Report for visualization of descriptive metadata that distinguishes properties with a "value"

[ndushay]

As part of #4493, we need a report akin to "descriptive_shapes" from https://github.com/sul-dlss/dor-services-app/blob/main/app/reports/descriptive_shape.rb, which will feed visualizations in sul-dlss-labs/cocina-shapes.

The deltas needed

• only count data when the cocina DescriptiveValue (or equivalent) has a value. For example, we don't care about the authority being LCSH if there is no actual value in a subject.
• two kinds of counts needed:
  • count once per object if there is one or more value
  • count all occurrences with values per object

we still want the desire to select for ils catalog links present or absent, as is in the descriptive shapes report.

Refs:

• https://consul.stanford.edu/display/DIGMETADATA/Complete+guide+to+Cocina+properties+and+spreadsheet+headers
• https://github.com/sul-dlss/cocina-models/blob/main/openapi.yml
• https://sul-dlss.github.io/cocina-models

Spec:

• 2 separate types of counts to track:

  • presence of one or more values in an object
  • number of occurrences with values in objects

  if object A has 2 subjects (with value) and object B has 3 subjects (with value):

  • presence of one or more subjects per object: 2: 1 for object A and 1 for object B
  • number of occurrences of subject: 5: 2 for object A and 3 for object B

How to determine if a cocina descriptive property has a value:

DescriptiveBasicValue

• https://github.com/sul-dlss/cocina-models/blob/main/lib/cocina/models/descriptive_basic_value.rb
• Considered to have a value if one of the following is true:
  • value property has a string or integer non-blank value
  • uri property has a non-blank value
  • valueAt property has a non-blank value
  • structuredValue, parallelValue or groupedValue child or descendent has a value (value, uri, valueAt) but there can be nothing but these three in the path

QUESTION: Note that code property with a value is only ok if we know the source -- punting for now. Note that these are all defined as type DescriptiveValue in the openapi doc

DescriptiveValue

• https://github.com/sul-dlss/cocina-models/blob/main/lib/cocina/models/descriptive_value.rb

DescriptiveBasicValue with appliesTo property added

Treat as DescriptiveBasicValue

Cocina Description Top Level Values

Title

Treat as DescriptiveBasicValue

Contributor

• https://github.com/sul-dlss/cocina-models/blob/main/lib/cocina/models/contributor.rb

only count value if direct children properties of name, note, or identifier have value per DescriptiveBasicValue

Event

• https://github.com/sul-dlss/cocina-models/blob/main/lib/cocina/models/event.rb

only count value if

• direct children properties of date, contributor, location, identifier, note have value per DescriptiveBasicValue
• direct child parallelEvent as above, e.g. parallelEvent[].date[].value

Form

Treat as DescriptiveBasicValue

Language

• https://github.com/sul-dlss/cocina-models/blob/main/lib/cocina/models/language.rb

as DescriptiveBasicValue with script property added:

• add language[].script.value as a value (parallelValue, structuredValue, groupedValue etc. can be in path after script)

Note

Treat as DescriptiveBasicValue

Identifier

Treat as DescriptiveBasicValue

Subject

Treat as DescriptiveBasicValue

Geographic

• https://github.com/sul-dlss/cocina-models/blob/main/lib/cocina/models/descriptive_geographic_metadata.rb

has only form and subject properties as direct children -- treat these as you would top level subject or form

Purl

It's a string. count if value is not blank

Access

• https://github.com/sul-dlss/cocina-models/blob/main/lib/cocina/models/descriptive_access_metadata.rb

Treat all immediate children (url, physicalLocation, digitalLocation, accessContact, digitalRepository, note) as DescriptiveBasicValue

RelatedResource

• https://github.com/sul-dlss/cocina-models/blob/main/lib/cocina/models/related_resource.rb

count it for any "top level" properties under relatedResource e.g. relatedResource[].title[].structuredValue[].value - counts as a value per recursion but relatedResource[].status - does NOT count as a value

AdminMetadata

• https://github.com/sul-dlss/cocina-models/blob/main/lib/cocina/models/descriptive_admin_metadata.rb

follow subproperties (note, event, contributor, identifier, language, metadataStandard) and use rules of indicated value, e.g.: adminMetadata.note[].value - count as value if present, but also add valueAt as a valid value

SUB-PROPERTIES

These should only be counted if the parent has a value per above

Standard

• https://github.com/sul-dlss/cocina-models/blob/main/lib/cocina/models/standard.rb has properties:
• code <-- count as value if source has a value
• uri <-- count as value
• value <-- count as value
• note
• version
• source

ValueLanguage

• https://github.com/sul-dlss/cocina-models/blob/main/lib/cocina/models/descriptive_value_language.rb

as Standard plus valueScript property, which should also count as a value

Encoding

exactly equivalent to Standard; treat same as Standard

Source

as Standard, but without source property ... so code cannot count

MarcEncodedData

• only present for RequestDescription Can ignore because we're going from Marc to MODS to Cocina, not directly Marc to Cocina

If we weren't ignoring: treat as DescriptiveValue

---

DECISION: the below is probably more human hours than it gains in computing hours

paths we can SKIP when looking for presence of value per object (these are useful only if there is a value:

• do not count for presence of value
• count occurences only when a value is present
  • encoding
  • source
  • standard
  • valueLanguage
  • valueScript

[ndushay]

So, I have a DRO that has 3 top level contributors with values. Am I counting ALL subvalues for each contributor with a value, for the report that counts all occurrences? And what am I counting for “presence of one or more values” ? Just the top level contributor fields? (How would I count presence or value for sub properties? Use all 3 occurrences with values to see if any of them use a sub property?)

If this spec is this hard ... that seems like a smell. This seems frighteningly complex for basic questions, like "does object A have a contributor in the desc metadata (that is a real value and isn't noise)?"

[ndushay]

@arcadiafalcone: event has a top level structuredValue, in addition to parallelEvent. Should that be allowed somehow?

Also, for event, if there is a top level subproperty of "contributor:" should it be treated as a descriptiveBasicValue, or as a contributor (only care about it if there is a name, note or identifier child property)?