from typing import List

import streamlit as st

from .shared import cached_csv_url_to_dataframe, remove_duplicate_client_records
from milestone_attribution_dataset_module import sanitize_funding_data
from constants_module import NEOSERRA_COLUMNS, OUT_COLUMNS

@st.cache_data
def cached_sanitize_funding_data(export_url:str, reportable_only:bool, allowed_centers:List[str] | None = None):
    funding_df = cached_csv_url_to_dataframe(export_url)

    funding_df = sanitize_funding_data(
        funding_df,
        col_neo_attribution_source=NEOSERRA_COLUMNS.milestone_attribution_source,
        col_neo_affirmation=NEOSERRA_COLUMNS.milestone_affirmation,
        col_out_documentation_level=OUT_COLUMNS.milestone_documentation_level,
        col_neo_center=NEOSERRA_COLUMNS.center
    )

    funding_df = remove_duplicate_client_records(funding_df)

    if allowed_centers is not None:
        funding_df = funding_df[funding_df[NEOSERRA_COLUMNS.center].isin(allowed_centers)]

    if reportable_only:
        funding_df = funding_df[funding_df[NEOSERRA_COLUMNS.reportable] == 1]

    return funding_df