`isoweek_to_datetime` not working with pl.DataFrame in a `with_columns` context

[pietroppeter]

I am trying to use isoweek_to_datetime in a dataframe to generate a new column, it seems the function is not working in a with_columns context.

Reproducible example (adapted from docs and generated by exporting a jupyter notebook to markdown):

import polars as pl
from iso_week_date.polars_utils import SeriesIsoWeek  # noqa: F401

s = pl.Series(["2022-W52", "2023-W01", "2023-W02"])
s.iwd.isoweek_to_datetime()

shape: (3,)

date
2022-12-26
2023-01-02
2023-01-09

df = pl.DataFrame(data=dict(isoweek=s))
df

shape: (3, 1)

isoweek
str
"2022-W52"
"2023-W01"
"2023-W02"

df.get_column("isoweek").iwd.isoweek_to_datetime().alias("week_date")

shape: (3,)

week_date
date
2022-12-26
2023-01-02
2023-01-09

df.with_columns(pl.col("isoweek").iwd.isoweek_to_datetime().alias("week_date"))

╭─────────────────────────────── Traceback (most recent call last) ────────────────────────────────╮
│ in <module>:1                                                                                    │
│                                                                                                  │
│ ❱ 1 df.with_columns(pl.col("isoweek").iwd.isoweek_to_datetime().alias("week_date"))              │
│   2                                                                                              │
│                                                                                                  │
│ /Users/pietropeterlongo/projects/poste-logistics/.venv/lib/python3.12/site-packages/iso_week_dat │
│ e/polars_utils.py:453 in isoweek_to_datetime                                                     │
│                                                                                                  │
│   450 │   │   '''                                                                                │
│   451 │   │   ```                                                                                │
│   452 │   │   """                                                                                │
│ ❱ 453 │   │   return isoweek_to_datetime(self._series, offset=offset, weekday=weekday)           │
│   454 │                                                                                          │
│   455 │   def isoweekdate_to_datetime(self: Self, offset: OffsetType = timedelta(0)) -> T:       │
│   456 │   │   """Converts `str` series or expr of ISO Week date format YYYY-WNN-D to a series    │
│                                                                                                  │
│ /Users/pietropeterlongo/projects/poste-logistics/.venv/lib/python3.12/site-packages/iso_week_dat │
│ e/polars_utils.py:186 in isoweek_to_datetime                                                     │
│                                                                                                  │
│   183 │   '''                                                                                    │
│   184 │   ```                                                                                    │
│   185 │   """                                                                                    │
│ ❱ 186 │   if not is_isoweek_series(series):                                                      │
│   187 │   │   msg = f"`series` values must match ISO Week format {ISOWEEK__FORMAT}"              │
│   188 │   │   raise ValueError(msg)                                                              │
│   189                                                                                            │
│                                                                                                  │
│ /Users/pietropeterlongo/projects/poste-logistics/.venv/lib/python3.12/site-packages/polars/expr/ │
│ expr.py:152 in __bool__                                                                          │
│                                                                                                  │
│     149 │   │   │   "- instead of `pl.col('a') in [y, z]`, use `pl.col('a').is_in([y, z])`\n"    │
│     150 │   │   │   "- instead of `max(pl.col('a'), pl.col('b'))`, use `pl.max_horizontal(pl.col │
│     151 │   │   )                                                                                │
│ ❱   152 │   │   raise TypeError(msg)                                                             │
│     153 │                                                                                        │
│     154 │   def __abs__(self) -> Expr:                                                           │
│     155 │   │   return self.abs()                                                                │
╰──────────────────────────────────────────────────────────────────────────────────────────────────╯
TypeError: the truth value of an Expr is ambiguous

You probably got here by using a Python standard library function instead of the native expressions API.
Here are some things you might want to try:
- instead of `pl.col('a') and pl.col('b')`, use `pl.col('a') & pl.col('b')`
- instead of `pl.col('a') in [y, z]`, use `pl.col('a').is_in([y, z])`
- instead of `max(pl.col('a'), pl.col('b'))`, use `pl.max_horizontal(pl.col('a'), pl.col('b'))`

workaround:

df["week_date"] = df.get_column("isoweek").iwd.isoweek_to_datetime().alias("week_date")
df

╭─────────────────────────────── Traceback (most recent call last) ────────────────────────────────╮
│ in <module>:1                                                                                    │
│                                                                                                  │
│ ❱ 1 df["week_date"] = df.get_column("isoweek").iwd.isoweek_to_datetime().alias("week_date")      │
│   2 df                                                                                           │
│   3                                                                                              │
│                                                                                                  │
│ /Users/pietropeterlongo/projects/poste-logistics/.venv/lib/python3.12/site-packages/polars/dataf │
│ rame/frame.py:1211 in __setitem__                                                                │
│                                                                                                  │
│    1208 │   │   │   │   "DataFrame object does not support `Series` assignment by index"         │
│    1209 │   │   │   │   "\n\nUse `DataFrame.with_columns`."                                      │
│    1210 │   │   │   )                                                                            │
│ ❱  1211 │   │   │   raise TypeError(msg)                                                         │
│    1212 │   │                                                                                    │
│    1213 │   │   # df[["C", "D"]]                                                                 │
│    1214 │   │   elif isinstance(key, list):                                                      │
╰──────────────────────────────────────────────────────────────────────────────────────────────────╯
TypeError: DataFrame object does not support `Series` assignment by index

Use `DataFrame.with_columns`.

(note that the workaround does not work)

[pietroppeter]

ah I found the workaround that works: df.with_columns(df.get_column("isoweek").iwd.isoweek_to_datetime().alias("week_date"))

[FBruzzesi]

Ciao Pietro!

That's certainly a bug. Everything that works with series' should work with expr's as well.

I wonder if the extensions registration ends in some kind of conflict:

@pl.api.register_series_namespace("iwd")
@pl.api.register_expr_namespace("iwd")
class SeriesIsoWeek(Generic[T]):
    ...

I will investigate this further.

Could you please share the polars version you are using?

[FBruzzesi]

Ignore all previous considerations. The error

TypeError: the truth value of an Expr is ambiguous

is triggered in the line:

return series.str.extract(pattern).is_not_null().all()

which for Expr seems to cause issues. I will think how to deal with this (most likely with a duck-typing approach)

[FBruzzesi]

@pietroppeter should be fixed in v1.4.0

[pietroppeter]

Will check and report back