Today I’m going to talk (rant) about runserver, a great inclusion of the Django software package. It’s designed to help you easily run a Django project in a local (development) environment. Well, like many people, I have Rails in my background, and I miss certain things from it. One of those things, is the shiny SQL output in your terminal while running Mongrel.
Since I don’t have enough open source projects to maintain, I decided today that I needed this. It turned out to be extraordinarily easy even. I just ripped out my code from django-debug-toolbar (SQL and Cache tracking), flipped up a generic framework, and voila, django-devserver was born.
Now let’s get down to it. django-devserver provides a simple drop-in runserver replacement. It allows you to run a command, python manage.py rundevserver, and to get some additional information. As of writing, that additional information includes real-time SQL logging (aka mass query spam in your terminal), and a summary of cache calls.
Let’s take a look at some of our sample output:
While the syntax coloring still has a lot to be desired, it’s definitely a nice usable output. The SQL module, at the moment, simply outputs the query pre-execution, and post-execution outputs the time taken. The cache module outputs some generic stats, such as total time taken, # of calls, and hits vs misses.
The beautiful thing about the solution is how extensible it came out, for example, our cache module is only a few lines of code:
from django.template.loader import render_to_string from django.shortcuts import render_to_response from django.utils import simplejson from django.core.cache import cache from devserver.modules import DevServerModule class CacheSummaryModule(DevServerModule): """ Outputs a summary of cache events once a response is ready. """ logger_name = 'cache' attrs_to_track = ['set', 'get', 'delete', 'add', 'get_many'] def process_init(self): from devserver.utils.stats import track # save our current attributes self.old = dict((k, getattr(cache, k)) for k in self.attrs_to_track) for k in self.attrs_to_track: setattr(cache, k, track(getattr(cache, k), 'cache')) def process_complete(self): from devserver.utils.stats import stats self.logger.info('total time %(time)s - %(calls)s calls; %(hits)s hits; %(misses)s misses' % dict( calls = stats.get_total_calls('cache'), time = stats.get_total_time('cache'), hits = stats.get_total_hits('cache'), misses = stats.get_total_misses_for_function('cache', cache.get) + stats.get_total_misses_for_function('cache', cache.get_many), gets = stats.get_total_calls_for_function('cache', cache.get), sets = stats.get_total_calls_for_function('cache', cache.set), get_many = stats.get_total_calls_for_function('cache', cache.get_many), deletes = stats.get_total_calls_for_function('cache', cache.delete), #cache_calls_list = [(c['time'], c['func'].__name__, c['args'], c['kwargs'], simplejson.dumps(c['stack'])) for c in stats.get_calls('cache')], )) # set our attributes back to their defaults for k, v in self.old.iteritems(): setattr(cache, k, v)
The entire thing supports all of the typical middleware processing, as well as the two additional methods: process_init() and process_complete().
I’m hoping to pull all of my usable work on the django-debug-toolbar into the devserver, but right now the two main additions I have planned are an SQLSummaryModule, and a ProcessTimeModule.
Hope you all enjoy, and get busy forking over at GitHub.
