Python client type hint error with column-oriented upsert

[affanshahid]

I'm trying to do an upsert without hard-coded data and I get some type hint errors. Everything works fine so I'm assuming its just a typing issue.

The error on ids is:

Pyright: Argument of type "list[int]" cannot be assigned to parameter "ids" of type "List[ExtendedPointId]" in function "__init__"
  "list[int]" is not assignable to "List[ExtendedPointId]"
    Type parameter "_T@list" is invariant, but "int" is not the same as "ExtendedPointId"
    Consider switching from "list" to "Sequence" which is covariant

I can fix this one by typing the ids declaration as ids: list[ExtendedPointId]. This may be just a pyright issue since ExtendedPointId is defined as int | str.

The error on vectors is:

Pyright: Argument of type "Iterable[ndarray[Unknown, Unknown]]" cannot be assigned to parameter "vectors" of type "BatchVectorStruct" in function "__init__"
  Type "Iterable[ndarray[Unknown, Unknown]]" is not assignable to type "BatchVectorStruct"
    "Iterable[ndarray[Unknown, Unknown]]" is not assignable to "List[List[StrictFloat]]"
    "Iterable[ndarray[Unknown, Unknown]]" is not assignable to "List[List[List[StrictFloat]]]"
    "Iterable[ndarray[Unknown, Unknown]]" is not assignable to "Dict[StrictStr, List[Vector]]"

Can't seem to get around this one without a cast. Like I said, everything works but the type errors hamper the DX.

[joein]

Hi @affanshahid
Thanks for reporting this, we'll look into it
Could you please tell us qdrant-client version you're using along with the version of pyright?

[affanshahid]

• qdrant-client 1.11.2
• pyright language server 1.1.381

[joein]

Hey @affanshahid
We took a look at the issue, we might need more time to fix the problem with ids
However, embeddings type hint issue seems to be correct, since it does not expect numpy arrays

[affanshahid]

I see, i imported TextEmbedding from fastembed in the example, what would be the correct way of doing this? Cast to regular list?