Links to TypeIndexes in seed URIs are not used

[RubenEschauzier]

Issue type:

• 🐛 Bug

---

Description:

When a URI is retrieved from the query, it is dereferenced during the optimize stage of ActorQueryProcessSequential. At this stage, the context does not contain KeysQueryOperation.operation. Consequently, the test for ActorExtractLinksSolidTypeIndex fails, preventing it from being invoked for the URI's content.

This becomes relevant when the query URI points to /profile/card#me, as seen in the example below:

PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX snvoc: <https://solidbench.linkeddatafragments.org/www.ldbc.eu/ldbc_socialnet/1.0/vocabulary/>
SELECT ?messageId ?messageCreationDate ?messageContent WHERE {
  ?message snvoc:hasCreator <https://solidbench.linkeddatafragments.org/pods/00000000000000000933/profile/card#me>;
    snvoc:content ?messageContent;
    snvoc:creationDate ?messageCreationDate;
    snvoc:id ?messageId.
  { ?message rdf:type snvoc:Post. }
  UNION
  { ?message rdf:type snvoc:Comment. }
}

As a result, the engine does not extract triples linking to publicTypeIndex or privateTypeIndex, meaning the TypeIndex content is not utilized. Furthermore, when the engine navigates to the /settings/ directory, which contains a link to the TypeIndex via ldp:contains, the TypeIndex file will be dereferenced. However, ActorExtractLinksSolidTypeIndex will not process it, as it requires an explicit link to the TypeIndex file rather than the file itself.

In addition (although this typically shouldn't matter), if the /settings/ URI is ever dereferenced before /profile/card#me, the TypeIndex will be ignored.

[rubensworks]

Very good catch!

One possible solution would be to include KeysQueryOperation.operation in the context during preprocessing as well.
This would not yet be the final query operation, but in this case this should not cause issues (as discussed offline).