[New] newline-after-imports: new option exactCount and docs update

[anikethsaha]

Fixes #1901. Fixes #514.

Checklist

• option implementation
• tests
• docs update

added option named exactCount of type Boolean default false that will enforce strict following of the count option where the number of newline has to be equal to count.

Updated the docs as mentioned in the linked issue accordingly.

[coveralls]

Coverage increased (+0.9%) to 82.365% when pulling bb84371 on anikethsaha:fixes-1913 into 7c1e8e4 on benmosher:master.

[golopot]

This also fixes #514.

[ljharb]

Seems legit.

[ljharb]

Looks like there's some test errors in eslint 2.

[bartzy]

@anikethsaha Are you planning on merging this?

[reosarevok]

I would love this - any way I can help move it forward?

[ljharb]

If someone wants to rebase this branch, fix the test errors, and post a link to the branch here (NOT a new PR, please!) then I can pull in the changes and we can get this merged.

[reosarevok]

I rebased, fixed conflicts, fixed eslint errors, and added a test using this in combination with the newer considerComments option. I ran npm test again and there's no issues with these specific files anymore (there's a few unrelated eslint warnings elsewhere which I didn't start fixing now). I don't know if there's more tests I could and should be running (I see your CI here runs a lot more tests than that). Anyway: https://github.com/reosarevok/eslint-plugin-import/tree/fixes-1913

[ljharb]

alrighty, if the tests pass we can un-draft this and move forward

[ljharb]

@reosarevok the tests are still failing in eslint 2, pretty consistently, about the autofix output. it would be ideal to figure out how to get them passing in eslint 2.

[reosarevok]

Oh, I think I figured out the issue - other places that don't autofix do return undefined (not sure if that's important, but it works anyway so I made the change) and have output matching the code rather than output: null as this PR originally had. I tried that with eslint 2 and it seems to work now (same branch, force-pushed). If I'm missing something else then do let me know, I haven't played with very old eslint before nor with writing my own rules :)

[ljharb]

The travis failure is because they're on an outdated nvm, and npm 10 breaks on node 14; appveyor's WSL failures are expected. Once the rest of the appveyor tests pass, I'll land this.

[FFdhorkin]

There's a comment related bug with this; I've filed a new bug since this is already merged.