Allow retrieval of preceding row

[NicolaCourtier]

Hi, I've been using a wrapper that includes the row immediately preceding a cycle or step as part of the dataset. It would be great to incorporate this functionality into PyProBE.

The following function can be used with filtering, for example:

• filter_with_preceding_row(procedure, experiment=experiment_name)
• filter_with_preceding_row(experiment, cycle=i)
• filter_with_preceding_row(experiment, phase="discharge", step=i).

import polars as pl


def filter_with_preceding_row(
    procedure,
    experiment: str | None = None,
    cycle: int | None = None,
    phase: str | None = None,
    step: int | None = None,
):
    """
    Obtain the PyProBE result including the preceding row of the data.
    """
    data = procedure
    if experiment is not None:
        data = data.experiment(experiment)
    if cycle is not None:
        data = data.cycle(cycle)
    if phase is not None:
        if phase == "discharge":
            data = data.discharge()
        elif phase == "charge":
            data = data.charge()
        elif phase == "rest":
            data = data.rest()
        else:
            raise ValueError("Unrecognised phase")
    if step is not None:
        data = data.step(step)

    # Add the preceding row
    start_time = data.lf.select("Time [s]").first().collect().item()
    preceding_row = procedure.lf.filter(pl.col("Time [s]") < start_time).last()
    data.lf = pl.concat((preceding_row, data.lf), how="diagonal_relaxed")

    return data

[tomjholland]

Hi NicolaCourtier, I've implemented this on this branch: #382. This allows you to set an include_preceding_row flag in any filter e.g. .discharge(1, include_preceding_row=True). For tidiness, this is a breaking change as you can no longer pass multiple index arguments to the method, and must do so in a list/tuple instead i.e. previously you could do .discharge(1, 3, 6) whereas now you must do .discharge((1, 3, 6)) or .discharge([1, 3, 6]). I'll therefore update the main release with this change alongside other breaking changes I am working on in #373 to natively support the BDF format. In the meantime, and if you don't want to make any major updates to your scripts, you can install from this branch directly.