Links are not crawlable with acceptable URL

[JPPhoto]

Provide the steps to reproduce

1. Run LH on https://www.cupcakeproject.com/best-vanilla-cupcake-recipe/

What is the current behavior?

I see "Links are not crawlable" for this element:

<a href="https://www.pinterest.com/pin/create/bookmarklet/?url=https%3A%2F%2Fwww.cu…" onclick="window.open(this.href,'targetWindow','toolbar=no,location=no,status=no,men…" data-recipe="21315" style="color: #333333;" class="wprm-recipe-pin wprm-recipe-link wprm-block-text-normal" target="_blank" rel="nofollow noopener noopener noreferrer">

What is the expected behavior?

Since the full link is crawlable, I wouldn't expect to see this error. The link has urlencoded parts, so perhaps that's triggering an issue; opening the URL in a browser shows the expected page.

Environment Information

• Affected Channels: All
• Lighthouse version: 6.3.0
• Chrome version: 87.0.4280.88
• Node.js version:
• Operating System: Windows 10 20H2

Related issues

[patrickhulce]

Thanks for filing @JPPhoto!

I'm really surprised this case of a valid href + onclick=window.open wasn't explicitly covered in #10590 or anywhere in our tests 😐 Conceptually I think this should be fine for the same reasons as #11443 (comment).

cc @AVGP in case this should still be considered failing for a reason I'm not aware of :)

[JPPhoto]

@patrickhulce I should add that the URI is valid, whether or not the onclick JS is invoked. I get that Lighthouse would report an error iff the onclick handler tried to deceive the user; perhaps this is rather hard to detect and it's easier to flag instances as errors? Having not looked at the code, I don't know what's going on under the hood.

[connorjclark]

1- we'd have to click every link and make sure the accompanying network request actually matches the provided href value...

2- we'd have to make sure that search/crawlers even try to resolve links with an onclick... or just ignore onclick and use href ? cc @AVGP any thoughts here?

ideally we could just check for href and ignore all else ... but only if crawlers actually work like that :)

[AVGP]

So as far as Google Search is concerned, look at the href and that's it. Googlebot only has an issue with hrefs like javascript:doStuff('home') or # as none of these are crawlable URLs (well the empty fragment is, but it's just the current page anyway)

[patrickhulce]

thanks @AVGP! We'll mark a valid href + onclick handler as passing then 👍

[svkrclg]

I can pick this up @patrickhulce. Upon looking at the codebase, this is where the anchor is getting marked as failed.
Few doubts here:

• why have we marked window.open as failed anchor tag?
• How can we verify onclick handler as valid?

[patrickhulce]

why have we marked window.open as failed anchor tag?

The purpose of this audit is to flag links that are not crawlable. <a onclick="window.open('http://example.com')"> is not crawlable.

How can we verify onclick handler as valid?

We aren't. This change is about considering an anchor valid if it has a valid href, regardless of onclick behavior.

[paulirish]

fixed by #13105. shipped in LH 8.6.0