timeprofiles 0.0.5

Easily profile time taken for methods to complete.

timeprofiles

This library is designed to assist in quickly debugging bottlenecks in both single-threaded and multi-threaded operations.

Installation

python -m pip install timeprofiles

Overview

It contains a class called TimeProfileCollection for easily storing the time taken for each method to complete, and displaying it as either a table or a diagram. The TimeProfiler class provides two decorator methods for profiling: one for profiling the entire class, and one for profiling individual methods.

@timeprofiles.profile_class_methods
@timeprofiles.profile_method

The display_profiles method is used to display all time profiles, ordered by a specified column.

Name        Calls    Average (ms)    Longest (ms)    Bottleneck (ms)
--------  -------  --------------  --------------  -----------------
method_a        1          203.50          203.50             203.50
method_b        5          104.82          155.19             155.40
method_c        5           58.86           78.35             140.00
method_d        1           45.83           45.83              45.83
method_e        1           93.76           93.76              93.76

The plot_profiles and plot_merged methods are used to plot all time profiles as ranges, ordered by earliest method call.

Example usages

The profile_method decorator is applied on individual methods or functions.

import timeprofiles as tp
    
class ExampleClass:
    @tp.profile_method
    def my_method():
        # Method content

    @tp.profile_method
    def my_other_method():
        # Method content

Alternatively, the profile_class_methods decorator can be used to apply the profile_method decorator on all class methods. This can be used in conjunction with profile_ignore to filter out methods you don't want to profile.

import timeprofiles as tp

@tp.profile_class_methods
class ExampleClass:
    def my_method():
        # Method content

    @tp.profile_ignore
    def my_other_method():
        # Method content

Another way of using the profiling decorators is to apply it directly onto an object. In the following example, only the methods called from the example_object will be profiled.

example_obj = tp.profile_class_methods(ExampleClass())

After adding the decorators and calling the methods, the display_profiles or the plot_profiles and plot_merged methods can be called to visualize the time profiles.

tp.display_profiles(order_by=tp.ORDER_BY_AVERAGE, reverse=False, full_name=True)
tp.plot_profiles()
tp.plot_merged()

Known issues

Despite being able to add profile_class_methods to an object, profile_ignore cannot be used since its implementation relies on attributes, which cannot be updated on objects.

# THIS WILL NOT WORK
tp.profile_ignore(example_obj)