How to locate Scala feature stats

[Jing Vergara]

Hi,

I'd like to locate the number of classes and interfaces for Scala programs. I tried doing this with this source but it's still returning output for other languages. What am I missing?

Thanks,

Jing

# Computes Number of Attributes (NOA) for each project, per-type

# Output is: NOA[Project Language][Project Name][TypeName][Type Kind] = NOA value

p: Project = input;

NOA: output sum[string][string][string][string] of int;

visit(p, visitor {

    # only look at the latest snapshot

    before n: Project -> ifall (i: int; !match(`^scala$`, lowercase(n.programming_languages[i]))) stop;

    before n2: CodeRepository -> {

        if (hasfiletype(n2, "scala")) {

        snapshot := getsnapshot(n2);

        foreach (i: int; def(snapshot[i]))

            visit(snapshot[i]);

        stop;

        }

    }

    before node: Declaration -> {

        if (node.kind == TypeKind.CLASS)

            NOA[p.programming_languages[0]][p.name][node.name]["class"] << len(node.fields);

        if (node.kind == TypeKind.ANNOTATION)

            NOA[p.programming_languages[0]][p.name][node.name]["annotation"] << len(node.fields);

        if (node.kind == TypeKind.ANONYMOUS)

            NOA[p.programming_languages[0]][p.name][node.name]["anonymous"] << len(node.fields);

        if (node.kind == TypeKind.DELEGATE)

            NOA[p.programming_languages[0]][p.name][node.name]["delegate"] << len(node.fields);

        if (node.kind == TypeKind.ENUM)

            NOA[p.programming_languages[0]][p.name][node.name]["enum"] << len(node.fields);

        if (node.kind == TypeKind.GENERIC)

            NOA[p.programming_languages[0]][p.name][node.name]["generic"] << len(node.fields);

        if (node.kind == TypeKind.INTERFACE)

            NOA[p.programming_languages[0]][p.name][node.name]["interface"] << len(node.fields);

        if (node.kind == TypeKind.OTHER)

            NOA[p.programming_languages[0]][p.name][node.name]["other"] << len(node.fields);

        if (node.kind == TypeKind.STRUCT)

            NOA[p.programming_languages[0]][p.name][node.name]["struct"] << len(node.fields);

    }

});

[Robert Dyer]

Hi Jing,

I believe you are running into a combination of two issues:

1) Projects manually set what programming languages they use and can select more than 1 language.  As such, some projects that select Scala (probably) also select Java, etc.  So your source code analysis would include non-Scala code as well, if those projects have it.

2) Currently our dataset only has source code information for Java source files (.java file extensions).

So these two issues, acting together, mean you are mining the *Java* source code from *Scala* projects!

We are hoping to add support for Scala source code in the near future.  We just need a good (read: fast and robust to errors) parser for Scala files! :-)

Note that in the future, if you want to filter a specific kind of file[1], you can do that by adding a visitor for ChangedFile.  For example:

  before node: ChangedFile -> if (node.kind != FileKind.SOURCE_JAVA_JLS2) stop;

You can also make use of the 'iskind' function[2] to easily match several kinds (in this case, any parseable Java file):

  before node: ChangedFile -> if (!iskind("SOURCE_JAVA", node.kind) stop;

- Robert

[1] http://boa.cs.iastate.edu/docs/dsl-types.php#enum+FileKind

[2] http://boa.cs.iastate.edu/docs/dsl-functions.php