Django for ISchoolers

From OpenHatch wiki

This is a tutorial for how to build a question-and-answer type site using Django and various other technologies, plus some lessons in general web dev concepts and best practices along the way.

Chunk one

Prerequisites

Assumptions

This tutorial assumes you know how to access the command line/terminal on your computer and how to move around (cd) and view files (ls) in it. It also assumes you're reasonably familiar with Python, the Best Programming Language Ever. (Okay, editorializing slightly.)

This tutorial also tends to be oriented towards people with some familiarity with HTML, CSS, and JS (since most of you guys are IOLabbers).

Stuff to install

TODO: Actually write my own dang installation instructions.

On Debian Wheezy: aptitude install python git python-virtualenv python-django-south python-django python-pip

Prior to starting this tutorial, please install Python, git, pip, virtualenv, Django, and South on your computer. There's instructions for all of those at http://pystar.org/setup_machine.html.

Setting up yer env

Once you've got pip and virtualenv installed, in your terminal, go into the folder you want to keep this tutorial code in and type (one line at a time):

$ virtualenv --no-site-packages django-tutorial-env

$ source django-tutorial-env/bin/activate

(django-tutorial-env)$ pip install django

(django-tutorial-env)$ pip install south

This will make and activate a virtual python environment for just this tutorial app. While this is a bit overkill if you only have one Django app at a time on your system, giving each Python app you work on its own environment is a good best practices just in case e.g. you're working on multiple apps, which use different versions of Django or other libraries, and you don't want them to conflict with each other.

To turn off the virtualenv (you can see it's active because the env name will appear in parentheses before your terminal prompt), type 'deactivate'. To reactivate the env, type the line that begins with 'source' again.

Spec'ing yer application

Your client, LulzTech, wants you to build them a Q & A website, called qandabear. ('Q and A bear', rhymes with panda bear.) For mysterious reasons, they don't want to use any of the open source Q&A website packages that are out there already; they want you to build one from scratch. Go figure.

There are whole classes on how to research and decide what to build, and how to state those requirements in a useful, clear way. (Bob's ISSD class, for one.) So we won't go into that. For our initial prototype for our client, then, our requirements are as follows:

  • On the website, users can see the questions that have been asked and when they were posted
  • Under each question, users can read the answers that have been given so far (if any) and when they were posted
  • Users can submit a new question
  • Users can submit a new answer to a question
  • Users can edit an answer to a question
  • Users can delete an answer to a question

Meeting these requirements will take us until the 'Bonus points' section. Once we get there, we can add more requirements.

Our prototype will have three pages: an index page of questions (and a form to ask a new question), a question page with the question and its answers (with a form to add a new answer), and a page to edit an answer to a question.

To support these pages, we will need two abstractions (objects, or models): Questions, and Answers.

Starting yer project

Activate your environment in your terminal

If you haven't already, run

$ source django-tutorial-env/bin/activate

in your terminal window.

Set up your git repository

If you're following this tutorial in class, in your terminal, run

$ git clone git://github.com/aldeka/django-for-ischoolers.git

. This won't download hardly anything interesting, but it will make it easy for you to sync up with the class later.

We'll be using the master branch (the default name for the main branch of code, which you're in right now) for syncing up throughout the class in case you get lost. You'll need to make your own branches for playing with the code on your computer, so that it doesn't interfere w. re-syncing later.

Think of branches in git as alternate timelines--you "branch" off of the main timeline of changes to the code into a separate timeline. You can try out a new feature or code setup in a branch, without overwriting or getting in the way of yourself or anybody else working on the code as-is. Sometimes a new idea doesn't work out, and you delete the branch or just let it sit around. Sometimes the new feature works, though, and if so you can merge it into the "real" timeline. You can make as many branches as you want, and branches of branches. You can even share branches with others (though we won't be doing that today). Git's branching system (and its non-centralized architecture generally) make it easy to try out new ideas in code without having to ask permission of everyone else on the project first.

To make your first branch, be sure you've changed directories to the django-for-ischoolers directory and then type

$ git branch my-chunk-1

to make your new branch, and

$ git checkout my-chunk-1

to switch into that branch. Type

$ git branch

to confirm that you're in your new branch (and that you only have two branches so far). If you're not following this tutorial in class, just run

$ git init

.

Create your Django project

Let's create your first Django project, called "mysite".

$ django-admin.py startproject mysite

You'll see we made a folder called "mysite", with some files in it. Let's check them out!

$ cd mysite
$ ls
    __init__.py
    manage.py
    settings.py
    urls.py

These files are:

  • __init__.py: An empty file that tells Python that this directory should be considered a Python module. Because of the __init__.py file, you can use import to import myproject.
  • manage.py: A command-line utility that lets you interact with this Django project in various ways. You can read all the details about manage.py in django-admin.py and manage.py.
  • settings.py: Settings/configuration for this Django project.
  • urls.py: The URL declarations for this Django project; a “table of contents” of your Django-powered site.

Start your local development server

  • Django, handily, comes with some local server software (warning: never use Django's built-in dev server for actually hosting a site in production! It's not meant for that.). This local dev server makes it easy to see the changes you make to your web app as you build it. Let's make sure our project works by verifying that the dev server will start.
    • Run the command:
python manage.py runserver
    • Review the output in your terminal. It should look similar to:
Validating models...
0 errors found.

Django version 1.2, using settings 'myproject.settings'
Development server is running at http://127.0.0.1:8000/
Quit the server with CONTROL-C.
  • Now that the server’s running, visit http://127.0.0.1:8000/ with your Web browser. You’ll see a “Welcome to Django” page, in pleasant, light-blue pastel. It worked!
  • Observe the logging that happens in the terminal where your server is running:
[24/Mar/2011 11:50:18] "GET / HTTP/1.1" 200 2057

which has the format:

DATE  METHOD URL  PROTOCOL  RESPONSE_CODE  CONTENTSIZE
  • Exit the server by returning to the terminal instance where the development server is running and pressing CONTROL-C on your keyboard.

