Ideas for Yasgui to help SPARQL-Anything query development

[MathiasVDA]

In #594 the idea was raised for a sparql-anything playground or query development environment. This issue exists to capture ideas for this environment so that we can work on them in Yasgui or elsewhere.

[MathiasVDA]

Idea 1: Autocomplete source file structure after adding the SERVICE call

When writing SERVICE <x-sparql-anything:location=/path/to/file.csv>{ an autocompleter would automatically add:

?row xyz:column_name1 ?column_name1;
         xyz:column_name2 ?column_name2;
...

Based on the source file. Advantage is that the developer will not have to look up the source file structure and will decrease the chance that structural errors would be made.

This should be pretty easy to do for CSV, might be a bit more tricky for XML/JSON but should still be feasible.

[luigi-asprino]

The query developed in this tutorial could be used to get an overview of the source and suggest possible triple patterns.

[luigi-asprino]

The challenge lies in obtaining a comprehensive overview without tripling the entire source. This is trivial for CSV input, but I'm not sure what can be done with the other formats. By the way, I would avoid having different strategies for each format.

[MathiasVDA]

Thanks for the reference to the tutorial.

By the way, I would avoid having different strategies for each format.

That would be very nice. Since it's all triples, I'm guessing that it should be possible to generalise the result of such a 'format independent structure retrieval query'.

EDIT: I would also add a LIMIT 1 (number configurable by the user maybe?) to this structure retrieval query to avoid triplifying the entire source.

[justin2004]

Advantage is that the developer will not have to look up the source file structure and will decrease the chance that structural errors would be made.

could address this too:
#347

which would make former TARQL users happy

[VladimirAlexiev]

I usually start with "echo.rq" that dumps "?s ?p ?o" and then massage it to the right query.
So I'd welcome such a Feature: -)

The challenge with "limit 1" is with nested structures. Consider a XML or HTML document. It has acsingke root, so "limit 1" won't buy anything.

Or consider a csv dump from Neo4j. It starts with say 10k nodes, followed by say 15k relations: no CSV fields in common.
Even for nodes alone, the CSV structure is very sparse: all props across nodes are collected in say 200 cols, and a given node has 10 of them. Each class (label) has different 10.
(I'm messing exactly with such a dump at present)

[MathiasVDA]

Idea 2: validate xyz: predicates

After typing xyz:my_header, the code editor should check if this predicate exists in the source file.

[luigi-asprino]

This relates to #434.

[luigi-asprino]

Once the input is in FX-RDF, predicates used in the source can be easily retrieved (SELECT DISTINCT ?p {...}). However, this would require the entire source to be transformed beforehand.

[MathiasVDA]

Ah I'm glad someone else also mistypes things from time to time :). I would implement this in yasgui with some background query that would retrieve the valid xyz: predicates. There is already a functionality to highlight invalid lines in the code editor, so I would reuse that to highlight the problem.

[MathiasVDA]

Idea 3: auto execute valid sparql queries

Add the possibility to automatically execute SPARQL queries after adding a new part and if the result is valid.

Add a default LIMIT 10 ?

[luigi-asprino]

Could you please provide more information on this idea?

[MathiasVDA]

I got this idea while looking at the playground of sparql generate. There is this 'run automatically' checkbox:

So this idea would correspond to implementing this feature into Yasgui

I'm not sure if I would use this a lot, since I have the habit of throwing in 'CONTROL+ENTER' quite a lot.