Is it possible to use another locale in `.sort()`?

[etiennebacher]

Is it possible to use another locale in `.sort()`?

[cjackal]

IIRC this kind of locale-aware collate is pretty nontrivial and not many programming language (including rust, the engine of polars) support it battery-included. My best shot so far:

import locale
locale.setlocale(locale.LC_ALL, "en_GB.utf8")

import polars as pl
df = pl.DataFrame({"x": ["a", "b", "A", "B"]})
df.with_columns(pl.col("x").map_elements(locale.strxfrm, return_dtype=pl.Utf8()).alias("x_cmp")).sort("x_cmp")
# shape: (4, 2)
# ┌─────┬───────┐
# │ x   ┆ x_cmp │
# │ --- ┆ ---   │
# │ str ┆ str   │
# ╞═════╪═══════╡
# │ a   ┆ š悝   │
# │ A   ┆ š晙   │
# │ b   ┆ Ů惹   │
# │ B   ┆ Ů暙   │
# └─────┴───────┘

[stinodego]

It's a fair request, but I'm going to say this is out of scope for now.

If someone can come up with a good API design for this, and show an elegant/efficient implementation, I'd be happy to take a look. But we will not be working on this ourselves.