Avoid double download buttons by using new data callback function

[prihoda]

Problem Statement

Streamlit now supports passing a function to st.download_button(data=...) so we don't have to use the double download button approach that first uses a regular button and then the confirm_download_button, such as done here for example:

ovo/ovo/app/components/download_component.py

Lines 97 to 114 in 58f1cef

	def download_descriptor_table(filename, design_ids, descriptor_keys=None, key="default", width="content"):
	if st.button(
	"Download descriptor table" if len(design_ids) > 1 else "Download descriptors",
	key=f"prepare_descriptors_{key}",
	width=width,
	):
	with st.spinner("Preparing descriptor table..."):
	# Get raw dataframe with single header, columns named with descriptor keys ("pipeline|tool_key|descriptor")
	df = get_wide_descriptor_table(
	design_ids=design_ids, descriptor_keys=descriptor_keys, nested=False, human_readable=False
	)
	excel_bytes = export_design_descriptors_excel(df)
	confirm_download_button(
	data=excel_bytes.getvalue(),
	file_name=f"{filename}.xlsx",
	mime="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
	key=f"download_descriptors_{key}",
	)

Proposed Solution

One limitation is that the data callback function cannot have any arguments. But it's possible to overcome using a lambda function. Minimal working example.

import streamlit as st
import time

def export_example(a, foo):
    return f"a,foo\n{a},{foo}".encode("utf-8")

def download_example(a):
    st.download_button(
        label="Download report",
        data=lambda: export_example(a, foo="bar"),
        file_name="report.csv",
        mime="text/csv",
        key=a,
    )

download_example('A')

download_example('B')

All functions calling confirm_download_button should now use this approach instead.

The data function should not call any streamlit functions such as st.spinner (the button already shows a spinner), they should just generate the data.

Alternatives Considered

No response