Templated strings

[scauligi]

• Closes 3.14 Template Strings #2687

Python 3.14 introduced t-strings as a new literal type.

This PR allows parsing t-strings, as well as introduces a new pragma bracketed-templates to additionally allow bracketed t-strings (a la bracketed f-strings).

[Kodiologist]

Looks like you gotta gate the tests to Python 3.14.

[Kodiologist]

I don't think you need pytest_ignore_collect for this. In an analogous way that I guard some files with do-mac, you can prefix your file with a reader macro like so:

(defreader r []
  (if hy.compat.PY3-14
    `(do ~@(.parse-forms-until &reader ""))
    (for [_ (.chars &reader True)])))

#r

[scauligi]

you can prefix your file with a reader macro like so

oh very cute, I like that

[Kodiologist]

I force-pushed to group some logically related commits together.

I think you might be gating for Python 3.14 in Hy itself a bit too aggressively. There's no reason that the Hy syntax t"foo" needs to be illegal in Python 3.13, for example. The Hy-level syntax and models for t-strings don't require support from Python. The only place we should need to check the Python version is in the compiler functions, when the user tries to compile a Hy t-string into a Python-level t-string. As a bonus, you'll no longer need a reader macro to gate the tests, and some tests won't need to be gated.

By the way, does it make sense for t-strings to by represented as hy.models.FStrings, instead of getting their own model type? That seems confusing.

It looks like hy-repr for t-strings isn't tested.

I think (= (type t1) (type t2) Template) would be better written with is instead of =.

Finally, of course, we'll need documentation.