Rule auto-fix variable name conflicts

The version of ESLint you are using.

v7.7.0

The problem you want to solve.

I have this code

function foo() {
	try {
	} catch (err) {
		console.log(err);

		if (test) {
			throw a;
		} else {
			throw b;
		}
	}
}

We added a rule unicorn/catch-error-name with auto-fix, it will fix err to error.

function foo() {
	try {
	} catch (error) {
		console.log(error);

		if (test) {
			throw a;
		} else {
			throw b;
		}
	}
}

We are adding a new rule prefer-ternary, we plan to fix this code to

function foo() {
	try {
	} catch (err) {
		console.log(err);

		const error = test ? a : b;
		throw error;
	}
}

Here is the problem, when we check variable name error in prefer-ternary rule and catch-error-name, they will both show error is not used, but when apply fixes, there will be conflicts.

function foo() {
	try {
	} catch (error) {
		console.log(error);

		const error = test ? a : b; // <-- error is already used.
		throw error;
	}
}

Any solution/suggestion?

Your take on the correct solution to problem.

N/A

Are you willing to submit a pull request to implement this change?

N/A

[mdjermanovic]

I reopened this issue to discuss if it would be useful to add something to public API (since FixTracker isn’t public), but now I realized there’s a simple way to extend the fixed range using the already existing features.

If the rule wants to fix something inside a node and make sure there will be no other fixes inside the node in the same pass, the fix() can just add empty strings around the node:

*fix(fixer) {

    // yield some fixes inside `node` ...

    // extend the range to `node`:
    yield fixer.insertTextBefore(node, "");
    yield fixer.insertTextAfter(node, "");
}

Then, the report translator will merge fixes and the unchanged code into one big fix having the range of node.

[nzakas]

All right, closing.