New Importer: IndentationLevelImporter

Feature request

Description

Create an importer that counts the indentation level of each line and generates an cc.json output with attributes like this:

{
    "il1+" : 1,
    "il2+" : 3
}

where il1 is the number of lines in the file with indentation level 1 or higher.

The indentation level is calculated by

(positionOfTheFirstCharacterInLine / tabWidth) + (positionOfTheFirstCharacterInLine % tabWidth == 0 ? 0 : 1)

The tabWidth needs to be estimated before.

The importer should be executable with: ccsh indentationlevelimporter FOLDER/FILE -p PROJECT_NAME -o OUPUT.CC.JSON

[alexhunziker]

One question, does this generalization conflict with ADR 6: Filters share nothing ?

I would suggest implementing it like the SCMLogParser, in other words all metrics are calculated by the same filter and thus the share nothing principle will not be violated. The potential drawback is, that there may be (too) many command line arguments (… that is if we make the metrics customizable (i.e. max indentation level, predetermined tab width, …). if we just hard-code/guess the parameters as done in the SCMLogParser the interface stays clean. It is a trade-off i guess)

How do we name the importer, as we already have the “SourceCodeParser”, which were an appropriate name for this one as well?

Good question, I asked myself the same and have found no good answer yet.

[alexhunziker]

@alexhunziker I am not into the code yet, but did think about the option of implementing these new custom metrics into the SourceCodeParser? Would that make sense or is the architecture rather very different?

The architectures are indeed very similar, and a combination technically possible. However I don’t know if this is conceptually desirable, since they do kinda different things; the (sonar) sourceCodeParser gets base metrics for certain languages (currently java only) while this thing here provides some special metrics for any file scanned (that may or may not be meaningful). I don’t know if the use case for this and the (sonar) sourceCodeParser are similar enough to fuse them together. Also the CLI options of a fused importer may get confusing.

Why put it into the SourceCodeParser? @alexhunziker what metrics does the IndendationLevelParser create that the SourceCodeParser also creates? What is actually the same? The ILParser does not have any Sonar and doesn’t it only create indendation level, while the SCParser creates all Sonar metrics + custom nesting level?

The metrics we are obtaining with the new Parser cannot be obtained by Sonar Directly, the only thing it does is computing the maximum nesting level (we created a custom metric for that). It would be possible to generate a custom metric for the sonar plugins that provides the same information as the indentation level parser does, however they would be only applicable to certain programming languages (i.e. Java only at this point)