Truncate description field in injected JSON-LD to 5000 characters

[amoeba]

@mbjones noted a few of our datasets on arcticdata.io were failing Google's schema:Dataset validation due to "Invalid string length in field "description". When we go into the Search Console, we can see that Google is seeing the description field. We then noticed that Google specifies that the description field should be between 50 and 5000 characters:

The summary must be between 50 and 5000 characters long.

From: https://developers.google.com/search/docs/advanced/structured-data/dataset#dataset

In at least one of the offending documents, the abstract is 5779 characters so the above rule seems like a likely cause for the failed validation. Let's truncate the description field to 5000 characters for all records and hopefully these errors will go away.

[amoeba]

As mentioned in DataONEorg/dataone#12 (comment), we should also pad descriptions that are too short. Something like "For complete metadata, visit..." would do the trick.

[robyngit]

Right now, the logic for managing JSON-LD is spread across multiple views:

• CatalogSearchView: injects JSON-LD for the data catalog
  • removes jsonld during the onClose method. Identifies with the id jsonld.
• AppView: In the render function, injects JSON-LD for the application itself based on a template.
• DataCatalogView (deprecated?): The method getLinkedData generates JSON-LD for the catalog but doesn’t seem to be actively used.
• MetadataView: The insertJSONLD method injects JSON-LD for datasets or metadata records. generateJSONLD generates it.
• router.js: removes jsonld (identifies with the id jsonld) when initializing.

If we want to add the general functionality to truncate or pad descriptions, we would need to add this logic in multiples places. So the first step is to consolidate the JSON-LD logic into a single model and view. Then we can add the functionality to truncate or pad descriptions in a single place.