Fix requireindex usage to use correct directory

[bmish]

There's an unreleased commit that switched the plugin to use requireindex for loading the rules: d3e0440. I noticed this broke the plugin for me.

eslint-plugin-qunit is included in a config in another linting plugin of mine. As a result, the requireindex call was checking the wrong current directory for lib/rules.

Oops! Something went wrong! :(

ESLint: 7.10.0

Error: Failed to load plugin 'qunit' declared in '.eslintrc.js » plugin:my-plugin/my-config': Cannot find module 'my-home-folder/eslint-plugin-qunit/lib'
Require stack:
- my-home-folder/eslint-plugin-qunit/node_modules/requireindex/index.js
- my-home-folder/eslint-plugin-qunit/index.js
- my-home-folder/my-repo/node_modules/eslint/lib/cli-engine/config-array-factory.js
- my-home-folder/my-repo/node_modules/eslint/lib/cli-engine/cascading-config-array-factory.js
- my-home-folder/my-repo/node_modules/eslint/lib/cli-engine/cli-engine.js
- my-home-folder/my-repo/node_modules/eslint/lib/eslint/eslint.js
- my-home-folder/my-repo/node_modules/eslint/lib/eslint/index.js
- my-home-folder/my-repo/node_modules/eslint/lib/cli.js
- my-home-folder/my-repo/node_modules/eslint/bin/eslint.js
Referenced from: my-home-folder/my-plugin/lib/index.js
    at Function.Module._resolveFilename (internal/modules/cjs/loader.js:1029:15)
    at Function.Module._load (internal/modules/cjs/loader.js:898:27)
    at Module.require (internal/modules/cjs/loader.js:1089:19)
    at require (my-home-folder/my-repo/node_modules/v8-compile-cache/v8-compile-cache.js:161:20)
    at my-home-folder/eslint-plugin-qunit/node_modules/requireindex/index.js:11:28
    at Array.forEach (<anonymous>)
    at module.exports (my-home-folder/eslint-plugin-qunit/node_modules/requireindex/index.js:9:15)
    at Object.<anonymous> (my-home-folder/eslint-plugin-qunit/index.js:13:12)
    at Module._compile (my-home-folder/my-repo/node_modules/v8-compile-cache/v8-compile-cache.js:194:30)
    at Object.Module._extensions..js (internal/modules/cjs/loader.js:1220:10)
error Command failed with exit code 2.

[coveralls]

Coverage remained the same at 100.0% when pulling 70ca2e2 on bmish:fix-require-index into 7fb119d on platinumazure:master.

[platinumazure]

Thanks for this. The change looks good generally.

Do you think it might be possible to add a test case? I would really like to catch this in CI if there's a regression.

[bmish]

A test would be great but I don't immediately know how to test this...

[platinumazure]

A test would be great but I don't immediately know how to test this...

@bmish It seemed like you could recreate it pretty easily outside of the package. I assume there's something about being inside the package that makes this difficult to test. Can you describe the issue you're seeing (or foreseeing) here? Maybe we can come up with something.

Off the cuff, the first thing I can think of is not being able to require("eslint-plugin-qunit") from inside the plugin (or rather, inside the test suite). If that's the case, could it be faked using a nested package.json inside a subdirectory of the tests directory?

[bmish]

I experimented with testing for a while and didn't have any luck. Tried a few things (using process.cwd() / process.chdir() before require('index.js'), eslint.calculateConfigForFile('index.js')) but was unable to repro from the issue from a test.

I did however check to confirm that other usages of requireindex on GitHub all appear to use __dirname: https://github.com/search?l=JavaScript&q=requireindex&type=Code

[platinumazure]

Thanks for taking the time @bmish, I appreciate it. Guess we can merge this as it is.