app.py

# %%

import io
import uuid
from dataclasses import dataclass
from typing import Optional

import matplotlib.pylab as plt
import numpy as np
import pandas as pd
import solara
import solara.lab
from matplotlib.figure import Figure
from scipy import stats
from solara.components.file_drop import FileInfo


def make_cdf_figure(
    values_left: np.ndarray, values_right: np.ndarray, stat_loc: float
) -> Figure:
    fig = Figure(figsize=(5, 5))
    ax = fig.subplots()

    ax.axvline(stat_loc, color="grey")

    # Cumulative distributions.
    ax.ecdf(values_left, label="left")
    ax.ecdf(values_right, label="right")
    ax.legend()
    ax.set_xlabel("Value")
    ax.set_ylabel("CDF")

    return fig


def make_pdf_figure(
    values_left: np.ndarray, values_right: np.ndarray, stat_loc: float
) -> Figure:
    fig = Figure(figsize=(5, 5))
    ax = fig.subplots()

    ax.axvline(stat_loc, color="grey")

    # Cumulative distributions.
    ax.hist(values_left, bins="fd", density=True, histtype="step", label="left")
    ax.hist(values_right, bins="fd", density=True, histtype="step", label="right")
    ax.legend()
    plt.show(fig)
    ax.set_xlabel("Value")
    ax.set_ylabel("PDF")

    return fig


# %%


def dropna(values: np.ndarray) -> np.ndarray:
    return values[~np.isnan(values)]


@solara.component
def KSTestResult(values_left, values_right):
    values_left = dropna(values_left)
    values_right = dropna(values_right)

    kstat = stats.ks_2samp(values_left, values_right)

    fig_cdf = make_cdf_figure(values_left, values_right, kstat.statistic_location)
    fig_pdf = make_pdf_figure(values_left, values_right, kstat.statistic_location)

    with solara.Card("Kolmogorov-Smirnov Test"):
        with solara.Columns():
            solara.FigureMatplotlib(fig_cdf)
            solara.FigureMatplotlib(fig_pdf)

        solara.Markdown("# Test Result")

        solara.Info(
            f"statistic: {kstat.statistic:.3g}",
        )

        solara.Info(
            f"p-value: {kstat.pvalue:.3g}",
        )

        solara.Info(
            f"location: {kstat.statistic_location:.3g}",
        )


@dataclass
class Selection:
    file: Optional[str] = None
    column: Optional[str] = None

    @property
    def is_set(self) -> bool:
        return self.file is not None and self.column is not None

    @property
    def columns(self) -> list[str]:
        if self.file is not None:
            return list(data_store.value[self.file].columns)
        return []

    @property
    def array(self) -> np.ndarray:
        if self.is_set:
            return data_store.value[self.file][self.column].to_numpy()
        return np.array([])


def all_set(selections: list[Selection]) -> bool:
    return all(s.is_set for s in selections)


@solara.component
def Selectors(selection: solara.Reactive[Selection]):
    solara.Select(
        label="Select file",
        values=list(data_store.value.keys()),
        value=selection.value.file,
        on_value=lambda x: selection.update(file=x, column=None),
    )

    solara.Select(
        label="Select_column",
        values=selection.value.columns,
        value=selection.value.column,
        on_value=lambda x: selection.update(column=x),
    )


file_info: solara.Reactive[list[FileInfo]] = solara.reactive([])
data_store = solara.Reactive({})

selection_left = solara.reactive(Selection())
selection_right = solara.reactive(Selection())


@solara.component
def KSApp():
    def load_data():
        d = {}
        for f in file_info.value:
            b_io = io.BytesIO(f["data"])
            df = pd.read_csv(b_io)
            d[f["name"]] = df
        data_store.set(d)

    _ = solara.use_memo(load_data, dependencies=[file_info.value])
    upload_key = solara.reactive(uuid.uuid4())

    def clear_all():
        upload_key.set(uuid.uuid4())
        selection_left.set(Selection())
        selection_right.set(Selection())
        file_info.set([])
        data_store.set({})

    with solara.ColumnsResponsive([3, 9]):
        with solara.Card("Input"):
            solara.FileDropMultiple(
                label="Upload CSV files",
                on_file=file_info.set,
                lazy=False,
            )  # .key(upload_key.value.hex)
            solara.Text("Select left:")
            Selectors(selection_left)
            solara.Text("Select left:")
            Selectors(selection_right)
            solara.Button(label="Clear", on_click=clear_all)

        if not all_set([selection_left.value, selection_right.value]):
            with solara.Card():
                solara.Text("Please upload data and select both files and columns")

        else:
            KSTestResult(selection_left.value.array, selection_right.value.array)


page = KSApp()