Add a builtins parameter

[Neyri]

Hello,
I am working on databricks notebooks and I use VScode to edit the content of the notebooks.
I make extensive use of builtin databricks functions such as dbutils or spark. However I find Pylance raises an UndefinedVariable on those functions as they are not locally declared.
I would like to be able to specify some builtins functions that pylance won't raise an undefined error on. I imagine this would be similar to flake8's approach on the subject, see here

Even better : I would like to be able to specify a databricks profile where all databricks' builtin functions will be already defined (and autocompletion compatible). But not sure that fits in the philosophy of the package.

[erictraut]

Pylance doesn't hard-code the definition of "builtins". Rather, it uses the type information from the builtins.pyi stub file from typeshed. You can download your own copy of the typeshed stdlib stubs and use the python.analysis.typeshedPaths setting to point to their path. You can then modify your local copy of builtins.pyi file and add additional symbols.

[erictraut]

We've received this request from a number of users, so I've implemented a less-cumbersome way to extend the builtins namespace. For details, refer to this documentation. This new capability will be included in next week's release of pylance.

[debonte]

This issue has been fixed in version 2022.2.1, which we've just released. You can find the changelog here: CHANGELOG.md

[bthorsted]

So, I have a question about using the __builtins__.pyi file. In the example from databricks, the builtins are not functions, but actually entire modules with submodules, classes and functions. How would I go about 'mocking' that structure with the __builtins__.pyi file?

[erictraut]

You should be able to import those other modules in __builtins__.pyi. To make these modules appear as "re-exported", you'll need to use a redundant form of import.

import foo as foo

[bthorsted]

You should be able to import those other modules in builtins.pyi.

So if the modules are not actually available outside of the databricks runtime, I would need to write a mock up module with the necessary submodules and classes and then import them into the __builtins__.pyi, is that correct?

[erictraut]

Yes, that's correct.