Initializing a struct series with an array type field using Numpy arrays results in nulls

[jkim-]

Checks

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest version of Polars.

Reproducible example

x = np.array([1,2,3], dtype=np.int32)
pl.Series(
    values=[{"x": x}], 
    dtype=pl.Struct({
        "x": pl.Array(inner=pl.Int32, shape=(3,)),
    }),
).struct.unnest()

results in

x
--
array[i32, 3]
null

Log output

No response

Issue description

I'm trying to create a series backed by Numpy arrays to avoid copying, and the resulting series only contained nulls.

Expected behavior

x = np.array([1,2,3], dtype=np.int32)
pl.Series(
    values=[{"x": list(x)}], 
    dtype=pl.Struct({
        "x": pl.Array(inner=pl.Int32, shape=(3,)),
    }),
).struct.unnest()

works fine:

x
--
array[i32, 3]
[1, 2, 3]

Installed versions

--------Version info---------
Polars:              1.7.1
Index type:          UInt32
Platform:            Linux-5.10.225-191.878.amzn2int.x86_64-x86_64-with-glibc2.31
Python:              3.11.9 | packaged by conda-forge | (main, Apr 19 2024, 18:36:13) [GCC 12.3.0]

----Optional dependencies----
adbc_driver_manager  <not installed>
altair               <not installed>
cloudpickle          2.2.1
connectorx           <not installed>
deltalake            <not installed>
fastexcel            <not installed>
fsspec               2024.5.0
gevent               24.2.1
great_tables         <not installed>
matplotlib           3.8.4
nest_asyncio         1.6.0
numpy                1.26.4
openpyxl             <not installed>
pandas               2.2.2
pyarrow              16.1.0
pydantic             2.7.1
pyiceberg            <not installed>
sqlalchemy           <not installed>
torch                2.3.0
xlsx2csv             <not installed>
xlsxwriter           <not installed>