep(0.3)  # not that slow

Total time: 13.0672 s
File: /Users/fabio/projects/line-profiler-decorator/examples/example.py
Function: function_that_calls_other_function at line 22

Line #      Hits         Time  Per Hit   % Time  Line Contents
==============================================================
    22                                           @profiler(follow=[other_function])
    23                                           def function_that_calls_other_function(n):
    24        11         97.0      8.8      0.0      for _ in range(n):
    25        10   13067104.0 1306710.4    100.0          other_function()

Grouped aggregations

Group aggregations using a key function. key is a function that receives the same arguments as the decorated function and returns a value. The profile results of each call of the decorated function will be grouped and aggregated together by the value returned by the key function

Example:

from time import sleep
from line_profiler_decorator import profiler


@profiler(aggregate=True, key=lambda x: x % 2 == 0)
def function_with_grouped_aggregations(number):
    """This function runs faster for even numbers."""
    if number % 2:
        # sleep for 1 second if number is odd
        sleep(1)
    else:
        # sleep for 0.2 second if number is even
        sleep(0.2)
    sleep(0.5)


def run_function_with_grouped_aggregations(n=10):
    for _ in range(n):
        function_with_grouped_aggregations(_)

Output

The key function returns value True for even numbers and False for odd numbers. That way the aggregated results will be grouped into 2 profile results.

Timer unit: 1e-06 s

Total time: 3.533 s
File: /Users/fabio/projects/line-profiler-decorator/examples/example.py
Function: function_with_filtered_aggregations at line 28

Line #      Hits         Time  Per Hit   % Time  Line Contents
==============================================================
    28                                           @profiler(aggregate=True, key=lambda x: x % 2 == 0)
    29                                           def function_with_filtered_aggregations(number):
    30                                               """This function runs faster for even numbers."""
    31         5         17.0      3.4      0.0      if number % 2:
    32                                                   # sleep for 1 second if number is odd
    33                                                   sleep(1)
    34                                               else:
    35                                                   # sleep for 0.2 second if number is even
    36         5    1015312.0 203062.4     28.7          sleep(0.2)
    37         5    2517674.0 503534.8     71.3      sleep(0.5)

Timer unit: 1e-06 s

Total time: 7.53658 s
File: /Users/fabio/projects/line-profiler-decorator/examples/example.py
Function: function_with_filtered_aggregations at line 28

Line #      Hits         Time  Per Hit   % Time  Line Contents
==============================================================
    28                                           @profiler(aggregate=True, key=lambda x: x % 2 == 0)
    29                                           def function_with_filtered_aggregations(number):
    30                                               """This function runs faster for even numbers."""
    31         5         15.0      3.0      0.0      if number % 2:
    32                                                   # sleep for 1 second if number is odd
    33         5    5017337.0 1003467.4     66.6          sleep(1)
    34                                               else:
    35                                                   # sleep for 0.2 second if number is even
    36                                                   sleep(0.2)
    37         5    2519231.0 503846.2     33.4      sleep(0.5)

Looking at the results above we can see that the first block only aggregated call with even number therefore spent no time at line 33. On the other hand, the second block shows the same for odd numbers where no time was used on even numbers.