Using version control

Before we do anything else, let’s commit our work. We'll be doing this with a piece of version control software called git.

What's this version control nonsense? Well, you probably know the value of saving your work. When writing an important term paper, you want to "Save early, save often" to make sure you don't lose your work if your computer crashes or something else bad happens! But most paper-writing software isn't very good at saving your history--if you decide to delete a section of your paper, save it, and then later decide that you want that section back, most document writing software doesn't have any way of helping you do that. There's only one saved version of the paper (unless you save multiple copies over time, yourself, by hand).

Programmers often have to go back into the history of a program to change things--whether it's to diagnose a bug or redesign how a feature works. Programmers also have to have an easy way of being able to edit the same application at the same time, and to share their edits with each other. Thus, besides saving their work the normal way, programmers commit their code using version control software.

Each commit is a checkpoint in time, containing the diff--the "difference", all of the changes you made to your code base -- between that commit and the commit before it. Different branches in git share the commits made prior to the branching-off point, then each have their own commit history.

  • To make a commit, first type
    git status
    
    into your terminal. This will let you know what changes git has noticed in your code and which of those changes, if any, are staged and ready to be committed.
$ git status
# On branch my-chunk-1
#
# Initial commit
#
# Untracked files:
#   (use "git add <file>..." to include in what will be committed)
#
#   __init__.py
#   manage.py
#   settings.py
#   urls.py
nothing added to commit but untracked files present (use "git add" to track)

"Untracked files" means that git has noticed some new files inside its folder, but you haven't told git explicitly that you want it to "listen" for, and track, changes in those files.

  • Add one file:
    git add manage.py
    
    . What does git status say now?
  • Add all the files to the repo, in the local directory:
$ git add *.py   # all .py files, using a wildcard match.

What does git status say now?

  • git commit to commit those files:
# -m -> what is the 'message' for the commit
git commit -m "Initial commit of Django project from the IOLab Django workshop"
  • Look at your changes with
    git log
    
    to see your history. Is your commit message there?

Huzzah! We committed our changes so far. Now let's make some more changes!

Set up settings

Now that we have a the scaffolding for our project in place, we can get to work! First, it needs to be configured.

Add yourself as an admin!

  • Open settings.py in your editor. settings.py is a Python script that only contains variable definitions. Django looks at the values of these variables when it runs your web app.
  • Find ADMINS and replace Your Name and [email protected] with your name and your email address.

Remove the pound mark from the front of the line to uncomment it out.

  • git add and commit it:
# even though git knows to "listen" to settings.py now, it still won't add it to the 'staging area' for your commit unless you tell it to explicitly with git add
git add settings.py
git commit -m "made myself an admin"

Set up the Database

Keep looking at settings.py: The DATABASES variable is a dictionary (note the ‘{}’ characters) with one key: default.

DATABASES = {
'default': {
'ENGINE': 'django.db.backends', # Add 'postgresql_psycopg2', 'postgresql', 'mysql', 'sqlite3' or 'oracle'.
'NAME': '',                      # Or path to database file if using sqlite3.
'USER': '',                      # Not used with sqlite3.
'PASSWORD': '',                  # Not used with sqlite3.
'HOST': '',                      # Set to empty string for 127.0.0.1. Not used with sqlite3.
'PORT': '',                      # Set to empty string for default. Not used with sqlite3.
}
}

Notice that the value of default is itself another dictionary with information about the site’s default database.

  • Set your app to use a sqlite database, in the ENGINE attribute. Sqlite is great for development because is stores its data in one normal file on your system and therefore is really simple to move around with your app. It's not sturdy enough to use for a website in production, though.
'ENGINE': 'django.db.backends.sqlite3',
  • Set your app to use a file called 'database.db' to store information for this project.
'NAME': 'database.db',

Does database.db exist right now? (No, but that's okay. It'll get created automatically when it's needed.)

Notice the INSTALLED_APPS setting towards the bottom of the settings.py. That variable (a tuple... note the ‘()’ symbols) holds the names of all Django applications that are activated in this Django instance. Apps can be used in multiple projects, and you can package and distribute them for use by others in their projects. What do you think these various apps do? Why does it make sense for them to come in a standard configuration?

  • Add South to our list of installed apps. (We'll need it later.)
INSTALLED_APPS = (
'django.contrib.auth',
'django.contrib.contenttypes',
'django.contrib.sessions',
'django.contrib.sites',
'django.contrib.messages',
# Uncomment the next line to enable the admin:
# 'django.contrib.admin',
# Uncomment the next line to enable admin documentation:
# 'django.contrib.admindocs',
'south',
)
  • Each of these applications makes use of at least one database table, so we need to create the tables in the database before we can use them. To do that, run the following command:
python manage.py syncdb

The syncdb command looks at the INSTALLED_APPS setting and creates any necessary database tables according to the database settings in your settings.py file. You’ll see a message for each database table it creates.

  • When prompted, you’ll get a prompt asking you if you’d like to create a superuser account for the authentication system. Say yes! Use ‘super’ as your password.

Does database.db exist right now? (Yes! Calling syncdb made Django realize it needed a sqlite database file, so it made one for us.)

  • Save and commit your work


Projects v. apps -- what's the difference?

We’ve talked a little about Django apps and projects. You might be wondering what that nonsense is all about. Here are the things to know:

  • An app is component of a website that does something. South is a Django app. So is qandabear (or will be, anyway). An app is:
    • single purpose - login, passwords, polls, forum, etc.
    • orthogonal to / independent of other apps - qandabear shouldn’t have to know the inside details of authentication, for example.
  • A project corresponds to a ‘website’: it contains a settings.py file, and it may have corresponding databases or other data stores that the apps interact with.

Most projects--those for more complex websites--should use multiple apps, one for each core piece of functionality.

