Incorrect results with pl.struct().value_counts().struct.unnest()

[etrotta]

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

test_df = pl.DataFrame({"x": [i for i in range(1, 6) for _ in range(i)]})
# Run it multiple times - the order of the results varies each time
test_df.select(pl.struct("x").value_counts().struct.unnest())

Log output

>>> test_df.select(pl.struct("x").value_counts().struct.unnest())
shape: (5, 2)
┌───────────┬───────┐
│ x         ┆ count │
│ ---       ┆ ---   │
│ struct[1] ┆ u32   │
╞═══════════╪═══════╡
│ {2}       ┆ 4     │
│ {5}       ┆ 5     │
│ {3}       ┆ 3     │
│ {1}       ┆ 1     │
│ {4}       ┆ 2     │
└───────────┴───────┘
>>> test_df.select(pl.struct("x").value_counts().struct.unnest())
shape: (5, 2)
┌───────────┬───────┐
│ x         ┆ count │
│ ---       ┆ ---   │
│ struct[1] ┆ u32   │
╞═══════════╪═══════╡
│ {1}       ┆ 5     │
│ {5}       ┆ 3     │
│ {3}       ┆ 2     │
│ {4}       ┆ 1     │
│ {2}       ┆ 4     │
└───────────┴───────┘

Issue description

The results vary each time you run the example code, as if it were randomly shuffling the columns separately from each other.

It works correctly if you do it in separate steps or use pl.col('x') instead of pl.struct('x')
(Reduced the minimal example as much as I could while still reproducing the issue - in my real case there were multiple columns like pl.struct('x', 'y'))

Expected behavior

It should return the count associated with the correct values for each row.

It works correctly if you do it in two separate steps, it still gets shuffled but with the correct value - count combinations instead of independently shuffling each column.

>>> test_df.select(pl.struct("x").value_counts()).select(pl.all().struct.unnest())
shape: (5, 2)
┌───────────┬───────┐
│ x         ┆ count │
│ ---       ┆ ---   │
│ struct[1] ┆ u32   │
╞═══════════╪═══════╡
│ {4}       ┆ 4     │
│ {1}       ┆ 1     │
│ {2}       ┆ 2     │
│ {5}       ┆ 5     │
│ {3}       ┆ 3     │
└───────────┴───────┘

Installed versions

--------Version info---------
Polars:              1.21.0
Index type:          UInt32
Platform:            Linux-5.15.167.4-microsoft-standard-WSL2-x86_64-with-glibc2.39
Python:              3.12.3 (main, Jan 17 2025, 18:03:48) [GCC 13.3.0]
LTS CPU:             False

----Optional dependencies----
Azure CLI            2.67.0
adbc_driver_manager  <not installed>
altair               <not installed>
azure.identity       <not installed>
boto3                <not installed>
cloudpickle          <not installed>
connectorx           <not installed>
deltalake            <not installed>
fastexcel            <not installed>
fsspec               <not installed>
gevent               <not installed>
google.auth          <not installed>
great_tables         <not installed>
matplotlib           <not installed>
numpy                2.2.2
openpyxl             <not installed>
pandas               2.2.3
pyarrow              <not installed>
pydantic             <not installed>
pyiceberg            <not installed>
sqlalchemy           <not installed>
torch                <not installed>
xlsx2csv             <not installed>
xlsxwriter           <not installed>