[Fix] guard against empty parent

[laurens-dg]

Without this fix I get a crash on my current linting set-up.
Not sure what is causing the parent to be empty, might be a bug in the parser, but I ran into this updating everything to the latest versions.

[ljharb]

We’d need a regression test to land this - can you figure out which code is crashing?

[laurens-dg]

Absolutely, but I don't know how to go about this, I get this crash for parents with type ExportNamedDeclaration I've got about 20 of those in my project.

The crash looks like this:
ESLint: 8.45.0

TypeError: Cannot read properties of undefined (reading 'indexOf')
Occurred while linting /foo/bar/baz.tsx:18
Rule: "import/newline-after-import"
at checkImport (/foo/bar/node_modules/eslint-plugin-import/lib/rules/newline-after-import.js:155:40)

[ljharb]

What's on line 18 of that tsx file?

[laurens-dg]

export import Foo = ns.baz.bar.Foo;

We have a bunch of namespaced classes which we rename and re-export in this way.

[ljharb]

woof, that is some gnarly TS syntax :-) i know about export = and import = but not export import =. thanks - it shouldn't be too tricky to come up with a test case based on that.

[ljharb]

Although in this case, the rule should presumably be enforcing a newline after the export import = line, not ignoring it?

[laurens-dg]

To be honest, I don't know.
I also believe this used to work but it might be a recent change to the @typescript-eslint/parser that causes this issue.
A bit further in the code this case is considered and skipped. But we see errors before that check with the indexOf, hence I created this PR.

[ljharb]

aha, #1830 indeed attempts to ignore them, treating them as exports rather than imports.

In this case, it sounds like the right bug fix, as long as there's a regression test.

[laurens-dg]

Thanks for the help and feedback in getting this merged!