A complete beginner's guide to starting a Jena project in Eclipse
I fairly often get email of the following form:
Dear Mr/Dr/Professor Dickinson
I am Jena newbie. I can't attach the jena to eclipse. could you please give me help to resolve this problem.
I'm not a prof, by the way, but it's flattering to be promoted! So this article is for all those who are new to Jena and Eclipse. This article is not:
- A Java programming tutorial
- Instructions for installing Eclipse
- A comprehensive Jena tutorial
Instead, my goals for this article are to show you how get the most basic possible Jena program set up in Eclipse. In other words, the Jena equivalent of hello world.
Preamble
Jena is a programmer's API (application programming interface) for Java semantic web applications. It isn't a program or tool that you run - if that is what you are looking for, I would suggest perhaps TopBraid Composer as a good option. So the primary use of Jena is to help you write Java code that handles RDF and OWL documents and descriptions. I use Eclipse as my Java development of choice, other environments are available, but I don't use them and so the experience I can pass along is based on Eclipse.
Prerequisites
You will need to:
- Download and install eclipse. This article was written with Eclipse 3.3.1, but the exact version number should not matter as the main Eclipse user interface has been stable for a while.
- Download and unzip Jena. At the time of writing, the current version of Jena is 2.5.5, but again it won't matter if you use a different version.
I'm going to use Windows for this tutorial, but the same steps apply to Linux and, I presume,
MacOS. In fact, on my Windows machine I use a directory structure similar to Linux: rather
than c:\Documents and Settings\ijd as the base directory, I use
c:\home\ijd. This is a personal choice that makes it easier to write scripts
that run on Linux and Cygwin on Windows, but there's no special reason for you to do the
same. However, it is often a good idea to avoid path names with spaces in them if you can.
So my Jena installation is in c:\home\ijd\projects\jena2. Wherever you see this
path in the notes below, substitute the location where you have installed your own copy
of Jena.
Step-by-step guide
This tutorial really is aimed at Eclipse newcomers, so I'm going to take it a step at a time. Feel free to skip ahead if you have some experience with Eclipse already.
Step 1 create a Java project
Eclipse organizes files into projects, so we need a project for this tutorial. Depending on the plugins installed, Eclipse may show a large number of different types of project. A plain Java project is fine for a basic Jena application. If you want to, for example, develop a Java servlet that uses Jena, then choose a suitable project type.
For clarity, I'm starting with an empty Eclipse workspace:
I need to create a new Java project:
Next I need to set up the internal structure of the project. There are many
schools of thought on this. I usually use a basic structure that is taken from
Maven.
Source code is in src/main, with Java code
in src/main/java, ontologies in src/main/owl, etc.
Compiled Java class files go into target/classes. So when I create a new project,
I edit the defaults to reflect these preferences. However, if the defaults suit
you, or you don't know enough to care about those details yet, just accept
the default settings. Alternatively, if your project or educational institution
has suggested guidelines, then go with those.
Step 2: create the hello world class
Java code is stored in files that correspond to the declaration of a Java class, so I need to create a class for my hello world example. In the project explorer pane (which is on the left by default, but you can move it around in Eclipse), I select the project's Java source folder and right-click to get the context menu to show:
Step 3: adding the Jena libraries
OK, so now let's write some Jena code. The first thing I'll need to work
with is a Model: a container for RDF statements. The Model
class is in package com.hp.hpl.jena.rdf.model, so I'll first
import that class, and then create an instance of it.
OK, so why the red wavy lines? These are Eclipse's way of indicating a problem.
If I tried to compile this code using javac on the command line,
I'd get an error saying that the package com.hp.hpl.jena.rdf.model
can't be found anywhere, and that the class Model isn't defined.
On the command line, I would fix this by setting the Java classpath. Essentially,
that's what I do in Eclipse too, but Eclipse makes it rather easier to
do. Notice that I haven't actually said anywhere yet that this is a
Jena project. All I've said is that it's a Java project. What's the difference?
