[SEO Audits] Structured data is valid

Audit group: Content Best Practices Description: JSON-LD structured data syntax is valid Failure description: JSON-LD structured data syntax is invalid Help text: Structured data contains rich metadata about a web page. The data may be used to provide more context to users in search results or social sharing. This audit is only checking for JSON-LD structured data syntax errors. For a checkup of Google search eligibility, try the Structured Data Testing Tool.

Success conditions:

• the JSON can be parsed
• the keys are not duplicated
• @ keywords are whitelisted in the JSON-LD spec
• @context URLs are valid
• validation using jsonld.js
• @types are whitelisted on schema.org
• object properties are whitelisted on schema.org

Notes:

• defers Google search eligibility and other more advanced validation logic to the Structured Data Testing Tool (SDTT)
• implementation logic is built into Lighthouse and not open sourced separately

[mattzeunert]

A few observations after running this on a few sites:

1. Case sensitivity

This is a failure because there’s no aggregateRating type, just a Thing.aggregateRating property. The type should be AggregateRating. It might not be clear to the user though, because the schema.org URL exists.

JSON-LD is case sensitive, but the Structured Data Testing Tool doesn’t complain about it.

2. Empty elements

There’s a few JSON-LD elements with no content or just {}. SDTT doesn’t care, and it probably just means there’s no data, so nothing we need to do I think.

3. Parser errors

I’ve only seen one example of this, but if you try to escape a single quote in the JSON our error message isn’t super helpful. (JSON only allows a few escape sequences.)

You get the same message if you have line breaks in your strings or a trailing comma somewhere. Maybe prefixing the message with "Parse error: " would make it clearer what the problem is.

[danbri]

operating over the parsed graph (set of node-edge-node triples) from json-ld would establish a model well suited to allow dropping in rdf, microdata parsers. This is what we do at Google and what other public tools do.

On Wed, 21 Aug 2019, 19:52 Jarno van Driel, [e-mail] wrote:

I’m happy to read about this initiative though I’m wondering whether microdata and RDFa have been considered as well?

Reason I ask is because even though JSON-LD is the ‘preferred’ syntax right now there’s still tons of microdata markup out there (and RDFa in lesser quantity as well) that’s actively maintained. It would be a real shame if these fall out of scope.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/GoogleChrome/lighthouse/issues/4359?email_source=notifications&email_token=AABJSGLN42PZCGQMV2PJEGDQFWFGLA5CNFSM4ENWC24KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD42YJXA#issuecomment-523601116, or mute the thread https://github.com/notifications/unsubscribe-auth/AABJSGLUBD3W6QWKBI6SN3LQFWFGLANCNFSM4ENWC24A .