Add basic support for TemplateStr

[cdce8p]

Description

Add the basic buildings blocks for ast.TemplateStr and ast.Interpolation nodes. Most of it is pretty similar to the f-string nodes.

The implementation isn't complete but good enough to not outright fail if a t-string is used anywhere in a file. Future PRs should look to address the following

• Additional tests for the as_string() logic. The basics work but there might be a lot of edge case which need to be handled properly. The existing f-string tests could provide a good starting point.
  

  astroid/tests/test_nodes.py

  Lines 295 to 307 in f9e109e

  	def test_f_strings(self):
  	code = r'''
  	a = f"{'a'}"
  	b = f'{{b}}'
  	c = f""" "{'c'}" """
  	d = f'{d!r} {d!s} {d!a}'
  	e = f'{e:.3}'
  	f = f'{f:{x}.{y}}'
  	n = f'\n'
  	everything = f""" " \' \r \t \\ {{ }} {'x' + x!r:a} {["'"]!s:{a}}"""
  	'''
  	ast = abuilder.string_build(code)
  	self.assertEqual(ast.as_string().strip(), code.strip())

• Add proper inference similar to JoinedStr._infer.
  

  astroid/astroid/nodes/node_classes.py

  Lines 4775 to 4781 in f9e109e

  	def _infer(
  	self, context: InferenceContext | None = None, **kwargs: Any
  	) -> Generator[InferenceResult, None, InferenceErrorInfo | None]:
  	if self.values:
  	yield from self._infer_with_values(context)
  	else:
  	yield Const("")

--
https://docs.python.org/3.14/library/ast.html#ast.TemplateStr
Ref #2789

[Pierre-Sassoulas]

Great ! And +1 for basic support now rather than perfect support too late for 3.14 release

[cdce8p]

@Pierre-Sassoulas Should we change the attribute name here? str is the one used by the ast node but it's a bit unfortunate in my opinion since it shadows the builtin str.

Not sure using an alias here would be better though.

https://docs.python.org/3.14/library/ast.html#ast.Interpolation

[Pierre-Sassoulas]

I agree, not ideal, but what should we use instead ? Any non adherence to ast's API is mental load on astroid's user and documentation burden on us.

[cdce8p]

Just wanted to discuss it first. I also think that str while not ideal is still the best option.

[Pierre-Sassoulas]

Before you merged I was writing: Let's wait to read other opinions. Some maintainers might still be ambitious enough to keep astroid a wrapper above the AST and its multiple interpreters :D

It's still possible to change this later if someone want, we're in alpha/betas atm. Being able to test this in pylint fast is good too.

[cdce8p]

Sorry about that. Guess I was too fast this time 😅

Yeah, we can still change it. Furthermore I think the biggest issue is actually in the postinit arg name as that's used on it's own. Accessing the attribute itself shouldn't really be an issue with node.str.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.29%. Comparing base (f9e109e) to head (6a17f53).
⚠️ Report is 97 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2793      +/-   ##
==========================================
+ Coverage   93.26%   93.29%   +0.03%     
==========================================
  Files          91       91              
  Lines       11009    11068      +59     
==========================================
+ Hits        10267    10326      +59     
  Misses        742      742              

Flag	Coverage Δ
linux	93.16% <100.00%> (+0.03%)	⬆️
pypy	93.29% <100.00%> (+0.03%)	⬆️
windows	93.27% <100.00%> (+0.03%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
astroid/__init__.py	100.00% <ø> (ø)
astroid/nodes/__init__.py	100.00% <ø> (ø)
astroid/nodes/as_string.py	96.94% <100.00%> (+0.15%)	⬆️
astroid/nodes/node_classes.py	94.96% <100.00%> (+0.07%)	⬆️
astroid/rebuilder.py	98.21% <100.00%> (+0.03%)	⬆️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Pierre-Sassoulas]

Do you want me to release an astroid beta ?

[cdce8p]

Do you want me to release an astroid beta ?

I still need to debug at least one astroid issue for 3.14. Might make sense to wait a bit.

[cdce8p]

Do you want me to release an astroid beta ?

I still need to debug at least one astroid issue for 3.14. Might make sense to wait a bit.

Looks like what I was seeing is an existing issue with astroid not being able to infer sys.version_info properly.
Even found an old issue for it #113. Maybe that's something I'll look into at some point.

Feel free to do a new beta release if you've some time.

[Pierre-Sassoulas]

I can't find the conversation about it but I remember saying somewhere that this kind of call should be uninferable (you don't know at static analysis time what is the value for the interpreter or the os, or...). If we want to do better, we would need to use a different constant for each, maybe we can discuss this in 113.