Add support for pseudo-classes in nested TCSS. #4252
Conversation
To be able to disambiguate between selector:pseudo-class and declaration:rule-value in nested TCSS (see #4039 for the original issue and #4163 for a first attempt at solving this) we establish that selectors with widget type names always start with an upper case letter A-Z or an underscore _ whereas declarations always start with a lower case letter a-z. When a user creates a widget subclass that doesn't conform to this, we issue a 'SyntaxWarning' to let the user know. Because we do this with the standard module 'warnings', the warning can easily be supressed if the user has a good reason to create a widget subclass with a name starting with a lower case letter (which is valid Python, just unhelpful to Textual).
src/textual/widget.py
Outdated
| if not name[0].isupper() and not name.startswith("_"): | ||
| warnings.warn( | ||
| SyntaxWarning( | ||
| f"Widget subclass {name!r} should be capitalised or start with '_'." | ||
| ), | ||
| stacklevel=2, | ||
| ) |
There was a problem hiding this comment.
Because this is from the std library, suppressing these warnings is “easy”.
Even so, do we add something like a Boolean textual.widget.SHOW_WIDGET_NAME_WARNING to make it even easier for users to suppress the warning?
There was a problem hiding this comment.
(just looking at the code in the comment above; so might be missing context)
Is it intentional that _foo seen as valid? (I'm mostly guessing yes, but wanted to be sure).
There was a problem hiding this comment.
_foo is valid and that's ok in the context in which this restriction was introduced.
Ideally, the wording of the warning (and of the comments around the code) should give that away.
I'm open to making it invalid as well if we think it makes sense to be a bit more restrictive with the goal of looking more coherent (given that foo raises the warning but Foo doesn't).
There was a problem hiding this comment.
Nah, I think it's fine; just wanted to be 100% sure that the whole package of code was doing what motivated it.
Relevant review comment: #4252 (comment)
All error messages are what you would expect. There are tests that check we didn't regress & I added tests that check in nested contexts the messages are the same. Here's a screenshot of the case that was unhelpful in the previous PR and which is now as good as it gets:
|
Fixes #4039 and supersedes #4163.
To be able to disambiguate between selector:pseudo-class and declaration:rule-value in nested TCSS (see #4039 for the original issue and #4163 for a first attempt at solving this) we establish that selectors with widget type names always start with an upper case letter A-Z or an underscore _ whereas declarations always start with a lower case letter a-z.
When a user creates a widget subclass that doesn't conform to this, we issue a 'SyntaxWarning' to let the user know. Because we do this with the standard module 'warnings', the warning can easily be supressed if the user has a good reason to create a widget subclass with a name starting with a lower case letter (which is valid Python, just unhelpful to Textual).