This tutorial is also available as a screencast. The video moves quite quickly to demo TurboGears and hold your interest, but a number of new users have reported having difficulty following certain parts. This tutorial initially was a direct translation of the video, but has been rewritten to be a gentler introduction.
This tutorial refers to TurboGears 1.1 which is outdated and no longer the recommended version to start with. If you are new to TurboGears, we recommend starting with the latest version of the TurboGears 2 branch based on Pylons. If you still want to use the older TurboGears 1 branch based on CherryPy, we recommend using its latest version TurboGears 1.5. The TurboGears homepage has links to the documentation and tutorials for these versions.
We recommend creating the code for this tutorial on your own, as you
follow each step of the tutorial. But you can also download the
final code for this tutorial
if you want to spare some typing or you’re getting stuck somewhere.
To go through this tutorial, you’ll want:
TurboGears 1.1 or higher. ;-]
A database. If you don’t have one, your best bet is sqlite 3.2+ with pysqlite 2.0+ which is part of the standard libs in Python 2.5+. If you are using an older Python version, you need to install it with:
SQLAlchemy 0.4.3 and Elixir 0.6.1 or higher to access your database.
docutils 0.4 or later, which is used for the wiki’s formatting. docutils is not a required part of TurboGears but is needed for this tutorial. You can install docutils with:
A web browser
Your favorite editor
Two command line windows (you only need one, but two is nicer.)
optional If you’re not aware of it, you may also find the ipython shell
to be helpful. It supports attribute tab completion for many objects (which
can help you find the method you’re searching for) and can display contextual
help if you append a question mark onto the end of an object or method. You
can do the same in the standard shell with the
functions, but ipython is more convenient. ipython has a number of other
convenient features (like dropping into the debugger on error), check the
ipython docs or online help.
This tutorial doesn’t cover Python at all. Check the Python homepage page for more coverage of Python.
TurboGears has a command line tool named
tg-admin, which provides a suite of
tools for working with TurboGears projects. A few will be touched upon in this
tutorial, check the tg-admin reference for a full listing. The
first tool you’ll need is
quickstart, which initializes a TurboGears project.
Go to one of your command line windows and run the following command:
tg-admin quickstart --elixir
--elixir options tells
quickstart to create a project using Elixir
instead of pure SQLAlchemy for its database layer. Elixir makes using SQLAlchemy
a bit more convenient and more similar to SQLObject that has been used by default
for the database layer in TurboGears 1.0.
You’ll be prompted for the name of the project (this is the pretty name that human beings would appreciate), and the name of the package (this is the less-pretty name that Python will like). For the identity prompt, answer ‘no’, since we’ll keep this wiki fairly simple, but when you need users/passwords in a future project, you’ll want to look up the identity management tutorial. Here’s what our choices for this tutorial look like:
Enter project name: Wiki 20 Enter package name [wiki20]: wiki20 Do you need Identity (usernames/passwords) in this project? [no] no ...output...
This creates a few files in a directory tree just below your current directory. Let’s go in there and you can take a look around:
You may have spotted a file called
start-wiki20.py. This script starts the
built-in web server. Let’s run it! Go to your second command line window and run:
Point your browser at http://localhost:8080/, and you’ll see a nice little welcome page with the current time. (If you’re on a Mac and have Bonjour bookmarks turned on in Safari, you’ll see your new server show up there!)
Easy indeed. You have a working project! If you take a look at the code that
quickstart created, you’ll see that there isn’t much involved in getting up
and running. In particular, you’ll want to check out the two files directly
involved in displaying this welcome page:
wiki20/controllers.py has the code that’s generating the welcome page.
CherryPy (TurboGears’ controller/app server) works using object publishing.
You write methods and
@expose() them to the web.
TurboGears adds template processing to the default CherryPy decorator, you just specify your template, return a dictionary, and TG takes care of the rest. This has a number of advantages over returning some sort of string object, which we’ll get to later in this tutorial.
wiki20/templates/welcome.html is the template you view on the welcome
screen. Did you notice that it’s standard XHTML with some simple namespaced
attributes? Very designer-friendly. You can even open it directly in your
If you’re not familiar with a wiki, you might want to check out the Wikipedia entry. The whole idea is that it’s an easily-editable collaborative web content system that makes it trivial to link to pages and create new pages.
TurboGears follows the Model-View-Controller paradigm, as do most modern web frameworks. Genshi templates are your view, your CherryPy classes are your controllers and basically any object can be your model. In practice, since we’re in a database-driven world, your model will be based on a relational database. Elixir over SQLAlchemy makes working with databases easy.
TurboGears quickstart gave us a
wiki20/model.py module with enough in it to
start creating model classes. Since a wiki is basically a linked collection of
pages, we’ll define a
Page class as the name of our model. Here’s what it
looks like, you’ll want to add it to your
wiki20/model.py right below the line
# your data model:
# your data model class Page(Entity): pagename = Field(Unicode(30), unique=True) data = Field(Unicode) # let Elixir do the setup setup_all()
Most TurboGears developers and users prefer to work with objects (and not SQL) as much as possible. Elixir does support creating objects based on the database definition, but if you work the other way and create the database definition based on the objects, your database information lives in one place and your table-declarations are database independent.
pagename, we specified
unique=True, which will guarantee
that the column is unique and allow us to select a
Page from our model using
pagename rather than having to do so by its’ (automatically created)
id (surrogate) primary key. Some databases require the length to be
specified on indexed columns, so we’ll just use the arbitrary length of 30 here
to prevent potential problems.