BYOD - Allow Polars

[MariusWirtz]

No description provided.

[vmitsenko]

Hey, I’ve been looking into how the interface can be generalised. Seems like Python’s Protocol may be a good fit here—it gives a clean way to define the minimal set of methods/attributes we expect from a “dataframe-like” object, without committing to a specific backend. This should make it easier to support pandas, polars, and others in a consistent way.

I tested this approach with both pandas and polars and it seems to be working. Here’s the example I used:

from DataFrameUtils import build_cellset_from_dataframe, convert_to_dataframe_like

import pandas as pd
import polars as pl

data = {
    "Name": ["Alice", "Bob", "Charlie", "Diana", "Alice",],
    "Age": [28, 34, 25, 30, 28],
    "Department": ["HR", "IT", "Marketing", "Finance", "HR",],
    "Salary": [55000, 72000, 48000, 61000, 45000]
}

df_pl = pl.DataFrame(data)
df_pd = pd.DataFrame(data)

cellset_pd = build_cellset_from_dataframe(convert_to_dataframe_like(df_pd))
cellset_pl = build_cellset_from_dataframe(convert_to_dataframe_like(df_pl))

Also, all unit tests are passed for the write_dataframe