PID minting should be done after checks so it's impossible to end up with a PID minted when checks fail.

Currently we are only checking a minimum cardinality. We should also check maximum/exact ones taking into account they mean maximum/exact per language.

• Check if ObjectProperties have only URI values
• Assure DatatypeProperties don't have literal values

Depends on acdh-oeaw/arche-lib-schema#2

Such resources must have:

• acdh:hasLicense
• exactly one acdh:hasCategory (to be mapped to edm:type)
• acdh:hasNextItem (to be mapped to edm:isShownBy)

according to our ontology e.g. hasCoverageStartDate has range http://www.w3.org/2001/XMLSchema#date but currently doorkeepers accepts dateTime strings as well; either reject them or transform them to proper iso-dates

At the moment the check determines if the content is a full biblatex record or only a set of properties by checking literally if the first character is @. This check fails for content beginning with whitespaces (which is common when it comes from a pretty-formatted XMLs). It should be addressed.

see https://redmine.acdh.oeaw.ac.at/issues/17610

running the script ids_and_binaries.php failed to create handle-PIDs

example resource

<rdf:RDF xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#" xmlns:acdh="https://vocabs.acdh.oeaw.ac.at/schema#" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xml:base="https://id.acdh.oeaw.ac.at/">
  <acdh:Resource rdf:about="https://id.acdh.oeaw.ac.at/rita/editions/inventar__1756-03-15.xml">
      <acdh:isPartOf rdf:resource="https://id.acdh.oeaw.ac.at/rita/editions"/>
      <acdh:hasTitle xml:lang="de">1756 III 15 – 1756 III 29, Saalen (in der Benefiziaten-Behausung) [Stainer Sebastian Joseph (Benefiziat)]</acdh:hasTitle>
      <acdh:hasNonLinkedIdentifier xml:lang="de">Archivsignatur: Südtiroler Landesarchiv (SLA) Inventare des Mittleren Pustertals Pos. Nr. 760 </acdh:hasNonLinkedIdentifier>
      <acdh:hasCoverageStartDate rdf:datatype="http://www.w3.org/2001/XMLSchema#date">1756-03-15</acdh:hasCoverageStartDate>
      <acdh:hasCoverageEndDate rdf:datatype="http://www.w3.org/2001/XMLSchema#date">1756-03-15</acdh:hasCoverageEndDate>
      <acdh:hasRelatedDiscipline rdf:resource="https://vocabs.acdh.oeaw.ac.at/oefosdisciplines/601"/>
      <acdh:hasMetadataCreator rdf:resource="https://d-nb.info/gnd/1043833846"/>
      <acdh:hasCurator rdf:resource="https://d-nb.info/gnd/1043833846"/>
      <acdh:hasContact rdf:resource="https://d-nb.info/gnd/1133094783"/>
      <acdh:hasDepositor rdf:resource="https://d-nb.info/gnd/1133094783"/>
      <acdh:hasPid>create</acdh:hasPid>
      <acdh:hasLicensor rdf:resource="https://d-nb.info/gnd/16332395-1"/>
      <acdh:hasLicense rdf:resource="https://vocabs.acdh.oeaw.ac.at/archelicenses/cc-by-4-0"/>
      <acdh:hasOaiSet rdf:resource="https://vocabs.acdh.oeaw.ac.at/archeoaisets/kulturpool"/>
      <acdh:hasOaiSet rdf:resource="https://vocabs.acdh.oeaw.ac.at/archeoaisets/clarin-vlo"/>
      <acdh:hasLanguage rdf:resource="https://vocabs.acdh.oeaw.ac.at/iso6393/deu"/>
      <acdh:hasOwner rdf:resource="https://d-nb.info/gnd/16332395-1"/>
      <acdh:hasRightsHolder rdf:resource="https://d-nb.info/gnd/36165-3"/>
      <acdh:hasCategory rdf:resource="https://vocabs.acdh.oeaw.ac.at/archecategory/text/tei"/>
      <acdh:hasCreator rdf:resource="https://d-nb.info/gnd/1133094783"/>
      <acdh:hasAvailableDate rdf:datatype="http://www.w3.org/2001/XMLSchema#date">2021-04-01+02:00</acdh:hasAvailableDate>
  </acdh:Resource>
</rdf:RDF>

same data source as mentioned in the script can be found here https://rita.acdh.oeaw.ac.at/archeutils/ids.xql

For all classes new acdh:hasTitle values should only overwrite existing values in the same language. Existing values in other languages should be kept (it's not a case now due to default RDF merging rules we use).

This change will allow us to express domains (for properties, e.g. for default values as well for cardinality rules) in terms of a resource having a binary payload.

With the new approach to the named entities handling we will by design leave "empty" resources in the repository. They should be accepted until they were covered by the checkRanges check. Unfortunately it's not easy to tell because their class is derived from the property relating to them and the namespaces check is regex-based so a precise at-the-end-of-transaction check would need to be pretty complex. Still we can't just release the check because we currently allow creation of "empty" resources for properties which range isn't defined in the checkRanges (or which are out of ontology).

It looks like a metadata as follows:

<sbj> <objectProperty> <non-standardized-URI> .

passes the doorkeeper check thanks to the URI standarization but still creates an ARCHE resource with the original non-standarized URI as an identifier.

Experienced by @csae8092 while ingesting arche_ttl.txt

If a resource belongs to the $cfg.doorkeeper.epicPid.clarinSet OAI-PMH set and PID generation is turned on, a separate PID should be generated for the CMDI record and stored in the $cfg.schema.cmdiPid. The PID should point to the additional ID created as {$cfg.schema.namespaces.id}/cmdi/{resource internal integer ID}.

Make use of arche-lib-schema PropertyDesc->defaultValue

acdh-oeaw/arche-schema#35
acdh-oeaw/arche-ref-sources#13

The main challenge here is the performance. We have a few big vocablaries (e.g. oefos and iso languages) and we don't want reading them to kill the doorkeeper performance (which is not perfect even now).

see https://redmine.acdh.oeaw.ac.at/issues/17612

acdh:isNewVersionOf can only form tree structure (there can be no more than one triple pointing to a given resource with acdh:isNewVersionOf)

see https://redmine.acdh.oeaw.ac.at/issues/17607