json_decode followed by .struct.field(<fld>) raises StructFieldNotFoundError in lazy mode

[alexander-beedie]

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

🟢 Eager (works)

import polars as pl

df = pl.DataFrame({
  "info": ['{"event":"UPDATE","event_id":"123456","provider":"SOMEBODY"}']
})
df = df.select(pl.col("info").str.json_decode())
# shape: (1, 1)
# ┌────────────────────────────────┐
# │ info                           │
# │ ---                            │
# │ struct[3]                      │
# ╞════════════════════════════════╡
# │ {"UPDATE","123456","SOMEBODY"} │
# └────────────────────────────────┘

df.select(pl.col("info").struct.field("provider"))
# shape: (1, 1)
# ┌──────────┐
# │ provider │
# │ ---      │
# │ str      │
# ╞══════════╡
# │ SOMEBODY │
# └──────────┘

🔴 Lazy (fails)

lf = pl.LazyFrame({
  "info": ['{"event":"UPDATE","event_id":"123456","provider":"SOMEBODY"}']
})
lf = lf.select(pl.col("info").str.json_decode())
lf.collect()
# shape: (1, 1)
# ┌────────────────────────────────┐
# │ info                           │
# │ ---                            │
# │ struct[3]                      │
# ╞════════════════════════════════╡
# │ {"UPDATE","123456","SOMEBODY"} │
# └────────────────────────────────┘

lf.select(pl.col("info").struct.field("provider")).collect()
# polars.exceptions.StructFieldNotFoundError: provider

Log output

Traceback (most recent call last):
  lf.select(pl.col("info").struct.field("provider")).collect()
File "/.../polars/lazyframe/frame.py", line 2056, in collect
  return wrap_df(ldf.collect(callback))
                 ^^^^^^^^^^^^^^^^^^^^^
polars.exceptions.StructFieldNotFoundError: provider
Resolved plan until failure:
---> FAILED HERE RESOLVING 'select' <---
SELECT [col("info").str.json_decode()] FROM
DF ["info"]; PROJECT */1 COLUMNS

Issue description

Lazy mode raises an error when unpacking a struct field following json_decode. Note: disabling the optimiser (.collect(no_optimization=True))does not resolve/avoid the error.

Expected behavior

Should successfully identify/return the requested struct field, as does eager mode.

Installed versions

--------Version info---------
Polars: 1.21.0
Index type: UInt32
Platform: macOS-15.2-arm64-arm-64bit
Python: 3.12.8 (main, Dec 3 2024, 18:42:41) [Clang 16.0.0 (clang-1600.0.26.4)]
LTS CPU: False
----Optional dependencies----
Azure CLI
adbc_driver_manager 1.4.0
altair
azure.identity
boto3 1.36.3
cloudpickle
connectorx 0.3.3
deltalake
fastexcel 0.12.1
fsspec 2024.12.0
gevent
google.auth 2.37.0
great_tables 0.16.1
matplotlib
numpy 2.0.2
openpyxl
pandas 2.2.3
pyarrow 19.0.0
pydantic 2.9.1
pyiceberg
sqlalchemy 1.4.54
torch 2.5.1
xlsx2csv
xlsxwriter 3.2.2

Replace this line with the output of pl.show_versions(). Leave the backticks in place.