pandas index & multiindex in TableReport

[jeromedockes]

supersedes #1074

adds displaying pandas index, handle correctly multiIndex (both for index and columns) and index/columns level names if any are present

allows controlling how many rows are displayed in the sample table

also some refactoring of how the html table is constructed

[jeromedockes]

todo:

• keep only one parameter for controlling number of displayed rows

[jeromedockes]

I reduced the number of rows from 10 to 5 in the srkub homepage.

BTW, not really related to this PR but in the home page if we want the report to take yet less vertical space one way is to give it a bit more horizontal space (either making the sk-landing-page a bit wider or giving the report more space compared to the code snippet on its right) because then the column filters select will drop to be on the same line as the tabs

[jeromedockes]

ok this one is ready for review

here is an example to generate a report with multiple column and index levels:

Details

# %%
import datetime

import pandas as pd
import numpy as np

from skrub import TableReport

df = pd.DataFrame(
    {
        "A": ["one", "one", "two", "three"] * 6,
        "B": ["A", "B", "C"] * 8,
        "C": ["foo", "foo", "foo", "bar", "bar", "bar"] * 4,
        "D": np.random.randn(24),
        "E": np.random.randn(24),
        "F": [datetime.datetime(2013, i, 1) for i in range(1, 13)]
        + [datetime.datetime(2013, i, 15) for i in range(1, 13)],
    }
)

df = pd.pivot_table(
    df,
    values="E",
    index=["B", "C"],
    columns=["A"],
    aggfunc=["sum", "mean"],
)

report = TableReport(df)
report

# %%
report.open()

[TheooJ]

Maybe top_rows and bottom_rows ? Or n_top_rows ?
Feels like max_top_slice_size is kind of long

[jeromedockes]

indeed it is long, but there are quite a few "rows" flying around, and we need to distinguish between the maximum number of displayed rows and the actual number so I wanted to be very explicit here.

note this is a parameter of a private helper, the parameter exposed to the user is just n_rows

[GaelVaroquaux]

LGTM, but I would prefer if the "!important" in CSS could be removed

[jeromedockes]

LGTM, but I would prefer if the "!important" in CSS could be removed

yep I removed it already. it's just that pure has a selector that is quite specific already (table class name + nth pseudo-class for alternating rows + elem) and I didn't want to have to come up with an artificially specific one but actually I remembered I wrapped the report in an element with an id which comes in handy for that

[TheooJ]

Merged, thanks @jeromedockes !