Skip to content

Conversation

@cormachallinanderilinx
Copy link

No description provided.

Markus92 and others added 30 commits December 2, 2024 19:26
Rdflib 7.1 introduced changes in [date parsing][1] that made some base
profile tests fail. Basically the previous rdflib versions incomplete
dates like

    <time:inXSDDateTimeStamp
         rdf:datatype="http://www.w3.org/2001/XMLSchema#date">
            1904
    </time:inXSDDateTimeStamp>

were expanded to `1904-01-01`. Of course this is not a valid date and
should be expressed using `gYear`:

    <time:inXSDDateTimeStamp
         rdf:datatype="http://www.w3.org/2001/XMLSchema#gYear">
            1904
    </time:inXSDDateTimeStamp>

and we should be expecting `1904`.
This should play nice with the time properties we are generating in CKAN as
they already handle automatically `gYear`, `gYearMonth`, `date` and
`dateTime`.
Sites importing external DCAT representations that use the wrong
encoding might need to check their parsers.

[1] RDFLib/rdflib#2929
Legacy `spatial_*` extras were not converted to the repeating subfield
`spatial_coverage`
Fix parsing of spatial properties when using scheming
Co-authored-by: Adrià Mercader <amercadero@gmail.com>
Co-authored-by: Adrià Mercader <amercadero@gmail.com>
amercader and others added 30 commits October 14, 2025 12:50
…-object

Fix AttributeError to be compatible with CKAN2.12
the MQA _MIGHT_ be sensitive to the exact format, even though it's legal
* added rdf.type class to _add_*_triple
* add codelist infra for HVD category
* refactor graph from resource
* R/O profile for catalog generation
* shaql shapes for HVD validation
This doesn't appear to be required by the BReg spec itsself,
and should be, as far as I can tell, a part of DCAT-AP.

That said, the BReg SHACL wants `dct:Distribution->dct:format`
to have `dct:MediaTypeOrExtent` as an explicitly registered class.
As with the previous commit, this doesn't appear to be a new constraint
in BReg, but the SHACL does validate it.

Specifically, the BReg SHACL wants `dct:LinguisticSystem` to be an
explicit class on the language entry, and it needed a url for that to
work well.
At present, this only handles dataset fields, and some associated classes.
In particular, no attempt is made at handling data service fields.
While this is required by the BReg spec, as the `dct:type` field is made
mandatory on `eli:LegalResource` therein, no field that we're handling as
part of BReg requires an `eli:LegalResource`, so I haven't put it in the
BReg profile.

The list of included ELIs will likely need to change.
This is structured as it is to avoid `group_show`ing a group more than once,
as would be necessary if the profiles were responsible for handling it
(as with resources).
Empty strings appear as fields when updated via the UI
"" being a `skos:Concept` is a shacl validation error
`dct:identifier` is required to be a `rdf:Literal`
This isn't entirely correct, as some groups aren't data services.
This is perhaps not the best place to put it, but for things like:

Contact_point:
  ...
  address:
    - address1
      - country
      - address
    - address 2
      - country

the dictionary for address was triggering a solr error.

de-524
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.