profile-decorator

Problem/Motivation

Find memory leaks in Python code quickly without significant code changes in a quickly digestible manner such that the root cause can be easily diagnosed.

Until open-telemetry Adds metrics API, I want a quick and dirty decorator that will give me the memory usage of the process in scope before and after a function is run.

Usage

# example.py
from profile_decorator import profile_decorator

profile_decorator.init()


@profile_decorator.profile_memory
def my_func():
    print("hello, world")


if __name__ == "__main__":
    my_func()

$ poetry run python example.py
hello, world
{
  "start_time": "2021-08-29T13:20:31.435036",
  "uss_memory_before": 6852608,
  "end_time": "2021-08-29T13:20:31.435892",
  "uss_memory_after": 6995968,
  "lines": [
    {
      "filename": "/home/harry/projects/GitHub/hkiang01/profile-decorator/example.py",
      "lineno": 7,
      "size": 752,
      "size_diff": 0
    },
    {
      "filename": "/home/harry/projects/GitHub/hkiang01/profile-decorator/example.py",
      "lineno": 12,
      "size": 560,
      "size_diff": 0
    }
  ]
}

Design

The decorator will collect the following information:

• name of function
• filepath
• line number
• memory usage before function is run
• timestamp when function is called
• timestamp when function is finished
• memory usage after function is run

It will expose this in json or other easily digestble format for reporting purposes.