A static type checker fails because of filter argument type.
Type "dict[str, str]" cannot be assigned to type "Dict[str, str | float | int | bool | List[Unknown] | dict[Unknown, Unknown]] | None"
"dict[str, str]" is incompatible with "Dict[str, str | float | int | bool | List[Unknown] | dict[Unknown, Unknown]]
Expected Behavior
The dictionary passed in the filter argument is not changed in the query function, therefore we could use covariant Mapping type instead of Dict, so the type checker doesn't throw.
4. Run Pyright: `pyright main.py`
### Relevant log output
```shell
% pyright main.py
/Users/****/main.py
/Users/****/main.py:17:12 - error: Argument of type "dict[str, str]" cannot be assigned to parameter "filter" of type "Dict[str, str | float | int | bool | List[Unknown] | dict[Unknown, Unknown]] | None" in function "query"
Type "dict[str, str]" cannot be assigned to type "Dict[str, str | float | int | bool | List[Unknown] | dict[Unknown, Unknown]] | None"
"dict[str, str]" is incompatible with "Dict[str, str | float | int | bool | List[Unknown] | dict[Unknown, Unknown]]"
Type parameter "_VT@dict" is invariant, but "str" is not the same as "str | float | int | bool | List[Unknown] | dict[Unknown, Unknown]"
Consider switching from "dict" to "Mapping" which is covariant in the value type
"dict[str, str]" is incompatible with "None" (reportArgumentType)
1 error, 0 warnings, 0 informations
Is this a new bug in the Pinecone Python client?
Current Behavior
I query Pinecone index like this:
A static type checker fails because of filter argument type.
Expected Behavior
The dictionary passed in the filter argument is not changed in the query function, therefore we could use covariant Mapping type instead of Dict, so the type checker doesn't throw.
Steps To Reproduce
pip install pinecone-client pyright
main.py
:pinecone = Pinecone(api_key=os.environ["PINECONE_API_KEY"]) pinecone_index = pinecone.Index(name=os.environ["PINECONE_INDEX_NAME"]) assert pinecone_index is not None
vector = [0.1, 0.2, 0.3, 0.4, 0.5] # sample embedding vector search_filter = { "username": "thyrst", } result = pinecone_index.query( vector=vector, top_k=10, filter=search_filter, includeMetadata=True, )
print(result)
Environment
Additional Context
I think covariant types should be used preferably for the input arguments everywhere where the passed objects are not changed.