Simply this: Eclipse has to know where to find the Jena classes I would
like to refer to from my program. Eclipse calls the locations where
it can find the supporting code I want to refer to as the build path.
There are actually a few different ways of setting the build path
in Eclipse. One way I could do it is to create a lib directory
in my project top-level folder,
then copy the Jena .jar files there, and then link that
directory to my project's build path. That works, but there's a better way: defining
a user library. A user library is a declaration of a library
(collection of supporting code) that I can reference from any project.
Once I have this set up once, I can use the same library definition in
multiple different projects. Moreover, if I subsequently update Jena to
a new release, then once the library is updated every project in my
Eclipse workspace will see
the new version. With the copy-files-to-the-lib-folder method, I have to re-copy to
every project that uses Jena. Here's how I create the Jena user library, starting
from the Preferences menu:
Click new to create a new user-library:
Now I click on add jars to add the .jar files from
Jena. .jar files contain the compiled Java libraries that
Jena uses, together with the Jena code itself in jena.jar.
The JAR selection dialogue that pops up allows me to select which
.jar files are in my user-library. I have selected all
of the .jar files in the lib/ directory of my
Jena install directory:
Result:
That's actually enough to allow me to use Jena in Eclipse, but there's a
couple of additional optional steps that make programming a bit easier. I
can tell Eclipse where to find the source code and the javadoc for the
Jena classes (I'll show how that's helpful later on). Next to the jena.jar
entry in the user library, there's a little + icon. Clicking that expands
the details of the .jar entry:
I can tell Eclipse that the source code is in the src/ folder
of my Jena install directory. I click on the Source attachment line,
then the Edit... button. In the source attachment configuration
dialogue, I click the External folder button
and then browse to the right location:
Similarly, I can notify Eclipse of the location of the Javadoc by first
selecting the Javadoc location line of the library entry, then
following a similar process. Notice here that the location path is a URL
(it starts file:). This is because the location can also be a Javadoc
web site, though I'm not using that capability here.
With the Jena user library configured, I click OK to close the library configuration dialogue.
Step 4: Finishing the hello world program
Now I can go back to my project, and configure the Java build path to use the library I just created. To start, I right-click on the project node in the explorer window to bring up the project properties menu, navigate to the build path menu option and add the library:
Having updated the build path, Eclipse will automatically rebuild the
project (i.e. recompile the Java code). With that, some of the
errors will go away, since the import statement can now
find the class to be imported, and so the Model class name
is meaningful to the compiler.
However, there is still a remaining error because ModelFactory
is not defined. What is needed is a suitable import statement.
This is easily fixed in Eclipse, either by clicking on the error symbol
(the red 'x' on the left margin), or by positioning the cursor just after
the ModelFactory class name, and pressing ctrl-space,
which triggers Eclipse to show the possible completions for the name:
When I select the first of the presented options (i.e. ModelFactory
rather than ModelFactoryBase), Eclipse will fill in the
import statement automatically:
Look, ma, no errors!
Eclipse's auto-complete feature is also useful when adding code. For
example, if I type Resou followed by ctrl-space I get the
possible completions that match that name:
Notice the Javadoc comment in yellow on the
candidate completion (Resource in this case). This extra
information comes from having added the source code and javadoc locations
when I specified the library in step 2, above. If you miss out specifying
the source code and javadoc, Eclipse can't be so helpful in describing
possible auto-completions. It also will affect the debugging view, though
I'm not discussing that in this article.
Here is the completed hello world program:
Step 5: running the hello world program
To run this program within Eclipse, I use the run menu, accessed from the
button showing a white triangle on a green circle. Since I haven't run any
code yet, I have to tell Eclipse what program to run. Click on the drop-down
menu to the right of the run button, and tell Eclipse to run HelloRDFWorld
as a Java application:
Which gives the following output in the Eclipse console:
And that's it: from empty Eclipse to a working RDF hello-world application. Obviously there is lot's more to learn (see the tutorial links at the top for some jumping-off points), but hopefully you can now have fun developing new semantic web applications!