Compute EU membership on a given date with Python by using data from Wikipedia

eu_membership.py

import pandas as pd
from datetime import date
from typing import List
from functools import cache
from warnings import warn
import re

MEMBERSHIP_URL = 'https://en.wikipedia.org/wiki/Member_state_of_the_European_Union'
MEMBERSHIP_URL_PERM_LINK = 'https://en.wikipedia.org/w/index.php?title=Member_state_of_the_European_Union&oldid=1184106841'
CANONICAL_WIKI_DATE = date(2023, 11, 11)

TREATY_OF_MAASTRICHT_EFFECTIVE_DATE = date(1993, 11, 1)

def _clean_names(df):
    df.loc[:, 'Country'] = df['Country'].str.replace('\[.*\]','', regex=True)

def _split_founders_joiners(df):
    is_founder = df['Accession'].str.match('Founder.*')
    founders = df[is_founder]
    joiners = df[~is_founder]
    joiners.loc[:, 'Accession'] = pd.to_datetime(joiners['Accession']).map(lambda dt: dt.date())
    return founders, joiners

@cache
def _wiki_data(canonical=False):
    if canonical:
        tables = pd.read_html(MEMBERSHIP_URL_PERM_LINK, match = 'Population', parse_dates=True)
    else:
        tables = pd.read_html(MEMBERSHIP_URL, match = 'Population', parse_dates=True)
    exited = [table for table in tables if 'Accession' in table.columns and 'Withdrawal' in table.columns][0]
    exited = exited[['Country', 'Accession', 'Withdrawal']]
    exited.loc[:, 'Withdrawal'] = pd.to_datetime(exited['Withdrawal']).map(lambda dt: dt.date())
    _clean_names(exited)
    exited_founders, exited_joiners = _split_founders_joiners(exited)
    remain = [table for table in tables if 'Accession' in table.columns and 'Withdrawal' not in table.columns][0]
    remain = remain[['Country', 'Accession']]
    _clean_names(remain)
    remain_founders, remain_joiners = _split_founders_joiners(remain)
    return ((remain_founders, remain_joiners), (exited_founders, exited_joiners))

def eu_members(dt: date, founding_date = TREATY_OF_MAASTRICHT_EFFECTIVE_DATE, canonical=False) -> List[str]:
    """
    Returns the list of EU members on a given date using Wikipedia data from https://en.wikipedia.org/w/index.php?title=Member_state_of_the_European_Union
    Useful in cases when data is tabulated both for EU and its member states, e.g. https://www.census.gov/foreign-trade/balance/.
    founding_date - default to Maastrich Treaty effective date, see https://en.wikipedia.org/wiki/Maastricht_Treaty
    canonical - use a frozen link Wikipedia data at the time of the writing of the code but have outdated data
    >>> eu_members(date(1992, 1, 1))
    []
    >>> eu_members(date(1998, 1, 1))
    ['Austria', 'Belgium', 'Denmark', 'Finland', 'France', 'Germany', 'Greece', 'Ireland', 'Italy', 'Luxembourg', 'Netherlands', 'Portugal', 'Spain', 'Sweden', 'United Kingdom']
    >>> eu_members(date(2019, 1, 1))
    ['Austria', 'Belgium', 'Bulgaria', 'Croatia', 'Cyprus', 'Czech Republic', 'Denmark', 'Estonia', 'Finland', 'France', 'Germany', 'Greece', 'Hungary', 'Ireland', 'Italy', 'Latvia', 'Lithuania', 'Luxembourg', 'Malta', 'Netherlands', 'Poland', 'Portugal', 'Romania', 'Slovakia', 'Slovenia', 'Spain', 'Sweden', 'United Kingdom']
    >>> eu_members(date(2023, 1, 1))
    ['Austria', 'Belgium', 'Bulgaria', 'Croatia', 'Cyprus', 'Czech Republic', 'Denmark', 'Estonia', 'Finland', 'France', 'Germany', 'Greece', 'Hungary', 'Ireland', 'Italy', 'Latvia', 'Lithuania', 'Luxembourg', 'Malta', 'Netherlands', 'Poland', 'Portugal', 'Romania', 'Slovakia', 'Slovenia', 'Spain', 'Sweden']
    """
    if canonical and dt > canonical_wiki_date:
        warn(f'Using Wikipedia data as of {canonical_wiki_date}, might be outdated')
    ((remain_founders, remain_joiners), (exited_founders, exited_joiners)) = _wiki_data()
    if dt < founding_date: 
        return []
    exited_founders = exited_founders[exited_founders['Withdrawal'] < dt]['Country'].tolist()
    remain_founders = remain_founders['Country'].tolist()
    remain_joiners = remain_joiners[remain_joiners['Accession'] <= dt]['Country'].tolist()
    exited_joiners = exited_joiners[(exited_joiners['Accession'] <= dt)&(exited_joiners['Withdrawal'] >= dt)]['Country'].tolist()
    return sorted([ *remain_founders, *exited_founders, *remain_joiners, *exited_joiners ])

if __name__ == "__main__":
    import doctest
    doctest.testmod()