Ideally, you should reuse existing free, open source Django apps that other people have written for things like user accounts, image galleries, and other common use cases. [Django Packages http://www.djangopackages.com] is a good resource for finding such apps. Django adheres to the principle of DRY -- Don't Repeat Yourself -- and in your real world coding adventures, you should strive to avoid repeating other, wiser developers as well. :)

Starting yer app

In this tutorial, we’ll create our Q&A app in the myproject directory for simplicity. In the future, when you decide that the world needs to be able to use your Q&A app and plug it into their own projects, and after you determine that your app plays nicely with other apps, you can publish that directory separately!

  • Open your terminal and navigate into the mysite folder
  • Make scaffolding for the app:
$ python manage.py startapp qandabear

That’ll create a directory qandabear to house our Q & A application.

  • Verify what is new.
$ git status
# should show 'qandabear/' in 'untracked'
  • Examine the layout of qandabear (we will do more of this in following sections).
# remember not to type the '$', it just means the prompt'.
$ ls qandabear
qandabear/
__init__.py
models.py
tests.py
views.py
  • Add and commit
    qandabear/*py
    
    .
  • Install the Q&A app into the project. Edit the settings.py file again, and change the INSTALLED_APPS setting to include the string ‘qandabear’ as the last entry.
  • Save and commit the settings.py file.

Chunk 2

At the end of each 'chunk', I'm going to push our progress so far to the tutorial git repository. If you're doing this tutorial in class, please switch back to the master branch, then pull my changes.

$ git checkout master
$ git pull origin master

This will make it so that everyone starts out each chunk with the same working code, so even if you get stuck or confused in one session, you can still keep up. (Since I don't have any TAs to help debug or answer questions, and we don't have that much time.)

Remember that *you* shouldn't be putting your modifications inside the master branch, though! (Otherwise, you'd have to overwrite them at the next chunk checkpoint, and that would be sad.) Instead, make a new branch for chunk 2 based off the master code you just downloaded, and let's get going!

$ git branch my-chunk-2
$ git checkout my-chunk-2


M-V-C separation

Model: how the data is organized. Controller: how the logic of the app works. View: How the app looks to the user.

(chart explaining model-view-controller separation in Django)

urls.py

The first step to writing your application is to design your URL structure. You do this by creating a Python module, called a URLconf. URLconfs are how Django associates a given URL with given Python code.

How URLconfs work (or, the joy of regular expressions!)

When a user requests a Django-powered page, the system looks at the ROOT_URLCONF setting, which contains a string in Python dotted syntax.

Django loads that module and looks for a module-level variable called urlpatterns, which is a sequence of Python tuples in the following format: (regular expression, Python callback function [, optional dictionary])

Django starts at the first regular expression and makes its way down the list, comparing the requested URL against each regular expression until it finds one that matches.

You might ask, “What’s a regular expression?” Regular expressions, or "regexes", are patterns for matching text. In this case, we’re matching the URLs people go to, and using regular expressions to match whole ‘groups’ of them at once. (If you’d like to learn more about regular expressions, read the [Dive into Python guide to regular expressions http://diveintopython.org/regular_expressions/index.html] sometime. Or you can look at this [xkcd http://xkcd.com/208/].)

In addition to matching text, regular expressions can capture text. Capturing means to remember that part of the string, for later use. Regexes use parentheses () to wrap the parts they’re capturing.

For Django, when a regular expression matches the URL that a web surfer requests, Django extracts the captured values (if any) and passes them to a function of your choosing. This is the role of the callback function above. When a regular expression matches the url, Django calls the associated callback function with any captured parts as parameters. This will much clearer after the next section.

Adding URLs to our urls.py

When we ran

django-admin.py startproject mysite

to create the project, Django created a default URLconf file called `urls.py`.

  • Write our URL mapping. Edit the file mysite/urls.py so it looks like this:
urlpatterns = patterns('',
(r'^questions/$', 'qandabear.views.index'),
(r'^questions/(\d+)/$', 'qandabear.views.question'),
(r'^questions/(\d+)/answers/(\d+)/edit/$', 'qandabear.views.edit_answer'),
# Examples:
# url(r'^$', 'myproject.views.home', name='home'),
# url(r'^myproject/', include('myproject.foo.urls')),

# Uncomment the admin/doc line below to enable admin documentation:
# url(r'^admin/doc/', include('django.contrib.admindocs.urls')),

# Uncomment the next line to enable the admin:
# url(r'^admin/', include(admin.site.urls)),
)

Suppose a visitor goes to http:// 127.0.0.1:8000/questions/23/.

  1. which regex pattern is tripped?
  2. what function is then called?
  3. what arguments is that function called with?
  • Save urls.py.
  • Start the dev server and try that url out! What happens?
  • Save and commit.

Review: When somebody requests a page from your Web site – say, “/questions/23/”, Django will load the urls.py Python module, because it’s pointed to by the ROOT_URLCONF setting. It finds the variable named urlpatterns and traverses the regular expressions in order. When it finds a regular expression that matches – r'^questions/(\d+)/$' – it loads the function question() from qandabear/views.py. Finally, it calls that module’s detail() function like so:

detail(request=<HttpRequest object>, '23')

The ‘23’ part comes from (\d+). Using parentheses around a pattern “captures” the text matched by that pattern and sends it as an argument to the view function; the \d+ is a regular expression to match a sequence of digits (i.e., a number).

The idea that a URL doesn’t have to map onto a file or a folder, or some other sort of static resource, is quite powerful. The URL is just a way of giving instructions to some server, somewhere.

(Rant: In Django, as in most modern frameworks, you have total control over the way your URLs look. People on the web won’t see cruft like .py or .php or even .html at the end of your URLs. There is no excuse for that kind of stuff in the modern era! (Though, putting .asp on the end of all your Django URLs, while pointless, is kind of hilarious and super easy.))

views.py

  • Start the development server:
    python manage.py runserver
    
  • Fetch “http://127.0.0.1:8000/questions/” in your browser. You should get a pleasantly-colored error page with the following message:
ViewDoesNotExist at /questions/

Tried index in module qandabear.views. Error was: 'module'
object has no attribute 'index'

Recall this line:

(r'^questions/$', 'qandabear.views.index')

.

