get active futures contract

get_active_contracts.py

#!/usr/bin/env python

from dateutil.parser import parse as parse_date
import datetime
import requests
import pandas as pd


def get_contracts(url):
    html = requests.get(url, timeout=5)

    """ CME switched to using ajax """
    prodDataUrl = html.text.split('component.baseUrl = "')[1].split(';')[
        0].replace('" + ', '').replace(' + "', '').strip('"')

    # get data
    url = 'https://www.cmegroup.com%s?tradeDate=%s' % (
        prodDataUrl, datetime.datetime.now().strftime('%m/%d/%Y'))
    data = requests.get(url, timeout=5).json()

    if len(data['settlements']) == 1:
        url = 'https://www.cmegroup.com%s?tradeDate=%s' % (
            prodDataUrl, parse_date(data['updateTime']).strftime('%m/%d/%Y'))
        data = requests.get(url, timeout=5).json()

    df = pd.DataFrame(data['settlements'])[:-1][['month', 'volume']]
    df.columns = ['expiry', 'volume']
    df.volume = pd.to_numeric(df.volume.str.replace(',', ''))
    df.expiry = df.expiry.str.replace('JLY', 'JUL').apply(
        lambda ds: parse_date(ds).strftime('%Y%m'))

    try:
        df = df.reset_index().drop_duplicates(keep='last')
    except Exception:
        df = df.reset_index().drop_duplicates(take_last=True)
    df.columns = ['contract', 'expiry', 'volume']
    df['contract'] = df['contract'].shift(-1).fillna(
        df['contract'].max()+1).astype(int)

    return df.dropna()
    
if __name__ == "__main__":
    url = 'http://www.cmegroup.com/trading/equity-index/us-index/e-mini-sandp500_quotes_settlements_futures.html'
    contracts = get_contracts(url)
    active_contract = contracts[:1]