Support recipe schema.org metadata from CAPI in LinkedData

[cemms1]

Co-authored-by: frederickobrien frederick.obrien@guardian.co.uk

Recipe structured data

Part of guardian/dotcom-rendering#10532, this is the last in a chain of PRs that will get Schema.org markup rendering on the website for recipes.

A schemaOrg field has been added to the content field of CAPI responses (see guardian/content-api-models#237 and https://github.com/guardian/content-api/pull/2858 for more context). What arrives at Frontend is mostly valid schema, though through a quirk of Thrift field naming '_atType' and '_atContext' need to be turned into '@type' and '@context'.

Once this is done the schema is converted to a LinkedData object and appended to the more generic article schema (or linked data as it's labelled here) already being generated and is passed on to Dotcom as before. Testing has been added to ensure data is being transformed as expected.

One upside of this approach is that we don't have to touch Dotcom, though an open question remains about where is the most appropriate place to transform CAPI data into Schema.org markup. Frontend? CAPI itself? There's a bit of both now. With time - and hopefully the SEO value of this established - it might be worth centralising the Schema.org/linked open data generation.

Most of the changed files - in data/database - are due to an unrelated issue concerning (non)generated data files.

To test deploy this branch to CODE and use https://validator.schema.org/ to check whether valid schema is being included in the ld+json blobs of articles containing recipe elements. #26931 (comment) contains a few example though there are hundreds by now.

[cemms1]

Update of things still outstanding:

• Update the CAPI models and CAPI scala client snapshot versions to include the recent work on sponsored packages
• Run dev-build locally, ensuring the .gu/frontend.conf file has entries for CAPI CODE environment (content.api.host ,content.api.key) a
• If testing using debug mode, make sure to set the --debug flag when running the ./sbt command and set breakpoints in LinkedData for the recipe case in the forArticle method.
• Try to access the CODE article http://localhost:9000/food/2024/feb/03/yotam-ottolenghi-alternative-pancake-day-recipes-rice-egg-harissa-cheese-crempogau (there's a basic test article without recipe schema here if needed

Ideally we'd also get a test file working to check the logic for returning recipe schema org data.

A possible improvement at this point might be to decode and encode the JSON object from schemaOrg.recipe rather than to stringify it and parse again.

[cemms1]

Where we've got to:

• We can recognise when an article has a Some schemaOrg.recipe field
• We need to convert every _atType and every atContext within the SchemaRecipe data structure to @type and @context because of a limitation with thrift not allowing the @ symbol.
  • One way of doing this might be to use Play Json but in order to do this we will need to fully define our output type (?) in order to define the Json.reads and Json.writes functions. This is not ideal due to wanting to pass the entire model response straight through, only changing these two fields throughout the object.
  • It would be useful to understand if this is possible from any Scala experts before we take too much time trying various methods in order to achieve this
  • If it will take too much effort/time or require a high level of schema definition within the frontend repo it would be worth evaluating whether CAPI is really the right place for this data, having frontend as the only real "client" of this

[github-actions]

Deploy build 1364 of dotcom:frontend-all to CODE

All deployment options

• Deploy build 1364 of dotcom:frontend-all to CODE
• Deploy parts of build 1364 to CODE by previewing it first
• What's on CODE right now?

---

From guardian/actions-riff-raff.

[frederickobrien]

Testing this in CODE and it's looking really good, valid schema coming through for a selection of single and multi-recipe articles:

• https://m.code.dev-theguardian.com/food/2021/sep/29/how-to-make-tiramisu-recipe-masterclass-felicity-cloake
• https://m.code.dev-theguardian.com/food/2024/mar/09/vegan-peanut-butter-ramen-recipe-meera-sodha
• https://m.code.dev-theguardian.com/food/2024/jan/27/30-minute-meal-yotam-ottolenghi-recipes-thai-deep-fried-omelette-tofu-broccoli-spicy-seaweed-sea-bass-spaghetti

Using https://validator.schema.org/ to confirm the ld+json blob is valid. No errors or warnings found as yet

[prout-bot]

Seen on ADMIN-PROD (merged by @cemms1 12 minutes and 13 seconds ago)

• Check your changes on www.theguardian.com ✔️
• Keep an eye on the deploy dashboard 📉

[prout-bot]

Seen on FRONTS-PROD (merged by @cemms1 14 minutes and 23 seconds ago)

• Check your changes on www.theguardian.com ✔️
• Keep an eye on the deploy dashboard 📉

[rtyley]

I'm not sure if this is important, but I wanted to share that I received this warning this morning from Google Search Console ("New Recipes structured data issues detected"):

Top critical issues* : Missing field 'name'

It looks like this is actually only on one url:

https://www.theguardian.com/food/2021/mar/21/joe-trivelli-recipes-asparagus-and-potato-schiacciata-cherry-ice-cream-sandwiches