szabgab has asked for the wisdom of the Perl Monks concerning the following question:

I am not sure exactly what am I looking for so let me just think aloud. Meditate about this.

I need some tool that will help me analyze the source code of an application written in Perl.

I think some of the analyzes should based on Perl::Critic but I'd probably also like to find duplicate pieces of code. Copies of functions or smaller parts of code. Cases where the copies are not 100% identical but that are very similar.

I'd like to be able to draw a map of the class hierarchy and where is each class, method or function being used.

I am sure there are other ways to analyze code. Ideas?

Replies are listed 'Best First'.
Re: Analyzing a Perl application
by Corion (Patriarch) on Jan 31, 2011 at 13:03 UTC

    There is B::Xref, which claims to create cross-reference sections.

    For class inheritance, I'm sure there is a module that lists you the tree, as Ovid uses fancy graphviz output in the paper, and I doubt he created that manually.

[reply]
[reply]
Re: Analyzing a Perl application
by andreas1234567 (Vicar) on Jan 31, 2011 at 13:55 UTC
    I'd probably also like to find duplicate pieces of code.
    You could hash all code lines and compare à la Eric Raymond.

    --
    No matter how great and destructive your problems may seem now, remember, you've probably only seen the tip of them. [1]
[reply]
Re: Analyzing a Perl application
by locked_user sundialsvc4 (Abbot) on Jan 31, 2011 at 21:28 UTC

    This is, un-fortunately, one of those cases where “TMTOWTDI” can turn-around and bite you in the (!).   There are many syntactically-different ways to express the same meaning in terms of source-code, and when you say, “are very similar,” you unfortunately (probably) mean, “means, or does, the same thing.”   And that kind of analysis ... which BTW I have done many, many times ... requires “the human eye.”

    I happen to have more-than-trivial knowledge of the arcane arts of “source code analysis,” having worked for an early-90’s startup that tried (in vain...) to do this sort of thing with COBOL.   (They went public anyway, then of course went bust.)   You can map out data-dependencies and control flows (as I have myself done recently, in another Meditation), but you cannot discern meaning.

    Certainly, you can automatically scan source-code to build up a data-dictionary and a code-dictionary ... and this, for an app that might involve many hundreds or even thousands of source-components, can certainly be useful.   These undertakings very quickly outstrip the limits of human visualization, even for the most spectacular of totally self-absorbed nerds.

      Finding code snippets that mean the same even though written differently would be of course super awesome but for now I would be happy to locate instances of "copy-paste development".
[reply]
Re: Analyzing a Perl application
by planetscape (Chancellor) on Feb 01, 2011 at 11:43 UTC
[reply]
Re: Analyzing a Perl application
by Jeppe (Monk) on Feb 01, 2011 at 14:38 UTC
    I've found that running the code through perltidy (using the same format) is a good idea, it will transform code to be uniformly formatted. That way, whitespace differences etc are eliminated as a source of difference between two pieces of code. Obviously, you should run the perltidy script before any other systems that searches your codebase for duplicate code or near-duplicate code. Using perltidy will improve the false-positive rate of the other systems.

    (You can even enforce a uniform coding format by using perltidy with the available hooks in source control products, but that's a different subject)

[reply]
Re: Analyzing a Perl application
by repellent (Priest) on Feb 06, 2011 at 19:45 UTC
[reply]

Back to Seekers of Perl Wisdom