[Feature Request] Add `addFromDir` option

[mrazauskas]

Currently .addFromDir() is method, but I would like to use it as an option. If that isn’t complicated to implement.

For instance, inputDir is somewhat similar option, so I though why not to have addFromDir as an option too? Also cleanCache or clean can be used as options or as methods.

---

I had two separate scripts to generate raw report and to merge coverage into codacy report. Since the codacy report was added, I merged both scripts into one. Now looking at the file it seems like addFromDir could be an option too. See this line: https://github.com/tstyche/tstyche/blob/8e16c416dc614226974bf8be29ab884fcf6b51e7/scripts/report-coverage.js#L53

[cenfun]

Yes, it's good idea to provide a option for .addFromDir() which is just a string type. Before do that let‘s consider how to name this option?

So currently we already have:

• outputDir - output reports inlcudes raw report
• inputDir - input multiple raw reports for merging report

To avoid confusing the users, what is name of this option for raw v8 coverage json files?

• addFromDir? - it is a method name, is it OK?
• rawV8CoverageDir? - A bit long
• coverageDataDir?
• dataDir?

What's your idea?

Additional explanation for the different data:

• raw v8 coverage data

  • provide by v8, it includes coverage data and source-map-cache for all files (intenal modules, node_modules, everything, it could be redundant)

  • Node.js: Simply use NODE_V8_COVERAGE or Inspector API

  • Browser: Playwright or Puppeteer coverage API (without source-map-cache provided)

  • CDP API for both Nodejs and Browser

  • use for add(data) or addFromDir(data-dir)

  • MCR raw report data

  • it's filtered data (by entryFilter) after added, which is user focused only, it includes coverage data and separated sources/sourcemaps (multiple coverage data but unique sources/sourcemaps)

  • use for generating all supported reports and also inputDir for merging report.

  • all those data cached in cacheDir to support multi-process mode. (for example Playwright parallelism mode)

[mrazauskas]

What if inputDir would be able to handle both types of data? The name of inputDir is so abstract. This can be any supported input data. Feels intuitive.

The input data type could be specified like this:

{
  inputDir: ['./v8-coverage', { type: 'v8' }]
}

{
  inputDir: ['./raw-coverage-unit', './raw-coverage-e2e', { type: 'raw' }]
}

// 'raw' is the default and can be omitted, so all works just like now
{
  inputDir: ['./raw-coverage-unit', './raw-coverage-e2e']
}

I think, the type must be specified for inputDir, not for each path. Otherwise config can get too complex to interpret.

[mrazauskas]

Alternatively input data type could be simply specified via some inputType option:

{
  inputDir: './v8-coverage',
  inputType: 'v8'
}

[cenfun]

Yes, your approach should work, but as you mentioned, it requires an additional specification of type. I personally think using a new option name might be better.

[cenfun]

I prefer to dataDir

[mrazauskas]

Sounds good too. Indeed specifying type is somewhat clumsy.

[cenfun]

please try new version 2.9.0 for new option dataDir

[mrazauskas]

Thanks! All works as expected.