Fix: Stop path analyzer on unknown nodes

[ilyavolodin]

Prerequisites checklist

• I have read the contributing guidelines.
• The team has reached consensus on the changes proposed in this pull request. If not, I understand that the evaluation process will begin with this pull request and won't be merged until the team has reached consensus.

What is the purpose of this pull request? (put an "X" next to an item)

[ ] Documentation update
[x] Bug fix (template)
[ ] New rule (template)
[ ] Changes an existing rule (template)
[ ] Add autofixing to a rule
[ ] Add a CLI option
[ ] Add something to the core
[ ] Other, please explain:

What changes did you make? (Give an overview)

Updated code path analyzer to bail if no known ESTree nodes are found in AST. I'm trying to get ESLint to work with syntaxes other then ESTree and this is one place where it throws an error, since all the AST nodes are hardcoded in code-path-analyzer and can't be supplied by the parser. Unfortunately, it would be very tricky to create a unit test for this, and since this is a one-line change that shouldn't affect happy path, I figured it would be OK without unit test.

Is there anything you'd like reviewers to focus on?

Nope

[jsf-clabot]

All committers have signed the CLA.

[mysticatea]

Thank you.

Hmm. I guess, the analyzer still may crash on a node that has the same type name as one of the ESTree nodes. I think better if we expose a flag for custom parsers to disable code path analysis and modify lib/linter/linter.js#L941.

[ilyavolodin]

Sure, that would probably be ideal way to fix this. This is just a very quick fix to unblock linting unknown nodes. Since a new flag from parser would require RFC and a bunch of code changes, could we merge this one in first, and I will create an RFC for supporting generic ASTs (was planning on doing that anyways), which will include this flag as well as few more things.

[mysticatea]

My concern is that this patch is incomplete; if the root node is not Program and the tree contains the node that is the same type as any of ESTree nodes, then the analyzer still can crash. I'd like to stop instantiating the analyzer rather than modifying the inside of the analyzer.

How about, stop instantiating the analyzer if the root is not a Program?

[ilyavolodin]

@mysticatea I updated the code to stop code path analysis all together if top level node is not Program. I also added unit test to cover this scenario. I also left the old code, just in case, top level node is Program, but followed by unknown nodes. Shouldn't hurt anything anyways.

[mysticatea]

Thank you.

Looks the linter does nothing at all if the root node is not a Program. I guess it's not intentional.

[ilyavolodin]

Ah, sorry about that. Not sure what I was thinking. Updated with the right code. Should be working fine now. Updated unit test also to verify that it's actually running rules.

[mysticatea]

LGTM, thank you.

[kaicataldo]

LGTM. Glad we were able to get to a more complete PR 👍

[kaicataldo]

Marking as accepted since we have the approval of 3/4 of the TSC.