  • Explore this using your django-shell:
    python manage.py shell
    
>>> import qandabear         # imports fine!
>>> import qandabear.views   # imports fine also!
>>> dir(qandabear.views) # what is in there!
>>> 'index' in dir(qandabear.views)
False
>>> import inspect
>>> inspect.getsourcefile(qandabear.views)
# something like
'/Users/adalovelace/gits/mysite/questions/views.py'

So, a mystery? Where is the view!? It’s nowhere! The URL parsing is going fine, but there is no one listening at the other end of the phone! This ViewDoesNotExist error happened because you haven’t written a function index() in the module questions/views.py.

Well, I guess we should do that!

  • Write some views. Open qandabear/views.py and put the following Python code in it:
from django.http import HttpResponse

def index(request):
    return HttpResponse("Hello, world. You're at the questions index.")

This is a very simple view.

  • Add a few more views by adding to the views.py file. These views are slightly different, because they take an argument or two (which, remember, is passed in from whatever was captured by the regular expression in the URLconf):
def question(request, question_id):
    return HttpResponse("You're looking at question %s." % (question_id,))

def edit_answer(request, question_id, answer_id):
    return HttpResponse("You're looking at the edit page for answer %s." % (answer_id,))
  • Save views.py.
  • Navigate to http://127.0.0.1:8000/questions/34/. It’ll run the question() method and display whatever ID you provide in the URL.
  • Add a little html to the ‘question’ view. Wrap the question_id in tags and verify that the view is indeed bold!
  • Add and commit your code. Remember to write a good commit message that mentioned what changed (in English) and more details below.

Databases and the ORM (or: finally something that HTML/CSS/JS couldn't do for you on its own)

Okay, being able to design your URLs however you want, without them having to correspond to your actual file structure, is pretty neat. But besides that, we haven't done anything yet that you couldn't do with just HTML, CSS, and JavaScript (or just HTML, as it happens!).

Time to fix that.

ORM = Object Relational Mapper/Mapping. Translating Python classes -- real object-oriented programming stuff -- into representations in a database on your server for you.

(models.py --> ORM --> DB chart)

models.py

Let's make it so we can store some real live data in our application!

In our simple Q&A app, we’ll create two models: Questions and Answers. As per our spec from the customer:

  • A question has:
    • The text of the question
    • A publication date.
  • An answer has:
    • The text of the answer
    • A publication date.

Each Answer is associated with a Question and each Question has associated Answers. We will respesent these concepts with python classes derived from django.db.models.

Edit the qandabear/models.py file so it looks like this:

from django.db import models

class Question(models.Model):
    text = models.CharField(max_length=200)
    pub_date = models.DateTimeField(auto_now_add=True)

class Answer(models.Model):
    question = models.ForeignKey(Question)
    answer = models.TextField()
    pub_date = models.DateTimeField(auto_now_add=True)
  • Save the models.py file.

All models in Django code are represented by a class that subclasses django.db.models.Model. Each model has a number of class variables, each of which represents a database field in the model. (cf: http://docs.djangoproject.com/en/dev/topics/db/models/)

Each field is represented by an instance of a Field class – e.g., CharField for character fields and DateTimeField for datetimes. This tells Django what type of data each field holds.

The name of each Field instance (e.g. answer or pub_date) is the field’s name, in machine-friendly format. You’ll use this value in your Python code, and your database will use it as the column name.

Some Field classes have required elements. CharField, for example, requires that you give it a max_length. That’s used not only in the database schema, but in validation as well.

Most Field classes also have optional elements that you can set. For example, setting auto_now_add to True on a datetime field means that by default, that field will be set to the date and time that the object is first created.

Finally, note a relationship is defined, using ForeignKey. That tells Django each Answer is related to a single Question. Django supports all the common database relationships: many-to-ones, many-to-manys and one-to-ones.

Database migrations and South, part one

When you create your models, you might not always know exactly what fields your models will need in advance. Maybe someday your qandabear app will have users, and you’ll want to keep track of the author of each question! Then you would want to add another field to the model to store that information. Maybe someday you'll decide that tracking all the pub_dates is silly, and want to delete that field.

Unfortunately, Django (and most database-using software) can’t figure out how to handle model changes very well on its own. Fortunately, a Django app called `south` can handle these changes–called ‘migrations’–for us.

Now that we’ve made our first version of our models file, let’s set up our qandabear app to work with South so that we can make migrations with it in the future!

  • On the command line, write:
$ python manage.py schemamigration qandabear --initial

As you can see, that’s created a migrations directory for us, and made a new migration inside it.

  • All we need to do now is apply our new migration:
$ python manage.py migrate qandabear

Great! Now our database file knows about qandabear and its new models, and if we need to change our models, South is set up to handle those changes. We’ll come back to South later.

IMPORTANT: You can't migrate an app if it's already been synced in the database using

python manage.py syncdb

. But you do need to run syncdb at least once before you use south (since south itself uses syncdb to give itself space in your database). That's why it's super important that when you run syncdb, south should be listed under INSTALLED_APPS, but none of your own apps should be, and after you add your app to INSTALLED_APPS, you must not run syncdb again until after you've already set up migrations with that app.

  • Add and commit all your work, including the migrations folder that South generated for you.

Chunk 3

You know the drill. Switch back to the master branch, pull down all the new code, and create a new branch called my-chunk-3. (Or something else, if you like. Your branch names are up to you!)

Let's add some data (via the Django shell)!

Now, let’s hop into the interactive Python shell and play around with the free API ("Application programming interface" -- APIs aren't just data doodads that hip web startups provide for you, they're an important concept in software architecture.) that Django gives you. To invoke the Python shell, use this command:

python manage.py shell

We’re using this instead of simply typing “python”, because manage.py sets up the project’s environment for you. “Setting up the environment” involves two things:

  1. Making sure qandabear (and any other apps you might have) are on the right path to be imported.
  2. Setting the DJANGO_SETTINGS_MODULE environment variable, which gives Django the path to your settings.py file.

