eprofiler

A lightweight, zero-dependency toolkit to monitor execution of functions or code blocks.

eprofiler provides decorators and context managers to observe execution time, cpu time, peak memory and arguments used for a function.

for a function or code block you can monitor and log;

• execution time
• peak memory usage
• CPU time
• parameters passed to a function

Installation

pip install eprofiler

Core Tools

• @audit: Execution logging with SUCCESS/FAIL status and error capturing.
• @timeit: Execution timing (microsecond precision).
• @memit: Tracks current and peak memory usage.
• @profile: The "All-in-One": Wall time, CPU time, and Memory.
• @profile_cpu: User vs System time (Unix) & Efficiency %.
• Timer: Context manager and/or decorator for granular blocks.

---

Usage

1. Function Auditing (@audit)

Good for monitoring what parameters are passed to a function/method. where you need to know if a function finished, how long it took, and why it failed and with which parameters.

⚠ Warning:
if parameters passed to function are not printable like str, int or a python object without __str__ or __repr__ that can be used in fstrings its better to use callback to handle them in logging

from eprofiler import audit

@audit(label="Audit", include_args=True)
def create_user(username, email):
    return f"User {username} created."

Output:

INFO: {'timestamp': '2026-03-05T18:33:37.651931', 'function': 'create_user', 'label': 'Audit', 'args': ('jdoe', 'jane@example.com'), 'kwargs': {}, 'status': 'SUCCESS', 'elapsed_seconds': '0.000003'}

2. Basic Timing (@timeit)

For quick performance checks during development. By default, results are printed to the console.

from eprofiler import timeit

@timeit(label="Computation")
def my_func():
    return sum(i**2 for i in range(100000))

my_func()

Output:

{'label': 'Computation', 'function': 'my_func', 'duration': '0.000074'}

3. Comprehensive Profiling (@profile & @profile_cpu)

Track wall-clock time, actual CPU usage, and memory (current and peak) simultaneously.

from eprofiler import profile, profile_cpu

# Standard profile (Wall Time + CPU Time + Memory)
@profile(label="Data Batch")
def process_data():
    return [x for x in range(1000000)]

# CPU profile (User/System breakdown + Efficiency)
@profile_cpu(label="Heavy Computation")
def compute_pi():
    return sum(1/i**2 for i in range(1, 1000000))

process_data()
compute_pi()

Output:

{'label': 'Data Batch', 'function': 'process_data', 'duration': '0.241128', 'cpu_time': '0.241104', 'peak': 40440488, 'current': 40440448}

{'label': 'Heavy Computation', 'function': 'compute_pi', 'user_time': '0.048526', 'system_time': '0.000074', 'cpu_time': '0.048600', 'duration': '0.048610', 'efficiency': '99.98%'}

4. Using Callback

Instead of printing to the console, you can pass a callback function to any decorator to handle the results programmatically (e.g., sending metrics to a database, Slack, or a logging service).

from eprofiler import profile_cpu

def metrics_handler(stats):
    """Custom function to process profiling data."""
    # Send to Datadog, CloudWatch, or an ELK stack
    print(f"TELEMETRY: {stats['function']} ran with {stats['efficiency']} efficiency.")

@profile_cpu(label="Production_Task", callback=metrics_handler)
def sync_data():
    # Logic here...
    return "Done"

sync_data()

Using your own metrics_handler you can do anything you want with stats.

5. Timer class for codeblocks

Timer can be used as a decoator, or can be used for code blocks for part of a function rather than whole function Note that Timer class does not have a callback

from eprofiler import Timer
import time

# Use as a Context Manager
with Timer(label="External API Call") as t:
    time.sleep(0.5)  # Simulate a network delay

print(f"Result: {t.stats['label']} took {t.stats['duration']:.6f}s")

# Also works as a decorator for simple timing
@Timer(label="Quick Check")
def short_task():
    pass

---

Links

• PyPI: https://pypi.org/project/eprofiler/
• GitHub: https://github.com/eyukselen/eprofiler
• Docs: https://eprofiler.readthedocs.io

---