Include the validation of templates in the SHACL shapes

[chrdebru]

I believe that checking the validity of templates should be included in the shapes. I'm not sure whether SPARQL's regular expressions allow for recursion, but it can be achieved by:

• removing the unescaped curly braces from the template
• checking whether the resulting string matches ^[^\{\}]*(?:\{[^\{\}]+\}[^\{\}]*)*$ (balanced and not nested)

This can be achieved for a SPARQL constraint component.

[DylanVanAssche]

SHACL can do regex, once the listed regex is tested and test cases are provided with valid and invalid templates, we can add it.

[dachafra]

Tasks required:

• Test the regex provided and ensure that the constraint is correct
• Define test-cases with valid templates (I think this is by default now and it's not required)
• Define a few test-cases with invalid templates.

@DylanVanAssche could you take care of this?

[DylanVanAssche]

Define a few test-cases with invalid templates.
AFAIK, these are the 'rules':

A string template is a format string that can be used to build strings from multiple components. It can apply [reference expressions](https://kg-construct.github.io/rml-core/spec/docs/#dfn-reference-expression) by enclosing them in curly braces ({ and }). The following syntax rules apply to valid [string templates](https://kg-construct.github.io/rml-core/spec/docs/#dfn-string-template):

    1. Pairs of unescaped curly braces MUST enclose valid [reference expressions](https://kg-construct.github.io/rml-core/spec/docs/#dfn-reference-expression).
   2. Curly braces that do not enclose [reference expressions](https://kg-construct.github.io/rml-core/spec/docs/#dfn-reference-expression) MUST be escaped by a backslash character (\). This also applies to curly braces within [reference expressions](https://kg-construct.github.io/rml-core/spec/docs/#dfn-reference-expression).
    3.Backslash characters (\) MUST be escaped by preceding them with another backslash character, yielding (\\). This also applies to backslashes within [reference expressions](https://kg-construct.github.io/rml-core/spec/docs/#dfn-reference-expression).
    4.There SHOULD be at least one pair of unescaped curly braces.

from spec.

Anything missing?
If not, we need to have 4 invalid test cases, one for each rule.

[dachafra]

I don't think so, please proceed :-)

[DylanVanAssche]

Escaping is tricky, especially since we use the same escape character in reference formulations as Turtle. If we write in something else then Turtle, we only need to escape once, other 2x.

In non-Turtle you can write \{ and escape it
In Turtle you must write \\{ to escape it.

However in 2 you might be confused if the \\ is escaping the { or means \ as token. Spoiler alert: you need to do \\\ in Turtle for \ in a reference formulation.

[chrdebru]

Is escaping really that tricky? You just need to load the file into a graph and indicate the format, no? Only then do we need to test the regex. That said, implementers can rely on ill-founded assumptions, so the test cases should provide redundant ones for RDF/XML, N-Quads, and even JSON-LD.

[DylanVanAssche]

Is escaping really that tricky? You just need to load the file into a graph and indicate the format, no?

If you use tools, they take care of that. It is mostly tricky if you write mappings manually. This is purely a developer experience thing. Since Turtle and the template-syntax use the same escape character / it becomes tricky when writing by hand.