* Implement lazy loading mechanism for expensive metadata providers
* Add support for lazy values in metadata matchers
* Fix type issues and implement lazy value support in base metadata provider too
* Add unit tests for BaseMetadataProvider
Co-authored-by: Zsolt Dollenstein <zsol.zsol@gmail.com>
which ensures we won't have inconsistent black-vs-isort errors
going forward. We can always format by running `ufmt format .`
at the root, and check with `ufmt check .` in our CI actions.
Because we'd consider `BaseMetadataProvider[int]` to be a subtype of
`BaseMetadataProvider[object]`, it should be covariant over its
typevar, rather than invariant.
This isn't entirely correct because we have a mutable data structure
(`_computed`) that depends on the typevar, and pyre points this out
(though with a really confusing error message). However, it's not
correct to say that `BaseMetadataProvider` is invariant either, so I
think this is the lesser evil.
I don't think it's practical to redesign this API to avoid the variance
issue, so I'm ignoring the new type error that results from this change.
I think this may resolve some of the issues we've seen internally with
D17820032.
This introduces a few new gotchas (namely attribute access and a bug with
Union[Callable]), but it also removes a whole host of pyre-fixmes and gets us
updated to a release of pyre that came out after May.
Standardize on the convention that private modules (those we don't expect people to directly import) are prefixed with an underscore. Everything under a directory/module that has an underscore is considered private, unless it is re-exported from a non-underscored module. Most things are exported from libcst directly, but there are a few things in libcst.tool, libcst.codegen and libcst.metadata that are namedspaced as such.
