Dozer
Dozer was originally a WSGI middleware version of Robert Brewer's Dowser CherryPy tool that displays information as collected by the gc module to assist in tracking down memory leaks. It now also has middleware for profiling and for looking at logged messages.
Tracking down memory leaks
Usage:
from dozer import Dozer # my_wsgi_app is a WSGI application wsgi_app = Dozer(my_wsgi_app)
Assuming you're serving your application on the localhost at port 5000,
you can then load up http://localhost:5000/_dozer/index to view the
gc info.
Profiling requests
Usage:
from tempfile import mkdtemp from dozer import Profiler # my_wsgi_app is a WSGI application wsgi_app = Profiler(my_wsgi_app, profile_path=mkdtemp(prefix='dozer-'))
Assuming you're serving your application on the localhost at port 5000,
you can then load up http://localhost:5000/_profiler to view the
list of recorded request profiles.
The profiles are stored in the directory specified via profile_path. If
you want to keep seeing older profiles after restarting the web app, specify a
fixed directory instead of generating a fresh empty new one with
tempfile.mkdtemp. Make sure the directory is not world-writable, as the
profiles are stored as insecure Python pickles, which allow arbitrary command
execution during load.
Here's a blog post by Marius Gedminas that contains a longer description of Dozer's profiler.
Inspecting log messages
Usage:
from dozer import Logview # my_wsgi_app is a WSGI application wsgi_app = Logview(my_wsgi_app)
Every text/html page served by your application will get some HTML and Javascript injected into the response body listing all logging messages produced by the thread that generated this response.
Here's a blog post by Marius Gedminas that contains a longer description of Dozer's logview.