SPARQL-star: Add embedded triple pattern to PrimaryExpression, define evaluation.

[afs]

Addresses issue #163.

---

Preview | Diff

[Tpt]

I am not sure it is possible to reuse the EmbTP rule here without restrictions: EmbTP allows blank nodes that are not allowed in SPARQL expressions. What about introducing a new rule:

EmbTPExpr ::= '<<' PrimaryExpression PrimaryExpression PrimaryExpression '>>'

and consider it an alternative writing of the TRIPLE function?

[afs]

Thanks for pointing that out.

PR set to "draft".

The idea is that it should be like a term in a graph pattern, so it is not full PrimaryExpression which allows ( expression ) and functions and built-ins.

I think it is going to have to be written out as a specific form of vars or constants, and also a for the predicate slot. (And so avoids going near related matching forms SPARQL 1.1 does not have ?str@en and ?lex^^:datatype.)

For evaluation definition, TRIPLE can be used as it covers all the cases (and more).

[afs]

Pushed an update for the grammar.

ExprVarOrTerm can be written using 177/DataValueTerm but that is currently specifically for VALUES or using iriOrVar but as rule names have implications so it seemed clearer to enumerate the cases.

It will need some stylistic reorganisation to fit with the grammar section. Rules119a and 119b are placed close to use for clarity for now but style is to group new productions at the end.

[afs]

This PR now has:

• Grammar rules for PrimaryExpression involving <<>>.
• Two new bad syntax tests for `<<>> in expressions (the "good" case was already covered).
• Definition of evaluation of <<>> in an expression context (added new section 4.4.6). It is defined as being the same as TRIPLE with the expressions allowed by the syntax for <<>>.

[afs]

@pchampin This now adds a section 4.4.6 for evaluation of << >> in expressions.