schemaorg.py duplicates maintainer contact point for both publisher and creator and ignores author in metadata

[tazchalmers]

I have found a small issue with schemaorg.py which lists the contact point of maintainer twice for both publisher and creator and ignores author in the resulting metadata.

I verified the field_names in the dataset_dict populated from the ckan_dataset.yaml default schema are:

• maintainer
• maintainer_email
• author
• author_email

However, in the graph_from_dataset method, "publisher" and "creator" are used as schema_property_prefix arguments to _agent_graph even though they do not appear as field_names in the ckan_dataset.yaml schema:

# Publisher
self._agent_graph(dataset_ref, dataset_dict, SCHEMA.publisher, "publisher")

# Creator
self._agent_graph(dataset_ref, dataset_dict, SCHEMA.creator, "creator")

I'm not a developer, so I know this is not a good solution, but for now for my use case, I have tweaked the following in the _agent_graph method which means both author and maintainer are now listed in the metadata.

if schema_property_prefix == "creator":
    name_key = "author"
    email_key = "author_email"
elif schema_property_prefix == "publisher":
    name_key = "maintainer"
    email_key = "maintainer_email"
else:
    name_key = f"{schema_property_prefix}_name"
    email_key = f"{schema_property_prefix}_email"

Results in:

I am well aware it's really poor form to directly edit a core file so I want my fix to be temporary. Is it possible to solve this in the right way?

[amercader]

@tazchalmers The semantic meaning of CKAN's default schema maintainer and author does not align 100% with Schema.org's publisher and creator and while maintainer_* and author_* fields are used as fallback (with the limitation you describe that just the first one found in used in both agents) if you really want to provide accurate descriptions is best to add the publisher and/or creator fields to your custom schema (ckan_dataset.yaml). You can copy them from the examples schemas provided in this extension (picking up just the subfields you are interested in):

ckanext-dcat/ckanext/dcat/schemas/dcat_ap_full.yaml

Lines 55 to 113 in f50a8e2

	- field_name: publisher
	label: Publisher
	repeating_label: Publisher
	repeating_once: true
	repeating_subfields:
	- field_name: uri
	label: URI
	- field_name: name
	label: Name
	- field_name: email
	label: Email
	display_snippet: email.html
	- field_name: url
	label: URL
	display_snippet: link.html
	- field_name: type
	label: Type
	- field_name: identifier
	label: Identifier
	help_text: Unique identifier for the publisher, such as a ROR ID.
	help_text: Entity responsible for making the dataset available.
	- field_name: creator
	label: Creator
	repeating_label: Creator
	repeating_once: true
	repeating_subfields:
	- field_name: uri
	label: URI
	help_text: URI of the creator, if available.
	- field_name: name
	label: Name
	help_text: Name of the entity or person who created the dataset.
	- field_name: email
	label: Email
	display_snippet: email.html
	help_text: Contact email of the creator.
	- field_name: url
	label: URL
	display_snippet: link.html
	help_text: URL for more information about the creator.
	- field_name: type
	label: Type
	help_text: Type of creator (e.g., Organization, Person).
	- field_name: identifier
	label: Identifier
	help_text: Unique identifier for the creator, such as an ORCID or ROR ID.

When using a dedicated publisher or creator field, the schema.org profile will pick up e.g. the publisher.email or creator.email as needed and will display both correctly.

Let me know if this makes sense