View Image

A framework to write static analysis tests.

Unlike tools such as checkstyle, PMD or findbugs, Freud attempts to give the user easy and full control of their static analysis tests. Freud does not attempt to give you out-of-the-box tests and tell you anything about what's needed to be tested. You can always use the examples, but once you get familiar with those, you will probably just want to write one yourself.

Using a DSL for every supported type of analysed entity, Freud allows users to define their own targeted test. What you get is the ability to write a test specifically tailored to your code and asserts your conventions.

Although some tools allow you to write your own tests, this is usually a complicated process of implementing an interface that requires you to understand the data structure and inner workings of the tool you're using.

In my experience - the most important static analysis tests I've used where the ones I had to write myself. The ones that are testing directly our own special patterns, conventions and contracts that are not protected in code by the compiler.

Yes - forcing something through signatures is the strongest way of defining a contract but sometimes its not possible. Existance (or not) of annotations on specific methods is one case - Related implementation of more than one method (equals and hashcode for example) is another case.

Freud's DSL is a group of static hamcrest matchers that implement common assertion about the tested items.

Freud is now completely pluggable meaning you can easily write your own hamcrest matcher and assert anything you like. Freud will accept any matcher - not just its own.

Although written in Java and initially targeted towards analysing Java sources, Freud is also not restricted to any specific source or content type.

The support provided within Freud is for:

1. Java sources
2. Java class objects. (i.e analysing the java.lang.Class object)
3. Java class files (i.e analysing the ".class" file)
4. CSS files
5. properties files
6. plain text files
7. spring xml configuration files

Different tests are much easier to implement using different content such as java sources / text files / class obejcts / class files). Also, a lot of java projects contain properties files... some may even have spring framework configuration files...

Freud was desiged to be content type neutral and currently contains DSL and parser support for all content types above.

Analysing another type of content means writing a parser for that content that will convert a resource such as a file to an object representation of the content.

Then all is left is to write Matchers for that object representation.