Once you’re in the shell, explore the database API:

  • Import the model classes we just wrote:
>>> from qandabear.models import Question, Answer
  • List all the current Questions:
>>> Question.objects.all()
[]

How many questions is this?

Zen koan: Can there be an Answer for a Question that doesn’t yet exist?

Add our first Question

>>> q = Question(text="What is the Weirdest Cookbook Ever?")
# We could specify the value for pub_date as well here, but we don't have to since when we wrote its model definition, we specified that pub_date has a default value (of when the question is created)
  • Try getting the ID number of our new question 'q' by typing
    q.id
    
    . What happens?
  • Save the Question instance into the database. You have to call save() explicitly.
>>> q.save()
  • Get the id of the Question instance. Because it’s been saved, it has an ID in the database now!
>>> q.id
1
  • Access the database columns (Fields, in Django parlance) as Python attributes:
>>> q.text
"What is the Weirdest Cookbook Ever?"
>>> q.pub_date
datetime.datetime(2011, 12, 1, 3, 3, 55, 841929)
  • Send the Question back in time:
# Change values by changing the attributes, then calling save().
>>> import datetime
>>> q.pub_date = datetime.datetime(2011, 4, 1, 0, 0)
>>> q.save()
>>> q.pub_date
datetime.datetime(2007, 4, 1, 0, 0)
  • Ask Django to show a list of all the Question objects available:
>>> Question.objects.all()
[<Question: Question object>]

Fix The Hideous Default Representation

Wait a minute! <Question: Question object> is an utterly unhelpful, truly wretched, beyond contemptable representation of this object. Let’s fix that by editing the Question model. Use your text editor to open the qandabear/models.py file and adding a __unicode__() method to both Question and Answer:

class Question(models.Model):
    # ...
    def __unicode__(self):
        return self.text

class Answer(models.Model):
    # ...
    def __unicode__(self):
        # return the answer text, or if the text is longer than 37 characters, return a summary with an ellipsis
        summary = self.text[:37]
        if len(self.text) > 37:
            summary = summary + '...'
        return summary

It’s important to add __unicode__() methods to your models, not only for your own sanity when dealing with the interactive prompt, but also because if you use Django's automatically-generated admin, these representations are used there too.

Adding custom methods to models

Enough of these normal python methods! Let's build in some useful functionality to our models in qandabear/models.py.

import datetime
# ...
class Question(models.Model):
    # ...
    def was_published_today(self):
        return self.pub_date.date() == datetime.date.today()

Note the addition of

import datetime

to reference Python’s standard datetime module. This allows us to use the datetime library module in models.py by calling it with datetime.

  • Save these changes to the models.py file.

Test that all those changes did something

Start a new Python interactive shell by running python manage.py shell:

>>> from qandabear.models import Question, Answer
Verify our __unicode__() addition worked:
>>> Question.objects.all()
[<Question: What is the Weirdest Cookbook Ever?>]
Search your database using the filter method on the objects attribute of Question.
>>> questions = Question.objects.filter(text="What is the Weirdest Cookbook Ever?")
>>> questions
[<Poll: What is the Weirdest Cookbook Ever?>]
>>> questions[0].id  # remember python lists start with element 0.
1

If you try to search for a question that does not exist, filter will give you the empty list. The get method will always return one hit, or raise an exception.

>>> Question.objects.filter(text="Who framed Roger Rabbit?")
[]
>>> Question.objects.get(id=1)
<Question: What is the Weirdest Cookbook Ever?>
>>> Question.objects.get(id=2)
Traceback (most recent call last):
    ...
DoesNotExist: Question matching query does not exist.

Add answers

Observe: there is a Question in the database, but it has no Answers! Let's fix that.

>>> q = Question.objects.get(id=1)
>>> q.answer_set.all()
[]
Create three answers:
>>> q.answer_set.create(text='To Serve Man')
<Answer: To Serve Man>
>>> q.answer_set.create(text='The Original Road Kill Cookbook')
<Answer: The Original Road Kill Cookbook>
>>> a = q.answer_set.create(text='Mini-Mart A La Carte')
>>> a
<Answer: Mini-Mart A La Carte>

Go in reverse! Find the question a particular answer belongs to:

>>> a.question
<Question: What is the Weirdest Cookbook Ever?>

Because a Question can have more than one Answer, Django creates the answer_set attribute on each Question. You can use that to look at the list of available Answer, or to create them.

>>> q.answer_set.all()
[<Answer: To Serve Man>, <Answer: The Original Road Kill Cookbook>, <Answer: Mini-Mart A La Carte>]
>>> q.answer_set.count()
3

Can one be a Answer for a Question that doesn’t yet exist?:

>>> koan = Answer(text="Is this even an answer")
>>> koan.question_id
>>> koan.question

Database migrations and South, part two

Changing your models

Oh no! Our client, LulzTech, has decided to change the spec for this prototype. They want to add a votes field to each Answer, to track how many upvotes it has. Which means we’re going to have to change our models.

  • Open qandabear/models.py and edit the Answer class:
class Answer(models.Model):
    question = models.ForeignKey(Question)
    text = models.TextField()
    votes = models.IntegerField(default=0)
    pub_date = models.DateTimeField(auto_now_add=True)

The default parameter lets us set a default value for this field if an answer's vote count isn't explicitly specified. Most new answers are going to have zero votes, so we set our default to 0.

  • Make a migration so we can let the database know that we added Answer.votes:
$ python manage.py schemamigration qandabear --auto
  • Apply the migration.
$ python manage.py migrate qandabear
Running migrations for qandabear:
 - Migrating forwards to 0002_auto__add_field_answer_votes.
 > qandabear:0002_auto__add_field_answer_votes
 - Loading initial data for qandabear.
No fixtures found.

