Data_editor issues when changing the index values

Using Streamlit
, ,
1

Hi all,

I’ve run into a problem when using st.data_editor to edit a Pandas DataFrame with a custom index. This code should reproduce the issue:

import streamlit as st
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

# Initialize df
if "df" not in st.session_state:
    index = [f"point_{i}" for i in range(10)]
    st.session_state.df = pd.DataFrame({
        "x": np.arange(10),
        "y": np.random.randn(10)
    }, index=index)

# Data editor
edited_df = st.data_editor(
    st.session_state.df,
    key="editor_key",
    num_rows="dynamic",
    use_container_width=True,
)

# Plot the result
st.subheader("Live-updated Plot")
fig, ax = plt.subplots()
ax.plot(edited_df["x"], edited_df["y"], marker='o')
ax.set_xlabel("x")
ax.set_ylabel("y")
ax.grid(True)
st.pyplot(fig)

For example, if you delete some rows and then proceed to rename the index entries, the deleted rows are unexpectedly restored, and the index changes are not properly registered.

I wonder if this is the expected behavior — perhaps modifying the index directly is not supported?

Kind regards,

2

This should work per the docs.

  • Almost all editable datatypes are supported for index editing. However, pandas.CategoricalIndex and pandas.MultiIndex are not supported for editing.

I simplified the example to the data editor, followed by a static dataframe display of the result. There certainly seems to be a bug here. Feel free to upvote or add any context here: Data editor does not correctly allow index editing · Issue #11434 · streamlit/streamlit · GitHub