How can I parse tokens but ignore some occurrences?

[felixjb]

I am using the @babel/parser to find the testing framework keywords and add VSCode codelens on them. However, it gives a false positive when the code contains a regex test method.

I would like to find the word "test" for the Jest and Mocha frameworks, but not for a regex test method.

Here's the code where do the parsing: Testify: code parser

[kaicataldo]

In the context of tokens, you could potentially check the previous token to see if it's a . or not. Otherwise, I think this would be easier in the context of the AST, as you can specifically check if the call is a MemberExpression or not.

[kaicataldo]

Can you clarify which option you're asking about? I would recommend traversing the AST rather than iterating over the tokens, since you already have it there.

[felixjb]

Sorry, I'll clarify.

Today I am iterating over the tokens to find occurrences of test keywords from unit testing frameworks. When I find them, I return their position.

How could I replace this implementation with traversing the AST?

[kaicataldo]

Here's generally how I would approach this (I haven't done this specifically, so you might have to do some tweaking):

1. Traverse the AST, visiting each node starting from the root
2. When you encounter a node with type CallExpression, check the callee property, verify that it's type: Identifier, and check the name value.
3. Use the location of the Identifier node.

[felixjb]

Nice! I'm gonna try this then.

[kaicataldo]

Forgot to mention, https://astexplorer.net/ is really useful for visualizing the AST!