add SPARQL endpoint and web interface

[KonradHoeffner]

What new feature do you want?

Now that @GregHanson added SPARQL support, it may be worth adding the additional functionality to publish an HDT file over a SPARQL endpoint and web interface. However as that is not the primary focus of the HDT library this should only be done if it is possible by simple means of reusing existing components.

Investigate what is available.

[KonradHoeffner]

HTML Interfaces

• Yasgui https://github.com/TriplyDB/Yasgui https://www.semantic-web-journal.net/system/files/swj1126.pdf last release from 2020, TypeScript

[KonradHoeffner]

SPARQL endpoints / REST APIs

• RAMOSE https://www.semantic-web-journal.net/system/files/swj2543.pdf Python
• Oxigraph (it has it's own database but can we reuse components of it?)

[KonradHoeffner]

@GregHanson: I remember you writing about your open-source SPARQL interface solution in the process of being published, is there something you can share yet?

[GregHanson]

Sorry for the delay @KonradHoeffner, holiday in the US last week. Delays on our end regarding the approval process. In the meantime, I'm trying to see if I can just drop the HDT implementation into the Store/Storage trait in upstream oxigraph so it can be plugged directly into the CLI in oxigraph in the serve read only command. Unfortunately the QueryableDataset trait doesn't appear to overlap, so it's taking a while to get the implementation working. My initial findings are that none of the QueryableDataset trait methods overlap so a lot of things are being duplicated

[KonradHoeffner]

@GregHanson no problem with the delays! It's just a side project for me so I can just fit it in whenever there is time so there is no rush. That would be great if you can plug it in directly.

[GregHanson]

@KonradHoeffner I have a working version of HDT integrated into oxigraph's CLI - which provides access to query and serve-read-only subcommands: oxigraph/oxigraph#1437

If you read the comments, it works for demo purposes but it will never be as optimized for querying as what's in hdt crate's QueryableDataset implementation since the underlying Store struct does not leverage any part of QueryableDataset