Success!

Save and commit

You know the drill!


Views with actual data

In Django, each view is responsible for doing one of two things: returning an HttpResponse object containing the content for the requested page, or raise-ing an exception such as Http404.

Your view can read records from a database, or not. It can use a template system such as Django’s – or not. It can generate a PDF file, output XML, create a ZIP file on the fly, anything you want, using whatever Python libraries you want. All Django wants is that at the end, it gets an HttpResponse or an exception out of your view function.

Most of the Django views in the world use Django’s own database API, which was touched on in the discuss of models, to write or show dynamic data to the user in some fashion. Right now, our views are very simple, and don't use the data in our database at all. Let's fix that.

Write a better index() view

To match the spec, our index page should displays the latest 5 questions in the system, separated by commas, according to publication date.

  • Edit views.py:
from qandabear.models import Question
from django.http import HttpResponse

def index(request):
    latest_qs = Question.objects.all().order_by('-pub_date')[:5]
    output = ', '.join([q.text for q in latest_qs])
    return HttpResponse(output)

Restart the dev server, and navigate to http://127.0.0.1:8000/questions/. You should see the text of the last 5 polls (or fewer than five, if you haven't made that many yet).

There’s a problem here, though: The page’s design is hard-coded in the view. If you want to change the way the page looks, you’ll have to edit this view code. That's silly.

  • Use Django’s template system to separate the design from Python:
from django.shortcuts import render_to_response
from qandabear.models import Question

def index(request):
    latest_qs = Question.objects.all().order_by('-pub_date')[:5]
    context = {'latest_qs': latest_qs}
    return render_to_response('qandabear/index.html', context)

To recap what this does:

  • Creates a variable called latest_poll_list. Django queries the database for all Poll objects, ordered by pub_date with most recent first, and uses slicing to get the first five.
  • Creates a variable called context that is a dictionary with one key.
  • Evaluates the render_to_response function with two arguments, and returns whatever that returns.

render_to_response loads the template called polls/index.html and passes it a value as context. The context is a dictionary mapping template variable names to Python objects.

If you can read this this view function without being overwhelmed, then you understand the basics of Django views. Now is a good time to reflect and make sure you do:

  1. What would you have to change to get 10 polls?
  2. What if you wanted the first 10 by name instead of by publication date?
TemplateDoesNotExist at /questions/
polls/index.html

Ah. There’s no template yet. Let’s make one!

  • Make a qandabear/templates/qandabear directory where templates will live, right alongside the views.py for the qandabear app. This is what I would do:
mkdir -p qandabear/templates/qandabear

Edit qandabear/templates/qandabear/index.html to contain:

{% if latest_qs %}
    <ul>
    {% for question in latest_qs %}
        <li><a href="/questions/{{ question.id }}/">{{ question.text }}</a></li>
    {% endfor %}
    </ul>
{% else %}
    <p>No questions are available.</p>
{% endif %}
  • Edit TEMPLATE_DIRS in settings.py to have the full path to the templates folder inside your new app. On my computer, this looks like:
 TEMPLATE_DIRS = (
    # Put strings here, like "/home/html/django_templates" or "C:/www/django/templates".
    # Always use forward slashes, even on Windows.
    # Don't forget to use absolute paths, not relative paths.
    '/karen/Code/django-for-ischoolers/mysite/qandabear/templates',
)
  • Reload http://127.0.0.1:8000/questions/ . You should see a bulleted-list containing up to five of your questions. There should also be link pointing to the question's detail page.
  • Save and commit.

Fix the detail view and handle user errors using a 404

Now, let’s tackle the question detail view – the page that displays one single question.

  • Edit the views.py file. This view uses Python exceptions:
from django.http import Http404
# ...
def detail(request, question_id):
    try:
        q = Question.objects.get(id=question_id)
    except Question.DoesNotExist:
        raise Http404
    return render_to_response('qandabear/detail.html', {'question': q})

Notice that view raises the Http404 exception if a poll with the requested ID doesn’t exist.

  • Create qandabear/templates/qandabear/detail.html with:
{{ question }}
  • Try visiting http://127.0.0.1:8000/questions/9000/. Oops, we forgot to make a template for our lovely 404 error! Create qandabear/templates/404.html (the qandabear template root dir) as:
<p>You have a 404.  Go back and try again.</p>
  • What? It says DEBUG has to be False? All right, set it (in settings.py), and try again! (note: Chrome ‘eats’ the 404. Safari will show our created page.)
  • Change DEBUG back to True
  • Save and commit.

Add more detail to the details!

Great, we've got a question detail page. But it isn't really that useful. We don't just want to see the question text -- we want to see the question's answers, too!

  • Edit the qandabear/detail.html template to add a question variable. question points the particular instance of the Question class that our view sent to the template.
<h1>{{ question.text }}</h1>
<ul>
{% for answer in question.answer_set.all %}
    <li>{{ answer.text }} -- <em>Answered at {{ answer.pub_date }} -- {{ answer.votes }} votes</em></li>
{% endfor %}
</ul>

The django.template system uses dot-lookup syntax to access variable attributes. Django’s template language is a bit looser than standard python. In pure Python, the . (dot) only lets you get attributes from objects, and we would need to use [] to access parts of list, tuple or dict objects. In this example, we are just doing attribute lookup, but in general if you’re not sure how to get data out of an object in django.templates, try dot.

Method-calling happens in the {% for %} loop: question.answer_set.all is interpreted as the Python code question.answer_set.all(), which returns a sequence of Answer objects and is suitable for use in the {% for %} template tag.

  • Save and commit your changes.

Chunk 4

$ git checkout master
$ git pull origin master
$ git branch my-chunk-4
$ git checkout my-chunk-4

Django templates 101

Okay, we've done a bit with Django's templating language already when we created those templates in the previous chunk. Now let's look at templating a bit more in-depth.

