Implement extensions rule

[lo1tuma]

This fixes #209 and fixes #204.

[jfmengels]

exception an allows --> exception and allows

[lo1tuma]

Thanks, fixed.

[jfmengels]

object-assign is duplicated (and linked to that: there seems to be missing a comma after pkg-up).

[lo1tuma]

Fixed now.
Seems like I did a bad job at resolving the conflicts after rebasing.

[jfmengels]

I think you can also add an entry to the changelog (and thank yourself).

Also, do you mind adding support for require? We had a discussion about a similar case here #245 (comment).

LGTM otherwise, nice job @lo1tuma!

[benmosher]

@jfmengels is there a utility function in what you've added to filter CallExpression to just CommonJS require calls? I've got something off in my utils branch, but I thought you had something similar.

I'm imagining something so CommonJS support is as easy as:

return {
  ImportDeclaration: checkFileExtension,
  CallExpression: filterRequire(checkFileExtension),
}

Anyway @lo1tuma if you don't want to worry about CommonJS, I'm cool with that, we can do it post-merge.

[jfmengels]

Yeah, I have a function for that, with a use of it here. You can't do something as simple as @benmosher proposed, as the same data (the source for instance here) as not at the same property path. It's more test code than code.

we can do it post-merge

👍, I can probably take a look at it if you wish.

[benmosher]

Ah, right. I made a moduleVisitor* to which you pass a function of the Literal node, here it is in use for no-unresolved: https://github.com/benmosher/eslint-plugin-import/blob/core-package/src/rules/no-unresolved.js

I really need to get that stuff merged back in. CJS vs. ES6 is cross-cutting and common enough that ideally, rules wouldn't need to worry about it explicitly.

*note: the JSDoc mistakenly documents the prototype of the visitor as Function(String), but it really does take the node object representing the string in the linted source.

[lo1tuma]

I’m not very familiar with AMD so I could only provide an implementation for commonjs.

Apart from that, I think it would be better to wait for the moduleVisitor and not implementing the same logic in every rule.

[jfmengels]

Sounds good to me

[lo1tuma]

I totally forgot about the CHANGELOG. Should be all good now.

[benmosher]

Just merged, thanks!