Non-string provider inputs must be projected as JSON formatted strings. The current codegen simply calls `json.dumps` for such properties, but this does not work for the new input types, which aren't JSON serializable.
To address this, make use of the new `pulumi.runtime.to_json` utility function, which is capable of serializing raw dicts and input types as JSON.
We currently emit array types as `List[T]` for Python, but `List[T]` is invariant, which causes type checkers like mypy to produce errors when values like `["foo", "bar"]` are passed as args typed as `List[pulumi.Input[str]]`. Instead, we should move to using `Sequence[T]` which is covariant, and does not have this problem.
We actually already do this for `Dict` vs. `Mapping`, emitting map types as `Mapping[str, T]` rather than `Dict[str, T]` because `Mapping[str, T]` is covariant. This change makes us consistent for array types.
If PyName(name) != PyNameLegacy(name), we panic, unless name is in an exclusion list.
Once all providers have been updated to this codegen (and we've added to `useLegacyName` and `excludeFromPanic` as appropriate), we can go back and remove the panic behavior along with the `excludeFromPanic` behavior.
If there's a config subpackage, we had explicitly added it to the map of modules because there *are* config properties, so we know it will be non-empty. This change ensures the config subpackage is imported in the top-level package's __init__.py and that the config subpackage itself has an __init__.py emitted.
Resource doc changes for Python:
- Types are included in constructor/function args
- The property names for input/output types are now always snake_case, regardless of the generated mapping tables, to match the new input/output classes
- Some other minor tweaks to function/constructor signatures (e.g. removed the `__props__` arg, as it's not meant to be used directly; use `@staticmethod` for static `get` methods).
In general, a package/module name in these targets is derived from the
module portion of a type token. If the type token is not already in an
expected form--namely, all lowercase--the generated package/module names
will also be in unexpected forms. These changes normalize the module
names to lowercase s.t. the generated package/module names conform to
expectations.
* Use the correct format for package name when module name is empty.
* Add an exclusion for Docker's Image component resource when generating the Python formal params.
* Prefix the type name with the package name when linking to Python function names.
The k8s SDK logs deprecations warnings as part of the
provider logic, and includes a flag to suppress these
warnings. Including them in the SDKs breaks this logic.
The PCL binder has supported resource options for some time, but these
options haven't been used or processed by the various code generators.
These options--particularly the parent and provider options0--are
critical for import codegen. These changes implement the basic set of
options, and add a note about fleshing out the rest as necessary.
One component of these changes is a new rewriter that rewrites property
references into property paths that are understood by the Pulumi engine.
This rewriter is used to preprocess the contents of the `ignoreChanges`
resource option.
These changes also hack around a weakness in the HCL2 type system:
In Go, references to resources should be typed as `hcl2.ResourceType`.
Unfortunately, this breaks the existing collection semantics associated
with resources. Because of this, the Go code generator does not have
enough information to know that it should generate a `[]pulumi.Resource`
for lists of resources. These changes hack around that limitation using
a Go-specific opaque type and some hardcoded comparisons in
`argumentTypeName`.
Fixes#4923.