Tags and variables

Django comes with its own templating language. A template is (mostly) valid HTML, with some special templating tags thrown in to handle dynamic content. Some of those tags you've already seen, and resemble basic Python operators -- e.g. if statements and for loops. Programmatic statements such as these are always found inside block tags -- {% %} -- which separate them from the surrounding HTML.

<div class="question {% if question.was_published_today %}new-question-highlight{% endif %}"></div>
{% comment %}This tag checks to see if the given question was published today (using the question model's class function). If it was, the div gets an additional class applied to it, 'new-question-highlight'.{% endcomment %}

Additionally, you have variables, which are not programming statements but rather print data that your view fed to the template so that it appears in the HTML code. Variables can have special filters applied to them, which allow you to e.g. lowercase a given piece of text, but that's the extent of their power. These variables are found inside variable tags -- {{ }}.

Example:

<h1>{{ question.text|upper }}</h1> 
{% comment %}This makes the text of the question appear inside the <h1> tag, and filters it so that it appears in all uppercase.{% endcomment %}

A reference to Django template filters and tags can be found here: https://docs.djangoproject.com/en/dev/ref/templates/builtins/

Note: One important thing to keep in mind is that the Django templating language is NOT as powerful as Python or Django proper! There are many functions that Python has that the templating language does not support. For instance, you can't add or subtract within the templating language. Similarly, there are limits to what data the templating language can access in the database -- it has to be passed that data by the view. This is on purpose. You should NOT be doing heavy-duty computing or data slicing and dicing in your templates! That logic should be in your views.py file, or a helper script of some kind if necessary. Remember M-V-C separation?

Template inheritance

When you make a web app with multiple pages, there are going to be a lot of elements shared by each of the pages. For example, each of the pages should have the same header and navigation UI, and the same footer. However, right now, we have two templates -- index.html and detail.html -- and each of them are entirely separate from the other. If we wanted to add a header to index.html, we'd have to add it separately to detail.html -- and every other template we have. If we ever decided to change our header, we'd be copying and pasting the whole day long. That's silly! And it violates Django's principle of DRY: Don't Repeat Yourself.

Fortunately, Django's template language helps us avoid this problem, by supporting template inheritance.

  • In qandabear/templates/, create a file named base.html. Put this code in the file:
<!DOCTYPE html> 
<html>
<head>
<title>QandaBear</title>
</head>

<body>
<div id="header">
<h1>QandaBear</h1>
<span class="subheader">All your questions answered</span>
</div>

<div id="content">
{% block content %}
<h2>Welcome to QandaBear!</h2>
{% endblock content %}
</div>

<div id="footer">
This web app is powered by Django!
</div>
</body>
</html>

What did we just do here? We made an HTML skeleton for our app including a header area, a content area, and a footer area. We also used the Django 'block' tag to make a block inside our content div, conveniently named 'content'.

  • Now that we've got this skeleton, let's modify index.html to inherit from this template.
{% extends "base.html" %}

{% block content %}
{% if latest_qs %}
    <ul>
    {% for question in latest_qs %}
        <li><a href="/questions/{{ question.id }}/">{{ question.text }}</a></li>
    {% endfor %}
    </ul>
{% else %}
    <p>No questions are available.</p>
{% endif %}
{% endblock content %}

What did we just do? We added an 'extends' tag to the top of our template, which tells Django that this template inherits from another one. We also put our code inside of a set of block tags, a block named 'content'. This tells Django, "Overwrite whatever the base template had inside this block and put this code there instead!"

Save your work and check out http://127.0.0.1:8000/questions/! Now your header and footer appear on the page, along with the list of questions inside your content div.

(We should modify detail.html to inherit from our base template as well. That's left as an exercise to the reader.)

You can have multiple layers of inheritance -- for instance, if you had some pages with a single-column layout and other pages with a two-column layout, you could have a base template, which is inherited by single-column.html and two-column.html, and each of those could be inherited by other templates. Each template can only have one immediate parent, however -- you couldn't tell a template to inherit both from single-column.html *and* two-column.html.

By default, calling a block inside a child template will overwrite any content inside that block in the parent template. Sometimes you don't want to do this, though. For instance, if you had all of your javascript code and import statements inside a block, and wanted to some additional javascript that applied to just one template, you wouldn't want to overwrite everything. Fortunately, there's a tag for that, too!

{% block content %}
{{ block.super }}
{% if latest_qs %}
    <ul>
    {% for question in latest_qs %}
        <li><a href="/questions/{{ question.id }}/">{{ question.text }}</a></li>
    {% endfor %}
    </ul>
{% else %}
    <p>No questions are available.</p>
{% endif %}
{% endblock content %}
  • Commit your work.

Includes

There's another way that Django enables you to reuse elements: the 'include' tag. Generally speaking, you should use template inheritance for the most common kinds of reuse. Using includes for everything would get pretty messy! However, sometimes you have a widget that you want to use on multiple pages, and there isn't a clean inheritance relationship between them.

  • In templates/qandabear, create a file called 'subscriber_widget.html'.
<div class="widget subscriber-widget">
This web app is totally awesome, don't you agree? <a href="#">Subscribe to our newsletter!</a>
</div>
  • Edit index.html's content block:
{% block content %}
{{ block.super }}
{% if latest_qs %}
    <ul>
    {% for question in latest_qs %}
        <li><a href="/questions/{{ question.id }}/">{{ question.text }}</a></li>
    {% endfor %}
    </ul>
{% else %}
    <p>No questions are available.</p>
{% endif %}
{% include "qandabear/subscriber_widget.html" %}
{% endblock content %}
  • Commit your work.

Oh, CRUD!

So, we've got a basic web app, with some templates and dynamic data. But right now the only way we have to edit that data -- add, edit, or delete questions or answers -- is from the command line. Our web app's users can't do it at all! Let's fix that.

A common acronym for these sorts of basic data operations is CRUD -- Create, Read, Update, and Delete. We've already got 'Read' covered. Time to do the rest!

