ymyzk/wsgi_lineprof

WSGI middleware for line-by-line profiling

wsgi_lineprof

wsgi_lineprof is a WSGI middleware for line-by-line profiling.

wsgi_lineprof has the following features:

• WSGI middleware: It can be integrated with any WSGI-compatible applications and frameworks including Django, Pyramid, Flask, Bottle, and more.
• Easily pluggable: All configurations for profiling in one place. Users don't need to make any changes to their application.

wsgi_lineprof is not recommended to be used in production environment because of the overhead of profiling.

At a Glance

You can use wsgi_lineprof as a WSGI middleware of existing applications.

$ pip install wsgi_lineprof

Apply wsgi_lineprof to the existing WSGI web application:

from wsgi_lineprof.middleware import LineProfilerMiddleware
app = LineProfilerMiddleware(app)

Start the web application and access the application. wsgi_lineprof writes results to stdout every time an HTTP request is processed by default. You can see the output like this in your console:

... (snip) ...

File: ./app.py
Name: index
Total time: 1.00518 [sec]
  Line      Hits         Time  Code
===================================
     9                         @app.route('/')
    10                         def index():
    11         1      1005175      time.sleep(1)
    12         1            4      return "Hello world!!"

... (snip) ...

Also, you can check the result on your web browser by accessing the special endpoint /wsgi_lineprof/.

Please check the documentation for more details.

wsgi_lineprof's Issues

Use setuptools_scm

Support regular expression in filters

Support regular expression in FilenameFilter and NameFilter.

Python 3.8 support

Schedule: https://www.python.org/dev/peps/pep-0569/

Improve documentation on how to check results on web browser

e.g., add a screen shot

Check test coverage on GitHub Actions

Use monotonic timer

• Apple: #27
• POSIX: #28
• Windows: already using QueryPerformanceCounter()

Ad-hoc profiling

By specifying a query parameter, enable profiling and return result via HTTP.

For example, /fib?wsgi_lineprof returns a profiling result when processing a request to /fib

Idea: use external libraries as profiling backend

It may be nice if wsgi_lineprof can use other profilers as backend.

'LineProfilerMiddleware' object has no attribute 'route' : Issue with Flask

I have a simple Flask app:

from flask import Flask
from wsgi_lineprof.middleware import LineProfilerMiddleware

app = Flask('zapp')
app = LineProfilerMiddleware(app)

@app.route('/')
def index():
    return 'Hello'

if __name__ == '__main__':
    app.run(port=8000)

Running it gives the following error message:

» python app.py
Traceback (most recent call last):
  File "app.py", line 13, in <module>
    @app.route('/')
AttributeError: 'LineProfilerMiddleware' object has no attribute 'route'

---

I tried changing app = LineProfilerMiddleware(app) to app = LineProfilerMiddleware(app).app. This lets me run the app but I no longer get any profiling results.

---

Other info:

Flask==1.1.1
wsgi-lineprof==0.8.0
Python 3.7.4

Run performance benchmark (ASV) on GitHub Actions

Drop Python 2.7 support

Configure .readthedocs.yml

Customizable stats output

If a user can select which columns to show/hide in the result, it may improve the readability.

Introduce pre-commit

Consider using defaultdict

If there's no performance degradation, using defaultdict would help to simplify some code.

Improve styling of HTML results

Improve styling of profiling results shown as HTML.

Show result of profiling via HTTP

Create packages using GitHub Actions

Use pyproject.toml and PEP 517/518 for package building

Support multiple streams to export logs

Drop Python 3.6 support

Drop Python 3.6 support as it reached EoL: https://peps.python.org/pep-0494/

Build 32bit Windows wheel using GitHub Actions

Colorize the output

Improve implementation of timer

Switch to use monotonic timer for profiling.

Fix docs builds and add .readthedocs.yaml