core(crawlable-anchors): ignore event listeners, validate url

[svkrclg]

Summary

Related Issues/PRs

fixes #11818

[svkrclg]

Not sure whether ignoring onclick behaviour completely is a good idea. Let me know.
TODO: fix smoke test

[connorjclark]

Not sure whether ignoring onclick behaviour completely is a good idea. Let me know.

Given we confirmed that search will only look at href, we're ok with ignoring click events entirely since this audit is only concerned about search engine crawlability.

---

Currently the audit does these things:

1. Ignore links with role set
2. Fail if href starts with file:
3. Fail if href is an empty string
4. Ignore links with a click event listener
5. Fail if onclick has javascript:void(0)
6. Fail if onclick has window.location/open

We want to drop items 4-6. Currently the PR drops 5-6 but not 4. Then you'll need to delete some tests which will newly fail.

---

Besides that, we can do a bit more to improve the audit (although if you don't want to tackle this part, we can do it in a new PR). As it is, this would pass:

<a href="http:://example.com">link</a>

Even though the URL is invalid (and thus definitely not crawlable). This is because we don't check href for anything but that it is non-empty and not file:. This should do the trick:

try {
	new URL(rawHref, URL.finalUrl); // URL can gotten from the first parameter of the `audit` function
} catch {
  return true;
}

[connorjclark]

can delete this too

[connorjclark]

Looking good!

For the smoke test, you can just flip these test cases from expecting "passing" to "failure":

1. Change PASS->FAIL in html https://github.com/GoogleChrome/lighthouse/blob/master/lighthouse-cli/test/fixtures/seo/seo-tester.html#L75 (doesn't do anything, just documentation)
2. Change expectation from pass to fail for this audit in seo-passing smoke

   lighthouse/lighthouse-cli/test/smokehouse/test-definitions/seo/expectations.js

   Line 237 in 9387e3a

   'crawlable-anchors': {

'crawlable-anchors': {
        score: 0,
        details: {
          items: {
            length: 2,
          },
        },
      }

yarn smoke seo-passing to confirm

[connorjclark]

Thanks for the assist!

[svkrclg]

Thanks for helping out @connorjclark .