Create the form

Recall that the prototype spec allows users to add answers to questions. We are going to use a form for that functionality. As an alternative, we could have used AJAX Requests or some other mechanism.

  • Update our question detail template (qandabear/detail.html) to contain an HTML <form> element:
{% extends "base.html" %}
 
{% block content %}
<h1>{{ question.text }}</h1>

<ul>
{% for answer in question.answer_set.all %}
    <li>{{ answer.text }} -- <em>Answered at {{ answer.pub_date }} -- {{ answer.votes }} votes</em></li>
{% endfor %}
</ul>

<h2>Add new answer<h2>
<form action="/questions/{{ question.id }}/answer/" method="post">
{% csrf_token %}
<textarea name="answer"></textarea>
<input type="submit" value="Enter" />
</form>
{% endblock content %}

There is a lot going on there. A quick rundown:

  • The above template displays a textarea below the list of previously-submitted answers, with an 'Enter' button.
  • We set the form’s action to '/questions/Template:Question.id/answer/ method="post"'. Normal web pages are requested using GET, but the standards for HTTP indicate that if you are changing data on the server, you must use the POST method. (Whenever you create a form that alters data server-side, use method="post". This tip isn’t specific to Django; it’s just good Web development practice!)
  • Since we’re creating a POST form (which can have the effect of modifying data), we need to worry about Cross Site Request Forgeries. Thankfully, you don’t have to worry too hard, because Django comes with a very easy-to-use system for protecting against it. In short, all POST forms that are targeted at internal URLs should have the {% csrf_token %} template tag inside the form.
  • The {% csrf_token %} tag requires information from the request object, which is not normally accessible from within the template context. To fix this, a small adjustment needs to be made to the detail view in the views.py file.
  • Fix views.py to protect against CSRF hacking:
from django.template import RequestContext
from django.shortcuts import get_object_or_404, render_to_response
# ...
def detail(request, question_id):
    q = get_object_or_404(Question, pk=question_id)
    return render_to_response('qandabear/detail.html', {'question': q}, context_instance=RequestContext(request))

Notice we also added a function that checks if a 404 is returned for us. This is a common pattern, so there is a pre-built shortcut function for it so we can use fewer lines of code! The details of how the RequestContext works are explained in the documentation for RequestContext.

Process the form

  • Add
    (r'^questions/(?P<question_id>\d+)/answer/$', 'add_answer'),
    
    to urls.py
  • Add add_answer() function to views.py:
from django.shortcuts import get_object_or_404, render_to_response
from django.http import HttpResponseRedirect, HttpResponse
from django.core.urlresolvers import reverse
from django.template import RequestContext
from qandabear.models import Answer, Question
# ...
def add_answer(request, question_id):
    q = get_object_or_404(Question, pk=question_id)
    answer_text = q.answer_set.get(pk=request.POST['answer'])
    a = Answer(text=answer_text, question=q)
    a.save()
    # Always return an HttpResponseRedirect after successfully dealing
    # with POST data. This prevents data from being posted twice if a
    # user hits the Back button.
    return HttpResponseRedirect(reverse('qandabear.views.detail', args=(q.id,)))

This code includes a few things we haven’t covered yet in this tutorial:

  • request.POST is a dictionary-like object that lets you access submitted data by key name. In this case, request.POST['answer'] returns the contents of our textarea (named 'answer'), as a string. request.POST values are always strings.
  • Note that Django also provides request.GET for accessing GET data in the same way – but we’re explicitly using request.POST in our code, to ensure that data is only altered via a POST call.
  • After making and saving our new Answer, the code returns an HttpResponseRedirect rather than a normal HttpResponse. HttpResponseRedirect takes a single argument: the URL to which the user will be redirected (see the following point for how we construct the URL in this case).
  • As the Python comment above points out, you should always return an HttpResponseRedirect after successfully dealing with POST data. This tip isn’t specific to Django; it’s just good Web development practice. That way, if the web surfer hits reload, they get the success page again, rather than re-doing the action.
  • We are using the reverse() function in the HttpResponseRedirect constructor in this example. This function helps avoid having to hardcode a URL in the view function. It is given the name of the view that we want to pass control to and the variable portion of the URL pattern that points to that view. In this case, using the URLconf we set up in urls.py, this reverse() call will return a string like
    '/questions/3/'
    
    where the 3 is the value of q.id. This redirected URL will then call the results view to display the final page. Note that you need to use the full name of the view here (including the prefix).
  • Restart your dev server.
  • Navigate to http://127.0.0.1:8000/questions/1/ in your browser and answer the question. You should see your new answer appear in the list of answers.
  • Save and commit.

Exercises for the reader

  • We just made it so users could add answers. Make another form, on the index page, to let users add questions!
  • Add an 'edit' link to each answer on the detail page, which when clicked takes you to another page where you can edit the answer using a form. When the edit form is filled out, it should take you back to the detail page for the question you came from.
  • Add a 'delete' link to each answer on the detail page that, when clicked, deletes the answer and takes you back to that detail page. Huzzah -- we now have full CRUD functionality for Answers!
  • Add a try/except to your view to check to see if the answer textarea is empty. If so, instead of making a new answer, return the user to the detail page right away with a warning message that their answer was invalid.
  • With more complex models, ModelForms are super helpful in generating a form that will fill out your model without having to match up all the fields from scratch yourself. Rewrite QandaBear to use ModelForms!

Bonus points

Static files (aka: Django for designers)

TODO: Write this!

https://docs.djangoproject.com/en/1.3/howto/static-files/

AJAX and Django

TODO: Write this!

For now, http://webcloud.se/log/AJAX-in-Django-with-jQuery/

Test-driven development, part two

TODO: (walk through how to write tests for a new feature)

http://dougalmatthews.com/articles/2010/jan/20/testing-your-first-django-app/

https://docs.djangoproject.com/en/dev/topics/testing/

Authentication and Users (and reusing apps)

https://docs.djangoproject.com/en/dev/topics/auth/

Deploying to a real live server

TODO