WebDriver: Partial implementation of Element Clear endpoint

[hanpham32]

This PR addresses issue #1040 :~)

Overview:

• Implemented sections 1, 2, and 3 of the Element Clear endpoint.
• Reordered Web::WebDriver::Error entries alphabetically.

[ladybird-bot]

Hello!

One or more of the commit messages in this PR do not match the Ladybird code submission policy, please check the lint_commits CI job for more details on which commits were flagged and why.
Please do not close this PR and open another, instead modify your commit message(s) with git commit --amend and force push those changes to update this PR.

[jdahlin]

I'm not a reviewer, but I would suggest that you either implement the whole line or keep FIXME.

The code below is untested, but something like it is what the spec expects.

Suggested change

	Web::DOM::Element* element = TRY(get_known_connected_element(element_id));
	if (!element) {
	return Web::WebDriver::Error::from_code(Web::WebDriver::ErrorCode::NoSuchElement, "Element not found");
	}
	Web::DOM::Element* element = TRY(get_known_connected_element(element_id));
	if (!element) {
	return Web::WebDriver::Error::from_code(Web::WebDriver::ErrorCode::NoSuchElement, "Element not found");
	}
	auto inner_html = TRY(element->inner_html());
	if (inner_html.is_empty()) {
	return ""sv
	}

[trflynn89]

We don't need to check the element for null. Take a look at the implementation of get_known_connected_element - it only returns non-null pointers, and already returns a NoSuchElement if the element was null.

(It returns a pointer instead of a reference because ErrorOr<> cannot hold references. We could update it to return NonnullGCPtr instead, but that's a separate issue.)

So, this is just:

auto* element = TRY(get_known_connected_element(element_id));

// 1. If element's innerHTML IDL attribute is an empty string do nothing and return.
if (TRY(element->inner_html()).is_empty())
    return JsonValue {};

(Not sure TRY will work exactly here, since inner_html return a DOM exception, but that's the gist of it).

[trflynn89]

We don't need to check the element for null. Take a look at the implementation of get_known_connected_element - it only returns non-null pointers, and already returns a NoSuchElement if the element was null.

(It returns a pointer instead of a reference because ErrorOr<> cannot hold references. We could update it to return NonnullGCPtr instead, but that's a separate issue.)

So, this is just:

auto* element = TRY(get_known_connected_element(element_id));

// 1. If element's innerHTML IDL attribute is an empty string do nothing and return.
if (TRY(element->inner_html()).is_empty())
    return JsonValue {};

(Not sure TRY will work exactly here, since inner_html return a DOM exception, but that's the gist of it).

[trflynn89]

This isn't really how this should be implemented. The steps starting here are "to clear a content editable element" and seem like they're meant to be implemented as a subroutine, which we can do with a lambda. So something like:

auto clear_content_editable_element = [&](auto& element) {
    // 1. If element's innerHTML IDL attribute is an empty string do nothing and return.
    if (TRY(element->inner_html()).is_empty())
        return JsonValue {};

    // and so on...
};

Then, the implementation of element_clear really begins at "the remote end" steps, invoking this subroutine as appropriate in step 11.

[hanpham32]

Hi! Thanks for the code review. It makes more sense to make clear_content_editable_element as a lambda function :~) I've incorporated your reviews into the new push